Child Language

Child Language
Acquisition and Development
2nd Edition
Matthew saxton
SAGE Publications Ltd
1 Oliver’s Yard
55 City Road
London EC1Y 1SP
SAGE Publications Inc.
2455 Teller Road
Thousand Oaks, California 91320
SAGE Publications India Pvt Ltd
B 1/I 1 Mohan Cooperative Industrial Area
Mathura Road
New Delhi 110 044
SAGE Publications Asia-Pacific Pte Ltd
3 Church Street
#10-04 Samsung Hub
Singapore 049483
© Matthew Saxton 2017
First published 2010
Reprinted 2010, 2011, 2012, 2015, 2016
Apart from any fair dealing for the purposes of research or private study, or
criticism or review, as permitted under the Copyright, Designs and Patents
Act, 1988, this publication may be reproduced, stored or transmitted in any
form, or by any means, only with the prior permission in writing of the
publishers, or in the case of reprographic reproduction, in accordance with
the terms of licences issued by the Copyright Licensing Agency. Enquiries
concerning reproduction outside those terms should be sent to the publishers.
Library of Congress Control Number: 2016962661
British Library Cataloguing in Publication data
A catalogue record for this book is available from the British Library
ISBN 978-1-4462-9561-8
ISBN 978-1-4462-9562-5 (pbk)
Editor: Luke Block
Editorial assistant: Lucy Dang
Production editor: Imogen Roome
Copyeditor: Sarah Bury
Proofreader: Christine Bitten
Indexer: Martin Hargreaves
Marketing manager: Lucia Sweet
Cover design: Wendy Scott
Typeset by: C&M Digitals (P) Ltd, Chennai, India
Printed in the UK
For Gary, Sue and Alex

Drawing of Alex Saxton by Colin Saxton

Contents
Acknowledgements
Publisher’s Acknowledgements
Notes on the Organization of this Book
Companion Website page
1 Prelude: Landmarks in the Landscape of Child Language
From burping to grammar in the pre-school years
Levels of language
Listen in mother
The cat in the hat in the womb
Some conclusions on sound
Word learning: From 0 to 14,000 in five years
Say ‘mama’
Estimating vocabulary size
The gavagai problem
Morphology: Bits and pieces
Syntax: Putting it all together
Language in context: Perceptual, cognitive and social development
The study of child language
The lie of the land
2 Can Animals Acquire Human Language? Shakespeare’s Typewriter
What is language?
The infinite monkey theorem
Language, talk and communication
The design of language
Teaching words to animals
Talking versus sign language
Lexigrams
Barking up the right tree: Word learning in dogs
Alex, the non-parroting parrot
Animal grammar
Combining words
Comprehension of spoken English by Kanzi
The linguistic limitations of animals
Is speech special?
Categorical perception in infants and primates
Statistical learning
 Back to grammar: Infants versus monkeys
The language faculty: Broad and narrow
3 The Critical Period Hypothesis: Now or Never?
What is a critical period?
A musical interlude
Lenneberg’s critical period hypothesis
Designing research on critical periods
Cats’ eyes: An example from animal development
How to identify a critical period
The effects of linguistic deprivation
The royal prerogative: Experiments on people
Feral children
Genie
Different critical periods for different aspects of language
A happier ending: The case of Isabelle
Conclusions from cases of deprivation
Age of acquisition effects in second language learning
Early versus late starters: Effects on language outcomes
Age effects may not be due to a critical period
Plastic fantastic: The receptive brain
Deafness and late language learning
Two more cases of linguistic deprivation: Chelsea and E.M.
Early versus late learning of American Sign Language
4 Input and Interaction: Tutorials for Toddlers
Talking to young children
Characteristics of Child Directed Speech
Phonology
Vocabulary
Morphology and syntax
A dynamic register
Individual differences and their effects
Child Directed Speech: Summary
Lack of interaction: Can children learn language from television?
Imitation
Linguistic creativity: Children make their own sentences
Skinner and Chomsky on imitation
Imitation as a mechanism in cognitive development
 Imitation: Who, when and how?
Individual differences in imitation
Corrective input
Recasts: Adult repetition of the child
The ‘no negative evidence’ assumption
Contrastive discourse
Negative feedback
Corrective input: Summary
Universality of CDS
Input and interaction in language acquisition
5 Language in the First Year: Breaking the Sound Barrier
Hunt the phoneme
In the beginning
Drops of sound in a river of speech
Categorical perception
Specialization towards the native language
Why I don’t speak Nthlakapmx
Loss or decline?
Enhancement of native contrasts
Individual differences in infant speech perception
Summary: Breaking the speech sound barrier
Word segmentation
The baby statistician
Learning in the real world
Prosodic cues to speech segmentation
Relative cue strength
Grammar from the babble
Phonemes, words and grammar: Summary
6 The Developing Lexicon: What’s in a Name?
Approaches to word learning
First words
Comprehension versus production
What do one-year-olds talk about?
Overextension
Categorically wrong
Lexical plugs: Pragmatic errors
Losing it: Retrieval failures
 Lexical processing
Up, up and away: The vocabulary spurt
Why so fast?
Spurt? What spurt?
The rate of word learning
Ten words a day?
Fast mapping
Slow mapping: The gradual accretion of meaning
Biases
The return of the gavagai problem
A noun bias in the child and in research
Nouns are easy
Verbs are hard
The shape bias
The rise and fall of word learning biases
Associative learning: The origin of biases?
Where do biases go?
Some lexical gaps
Computational modelling based on probability theory
7 The Acquisition of Morphology: Linguistic Lego
Inflection
The acquisition of inflection
Whole word learning
The past tense debate: Rules or connections?
A dual-route account: Words and Rules theory
The acquisition of words and rules
The blocking hypothesis
Words and Rules: The story so far
Connectionism and a single-route account
Problems with connectionist models
Crosslinguistic evidence
Summary: One route or two?
Compounding and derivation
Derivation
Compounding
Derivation: The influence of productivity
Early compounds
 Relating the parts to the whole in compounds
Complex compounds: Three processes combined
Morphology in the school years
Morphological awareness
Connections with vocabulary, reading and spelling
Morphological awareness in teachers
8 Linguistic Nativism: To the Grammar Born
Universal Grammar
The problem of linguistic diversity
Core versus periphery
Parameters of variation
Setting parameters: Triggers
Arguments for linguistic nativism
Some initial observations
Limited exposure to linguistic input
No direct instruction
Ease and speed of language acquisition
The poverty of stimulus argument
Plato’s problem
Degenerate input
Negative evidence: Corrective input for grammatical errors
Knowledge in the absence of experience: The case of structure
dependence
The origins of structure dependence
The imitation of grammatical structures
Evidence from children
Poverty of the stimulus: Summary
The contents of UG: What precisely is innate?
Conclusion
9 The Usage-based Approach: Making it Up as You Go Along
Language knowledge from language use
Social cognition
Dyadic and triadic interaction
Collaborative engagement and intention-reading
Collaborative engagement as a basis for language
development
Early constructions: A route into grammar
 In the beginning was the utterance
From single-unit to multi-unit speech
Does the child go from fully concrete to fully abstract?
The productivity puzzle
The transitivity bias
Pattern finding
Type frequency: A route to productivity
Sounds familiar: The role of frequency
Early productivity: Syntactic bootstrapping
Constraining productivity
Conservative learning
Entrenchment
Pre-emption
Summary: Reining back on productivity
10 You Say Nature, I Say Nurture: Better Call the Calling Off Off
Nature and nurture in the study of child language
The genetic basis of language development
Integrating ‘nurture’ into theories of syntax acquisition
Some basic facts
Dietrich Tiedemann (1787)
Child language: A timeline
The ‘nature’ in nature–nurture: Something must be innate
Learning mechanisms
Domain-general learning: How general?
Linguistic nativism: The need for learning mechanisms
Methodology: Limitations and possibilities
Language acquisition: The state of the art
Child language: Acquisition and development
Answers to Exercises
Appendix 1: Observations on Language Acquisition Made by Dietrich
Tiedemann (1787)
Appendix 2: Pronunciation Guide: English Phonemes
Glossary of Linguistic Terms
References
Author Index
Subject Index
Acknowledgements
I have had much help and encouragement in the writing – and updating – of
this book from family, friends and former colleagues in academia. It has also
been gratifying to receive help from colleagues further afield, in the wider
world of child language research, many of whom I have yet to meet, but all of
whom responded to my queries with considerable generosity of spirit. In
particular, I should like to thank Ayhan Aksu-Koç, Shanley Allen, Ben
Ambridge, Misha Becker, Heike Behrens, Ruth Berman, Raymond Bertram,
Joan Bybee, Robin Campbell, Shula Chiat, Anne Christophe, Alex Clark, Eve
Clark, Gina Conti-Ramsden, Annick de Houwer, Holly Garwood, Jonathan
Ginzburg, Roberta Golinkoff, Marisa Teresa Guasti, Bart Guerts, Margaret
Harris, Maya Hickmann, Josephine Howard, Christine Howe, Dick Hudson,
Jane Hurry, Evan Kidd, Sarah King, Mike Kirkman, Shalom Lappin, Elena
Lieven, Brian MacWhinney, Michael Maratsos, Theo Marinis, Chloe
Marshall, Cecile McKee, Evelyne Mercure, David Messer, Gary Morgan,
Vicki Murphy, Letitia Naigles, Keith Nelson, David Olson, Mitsuhiko Ota,
Anna Papafragou, Lisa Pearl, Colin Phillips, Li Ping, Dorit Ravid, Maritza
Rivera Gaxiola, Stuart Rosen, Caroline Rowland, Jenny Saffran, Colin
Saxton, Susan Sciama, Yasuhiro Shirai, Melanie Soderstrom, Morag Stuart,
Mike Swan, Michael Tomasello, Michael Ullman, Angelika van Hout,
Athena Vouloumanos, Catherine Walter, Dan Weiss and Charles Yang.
Beyond the professional, there is the personal. My husband, Gary Yershon, is
constant in his support. My son, Alex – a grown man now – features
throughout the book with examples from a diary study I did with him during
my doctoral studies. In the field of child language, it is not enough simply to
breed one’s own data. One needs also a son with a generous spirit who is
happy to share his early ventures into language with the wider world. My
father, Colin Saxton, drew the picture of Alex as a newborn which features as
the frontispiece. And my mother, Josephine Howard, created the painting
which has been used for the book cover. Finally, friends, family and
colleagues make numerous appearances in these pages, embedded in the
examples of linguistic structures. To them all, I offer my heartfelt thanks.
Publisher’s Acknowledgements
The author and publisher wish to thank the following for the permission to
use copyright material:
We thank APA for granting us permission to reproduce:
Baillargeon, R. (1987). Object permanence in 3½-month-old and 4½-month-
old infants. Developmental Psychology, 23(5), 655–664. Fig. 1, p. 656.
Ganger, J. & Brent, M.R. (2004). Reexamining the vocabulary spurt.
Developmental Psychology, 40(4), 621–632. Fig. 1. A spurtlike function
(logistic) superimposed on slightly modified data from Child 041B (p. 623).
Fig. 2. A nonspurtlike curve (quadratic) superimposed on the same data
shown in Figure 1, p. 623.
We thank Cambridge University Press for granting us permission to
reproduce:
Gershkoff-Stowe, L., Connell, B. & Smith, L. (2006). Priming
overgeneralizations in two- and four-year-old children. Journal of Child
Language, 33(3), 461–486. Figure 1, p. 464. Levels of processing and lexical
competition involved in naming a perceived object.
Dromi, E. (1987). Early lexical development. Cambridge: Cambridge
University Press. Figure 1, p. 111. Keren’s cumulative lexicon at the one-
word stage.
Bates, E. & Goodman, J.C. (1997, p. 517, Fig. 2) in E. Bates, I. Bretherton &
L. Snyder (1988). From first words to grammar: Individual differences and
dissociable mechanisms. New York: Cambridge University Press.
Smith, L.B. (2001). How domain-general processes may create domain-
specific biases. In M. Bowerman & S.C. Levinson (Eds.), Language
acquisition and conceptual development (pp. 101–131). Cambridge:
Cambridge University Press. Figure 4.1. Sample stimuli from Landau, Smith
& Jones (1988). All stimuli were three-dimensional objects made of wood,
wire, or sponge.
We thank Elsevier for granting us permission to reproduce:
A figure from Johnson, J.S. & Newport, E.L. (1989). Critical period effects in
second language learning: The influence of maturational state on the
acquisition of English as a second language. Cognitive Psychology, 21(1),
60–99.
Nicoladis, E. (2003). What compound nouns mean to preschool children.
Brain and Language, 84(1), 38–49. Fig. 2. ‘Sun bag’ target for
comprehension, p. 43.
Benasich, A.A., Choudhury, N., Friedman, J.T., Realpe-Bonilla, T.,
Chojnowska, C. & Gou, Z.K. (2006). The infant as a prelinguistic model for
language learning impairments: Predicting from event-related potentials to
behavior. Neuropsychologia, 44(3), 396–411. Fig. 1. Photograph of a 6-
month-old child seated on his mother’s lap during an ERP testing session
using a dense array Geodesic Sensor Net system (Electric Geodesic, Inc.,
Eugene, Oregon, USA), p. 399.
We thank the Linguistic Society of America for granting us permission to
reproduce Brooks, P.J. & Tomasello, M. (1999). How children constrain their
argument structure constructions. Language, 75(4), 720–738. Figure 1. Novel
‘directed motion’ and ‘manner of motion’ verbs, p. 724.
We thank MIT Press for granting us permission to reproduce Hirsh-Pasek, K.
& Golinkoff, R.M. (1996). The origins of grammar: Evidence from early
language comprehension, Figure 6.1, © 1996 Massachusetts Institute of
Technology, by permission of the MIT Press.
We thank Wiley-Blackwell for granting us permission to reproduce:
Gertner, Y., Fisher, C. & Eisengart, J. (2006). Learning words and rules:
Abstract knowledge of word order in early sentence comprehension.
Psychological Science, 17(8), 684–691. Figure 1, p. 686.
Saffran, J.R. (2003). Statistical language learning: Mechanisms and
constraints. Current Directions in Psychological Science, 12(4), 110–114.
Figure 1, p. 111.
Fenson, L., Dale, P.S., Reznick, J.S., Bates, E., Thal, D.J. & Pethick, S.J.
(1994). Variability in early communicative development. Monographs of the
Society for Research in Child Development, 59(5). Figure 1, p. 35 and Figure
2, p. 38.
Notes on the Organization of this Book
This text is aimed principally at students of psychology with an interest in
child language. It is suitable for use at undergraduate level, and also at
postgraduate level, in cases where the field is new. In both cases, I am keenly
aware that most psychology students have no prior training in linguistic
theory. In fact, if you’re like me – the member of a lost generation – you may
not even have learnt very much at all about language at school. For this
reason, I have tried to take nothing for granted as far as linguistic
terminology is concerned, not even with common items like noun or verb. Of
course, you can always skip over the linguistic interludes, if it’s all old hat,
and stick with the main event. Either way, the aim of this book is to equip
you to appreciate more fully the arguments and evidence advanced in the
study of child language. The following menu of pedagogic features should
sustain you on the journey ahead.
Glossary of linguistic terms
Linguistic terms are highlighted in bold to indicate their appearance in
the glossary, towards the end of the book. There you will find
definitions of all things linguistic. You can test your knowledge of
terminology via the related website, where you will find eFlashcards to
help you.
Pronunciation guide: English phonemes
A list of the special symbols used to represent the consonant and vowel
sounds of English. And yes, the terms phoneme, consonant and vowel all
feature in the glossary.
Boxes
Boxes have been used for two kinds of diversion from the main text: (1)
to expand on essential terminology from linguistic theory; and (2) to
provide extra information on key background concepts.
References and further reading
As well as the list of references at the end of the book, I have ended each
chapter with a few suggestions for further reading. These latter are
annotated with potted reviews and notes.
Website addresses
The internet makes life easy for students in all kinds of ways. But with
regard to reading material, articles found on the internet can be
intrinsically unreliable. The crux of the matter is this: one cannot always
tell, with any certainty, who wrote a given internet article. Nor can one
always be sure if the claims made in internet sources are reasonable,
valid, and backed up by reference to genuine and appropriate research.
That said, many sources are perfectly respectable – as testified by the
burgeoning number of electronic journals now available. I have been as
careful as I can in my listing of websites, but approach with caution. I
have included the academic web pages of some key child language
researchers, and these should be pretty reliable. In particular, many
academics now post downloadable versions of research articles on their
university homepages.
Discussion points
Discussion points are sprinkled throughout the book wherever they seem
like a Good Thing. They can be used in seminars or in student self-study
groups (never tried the latter? – give them a go). For some of the
Discussion points, you should equip yourself by reading the relevant
chapter in advance and/or reading an item from the Further Reading
section.
Exercises on linguistic concepts (with answers)
Like cod liver oil, linguistic exercises are unpalatable, but very good for
you. The idea is to limber up with some practice on unfamiliar concepts,
before tackling the literature. Model answers are provided at the end of
the book.
PowerPoint slides
These slides are intended for use by your lecturers. They fillet the main
points from each chapter, allowing room to expand on the main points in
classroom teaching sessions. But they could also quite easily be used for
private study, as the basis for identifying key points for revision and
reflection.
Multiple choice questions
You will find a set of MCQs for each chapter on the companion website.
They will help sharpen your understanding of key concepts. Or maybe
they’ll just reveal how lucky you are.
Author index
Find your favourite authors, as mentioned in the text, and source their
work in the list of references. Then challenge yourself to find other
work, especially recent research, by the same authors (your university
library will help if you’re new to the sport of Reference Hunting).
Subject index
Separate from the author index, because it makes life a little less
cluttered. Relevant topics from each chapter are included to enhance the
sum total of your learning happiness.
Companion Website page

The second edition of Child Language is supported by a wealth of online

resources for both students and lecturers to aid study and support teaching,
which are available at https://study.sagepub.com/saxton2e.
For students
Multiple choice questions to test your knowledge of key concepts and make
sure you have retained the most important parts of each chapter.
Weblinks direct you to relevant resources to broaden your understanding of
chapter topics and link to real conversations about social research.
Flashcard glossary so you can test yourself on all the key terms from the
book and check your understanding.
For lecturers
PowerPoint slides featuring figures, tables, and key topics from the book can
be downloaded and customized for use in your own presentations.
1 Prelude: Landmarks in the Landscape of
Child Language
Contents
From burping to grammar in the pre-school years  2
Levels of language  4
Listen in mother  5
The cat in the hat in the womb 5
Some conclusions on sound 7
Word learning: From 0 to 14,000 in five years  7
Say ‘mama’ 7
Estimating vocabulary size 8
The gavagai problem 9
Morphology: Bits and pieces  10
Syntax: Putting it all together  13
Language in context: Perceptual, cognitive and social development  16
The study of child language  18
The lie of the land  22
Overview
By the end of this chapter you should have some appreciation of the
challenges facing the newborn infant in the acquisition of language.
Major landmarks in language development are presented at each of the
following four levels of linguistic analysis:
phonology (the sound system)
vocabulary
morphology (parts of words, especially those parts used for
grammar)
grammar
We will consider some of the philosophical problems facing the child
(for example, how does the child know what a word is, or what it might
refer to?). And we will introduce the nature–nurture problem: to what
extent do genetic factors determine the development of language? To
set the child’s language learning achievements in context, an overview
is provided of the child’s achievements in other developmental
domains (cognitive, perceptual–motor and social), before sketching out
the contents of the chapters that follow.
From burping to grammar in the pre-school years
Have you ever had a chat with a toddler? A rather precocious two-year-old,
known as Eve, came out with the following one day, while in conversation
with her mother:
(1) Eve aged two years (Brown, 1973):
he go asleep
want more grape juice
putting sand in the pail
I have get my crayons
where other baby?
We can see straight away that Eve is fairly savvy about a number of different
topics. But however impressed we are by Eve’s knowledge of crayons, sand
and juice, it is clear that not one of the sentences would pass muster if they
were uttered by an adult in the same setting. If nothing else, this reminds us
that language develops.
As we shall discover in this book, though, Eve has already come a long way
by the age of two years. A typical newborn is capable, vocally, of no more
than reflexive crying and fussing, plus a small repertoire of vegetative
sounds, principally, burping, spitting up and swallowing (Stark, 1986). This
repertoire is lent some charm at about eight weeks, with the emergence of
cooing and laughter. But if we fast forward to the typical five-year-old, then
we suddenly find ourselves in the company of a linguistic sophisticate,
someone with an extensive vocabulary who is able to put words together in
interesting, complex sentences that, for the most part, are perfectly well
formed.
(2) Ross aged 5;1 (MacWhinney, 2000):
I had the worst dream of my life
I wish I could let you in here, but there’s no room
You thought he couldn’t go to school because we didn’t have the
medicine
Box 1.1 Notation for the Child’s Age
There is a standard convention for denoting a child’s age in the
child language literature. A child aged two years, three months
would be recorded as 2;3. The child of four years, six months
appears as 4;6, and so on. Note the use of the semi-colon to
separate years from months. When even more fine-grained
 analyses are required, we can also add the number of days after a
period (.). For example, 1;9.10 is read as one year, nine months
and ten days.
This shorthand, in which we note months as well as years, is very
useful. Things can move fast in child language and important
distinctions might otherwise be lost. For example, take two
children, both aged one year. The first child, aged 1;0, might not
yet have produced her first word, whereas the second child, aged
1;10, might already be stringing multi-word utterances together.
Eve aged 1;10 (Brown, 1973):
Sue make some
oh my Graham cracker broke
here Fraser briefcase
have to drink grape juice first
The acquisition of language is a staggering feat. It is all too easy to overlook
the monumental nature of this achievement, because language learning seems
to come so easily to all typically developing children. Perhaps we take the
miracle of language learning for granted because, as adults, we typically take
the possession of language itself for granted. Every cognisant reader of this
book has an extensive, complex, rich knowledge of language. But this
knowledge is such universal currency – so very much part of everyday life –
that we often fail to notice or appreciate the great gift it affords the human
species. Exercise 1.1 (below) throws a spotlight on the position of language
in human society.
Exercise 1.1
Imagine a world without language. Consider the world we live in and
consider the ways in which we depend on language. In some ways, this
is an incredibly easy task. In others, it is quite overwhelming. Once you
begin, there seems to be no end to the ways in which we rely, either
directly or indirectly, on our ability to communicate with language.
Write down two or three topics that characterize some aspect of the
human experience (I provide a few suggestions below, but don’t feel
restricted by these). Consider the ways in which they depend on
language for their existence.
bridges
governments
 a family meal
travelling from London to Paris
football
laws
taking a shower
gardens
Levels of language
In this section, we will try to get a sense of the magnitude of the task facing
the newborn child. The first thing to note is that the child is battling on
several different fronts at once. Language has different components, or levels,
each of which must be tackled.
The Big Four levels of language:
phonology: concerned with the sounds of speech
vocabulary: the storehouse of meaning (words)
morphology: bits of meaning encoded in the grammar, like the plural
ending, -s in dogs
grammar: the rules dictating how words are put together into sentences
By way of hors d’oeuvre, the current chapter will sample from each of these
key areas of language, to get some flavour of how the child tackles them.
Our division of language into four levels stems from linguistics, the study of
language. In academia there are typically many different ways to dissect an
issue in analysis. It all depends on one’s theoretical perspective. For example,
we could add pragmatics to the list above, if we believe that the way
language is used is especially important (see Chapter 4 for more on this). Or
we might reduce our list to just two factors: meaning and sound. We know
that language is used to communicate meanings. And we know that,
typically, the human vocal apparatus is used to transmit those meanings via
sound. The study of language (and also child language) could thus be reduced
to working out how meaning and sound are connected (Chomsky, 1995).
From the perspective of child language, our problem is not just one of
deciding how to cut up language into its component parts (see Chapter 2). It
is also one of working out how they interconnect and influence each other.
Highlighting different aspects of language is useful. At the same time,
though, there are times when the divisions seem artificial, especially if we
focus on one level at the expense of another. For example, we could examine
the development of phonology in isolation. How does the infant come to
distinguish one sound from another in the torrent of speech assailing their
ears? This is a big question. But if we concentrated solely on the problem of
phonology, we would fail to notice that the child is also learning something
about words, and even grammar, in the first year of life, long before they ever
produce speech of their own (see Chapter 5). Unfortunately, research has
tended to consider each level of language independently, as though the child
had a series of different boxes to tick off in a particular order. Traditionally,
this has been seen as phonology first, followed by words and then
morphology and finally, grammar (Bates & Goodman, 1997). We can still
examine each level of language in its own right (and the chapters in this book
follow that general pattern), but we will be mindful that the levels of
language do not comprise the rungs of a ladder for the child to ascend.
Simultaneous development and mutual influence are more characteristic of
language acquisition.
Listen in mother
Imagine what it is like to hear language for the first time. No, really: imagine.
Did you picture yourself in a crib listening to your mother? Understandable,
but think again. We hear language before we are even born. As you will
know from those noisy neighbours who drive you mad, sound passes through
solid barriers – not just the walls of houses but also through the wall of the
womb. And it has long been known that the human ear begins to function
several weeks before birth, in the third trimester (or third) of pregnancy at
about seven months (Sontag & Wallace, 1936). The foetus can respond to the
sound of bells, and can even discriminate between different tones. But
sensitivity to sound is not the same as sensitivity to language. Can the foetus
distinguish noises, like a power drill or the banging of a door (those
neighbours again), from the sound of their own mother’s voice? Remarkably,
the answer is ‘yes’. Moreover, the foetus has already begun to recognize the
distinctive properties of their native language (May, Byers-Heinlein, Gervain
& Werker, 2011). The brain responses of newborn infants (0–3 days old)
differ when they hear a foreign language (in this case, Tagalog from the
Philippines) versus the language heard in the womb (English). Even more
remarkable, the unborn baby can learn to recognize the telling of a particular
story.
The cat in the hat in the womb
In an ingenious experiment, DeCasper & Spence (1986) asked women in the
later stages of pregnancy (7½ months) to make recordings of three different
children’s stories. One was the first part of the classic 1958 Dr Seuss story,
The cat in the hat. The second story, The dog in the fog, was created by
adapting the last part of The cat in the hat, including, as you can see from the
title, some changes to vocabulary. The third story, meanwhile, The king, the
mice and the cheese, was unrelated to the first two, but all three stories were
of very similar length, with equal-sized vocabularies. Even a good proportion
of the actual words in all three stories was shared (about 60–80 per cent of
the total, depending on which two stories are compared). But note, The dog in
the fog shared the rhythmic properties of The cat in the hat, while The king…
story did not. The influence of rhythm was thus controlled for in a study that,
in many ways, provides an excellent model of well-designed research.
The pregnant women were asked to read just one of the three stories out loud
twice a day which meant that, on average, they recited their particular story
67 times prior to giving birth. At just three days old, these neonates displayed
a clear preference for the particular story they had heard in the womb. You
might wonder how a newborn baby tells you which story they want to hear.
Infant interest was measured via their sucking behaviour on a nipple, with
high rates of sucking being taken as a sign of increased attention (see Box
5.1, Chapter 5). For example, babies exposed in the womb to The cat in the
hat sucked more fervently when they heard this story played to them than
when either of the other two stories were played. The same result was found
irrespective of the story: newborns prefer the particular story they have
already heard in the womb. This finding is remarkable, but what can we
conclude about the linguistic abilities of the foetus? No-one would claim that
these infants knew anything about the characters in the story or the
development of the plot. But clearly, something rather specific about the
story was familiar to them.
Infant preference was unlikely to be influenced by either the length of the
story, or its rhythmic qualities, or the particular vocabulary used. As noted
above, these factors were controlled for. We can also rule out voice quality as
the source of infant interest. As you will know, one human voice is distinct
from another. The unique qualities of each person’s voice allows us to
recognize family, friends, or even celebrities’ voices (Philippon, Cherryman,
Bull & Vrij, 2007). Given this ability, it is conceivable that the infant’s
display of interest (high amplitude sucking) is driven by preference for the
mother’s voice (not the story). As it happens, infants do prefer to listen to
their own mother’s voice (Mehler, Bertoncini, Barrière & Jassik-
Gerschenfeld, 1978), but that was not the critical factor here. We know this
because every story heard by every infant in every condition was read to
them in their own mother’s voice. So their marked preference for one story
over the others was not based on voice quality or recognition of the mother.
Instead, we are left with intonation as the critical factor. One might think of
intonation as the ‘music’ of speech. Modulations in pitch, up and down,
during the course of an utterance provide speech with its melody. It turns out
that the unborn child can perceive and later recall very specific information
from the intonation contours produced in the telling of each story.
A further point of interest is that foetal sensitivities to speech intonation seem
to be confined to low-frequency sounds (think bass guitar). This restriction
might point to an immature hearing system in the foetus. But the more likely
explanation is that the quality of sound that the foetus hears through the
barrier of maternal body tissue (including the womb wall) is degraded. In
particular, high-frequency sounds (think school recorder) do not penetrate the
womb very well. We know this from a further study by Spence & DeCasper
(1987). They found that newborns prefer to hear a low-frequency version of
their mother’s voice. This special version was created by filtering out high-
frequency sounds to simulate what the baby would have heard in the womb.
This latter study thus refines the discovery that children can tune into
language and form memories of their experience even before birth. The
prenatal experience of the mother can also have an impact on the child’s
language development. A recent study on a large cohort (>34,000) mothers in
Norway reveals that prenatal distress in the mother is associated with
negative language outcomes for the child at three years of age (Ribeiro,
Zachrisson, Gustavson & Schj⊘lberg, 2016). Fortunately, though, this effect
is not very large.
Some conclusions on sound
So far, we can make at least four interesting observations about language
development. First, it starts before the child is born. Second, the foetal brain
is well equipped to process human speech sounds. Third, whatever
specialization or ‘mental equipment’ the child is endowed with, specific
experiences have a large impact on learning and memory, again, before the
child is born. We can see this both in the preference for the mother’s voice
and also in the preference for a particular story. Fourth, DeCasper & Spence’s
study provides a really good demonstration that, when it comes to unlocking
the secrets of child language acquisition, considerable ingenuity is required.
Psychology, in general, faces the fundamental problem of understanding what
goes on inside someone’s head. Even if we cracked open the skull for a close
look at the brain, it would not get us very far in understanding how the mind
works or develops. In child language research, we have the added frustration
that we lack one of the most obvious tools for investigating the mind:
language. In adult psychology, I might ask you directly about your attitudes
or personality. But we cannot use language to ask the foetus or the newborn
or even a toddler to use language themselves, to tell us anything about how
they acquire language. Thinking about language, so-called metalinguistic
skill, does not emerge in any sophisticated way until the child is about five
years old (Gombert, 1992). Ironically, then, language is not a tool we can use
to examine language learning in the young child. This raises the bar very high
for empirical study. But happily, you will discover that there are numerous
examples of both great ingenuity and good science that overcome this
problem.
Word learning: From 0 to 14,000 in five years
Say ‘mama’
One of the great landmarks in a baby’s life happens around the time of their
first birthday: the moment when they produce their first word. Parents are
typically captivated on this occasion and make a great fuss over what is,
indeed, a major achievement. As you might guess, the first word is very often
the name for ‘mother’ or ‘father’ (including mummy, mama, daddy, papa). A
word like mama is relatively easy for the 12-month-old to pronounce. In fact,
it often arises spontaneously in the child’s babbling some time before its
appearance as a word. This may happen because mama is composed of
simple sounds, arranged into repetitive strings of simple syllables (Goldman,
2001). Parents are quick to ascribe meaning to the baby’s vocal productions
and will be very easily convinced that mama means ‘mother’. Although some
version of ‘mother’ is a common first word, it should be pointed out that
children differ. Some children produce their first word a little earlier than 12
months: an acquaintance of mine, Arthur, started saying bye at the age of 10
months. Many other children, meanwhile, do not speak until some time later.
As we see from Arthur, the child’s first word is not inevitably some version
of ‘mother’ or ‘father’. My own son, Alex, produced his first word, juice, at
the age of 15 months. His second word only appeared three months later, at
Christmas time, when he joined us in saying cheers. He was still on the juice,
by the way: there’s no point wasting good champagne on toddlers. The idea
that children differ with respect to language learning is an important one.
Accordingly, we will pick up the theme of individual differences, where
appropriate, as we proceed.
Once the child has cracked the problem of the first word, many other words
follow. At first, new words are acquired at a fairly gentle rate, something like
one per week on average. But then things speed up, once the child has passed
another, somewhat more intriguing milestone: the accumulation of roughly
50 words. For many children, a step-change in the rate of word learning takes
place at this point, with words being added to the total store at the rate of one
or two per day (Tamis-LeMonda, Bornstein, Kahana-Kalman, Baumwell &
Cyphers, 1998). Between the ages of two and six years, many children then
step on the gas again, with something like 10 new words per day being
learned (Clark, 1993). This huge increase in rate of word learning is often
described as the vocabulary spurt. As we shall see in Chapter 6, there are
individual differences among children in both the rate and pattern of word
learning (Bloom, 2000). But the notion of a vocabulary spurt enjoys
widespread currency as a general feature of child language. Ten words per
day is a lot. Think of the last adult evening class you attended where you
tried to learn a new foreign language (or why not give it a go, if you’ve never
tried). Think how hard you would find it to acquire 10 new words every
single day for four years. And when I say ‘acquire’, I mean learn, remember,
retrieve at will and use with facility and accuracy in a range of settings. Seen
in this light, the child’s word learning prowess in the early years is deeply
impressive, resulting in a vocabulary somewhere between 10,000 (Bloom &
Markson, 1998) and 14,000 (Clark, 1993) by the age of six.
Estimating vocabulary size
10,000 or 14,000 words? The wide discrepancy in these two estimates hints
at a tricky empirical problem: how does one begin to estimate the number of
words in someone’s head? To get some idea of the problem, open a
dictionary at random and scan the words on the page. My guess is that the
words which catch your eye will be the new or unfamiliar words. But I would
also guess that you do, in fact, know a good many words on any given page,
at least to some degree. Given that dictionaries often contain thousands of
pages, on each of which you will know many words, you can begin to
calculate the extent of your vocabulary (see the website listed at the end of
this chapter for information on a more rigorous method). Arriving at this
estimate is not straightforward. Among many factors that have to be
entertained, there is the fundamental question: what constitutes a word? In a
dictionary, the verb cogitate will typically appear under its own heading. It is
unlikely, though, you will find related forms like cogitates, cogitating or
cogitated. But each one of these versions looks like a separate word.
Whichever way we decide to count, it should be clear that your vocabulary is
enormous (Kučera, 1992). At birth, your vocabulary was zero. By the time
you left school it was more like 50,000 (Nagy & Anderson, 1984). Individual
differences aside, we all make quite a leap. In Chapters 5 and 6, we will
consider how the child manages the feat of word learning. But first, we will
consider an even deeper mystery. How does the child ever manage to work
out the meaning of even a single word?
The gavagai problem
We have seen that it is no straightforward matter to define what a word is.
Equally vexing is the issue of what words mean. Take rabbit. In pronouncing
this word, rabbit is a sequence of sounds that comes out of my mouth. Its
special status as a word derives from my ability to use this sound-string
symbolically. The sounds are used to ‘stand for’ an object in the world. In
this way, words provide the vehicle for pairing sound with meaning.
Critically, the pairing we have made between the sound string and the object
is arbitrary: there is no intrinsic or necessary relationship between the
particular sounds and the particular object denoted (de Saussure, 1916/1974).
This becomes obvious when you consider that, in languages other than
English, reference to the object would be made using different combinations
of speech sounds (for example, lapin in French, krolik in Russian, and so on).
So far, our definition of word comprises a simple pairing between a sound
pattern and an object (or referent). But maybe it is not quite that simple.
W.V.O. Quine, an American philosopher, complicated matters (as
philosophers are wont to do), when he conjured up the image of an intrepid
linguist who comes across a culture whose language is entirely foreign to
him. The linguist tries to learn all he can about this new language. Then one
day …
a rabbit scurries by, the native says ‘Gavagai’, and the linguist notes
down the sentence ‘Rabbit’ (or ‘Lo, a rabbit’) as tentative translation.
(Quine, 1960: 29)
But how good is this translation? Quine demonstrated that we cannot take the
linguist’s translation for granted. Conceivably, the ‘native’ was using gavagai
in a predictable way, to mean the type of animal that scurried by. But gavagai
might equally well refer exclusively to white rabbits (not grey), or it might
mean rodent or even a special sub-class of rodents (important to the native’s
culture) comprising only rabbits, gerbils and coypu. Gavagai might
conceivably refer to the rabbit’s eyes, or to the rabbit’s feet. Gavagai might
refer to the rabbit as it appears in a particular light (sunshine rather than
shade, say). Gavagai might be confined to a rabbit in mid-scurry, not when it
is stationary. And so on, ad infinitum. From Quine’s discussion of the
translation problem, we learn something important about word learning. How
does the infant work out what an adult means when they use a particular
word? Does mummy refer exclusively to the woman who gave birth to the
infant? Or to both parents? Or does it mean the hair on a particular woman’s
head? Or the hair plus her left ear, viewed in the autumn light? Or …? As you
see, the infant faces the gavagai problem. We have no good grounds, in
advance, for assuming that the child will ascribe the same meaning to a word
that we would.
Quine’s gavagai raises several more fundamental challenges in the study of
child language (Bloom, 2000). First, how does the child know that a word
such as rabbit is the name for something? Without prior information about
what a word is, one might simply assume that the speaker was making a
random noise or expressing emotion. If we ignore that problem for a moment,
a second issue is to identify individual words. How does the child come to
know (without prior knowledge or help) that rabbit is one word and not two
(rab and it, perhaps)? This second issue is an example of the segmentation
problem: we need to explain how the child cuts up a continuous stream of
speech into appropriate linguistic units (see Chapter 6). Third, Quine’s
gavagai is untypical because it is a single-word utterance. Although perfectly
possible, words are not generally used in isolation. We are much more likely
to string words together into phrases and sentences. We need to consider,
therefore, how the child extracts each new word and its meaning from
running discourse. We must also deal with the issue of generalization. You
may have noticed, above, that my notion of a typical meaning for gavagai
was not confined to the particular rabbit that scurried past. It embraced the
type of animal, that is, all rabbits. In other words, I would normally expect to
extend (or generalize) my use of gavagai to refer to different rabbits that
might crop up in the future. What about the infant? Even if the child is able to
extend the use of gavagai, how do they know what to extend it to? Again,
there is an infinity of possibilities. The word could extend to just white
rabbits, or rabbit ears, or rabbits with eating disorders, and so on. All in all,
words and their meanings seem to present the child with a massively
daunting learning problem.
Are the problems of word-learning insurmountable? Evidently not. Every
typical child solves the gavagai problem with alacrity. We can assume this
from the simple observation that children acquire words quickly and easily,
once they get started (though see Chapter 8 on what might count as ‘quick’ in
language learning). Put another way, if children acquired word meanings in
ways that, according to Quine, are logically possible, then we would expect
linguistic chaos to break out. For example, the child who decided that rabbit
is confined exclusively to ‘the pet we keep in a hutch in the garden’ would
soon find themselves incredibly restricted in their ability to use that word.
Rabbit would not be much use when watching Bugs Bunny, to name but one
limitation. At the same time, the child would presumably hear people using
the word, but in different (conventional) ways that would not make any sense
to the child. Communication would be impossible. As noted, though, we can
be pretty confident that this kind of chaos does not break out. We must
conclude, therefore, that something clearly prevents the child from
entertaining wild hypotheses about word meaning, even though they are
logically possible. What is it that constrains the child’s word learning? We
will explore this issue in Chapter 6.
Morphology: Bits and pieces
Words seem like neat, self-contained linguistic bundles. In speech, I can
pronounce any word in isolation (even if, as we’ll see in Chapter 5, this is
quite rare). On the page, each word is surrounded, and therefore demarcated,
by white space on either side. Hence it would be easy to mistake words as the
smallest meaningful units of language. In fact, though, we can dig deeper.
Words can often be subdivided into separate parts, known as morphemes,
and it is the morpheme that constitutes the smallest unit of meaning in a
language.
As the examples in Table 1.1 show, some morphemes can stand alone as
single words (tree or flower). But others cannot exist independently and must
be attached (or bound) to a lexical stem. An example of a bound morpheme is
the -s that can be added to tree to make trees, where the -s morpheme denotes
a plural meaning. The existence of morphemes is one reason why the concept
of word is so difficult to define. If you recall, we cited above the example of
cogitates, cogitating and cogitated, wondering, in the process, if we were
dealing with three words or just one. One solution is to recognize that there is
one lexeme (cogitate), and three word forms, each one inflected with a
different morpheme (see Figure 1.1). In each case, the inflection serves a
different grammatical function, specifying more precisely the meaning to be
conveyed (see inflectional morpheme for more on this).
Even from this very brief sketch, we can see that morphology complicates
matters. For the language-learning child, it is not enough to learn a verb like
jump. The child must also be able to distinguish between the different
meanings and uses of different forms of the verb (jump, jumps, jumping,
jumped). In English, this is difficult enough. But some languages have far
more complex systems of inflectional morphology than English. Russian, for
example, would make your eyes water (Wade, 1992, will really make the
tears flow). Russian nouns can be marked in a bewildering number of ways,
because, to start with, there are at least six cases: nominative, accusative,
genitive, dative, prepositional and instrumental. These different noun forms
each fulfil a different grammatical function, like subject or object. Beyond
case, nouns select from two forms for number (singular or plural) and three
forms for gender (masculine, feminine and neuter). And then there are
different markings for adjectives and verbs, too. We do not need to dwell on
the grammar of Russian nouns (sighs of relief all round), but we can at least
pause to consider the monumental task facing some of the world’s children.
Do children really have to learn all these different morphemes? Well, maybe
not. Conceivably, each inflected form could be learnt as a separate word,
together with its specific meaning. Provided the child learns the difference in
meaning between jumps and jumping, for example, then internal analysis of
each word might not be necessary. Of course, there is a lot of redundancy in
this approach. It would mean learning a huge number of separate word forms,
especially in a language like Russian. But then again, a huge capacity for
word-learning is something we possess, both as children and as adults. What
do children do? Are they whole-word learners? Or can they dissect words
into their constituent parts and extract the separate morphemes?
Figure 1.1 The meaning of three inflectional morphemes in English

Jean Berko Gleason addressed this question some sixty years ago in what is
probably the first experiment in child language research (Berko, 1958). Berko
reasoned that, in the ordinary run of things, we cannot tell if a child has
acquired inflectional morphemes or not. We have seen that, in English, we
can mark the third person singular of a verb by adding -s (jumps). We can
also add -s to the end of a noun, this time to denote plurality (one apple
versus seven apples). How do children learn to mark plurality? They could
assemble the word by applying a rule: apple + s → apples. Or they could
learn apples as a single, indivisible unit. To distinguish between these two
possibilities, Berko devised a series of novel creatures, depicted in line
drawings, each with a novel name, like wug or niz (for more on nonsense
words see Box 4.2). The child is first shown a picture of a single creature and
hears: This is a wug. Next the child sees a picture of two creatures and hears
the following: Now there is another one. There are two of them. There are
two _. The child is thus invited to complete the sentence. Observe that the
child has only ever heard the uninflected form wug (and then only once).
Because the experimenter is controlling the input to the child, at no point
does the child hear the plural form, wugs. In consequence, if the child can
complete the test sentence successfully, then they must be compiling the
correct plural form, wugs, online, that is, at the time of production.
The child cannot have simply dragged wugs out of memory as a single,
unanalysed chunk of language, if only because it cannot have entered
memory in the first place. In the event, Berko found that children were very
good at this task. Ninety-one per cent of children aged 4–7 years produced
the correct plural form. By the time children reach school, therefore, they
seem to have a productive morphological system which they can apply to
words never before encountered. A character in James Joyce’s Ulysses,
Stephen Dedalus, remarks that ‘I fear those big words which make us so
unhappy’ (1922/1960: 37). Perhaps Dedalus could have cheered himself up
with morphology. All he had to do was break down those daunting big words
into their separate morphemes. We will guarantee our own happiness by
picking up this theme again in Chapter 7, where we consider how children
acquire and access the productive power promised by morphology.
Syntax: Putting it all together
Words by themselves, even without the aid of morphology, have considerable
expressive power. But consider what happens when we start putting them
together. At 1;6, the little girl we met at the start of this chapter, Eve,
produced the following utterances:
One word Two words Three words
baby Eve writing doll eat celery
there block broke Eve find it
coffee Fraser water read the puzzle
eye more cookie man have it
hat that radio man no spoon
Notice how Eve can be much more precise in her meaning when she
combines even just two or three words. In contrast, the single word utterance,
baby, is extremely vague. It might be taken to mean I’m a baby or The baby’s
on the loose or Do you think they’ll ever give babies the vote? and so on.
Without being in the room with Eve, it’s difficult to discern her intention. But
with two or three words, the picture is much clearer and contextual support is
less vital for working out Eve’s meaning. Eve writing probably means Eve is
writing, while Eve find it might well be glossed as Eve can find it. Putting
words together into utterances is therefore useful for communicating specific
ideas. It is no wonder, then, that ‘the words of the world want to make
sentences’ (Bachelard, 1960/1971: 188). And to make sentences, we need
syntax, the set of principles or rules which dictate how words can combine.
See Box 1.2 on syntactic categories, and then take a breather with Exercise
1.2 below.
Box 1.2 Syntactic Categories
The words of a language belong to different syntactic categories,
like verb, noun, adjective and preposition (explained
individually in the Glossary). One way of distinguishing these
different categories is to think about the kinds of meanings each
one conveys. Thus, verbs like run, fly and chase might be thought
of as ‘action words,’ while nouns are often associated with objects
(tree, book, house). But this only gets us so far. The category of
verb also includes words like anticipate and procrastinate, which
are not easily described as actions. Similarly, purity and truth are
classed as nouns, though they do not denote concrete objects.
To get round this problem, linguists rely on two further criteria for
identifying syntactic categories: (1) possible word forms; and (2)
behaviour in phrases. Word form (an aspect of morphology)
refers to the way we can change the shape of words in particular
ways. Each syntactic category permits its own set of characteristic
endings (inflections). The verb shout can appear as shouted or
shouting; a noun like table can also occur as tables; and an
adjective like red can appear as redder or reddest. We cannot just
add any inflection to any category (for example, shoutest).
We can also determine which category a word belongs to by
looking at its behaviour in phrases, that is, the combinations that
are allowed with other words. Words that can substitute for one
another in the same slot within a sentence belong to the same
syntactic category.
Hanadi does nothing but_________
The empty slot calls for a verb, so I could insert complain, belch,
or procrastinate and come up with a grammatical sentence. The
outcome would not be grammatical if I tried inserting a noun
instead (e.g., Hanadi does nothing but tree). We can take this
observation one step further by considering whole phrases, that is,
groups of words that can function as a single unit, and which
therefore constitute syntactic categories themselves. For example,
a Noun Phrase constitutes a group of words with a noun as the
key constituent (or head), for example, a tiger or the old dark
house. Like our single-word categories, noun phrases can only
 occupy certain slots within sentences.
Ben wants to buy_________
Both of our noun phrases, a tiger and the old dark house, could
happily fill the empty position in this sentence frame. But other
syntactic categories, like verb, would fare less well (for example,
Ben wants to buy think).

Exercise 1.2: Grammatical categories

You may want to start by checking the Glossary to make sure you’re
up to speed with the following four terms: noun; verb; adjective;
preposition. Then use these four categories to sort the following
words. If you have trouble, try using the tests of category membership
described in Box 1.2. Answers at the back of the book.
dog in eat hot
by happiness Somerset from
inculcate divorce pusillanimous happy
acceptance complete up charming
Back to Eve. The imperfections in her utterances are due, in part, to missing
morphemes. Compare doll eat celery with The doll is eating celery (more on
this in Chapter 7). One thing Eve does seem to be getting right, though, is the
grammatical feature of word order. We cannot just put words together in any
old order. Think back to Example (2). Instead of saying I had the worst
dream of my life, Ross might have thrown the words together at random: *The
of dream life had worst I my. (The asterisk (*) is used to denote an
ungrammatical sequence.) Word order is especially important in English, as
we can see from the study illustrated in Figure 1.2.
If I was two years old, I’d be terrified by these characters. They look like
something out of a bank heist movie. Happily, though, the two-year-olds
doing this study were unfazed. Observe that we have another novel word, the
verb gorp. If I were to tell you: The duck is gorping the bunny, which picture
would you choose as most appropriate? I’m guessing the one on the right.
But why? The answer is to do with agency. On the right, the duck seems to
be actively doing something to the rabbit (gorping, in fact). But on the left, it
is the rabbit who seems to be the agent (the doer) of the action (notice how
the meaning of gorp is different in each case). For The duck is gorping the
bunny, we choose the right-hand picture because the default in English syntax
dictates that the agent of the verb comes first in the sentence. So we look for
a picture where the duck is the agent. Children as young as 2;1 do the same
thing, provided they are presented with typical, default sentences (Hirsh-
Pasek & Golinkoff, 1996; Gertner et al., 2006; Dittmar, Abbot-Smith, Lieven
& Tomasello, 2011). Consider what happens if we change our test sentence
to The rabbit is gorping the duck. Suddenly, the left-hand picture becomes
the appropriate illustration. All we have done is switch duck and rabbit
around in the sentence. But see what a profound effect this change in word
order has on the meaning. How do children acquire this knowledge of syntax
so early in life? We will explore this question further in Chapters 8 and 9.
Figure 1.2 ‘The duck is gorping the bunny’ (adapted from Gertner, Fisher &
Eisengart, 2006)

Language in context: Perceptual, cognitive and

social development
How do millions of typically developing children get to the point where they
can produce and understand an infinite number of different sentences? Let’s
find out. But before we get started, have a glance at Table 1.2, where you will
find some of the key milestones in child development over the first few years
of life. This table allows you to compare progress in physical, cognitive,
social and linguistic development. The first thing to say is that these are not
separate streams in development, running in parallel. There are numerous
overlaps along the way. To take just one example, it has been shown recently
that once the child starts walking, their language development takes off,
much like the child (Walle & Campos, 2014). Table 1.2 might give the
impression that the child’s achievements in language are especially
impressive. For example, at the age of four years, the average child has a
digit span of just four (Kail, 1990). That is, they can reliably repeat back a
random list of four single-digit numbers, for example, 6 – 2 – 3 – 9. But if we
added just one more digit, the same four-year-old would struggle to repeat
them back correctly. Our average four-year-old also assumes, erroneously,
that other people share their own perspective on the world. They believe that
other people see (and know) the same things that they see, from the same
vantage point (Happé, 1994). In some rather fundamental sense, therefore,
the child’s cognitive representation of the world is qualitatively different
from that of an adult. In linguistic terms, though, the four-year-old is much
closer to an adult. The differences are more a matter of degree than kind. For
example, an obvious difference between child and adult is vocabulary size. In
terms of grammar (both syntax and morphology), there is perhaps less to
distinguish a four-year-old child and an adult.

Exercise 1.3
I began this chapter by asking if you had ever chatted with a toddler.
Even if you have, it would be a good idea to repeat the experience, this
time focusing on the child’s language. There is no substitute for real
data and talking to a young child will bring the subject to life in a way
that no textbook can. But you need to take seriously the sensitivities of
this task. Consult established guidelines for research (e.g., the Code of
Ethics and Conduct published by the British Psychological Society,
www.beta.bps.org.uk). Unless you enjoy the luxury of being parent to a
young child, access should be considered very carefully, even if you
 are related to the child in question. The mother (or primary caregiver)
should be present at all times for fairly obvious reasons, but also to
better ensure that the child feels sufficiently comfortable to talk freely
with you. If you can secure a conversation with a toddler, consider the
following:
What kinds of words and phrases does the child know?
How easy is it to make yourself understood?
What limitations are there in the child’s ability to communicate?
Are there any differences in the quality of interaction between
mother and child, on the one hand, and between yourself and the
child, on the other?
Now compare the way you talk with the child with the way you
talk with the parent. How does the interaction differ,
linguistically, from one conversational partner to the other?
The examples raised here make the child look linguistically fast, but
cognitively slow. But this is partly because of the examples that have been
chosen. Consider, instead, object permanence. Object permanence is the
knowledge that an object remains the same object when viewed from
different angles or under different lighting conditions, and that an object
continues to exist even when out of sight. Many authors argue that infants
attain an understanding of object permanence much earlier than the seven
months suggested in Table 1.2 (for example, Rochat, 2001). But I have been
cautious, because seven months is the earliest point at which the infant can
demonstrate their knowledge of object permanence in their own actions. But
even with this caution, it is apparent that the infant has a sophisticated
understanding of an object’s physical properties long before they utter their
first word. Now, all of a sudden, the child’s linguistic development looks
comparatively slow. Evidently, it is no easy matter deciding how rapid and
easy child development is, in any domain, be it linguistic, perceptual,
cognitive or social. We return to this theme in Chapter 8.
The study of child language
The overview of child language sketched out in this chapter sets the scene for
the rest of the book. Before we plunge into detail, though, it is worth
remarking on how far the study of language acquisition has come over the
years. Perhaps the first published study on child language was by a German
biologist, Tiedemann, in 1787 (see Murchison & Langer, 1927, and
Appendix 1). But until the late 1950s, it is fair to say that child language
research was not especially systematic or scientific in its approach. Research
was often confined to fairly informal observations (for example, Lukens,
1894) or diary studies, conducted by researchers on their own children (for
example, Taine, 1877; Marsden, 1902). And the total output of research on
child language was rather limited. Leopold (1952) compiled a bibliography of
child language studies, comprising every possible source he could find,
including work on several different languages (not just English). But the total
number of studies from more than a century of endeavour amounted to just
746 published works. This figure has now been dwarfed, owing to a seismic
shift that occurred in the late 1950s, presaged by two academics – Roger
Brown and Noam Chomsky – both based in Boston in the USA. Roger
Brown introduced a fresh rigour and thoroughness in the collection of child
language data that remains an excellent role model and inspiration decades
later. Brown and his students also generated a wealth of theoretical ideas and
insights that, again, continue to resonate to the present day (see Kessel,
1988). Incidentally, the child known as Eve, quoted in (1) above, was one of
Brown’s early research participants, recorded on a regular basis for nine
months between the ages of 1;6 and 2;3. If you want to mull over the full set
of transcripts of these sessions for yourself, it is now possible online (see Box
1.3).
Box 1.3 The Child Language Data Exchange System (CHILDES)
The excerpts of child speech we have seen are part of a much
larger corpus known as CHILDES, a magnificent resource for
serious students of child language. Under the auspices of Brian
MacWhinney at Carnegie Mellon University, CHILDES
comprises an extensive database of child language transcriptions.
Many researchers have donated their original data to this
database. The idea is that child language researchers can access
and share their data and even, where appropriate, perform
analyses on data that were originally gathered by other
researchers. Data are available from more than 25 different
languages with children of different ages in a wide range of
settings. These include data on bilingual acquisition, and on
children with language delays or disorders. Also available is a
sophisticated software package known as CLAN (Computerized
Language Analysis) that allows one to do all manner of automated
analyses, including word searches, frequency counts and
interactional analyses. In some cases, original audio and video
recordings are available, too. On top of this, there is a very useful
online discussion forum (known as InfoCHILDES). Access to
CHILDES is not automatic, and may well be one step beyond
what you need if you’re new to the world of child language, but if
you have a serious interest, check out the CHILDES website
(http://childes.psy.cmu.edu).
Box 1.4 Noam Chomsky
When: Born in Philadelphia, USA, in 1928.
Where: Based since 1955 at the Massachusetts Institute of
Technology, Boston, USA.
What: Linguistics, philosophy of mind and politics.
Famous for: In the world of child language, Chomsky is best
known for the idea that much of what we know about grammar is
innate, that is, genetically determined.
Noam Chomsky is perhaps the most influential academic alive
today. We know this because huge numbers of people throughout
the world have cited, and continue to cite, his work. More than
that, the influence of Chomsky’s ideas can be traced directly in a
vast amount of research, especially in the fields of linguistics,
psychology and philosophy. The whole research enterprise now
known as cognitive science owes a particular debt to Chomsky’s
work. In particular, he has changed the way we think about
language and the way it is acquired.
Chomsky first made an impact in linguistics in the 1950s. Before
Chomsky, the study of language was something akin to solving a
newspaper crossword puzzle. Language was treated like an object
‘out there’ (on paper, in recordings), with patterns, regularities
and peculiarities to be identified. With the advent of Chomsky,
though, linguistics was rapidly turned into a branch of the
biological sciences. The focus ever since has been on explaining
how language is acquired and represented in the minds of human
beings. The quintessence of Chomsky’s position with regard to
child language is his nativism: he believes that much of the
child’s knowledge of grammar is genetically determined (see
Chapter 8).
 Chomsky has also had a parallel career as a political maverick,
turning out books, essays, speeches and interviews in a sustained
campaign against what he maintains is the malign influence of the
US Government and its allies. Chomsky paints a deeply sceptical
picture of the motives of the United States. He argues that the US
has used every means at its disposal to maintain its position as the
richest and most powerful country on earth, including subversive
means such as assassination, covert support for military coups and
terrorist assaults. As the invasion of Iraq in 2003 demonstrates,
the US also adopts openly aggressive manoeuvres, in what
Chomsky (1985: 47) regards as a self-appointed ‘freedom to rob
and exploit’. Chomsky has also been concerned to debunk the
myths and obfuscations that the US Government has, in his view,
generated to justify its actions and keep the general populace in a
state of ignorance and confusion. If you need a break from child
language (not yet, surely!), you will find a recommendation for
further reading on Chomsky’s political ideas at the end of this
chapter.
The second figure I mentioned, Noam Chomsky, does not do empirical
research on child language. But Chomsky’s impact on research in the field
has been phenomenal (see Box 1.4; see also innateness of language). There
is one Big Idea for which Chomsky is especially well known, namely, that
language is innate. This is a fascinating idea, but before we dip a toe into the
waters of nature–nurture, let me offer you a flotation device: Chomsky does
not, to my knowledge, baldly say that ‘language is innate’ (though see Smith,
2004: 167). As a typical academic, Chomsky qualifies much of what he says,
in ways that will become apparent. So you may be disappointed to find our
Big Idea shrinking a little:
the form of the language that is acquired is largely determined by
internal factors (Chomsky, 1966: 64, my emphasis)
But even after this qualification, I hope you will come to appreciate just how
radical and even counter-intuitive Chomsky’s position on language
acquisition remains. If nothing else, it seems odd to suggest that we are all
born with the same knowledge of language when something like 7,000
different languages are currently spoken throughout the world (Gordon,
2005). Did I inherit English, while my neighbour inherited Bengali? Patently,
no. Chomsky has something quite different in mind. Hold that thought for a
moment (actually, hold it until Chapter 8), and let me just slip in a further
word of caution: students of child language should take care in the way they
represent Chomsky’s ideas. It is very easy both to overstate and misinterpret
Chomsky’s position (see Botha, 1989). I shall try not to fall into this trap
myself, but it is always a good idea to check the original sources for yourself
(try Chomsky, 1975, for a classic position statement).
Box 1.5 Language Development or Language Acquisition? The
Culture of Child Language Research
Is language acquired? Or does it develop? You will have noticed
both words in the title of this book and might assume that they are
interchangeable. Both words call attention to how children end up
knowing their native language (or languages). But, in fact, there is
a difference, somewhat subtle, but nevertheless useful for the
student of child language to be aware of. In brief, nativists tend to
avoid the phrase language development, preferring the notion of
language acquisition. I cannot actually find an explicit nativist
repudiation of the term language development in the literature.
But at the same time, it is difficult to find a nativist with a
preference for the term. See for example, the indexes in
Jackendoff (1993) or Smith (2004). Non-nativists, on the other
hand, sometimes show the opposite tendency, avoiding the phrase
language acquisition (Dąbrowska, 2004, has a telling index in this
case, though see Rowland, 2014, for a shift away from this
position).
The difference seems to be tangled up with a third concept, that of
learning. For Chomsky, language is not something that we learn.
He argues that ‘we misdescribe the process when we call it
“learning”’ (1980a: 134). Hence, ‘we do not really learn
language; rather grammar grows in the mind’ (ibid.: 134). This
idea relates directly to the characterization of language as a
mental organ. In the newborn baby, the organs of the body, such
as the liver, grow, largely according to a genetic blueprint. As a
mental organ, language – for Chomsky – is subject to its own,
genetically determined processes of growth. As it happens, the
idea that language simply grows is quite old: ‘the speech learning
of children is … a growth of speech capacity via maturation and
practice’ (von Humboldt, 1836/1988: 37).
 Given this distaste for the concept of learning, one begins to see
why the term development is also avoided by nativists. In essence,
the term development tends to imply some form of learning
(though, strictly speaking, it includes genetic factors also). Thus,
the notion of development in psychology is associated with
figures like Piaget, who considered the child’s experience, and the
knowledge that the child constructs from that experience, to be of
fundamental importance (see Piattelli-Palmarini, 1980). In the
field of second language learning, Larsen-Freeman (2015) has
recently called for the repudiation of the term acquisition. Second
language learners typically arrive at different endpoints in their
learning; they do not ‘acquire’ language in toto like some item in
a shop.
As a shorthand for spotting an author’s theoretical bias, the
preference for language development versus language acquisition
might prove useful. My own preference is to trample all over
these cultural niceties and use both terms in a directly
interchangeable fashion. I will even resort to language learning at
times.
Chomsky’s position on language has polarized opinion within the child
language research community. Researchers tend to align themselves very
clearly either for or against the idea that language – especially grammar – is
innate. It is as well to remember this when you approach the primary
literature (see Box 1.5). Although there is some communication across the
divide, it can be fractious and unproductive at times. As we shall see, there
are occasions when competing and quite different explanations are pretty
much equally good at accounting for particular facts of child language as we
currently understand them. In other cases, nativists and non-nativists are
interested in different phenomena and are therefore unlikely to coincide in
debate with one another. The lack of communication is problematic, but
should not be overstated. Child language research is a thriving field of
enquiry that has blossomed enormously over the past 50 years. In the past 15
years, in particular, a number of alternatives to the nativist credo have been
developed that provide genuine and serious challenges to Chomsky’s theory.
Meanwhile, Chomsky’s own ideas have undergone a major process of change
and development throughout the 1990s and into the current century. There is,
then, a wealth of ideas that promises to drive the field of child language
forward.
The lie of the land
As we progress through the rest of the book, the nature–nurture issue will
come in and out of focus, depending on the particular topic under discussion.
It is by no means the only issue of interest in child language research. But in
common with the rest of developmental psychology, the contribution of
genetics to behaviour is a prominent theme. Accordingly, Chapter 2 deals
with the question of whether language is a uniquely human trait. It does this
by asking if animals can acquire language when given the opportunity. Then,
in Chapter 3 we will consider if the human ability to acquire language is
confined to a specific period of time early on in development, a so-called
critical period. If there is a universal timetable for the development of
language, then genetic factors might well be implicated. The topics of animal
learning and critical periods have attracted a good deal of attention, in part
because of the fascination people have for stories of animal and human
learning in unusual circumstances. It would be fair to say, though, that active
research in both these areas is not so intense at present. I have included them
here, though, because they provide enduring evidence on the biological basis
of language. They also provide good opportunities for considering two other
vital issues in research on child language. The first issue is very basic: we
must be clear about what language is. Chomsky has argued that this question
takes priority, because ‘there is little point in speculating about the process of
acquisition without a much better understanding of what is acquired’ (1959:
55). Although fundamental, it will become apparent that questions about
what language is and questions about what is acquired are not easily
answered. Beyond this issue, we will consider research methods in child
language. Our discussion of The cat in the hat (above) has touched on the
need for ingenuity and rigour in research on child language. This theme is
expanded in Chapter 3.
In Chapter 4, we will shift the focus away from the child’s genetic
endowment towards the child’s environment. We will examine the linguistic
input that the child encounters and consider its role in facilitating language
learning. Chapter 5 looks at language learning in the first year of life,
culminating at about the age of 12 months in the child’s first word. We will
see that, even though the child does not produce much recognizable language,
the child’s knowledge of language expands enormously in the first year.
Chapter 6 looks at word learning, in particular, the constraints on word
learning that somehow allow the child to overcome the gavagai problem.
Chapter 7 moves on to morphology and considers how research has informed
our understanding of the representation of language in the mind. Chapters 8
and 9 deal with what is, for some, the holy grail of language acquisition:
grammar. How does the child acquire something so complex and abstract as
grammar? In Chapter 8, the answer we consider comes from the nativist
perspective: knowledge of grammar is largely innate. In Chapter 9, we
consider what is currently the most prominent non-nativist alternative: usage-
based theory. Human beings are intelligent creatures, capable of acquiring all
manner of skills and knowledge. Maybe grammar is just another (admittedly
complex) system of knowledge that we pick up using our general, all-purpose
learning capacities.
In a Nutshell
Children seem to acquire language quickly and effortlessly.
By the time they reach school, they have acquired a rich
system of knowledge, including an abstract system of
grammar that allows them to combine words in complex,
grammatically correct sentences.
There are several different levels of language, including:
phonology (speech sounds); vocabulary; morphology (parts
of words); and grammar. They are acquired at the same time,
and interact with one another in development.
Language learning begins in the womb. The foetus can learn
to recognize a story from its intonational properties.
Vocabulary learning seems rapid. By the age of six years, the
average child has a vocabulary somewhere between 10,000
and 14,000 words.
In working out the meaning of a word, the child faces
Quine’s gavagai problem: how to determine what a word
refers to, given the infinite possibilities.
We know that children can analyse the internal structure of
words from Berko’s wug test, in which the child is prompted
to supply inflectional morphemes for nonsense words (‘One
wug, two …’ / ‘wugS’).
Children demonstrate some understanding of syntax – the
rules governing how words are put together in sentences – as
early as the second year of life. For example, children
 acquiring English realize that changes in word order can
fundamentally alter the meaning of a sentence.
Language development can seem rapid and effortless,
especially when compared with certain milestones in other
developmental domains (perceptual, cognitive and social).
But we should be cautious. Comparisons are difficult to
make and it is not even clear how to determine what would
count as ‘rapid learning’.
Further Reading
Aitchison, J. (2010). Aitchison’s linguistics (7th ed.). London:
Hodder Headline.
This book has been around, in one format or another, since 1972
and is written by the pre-eminent writer of student-friendly books
on language-related topics. I provide definitions of linguistic
concepts (even exercises) throughout the book. But Aitchison’s
book provides more detail and could therefore prove useful when
accessing original research on child language.
Chomsky, N. (2005). Imperial ambitions: Conversations with
Noam Chomsky on the post 9/11 world: Interviews with David
Barsamian. London: Hamish Hamilton.
For all those whose horizons extend beyond child language, this
book provides an insight into Chomsky’s political views.
Chomsky’s output in politics is, like his work in linguistics and
philosophy, prodigious, making him seem like a latter-day
Renaissance man. Put it this way: I doubt Professor Chomsky
spends much time on the golf course.
Websites
Estimating vocabulary size: www.testyourvocab.com
There are several websites which offer a free assessment of
your vocabulary size. This one is as good as any other, and
includes an interesting description of the rationale used to
arrive at the estimate.
Child language data: www.childes.talkbank.org
The Child Language Data Exchange System (CHILDES)
affords access for serious students to many original
transcripts, in addition to a range of automated tools for
 analysing child language data (see Box 1.3).
Still want more? For links to online resources relevant to this
chapter and a quiz to test your understanding, visit the companion
website at https://study.sagepub.com/saxton2e
2 Can Animals Acquire Human Language?
Shakespeare’s Typewriter
Contents
What is language?  28
The infinite monkey theorem 28
Language, talk and communication 29
The design of language 30
Teaching words to animals  36
Talking versus sign language 36
Lexigrams 37
Barking up the right tree: Word learning in dogs 38
Alex, the non-parroting parrot 40
Animal grammar  41
Combining words 41
Comprehension of spoken English by Kanzi 43
The linguistic limitations of animals 44
Is speech special?  45
Categorical perception in infants and primates 45
Statistical learning 47
Back to grammar: Infants versus monkeys 48
The language faculty: Broad and narrow 49
Overview
We start this chapter with a basic question: what is language? Taking
care to distinguish between three separate concepts – language, talk
and communication – we go on to consider some of the key differences
between animal communication systems and language. No animal
acquires language spontaneously, but can they be taught? We review
evidence of attempts to teach a range of animals, including
chimpanzees, monkeys, dogs and parrots. In some ways, animals have
surprising linguistic gifts. Yet they fall far short of being able to string
words together into grammatical sentences. We will see that attempts
to teach the grammatical rules of a natural human language are overly
ambitious. But the ability to refer to objects with words is now well
established for many animals. Recent research also looks at basic
psycholinguistic mechanisms, in particular, the processing of speech
 sounds. This research has shown that cotton-top tamarin monkeys can
identify word-like units in a continuous stream of speech, in a way that
is similar (though not identical) to the way human infants go about it. It
emerges that humans and animals share many basic capacities that can
be harnessed in the acquisition of language. At the same time, there
remain clear differences between humans and non-humans. Language
is a very human phenomenon.
What is language?
The infinite monkey theorem
The legend goes that if you sit a monkey down at a typewriter for an infinite
length of time, then eventually, by hitting the keys at random, the monkey
will type out the complete works of Shakespeare. This is one version of the
infinite monkey theorem. You will not be surprised that the shelf life of this
theorem is something less than infinity, once we hold it up to even the most
casual scrutiny. First, monkeys are not random letter generators. Second, we
are seduced by this problem into thinking about infinity as a vast but finite
number. Third, it seduces me into showing you a picture of a chimpanzee (not
a monkey) sitting at a typewriter (Figure 2.1). Fourth, the monkey would, in
all likelihood, be dead long before the first sentence had been typed. And
fifth, even Shakespeare would have trouble producing the complete works of
Shakespeare, given that he couldn’t type.
Our sense of the ridiculous is provoked by this scenario. We find it laughable
to think that a monkey could generate Shakespeare’s plays or even a simple
‘Hello, how are you?’ Monkeys do not have language. But this theorem
suggests that they could acquire language, given sufficient time and an
appropriate means of expression. The underlying assumption, therefore, is
that humans and monkeys are not qualitatively different when it comes to
language. The difference is merely quantitative. From this line of thinking it
follows that animals (perhaps the more intelligent ones, at least) could
acquire language, if they went about it in the right way. In fact, there have
been several attempts to talk to the animals, and to get them to talk back to
us. We will examine some of these efforts in this chapter and, in the process,
we will discover how this comparative approach can help clarify what,
precisely, is unique, and possibly biologically determined, about the human
capacity to acquire language.
Figure 2.1 The infinite monkey theorem
 Source: Wikipedia: Infinite Monkey Theorem
http://en.wikipedia.org/wiki/Infinite_monkey_theorem
Language, talk and communication
So … can animals talk? At first glance, this question seems absurd. We do
not find sheep debating the pros and cons of a vegetarian diet. Nor do we
hear pigeons chatting about life in Trafalgar Square. But this question is
deceptively simple. A number of bird species, including mynah birds and
parrots, have astonishing abilities to mimic the human voice. Do these birds
talk? Evidently talk refers to the physical act of producing human speech
sounds, but its meaning can be wider, including the communication of ideas
via speech. Some parrots can talk in the former sense, but can they hold a
conversation? In contrast, many deaf people cannot talk, in the sense of being
able to produce speech sounds with great accuracy or fluency, but they can
hold conversations every bit as sophisticated as those of hearing people,
through the medium of sign language. When we consider the biological basis
of language, it is this latter, conversational, meaning of talk that holds more
interest. Perhaps we should abandon all talk of talk and go straight to the
heart of the matter: do parrots (or any other animal) have language?
But before we get to language, we must dispense with communication. Many
people believe that (at least some) animals have language, because they use
the terms language and communication interchangeably. But they are not the
same thing. Many, possibly all, animals communicate, in the sense that they
intentionally convey information to each other. To this end, dogs bark, sheep
bleat and lions roar. But they do not have language. The upshot is that we
must be careful to distinguish the concepts of talk, communication and
language. It is the biological basis of language that we are concerned with
here.
The design of language
Language has a number of characteristics that sets it apart from animal
communication. The American linguist, Charles Hockett, worked on this
problem for many years, finally coming up with a list of 16 defining
characteristics (Hockett, 1963). As we shall see, most of these so-called
design features are shared to some extent with animal communication
systems. On this approach, it is the combination of all 16 features that
distinguishes human language. We will focus on just four of Hockett’s design
features here (but see Box 2.1 for the complete list):
creativity
semanticity
arbitrariness
displacement
Hockett (1963) had two kinds of linguistic creativity in mind. The first refers
to the creation of new words and idioms, a process that takes place all the
time, and with especial enthusiasm among young people (Cheshire, 1982; see
also Discussion point 2.1 below). The second kind of creativity refers to our
ability to produce and understand new messages. In recent times, Chomsky
(1965), in particular, has emphasized the importance of this latter kind of
creativity, reviving a very old observation in linguistics (von Humboldt,
1836/1988): the rules of grammar (a limited, or finite, set of rules) allow us
to construct an infinite variety of new utterances. New sentences are created
all the time.
Two aspects of child speech demonstrate that linguistic creativity is in
evidence very early on: (1) novel utterances; and (2) grammatical errors.
With regard to novelty, Chomsky (1959: 42) pointed out that ‘a child will be
able to construct and understand utterances which are quite new, and are, at
the same time, acceptable sentences in his language’. The following
sentences from Alex, aged four years, are in all likelihood, novel:
No, he bit this little tiger from this set.
You stood on my botty.
Yes, he slid down to the bottom with Panda.
Frog fell over and Bee fell over and Soldier fell over.
 When I was building the zoo, I jumped down and hurt myself.
The examples above are grammatical sentences. The ones below (also from
Alex) contain errors and, therefore, we can be even more confident that they
are novel utterances, because young children are very rarely exposed to
ungrammatical speech (Newport, Gleitman & Gleitman, 1977; see also
Chapter 4). This means that errors have not been imitated.
She’s too wide asleep, I think.
I like Shinji because he’s the goodest.
Last day, when you had some toast, how many did you have?
Look, my scratches are going who I got at Colin’s house.
I had a happy fry tuckey.
Incidentally, the ‘happy fry tuckey’ was a Kentucky Fried Chicken Happy
Meal. Naturally, Alex was given this by his mother. I only ever gave him
organic fish and vegetables. Cough. Both novel utterances and errors provide
strong evidence that the child is an active producer of their own language,
using a system of rules to assemble their own sentences. Throughout our
lives, we create entirely new sentences at will, and the people listening to us
can, for the most part, understand them. Creativity appears to be one of the
few characteristics of language that is unequivocally unique to human
language. Animal communication systems simply do not yield creative
potential.
Box 2.1 Hockett’s Design Features
Charles Hockett set out 16 different characteristics which together
help define human language and crystallize what sets it apart from
animal communication systems (Hockett & Altmann, 1968).
1. Vocal-auditory channel. The conduit for language is sound
(the human voice), transmitted to a hearing mechanism (the
ears and auditory channel).
2. Broadcast transmission and directional reception. Sound is
transmitted (broadcast) in every direction, but the listener
perceives it as coming from a particular source.
3. Rapid fading. Speech sounds do not last very long; they are
transitory.
4. Interchangeability. We can switch between transmitting and
receiving linguistic signals.
5. Complete feedback. Speakers can hear what they are saying:
they receive feedback on the signal they are sending out.
 6. Specialization. The organs used for producing speech
(including lips, tongue, larynx and throat) are specially
adapted for language; they are not simply used for breathing
and eating.
7. Semanticity. Linguistic signals (words) can be used to refer
to actions, objects and ideas.
8. Arbitrariness. There is no intrinsic, necessary relationship
between a word and the concept it denotes. There is nothing
(in principle) to prevent us from dropping the word cat and
switching to zug to denote the same object.
9. Discreteness. Language makes use of discrete units, for
example, the two speech sounds (phonemes) /p/ and /b/. The
point at which the vocal cords begin to vibrate can vary, with
more /p/-like sounds shading gradually into more /b/-like
sounds. But we perceive a sharp cut-off between the two
sounds (they are discrete), even though, physically, they
figure on a single continuum (see Chapter 6).
10. Displacement. We can talk about things that are remote in
time or space.
11. Creativity. We can extend the repertoire of language by
coining new words and expressions. Also, we can produce
and understand sentences that have never before been
uttered. (Hockett used the term openness, but creativity is
preferred these days).
12. Cultural tradition. Language is transmitted from one
generation to the next by teaching and learning.
13. Duality of patterning. Languages have two levels:
meaningful units at one level (words or morphemes) are
created by combining meaningless units at the second level
(phonemes).
14. Prevarication. Human beings can lie. They also write novels
and talk about entirely hypothetical entities.
15. Reflexiveness. Language can be used to think and talk about
language: we have so-called metalinguistic skills.
16. Learnability. We have the capacity to learn new languages
beyond our native tongue.
 Discussion point 2.1: Creation of new words
Identify words and phrases that have only recently come into popular
usage. They could well be expressions that you use with your friends,
but would hesitate to use with your grandparents. You might think of
this as slang or casual speech and would struggle to find such words in
a dictionary. List five different words of this kind. For each one,
consider the following:
Think of how the word is pronounced (the sounds). Is it a new
sequence of sounds? Or is it an old word form that has been given
a new meaning (like cool to mean ‘admirable’ rather than ‘low
temperature’). Or has it been pieced together from existing words
and English sound sequences (like cowabunga, an expression of
exhilaration for 1960s surfers and Teenage Mutant Ninja Turtles)?
Does the word denote a new concept (have a new meaning)? If so,
consider how you would express this new concept if you did not
have this new word.
If the word is, essentially, a synonym for a word that already
exists, why do you think it has come into being?
The second of Hockett’s design features, semanticity, describes our use of
symbols to refer to (or ‘mean’) something. These symbols are words,
ostensibly sequences of speech sounds that can be used to refer to objects,
actions or ideas. The word cat refers to a furry domestic animal with
whiskers and a tendency to sit on just that part of the paper you’re trying to
read. What is more, the word cat can be used to pick out (or ‘mean’) all
exemplars of this kind of animal. Cat denotes a category of objects. Do
animals have anything that equates to words? One example might be the
danger cry of the white-handed gibbon, issued to warn other gibbons about
approaching predators, including the clouded leopard (Clarke, Reichard &
Zuberbühler, 2006). We might argue that the cry ‘means’ danger in the way
that cat ‘means’ my pet, especially since these gibbons use different calls in
different situations. But the gibbon cry might just as easily be a violent
emotional response that only incidentally communicates information to other
gibbons. Even if the gibbon cry has meaning, it is highly general in its scope,
which, if your life is in peril, might not be all that helpful. The type and
number of predators, their proximity, the direction they are moving in: none
of these are specified by the danger cry. Instead, the gibbon call seems to
apply to a total situation. And although other gibbons can obtain information
from these calls, it is not clear that there has been any intention, in the human
sense, to communicate that information to them (Seyfarth & Cheney, 2003).
The (potential) semanticity of animal calls remains a matter for debate. But
whichever way one looks at it, human beings far outstrip animals in their
deployment of semanticity (Chomsky, 2011). There are many (many)
thousands of words in English, compared to the extremely limited repertoires
of animal vocal signals. So human beings exploit the benefits of semanticity
to a far greater degree than any animal. We shall consider below whether
animals can be encouraged to extend their repertoires and behave more like
humans with regard to semanticity.
In the meantime, let’s consider arbitrariness, another key design feature of
language. The central idea here is that word forms – the packages of sound
mentioned above – bear no intrinsic relation to their referents (de Saussure,
1916/1974). There is nothing in the nature of a domestic feline animal that
forces me to call it cat. In Paris, I would use chat, while in Tokyo, it would
be neko. The relation between the referent (the object in the world) and the
word form is arbitrary, and the same holds true for the vast majority of words
in any language. A tiny handful of exceptions can be found in the form of
onomatopoeic words, like pop, crack and squash. Animal cries, on the other
hand, are often tied closely in form to the ‘meaning’ they express. The gibbon
danger cry is invested with the heightened emotion of arousal. But if we
scour the animal world thoroughly, examples of arbitrary signals can be
found. Western gulls, for example, have a wide repertoire of behaviours to
denote aggression (Pierotti & Annett, 1994). Most of them, such as striking
and grabbing other gulls, or pursuits on the ground with upraised wings, are
non-arbitrary. But these gulls also turn away from an opponent and uproot
grass with their beaks to display aggression. There is nothing about grass-
pulling per se that inevitably connects it with aggression. It appears to be an
arbitrary signal. The moral: never invite two Western gulls to a garden party.
They’ll fall out and make a terrible mess of your lawn.
Our final star feature is displacement. This simply refers to our ability to talk
about things beyond our immediate situation. Human conversation can be
displaced to matters that are remote in both time and space. I can talk about
events both past and future, near and far. I am not confined to talking about
my immediate context. In contrast, consider once again the gibbon danger cry
(poor gibbon, we should get him to a place of safety). The danger signal is
only triggered by the sight of an actual predator that is physically present.
The gibbon has no signal to inform fellow gibbons that, say, they ‘spotted a
leopard about half a mile south of here three hours ago’. Nonetheless, danger
signals triggered by the sight of a leopard can persist some time after the
leopard has disappeared from view. Does that count as displacement? It is
debatable. Other animals show more definite signs of displacement in their
communication. Most famous in this regard are Karl von Frisch’s bees.
Figure 2.2 The wagging dance of the honey bee

Source: von Frisch, K. (1954). The dancing bees: An account of the life
and senses of the honey bee (translated by D. Ilse). London: Methuen,
Figure 40: The wagging dance, p. 117.
Von Frisch (1954) observed that when a bee discovers a good source of
nectar, it communicates this information to other bees via a special ‘waggle
dance’ (see Figure 2.2). The dance is performed vertically on a honey comb
and comprises a simple figure of eight pattern drawn around a central straight
line. During the straight line part of the run, the bee waggles its abdomen
before turning left or right. The duration of the waggle phase corresponds to
the distance of the food from the hive. Meanwhile, the angle at which the bee
departs from the vertical, relative to the sun, during the straight line part of
the dance corresponds to the direction of the food source. Amazingly, the bee
alters the angle of its dance to accommodate the changing angle of the sun
throughout the day. Bees therefore communicate both the direction and
distance of remote food sources to other bees, a clear case of displacement.
Karl von Frisch won the Nobel Prize for his ingenious work on the bee dance
in 1973, and was never short of honey for tea. For our purposes, we can
observe that, once again, one of Hockett’s design features is not the exclusive
preserve of human language. At the same time, it is worth reflecting that
displacement is very rare in the animal world. And the kind of displacement
we observe in bees, while impressive, is easily outstripped by our own
accomplishments. We can talk about anything remote from our current
location, whereas bees are confined to food sources within flying distance of
the hive. They cannot even tell other bees how high a particular food source
is. And humans can talk about things remote in time as well as space, and
even things that are displaced in the sense of being entirely imaginary. The
flexibility and range of the displacement witnessed in human language is
massively greater than anything an animal communication system can offer.
Discussion point 2.2: Defining language
We have considered four of Hockett’s design features in some detail.
In Box 2.1 you will find the complete set. For each one of the 16
design features, consider the questions below. You might want to split
this task among different groups.
Can you think of any animal communication systems that have a
similar feature?
If the answer is yes, do animals nevertheless differ from humans
in the way they exploit this feature?
Returning to our original question – do animals have language? – the answer
is clearly ‘no’. Animal communication systems share many of their core
features with language, but the overlap is highly selective. Bees might show
displacement in their communication. But the extent of the displacement they
are capable of is strictly limited. Moreover, bees do not use the vocal-
auditory channel. They do not demonstrate creativity. Their communication
is not arbitrary (there is a direct relation between the angle of the bee dance
and the source of nectar). And so on. There are qualitative differences
between animal communication and human language. But there are
quantitative differences also, often quite dramatic. Humans exploit the
potential of the design features to a much greater extent than animals. The
‘words’ found in gibbon communication could be listed on a single page. Set
this against the multiple volumes of the Oxford English Dictionary: the
contrast in our mutual exploitation of semanticity is readily apparent.
Animals may not have language, but there is a second question we can ask in
our exploration of the biology of language: Do animals have the potential to
acquire language? If so, Hockett’s feature of cultural tradition might be the
only barrier keeping animals from language. Whatever its biological basis,
humans clearly do pass on many aspects of language from one generation to
another. These include both the particular phonemes of a given language and
the vast and arbitrary lexicon that must be learned by every child. If language
is a cultural artefact, much like fire or clothing, then it might be possible to
introduce it to animals. After all, there are several highly impressive
cognitive achievements which do not emerge spontaneously in human beings,
but which we nevertheless have the capacity to acquire. Consider
mathematics or music. Perhaps animals simply need the right opportunities.
If an animal could be taught language, then the argument that language is the
sole province of human beings would collapse. It would also be much more
difficult to argue that language per se is coded for in the human (but not any
animal) genome. As it happens, a parade of animals, including chimpanzees,
gorillas, grey parrots, dogs and dolphins, have signed up for Language
School. We shall consider how well they got on in the next section.
Teaching words to animals
Talking versus sign language
If you want to teach an animal to talk, it makes sense to try one of our more
intelligent neighbours in the animal world. Not only is success more likely,
the calibre of conversation should be better than with less bright animals.
What would you talk about with a crayfish or a beetle? Accordingly, most
efforts to teach language have been directed towards our close evolutionary
cousins, the great apes, including chimpanzees, gorillas and orang-utans (see
also Herman, Richards & Wolz, 1984, on artificial language learning in
dolphins). From the 1930s until the end of the last century, several such
attempts were made, beginning with a chimpanzee named Gua who was
raised like a human infant by Winthrop and Luella Kellog (1933). The
Kellogs wanted Gua to talk in both senses described above: to communicate
ideas via language, but also to engage in the physical act of producing speech
sounds. This latter enterprise, in particular, was doomed to failure. It turns
out that the chimpanzee’s vocal apparatus – in particular, the proportions of
the tongue, the angle of the airway and the relatively high position of the
larynx – render human speech sounds all but impossible (Nishimura, 2005).
Another chimpanzee, Viki, did produce three words: mama, papa and cup.
She pronounced them ‘softly, and hoarsely, but quite acceptably’ (Hayes &
Hayes, 1951: 107). But this meagre outcome was achieved only after three
years of intensive training. Astonishingly, an elephant called Koshik has
produced human speech recently. Koshik imitated a few words of Korean
well enough for native speakers to transcribe accurately (Stoeger, Mietchen,
Oh, De Silva, Herbst, Kwon & Fitch, 2012). He did so by placing his trunk
inside his mouth to alter the acoustic properties of the vocal tract. As yet,
though, there is no indication that Koshik understands what he is saying.
Returning to chimpanzees, research moved on to sign language, which does
not depend on the production of speech sounds. Moreover, primates,
including chimpanzees, gorillas and orang-utans, have all proved to be
manually dextrous enough to communicate through the use of gesture. The
first of the signing apes, Washoe, was born some time in 1965 and started
learning an adapted version of American Sign Language (ASL) at about 10
months of age. By 36 months, Washoe was credited with knowing 85
separate signs (Gardner & Gardner, 1969). Her vocabulary eventually rose to
132 signs after four years of training (Fouts & Fouts, 2004). Washoe’s use of
signs was noteworthy in two respects: (1) spontaneity; and (2) generalization.
First, Washoe did not always need to be prompted to use signs once they had
been learned, but could do so spontaneously. And second, she sometimes
generalized their use to a range of objects and events around her. The ability
to generalize is an important aspect of reference, allowing us to apply a word
to all members of a category. The word apple can be applied to all apples –
red, green, large, small – not just one particular object in my fruit bowl. This
is because the word apple is used to refer to a category of objects. We will
consider how children tackle generalization of word meanings in Chapter 6.
For the moment, we can note that chimpanzees are able to generalize their
use of words beyond the immediate context in which they are learned.
Lexigrams
Sign language is not the only medium available for teaching language to
apes. This sentence demonstrates the potential to represent linguistic
information visually via printed symbols (letters). And while no chimpanzee
has been taught to read and write, efforts have been made to use visual
symbols (called ‘lexigrams’) that represent whole words (Savage-Rumbaugh,
Murphy, Sevcik, Brakke, Williams & Rumbaugh, 1993). This system has a
direct counterpart in Chinese characters: a single symbol can represent an
entire word, whereas in English, we construct written words from one or
more letters, each representing different phonemes. In teaching lexigrams,
the chimpanzee is presented with a keyboard containing numerous different
symbols constructed from different combinations of shapes and colours. By
pressing one of the symbols, a speech synthesizer produces the spoken
version of the word. The greatest success with the lexigram keyboard has
been achieved with bonobo (or pygmy) chimpanzees, a species that is noted
both for high intelligence and prolific sexual activity (de Waal, 1995). Kanzi,
a male bonobo born in 1980 was exposed to lexigrams from the age of six
months. Four years of effort had previously gone into trying to teach Kanzi’s
adoptive mother, Matata, how to use the lexigram keyboard, but to no avail.
Kanzi, on the other hand, did much better, acquiring signs by observing his
mother’s training sessions. Kanzi’s first 10 words were orange, peanut,
banana, apple, bedroom, chase, Austin, sweet potato, raisin and ball. Kanzi
would get on well with Jamie Oliver.
In the more popular end of the scientific press, Sue Savage-Rumbaugh claims
that Kanzi can recognize and use 348 different lexigram symbols (Raffaele,
2006). More astonishing, in the same article it is claimed that Kanzi
understands more than 3,000 spoken English words. We will consider
Kanzi’s comprehension of spoken English below, but these figures have not
been corroborated and are, in all likelihood, highly inflated. For one thing,
this figure seems to be based simply on the number of words that Kanzi was
exposed to. That is a far (ape) cry from systematically testing comprehension
of each word in isolation. More reliable estimates point to a stark contrast in
the size of ape and child vocabularies. While apes seem to peak at a few
hundred words, the language learning child soon dwarfs this figure.
The vogue for teaching apes to talk was at its height in the 1970s, and there
was a point during that decade when it looked as though language could no
longer be seen as the sole preserve of human beings. If apes can acquire
words and combine them into novel utterances, one might assume (as the vast
majority of their trainers did assume) that they had conquered language. But
there are several factors that must temper this enthusiasm. First, the word
learning of apes is slow and effortful compared to children. Children can
acquire words on a single exposure, a process known as fast mapping (Carey,
1978; see also Chapter 6). This contrasts with the efforts of Kanzi and
another bonobo, Panbanisha, who required in the region of 32–65 exposures
to learn each new word (Lyn & Savage-Rumbaugh, 2000). What is more, the
bonobos were tested within one hour of training for signs of comprehension.
In consequence, it is not clear how long the bonobos’ learning lasted,
whereas for the child, of course, words can be lodged in the mind for a
lifetime. Word learning in the chimpanzee is clearly a laborious task that they
do not take to naturally. The motivation is often driven by reward – food,
play, toys – rather than a desire to communicate meanings and hold a
conversation either with their trainers or each other. In addition, the
categories of words acquired are confined almost exclusively to objects and
actions, with none of the wide variety of semantic categories witnessed in
human vocabularies (Seidenberg & Petitto, 1979). But even on the most
sceptical interpretation, we must still allow that chimpanzees have the
capacity to acquire words. The arbitrary nature of words, their semanticity,
and their spontaneous, generalized use in displaced communication have all
been reliably observed.
Barking up the right tree: Word learning in dogs
Before we move on to consider the holy grail of language learning – grammar
– let’s consider how dogs and parrots have tried to get in on the word-
learning act. Kaminski, Call & Fischer (2004) report that Rico, a border collie
dog, acquired the meanings of more than 200 items and can retrieve named
objects up to four weeks after the original training session. Moreover, Rico
seemed to be using exclusion learning, also witnessed in young children
(Heibeck & Markman, 1987). Presented with an array of objects, for which
the names of all but one are already known, children infer that a novel word
must refer to the single nameless object on display. Thus, if a two-year-old is
shown an apple, a ball and a whisk and is asked to retrieve the whisk, they
can do so, even if, previously, they only knew the labels ball and apple.
Kaminski et al. (2004) claim that Rico can do this too, reporting an accuracy
level of 93 per cent on 40 trials. It is unlikely that Rico simply has a novelty
preference, because he is also able to retrieve familiar objects as well as new
ones. As Bloom (2004: 1605) remarks, ‘for psychologists, dogs may be the
new chimpanzees’. But Bloom also sounds a note of caution (see also
Markman & Abelev, 2004). Rico may be doing nothing more than
responding to an entire situation (‘fetch X’). There is a difference between a
simple association between two items (a word and an object) and the act of
reference. This observation was made in the nineteenth century by Taine
(1877: 254) who remarked: ‘There is nothing more in this than an association
for the dog between a sound and some sensation of taste’. In contrast to this
kind of associative learning, Bloom (2004) points out that, for the two-year-
old child, the word sock does not mean ‘fetch the sock’ or ‘go to the sock’.
Instead, the child appreciates that the word refers to a category of objects and
can be used to talk about different socks, request a sock, or notice the absence
of one.
Bloom’s concerns have been addressed more recently by Pilley & Reid
(2011), who started out by shopping for hundreds of soft toys (yes,
hundreds), plus a number of Frisbees and balls thrown in. So to speak. Each
object was distinct and was labelled in marker pen with a unique name. Over
the course of three years, Chaser was trained for 4–5 hours per day and
learned the names for 1,022 of these toys. Remarkably, even several months
after training was complete, Chaser was 92 per cent accurate when asked to
retrieve a toy from an array of eight objects. There is also evidence that
Chaser’s understanding of words goes beyond a global response to an entire
situation (‘fetch the ball’). Chaser could respond appropriately to three
different commands (take, paw and nose) when combined with an object
name (Pilley, 2013). She seemed to understand the notion of reference, in
particular, that words can refer to a given object quite independent of any
actions which that object might be associated with. Chaser also seemed to
understand that common nouns refer to an entire category of objects, even
though most of the words she was trained on were treated as proper nouns.
For example, a single object could be labelled in three different ways: as a
toy; a ball; or by its specific name.
Like children (and Rico), Chaser was able to infer the name of a new object
by exclusion (cf. Greer & Du, 2015). For example, Chaser was presented
with eight objects, all of which were familiar (Sugardaddy, Valentine, Wow,
Pregnant, Backpack, Tortoise, Red Riding Hood and ABC – there’s a sad
story struggling to get out in that list). A novel object was then added to this
collection, and Chaser was asked to fetch Airborne, a word which was also
novel to her. Because she knew all the other object names, she inferred that
the new name – Airborne – must refer to the only novel object. This kind of
learning is not confined to language. Nor is it confined to human beings.
Learning by exclusion in non-linguistic domains has been observed in parrots
(Pepperberg, Koepke, Livingston, Girard & Leigh, 2013), macaques,
baboons, capuchins and squirrel monkeys (Marsh, Vining, Levendoski, Judge
& Call, 2015). Evidently, both children and dogs can recruit a domain-
general learning mechanism in the acquisition of new word meanings (more
on this in Chapter 10). Even more startling, there are parallels between the
way pigeons learn to categorize objects and the way children acquire words.
Wassermann, Brooks & McMurray (2015) trained pigeons to sort 128 objects
into 16 human language categories. Like word-learning children, the pigeons
were guided by the perceptual coherence of the different object categories
(e.g., shape), and moreover, they generalized their learning to novel objects.
Chaser’s lexical prowess puts her fellow collie, Rico, in the shade, but more
startling, she outstrips even Kanzi by a wide margin, having learned roughly
three times as many words. This is puzzling, if one considers that
chimpanzees are renowned for their intelligence (Mandalaywala, Fleener &
Maestripieri, 2015) and, what’s more, are much closer to humans,
phylogenetically, than dogs. But there is a great disparity between chimps
and dogs in their willingness to be directed by humans. Domestic dogs can
recognize when particular actions are intended as communicative signals,
including pointing, the direction of someone’s gaze and high-pitched
vocalizations (Tomasello & Kaminski, 2009). Of note, human infants also
tune in with alacrity to these self-same signals (see Chapter 4). These skills
appear to be largely untaught in dogs, since they are witnessed in puppies as
young as six weeks old (Kaminski, Shulz & Tomasello, 2012). Dogs have
been domesticated for thousands of years. It is possible, therefore, that the
long, intertwined history of humans and dogs underpins canine sensitivity to
the communicative intentionality of human behaviours (Hare, Rosati,
Kaminski, Bräuer, Call & Tomasello, 2010). Unlike chimps, dogs are
dependent on humans for food and shelter. It is perhaps not surprising,
therefore, that they are so responsive to the communicative signals of human
beings. In contrast, chimpanzees are much less inclined to communicate
spontaneously with a human, even when supplied with the means to do so.
The cognitive skills needed to acquire language are of little use if the
motivation to learn and use language is missing in the first place. Unlike
Chaser, Kanzi was not a willing player of the Language Game.
Alex, the non-parroting parrot
Perhaps even more impressive than Chaser is the case of Alex, an African
grey parrot, bought from a pet shop in 1977 when he was about one-year-old.
For 30 years, until his death in September 2007, Alex was involved in a
programme of research that transformed our view of animal learning
(Pepperberg, 2006). While Rico can understand English commands, Alex
could both understand and speak for himself, making use of parrots’
celebrated vocal mimicry skills. Alex could label 50 different objects,
including seven colours and five shapes, as well as being able to discriminate
different quantities up to and including six. Remarkably, Alex could answer
different questions about the same object (Pepperberg, 2010). Shown a scrap
of yellow wool, Alex could answer correctly if the question was either ‘What
matter?’ or ‘What colour?’ Alex even mastered the number words from one
to eight, demonstrating knowledge of cardinality – the fact that each number
word denotes a collection of a given size (Pepperberg & Carey, 2012). Alex
learned via the so-called Model/Rival technique in which he first observes
two humans, one ‘training’ the other, with success being rewarded by
receiving the object that is being named. Alex thus observes and competes
with a human being for the objects being labelled. Rewarding Alex with the
actual object proved far more successful than providing something extrinsic
to the learning situation, like food. This is different from Rico, of course, who
must have to watch his weight, given the number of snacks on offer.
We can see, then, that both dog and parrot learned labels for objects in ways
that differ quite fundamentally from the way a young child learns words.
Pepperberg (2006: 78) asserts that Alex ‘demonstrates intriguing
communicative parallels with young humans, despite his phylogenetic
distance’. In assessing this claim, we should perhaps focus on the judicious
use of the word ‘parallel’. Alex’s use of his 50-word vocabulary bears some
resemblance to that of a young child. But the route he took to get there is very
different and his mental representation of words is also very likely quite
different from that of a human learner. Shown a wooden block, Alex can
correctly answer a range of different questions about it, including: ‘What
colour?’, ‘What shape?’, ‘What matter?’ (wooden) and ‘What toy?’ But each
answer has to be trained separately. When different attributes, like colour and
shape, are presented together, learning does not take place. When asked
‘What colour?’, another parrot called Griffin answers with an object name,
like ‘cup’. In contrast, the young child can integrate the different attributes of
an object into a coherent network of interrelated meanings.
Animal grammar
Combining words
As soon as a language learner begins to combine two or more words together
into single utterances, it becomes possible to broach the issue of grammar.
And it is here, perhaps, that the real battles have been fought in what the
Gardners ruefully refer to as the ‘chimp-language wars’ (Gardner & Gardner,
1991). We can start by confirming that chimpanzees do produce multi-word
sequences. Washoe came out with phrases like gimme tickle (‘come and
tickle me’), open food drink (‘open the fridge’) and sweet go (‘let’s go to the
raspberry bushes’). Moreover, this language learning feat is not confined to
Washoe. Several other primates have been observed producing sign
combinations, including other chimpanzees (Terrace, 1979; Asano, Kojima,
Matsuzawa, Kubota & Murofushi, 1982), a gorilla (Patterson, 1978;
Patterson, Tanner & Mayer, 1988) and an orang-utan (Miles, 1983).
Rivas (2005) analysed a large corpus of signing behaviour from five
chimpanzees and reports that 33 per cent of their output comprised two or
more signs. The question that then follows concerns rule-governed behaviour.
Are words being combined in a systematic way that conforms to the rules of
grammar? The easiest way of tackling this question is to look for consistency
in the way words are combined. Some combinations look promising in this
regard, for example, eat apple and drink coffee. Children’s earliest two-word
utterances look very similar: Mommy letter; finger stuck; and empty garbage.
In both ape and child, we can see that the basic word order of English has
been preserved, even though there are several words (and bits of words)
missing. Thus, eat apple might be glossed as I want to eat an apple, while
finger stuck might be more fully rendered as my finger is stuck. So far so
good. But whereas the child maintains the correct word order throughout
their language learning, chimpanzees are far less reliable. For example,
although Kanzi had a preference for the correct word order, he was
nevertheless quite happy to switch words around: bite tomato alternating with
tomato bite. But think of the famous contrast in newspaper headlines: ‘Dog
bites man’ is not news, whereas ‘Man bites dog’ is news. Switching the word
order makes a big difference in how the sequence of words is interpreted.
Nim, however, switched the order of words around with alacrity: eat Nim
occurred 302 times, compared with 209 instances of Nim eat, seemingly with
no regard for the potential impact it might have on the message conveyed.
Across a range of chimpanzees, object and action signs tend to occur in initial
position, with request markers like that, there, and good occurring in final
position. This makes sense if one considers that the crux of a message, a
particular object or action that is being requested, is most salient to the
chimpanzee and hence more naturally takes first position in an utterance.
Hence, the order of words is dictated by pragmatic, rather than syntactic,
factors. In other cases, two-word combinations seem to be completely
haphazard in a way that child utterances are not: chase toothbrush, flower
peekaboo, sodapop brush, drink gum. Longer combinations look even less
like grammatical sentences:
Multi-sign combinations made by chimpanzees (Rivas, 2005)
nut brush toothbrush
cheese drink Tatu
flower gimme flower
banana toothbrush that
flower drink gum toothbrush
drink there brush clothes
As the signing sequences increase in length, they rely increasingly on
repetition:
Four-sign sequences by Nim (Terrace, 1979)
eat drink eat drink
drink eat drink eat
Nim eat Nim eat
me Nim eat me
Drink eat me Nim
eat Grape eat Nim
Anyone would think the poor ape was being deprived of food. Maybe if he
learnt to ask nicely, he’d have more success. But it would seem that this feat
is beyond both Nim and all other chimpanzees. The longer the sequences, the
more repetitious the ape becomes. Hence, no extra meanings are being
conveyed in longer utterances, and any semblance of regularity or
consistency is diminished. Rivas (2005) reports one sequence of 22 signs
from a chimpanzee, and while he does not spell it out, two sequences, each
comprising 16 signs are detailed:
Sixteen-sign sequences by Washoe and Nim (Rivas, 2005)
Washoe: flower hurry flower hug go flower book flower gimme flower
gimme flower hug hurry drink gimme
Nim: give orange me give eat orange me eat orange give me eat orange give
me you
What we see, then, is a complete lack of regard for the niceties of grammar
(Yang, 2013). And this is the case even when, arguably, the demands placed
on chimpanzees are considerably curtailed in comparison with young
children. This is because Washoe, Nim, Kanzi and the rest have never been
taught morphology, the aspect of grammar concerned with the internal
structure of words (see Chapter 7). Thus Nim could only ever produce the
bald utterance Nim eat, not having been exposed to the subtleties of Nim is
eating or Nim ate. In other ways, too, the language that chimpanzees have
been exposed to constitutes a stripped down version of a fully fledged natural
language. Given that the linguistic demands placed on them have been
reduced in this way, one might have expected more from the learning they
did undertake. But even with much of the grammatical clutter brushed to one
side for them, chimpanzees clearly cannot acquire even basic aspects of
grammar, like consistent use of word order.
Comprehension of spoken English by Kanzi
Before we finally give up the ghost, we should take into account the
distinction between production and comprehension. So far our emphasis has
been on the language that chimpanzees can produce themselves. But we
might also consider what they can understand. Perhaps the neuromotor co-
ordination required to produce signs or press lexigram symbols obscures the
chimpanzee’s true level of linguistic competence. It is certainly well known
that young children experience a lag between comprehension and production.
Two-year-olds generally understand far more words than they can actually
utter themselves (Griffiths, 1986; see also Chapter 6). It is conceivable, then,
that chimpanzees are in the same boat. Accordingly, Savage-Rumbaugh et al.
(1993) tested Kanzi’s ability to follow complex instructions, issued in spoken
English. Recall that Kanzi’s language learning was always, in a sense,
bilingual, involving exposure to both lexigrams and spoken English. Typical
instructions are shown below:
Go to the microwave and get the shoe
Give the monster mask to Kelly
 Turn the flashlight on
Give the big tomato to Liz
Take the ice back to the refrigerator
Savage-Rumbaugh et al. (1993) report an overall success rate of 71 per cent
in Kanzi’s ability to carry out instructions of this kind. But the criteria for
judging success are extremely liberal. More to the point, there is no evidence
that Kanzi can understand every word addressed to him. And the basis on
which he acts can very often be ascribed to pragmatic factors, rather than his
interpretation of grammatical rules to get at meaning. For example, if you
hand me a flashlight I will probably turn it on, because that’s what one does
with flashlights. If I hear Give the big tomato to Liz, I would only need to
know the words tomato and Liz to guess that these two ‘objects’ should be
married up. As it happens, Kanzi took two tomatoes of different sizes, and
gave them both to Liz. This response was scored as ‘Partially Correct’, even
though the intention of the sentence was to distinguish between large and
small exemplars of a category (tomato). Using a stricter definition of
‘correct’, where Kanzi performs the requested action immediately, and where
we include only blind trials in which the instructor is out of sight, Kanzi’s
success rate drops to 59 per cent. And, as noted, this success may be achieved
without a knowledge of grammar. Instructions can often be carried out
correctly from an understanding of one or two individual word meanings.
Overall, then, the work on Kanzi’s comprehension does not lend any extra
confidence to the idea that apes can acquire grammar.
The linguistic limitations of animals
Research on ape language dwindled fast in the 1980s, largely in the wake of
the scepticism expressed by Terrace (1979), a ‘chimp insider’ who set in
motion an increasingly acrimonious debate (for example, Wallman, 1992,
1993; Lieberman, 1993). In consequence, it is difficult to find a balanced
approach to this topic, so reader beware. Nativists tend to be extremely
dismissive of apes’ linguistic abilities (for example, Pinker, 1994) while
others, especially ape researchers, tend towards excesses in the opposite
direction (for example, Savage-Rumbaugh et al., 1993). In fact, both sides of
the argument have some merit. Apes – and other animals – have
demonstrated abilities that no-one dreamed of a century ago. As we have
observed, they are capable of acquiring information about individual words.
In so doing, some of Hockett’s (1963) defining features of language have
been assailed. Čadková (2015) suggests that these abilities evolved in parallel
across different species, independently from one another. She further argues
that if language cannot be considered the exclusive ‘quality’ of a single
species, then Hockett’s design features are undermined and in need of a total
rethink: language should no longer be considered the sole preserve of human
beings. Hockett overemphasized spoken language (vocal-auditory channel,
broadcast transmission, rapid fading, specialization) at the expense of
alternative media for language, including sign language and writing
(Čadková, 2015). But Hockett’s features retain their value for research. It is
the combination of all 16 design features which together identify language.
Moreover, some of Hockett’s features, like duality of patterning, do seem to
remain the sole preserve of human language.
When provided with an appropriate medium (e.g., lexigrams), chimpanzee
words show semanticity, arbitrariness and, sometimes, displacement. And the
essential symbolic function of words appears to be within the grasp of
chimpanzees. There has even been at least one case of cultural transmission,
where one chimpanzee (Kanzi) has learned words from another chimpanzee
(his mother), rather than from a human trainer. At the same time, though, we
should be clear that chimpanzee word learning does not have the same
flavour as that of a child engaged in the same task. It is slow, effortful, and
very limited in scope. And the lexical knowledge that the chimpanzee
acquires is far less rich than that attained by the human child. Children
acquire information about the grammatical categories of individual words,
including information about what kinds of words can go together in words
and phrases (for example, the article the is followed by a noun in phrases
like the dog, the biscuit, and so on). Beyond word learning, no animal has
succeeded in acquiring grammar, not even the star turn, Kanzi. And with
respect to the purpose of language, it is also apparent that no non-human
animal really gets it. No amount of intensive training will ever induce a
chimpanzee to express an opinion, ask a question, or phone a friend for a
chat. Their use of language is almost tool-like, an instrument used to obtain
food or some other desired goal. There is no compunction on the part of a
chimpanzee to engage in conversation.
Maybe we have been asking too much of our primate cousins. The Hayes
wanted Viki to speak, while the Gardners wanted Washoe to sign like a deaf
person. The methods of training chimpanzees did improve with each attempt,
most notably with the lexigram keyboard introduced by Savage-Rumbaugh.
However, it is one thing to adapt the medium of communication to
chimpanzee tastes – it is quite another to consider how suitable the learning
task is in the first place. The aim has been to teach language to a primate –
words and grammar – the whole kit and caboodle. In this respect, we have
certainly expected too much of chimpanzees. Weiss & Newport (2006: 247)
point out that ‘this paradigm disadvantages primates in that it requires them
to acquire a heterospecific (another species’) system of communication’. The
fact that chimpanzees can acquire any aspects of language at all is therefore
surprising and leads us to suspect that some similarities do exist across
species in the way that linguistic information is processed. We will pursue
this idea further in the next section.
Is speech special?
Categorical perception in infants and primates
As noted, efforts to teach Language (with a capital L) to chimpanzees faded
away with the twentieth century, the last major attempt being Savage-
Rumbaugh et al. (1993). However, comparative research has been
reinvigorated by a change of direction. This new line of enquiry examines the
extent to which basic psycholinguistic mechanisms are shared across species,
in particular, the rapid, automatic processing of speech sounds. Very early in
life, human infants demonstrate an ability to process speech in a very adult-
like manner. In particular, infants discriminate between different speech
sounds, or phonemes, in a categorical manner (Eimas, Siqueland, Jusczyk &
Vigorito, 1971; see Box 2.2). For example, the phoneme /p/ can be
pronounced in many different ways. But we do not perceive any of these
differences. Our perceptual systems place all the different /p/s in a single box
(or category) marked /p/. We will examine categorical perception in much
more detail in Chapter 5. For the moment, we maintain the focus on our
comparison between humans and non-humans.
The discovery of categorical perception in 1971 pointed to a language-
specific, uniquely human, inborn ability. However, it was not long before
human infants were placed in an unlikely menagerie that included chinchillas,
macaques, budgerigars and even Japanese quails. These animals, too, can
make categorical phonemic distinctions, a finding which suggests a general
property of avian and mammalian auditory systems (Kuhl & Miller, 1975;
Kuhl & Padden, 1983; Dooling & Brown, 1990; Kluender, 1991). This
finding seems rather odd at first, because only humans have speech. Why
would chinchillas need to distinguish /p/ from /b/? Kuhl (2004) suggests that,
in evolution, categorical perception for sound came first. As language
evolved, it exploited a domain-general auditory capacity that is shared with
other animals. The processing of sound for communication might then have
become more specialized. Of note, adults use different brain regions for
processing speech versus non-speech sounds (Vouloumanos, Kiehl, Werker
& Liddle, 2001). But then again, monkeys recruit different brain regions for
processing communicative calls versus other kinds of auditory stimulus.
Moreover, monkey calls are processed in regions of the left hemisphere that
correspond well with brain regions used by humans for speech processing
(Poremba, Malloy, Saunders, Carson, Herscovitch & Mishkin, 2004).
Box 2.2 Phonemes and the Notation for Speech Sounds
In order to represent the speech sounds in words (the phonemes),
there is a special notation system. Phonemes are not the same as
letters, but confusingly, there is some overlap in the notation
systems. Some phonemic symbols look like ordinary English
letters (for example, / p, b, t /), while others require symbols that
are not found in the conventional alphabet (for example, θ, %, æ).
The easiest way to identify a phonological representation is
through the use of slash brackets. Thus, the word cat can be
spelled with the letters c-a-t, or represented phonemically as / k æ
t / within slash brackets. Observe the use of /k/ to represent what
is often called a ‘hard C’ in English. The use of /k/ to represent
this phoneme allows us to distinguish the hard C in cat, /k/, from
the ‘soft C’ in ceiling, /s/. And it also allows us to recognize that
the words cat and king start with the same sound, or phoneme, /k/,
despite the idiosyncrasies of the English spelling system.
Note also that the vowel in cat is represented by the symbol /æ/.
Vowel sounds, in particular, tend to have curious symbols, not
found in the English alphabet, but this can apply to consonants
also. For example, in king, the final sound is spelled –NG (yes,
it’s one sound, not two). The phonemic symbol for this phoneme
is /ŋ/.
The full range of possible human speech sounds is encompassed
within a single notation system known as the International
Phonetic Alphabet (IPA) (see the website listing at the end of
Chapter 5). The IPA system allows any word from any language
to be transcribed, providing an accurate record of how to
 pronounce it. A simplified version, listing the phonemes of
English, is provided in Appendix 2.
But humans and non-humans are not identical when it comes to speech
processing. First, although animals perceive speech categorically, animal
phoneme boundaries are different from those perceived by humans in about
one third of cases studied so far (Sinnott, 1998). Second, it has been argued
that infants have a preference for listening to speech over other sounds. For
example, newborn infant brains respond more strongly to samples of Infant
Directed Speech than they do to the same samples played backwards (Peña,
Maki, Kovačič, Dehaene-Lambertz, Koizumi, Bouquet & Mehler, 2003).
Infant preferences have also been examined by manufacturing synthetic
sound samples that resemble speech on several dimensions. These include the
fundamental frequency (which provides information on pitch), duration,
pitch contour, amplitude and intensity (Vouloumanos & Werker, 2007).
Given a choice between real speech and the non-speech analogue, infants
preferred human speech. But despite the strenuous efforts to come up with a
speech-like alternative for infants to listen to, Rosen & Iverson (2007) point
out that the simulation of voice pitch remains very crude. It is very difficult to
establish whether infants prefer speech over other sounds. This follows from
the difficulty in constructing a sound stimulus that has the qualities of speech
without actually being speech. What we do know is that the preference for
speech is not present at birth. Neonates are equally happy listening to either
human speech or rhesus monkey vocalizations, though both of these are
preferred to synthetic (non-primate) sounds (Vouloumanos, Hauser, Werker
& Martin, 2010; Shultz & Vouloumanos, 2010). By three months, any bias
for monkey vocalizations has dropped away. Hence, speech may be
something that infants focus on very early in life due to a massive amount of
exposure.
But there is a third way in which speech can be seen as special. Human
infants demonstrate categorical perception effortlessly, whereas it takes many
thousands of training trials of supervised learning (operant conditioning) to
induce categorical perception in animals. We must also set this great learning
effort against the fact that ‘humans do not just make one-bit discriminations
between pairs of phonemes. Rather, they can process a continuous,
information-rich stream of speech’ (Pinker & Jackendoff, 2005: 207). It
seems that once again we have an example of overlap (not identity) between
human and animal cognition: animals can process speech sounds in a way
that is analogous to human perception. The contrast lies in the facility with
which humans execute the processing and learning of linguistic stimuli. This
observation echoes our conclusions (above) about vocabulary learning in
chimpanzees. It all comes much more naturally to us. There may not be
anything inherently special about speech as an acoustic experience, but there
is certainly something special about what we do with speech in the
development of a linguistic system. This point has been underscored recently
by Marcus, Fernandes & Johnson (2007), who showed that infants learn rules
more easily from speech stimuli than from non-speech stimuli (though see
Saffran, Pollak, Seibel & Shkolnik, 2007, for contrasting evidence). Overall,
the picture is somewhat mixed. There is evidence that speech is processed
differently from other acoustic signals from very early on in life. At the same
time, there are signs that some animals can also process speech categorically.
Some animals also dedicate brain regions to communicative calls
corresponding to brain regions that humans use for speech processing. We
may not yet be able to declare unequivocally that speech is special to humans
but, clearly, what we do with it, in terms of processing and learning, is
special.
Statistical learning
Recent comparative research has examined our ability to process and
remember information from extended sequences of speech. Infants can learn
to recognize particular syllable patterns that occur repeatedly within a long,
uninterrupted sequence of syllables (Saffran, Aslin & Newport, 1996). For
example, an infant can learn to recognize the sequence golabu within a longer
sequence, like the following:
padotigolabupadotibidakutupirotupirogolabubidaku
This remarkable ability is often referred to as statistical learning and
provides one key to explaining how young children begin to isolate
individual words from the torrent of speech assailing their ears. How do
infants do this? Why is it called statistical learning? We shall return to these
questions in Chapter 5. For the moment, we want to know if animals also
possess the ability to recognize specific, word-like sequences of speech. For
those of you who have to go out now, the answer is yes. For those of you who
can stay a little longer, we must qualify that yes: humans and primates are
similar, but not identical.
In one respect, cotton-top tamarin monkeys perform like human infants
(Hauser, Newport & Aslin, 2001). They can isolate predictable syllable
sequences (nonsense words, like golabu) from a continuous stream of speech.
Two basic processing abilities therefore seem to be shared between humans
and monkeys. First, there is the ability to distinguish one syllable from
another and treat them as elementary units. Second, we share an ability to
keep track of the serial order in which those syllables occur, together with
information about which syllable sequences co-occur regularly. However,
subsequent research points to subtle differences between human adults and
tamarins (Newport, Hauser, Spaepen & Aslin, 2004). Syllables might not, in
fact, be the basic unit of analysis. Accordingly, this later study teased apart
two factors in the speech stream that could be used as the basis for learning
about the serial order of word-like sequences. The first was the syllable (for
example, go, la, bu) and the second was the phoneme, either consonants (for
example, /g, l, b/) or vowels (for example, (/o, æ, u/). When we listen to a
continuous stream of speech, do we pay attention to whole syllables or
individual phonetic segments? The answer seems to be that humans prefer
phonemes, while tamarins have a penchant for syllables. Both monkeys and
adults were exposed to 21 minutes of continuous speech, of the padotigolabu
kind. Humans did well at identifying words (predictable stretches of speech)
when the stimuli contained predictable relations between either consonants or
vowels. The tamarins did well when the predictability of the stimuli lay either
in the syllable structure or in the vowels. Vowels are sometimes described as
the nucleus of the syllable and it might be argued that vowels and/or syllables
are relatively large, holistic units of speech. In contrast, humans can cope
with the more fine-grained phonetic distinctions to be found between
different consonants. The basis for this cross-species difference is not entirely
clear yet, but the fact that it exists is important in helping explain the human
facility for language.
Back to grammar: Infants versus monkeys
The new approach to comparative learning has recently ventured into the
domain of grammar (Saffran, Hauser, Seibel, Kapfhamer, Tsao & Cushman,
2008). Concordant with this new approach, the task presented to learners is
highly focused. No attempt was made to teach the whole of Grammar (with a
capital G) as in Savage-Rumbaugh et al. (1993). Instead, a language was
constructed with very simple grammatical rules. The aim was to see if
tamarins can acquire any kind of rule characteristic of human language. The
focus was on so-called predictive dependencies, whereby the presence of one
word can be used to predict the presence of another in a phrase. For example,
the presence of an article like the means we can predict fairly certainly that a
noun (nematode, perhaps) will follow, either directly (the nematode) or very
soon (the white nematode). Articles do not occur by themselves. Whereas I
can predict a noun from the presence of an article, the reverse is not true.
Nouns can occur with or without an article (I like nematodes, My friend
named his baby Nematode, and so on). Saffran et al. (2008) constructed two
artificial, miniature languages, one with, and one without, predictive
dependencies. Following training, they found that 12-month-old infants could
distinguish between grammatical and ungrammatical sentences from the
Predictive language, but not from the Non-predictive one. Tamarins,
meanwhile, performed in a similar fashion to human infants, but only on
simple dependency relationships. When the complexity of test materials was
increased, tamarins dropped out of the race, leaving human infants out in the
lead. We find another example, therefore, of subtle differences between
humans and non-humans that have very large ramifications. Tamarins are
capable of learning simple rules, even rules that look language-like
(Neiworth, 2013). But the degree of complexity is strictly limited. In this
vein, another study has found that rats can learn a simple rule (Murphy,
Mondragon & Murphy, 2008). The rule determines the order of two kinds of
element (A and B) in three-element sequences. Rats can distinguish AAB
from ABA sequences. Zebra finches have also been trained to discriminate
patterns of this kind, using elements of songs (Chen, Rossum & ten Cate,
2015). Both rats and finches attend to repeated A or B elements and
demonstrate rule-governed behaviour. But the complexity of rules which can
be acquired, and the ability to generalize them to novel elements is strictly
limited (ten Cate, 2014). Overall, we can see that animals are capable of
learning rules, even the kinds of rules that appear in human language. They
are strictly limited, however, with regard to the complexity of the rules that
can be acquired.
The language faculty: Broad and narrow
In our day trip to the zoo, the overriding interest has been in discovering
what, if anything, is special about language. A recent approach to this issue
divides the faculty of language into two components: broad and narrow
(Hauser, Chomsky & Fitch, 2002). The faculty of language – broad (FLB)
embraces the biological capacity that we have, as human beings, to acquire
language. FLB is one of the defining characteristics that distinguishes human
beings from other animals. At the same time, much of the biological capacity
underpinning the broad faculty is assumed to derive from shared origins with
animal communication. For example, parts of the human conceptual system,
in particular those aspects of cognition that allow causal, spatial and social
reasoning, are present in other animals, especially other primates (Buttelman,
Carpenter, Call & Tomasello, 2007). The mechanisms of FLB, therefore, are
shared with other animals, ‘in more or less the same form as they exist in
humans, with differences of quantity rather than kind’ (Hauser et al., 2002:
1573). Embedded within FLB, as a subset, is the faculty of language –
narrow (FLN). FLN comprises just those phenomena that are unique, both to
language and to human beings (see Figure 2.3).
What does FLN consist of? The answer is an open question. Some authors,
like Pinker & Jackendoff (2005) argue that there are numerous aspects of
language that are uniquely human, including: speech perception, speech
production, phonology, words and grammar. They acknowledge that, for
several of these phenomena, animals display an ability to learn at least
something. But Pinker & Jackendoff emphasize the gulf that generally exists
between humans and non-humans. The speed and facility of human learning
goes way beyond any animal achievement, and the full complexity of what
can be learned also leaves animals way behind in the slow lane. Pinker &
Jackendoff (2005) emphasize that many aspects of human conceptual
understanding would be impossible without language. These range from
seemingly simple concepts like a week, to the possibility that numbers could
not exist without language. The narrow faculty of language does not look so
very narrow on this view. However, other authors have a much more
restricted view of FLN. Chomsky has argued that a single, very simple
property of language underpins what is both unique to language and uniquely
human (Berwick & Chomsky, 2016). This property, known as Merge, allows
two linguistic units to combine to form a distinct new unit. When the
adjective crazy is combined with the noun Trump, we get a new unit, the
noun phrase crazy Trump (we also get a shiver down the spine). This new
unit, with its internal structure, allows for the possibility of hierarchical
structure in language.
Figure 2.3 The faculty of language: broad (FLB) and narrow (FLN)
So that’s the long and the short of it. Or rather, the broad and the narrow. The
distinction between FLB and FLN may have some mileage for
conceptualizing the human faculty of language. At present, there are clearly
some rather strong disagreements as to which aspects of language should be
allocated to each component, broad versus narrow. And therefore, there
remain disagreements about what, if anything, is unique to humans about
language. That no non-human acquires language spontaneously we can all
agree on. There is also broad agreement that some aspects of language can be
acquired by animals, but in a different way from human infants. The huge
effort chimps expend on acquiring just a few dozen words is one example of
this difference. But recent comparative research has shown that humans and
primates share some basic mechanisms for processing speech, albeit with
certain differences. Echoing Darwin, Bekoff (2013) reminds us recently that
‘the differences among various animals are differences in degree rather than
kind’. We see, then, that the linguistic gulf between humans and other
animals is not as wide as might have been expected.
In a Nutshell
Language, talk and communication are distinct concepts: the
central question concerns animals’ capacity to acquire
language.
Animal communication systems overlap in various ways
with language. But animal communication remains strictly
limited in key ways, including its lack of creativity,
semanticity, arbitrariness and displacement.
Numerous attempts have been made to teach language to
animals, including chimpanzees, gorillas, orang-utans,
dolphins, dogs, parrots and cotton-top tamarin monkeys.
Chimpanzees, dogs and parrots are capable of acquiring
words, at least with regard to the pairing of a word form with
a concrete concept. Animal vocabularies do not exceed a few
hundred items.
Animals cannot acquire grammar. Chimpanzees can combine
words into multi-word utterances, but there is no sign that
they do so in any consistent, rule-governed manner.
However, some animals, including rats and zebra finches,
can learn simple rules based on the order of elements and can
distinguish AAB from ABA sequences.
Recent research examines the extent to which basic
psycholinguistic mechanisms are shared. In particular,
research has focused on the ability to process and learn from
rapid sequences of speech sounds.
Cotton-top tamarins can perceive phonemes and identify
word-like units in continuous speech. The way they do so
may differ somewhat from humans. Some phoneme
boundaries differ and, whereas monkeys identify whole
syllables as the unit of analysis, humans prefer phonetic
segments.
A recent way of conceptualizing language draws a
distinction between broad and narrow components of the
language faculty (FLB and FLN). One controversial view
claims that FLN comprises a single feature, the grammatical
property of Merge (Berwick & Chomsky, 2016). On this
hypothesis, all other aspects of language belong in FLB and
the mechanisms that support them are shared with animals.
Further Reading
Hauser, M., Chomsky, N. & Fitch, T. (2002). The faculty of
language: What is it, who has it, and how did it evolve? Science,
298(5598), 1569–1579.
This is the article that started a furore about what is, and is not,
special about language. The authors suggest that a single property
of grammar (recursion) is unique to humans. Other properties of
language are shared, to some extent, with animals.
Pinker, S. & Jackendoff, R. (2005). The faculty of language:
What’s special about it? Cognition, 95(2), 201–236.
And this is the furore that was started. Pinker & Jackendoff reply
to Hauser et al. (2002), arguing that numerous aspects of language
are special to humans.
Websites
Friends of Washoe: www.friendsofwashoe.org/
Google friends of Washoe. This site provides interesting
background information on Washoe, the first
chimpanzee to acquire (aspects of) human sign
language.
The Alex Foundation: www.alexfoundation.org
Dedicated to the most famous talking parrot in the
world. Even though he died in September 2007, his
legacy lives on in an organization dedicated to the
advancement of study on the cognitive and linguistic
abilities of parrots.
Video
A Conversation With Koko The Gorilla – PBS Nature (1999)
www.youtube.com/watch?v=teB-Icrog6E
Washoe, Koko, and the social exchange of language.
www.youtube.com/watch?v=3V_rAY0g9DM
Kanzi with lexigram
www.youtube.com/watch?v=wRM7vTrIIis
Kanzi understands spoken language
www.youtube.com/watch?v=TaNtf-MviLE
South Korean elephant can talk
www.youtube.com/watch?v=vLUz7E5gU2c
Still want more? For links to online resources relevant to this
chapter and a quiz to test your understanding, visit the companion
website at https://study.sagepub.com/saxton2e
3 The Critical Period Hypothesis: Now or
Never?
Contents
What is a critical period?  58
A musical interlude 58
Lenneberg’s critical period hypothesis 59
Designing research on critical periods  59
Cats’ eyes: An example from animal development 59
How to identify a critical period 61
The effects of linguistic deprivation  62
The royal prerogative: Experiments on people 63
Feral children 63
Genie 64
Different critical periods for different aspects of language 66
A happier ending: The case of Isabelle 72
Conclusions from cases of deprivation 72
Age of acquisition effects in second language learning  73
Early versus late starters: Effects on language outcomes 74
Age effects may not be due to a critical period 75
Plastic fantastic: The receptive brain 78
Deafness and late language learning  79
Two more cases of linguistic deprivation: Chelsea and E.M. 80
Early versus late learning of American Sign Language 81
Overview
By the end of this chapter you should be able to summarize Eric
Lenneberg’s hypothesis that there is a critical period for language
development. You should be able to evaluate the relative importance of
five key features that have been used to define a critical period:
1. period of peak plasticity
2. cut-off point
3. occur early in development
4. brief
5. deprivation has permanent and irreversible effects
You should have a good idea about how to design a scientifically valid
test of the critical period hypothesis, and use this knowledge to assess
 the value of research on language development. In this regard, you
should be able to evaluate what different populations of late language
learners can tell us about a critical period for language, including feral
children, isolated children, second language learners and deaf language
learners. This evidence reveals the need to distinguish multiple critical
periods for different aspects of language, in particular phonology and
syntax. While there is broad (though not universal) support for the
critical period hypothesis in first language acquisition, it will become
apparent that support is far less clear-cut when we examine the case of
second language learning in adults.
What is a critical period?
A musical interlude
At the age of nine I had my first piano lesson. The excitement of this
occasion was punctured by my teacher, who announced that I should have
started learning four or five years earlier. It was made pretty clear that I now
had no chance of becoming the next Horovitz (a famous pianist, not a
breakfast cereal). Neither strenuous efforts on my part, nor the ministrations
of my piano teacher, could ever compensate for the advantage of starting
early. On this view, the same teaching provided to both a four-year-old and a
nine-year-old would benefit the younger child more. My piano teacher may
have been right, since I’ve yet to make my debut at Carnegie Hall. Perhaps
unconsciously, my teacher was advocating her belief in a critical period for
music.
Beyond music, the general idea is that a particular kind of experience has a
greater effect on development when supplied during a given (critical) period
than at other times. If the key experience is not available during the critical
period, there is a negative effect on development, more strongly negative than
if the experience is missing outside the critical period. Thus, according to the
kind of experience, or lack of it, the effects on development can be either
positive or negative. Interest in critical periods has been widespread and not
confined to music. Over the past 80 years, a whole slew of critical periods
has been identified for a wide range of biological, social and behavioural
aspects of development. Let’s begin by seeing how this idea might apply to
language. We shall then take a detour into physiology, to discover how to
design a rigorous critical period study. We shall then come back to language
development to see if we can match the physiologists for rigour. You will
discover that, on the whole, we can’t. But you will also see that much has
been learnt about the critical period for language acquisition, and, in
consequence, something about the biological basis of language.
Lenneberg’s critical period hypothesis
The idea that there might be a critical period for language first took flight in
the 1960s. Eric Lenneberg (1967) – inspired by the explosion of interest in
Chomsky’s ideas on language – marshalled a wide range of evidence on
language learning and brain development in support of his thesis:
Between the ages of two and three years language emerges by an
interaction of maturation and self-programmed learning. Between the
ages of three and the early teens the possibility for primary language
acquisition continues to be good … After puberty, the ability for self-
organization and adjustment to the physiological demands of verbal
behavior quickly declines. The brain behaves as if it had become set in
its ways and primary, basic, skills not acquired by that time usually
remain deficient for life. (Lenneberg, 1967: 158)
Lenneberg’s hypothesis has been around for a long time but continues to
stimulate research. We should note straightaway, though, that, in pretty much
every detail, Lenneberg turned out to be wrong (Snow, 1987). For example,
language acquisition begins before birth, not at two years (see Chapter 1).
And the offset of any critical period – the point at which sensitivity starts to
decline – must be more like five years, not the early teens, since most of the
basic grammar we acquire is in place by age five (see Chapters 8 and 9).
Disputes about the details of Lenneberg’s proposals have not detracted from
their core validity. It remains plausible that language development is subject
to critical period effects. But we begin our exploration, not with language in
humans, but with vision in cats. The scientific rigour afforded by animal
research provides a useful benchmark for judging the value of empirical
findings in the domain of child language. It also provides an excellent
framework for exploring the characteristic features of critical periods in
general, before we consider the special case of language.
Designing research on critical periods
Cats’ eyes: An example from animal development
We begin by taking a close look at research on the development of vision in
kittens. Our interest in this case is twofold. First, the series of experiments in
question provides a model of how to go about doing research on critical
periods; so good, in fact, that the two authors, David Hubel and Torsten
Wiesel, won the Nobel Prize in 1981. A second reason for looking at Hubel
& Wiesel’s work is that it provides an example of how development can be
arrested, or seriously ruptured, when essential experiences are absent during
a critical period. In this respect, there is a direct parallel with research on
language acquisition.
Hubel & Wiesel (1963) monitored the effects of visual deprivation on neural
activity in the cat visual cortex. Kittens were deprived of normal visual
experience via surgical closure of the right eye. Not to put too fine a point on
it, the eyelids were sewn (or sutured) together (see Discussion point 3.1,
below). Two factors were systematically varied: (1) the age at which visual
deprivation occurred; and (2) the duration of that deprivation. Thus, the age
at the time of eye closure ranged from 10 days through to adulthood, while
the length of time that the right eye remained closed ranged from three days
through to 180 days. Following the period of deprivation, the sutured eye was
re-opened and the kitten’s responses to visual stimuli were recorded.
Measurements were taken from groups of cells in the striate cortex, which is
part of the ‘grey matter’ of the brain’s outer surfaces, in this case towards the
‘rear’, where brain areas controlling vision are located. Groups of cells in this
region are organized into so-called ocular dominance columns. Most of the
cells in a column tend to respond to visual input from one eye or the other. So
the aim was to gauge the effects of deprivation on columns that would
normally respond to input from the sutured (right) eye. Measures were taken
by first anaesthetizing the kittens, a procedure which prevented eye
movements and which also allowed recordings of electrical activity to be
made from individual cells via tungsten microelectrodes inserted into the
cortex.
In a series of studies, Hubel & Wiesel (1963, 1970) found that if the right eye
is kept closed for the first 10–12 weeks after birth, it renders the kitten blind
in that eye. In effect, inputs to the deprived eye become disconnected from
the brain, so visual stimulation produces no effect when the right eye is re-
opened. What is more, this effect seems to be permanent. Even very brief
periods of deprivation, as little as three days in some cases, can have drastic
effects. There is a ‘use it or lose it’ principle at work here. Susceptibility to
deprivation begins in the fourth week after birth, at about the time that kittens
begin using their eyes. And the effects of deprivation were found to be
strongest six to eight weeks after birth, declining thereafter until about 16
weeks. Tests on older cats, though, revealed that the visual system becomes
remarkably robust when faced with deprivation, provided the deprivation
starts beyond the critical period. In one case, eyelid closure was maintained in
an adult cat for 16 months with no adverse effects when the eye was re-
opened. It appears very clearly, therefore, that the same experience has
dramatically different effects, depending on when it occurs in life. In the case
of cat vision, the critical period is very limited and occurs early on in
development.
Discussion point 3.1: Ethics in science
George Bernard Shaw once argued that ‘if you cannot attain to
knowledge without torturing a dog, you must do without knowledge’
(Shaw, 1911: xlvii). To consider this position, the following website
may prove useful: www.peta.org. Consider the following:
Scientific testing with animals has the potential to save millions of
lives, as when new drugs and medical procedures are discovered.
In such cases, is animal testing justifiable?
Can animal testing ever be justified in psychology?
How to identify a critical period
At this point, it is useful to consider the methodology of critical period
research, since it will better enable us to appreciate the value of research
devoted to language. We can again take Hubel & Wiesel (1963) as the
benchmark of rigour in the way they designed their experiments. Bruer
(2001) provides a useful analysis of their design and points out that there are
two sensible approaches:
1. Give the participants the same experience (in both quality and duration),
but at different stages in development; or
2. Systematically vary the duration of the target experience, but ensure that
the starting point is the same for all participants.
Both of these research designs avoid the confound of a practice effect. To
illustrate this point, let’s return briefly to my early life at the piano. I started
learning the piano at the age of nine. Mozart, meanwhile, was picking out
tunes on the harpsichord at the tender age of three. He learned his first piece,
a scherzo by Wagenseil, a few days before his fifth birthday. And by the age
of 12, Mozart was entertaining the crowned heads of Europe, at which age, I
was playing like I’d traded my hands in for flippers. But just a second, maybe
this is not a fair comparison. Natural genius aside, Mozart had six more years
of practice than me. This means that both kinds of design, above, are
violated. With respect to Design (a), Mozart and me differ in terms of both
duration and quality of experience (Mozart’s father was no musical slouch
himself and provided his son with a rich musical experience). With respect to
Design (b), of course, Mozart and me started learning at different ages.
Hubel & Wiesel conducted careful experiments that conformed to both kinds
of design outlined above. They were also careful about two other factors that
should be taken into account in work of this kind. The first consideration is
the need to be clear about what the target experience is. In the case of cat
vision, it was simply normal exposure to objects and light in the visual field.
The second point of concern is the outcome measure. That is, how do we
measure the effects of experience on the participants? Again, for Hubel &
Wiesel, the answer is very straightforward. How well did the cats see at the
end of the deprivation period? The answer could be determined
unequivocally, by counting the number of cells in the ocular dominance
columns that responded to stimulation. In physiological studies of this kind,
both the experience and the outcome measure can be clearly defined. As we
shall see below, the same cannot be said for language.
Just before we launch off into linguistic waters, it is worthwhile summarizing
the key features of critical periods from classic research like that of Hubel &
Wiesel:
1. period of peak plasticity, when the system is especially open to
experience
2. cut-off point, beyond which plasticity is greatly reduced – the
environment no longer has any effect
3. tend to occur early in development
4. tend to be brief
5. effects of deprivation are permanent and irreversible
It is worth pointing out that none of these five features is entirely sacrosanct.
In fact, there are problems with all of them. For example, critical periods are
not always brief and are not always characterized by a sharp cut-off point
(Bruer, 2001). Instead, there is often a gradual decline in plasticity. As we
shall see, this latter point has special relevance for research on language
learning. With regard to timing, critical periods do seem to be brief and
feature early in development (though 13 years for language could be seen as
stretching the point somewhat). The general point is that critical periods are a
developmental phenomenon. They are associated with genetically determined
periods of growth during which the system is especially plastic, in the sense
that it is especially receptive to particular environmental inputs. Once
development is complete, there is no need for the system to be susceptible to
key learning experiences any longer. Hence, one encounters critical periods
early in an organism’s life.
The effects of deprivation during the critical period are not always absolutely
permanent, even in the case of cat vision (Chow & Stewart, 1972; Harweth,
Smith, Crawford & van Noorden, 1989). When it comes to language learning,
we shall see that our ability to acquire a second language as adults presents a
special puzzle. Given that our five defining features are not chiselled in
marble for us, some scientists prefer the phrase sensitive period, instead of
critical period. The term sensitive period carries with it the idea of an
extended period during which receptivity to experience peaks and then very
gradually declines (for example, Immelmann & Suomi, 1981). Despite this
care in defining terms, the phrase critical period tends to be far more popular,
especially among language researchers. And the concept of a sensitive period
brings its own complications (see Freedman, 1979, as your guide through the
minefield). In the domain of language acquisition, the term critical period
dominates the literature, so we shall retain it here.
The effects of linguistic deprivation
Now that we know how to design a good test of the critical period hypothesis,
what would be the perfect experiment in the domain of language acquisition?
If we follow the precepts laid down above, we might vary the age of exposure
to language for different children, while keeping the length of exposure
constant. This would mean keeping children waiting somewhere, for different
lengths of time, until we were ready to start talking to them, or rather,
supplying linguistic experience. Alternatively, we could vary the amount of
exposure to language, while keeping the age at which we start constant. Of
course, no-one in their right mind would attempt a study like this. Sadly,
though, the callousness and depravity that one human being can inflict on
another has yielded several cases where children appear to have been
deliberately denied access to language.
The royal prerogative: Experiments on people
Every so often, history relates how some bored, sadistic monarch decides to
experiment on the effects of linguistic deprivation. At least four cases are
well known, if not well attested, thanks to Campbell & Grieve (1982):
 Psamtik I, Saitic king of Egypt (7th century BCE)
Frederick II of Sicily (1192–1250)
James IV of Scotland (1473–1513)
Akbar the Great of India (1542–1605)
In each case, one or more children were deliberately deprived of language in
order to gauge the effects. Psamtik I thought he would establish beyond doubt
what the original language of the world was. He assumed that the first word
uttered by a child, if they had never heard anyone speak, would be in this
original language. Psamtik naturally placed his money on Ancient Egyptian
(though, of course, the language was not so ancient in his day). But he was
wrong: the winner was Phrygian, the language of an Indo-European people
settled in Anatolia (part of modern Turkey). See Campbell & Grieve (1982)
for a fascinating review of the four cases listed above. Cutting to the chase,
we can conclude that these Royals were not so hot at science, and also that
science reporting left much to be desired in days gone by. For example, the
story of Psamtik I is not a contemporary account, being retold a couple of
centuries later by Herodotus (485–425 BCE). What is more, some of the
details Herodotus includes betray the story as apocryphal, including the use
of goats’ milk to suckle infants. It turns out that in Egypt at this time, the only
use for goats’ milk was in the relief of anal dysfunction! All four cases listed
above could have been rigorous, albeit deeply unpleasant experiments on
linguistic deprivation, but apart from the case of Akbar, we do not know for
certain that they even took place.
Feral children
Deliberate experiments aside, numerous cases have been reported of children
who have been deprived of normal social contact and exposure to language.
Perhaps the earliest report is the legend of Romulus and Remus, twin boys
raised by wolves. Many such children have been discovered in the wild,
having been abandoned by their parents, with animals like wolves, jackals
and leopards taking over the parental role. It remains a mystery what brings
out the nurturing side of these animals in the face of a perfectly good meal
like a human infant. On discovery, these children are notably reluctant to
relinquish either their animal habits or diet. Typically, also, they have an
aversion to human company and can endure extremes of temperature. A
systematic description is provided by Linnaeus in the tenth edition of his
Systemae Naturae, published in 1758. Linnaeus classified these children as
Homo Ferus, a separate subdivision of Homo Sapiens. Homo Ferus was
distinguished by the features of tetrapus, hursutus and mutus, which for those
of us with little Latin and less Greek, means that they walked on all fours,
were hairy, and were without speech. Of course, it is the lack of speech that
interests us, and the fact that all of these children suffered a period of
linguistic deprivation in early childhood. See Table 3.1 for a few examples of
feral children who have been discovered over the past three centuries.
Do these cases shed any light on the critical period hypothesis? In many
cases, no, since they often turn out to be hoaxes. A case in point is the
Nullarbor Nymph who was supposedly found hopping about with kangaroos
in the Australian outback in 1971. Other cases may be true, but are difficult
or impossible to verify, including the Irish sheep-boy of 1672 and the
Mauritanian gazelle boy of 1900. There is, though, a small number of well-
attested cases. One of these is Victor, the Wild Boy of Aveyron, and in our
review of his case we need to consider the following factors: the period of
deprivation; his age on discovery; and how well he did subsequently in the
task of acquiring language. Victor was discovered in France at the turn of the
nineteenth century, at the age of something like 10 or 12. He was eventually
taken to Paris under the care of a young doctor called Itard. But despite
concerted efforts, attempts to civilize Victor and teach him language were
almost entirely unsuccessful. He acquired only two spoken expressions –
‘Oh, Dieu!’ and ‘Lait!’ – and while calls for God and milk can be useful, they
do not constitute a full-blown command of French.
Itard’s first report on Victor was written in 1801 and promised much, but five
years later, the tone of his second report was much more pessimistic. Itard
concluded, with remarkable prescience, that the ‘imitative force … and the
apprenticeship of speech, … which is very energetic and active during the
first years of … life, wanes rapidly with age’ (cited in Lane, 1976: 129). On
the face of it, Victor seems to confirm Lenneberg’s hypothesis. But precisely
how old was Victor when he turned up in Aveyron? Also, how old was he
when he was abandoned? Clearly, it is impossible to specify the length of
linguistic deprivation, which means that one of the fundamental maxims of
critical period research is violated. In addition, we have very little idea
whether Victor’s other cognitive faculties, like hearing, vision and general
intelligence, were normal. Impairments to any of these might account for
poor language learning, independent of the effects of a critical period. So
although many of the details of Victor’s case are well attested, we
nevertheless lack vital information when viewing his case as a test of the
critical period hypothesis.
Genie
Feral children aside, we might also consider children who have been brought
up in isolation. These children have not been abandoned, but have been
brought up in circumstances of extreme deprivation, including lack of
exposure to language. Given the terrible cruelty involved, it is fortunate that
such cases are surpassingly rare. The most detailed account we have is of
Genie, a girl who came to light in 1970, aged 13, in Los Angeles. Genie had
been raised in circumstances of truly awful deprivation and maltreatment.
She spent her days physically restrained, harnessed to a potty seat, unable to
move anything but her hands and feet. When not forgotten, Genie would be
transferred at nights into an equally restrictive sleeping bag, designed by her
father to prevent movement of her limbs. Her days and nights were spent
confined in this way in a bare room with almost no tactile or visual
stimulation. Genie was rarely spoken to and any attempts on her part to
vocalize were met with a beating from her father. She was therefore raised in
virtual silence. Interest in Genie was intense. Chomsky’s ideas on the
biological origins of language were beginning to take hold and Lenneberg’s
hypothesis was hot off the press (Chomsky, 1965, 1966; Lenneberg, 1967).
Genie seemed to present an ideal opportunity to test out the critical period
hypothesis. Given that Genie was found at puberty, she should not have been
able to catch up with her peers. In the event, a great deal of value and interest
was discovered in the study of Genie (see below). At the same time, though,
you may not be surprised to learn that, in common with other cases of
deprivation, a lack of scientific rigour strictly curtails the scope of any
conclusions we can draw about critical periods.
It is usually reported that Genie had absolutely no language when she was
found (for example, Hoff, 2001). However, some sources report that Genie
did understand a handful of words when discovered (Rymer, 1993). These
included Mother, walk, and go, together with colour terms like red, blue and
green. Recordings of Genie reveal a very strange, high-pitched fluting
quality, far removed from normal speech. Rymer (1993) suggests that she
could produce two expressions when first found: stopit and nomore. Observe
that these expressions are written as single words, because there is no
evidence that Genie could analyse them into their component parts (see
Chapter 9 on holophrases).
Different critical periods for different aspects of
language
A concerted effort was made to teach Genie language (Curtiss, 1977). Before
we consider how well she did, we must tackle a major problem. So far, we
have the prediction that language is subject to critical period effects. But you
will recall from Chapters 1 and 2 that language is a nebulous term. The
researchers working with Genie were well aware of this problem, observing
that we lack an ‘adequate definition of the innate behavior to be elicited’
(Curtiss, Fromkin, Krashen, Rigler & Rigler, 1974: 540). We must therefore
be much more precise. Which aspects of language, in particular, are subject
to critical period effects? It is entirely possible that multiple features of
linguistic development are involved (Long, 2005).
We can begin to tackle this issue by considering each major level of language
separately: phonology, vocabulary, morphology and syntax. With regard to
phonology (speech sounds), several authors have suggested that the
development of a native speaker-like accent is subject to its own critical
period effect, quite independent of any other aspect of language (for example,
Bongaerts, Planken & Schils, 1995; Spaai, Derksen, Hermes & Kaufholz,
1996; Pallier, Bosch & Sebastián-Gallés, 1997; though see Yeni-Komshian,
Flege & Liu, 2000, for a contrary view). Adults learning a foreign language
are almost always betrayed by their non-native accent, regardless of how
good their grammar and vocabulary might be (see below). In fact, it has been
suggested that our ability to acquire a flawless accent in a second language
has lapsed by the age of five or thereabouts (Scovel, 1988). Beyond accent,
there is now good evidence that numerous aspects of early phonological
development are subject to critical period effects (Werker & Hensch, 2015).
These include the attunement of the infant to properties of the native
language such as rhythm, stress, melody and the particular phonemes and
phoneme combinations which are permitted. We can see these effects quite
clearly in babies who are born pre-term. These babies start to experience the
world in advance of babies born at full-term (40 weeks), including listening
to and processing speech sounds. Does this extra experience make any
difference to the infant’s ability to discriminate between different phonemes
(for example, /b/ versus /g/)? The answer seems to be no, at least for infants
with a gestational age of less than 30 weeks (Key, Lambert, Aschner &
Maitre, 2012). Instead, a certain level of maturation is required before infants
can make certain phoneme discriminations. The notion of multiple critical
periods is lent weight in a recent study by Huang (2014). She reports separate
critical period effects for accent and grammar within a single group of
individuals, 118 Mandarin-speaking immigrants to the US.
What about the lexicon? If vocabulary learning is taken as the target
behaviour, then we might argue that the case of Genie refutes the critical
period hypothesis, since she acquired a vocabulary of several hundred words
that she could both understand and produce. At the same time, Genie’s word-
learning pales into insignificance when compared with the prowess of a
typical two-year-old, both in terms of vocabulary size and the speed at which
it is acquired. The plasticity of Genie’s word-learning mechanisms has
clearly declined, relative to a typical toddler. There is some independent
evidence that word learning in typically developing children is subject to
critical period effects (Hernandez & Li, 2007). Recall for words learned early
in life tends to be both faster and more accurate (Morrison & Ellis, 2000).
Before we shift the focus to syntax, let’s first insert a note on morphology,
which is concerned with the internal structure of complex words. If need be,
shoot back to Chapter 1 for a quick reminder and if that doesn’t do the trick,
lurch to the end of the book, for more information from the Glossary. Then
settle down and consider the following sentence: Genie’s morphology is very
poor.
Small two cup
Like chew meat
These typical utterances are notable for their lack of inflectional
morphemes: *two cup instead of two cups, *chew meat instead of chewing
meat. As we shall see below, other research on critical periods supports the
idea that morphology suffers when language learning starts late in life.
Turning to syntax, the two examples above demonstrate that Genie did
acquire multi-word speech. In fact, some of Genie’s productions are perfectly
acceptable. These include Grandma gave me cereal, and I like Dave’s car, as
well as phrases like In the backyard, My house, and Mark’s room. However,
her route to this achievement was not typical. Genie took some time to reach
the two-word stage and hung around for quite a while before moving on. A
typically developing two-year-old reaches this milestone once they have
acquired about 50 words (Bates & Goodman, 1999). Genie, meanwhile,
commanded some 200 words before she began to combine them into two-
word utterances. Also, the two-word stage lasts just six to eight weeks in a
typical toddler. But in Genie, the two-word stage lasted much longer – more
than five months – before she developed further.
Box 3.1 Content versus Function Words
The words in a sentence can be divided into two major categories:
content words and function words. Content words carry the
burden of meaning in a sentence. Of note is the fact that they can
be interpreted in isolation. That is, we can work out their meaning
without referring to the other words in the sentence. Nor do we
need to take note of who is talking, or anything else about the
context, in order to interpret them. Content words typically fall
into one of the major sub-classes of parts of speech: verbs;
nouns; adjectives; and adverbs.
verbs: run, talk, govern, think
nouns: car, lion, museum, truth
 adjectives: hot, desperate, green, clever
adverbs: quickly, happily, deliberately, badly
In contrast, function words are like the icing on the cake. They
give us a more precise idea about who is doing what to whom in a
sentence. Typical function words include the following:
articles: a, the, some
prepositions: in, under, above, for, with
quantifiers: some, many, all
pronouns: I, you, he, she, it, we, they
auxiliary verbs: can, could, may, might, should, will
To fully understand function words, we often need to refer to the
context of utterance. When someone says: I like New York in
June, we cannot determine what I refers to unless we know who is
talking. When I say I, I’m referring to me, but when you say I, the
word suddenly refers to you. A rough rule of thumb for deciding
which words are content words and which are function words is to
think of a text message (30 years ago, I would have said a
telegram). To avoid repetitive strain injury, when keying text in
on a mobile phone, it is commonplace to eliminate all but the
most essential words in a message. So instead of texting My train
has been delayed by 30 minutes, you might send Train delayed 30
mins. As you can see, the words that get transmitted tend to be
content words, because they carry the essence of our meaning.
The existence of multi-word speech is important, because it means that
syntax was a possible acquisition for Genie. The next consideration,
therefore, is whether she combined words according to the rules of syntax. To
help us decide, here are some more of Genie’s multi-word utterances:
Like good Harry at hospital
No more have
Dentist say drink water
Tell door lock
Glass is break
Another house blue car
Unfortunately, this looks like linguistic spaghetti. To illustrate what is wrong,
we can focus on two aspects of syntax: function words and word order.
Function words are generally small words, like a, the, we, of, some, that tend
not to have much meaning by themselves (see Box 3.1). Evidently, Genie
displays poor control over function words, often omitting them:
Cookie sleep car
Bus have big mirror
In both examples, function words have been omitted. Cookie sleep car, is
probably better rendered as: Cookie is sleeping in the car (on the assumption
that Cookie is someone’s name). Bus have big mirror would be better as The
bus has a big mirror. Another aspect of Genie’s syntax we can consider is
word order. The main components of a prototypical sentence are subject,
object and verb and many languages possess a default way of ordering them.
English grammar is very strict about word order, the default being Subject–
Verb–Object. Hence, English is known as an SVO language. If you’re a bit
rusty on subject, verb and object, limber up on them in the Glossary
Gymnasium and then try the following Exercise.
Exercise 3.1: Subject, verb and object
1. Identify the subject of the sentence in the following examples:
1. Gordon’s hair looks like nylon.
2. I couldn’t help noticing the baby elephant in your fridge.
3. The man who lives next door to me has a fondness for figs.
4. The end of the world is nigh.
5. Have you ever seen a Buster Keaton movie?
2. Now find the verb:
1. Nigella licked her spoon in a rather suggestive fashion.
2. Now is the winter of our discontent.
3. Henrietta scoffed a big bag of toffees on the bus.
4. Fly me to the moon.
5. Play it again, Sam.
3. Finally, locate the object:
1. That really takes the biscuit.
2. Frankly, my dear, I don’t give a damn.
3. I simply adore champagne.
4. How many cream buns can you eat at one sitting?
5. I’ll make him an offer he can’t refuse.
answers at the back of the book
Before that ocean of learning evaporates, let’s harness it to evaluate Genie’s
control of English word order. I have parsed three of Genie’s sentences,
below, with regard to subject, verb and object. You might want to extend this
exercise for yourself with some of Genie’s other multi-word productions (my
own analyses are in the Answers section).

What we see is that Genie’s basic word order is generally correct. That is,
when she produces at least two of the three main sentential building blocks,
they tend to be placed in the right position, relative to one another.
Evidently, Genie’s ability to acquire syntax was seriously impaired. But we
have also seen that in at least one respect – basic word order – Genie does not
do too badly. Some aspects of syntax may thus be immune to critical period
effects. Or, alternatively, there may be multiple critical periods that
conceivably apply within (as well as across) different levels of language.
There are several different components of syntax (the same applies to
phonology, vocabulary and morphology). So we may have to ask which
particular aspects of syntax, if any, are subject to critical period effects. We
might consider knowledge of grammatical categories, like noun and
preposition (see Box 1.2). Or knowledge of basic word order. Or knowledge
of particular complex constructions, like relative clauses. Although several
studies examine specific aspects of language (for example, Johnson &
Newport, 1989; Sorace, 1993), none has yet considered syntax at this level of
detail. In fact, there is something fairly arbitrary in the selection of those
aspects of grammar that have been examined. But unless we specify precisely
which aspects of language might be affected by critical period effects,
assessing the effects of linguistic deprivation becomes problematic. In effect,
we are left without a proper measure of the behaviour we are interested in. As
Medawar (1967: 109) observes, ‘it is not informative to study variations of
behaviour unless we know beforehand the norm from which the variants
depart’.
How well does Genie support the critical period hypothesis? From a scientific
standpoint, the really interesting aspect of Genie’s case is that she was
discovered at the age of 13, which matches Lenneberg’s cut-off point (the
onset of puberty). Language learning should be impossible. In fact, though,
we have encountered something of a mixed bag. Genie did learn to produce
speech sounds, though with a distinctly odd pronunciation. She did acquire a
few hundred words, before reaching a plateau, though her word learning did
not mirror the speed and ease enjoyed by a typical child. Genie also acquired
some rudimentary aspects of syntax, most notably basic word order. In other
respects, though, Genie’s syntax was nothing like that of a mature native
speaker, even after several years of effort were put into supporting her
language development.
As it happens, there is a further complication with Genie’s history that makes
her case less than ideal as a test of the critical period hypothesis. The brain
comes in two halves, or hemispheres (left and right), and in most people,
language tends to be processed in the left hemisphere (Broca, 1865; Obler &
Gjerlow, 1999). Unusually, it was found that Genie processes language on
the right side of her brain. The indications are that Genie’s left hemisphere
was atrophied. It is possible, therefore, that her impaired language learning
was due to a general cognitive deficit, and not related to her language faculty
per se. All in all, the case of Genie does not provide a definitive test of the
critical period hypothesis. But the fact that she acquired some aspects of
language provides a salutary lesson. There may be multiple critical periods
affecting different aspects of language. We must therefore be precise about
which particular aspects of language we predict are subject to critical period
effects. In this vein, current research points to at least two separate critical
periods, for phonology and syntax, respectively.
Discussion point 3.2: Genie and the critical
period hypothesis
Review the evidence on Genie and consider the following points:
What aspects of language did Genie acquire?
What aspects of language did Genie fail to acquire?
Given that Genie did acquire some aspects of language, what are
the implications for Lenneberg’s original hypothesis?
How plausible is the idea that there are multiple critical periods,
each one dedicated to a different aspect of language?
What is the relevance of the critical period hypothesis for the
 nature–nurture debate?
A happier ending: The case of Isabelle
We can contrast Genie’s case with that of Isabelle, a girl who also suffered
abnormal deprivation in her early years (Brown, 1958). Isabelle was
discovered in Ohio in the 1930s at the age of six-and-a-half. An illegitimate
child, Isabelle was kept hidden from the world in a darkened room. This
cruelty was compounded by the fact that Isabelle was deprived of linguistic
input of any kind, because her mother was both deaf and mute. When
discovered, the only sounds Isabelle made were a kind of croaking noise, and
at first, it was thought that, like her mother, she was deaf. However, Isabelle
made remarkable progress. Her language learning was so rapid that she began
to put words together in sentences within two months. Within a year, she was
beginning to learn how to read. And by the age of eight-and-a-half, Isabelle’s
language was so sophisticated that she appeared to have fully caught up with
her peers. Her IQ also showed rapid improvements. Having initially scored so
low that her IQ was almost off the scale, within two years, Isabelle had a
normal IQ.
We suggested above that the offset for the critical period may be more like
five years than 13. But recall that ‘offset’ simply means the point at which
sensitivity begins to wane. The terminus is a separate concept, denoting the
very end of the line for a critical period. Beyond the terminus (still often
designated as puberty), first language learning should be impossible. Hence,
the case of Isabelle shows that even in the period when sensitivity to
language experience is in decline, acquisition is nevertheless still possible.
Firsthand details of Isabelle’s language are not readily available, and we must
rely on Brown’s summary from 1958. It would be fascinating, though, to
probe the state of Isabelle’s language abilities in more detail. With an offset
for the critical period at five years, we would predict subtle deficits in
Isabelle’s language. As with other cases of deprivation, though, the case of
Isabelle is far from ideal as a rigorous test of Lenneberg’s hypothesis. As it
stands, we can say that what information we have on Isabelle is consistent
with the critical period hypothesis.
Conclusions from cases of deprivation
Over the centuries, numerous cases of feral and isolated children have been
reported. Many such cases are vague and uninformative, while others turned
out to be hoaxes. However, that still leaves several genuine cases. But even
these desperately unfortunate children do not provide the basis for good
science. There are simply too many unknown factors that cannot be
accounted for. These include the precise period of deprivation and the precise
kind of deprivation. It is often assumed that these children had no exposure to
language at all during their early years, but there is generally no way of being
certain. As Lenneberg himself observed: ‘The only safe conclusions to be
drawn from the multitude of reports is that life in dark closets, wolves’ dens,
forests, or sadistic parents’ backyards is not conducive to good health and
normal development’ (1967: 142). The very fact that cognitive and emotional
development are likely to be abnormal means that poor linguistic
development may be due to these factors, rather than due to anything
exclusively language-related.
And yet. There remains something tantalizingly plausible about the basic idea
that critical periods exist for particular aspects of language. What we need,
though, is a more reliable evidence base than abused and neglected children.
To this end, psychologists have turned their attention to two other
populations of language learners: deaf people and people who learn a second
language. As we shall see, in both cases, the start of language learning is
often delayed, according to individual circumstances. We can therefore
compare the effects of early versus late acquisition. The critical period
hypothesis predicts that the later in life language learning begins, the less
successful it is likely to be. We can start by looking at the case of second
language learning, before concluding with the case of deaf children acquiring
their first language. A major advantage of these two populations is that they
have not deliberately been denied access to a first language. And other
aspects of development tend to be normal, including social, emotional and
cognitive development. Effects of late language learning can therefore not so
easily be ascribed to non-linguistic abnormalities in development.
Age of acquisition effects in second language
learning
Like many people condemned to life in a monolingual culture, my experience
of second language learning has been less than triumphant. A life speaking
English was interrupted at the age of 12, when lessons in French, German and
Latin were introduced at school. There followed four years reminiscent of life
at Bletchley Park during World War II. The cryptographers at Bletchley Park
were faced with breaking the code of the German Enigma machine, an early
example of Vorsprung durch technik that could encode a given message in
150 million million million different ways. Feeling like a cryptographer, I
was faced with cracking the code of three European languages (one of them
dead). Vocabulary was deduced from miscellaneous lists of words, written
out with translations. Grammar came via the use of coloured inks to denote
different verb endings and case markings. And any attempt to speak these
languages was an act of declamation, not communication. After four years, I
would have struggled to order a glass of absinthe in Paris. And not just
because I was still underage.
Why was I so bad at learning new languages? Was I past it at the age of 12?
According to Lenneberg (1967), the answer is yes. On this view, attempts to
learn a new language post-puberty must proceed without the benefit of the
specialized faculty for language acquisition. Therefore, whatever we do learn
must be achieved by making use of general cognitive learning mechanisms.
This position has intuitive appeal. It chimes with the experience of many
adults who migrate from one country to another. For many such people, even
decades after settling in a new country, their mastery of the host language
remains distinctly non-native. In this section, we will examine age effects in
language acquisition in populations that have not been abused or otherwise
deprived in terms of emotional, cognitive and educational development.
Apart from starting to learn language at different points in life, these people
can be considered normal. We are therefore provided with a useful new
method of tackling the critical period hypothesis. One quick note. To save
excessive wear on my typing fingers, I will alternate at times between L1 for
‘first language’ and L2 for ‘second language’.
Early versus late starters: Effects on language
outcomes
The effects of age of acquisition were investigated by Johnson & Newport
(1989, 1991). Their first study investigated 46 people with Chinese or Korean
as a first language. These people had moved to the United States at different
ages, ranging from 3 to 39 years. Thus, the age at which they began to learn
English varied widely, extending quite far either side of the critical period
boundary. Although age of acquisition varied, the length of exposure to L2
English was controlled for. Participants were allocated to two groups,
corresponding to early versus late arrival in the US. For the early arrivals, the
average length of stay in the US was 9.8 years, compared with 9.9 years for
the late arrivals. This is encouraging, because, as a design feature, this study
seems to adhere to one of the research maxims outlined above: give
participants the same length of experience, but at different stages in
development. Our only quibble would be that we are dealing with averages
here. Within the sample as a whole, length of exposure ranged quite widely,
from 3 to 26 years. However, it is possible, statistically, to separate out the
effects of experience from the variable of interest, which is the age at which
language learning began.
Johnson & Newport (1989) asked their participants to judge 276 English
sentences on 12 different aspects of morphology and syntax. Half of the
sentences were perfectly grammatical, while the remainder were tampered
with in some way, as below:
The farmer bought two pig at the market.
Three bat flew into our attic last night.
I hope you agree that both of these sentences sound wrong. More to the point,
they are wrong for the same reason: the absence of the plural morpheme -s
(two pigS; three batS). A native speaker should have no trouble rejecting
these sentences. As it happens, a control group of 23 native speakers did do
extremely well, though, curiously, their performance is not absolutely perfect,
as Figure 3.1 reveals. We also see a strong age of acquisition effect for the
Korean and Chinese learners. Intuitions about L2 English grammar are
native-like only for those people who arrived in the US before the age of
seven. Thereafter, there is a marked decline in performance. It is unlikely that
late-arrivals are simply guessing about grammaticality, since their
performance is well above chance. Instead, they are being inconsistent. They
are more prone than native speakers to vacillate between accepting or
rejecting test sentences. Looking at Figure 3.1, it looks as though
inconsistency increases with age of arrival. But that is not quite right. All of
these learners are inconsistent (apart from the native speakers and early
arrivals). This leads to problems in interpreting these data. How do we
compare Learner A, who scores 30 per cent on a grammar test, with Learner
B, who scores 70 per cent? It seems nonsensical to conclude that Learner B
knows 40 per cent more grammar than Learner A. Perhaps it is not the ability
to acquire grammar that is subject to the critical period phenomenon. Instead,
it might be an issue of control. People who start learning early in life seem to
have greater control and are more rigorous in applying the rules than late
learners. Plausibly, therefore, there is a critical period for the development of
our ability to process language (cf., Herschensohn, 2007: 226).
For Johnson & Newport, the key finding is the fairly sharp decline in
grammatical ability beyond puberty. If a second language is acquired at any
age in adulthood (20, 40, 70), the outcome, they argue, will be similar, and
similarly disappointing. This pattern is, of course, entirely consistent with the
critical period hypothesis. During the critical period, learning is possible, but
the efficiency of that learning declines as the period of maximum receptivity
to the input wanes. Johnson & Newport (1989) controlled for a number of
possible confounding factors, including amount of formal instruction, and the
attitudes and motivation of participants with regard to learning English. But
none of these factors were significant. What we are left with is a clear age of
acquisition effect that has been repeated in several other studies (for example,
Coppieters, 1987; Schachter, 1990; Johnson & Newport, 1991; Lee, 1992;
Sorace, 1993; DeKeyser, 2000).
Age effects may not be due to a critical period
Before we pack our bags and go home, we have to face the fact that
considerable controversy persists in the interpretation of findings on second
language research (Hyltenstam & Abrahamsson, 2000). There are three main
sticking points:
1. There may be no qualitative changes in learning after the cut-off point.
2. Some late learners do seem to achieve native-like competence in a
second language.
3. There may be no sharp cut-off point at puberty in L2 ability (despite the
picture painted in Figure 3.1).
The first issue concerns the kind of learning that takes place, both during and
beyond the critical period. If we compare the L1 toddler and the L2 adult
(post-puberty), the language learning processes deployed by each should be
different in kind. While the L1 toddler is often characterized as a sponge,
naturally soaking up language (Murphy, 2010), the L2 adult is painted more
like me at school, engaged in effortful, conscious learning. Hence, some
authors argue that the end of the critical period should be marked by
qualitative differences in the kinds of learning that take place (for example,
Hakuta, 2001). The argument goes like this: if the biologically determined
learning mechanisms, available during the critical period, are no longer
available, then any learning that takes place will depend on alternative
learning mechanisms. As a result, the kind of learning that takes place in
adulthood will be qualitatively different from that which takes place during
childhood. However, available evidence suggests that early and late starters
learn in similar ways, and make similar kinds of errors (for example,
Bialystok & Hakuta, 1994, 1999; Juffs & Harrington, 1995; Bialystok &
Miller, 1999). In other words, the kind of learning that takes place looks
remarkably similar either side of the purported critical period barrier. This is
an interesting observation, but we need to exercise caution in how we
interpret it. The product of learning (language performance on
grammaticality judgement tasks) does indeed look similar. But we currently
have no way of knowing if the mechanisms by which that learning came
about are the same or different for children and adults. In this vein, it has
been argued that adults inevitably learn languages in a different way from
children (Newport, Bavelier & Neville, 2001).
My memories of language learning at school illustrate a common story of
deliberate, strategic and effortful learning that bears no resemblance to the
way a toddler goes about it. It is difficult to argue against this image of the
adult language learner. At the same time, it does not mean that less conscious
processes, possibly more akin to those of a child, are not also working in
tandem with more explicit processes. As yet, though, no attempt has been
made to distinguish among these possible mechanisms in adults. We should
also point out that, scientifically, we are faced with two problems: (1) we
may apply different learning mechanisms to language learning in early versus
later life; and (2) both the quantity and quality of input for learning may
differ in early versus later life. We will examine the input to L1 learning in
Chapter 4. But we can note, in passing, that the input and settings in which
second languages are learned (both for children and adults) have a profound
influence on the learning outcomes (see Murphy, 2010, for a review).
Figure 3.1 Number of correct judgements of English sentences (out of 276
total) according to age of first exposure to English
 Source: Johnson, J.S. & Newport, E.L. (1989). Critical period effects in
second language learning: The influence of maturational state on the
acquisition of English as a second language. Cognitive Psychology, 21,
60–99.
The second argument against critical periods from our list concerns native
speaker status. If we accept the critical period phenomenon as real, but also
accept that many adults learn second languages to a very high standard, we
are left to ask how they do it. It has often been suggested that ‘humans bring
many high-level cognitive abilities to the task of learning a language, and
might be capable of using systems to acquire a second language other than
the ones they use for primary language acquisition’ (Newport et al., 2001:
496). For example, some authors point to a general analytic verbal ability
(think crosswords and Scrabble) as the only significant predictor of second
language success beyond puberty (Harley & Hart, 1997). These abilities must
be so good that, in a limited number of cases, they lead to native-like levels
of performance (for example, Birdsong, 1992; Ioup, Boustagui, El Tigi &
Moselle, 1994; White & Genesee, 1996; Bialystok, 1997). We might quibble
about how perfect perfection is, because native-like is not the same as native
(for example, DeKeyser, 2000), but the main point is that sightings of native-
level late second language learners are extremely rare. What we do find,
though, are considerable numbers of late learners who do very well indeed in
their second language. Whichever way we look at it, therefore, humans must
have remarkable general learning abilities to compensate for the loss of any
natural, dedicated disposition for language learning. If you examine Johnson
& Newport’s data again (Figure 3.1), you will see that even the worst adults
achieve roughly 75 per cent accuracy on the grammar tests. And although
there is individual variation in performance, most people do pretty well. It is
worth considering what this means for the portrait painted by Chomsky,
where language is described as an object of learning so abstract, complex and
difficult that specialized, innate resources are necessary to account for its
acquisition.
We come now to our third point, which concerns the definite cut-off point at
puberty predicted by Lenneberg (1967). Some authors argue that there is, in
fact, no sudden plummet in ability. Instead, they emphasize the more gradual,
long slow decline of our cognitive faculties that comes to us all with
increasing age. Hearing, vision, general problem-solving skills and memory
(both long-term and short-term) get progressively weaker with age (Seifert,
Hoffnung & Hoffnung, 2000). It is conceivable, therefore, that receptivity to
language remains strong throughout life, but that deterioration in these other,
non-linguistic, functions mediates increasingly poor language learning. These
effects are in force throughout adult life, which means that late L2 learning,
in particular, may be prey to this slide into the Pit of All-Round Poor
Learning. What we have, then, are two factors, one or both of which might
influence later language learning: (1) a critical period; and (2) a decline in
general cognition. If the second factor alone is at work, then the dramatic cut-
off point, so vivid in Figure 3.1, should not exist. Put another way, perhaps
Johnson & Newport (1989) would not find this definite kink in the graph if
they had done a bigger, better, more powerful study.
Hakuta, Bialystok & Wiley (2003) attempted just such a study. They used
data on two million speakers of Spanish and over 300,000 Chinese speakers
from the 1990 US census. The census provided information on age,
education, and age of arrival in the United States. In addition, respondents
had also been asked to rate their own English ability. Hence, Hakuta et al.
(2003) had the same ingredients in their study as found in Johnson &
Newport (1989), only on a massively greater scale. Hakuta et al. found a
typical age-related decline in English ability (English being the second
language of respondents). But there was no evidence for a significant break
or cut-off point, when either 15 or 20 years of age were taken as possible
end-points for a critical period. In other words, the statistical model that best
described the data showed a gradual decline with age. Moreover, it was
found that the amount of formal education was important for explaining
reported levels of English proficiency. Once in America, the more time spent
in school and college, the better one’s English becomes. For this reason,
Hakuta et al. (2003) argue that, at least for adult L2 learning, there is no
critical period, only a general tailing off caused by an age-related decline in
general cognition. Hakuta et al.’s results have recently been replicated in an
experimental study by Huang (2014). She found that Mandarin-speaking
immigrants to the US show a steady decline in grammatical ability in
English, depending on their age of arrival in America. Children who were
five years old on arrival show near native-like proficiency, while those
arriving at the age of 30 years performed much less well. Like Hakuta et al.,
Huang’s study controlled for factors like amount of education and length of
residence. Huang (2015) has also reviewed numerous studies on age of
acquisition effects and finds no clear advantage for an early start in life when
learning a second language.
What are we to make of the clash between Johnson & Newport, on the one
hand, and Hakuta et al. and Huang, on the other? First, there are huge
differences in sample sizes. The data from Hakuta et al. (2003), with more
than two million participants, are bound to be more reliable than the 46 in
Johnson & Newport (1989) (see Gravetter & Walnau, 1992, on sample sizes).
But the measure of English proficiency adopted in Hakuta et al. (2003) seems
extremely weak. Respondents rate themselves (not a good start) and do so on
a very simplistic scale. They are asked to say if they speak English ‘not at
all’, ‘not well’, ‘well’, ‘very well’ or ‘speak English only’. Hence, there is no
adequate definition of the target behaviour (language) that may, or may not,
be subject to a critical period effect. In particular, the top of the scale – ‘speak
English only’ – can, ostensibly, encompass a multitude of linguistic virtues
and sins. I could speak anything between really bad English all the time
(possibly without even knowing it) right through to speaking perfect English
all the time, with native-like fluency. Overall, the evidence lined up on each
side of this issue suffers from its own weaknesses. As Bialystok & Miller
(1999: 127) point out, critical period studies ‘use different methodologies,
engage different subject groups, administer different types of tasks, and
assess different linguistic features. The lack of consensus in outcomes is
hardly surprising’.
Abrupt cut-off point or gradual decline? As we have seen, research on adult
L2 learning is not yet decisive on this point, possibly because it has been
looking in the wrong places. Instead of a dramatic cut-off at 15 or 20 years,
as sought by Hakuta et al. (2003), there may instead be an off-set in ability at
about five years. The notion of an off-set is different from that of a cut-off
point. By off-set is meant the point where the system ceases to be maximally
receptive to the effects of the environment. That is, the period of peak
plasticity ends. Thereafter, there will be a decline (not a complete cessation)
in receptivity to language experience. Some time beyond this off-set, there
may be a definite cut-off point for the critical (or sensitive) period. Puberty
still seems like the best bet for such a final cut-off point. Subsequent to the
critical period cut-off, a more general decline in cognitive functioning begins
to reveal itself. Adults who exploit general learning mechanisms to acquire an
additional language often do remarkably well, despite the end of the critical
period. But they perform increasingly poorly over time, as general cognition
deteriorates.
Plastic fantastic: The receptive brain
Evidently, the evidence for a sudden cut-off is not yet clear. But more
important, the significance of any definite cut-off is also less than clear.
Advocates of the critical period hypothesis now suggest that the existence of
a critical (or sensitive) period is not dependent on the existence of this kind of
cut-off. Newport et al. (2001) take this tack and cite evidence from animal
critical periods – in particular the hearing systems of owls – in support of this
view. More important than a sudden cut-off is evidence that the child
undergoes a period of peak plasticity. This, at least, is a sine qua non
(‘without which nothing’) of critical periods. Of course, plastic in this context
does not mean ‘Tupperware’ or ‘polythene’. It means something more like
‘mouldable’, ‘flexible’ or ‘adaptable’. A plastic learning system (the brain) is
one that can be moulded in different ways according to the particular inputs it
is exposed to (or deprived of). Hence the familiar idea that the child is
especially receptive to language during the projected critical period. We can
look for signs of brain plasticity in patients who undergo radical brain
surgery, especially in those rare cases where an entire brain hemisphere must
be removed (hemispherectomy). We noted above that, in most people,
specialization for many language functions resides in the left hemisphere. If
this hemisphere is removed in an adult, the effects on language are
catastrophic and permanent. But Lenneberg (1967) suggested that language
learning is possible in cases where the hemispherectomy takes place before
the age of two. This follows from an assumption that the brain, as a whole, is
still sufficiently plastic for the right hemisphere to take over the language
functions normally engaged by the left. In essence, Lenneberg was right.
Recovery from severe left hemisphere damage is both possible and strongest
for the very youngest children (see Herschensohn, 2007). Plasticity is at a
peak in the first few years of life, and then wanes slowly as brain
development proceeds into adulthood. In fact, the human brain is not fully
mature until 15–20 years of age, ‘a long developmental period during which
environmental input could shape brain systems’ (Neville & Bruer, 2001: 152;
cf., Pallier, Dehaene, Poline, Le Bihan, Argenti, Dupoux & Mehler, 2003).
Overall, we see that a period of plasticity, with an off-set and subsequent
decline, is crucial to the definition of a critical period. And as we have seen,
the concept of an off-set is far less drastic than Lenneberg’s (1967) original
belief in a screeching, dead halt cut-off point.
Deafness and late language learning
Deaf children and adults comprise a further population of language learners
that have contributed to the critical period debate. We will have less to say
about the research in this area, because it parallels the research on both
isolated children and second language learners. The essential ingredients are
the same, with the focus on the effects of early versus late language learning.
Deaf children provide a testing ground because they often start life in a
linguistic vacuum. This is because more than 90 per cent of deaf children are
born to hearing parents (Meadow, 1980). Access to language is therefore
severely limited. In the past, it often took many months, even years, before
children were diagnosed. Happily, serious hearing loss can now be identified
in the first week of life (Google: newborn hearing screening programme for
information on the NHS programme). Not surprisingly, early diagnosis goes
hand in hand with enhanced language development (Yoshinaga-Itano, 2003).
But even with the best will in the world, hearing parents of deaf children have
a steep hill to climb in providing adequate language input. These parents can
only start learning sign language once their child has been diagnosed with
hearing loss. Effectively, these parents are learning a second language.
Imagine if I had been forced to teach my son, Alex, Japanese instead of
English when he was born. As it happens, we did cart Alex off to Japan when
he was 10 weeks old. But in the following year, the only Japanese he heard
were screams of ‘kawai!’ (‘cute’) from the locals. It is not surprising,
therefore, that Alex did not learn much Japanese. Nor is it surprising that
hearing parents of deaf children face a monumental task if they want to gain
fluency in what is, in effect, a foreign language for them.
Two more cases of linguistic deprivation: Chelsea
and E.M.
It can often be several years before deaf children gain access to native
speakers of sign language and begin properly acquiring their first full-fledged
language. For some children, access to sign language only begins when they
start school at about the age of five. For others, they may have to wait until
the age of 11 before going to a specialist secondary school. There are also
individuals who reach maturity without having acquired language at all. A
case in point is Chelsea, a deaf woman who did not start learning language
until her early thirties (Curtiss, 1988). The results were similar to Genie, in
the sense that Chelsea managed to acquire items of vocabulary, but stringing
them together into grammatical sentences was beyond her. Chelsea produced
utterances like Orange Tim car in, The woman is bus the going, and Banana
the eat. Despite their ungrammaticality, note that these verbal concoctions are
not random. For example, the article the is correctly associated with a noun
(woman, bus, banana), though, of course, the order is not consistent, even
within the same utterance (the woman versus bus the).
A further example of linguistic, but not social, deprivation is provided by
Ramirez, Lieberman & Mayberry (2013). They studied three deaf children
who were not exposed to language of any kind until they were about 14 years
old. Between one and two years later, after an immersive experience with
American Sign Language, it was found that they were acquiring vocabulary
in a manner which paralleled that of normally developing toddlers. In
particular, they acquired a high proportion of concrete vocabulary items
(words for things which can be seen and/or touched), with significantly fewer
class items like prepositions (for example, in English by, from, up, on). It
will be interesting to chart how their language continues to develop. A hint is
provided by the case study of E.M., a deaf boy studied by Grimshaw,
Adelstein, Bryden & MacKinnon (1998). E.M. was raised in a rural area,
where he received little formal education before the age of 12, and where he
was cut off from the deaf community. At the age of 15, he was fitted with
hearing aids that enabled him to hear speech at conversational level, and at
this point, he began to learn Spanish. After four years, Grimshaw et al. (1998)
tested E.M.’s comprehension on a range of 12 grammatical features. For
some of the structures, like the use of pronouns (for example, he, she, it),
E.M. performed at chance levels. For other features, E.M.’s performance
differed from chance, but was nevertheless not perfect. For example, his
understanding of the singular–plural distinction (cat versus cats) was good,
but not perfect, at roughly 85 per cent. What is fascinating about E.M. and
many, if not all, of the late language learners reported in the literature, is that
they have clearly learned something about grammar. Scoring 85 per cent on a
grammar test would attract accolades at school, but in the critical period
literature, it is taken as evidence that specialized language learning
mechanisms have been switched off. E.M.’s problem is not a lack of
grammar, but a lack of consistency in applying his knowledge of grammar.
His control over the rules does not seem to be as well developed as that of a
native speaker, but he does at least have appropriate grammatical features in
his repertoire. Of course, this is exactly what we saw with the late second
language learners described above. Evidently, linguistic deprivation seems to
have had a permanent effect on E.M.’s ability to acquire and/or use the rules
of grammar. And it is worth bearing in mind that E.M. did not suffer the
physical and psychological abuse that Genie was subject to. His social and
cognitive development appear to have been normal, so there must be
something about grammar learning per se that is affected by late exposure.
Early versus late learning of American Sign
Language
Single case studies on deaf children have also been supplemented by more
wide-ranging studies that have compared the effects of different starting ages
on language learning. Elissa Newport conducted a series of studies that
complement her work on second language learning reported above (Newport,
1988, 1990). Three groups of American Sign Language (ASL) learners were
identified, so-called native, early, and late. The native learners were exposed
to ASL from birth by deaf parents. Early learners began learning ASL
between the ages of four and six years, while late learners did not start until
the age of 12. On average, the participants in all three groups had been using
ASL for 30 years at the time of testing, so the amount of experience with
ASL was controlled for. As with second-language learners, Newport found
that late language learners in particular did less well, especially on tests of
morphology. Even early learners showed some deficits compared to the
native signers. Newport’s findings have been replicated in a number of
studies (for example, Mayberry & Eichen, 1991; Emmorey, Bellugi,
Friederici & Horn, 1995). It has also been found that late learners process
vocabulary in real time less efficiently than native sign language users
(Lieberman, Borovsky, Hatrak & Mayberry, 2015).
In addition, it has been shown that native and late deaf signers differ in terms
of brain organization. There are several sophisticated techniques for detecting
which parts of the brain are especially active during particular activities (see
Obler & Gjerlow, 1999, on matters grey). A number of studies support the
idea that language is organized differently in the brains of native versus late
ASL learners (for example, Wuillemin, Richardson & Lynch, 1994; Neville,
Coffey, Lawson, Fischer, Emmorey & Bellugi, 1997; Newman, Bavelier,
Corina, Jezzard & Neville, 2002; though see Perani et al., 1998, for an
opposing view from the second language literature). The feature of language
that distinguishes late from early second language learners appears to be
grammar. This finding has emerged by examining brain activity in Broca’s
area, the region of the left hemisphere that is especially associated with
grammar processing (among other functions). In late L2 learners, activity in
Broca’s area is especially high for L2 compared with L1. Significantly, this is
the case regardless of how proficient the learner is in their second language
(Wartenburger, Heekeren, Abutalebi, Cappa, Villringer & Perani, 2003). In
contrast, people who acquire their second language early in life (early
bilinguals) show similar levels of activity in Broca’s area for both L1 and L2.
Vocabulary is a further domain where differences can be observed in cerebral
activity between late and early deaf learners (Ramirez, Leonard, Torres,
Hatrak, Halgren & Mayberry, 2014).
Research on brain activity adds to our conviction that, when it comes to the
acquisition of both syntax and phonology, age matters. These findings are
clear. Their interpretation is much less clear. The problem we face is that an
age effect is not necessarily the same thing as a critical period effect. As we
discovered, the thrust of Hakuta et al. (2003) is that we experience a long
slow decline in the power of our general learning mechanisms that
progressively hinders our ability to acquire language. Hence, age effects may
not be caused by a language-specific critical period. But age effects are at
least compatible with the critical period hypothesis, especially if we
relinquish Lenneberg’s expectation that there is a clearly demarcated
termination point (Newport et al., 2001). We have also shifted the goal posts
with regard to the start and end points, for both syntax and phonology.
Critical periods for language probably begin before birth. Their off-set, the
age at which peak sensitivity ends, seems to be around age five. And
thereafter, there is a gradual decline in sensitivity to language. Some authors
continue to believe in a complete termination at about puberty, while others
adhere to the idea of a continual decline into adulthood and old age.
To conclude, it is difficult to deny the intuitive appeal of the idea that
language learning in later life is qualitatively different from the effortless
blooming of the toddler’s linguistic capacity. The following conclusion
seems reasonable, therefore:
The development of neural architecture dedicated to native language
grammar, lexicon and processing is a clearly biological characteristic of
typical human growth during the first few years of life that presents the
onset and peak of sensitivity of a representative critical period.
(Herschensohn, 2007: 226)
However, this chapter has revealed that there is still a great deal we need to
confirm, before we can fully accept Lenneberg’s original proposal. The
perfect critical period experiment is beyond our grasp, while the evidence that
is at our disposal is either incomplete or inadequate, or both. What good
evidence we do have is at least consistent with the idea that critical periods
exist in first language acquisition for both phonology and syntax. Future
research on brain functioning and brain development, together with more
substantial behavioural data, will surely provide us with more decisive
evidence on critical periods for language.
In a Nutshell
Critical periods in development are biologically determined.
The effects of the environment on development are
especially strong during the critical period and have a much
weaker effect once the critical period has ended.
The critical period is like a ‘window of opportunity’ for
development. If development does not take place during this
period, it may not be possible at all thereafter.
Environmental deprivation has its strongest effects during
the critical period. If the organism is deprived of vital input
during this period, development may be permanently
hindered, even if the input is supplied subsequently.
In the domain of language, different populations have been
studied: feral, abused and neglected children, in addition to
deaf people and second language learners.
Research on language as a critical period phenomenon
 suffers from a lack of rigour:
it is not clear precisely which aspects of language are
subject to critical period effects
most research does not have sufficiently tight control
over the length and quality of environmental experience
many of the subject populations studied may have
cognitive deficits that affect language learning, quite
apart from any critical period effects (for example,
Genie).
It is controversial whether there is a clear cut-off point at
puberty in language learning ability. The alternative concept
of an off-set may be more valid. An off-set marks the end of
a period of peak plasticity, after which receptivity to the
environment declines (rather than terminates altogether).
Evidence for critical period effects in adult second language
learning is far from decisive. The influence and gradual
decline of general cognitive learning mechanisms greatly
complicates empirical enquiry.
Different aspects of language are probably subject to their
own critical period effects, in particular phonology, syntax
and language processing.
On balance, available evidence favours the existence of
critical periods in first language development.
Further Reading
Curtiss, S., Fromkin, V., Krashen, S., Rigler, D. & Rigler, M.
(1974). The linguistic development of Genie. Language, 50(3),
528–554.
A golden oldie, but still worth reading for the detailed examples
of Genie’s actual speech. You may be surprised at the positive
spin put on Genie’s grammatical prowess, in light of the
discussion in this chapter. A chance to make up your own mind
by looking at some of the original data.
Rymer, R. (1993). Genie: A scientific tragedy. Harmondsworth:
Penguin.
A highly accessible account of Genie: how she came to light and
how her story unfolded subsequently. Light on the science of the
critical period, but rich in insights about a remarkable person.
Werker, J.F. & Hensch, T.K. (2015). Critical periods in speech
perception: New directions. Annual Review of Psychology, 66,
173–196.
A thorough review of research on speech perception in particular,
with an emphasis on the biological underpinnings of critical
periods.
Blom, E. & Paradis, J. (2016). Introduction: Special issue on age
effects in child language acquisition. Journal of Child Language,
43(3), 473–478.
These authors introduce a wide range of up-to-date articles on age
effects in language acquisition, including the impact of cochlear
implants on hearing-impaired children and issues raised in
bilingual settings.
Websites and Video
Genie
Search on YouTube for Genie (Secret of the Wild Child)
for an interesting documentary about Genie.
Testing on animals in scientific research: www.peta.org
and http://www.ca-biomed.org/csbr/pdf/fs-whynecessary.pdf
These two sites provide strongly conflicting views on a
deeply controversial topic.
Still want more? For links to online resources relevant to this
chapter and a quiz to test your understanding, visit the companion
website at https://study.sagepub.com/saxton2e
4 Input and Interaction: Tutorials for
Toddlers
Contents
Talking to young children 86
Characteristics of Child Directed Speech 88
Phonology 88
Vocabulary 89
Morphology and syntax 89
A dynamic register 91
Individual differences and their effects 92
Child Directed Speech: Summary 94
Lack of interaction: Can children learn language from television? 95
Imitation 97
Linguistic creativity: Children make their own sentences 97
Skinner and Chomsky on imitation 98
Imitation as a mechanism in cognitive development 98
Imitation: Who, when and how? 99
Individual differences in imitation 101
Corrective input 102
Recasts: Adult repetition of the child 102
The ‘no negative evidence’ assumption 103
Contrastive discourse 105
Negative feedback 110
Corrective input: Summary 111
Universality of CDS 112
Input and interaction in language acquisition 114
Overview
By the end of this chapter you should be able to describe Child
Directed Speech (CDS). This is the special register adopted by adults
and older children when talking to young children. You will gain some
appreciation of specific modifications which are made at the levels of
phonology, vocabulary, morphology and syntax. It will become
apparent that, in several instances, CDS functions to facilitate language
development. You will also gain an understanding of individual
differences in CDS. In particular, you will be able to describe how
 socioeconomic status affects the amount and quality of language
children hear, and also, how these differences impact on language
development.
This chapter will also introduce you to the difference between input
(the language forms children hear) and interaction (the way language is
used in conversation). With regard to interaction, you should be able to
assess the importance of imitation (both verbal and non-verbal) for
language acquisition. You should also be able to describe a special
form of verbal imitation, the adult recast, and describe its potential as a
form of corrective input for child grammatical errors. You will also
gain some understanding of cross-cultural research on the child’s
linguistic environment and be able to evaluate the claim that at least
some aspects of CDS may be an inevitable feature of adult–child
conversation.
Talking to young children
To many people, there is nothing special or mysterious about child language
development. If you grab a representative of the people – the man on the
Clapham Omnibus – and ask him how children learn language, it’s an even
bet he’ll tell you that they learn it from their parents. And who will contradict
him? A child born to Tagalog-speaking parents in the Philippines will grow
up speaking Tagalog. The child born to Inuktitut-speaking parents in the
Arctic region of Canada will learn Inuktitut. My own parents are monolingual
English speakers, and here I am, thinking, speaking, reading, and
occasionally writing, in English. In one sense, therefore, no-one can argue
with the man on the bus. Children learn language from their parents.
Unfortunately, though, this formulation does not get us very far.
The real problem is to understand how the child learns. We need to find out
what learning mechanisms and conceptual knowledge the child applies to the
task of language acquisition. In other words, we want to lift the lid on the
machinery of language acquisition, all of which resides in the child. The
language children hear from parents and others is merely the fuel that allows
the machine to run and produce a mature knowledge of language as its end
product. Linguistic input is certainly an essential ingredient, which is why we
will examine it in this chapter. But we should not forget that, ultimately, we
want to establish how the child’s language learning capacities engage with
and learn from the available linguistic environment. If we torture the fuel
metaphor just a little further, one question, of considerable interest, crops up
straightaway. Does our language learning machine require low-grade fuel or
a high octane deluxe grade? Some machines run very well on low-grade fuel.
When it comes to language acquisition, we find that nativists, in particular,
adopt this ‘low grade’ view. In a well-known statement, Chomsky (1965: 31)
described the input available to the language learning child as ‘fairly
degenerate in quality’, characterized by ‘fragments and deviant expressions
of a variety of sorts’ (ibid.: 201). However, as soon as researchers began to
examine the way parents talk to their children, it became apparent that they
adapt their speech in numerous ways at every level of linguistic analysis
(Snow, 1972; Phillips, 1973). In other words, parents seem to provide high
octane input as the fuel for language acquisition.
The way in which parents speak to their children constitutes a special register
or style, which has been given many names over the years (see Box 4.1). We
will favour Child Directed Speech (CDS), with a brief diversion into Infant
Direct Speech (see Saxton, 2008, for a justification). A critical distinction is
between input and interaction. Input constitutes the particular language forms
that the child hears, while interaction refers to the way in which those forms
are used in adult–child discourse. Imitation, broadly construed, is a critical
form of interaction for language development which includes both verbal and
non-verbal behaviour. We will discover that CDS facilitates language
acquisition. But many researchers downplay the importance of ‘high octane’
CDS, on the grounds that it is not universally available to all children. We
examine the evidence for this assumption in what follows.
Box 4.1 Terms of Engagement
The register used by parents and others to talk to young children
has been given many names since the late nineteenth century.
Here are 15 of the most prominent:
1. baby talk (Lukens, 1894)
2. nursery talk (Jakobson, 1941/1968)
3. motherese (Newport, 1975)
4. caregiver speech (Ochs, 1982)
5. caretaker talk (Schachter, Fosha, Stemp, Brotman & Ganger,
1976)
6. verbal stimuli (Skinner, 1957)
7. exposure language (Gillette, Gleitman, Gleitman & Lederer,
1999)
 8. input language (Ninio, 1986)
9. linguistic input (Schlesinger, 1977)
10. primary linguistic data (Chomsky, 1965)
11. Infant Directed Speech (Cooper & Aslin, 1990)
12. Child Directed Speech (Warren-Leubecker & Bohannon,
1984)
13. caregiver talk (Cole & St. Clair Stokes, 1984)
14. verbal environment (Chomsky, 1980a)
15. parentese (Ramírez-Esparza, García-Sierra & Kuhl, 2014)
There are prizes for those who can dig up any more. Not big
prizes, mind you – I’m not a banker.
One might ask if it matters which term one chooses. In fact, it
does matter. The people doing the talking (e.g., parent or elder
sibling) and the person being addressed (e.g., infant or toddler)
can have a significant influence on both the quantity and quality
of language used. Probably the most useful term is Child Directed
Speech, hence its use in this chapter. For a guided tour through
the jargon jungle, see Saxton (2008).
Characteristics of Child Directed Speech
Child Directed Speech is a special register, which means that it constitutes a
distinct mode of speech. The term register is borrowed from sociolinguistics,
and overlaps with the concepts of style and dialect (Saxton, 2008). Although
it is not well defined, the central idea is nevertheless fairly straightforward.
We talk in different ways to different people in different settings. Consider
how you might converse with the Queen at a palace garden party (converse,
mind you – not talk). Now compare that with the kind of chat you might have
with a close friend. You would, in all likelihood, adopt a separate register for
each occasion. We can establish that CDS is a distinct register by comparing
it with Adult Directed Speech (ADS) (Hills, 2013). This comparison can
provide a scientific control. The ideal study would take a group of adults and
record each person in conversation with another adult and also, on a separate
occasion, in conversation with a young child. This method allows one to
identify what is special or distinctive about Child Directed Speech.
Fortunately, this approach was taken by Snow (1972), who provided the first
major study in this field, and therefore the first challenge to Chomsky’s belief
that the input is ‘degenerate’.
Phonology
If you have ever spoken to a baby, you may well have found yourself riding a
linguistic rollercoaster. People tend to exaggerate their intonation, producing
great swooping curves of sound over an extended pitch range. At the same
time, the overall pitch tends to be higher than normal (Garnica, 1977; Stern,
Spieker, Barnett & MacKain, 1983; Fernald, Taeschner, Dunn, Papousek,
Deboysson-Bardies & Fukui, 1989; Werker & McLeod, 1989). In particular,
mothers are especially prone to raise their pitch when the infant shows signs
of positive emotional engagement (Smith & Trainor, 2008). Speech also
tends to be slower, with syllable-lengthening, longer pauses and fewer
dysfluencies (Broen, 1972; Fernald & Simon, 1984; Fernald, 1989; Albin &
Echols, 1996). In addition, there is evidence that individual phonemes are
pronounced with greater clarity than is the case in Adult Directed Speech. For
example, in ADS, the phonemes /t, d, n/ are often not pronounced fully when
they are followed by certain phonemes (/b, p, m, g, k/). Instead, they
assimilate to these following phonemes. This means that the sounds take on
each other’s acoustic properties in certain respects. Infants, by contrast, are
more likely to hear canonical, fully pronounced versions of the /t, d, n/
phonemes (Dilley, Millett, Devin McAuley & Bergeson, 2014). Parents may
well gravitate towards this manner of speech because infants prefer to listen
to it. In fact, some infants have a stronger preference for listening to ADS
than others. And those infants who pay particular attention at 12 months have
acquired relatively large vocabularies by the age of 18 months (Vouloumanos
& Curtin, 2014). From the adult’s point of view, ADS provides the best
method to grab the infant’s attention and, in the process, convey a mood of
positive affect (Fernald, 1989). On occasion, this exaggerated intonation is
lampooned as a minority pursuit, indulged in only by the privileged Western
mothers mentioned above. But this kind of speech has also been observed in
languages as diverse as German, Hebrew, Italian, Japanese, Luo (an East
African language), and Spanish (see Soderstrom, 2007, for a summary). The
phonological adaptations sketched out here figure most prominently during
the child’s first year. For this reason, some authors refer to Infant Directed
Speech (IDS) as a special register, distinct from the Child Directed Speech
addressed to older children (Golinkoff, Can, Soderstrom & Hirsh-Pasek,
2015).
Vocabulary
Adult–child conversation with a toddler tends to be about the here-and-now,
rather than topics distant in time or space. This means that the words chosen
by the adult are likely to be more easily comprehended by the child. In this
vein, there is a marked emphasis on concrete, rather than abstract, concepts.
Cup, juice and tree are far more likely in speech addressed to two-year-olds
than purity, truth or beauty. Moreover, adults tend to place object words at
the end of sentences and pronounce them relatively loudly, thus giving them
special prominence (Messer, 1981). Five topics in particular seem to
dominate the conversation: members of the family; animals; parts of the
body; food; and clothing (Ferguson, 1977; Ferguson & Debose, 1977). These
topics are dictated by the interests of the child, which indicates that, if you
want to maintain a toddler’s attention, you have to talk about what interests
them. It is no use launching into a discussion about falling share prices, fiscal
prudence and quantitative easing. You will be met with a blank stare.
Actually, if you tried that with me, you’d be met with a blank stare. As with
phonological adaptations, parental lexical choices respond to the needs and
interests of the child.
Morphology and syntax
With regard to morphology, the frequency with which children are exposed
to particular morphemes is reflected in their own speech (Warlaumont &
Jarmulowicz, 2012). Some languages, like Russian, have a dauntingly
complex system of word endings. But Russian mothers tend to subvert these
complexities and, instead, rely heavily on diminutive forms. Equivalents in
English might be diminutive forms like horsie, doggie and duckie. Russian
mothers alternate between diminutive and morphologically simplified forms
within the same stretch of conversation (Kempe, Brooks, Mironova,
Pershukova & Fedorova, 2007). In so doing, these mothers may provide an
accessible demonstration of the morphological variety typical of Russian,
without overwhelming the child with the massive variety witnessed in ADS.
Turning to syntax, sentences in CDS are remarkable for being well-formed
grammatically. Although children hear a considerable number of incomplete
sentences, these nevertheless comprise legitimate syntactic phrases, for
example, noun phrases like: more grape juice; a cup; or another blue one
(Brown, 1973). One survey by Newport et al. (1977) found just one genuine
adult error in a corpus of 1,500 utterances. CDS sentences are not only
grammatical, but also relatively short (Barnes, Gutfreund, Satterly & Wells,
1983). In jargonese, we could say that the MLU of CDS is lower than in
ADS. Mean Length of Utterance (MLU) provides a rough and ready
measure of syntactic complexity, whereby shorter sentences tend to be
simpler. The impression of simplicity is confirmed by the relative scarcity of
certain complex structures. Thus, Sachs, Brown & Salerno (1976) report few
subordinate clauses, relative clauses, sentential complements and
negations in CDS.
The subject of CDS sentences also has a strong tendency to be an agent. We
encountered the category of subject in Chapter 3. Agent, or ‘doer’ of an
action is one semantic role that the subject can take on. In fact, we might
argue that ‘agent-as-subject’ is prototypical: Venus in Venus knocked on our
door, or Paula in Paula taught me British Sign Language. Critically, subjects
are not always agents: The oysters were washed down with champagne or
The meeting came to order at 10 o’clock. Neither the oysters nor the meeting
are acting as the agent of any action, yet they each function as the subject.
Coming back to CDS, consider the advantage to the child if the majority of
subjects they hear are agents. Eve’s mother produced the following
examples, when Eve was 1;7:
Agent as subject of the sentence:
you’re dancing
Eve can get the banjo
you broke it
shall we change your diaper?
you make a train
Initially, the child’s experience is confined mostly to just one kind of subject
and, moreover, it is the prototypical (agent) subject. By confining sentence
subjects to agents, the adult makes the relationship between meaning and
grammar especially clear for the young child (Rondal & Cession, 1990). The
one-to-one mapping between agents and subjects in CDS may well provide
the child with an entry point for discovering the grammatical category of
subject, a process known as semantic bootstrapping (Pinker, 1984). The
complexities involved in a range of different subject types are thus introduced
later. In fact, it takes several years before children make the connections
between different kinds of sentential subject and treat them as a single
grammatical category. Four-year-olds have trouble in this regard, and it is not
until the age of nine that children demonstrate full control over the subject
category (Braine, Brooks, Cowan, Samuels & Tamis-LeMonda, 1993). It is
fascinating to discover, therefore, that the child’s introduction to the subject
category, via CDS, is simplified.
A dynamic register
The register used with children is dynamic. Rather than being a fixed mode
of address, adult speech changes continually in concert with the child’s
developing language. Perhaps the clearest sign of these changes is the shift
from infant-directed to child-directed speech. The different names – IDS and
CDS – indicate that the changes in adult speech are substantial. But there is
no step change from one mode to the other. On a day-to-day basis, the
changes are subtle and difficult to detect, but they continue all the way
through childhood until a point is reached where the speech addressed to
one’s offspring can genuinely count as an instance of Adult Directed Speech.
Some of the changes over time have been monitored (e.g., Buchan & Jones,
2014), though it must be said that there is far more to learn about the dynamic
nature of CDS. One change we do know about concerns tone of voice. From
birth to three months, infants respond well to a comforting tone of voice, but
subsequently, from three to six months, they come to prefer a more approving
tone of voice. Finally, there are signs that, by nine months, infants respond
more readily to a directive tone of voice (Kitamura & Lam, 2009). Adults are
able to satisfy these changes in infant preferences because they are sensitive
to the infant’s emotional needs on a moment-to-moment basis (Smith &
Trainor, 2008). Snow (1995) refers to these changes over time as a process of
‘finetuning’ to reflect the continual sensitivity of caregivers to the child’s
communicative needs. In fact, the notion of finetuning may overstate the
case, with some authors arguing that the ‘input is only grossly, not finely,
tuned’ (Hoff, 2004: 924). But the fact that the adult input is tailored in any
way at all to the infant’s language level deserves our attention.
Exercise 4.1: Adult Directed Speech
Make a short audio recording of a conversation between yourself and a
friend. Tell your friend that you are doing this, to guarantee both
eternal harmony and a proper regard for the ethics of research (see
www.beta.bps.org.uk for more information on the ethics of psychology
research). Nowadays, you won’t need a big reel-to-reel tape recorder;
many mobile phones have an audio recording capacity that should suit
our rough-and-ready purposes. Try not to be self-conscious and talk
about something that you really want to talk about.
Once you have made your recording, transcribe part of your
 conversation word (and even part-word) for word. Listen with great
attention to precisely what was said. We are interested in your speech
performance, so do not edit out the unintended glitches. You will
probably need to listen to each utterance several times before you can
get an accurate record. Now consider the following:
If you were a toddler, what aspects of your conversation would be
difficult to follow? Consider phonology, vocabulary and syntax.
Consider also any errors, hesitations, repetitions and interruptions
to the flow.
How adequate is your language as a model for learning by a
toddler?
Would it be possible to modify the content of your conversation in
any way to improve its ‘language teaching’ potential? If so, what
would you change?
Individual differences and their effects
So far, we have sketched a picture of Child Directed Speech that marks it out
quite clearly as a special register, quite different in its characteristics from
Adult Directed Speech. In terms of average tendencies, this picture is
accurate. But we should also be aware that individual children have different
experiences of language. There is considerable variation in both the amount
and quality of CDS provided. In this regard, two points are often raised: (1)
CDS may be a minority phenomenon, not universally available; and (2)
parents who do supply CDS nevertheless differ, in terms of both quantity and
quality, in how they talk to children. The assumption that CDS is not
available to all children is widely retailed in the child language literature. But
as we shall see, below, this assumption is very poorly supported, empirically.
Far more convincing is the evidence on the second point – individual
differences among parents – and this provides the focus in this section.
Hart & Risley (1995) provide the most substantial demonstration that
children differ in the sheer amount of language they hear. They grouped
parents into three bands according to socioeconomic status (SES), a variable
that generally takes into account level of education, income and job prestige.
High-SES parents were professionals, Mid-SES parents were working class,
while Low-SES parents were generally on public assistance, a form of social
welfare provided in the US.
As Table 4.1 reveals, there are substantial differences in the amount of
language different children hear. In fact, both the quantity and quality of the
input varied among groups. High-SES children are exposed not only to more
words, but to a greater variety of words also. This group also receive fewer
prohibitions and directives: utterances designed to control child behaviour in
some way. High levels of parental prohibition are associated with relatively
poor language growth on a range of measures, including MLU, diversity and
complexity of vocabulary, and use of different language functions (Taylor,
Donovan, Miles & Leavitt, 2009). Furthermore, High-SES parents use
longer, syntactically more complex utterances with their children. Generally
speaking, more talkative parents tends to use a richer vocabulary and express
more complex ideas, regardless of SES (Rowe, 2012). And this input has a
beneficial impact on the child’s own acquisition of complex grammar
(Vasilyeva, Waterfall & Huttenlocher, 2008). High-SES children produce
complex structures themselves at an earlier stage than Low-SES children (22
months) and this advantage persists over the next 18 months or more (see
also Hoff, 2006, who reviews the impact of a wide range of social factors on
language development). Finally, High-SES parents in China, as measured by
parental education level, have children with more advanced vocabulary
development (Zhang, Jin, Shen, Zhang & Hoff, 2008).

There are signs that a simple equation – High-SES = ‘high talking’ – may not
hold, not even within the United States. Low-SES Spanish-speaking mothers
in San Francisco produce 17.5 utterances per minute, on average, when
talking with their 18-month-old children (Hurtado, Marchman & Fernald,
2008). This figure compares with an average of 14.6 per minute for another
US sample, this time comprising both Mid- and High-SES speakers of
English (Hoff & Naigles, 2002). Low-SES mothers are more talkative than
High-SES mothers on this comparison. Apart from SES, these groups also
differ in terms of the native language spoken and membership of different
sub-cultures. These latter factors probably exert separate influences on the
amount of parental speech. It is also worth emphasizing that there can be
considerable variation in talkativeness within SES bands (Hurtado et al.,
2008). The children of talkative mothers heard seven times as many words,
and three times as many different words, as children of less talkative mothers.
In fact, individual parental loquaciousness predicts the child’s language
development better than the SES group they belong to (Weisleder, Otero,
Marchman & Fernald, 2015). Regardless of SES, lexical frequency and
lexical diversity co-occur (Song, Spier & Tamis-LeMonda, 2014). And both
of these factors have an impact on the child’s vocabulary development, with
children of talkative mothers developing larger vocabularies (Hart & Risley,
1995; Hoff & Naigles, 2002; Hurtado et al., 2008; Fernald & Weisleder,
2015). In a similar vein, children whose parents engage in more episodes of
joint attention and shared book reading develop larger vocabularies by the
age of six years (Farrant & Zubrick, 2013).
Hurtado et al. (2008) report a further consequence of being talkative with
one’s child. They found that the children of talkative mothers were more
efficient at processing speech. Pictures of two familiar objects were presented
side by side, and one of them was named. Children of talkative mothers
looked more quickly to the named target. Lexical knowledge and processing
efficiency work together in a synergistic fashion. As vocabulary grows, so
too do the processing skills needed to make fine discriminations among
words on the basis of their phonological, morphological and semantic
characteristics. It makes sense, therefore, that children with large
vocabularies are better able to learn words on a single exposure (Gershkoff-
Stowe & Hahn, 2007; see also Chapter 6 for more on so-called fast mapping).
Overall, we have observed a tendency for High-SES parents to be relatively
talkative. It turns out that they also use more gestures, with a concomitant
beneficial impact on later child vocabulary learning (Schmidt, 1996; Rowe &
Goldin-Meadow, 2009a, 2009b; Hall, Rumney, Holler & Kidd, 2013;
Stanfield, Williamson & Özçalişkan, 2014). But why should wealthy, well-
educated people chatter and gesticulate so much? The answer is not yet clear,
though Rowe (2008) does report that attitudes towards child-rearing practices
differ according to SES status. High-SES parents hold beliefs that reflect the
information on offer from textbooks, experts and paediatricians. Of course,
now that you’ve read this chapter you too can feel perfectly justified, when
you next meet a toddler, in jabbering on at ten to the dozen while waving
your arms about like a windmill.
Child Directed Speech: Summary
If one was asked to design a language course for infants, one might very well
come up with something resembling Child Directed Speech. The numerous
modifications on display are, without fail, geared towards simplifying and
clarifying the object of learning. As a language course, CDS benefits from
confining the syllabus to topics that interest the learner. We might also point
to the dynamic nature of CDS, because, like any well-designed course, it
presents the child with a graded series of lessons. Or, more subtly, the child
elicits from parents the input that meets their language learning needs. This
view is expressed by Bohannon & Warren-Leubecker (1985: 194), who
observe that, ‘since the complexity of speech addressed to children is largely
determined by cues from the children themselves …, one might think of
language acquisition in this view as a self-paced lesson’. One thing is certain.
The input to the child is not, as assumed by Chomsky (1965), degenerate. At
the same time, there is variation in the amount and quality of CDS available,
with concomitant effects on the rate of language development. Evidently,
Child Directed Speech has a facilitative influence on child language
development. But facilitative is not the same as necessary. Many researchers
assume that CDS cannot be necessary for language development, because
they believe that CDS is not supplied to all children everywhere. We shall
consider the validity of this assumption. First, though, we turn our attention
to the neglected topic of interaction. CDS does not simply comprise a special
set of language forms. Language is presented to the child via particular kinds
of interaction, including imitation. We begin, though, by considering what
happens when the child has linguistic input, but no interaction, at their
disposal.
Discussion point 4.1: The input in language learning
Consider your own experience of learning a second language.
What age did you start learning?
What kind of input and/or teaching did you receive?
What are the similarities in the input available to both second
language learners and toddlers acquiring their first language?
What are the differences?
Hop back to Chapter 3 to help with the following two questions:
To what extent can differences in the input explain
differences in the outcome of learning, when comparing
native first language learning with adult second language
learning?
Did the age at which you started learning have an effect on
 the outcome?
Lack of interaction: Can children learn language
from television?
We can demonstrate the difference between input and interaction by
switching on the TV. All of a sudden, the child is exposed to linguistic input,
but without the benefit of a conversational partner to interact with. We can
therefore investigate language acquisition in the absence of interaction. On
the whole, the results are disappointing. It has been found that before the age
of two years, children are not capable of learning new words from television
(Kuhl, Tsao & Liu, 2003; Mumme & Fernald, 2003; Anderson & Pempek,
2005; Krcmar, Grela & Lin, 2007). And if we compare the power of TV with
live interaction, then live interaction is clearly superior. In this vein, Patterson
(2002) showed that children aged 21–27 months learned new words from a
shared book reading activity, but none at all from TV viewing. Moreover,
television has no impact on second language development in young children
(Snow, Arlman-Rupp, Hassing, Jobse, Joosten & Vorster, 1976). The
limitations of television are especially clear in the case of Jim, a hearing child
born to deaf parents (Sachs, Bard & Johnson, 1981). Jim’s only exposure to
English was via television, which he spent a lot of time watching. By the age
of 3;9, Jim’s language was not merely delayed, it was distinctly odd:
This is how plane
I want that make
House chimney my house my chimney
Rather like the case of Genie discussed in Chapter 3, Jim was able to learn
some words, but very little about how to put words together to form
grammatical sentences.
Beyond the age of two years, it is possible to learn some vocabulary from TV
viewing (Rice & Woodsmall, 1988; Barr & Wyss, 2008). But it depends what
you watch. Programmes that are not designed expressly for toddlers have no
discernible effect (Rice, Huston, Truglio & Wright, 1990). Even some
programmes that are designed for young children do not have much impact.
For example, regular viewing of Teletubbies from the age of six months is
associated with relatively low vocabulary scores at three years of age
(Linebarger & Walker, 2005). If you’re a Teletubbies fan, then you will not
be surprised by this. You will probably also have a small vocabulary.
Teletubbies does not contain very much language and the characters are
prone to talk in a rather bizarre, fake kind of ‘baby talk’. In contrast, other
programmes, including Dora the Explorer, Blue’s Clues and Dragon Tales,
did have a beneficial effect on vocabulary and expressive language scores
(Linebarger & Walker, 2005; cf., Uchikoshi, 2005). In one study, children
aged 2;6 learned new words from video only when there was some form of
reciprocal social interaction with an adult, either on or off screen (O’Doherty,
Troseth, Shimpi, Goldberg, Akhtar & Saylor, 2011). Given an optimal
programme style, it is perhaps not surprising that three-year-olds can learn
some vocabulary from television. By this age, language acquisition has long
since taken off in an exponential fashion, with rapid growth in both
vocabulary and syntax (Bates, Dale & Thal, 1995). The child is therefore
increasingly well equipped, both cognitively and linguistically, to infer word
meanings for themselves, even in the absence of interactive support.
Television is not the only source of non-interactive input that might influence
the child. Overheard conversations, radio and song lyrics all expose the child
to linguistic information. There is some indication that children as young as
18 months can learn new words that they have overheard being used by
adults (Akhtar, 2005; Floor & Akhtar, 2006; Gampe, Liebal & Tomasello,
2012). But as with television, the language acquired is confined to a limited
range of vocabulary items, at an age when vocabulary learning has already
taken off. Somewhat older children (mean age 27 months) can learn
rudimentary aspects of word meaning in the absence of social and
observational cues (Arunachalam, 2013). However, the amount of speech
overheard by the child does not predict vocabulary size, whereas the amount
of speech targeted specifically at them does (Shneidman, Arroyo, Levine &
Goldin-Meadow, 2013; Weisleder & Fernald, 2013). Of course, many
children grow up in a world where the TV is switched on for hours at a time
as a backdrop to daily life, thus exposing them to a large amount of overheard
language. However, constant background TV interferes with adult–child
interaction, with a negative effect on subsequent vocabulary growth (Masur,
Flynn & Olson, 2016).
Chomsky (1959: 42) suggested that ‘a child may pick up a large part of his
vocabulary and “feel” for sentence structure from television’. But most of
Chomsky’s early assertions about language acquisition are the result of
armchair speculation, not empirical enquiry. And most of them, like this one,
are wrong. In the nativist approach, the role of the linguistic environment is
reduced to a matter of simple and limited exposure to key linguistic forms.
The assumption is that limited exposure of this kind will suffice to trigger
language acquisition (Lightfoot, 1989, see also Chapter 8). In principle,
therefore, the nativist might be content to leave the child, unaccompanied, in
front of the TV screen. But as any guilty parent knows, if the child must
watch TV, it is better that someone watch with them (Naigles & Mayeux,
2000; Masur & Flynn, 2008). In essence, parents should consider TV viewing
in the same way as shared book reading – as an opportunity to interact with
the child and provide the framework for extending the child’s current state of
linguistic knowledge (Vygotsky, 1934/1962; Cameron-Faulkner & Noble,
2013). Left alone, with no more than exposure to linguistic forms, the child
cannot acquire syntax. Interaction is essential.
Imitation
We started out this chapter on the Clapham Omnibus, where we heard the
common sense view that children learn language from their parents. Even if
we accept this view, we still need to know how children learn from parental
input. To the man on the bus, and many like him, the answer is, once again,
entirely obvious: children learn language by imitating their parents.
Surprisingly, though, researchers have largely ignored imitation as a serious
factor in child language acquisition. Relatively little research effort has been
devoted to the issue. Moreover, imitation rarely figures in theoretical debates
and does not even feature as an index topic in several recent books in the
field (e.g., Cattell, 2000; Karmiloff & Karmiloff-Smith, 2001; Clark, 2003;
Hoff & Shatz, 2007). It was not always thus. Imitation of language is
mentioned several times by Tiedemann (1787) (see Appendix 1). Tiedemann
observed a child of six months and noted that ‘his mother said to him the
syllable “Ma”; he gazed attentively at her mouth, and attempted to imitate the
syllable’ (see Murchison & Langer, 1927: 218). Incidentally, infants who are
good at observing the mother’s mouth (and following her gaze) acquire
vocabulary with particular alacrity (Tenenbaum, Sobel, Sheinkopf, Malle &
Morgan, 2015).
In this section, we shall demonstrate that imitation, as a special form of
interaction between parent and child, is fundamentally important in the study
of language acquisition. But before we get started, a note on terminology: I
will mostly use the term imitation in this section, but when you access the
child language literature directly, you will find that repetition is often
preferred (e.g., Gathercole, 2006; Clark & Bernicot, 2008). The points to be
made are not substantively affected in the alternation between these two
terms. Repetition can be seen as a particular kind of imitation, one that is
often associated with language.
Linguistic creativity: Children make their own
sentences
As mentioned, imitation is often downplayed in theories of language
acquisition because the child possesses linguistic creativity. As speakers of a
language, we create genuinely novel sentences all the time, putting words
together in sequences that have not been uttered before and which may never
occur again. We can do this because grammar allows for infinity in the
number of different sentences we can put together, even though the system of
grammatical rules is itself finite (von Humboldt, 1836/1988; Chomsky,
1965). Two consequences follow from this observation that have a direct
bearing on imitation: (1) it is logically impossible to imitate an infinite
number of sentences; and (2) we could not rely on imitation as the source for
producing our own sentences. Think how odd it would be to sit around
waiting for someone to say a sentence, so that you, as a learner, got an
opportunity to use it for yourself, through imitation. You might literally have
to wait forever for the right sentence to come along. Not much fun if the
sentence is: Can you tell me where the toilet is, please? Even if the child were
adept at imitating each and every sentence they heard, it would not provide
the information needed, in and of itself, to construct new sentences for
themselves. In other words, the child cannot imitate grammar (the rules), only
the output from those rules (sentences). We will expand on this point in
Chapter 8. For now, we can confirm that imitation of sentences does not
provide a feasible route into the acquisition of syntax.
Skinner and Chomsky on imitation
It is often wrongly assumed that Chomsky (1959) dismissed imitation as part
of his campaign against the behaviourist approach to language acquisition.
Chomsky (1959) provides a book review of Verbal Behavior (1957), written
by B.F. Skinner, the doyen of twentieth-century behaviourism. Ironically, this
book review has been far more widely read than the actual book, in which
Skinner argued that child efforts to speak are rewarded by parents. Skinner
based his position on a form of learning studied extensively by behaviourists,
termed operant conditioning. Each time the child produces an utterance that
comes close to sounding like an acceptable word or sentence, the parent
offers a ‘reward’ in the form of praise or encouragement. On successive
occasions, closer approximations to the adult model receive yet further
parental rewards. Thus, operant conditioning relies on the learner producing a
linguistic behaviour that is progressively shaped through rewards, until the
desired behaviour is achieved. Punishments can also be used to ‘dissuade’ the
learner, when behaviours veer away from the desired learning outcome.
Operant conditioning is very different from imitation. In fact, Skinner had
very little to say on the subject of imitation, other than to reject the idea that
humans have an instinct to imitate. Instead, he preferred the phrase echoic
behaviour, whereby a ‘response generates a sound-pattern similar to that of
the stimulus. For example, upon hearing the sound Beaver, the speaker says
Beaver’ (Skinner, 1957: 55). Skinner does not call this behaviour imitation,
on the grounds that there is ‘no similarity between a pattern of sounds and the
muscular responses which produce a similar pattern’ (ibid.: 59). This is an
astute point, although it does no violence to standard definitions of imitation.
For most people, Skinner’s ‘beaver’ example is a case of imitation.
The champion of imitation in the Skinner–Chomsky exchange was Chomsky
(1959: 42), who insisted that ‘children acquire a good deal of their verbal and
nonverbal behavior by casual observation and imitation of adults and other
children’. Elsewhere, Chomsky talks about children’s ‘strong tendency to
imitate’ (ibid.: 43) and castigates Skinner for rejecting the ‘innate faculty or
tendency to imitate’ (ibid.: ff51). Unfortunately, this clear position has been
completely reversed in the re-telling. Some recent authors now suggest that it
was Chomsky (not Skinner) who rejected imitation as a significant factor (for
example, Cattell, 2000; DeHart, Sroufe & Cooper, 2000; Stilwell-Peccei,
2006). In this vein, Owens (2008: 33) asserts that ‘an overall imitative
language acquisition strategy would be of little value, according to Chomsky
(1957), because adult speech provides a very poor model’. This not only
misrepresents Chomsky on imitation, it gets the date wrong (there is nothing
at all on imitation in Chomsky, 1957). Alarm bells should ring for students at
this point: read the original sources for yourself, if you possibly can. None of
this would matter, perhaps, if our initial critique of imitation, vis-à-vis
linguistic creativity, was the last word on the subject. Arguably, though, there
is more to say and far more to learn about the function of imitation in
language acquisition.
Imitation as a mechanism in cognitive development
The definition of imitation is less straightforward than it might seem. A
simple definition might be: the reproduction of another person’s behaviour.
In the context of language, this could mean that, if you first say something, I
would repeat it back verbatim. But no matter how good a mimic I am, I will
not be able to reproduce precisely your accent, your voice quality, the
emotions conveyed in your tone of voice, the speed and rhythms of
production, nor the fine-grained acoustic detail in the way each word and
phrase is articulated. If you nevertheless recognize that I am imitating you
successfully, then that recognition must derive from some abstraction of
properties common to your model and my response. Furthermore, this
abstraction requires the integration of information from different sensory
modalities. As Skinner (1957) observed, the act of producing a burst of
speech sound is very different from the act of listening to it, prior to
attempting one’s own production. Cross-modal co-ordination is therefore
required to bridge the gulf between perception and performance (Meltzoff,
2002).
The act of imitation is complex. But these complexities are overcome, to
some extent, because the ability to imitate may well be inborn. Infants can
imitate adult tongue protrusion within minutes of birth (Meltzoff & Moore,
1983). Within the first week, they can even imitate an adult who raises either
one or two fingers (Nagy, Pal, & Orvos, 2014). This ability is all the more
remarkable because the newborn infant has not yet seen their own face
(Meltzoff & Moore, 1997). In other ways, too, infants have been described as
‘prolific imitators’ (Marshall & Meltzoff, 2014). The genetic basis of
imitation has been challenged recently (Cook, Bird, Catmur, Press & Heyes,
2014). But none deny that the primate brain possesses a dedicated functional
capacity for imitation, operating via so-called mirror neurons. Research on
monkeys shows that these special neurons (in the premotor cortex) discharge
both when an action is observed and also when it is performed (Gallese,
Fadiga, Fogassi & Rizzolatti, 1996). In monkeys, the sight of another monkey
grasping, placing or manipulating an object will cause mirror neurons to fire.
Mirror neuron activity is associated with the monkey’s own motor response,
which often constitutes an imitation. Mirror neurons therefore form a link
between an observer and an actor. Mirror neurons have also been discovered
in the human brain. At one point, it was believed that mirror neurons could be
detected in Broca’s area (Rizzolatti & Arbib, 1998), an area in the left
hemisphere long known for its importance in the motor planning of speech
and probably also in the processing of syntax (Obler & Gjerlow, 1999).
However, more recent studies locate the human mirror neuron system more
accurately in a region known as premotor BA6 (a catchy name, I think you’ll
agree) (Cerri et al., 2015). Whatever the precise location of mirror neurons,
their existence points to the intrinsic importance of imitation in human
behaviour.
Imitation: Who, when and how?
In the study of imitation, three factors demand attention: (1) the actors: who
is imitating whom? (2) the time lag: how long is the delay between model
and imitation? and (3) precision: how closely does the imitation match the
model? With regard to the actors, imitation can work both ways. The child
can imitate the adult, but equally well, the adult can (and does) imitate the
child (Olson & Masur, 2012). We will consider one form of adult imitative
behaviour in our discussion of corrective input, below. The second factor to
consider concerns the time lag between presentation of a model and its
subsequent imitation. Imitation can be either immediate or deferred (Stern,
1924). It is easy to establish that an imitation has occurred when it is
immediate. But imitation may also be deferred until quite some time after
presentation of the model (Meltzoff, 1988). In this case, researchers need to
be more resourceful in confirming that a given behaviour does indeed
constitute imitation. Perhaps for this reason, child language research on
deferred imitation is limited. But there is some evidence that children can
imitate utterances first presented several hours or even days earlier (Clark,
1977; Moerk & Moerk, 1979; Snow, 1981). Logically, of course, every time
a child uses a new word for the first time, in the immediate absence of a
parental model, we witness a case of deferred imitation. The child must have
heard each word spoken by a parent at some point previously. Imitation must
therefore play an important role in lexical development.
A third factor that has proven important is the fidelity of the imitation to the
original model. As noted above, it is pretty much impossible to imitate
someone else’s verbal behaviour precisely. Nevertheless, most researchers
would describe an imitation as ‘exact’ if the same words are reproduced in
the same order. It turns out that this kind of verbatim repetition is rare,
accounting for no more than 10 per cent of the child’s imitative behaviour
(Snow, 1981). It is far more common for the child to incorporate just part of
an adult utterance into their own response, which may also embody further
additions, reductions or modifications. Strictly speaking, therefore, the child
imitates selectively, as shown in the following examples from Nathaniel aged
2;3–2;7 (Snow, 1981):
Mother: A Dutch house.
Nathaniel: Nathaniel Dutch house.
Mother: What’s this?
Nathaniel: What’s this a boat.
Mother: The pigs are taking a bath.
Nathaniel: Taking a bath and make juice.
Mother: You like jingle bells.
Nathaniel: Like other jingle bells.
Imitations of various kinds are frequent in adult–child conversation. Clark
and Bernicot (2008) analysed nine hours of data from 41 French children and
report that both children and adults are very repetitive, with adults having a
slight edge on their children (Table 4.2).
Adults also repeat themselves a great deal, tending to rely on a restricted
repertoire of verbal routines of the form: Look at NP, Here’s NP and Let’s
play with NP (where NP stands for noun phrase like the dog or Auntie
Mary). These routines are used to introduce new information, with the noun
in the noun phrase often being produced with exaggerated intonation and
heavy stress (Broen, 1972). New information appears sentence-finally 75 per
cent of the time in CDS, as against 53 per cent for ADS (Fernald & Mazzie,
1991). This is significant, because new information is more exposed in
sentence-final position (Slobin, 1973), and young children respond more
accurately to new information presented in this way (Shady & Gerken, 1999).

Individual differences in imitation

Research is beginning to show that our ability to imitate varies and,
moreover, that imitation ability has an impact on language development. This
topic has been investigated as part of the Twins Early Development Study in
the UK (McEwen, Happé, Bolton, Rijsdijk, Ronald, Dworzynski & Plomin,
2007). McEwen et al. (2007) examined more than 5,000 twins for their ability
to imitate non-verbal behaviours like gestures and facial movements. They
found that good non-verbal imitators had higher vocabulary scores. This
finding is intriguing because the link is between non-verbal imitation and
language development. McEwen et al. (2007) also used their twin sample to
examine the genetic basis of imitation. By comparing identical twins with
non-identical twins, they could estimate the heritability of imitative capacity
(Plomin, 1990). They found that 30 per cent of the variance in imitative skill
could be attributed to genetic factors. This is a fairly modest figure and means
that environmental factors play an important role in distinguishing between
good and poor imitators. It remains for future research to determine more
precisely how the child’s upbringing affects their capacity to imitate. Perhaps
some parents encourage this kind of behaviour through their own example,
by providing high levels of imitations themselves.
Individual differences also exist in verbal imitative ability (Bloom, Hood &
Lightbown, 1974). An important measure of these differences is the nonword
repetition task (Gathercole, 2006). Children are presented verbally with a
series of nonsense words, like prindle, frescovent and stopograttic, and are
asked to repeat each one back to the experimenter. The ability to imitate (or,
as the authors prefer, repeat) these words is highly correlated with children’s
vocabulary level at four, five and six years (Gathercole, Willis, Emslie &
Baddeley, 1992; Hoff, Core & Bridges, 2008). Good nonword repetition
skills are also found in children who produce relatively long, syntactically
complex utterances (Adams & Gathercole, 2000). At the other end of the
spectrum, children with serious language impairments do very poorly at
repeating back to another person polysyllabic nonwords (Bishop, Adams &
Norbury, 2006). The nonword repetition task is generally discussed within
the context of individual differences in working memory capacity (Archibald
& Gathercole, 2007). But performance on this task also relies on the child’s
imitative capacity. It is possible, then, that imitative skill is partly constrained
by working memory capacity.
Masur & Olson (2008) provide another example of how individual
differences in imitative ability have an impact on child language
development. These authors examined imitation of both verbal and non-
verbal behaviours, by mothers and children. In a longitudinal design, children
were tested at four different points between the ages of 10 and 21 months.
Mothers imitate their infants’ behaviours and infants demonstrate an
increasing responsiveness to, and awareness of, being imitated. In response to
being imitated by an adult, the child may repeat the imitation back, in return.
Thirteen-month-olds who engaged more in this chain of imitation were more
lexically advanced at 21 months. And generally speaking, infants who were
most responsive to maternal imitations at 10 months were also those children
with more advanced vocabularies at 17 and 21 months (see also Olson &
Masur, 2015). In a similar vein, verb learning was enhanced in children aged
30 and 36 months when they actively imitated the action associated with a
verb, rather than just passively observed it being performed (Gampe, Brauer
& Daum, 2016).
Another source of individual differences in imitation is the motivation of the
child to engage with other people. Imitation is a fundamentally social
behaviour, one that assumes a strong impulse to interact with other people.
But this impulse is weaker among certain individuals, including those with
autism. Of interest, children with autism do not perform well on non-verbal
imitation tasks (Vivanti, Nadig, Ozonoff & Rogers, 2008). In fact, these
children can copy another person’s behaviour, but their motivation to do so
seems to be depressed by a relative lack of interest in the goals and intentions
of other people. Even within this population, though, there is considerable
variation in the rate at which children imitate other people. Significantly,
among children with autism, the ability to imitate simple sounds at the age of
three years predicts general spoken language ability at the age of five years
(Thurm, Lord, Lee & Newschaffer, 2007). Good imitators turn out to be good
talkers.
Generally speaking, we see that individual differences in imitative behaviour,
both verbal and non-verbal, have an impact on language development. But
despite individual differences, we have also seen that imitation is a
fundamental human capacity. It has a partly genetic basis and exploits a
dedicated neurological resource in mirror neurons. With regard to language
acquisition, we have argued that it deserves our close attention, as a critical
form of interaction between adult and child. In the next section, we will focus
on imitation of the child by the adult and consider how it might influence
language acquisition. In particular, we consider its potential as a form of
correction for child grammatical errors.
Corrective input
We begin this section by describing a special form of adult imitation, the
recast. In what follows, we zoom in on one particular kind of adult recast that
might function as a form of corrective input for child grammatical errors.
More broadly, we consider the significance of findings on corrective input for
theories of child language acquisition.
Recasts: Adult repetition of the child
Adult imitation of child speech is often embellished in one way or another. In
fact, repetitions with minor variations to the original utterance are the
hallmark of adult–child discourse. This kind of adult imitation is often
referred to as a recast, and was first studied by Roger Brown and his students
(Brown & Bellugi, 1964; Cazden, 1965). The following examples are from
Brown (1973):
Eve: Eve get big stool.
Mother: No, that’s the little stool.
Eve: Milk in there.
Mother: There is milk in there.
Eve: Turn after Sarah.
Mother: You have a turn after Sarah.
Recasts fall out naturally from conversation with a two-year-old. It is unlikely
that adults consciously recast child speech, but they do it a lot (see below).
The prime function of recasts is to maintain the flow of conversation with a
partner who is cognitively and linguistically immature. This is achieved, in
part, by reproducing some or all of the child’s own words and structures. In
so doing, the adult increases the chance of being understood by the child.
And, of course, the conversation topic is of interest and relevance to the
child, since it follows the child’s lead. In essence, the adult adopts the
linguistic framework supplied by the child, so any additions or changes they
introduce will place the minimum burden on the child’s processing and
memory resources. In consequence, it is likely that any new or unfamiliar
language will be more readily assimilated by the child. In this vein, studies
with adults have shown that grammatical forms that have just been used by
another speaker are more easily accessed and produced (for example,
Kaschak, Loney & Borreggine, 2006). The value of recasts has been
established in numerous studies that demonstrate their association with
language growth, in both typically and atypically developing children (for
example, Seitz & Stewart, 1975; Nelson, Denninger, Bonvillian, Kaplan &
Baker, 1984; Hoff-Ginsberg, 1985; Forrest & Elbert 2001; Eadie, Fey,
Douglas & Parsons, 2002; Swensen, Naigles & Fein, 2007). Recasts are one
sign that adults are being responsive in conversation with children, and
parental responsiveness in general is closely linked with language
development (Tamis-LeMonda, Kuchirko & Song, 2014).
The ‘no negative evidence’ assumption
All language learners make errors. In fact, errors are the hallmark of language
development. They demonstrate that a recognized, mature end-state has not
yet been reached. From about the age of 15 months, when the child begins to
put two or more words together in multi-word utterances, very little of the
child’s output looks adult-like in its grammatical form:
Eve aged 1;6 (Brown, 1973):
more cookie
Mommy read
Fraser water?
that radio (following the mother’s question: What is that?)
dollie celery
All typical children produce errors. But eventually, and just as surely,
children retreat from error as they acquire an adult-like, mature knowledge of
grammar. The problem, then, is to explain how errors are expunged. The
most obvious answer, explored below, is that parents correct their children’s
errors. But many child language researchers reject this suggestion. Nativists,
in particular, have argued that the child receives no help from the linguistic
environment in eradicating errors (for example, Weissenborn, Goodluck &
Roeper, 1992). In nativist terms, the child receives ‘no negative evidence’,
that is, no information about what is, and is not, grammatical. Adherence to
this assumption has a radical effect on one’s view of the task that faces the
language-learning child. If there is no guidance in the input on
grammaticality, how does the child acquire this knowledge? The nativist
answer is that this knowledge must be innate and will, at some point in
development, come to the child’s rescue (see Chapter 8). Many non-nativists
have also expressed support for the ‘no negative evidence’ assumption
(Ambridge, Pine, Rowland & Young, 2008). These researchers rely on the
child’s general learning mechanisms to explain how children retreat from
error (see Chapter 9). In this section, though, we will argue that the ‘no
negative evidence’ assumption is unfounded. There is evidence that parents
do correct their children’s grammatical errors.
We begin our review of the evidence by returning to the adult recast. Brown
& Hanlon (1970: 197) were the first to recognize the corrective potential of
adult recasts with their observation that ‘repeats of ill-formed utterances
usually contained corrections and so could be instructive.’
Eve aged 1;6 (Brown, 1973)
Eve: Want lunch.
Mother: Oh you want lunch then.
Eve: Mommy gone.
Mother: Mommy’s gone.
Eve: Fraser coffee.
Mother: Fraser’s coffee.
Brown & Hanlon’s suggestion concerning the corrective potential of recasts
has been ignored for many years. Instead, the focus has been on a different
kind of adult response, also investigated by Brown & Hanlon (1970), which
they call a Disapproval. Behaviourist in inspiration, it was thought that
Disapprovals might function as a form of corrective input. The idea was that
parents might explicitly signal their displeasure with ungrammatical child
sentences with injunctions like no, that’s wrong, or don’t say that. But it
turned out that Disapprovals were not contingent on child grammatical errors.
Instead, Disapprovals were contingent on the meaning of child speech (No,
that’s a purple sweater, not a blue one). The influence of Brown & Hanlon’s
study has been extensive. Researchers continue to cite the evidence on
Disapprovals as evidence that parents do not correct child errors (Tomasello,
2003). But there is more than one way to provide corrective input.
Disapprovals are just one possibility, a rather implausible one at that. Far
more promising are the special recasts – ‘repetitions with corrections’ – that
Brown & Hanlon (1970) observed, but unfortunately neglected. Attempts to
revive interest, beginning with Hirsh-Pasek, Treiman & Schneiderman
(1984), have not been universally welcomed (Tomasello, 2009). But we shall
give them more leeway than is typical in what follows.
Contrastive discourse
Look again at the three exchanges between Eve and her mother, above. In
each case, Eve’s utterance is ungrammatical. And in each case, Eve’s mother
provides grammatical versions of the erroneous elements in Eve’s speech. As
you can see, the adult response, in each case, is a particular kind of recast.
Recall that recasts preserve some of the child’s original words, but changes
are also introduced. The adult ‘expands, deletes, permutes, or otherwise
changes the … [child utterance] while maintaining significant overlap in
meaning’ (Bohannon, Padgett, Nelson & Mark, 1996: 551). Hence, recasts
can take many forms. When it comes to corrective input, though, we are
interested only in those recasts where the change constitutes the provision of
a correct form directly following a child error. It is possible that this kind of
recast counts as a form of negative evidence for the child. Negative evidence
is a topic that I have investigated myself. The following examples come from
a diary study I conducted with my son Alex, when he was four-years-old
(Saxton, 1995):
Alex: I had all my breakfast and I drinked up all the milk
Matthew: You drank the whole bowl?
Alex: Listen to me! I was talked first
Matthew: You weren’t talking first!
Alex: I’m easy to eat you up.
Matthew: You can eat me up easily?
Alex: All by her own. All by her own. All by her own.
Matthew: All by herself?
Incidentally, bold has been used to highlight errors and correct alternatives. It
does not indicate stress, which may well lie elsewhere. To explain how the
child might identify the corrective potential in these responses, I formulated
the so-called direct contrast hypothesis (Saxton, 1997). The prediction is that
the correct adult form is especially conspicuous when it directly follows a
child error. In the first example above, Alex uses drinked, but I counter this
directly with the correct version, drank. The contrast between the two forms
is therefore especially noticeable. It is predicted that the child will recognize
their own selection (in this case, drinked) as erroneous, and furthermore, will
recognize that the adult alternative (drank) constitutes the appropriate form.
It has been discovered that contrastive discourse of this kind is very common.
In one study, as many as 65 per cent of all child errors met with this kind of
correction (Chouinard & Clark, 2003). A large number of other studies also
report high levels of contrastive discourse (for example, Hirsh-Pasek et al.,
1984; Demetras, Post & Snow, 1986; Penner, 1987; Bohannon & Stanowicz,
1988; Morgan & Travis, 1989; Moerk, 1991; Farrar, 1992; Furrow, Baillie,
McLaren & Moore, 1993; Post, 1994; Morgan, Bonamo & Travis, 1995;
Strapp, 1999; Strapp & Federico, 2000; Saxton, Backley & Gallaway, 2005).
This high prevalence is perhaps not surprising. Young children produce
numerous errors, whereas adults do not. The chances are therefore high that
errors and their correct counterparts will often be found sitting side-by-side in
adult–child conversation.
It is one thing to show that negative evidence is supplied to the child. We also
need to determine whether it works. That is, do children pick up on the
corrective information on offer, and use it to modify their own developing
grammars? If corrective recasts really do function as a form of negative
evidence, the child should respond to them by adopting the correct forms
offered by the adult. There is evidence that children can, in fact, do this.
Some more examples from Alex:
Alex (aged 4;1–4;9)
Alex: It’s even gooder than anything.
 It’s gooder, isn’t it?
Matthew: Yes, it is better.
Alex: Better, yeah.
Matthew: What did he do?
Alex: He wiped him.
Matthew: He wiped himself.
Alex: Yes, he wiped himself.
Alex: He’s got little nice feet.
Matthew: Oh, he has got nice little feet.
Alex: Yes, he’s got nice little toes.
Alex: It’s bored of being on the bike.
Matthew: It’s not boring.
Alex: Yes, it’s boring on the bike.
Alex: That’s what happens to Tarzan Dog. He gets falled over by Tarzan
Man.
Matthew: I make him fall over, do I?
Alex: What?
Matthew: I make him fall over, do I?
Alex: Yes, you make Tarzan Dog fall over with your sword.
If you have ambitions to be a child language researcher, then it helps to breed
your own source of data. Thanks, Alex. We can see in these examples that
Alex is sensitive to the linguistic form (not just the meaning) of my utterance.
He is capable, on some occasions at least, of switching, from his own
erroneous uses to the ones offered by me, in the form of a direct contrast.
Experiments and studies using naturalistic conversational data confirm that
children are responsive to this kind of correction (Saxton, 1997; Saxton,
Kulcsar, Marshall & Rupra, 1998; Saxton, 2000; Strapp, Bleakney, Helmick
& Tonkovich, 2008; Laakso & Soininen, 2010; Holtheuer & Rendle-Short,
2013; see also Box 4.2). It has also been shown that the child’s immediate
responsiveness to negative evidence is not ephemeral: the beneficial effects
on the child’s grammar can be observed, in some cases, several weeks later
(Saxton et al., 1998). In other cases, effects are apparent for some structures,
but not for others (Saxton, Houston-Price & Dawson, 2005). Teasing apart
the effects of one kind of adult response from all other sources of input
influence is no easy matter, empirically. What evidence we have, though, is
broadly supportive of the idea that negative evidence can facilitate the
acquisition of more adult-like states of grammatical knowledge.
Box 4.2 Novel Words (Saxton, 1997)
The beauty of novel words is that one knows exactly how many
times children have heard them and in what context. The novel
words in the study described here are irregular past tense verbs
(Saxton, 1997). Children are first taught the present tense of verbs
like streep and pell. It is important to give children lots of practice
at recognizing the new words and getting their tongues round
them in different forms (pell, pells, pelling). A schedule of
multiple training sessions – little and often – works well. The
characters shown here were modelled on glove puppets that the
children used to act out the verbs.
The next stage was to elicit a past tense form from the child. This
was done by showing a video featuring the familiar puppet
characters. A dragon is shown asleep under a tree, snoring loudly.
To stop him snoring, one of the characters suggests they sneak up
on the dragon and wake him by performing one of the actions.
The poor dragon wakes up, roaring loudly, at which point the
video is paused and the child is asked: What happened? One has
thus created a past tense context. The child has never heard the
past tense form of these verbs, so they have no choice but to treat
the verbs as regular: He pelled his nose. But little do they know:
these are irregular verbs. We have not only induced a past tense
form, but an error also.

sty / stought
prodding action performed with a plastic concertina that makes a
honking noise (cf., buy / bought)

streep / strept
ejection of a ping pong ball from a cone-shaped launcher towards
a target (cf., creep / crept)

pell / pold
a beanbag on the end of a string is swung like a bolas at the target
(cf., sell / sold)

pro / prew
twisting motion applied with a cross-ended stick (cf., throw/
threw)
jing / jang
a beanbag is catapulted from a spoon at a target (cf., sing / sang)

neak / noke
repeated clapping motion in which target is trapped between the
palms (cf., speak / spoke)
This means that the child can be corrected:
Child: He pelled his nose.
Adult: Oh yes, he pold his nose.
Alternatively, the correct irregular past tense form can be
modelled as soon as the video is paused, before the child has had
a chance to make an error: Oh look! He pold his nose. In both
cases, the adult models the correct past tense form. The difference
is simply that in one condition only, the correct model is supplied
in a contrastive way, directly following a child error. If direct
contrasts do indeed function as a form of correction for the child,
then they should be more likely to reject their own, erroneous
version in favour of the correct adult version. And this is precisely
what was found.
 Note that each novel word sounds like a genuine word (they are
phonologically plausible). And the meaning of each verb is also
novel. This prevents any form of competition with meanings that
already exist in the child’s repertoire. Novel words have a long
history in child language research (Berko, 1958). As noted,
control over the input is their chief virtue. A major limitation is
that it is usually only possible to expose children just a handful of
times to novel words. Practical considerations prevent children
being exposed hundreds (or even thousands) of times, as with
many real words. Nevertheless, novel words continue to be a
valuable tool in the armoury of child language researchers.
Negative feedback
Another form of corrective input can also be identified, which I term negative
feedback (to distinguish it from negative evidence).
Alex (aged 4;1–4;9)
Alex: Knights have horse, they do.
Matthew: They what?
Alex: He slided down the door.
Matthew: He what?
Alex: Ouch! It hurted.
Matthew: Eh?
Negative feedback is provided when adults seek to clarify what the child has
said, following a grammatical error. My responses here can be classified in
each case as an error-contingent clarification question (CQ). CQs feature as a
significant part of conversation, accounting for something like 5 per cent of
all turns in adult–adult discourse (Purver, Ginzburg & Healey, 2001). We
constantly need to check on each other’s meaning, but my suggestion is that,
incidentally, they may help the child check on the syntactic form of their
speech. That is, error-contingent clarification questions may function as a
form of negative feedback. Children as young as 12 months old respond
appropriately to clarification questions, by returning to their original
utterance and either repeating or repairing it in some way (Golinkoff, 1986).
There is some evidence also that children correct themselves, following the
intervention of an error-contingent CQ. They switch from erroneous to
correct forms for a wide range of grammatical errors, when supplied with
negative feedback (Saxton, 2000; Saxton et al., 2005). If we think about how
negative feedback works, then it is quickly apparent that it provides a weaker
form of corrective information than negative evidence. This is because the
adult provides no correct alternative to the child error. The prompt hypothesis
predicts that negative feedback provides a cue for the child that, essentially,
jogs their memory about language forms they have already learned. Overall,
we see that parents provide at least two kinds of corrective signal, both
negative evidence and negative feedback.
Corrective input: Summary
The problem of explaining how children overcome their language errors must
be addressed in all approaches to language acquisition. Corrective input
provides the most straightforward solution to this problem. And the evidence
is now quite compelling that parents not only correct grammatical errors, but
that children respond accordingly. We know this from a recent meta-analysis
conducted by researchers with an interest in therapeutic interventions for
children with language delays (Cleave, Becker, Curran, Van Horne & Fey,
2015). In essence, meta-analyses provide a statistical method for throwing the
data of numerous studies into a big metaphorical pot to determine, overall,
the fundamental trends in the findings. This kind of analysis is especially
useful for investigating controversial topics, where conflicting evidence
arises over the years. Cleave et al. (2015) deduce that error-contingent recasts
are indeed a valuable source of learning for the child. Even so, many
researchers continue to adhere to the ‘no negative evidence’ assumption. For
example, a devout non-nativist, Tomasello (2009: 82) remarks that ‘adults do
not explicitly correct child utterances with any frequency’. But the basis for
this assertion is limited to the behaviourist-inspired notion of Disapproval
reported by Brown & Hanlon 40 years ago. When I published the first edition
of this book, Tomasello was one of many non-nativists who espoused the
view that negative evidence is either not available to the child or of little
practical use. Since then, though, the ground has begun to shift. For example,
Ambridge (2013: 510) explicitly accepts that negative evidence ‘likely plays
an important role in children’s learning’. Ironically, one researcher who has
long supported the idea of parental corrections is Chomsky:
certain signals [child sentences] might be accepted as properly formed
sentences, while others are classed as nonsentences, as a result of
correction of the learner’s attempts on the part of the linguistic
community. (Chomsky, 1965: 31)
As with imitation, it is widely – but wrongly – assumed that Chomsky
rejected the notion of corrective feedback (e.g., Brooks & Kempe, 2012). But
before we get too enthusiastic, it is unlikely that corrective input is either
necessary or sufficient to guarantee the successful acquisition of grammar.
Other mechanisms are undoubtedly at work. The ‘no negative evidence’
assumption has created an abiding puzzle, and hence, an abiding prompt to
theorizing. It has encouraged the exploration of multiple solutions to the
same learning problem, some or all of which may be available to the child
(cf., Golinkoff & Hirsh-Pasek, 2008). We shall consider some of these
proposals in Chapters 7, 8 and 9. The more suggestions we have to consider,
the more likely it is that we shall discover how the child succeeds in
acquiring an adult sense of grammaticality.
Universality of CDS
While linguistic input is clearly necessary for language learning, it is less
obvious whether Child Directed Speech is necessary. It is conceivable that
the many fine adjustments and simplifications typical of CDS are simply
icing on the cake: a welcome addition, that may facilitate language
development, but which are not in any way necessary. To test this idea, all we
need to do is find a single child who has been deprived of this special input,
but who has nevertheless acquired language normally. In fact, it is widely
assumed that many children are so deprived. Pinker (1994: 40) asserts that
‘in many communities of the world, parents do not indulge their children in
Motherese’. He further adds that ‘the belief that Motherese is essential to
language development is part of the same mentality that sends yuppies to
“learning centers” to buy little mittens with bull’s-eyes to help their babies
find their hands sooner’ (ibid.: 40). In fact, Pinker is setting up a straw man
here, since no-one, to my knowledge, has argued that CDS is necessary,
rather than simply facilitative. More to the point, when we look at the
empirical evidence, Pinker’s assertion is not supported. What evidence we
possess indicates that children around the world are exposed to at least some
of the characteristic features of Child Directed Speech.
A limited number of studies are continually cited in defence of the idea that
CDS is not universally available (see Saxton, 2009). We will focus on the
best known example, the African-American community of Trackton in South
Carolina studied by Heath (1983). The Trackton adults studied by Heath were
dumbfounded by the idea that parents should modify their speech when
talking to infants:
Now just how crazy is dat? White folks uh hear dey kids say sump’n,
 dey say it back to ‘em, dey aks ‘em ‘gain ‘n’ ‘gain ‘bout things, like they
‘posed to be born knowin’. You think I kin tell Teegie all he gotta
know? Ain’t no use me tellin’ him: learn dis, learn dat. What’s dis?
What’s dat? He just gotta learn, gotta know. (Heath, 1983: 84)
Heath (ibid.: 7) is the first to acknowledge that she is not a psychologist. And
this is important because what we have here is an informant who is not the
main caregiver of the child and who is reporting on her beliefs about child
rearing. What her actual practices are remain unclear. In fact, there is a
marked contrast between what people say they do in talking to children and
what they actually do. Haggan (2002) conducted interviews with 82 Kuwaiti
adults and discovered that 18 of them were adamant that they made no
concessions when talking to children. Some even suggested that the use of
‘special ways’ to communicate would be detrimental to language
development (ibid.: 22). Haggan then observed each of these ‘CDS sceptics’
interacting with a child aged 2–3 years. She found that every single one of
these sceptics modified their speech in ways entirely typical of Child
Directed Speech. These include the use of short, semantically simple
sentences, concrete referents based on the child’s own interests and extensive
repetitions. Had Haggan stopped her study at the interview stage, she would
have been left with a fundamentally mistaken view of how Kuwaiti adults
interact with their young children. In a similar vein, Birjandi & Nasrolahi
(2012) report that, in the Iranian city of Babol, parents generally deny
correcting their children’s errors. Nevertheless, a majority of parents do, in
fact, supply negative evidence.
As it happens, there is a strong indication that the parents in Trackton do use
elements of CDS in their talk with young children. For example, ‘when adults
do not understand what point the young child is trying to make, they often
repeat the last portion – or what is usually the predicate verb phrase – of the
child’s statement’ (Heath, 1983: 93). This kind of interaction sounds
remarkably like the recasts described above. More research of a less
anthropological flavour, with a stronger psychological bite, would be
welcome here. But it is apparent that key pieces of evidence from Heath’s
study have been overlooked. Instead, much attention has been paid to Heath’s
observation that ‘during the first six months or so, and sometimes throughout
the entire first year, babies are not addressed directly by adults’ (ibid.: 77).
This is a dramatic assertion. Why would parents not talk to their babies? It is
suggested that ‘for an adult to choose a preverbal infant over an adult as a
conversational partner would be considered an affront and a strange behavior
as well’ (ibid.: 86; cf., Pye, 1986). Again, though, what we need is some hard
evidence on what Trackton parents actually do, both in private and in public.
What we do know for certain is that parents and others must, and indeed do,
start talking to the child at some point. We also know that parents are not the
only source of CDS. Children as young as 4;0 modify their speech when
talking to toddlers (Shatz & Gelman, 1973; Weppelman, Bostow, Schiffer,
Elbert-Perez & Newman, 2003). In Trackton, there is evidence that older
children may supply specially modified input for their younger siblings and
peers (Heath, 1983: 93).
Every possible feature of CDS does not appear in the speech of every parent
throughout the world (Pye, 1986; Lieven, 1994). More likely, there is a
‘smorgasbord effect’. The speech of parents in different cultures will express
different combinations of CDS features, ‘selected’ from the full menu. Cross-
cultural research does reveal the widespread occurrence of one important
CDS feature: the practice of repeating child speech back to them in one form
or another. Repetitions, elicited repetitions and recasts have been recorded in
the adult input in a wide range of languages. Thus, Mead (1930: 35) remarks
of the Manus people of New Guinea that the adult’s ‘random affection for
repetitiousness makes an excellent atmosphere in which the child acquires
facility in speech’. In a Danish context, Jespersen (1922: 142) made a similar
observation: ‘understanding of language is made easier by the habit that
mothers and nurses have of repeating the same phrases with slight
alterations’. In addition to Manus and Danish, speakers of many other
languages provide recasts for their children, including French (Chouinard &
Clark, 2003), K’iche’ Maya from Guatemala (Pye, 1986), Hebrew (Berman,
1985), Mandarin (Erbaugh, 1992), Persian (Birjandi & Nasrolahi, 2012),
Japanese (Clancy, 1985), Korean (Clancy, 1989), Samoan (Ochs, 1982), and
Warlpiri from Australia (Bavin, 1992). And, of course, we can also add
Trackton in South Carolina to this list.
Generally speaking, the idea that CDS is confined to well-educated, white,
middle-class mothers is without foundation. There has not been a great deal
of research, and much of it is, in any case, anthropological in nature. It was
not designed for the study of language acquisition. Moreover, this research
has often been communicated to the child language research community in a
partial, inaccurate manner. For example, the recasts found in the speech of
Trackton parents are routinely overlooked. At the same time, errors have
crept in that betray a lack of direct engagement with the literature. For
example, ‘Trackton’ sometimes appears as ‘Tracton’ (Hamilton, 1999: 430),
while the informant quoted above has been wrongly described as ‘Aunt Mae’
(Pinker, 1994: 40). She is, in fact, ‘Annie Mae’. Moreover, she is not the
child’s aunt, but his grandmother. Repeat warning: read the original sources
for yourself. All in all, the assumption that CDS is absent in some cultures
lacks support. In fact, the available evidence points in the opposite direction
with at least some standard features of CDS being reported in all cases so far.
Undoubtedly, though, we require better evidence on this issue. Future
research should tell us much more about the linguistic environments of
children throughout the world.
Input and interaction in language acquisition
The child’s linguistic environment was a major focus of interest during the
1970s, but since then, interest has declined. We can identify at least three
reasons for this decline. First, many of the basic facts about Child Directed
Speech were established during this period. And although subsequent studies
have refined these findings, they have not substantially altered our view
about the nature of CDS. Second, the significance of a special register,
directed at young children, was undermined by the assumption that CDS is
largely confined to a privileged minority of Western mothers. This
assumption still prevails, but we have challenged it here, in our review of
cross-cultural research. Third, the locus of interest, in both nativist and non-
nativist theories, has shifted firmly to the child. The central aim is to describe
and explain the learning mechanisms and knowledge that the child brings to
the task of language acquisition. This shift towards the child, away from the
environment, is entirely justified. We want to discover how the child acquires
language, after all. But the environment still deserves our attention.
As we shall discover in Chapter 9, present theorizing reduces the role of the
input to a matter of frequency. The number of times that a child hears a
particular linguistic form is held to be the critical environmental factor that
contributes to language development. Of course, frequency is important
(Cameron-Faulkner, 2012; Ambridge, Kidd, Rowland & Theakston, 2015).
Children’s experience of language differs quite widely with regard to input
frequency, with a direct impact on language development. But frequency is
not the only factor of interest. Interaction is also critically important, both
verbal and non-verbal. In this regard, imitation – as a form of interaction –
stands out as a fundamental human behaviour that is implicated in language
development. The study of corrective input can be seen as one branch of the
broader topic of imitation studies. Language is acquired in the context of
interaction with other people and input alone – simple exposure to linguistic
forms – will not suffice.
Even if we accept that both input and interaction are essential ingredients for
language development, that does not commit us to the need for a special
register. In principal, we could speak to a baby or toddler in precisely the
same way that we speak to an adult. The assumption that CDS is not
universally available leads directly to this conclusion. And yet, my guess is
that very few researchers, if any, would advocate withholding CDS and
replacing it with Adult Directed Speech. In any event, our review of cross-
cultural research revealed that assumptions about the ‘non-universality’ of
CDS are unfounded. At the same time, much more cross-cultural research is
needed, designed by psychologists with a direct interest in the environment of
the language learning child. Personal guess Number 2 is that parents
throughout the world will adopt at least some of the characteristic features of
CDS at least some of the time. This belief is based on the assertion that Child
Directed Speech is actually inevitable (Saxton, 2009).
Arguably, the only way to engage a young child in conversation is to follow
their lead and this includes the choice of conversation topic. Try talking to a
two-year-old about your council’s recycling policy and see how far you get.
Now try talking about the book that the child is holding in their hands. Notice
the difference? The young child is constrained, both linguistically and
cognitively, and these constraints compel adults (and older children) to adapt
to these limitations. Otherwise, communication will not succeed. An adult
who fails to engage the child will only succeed if their next attempt dovetails
more closely with the child’s limited capacity and particular motivation to
communicate. It is not surprising, therefore, that parents around the world
spend so much time recasting child speech, effectively acting like a mirror,
reflecting back the child’s own speech with a range of modifications. By
following the child’s lead in this way, the adult can more easily maintain a
conversation that is geared to the level and interests of the child. Observe that
the motivation of the adult, conscious or otherwise, is to communicate with
the child, not teach language. It just so happens that a special form of
facilitative input and interaction – Child Directed Speech – falls out naturally
from communication with a young child. It is in this sense that CDS might be
inevitable. Given that both input and interaction are necessary for language
development, it remains possible that at least some aspects of the special
register, Child Directed Speech, are essential to guarantee successful
language acquisition.
In a Nutshell
Adults and older children use a special register, known as
Child Directed Speech (CDS), when talking to young
children. They simplify and clarify their speech in numerous
ways, at every level of language, including phonology,
vocabulary, morphology and syntax.
CDS is a dynamic register, with adult speech being tuned, to
some extent, to the language level of the child as it changes
over time.
The amount of linguistic input available to children varies
quite widely. Children of ‘high-talking’ parents tend to be
relatively advanced in terms of vocabulary and grammar
development.
The child’s linguistic environment comprises both input
(language forms) and interaction (the way those forms are
used in conversation).
Children cannot acquire language from watching TV, a
situation in which they are exposed to input in the absence of
interaction. Interaction is therefore essential for language
acquisition.
Imitation, both verbal and non-verbal, is a fundamental
aspect of human interaction. Infants who are especially
responsive to imitation develop relatively advanced
vocabularies.
Parents frequently imitate their children, repeating back child
utterances with modifications (recasts).
Some adult recasts can function as a form of corrective input
for grammatical errors (negative evidence). They present a
direct contrast between a child error and a grammatical
alternative offered by the adult.
It is widely assumed that CDS is not universally available,
but cross-cultural research does not support that assumption.
Child Directed Speech may be an inevitable consequence of
adapting to the communicative needs of a conversational
 partner who is both cognitively and linguistically immature.
Further Reading
Hoff, E. (2006). How social contexts support and shape language
development. Developmental Review, 26(1), 55–88.
This article provides a thorough review of the social factors that
affect the amount and quality of talk to young children. It goes
beyond the factor of socioeconomic status considered here to
include numerous other factors, including ethnicity,
multilingualism, and childcare experience.
Matthews, D. (Ed.) (2014). Pragmatic development in first
language acquisition. Amsterdam: John Benjamins.
Although topics in interaction are touched on in this chapter, we
do not explore in any depth how the child comes to use language
as a tool for communication. This recent collection does an
excellent job of filling the gap.
Websites
Effects of TV viewing on child development:
Google: How TV affects your child – KidsHealth for the
following site: http://kidshealth.org/en/parents/tv-
affects-child.html. If you still need convincing about the
Evils of Television, this site takes you one step beyond
language acquisition to summarize the effects of TV
viewing on other aspects of child development. A note
of warning: this site is not intended for a scientific
audience, so be prepared to follow up any points of
interest by checking with the original research literature.
Still want more? For links to online resources relevant to this
chapter and a quiz to test your understanding, visit the companion
website at https://study.sagepub.com/saxton2e
5 Language in the First Year: Breaking the
Sound Barrier
Contents
Hunt the phoneme 120
In the beginning 120
Drops of sound in a river of speech 120
Categorical perception 122
Specialization towards the native language 126
Why I don’t speak Nthlakapmx 126
Loss or decline? 127
Enhancement of native contrasts 127
Individual differences in infant speech perception 130
Summary: Breaking the speech sound barrier 130
Word segmentation 131
The baby statistician 132
Learning in the real world 137
Prosodic cues to speech segmentation 138
Relative cue strength 140
Grammar from the babble 141
Phonemes, words and grammar: Summary 143
Overview
In this chapter we consider how the child gets started on the task of
language acquisition. We focus on infant speech perception in the first
year of life and show how the child breaks the sound barrier to
discriminate individual speech sounds and words. By the end of the
chapter, you will know more about:
categorical perception: the child’s innate mechanisms for
distinguishing one phoneme from another.
specialization towards the native language: how infants tune in to
the particular phonological properties of the input language, and
why they lose the power to tune in to foreign languages.
word segmentation: how the child uses the statistical properties of
the input and information on prosody to work out where one word
stops and the next one begins.
early grammar: how the infant learns something about
 grammatical organization in the first year of life before they have
uttered their first word.
Hunt the phoneme
In the beginning
As we saw in Chapter 1, language learning begins before the baby is even
born. Newborn infants can recognize a story (The cat in the hat) that they
first heard while still in the womb (DeCasper & Spence, 1986). They do this
by hearing and remembering information about the prosodic (musical)
qualities of the story. In this chapter, we will pick up the baton from the
moment of birth and follow the infant through to their first birthday. For most
babies, it takes a whole year of life before they produce their first word. But
infants are not idle during this period. They are avid listeners. We will see
that infants possess remarkably powerful mechanisms for processing speech.
This is just as well, given that the infant is faced with the daunting task of
tuning in to the specific properties of their native language (or languages).
The ‘tuning in’ process includes the need to discriminate individual speech
sounds (phonemes) and information about how phonemes combine with one
another into words. As we shall see, the perceptual abilities of the child are
used to identify individual sounds, individual words and even some
information about the grammatical organization of language. By 12 months,
therefore, the typical child is well equipped for the rapid progress in lexical
and grammatical development that follows.
Drops of sound in a river of speech
London is one of the most linguistically diverse cities in the world, with more
than 300 different languages being spoken by children in schools (Baker &
Eversley, 2000). Living in London, I constantly find myself eavesdropping
on languages I do not even recognize. It can be quite bamboozling. Where
does one word stop and the next one begin? What are the individual sounds
that make up each word? Perhaps if I wasn’t just idly eavesdropping this
would not seem so daunting. After all, within each word, one sound follows
another, and, by the same token, one word follows another. It looks as though
all I need do is identify the separate elements, like beads on a string. This is
quite a nice image, but unfortunately, it is not quite right. Let’s throw the
necklace out and try a different image. The words we hear in everyday
conversation flow together in a fast-moving river of sound. In a river, I
cannot tell you where one drop of water ends and the next one begins. In a
similar way, within each word, it is generally impossible to tell where one
phoneme stops and the next one begins. And the same is true for the
boundaries between words. A spectrographic analysis of fluent speech shows
that individual phonemes overlap in their physical properties, even at the
boundaries of words (Ladefoged, 2006). The infant language learner is thus
faced with two segmentation problems. In the river of speech, where are the
phonemes and where are the words? We will examine each of these problems
in more detail, and then consider how the child tackles them (see Figure 5.1).
Figure 5.1 A speech waveform of the sentence ‘Where are the silences
between words?’ (from Saffran, 2003: 111)

Let’s first think about the task of individuating the phonemes within a word.
The languages of the world contain something like 600 consonants and 200
vowels (Ladefoged, 2004). However, each language uses only a fraction of
this total number. For example, English has about 45 different phonemes,
while Igbo (a West African language) uses more like 35. Meanwhile,
Japanese and Warlpiri (an Australian Aboriginal language) make use of
roughly 25 each. The infant must therefore hone in on a particular small
subset of phonetic units from the full range of possible speech sounds. Let’s
take a single example, the phoneme /p/. It turns out that there are many
different kinds of /p/. The physical properties of the /p/ phoneme vary
depending on the age, gender, dialect and emotional state of the speaker. Less
obviously, /p/ is altered by the sounds it co-occurs with. Thus, the /p/ sounds
in lap, lip, lop and loop all differ from one another. This variation is caused
by coarticulation, the effects of the physical properties of neighbouring
sounds (Ladefoged, 2006). In this case, the vowel in each word affects the
way the lips, tongue and other articulators position themselves in readiness to
pronounce the following /p/ sound. Variation in the /p/ sound is also
introduced according to its position within a syllable. In lap, /p/ is syllable-
final, while in pal, it is syllable-initial. You can see for yourself how different
these two sounds are if you tear off a small strip of paper (not from this
page!) and hold it up close to your lips. When you say lap, not much
happens, but with pal, you should blow the paper over. This is because the /p/
in pal is produced with aspiration as air is released in a sudden burst. With
lap, this does not happen, because the syllable-final /p/ is not fully released.
We can see, therefore, that there are many kinds of /p/ sounds. But to the
untrained ear, a /p/ is a /p/ is a /p/, whatever its position in a word, regardless
of the sounds it co-occurs with, and regardless of who produces it. Generally,
therefore, adults do not perceive the differences within the /p/ category. They
only hear differences across categories, for example, /p/ versus /b/.
Categorical perception
We need to consider how the infant comes to recognize all the different /p/
sounds as a single phoneme, and, further, how /p/ is distinguished from every
other phoneme. One answer to this question was provided by Peter Eimas and
his colleagues, more than 40 years ago (Eimas et al., 1971). In an ingenious
experiment, Eimas et al. (1971) demonstrated that infants perceive speech
sounds categorically, in much the same way that adults do. Their study
focused on the contrast between /p/ and /b/, two sounds that are almost
identical in the way they are produced, apart from the feature of voicing.
We can demonstrate the concept of voicing most easily by taking a sideways
step away from /p/ and /b/ to consider another contrasting pair that allow a
more leisurely inspection: /s/ versus /z/. To detect the contrast between
voiced and voiceless sounds, place your fingers on your Adam’s apple
(therein lie the vocal cords). Take a deep breath and produce a continuous /s/
sound. Without stopping, switch to a /z/ sound. You should notice that, with
the shift to /z/, your Adam’s apple starts vibrating – an indication of voicing.
Next, try alternating between long /s/ and /z/ sounds. You will feel the
voicing switch on and off. Now look behind you. If you see someone looking
at you strangely, it’s time to stop this exercise and leave the room (take the
book with you).
If we now turn our attention back to /p/ versus /b/, there is one further detail
we need in order to appreciate the study by Eimas et al. (1971). Our
perception of voicing is not simply a matter of whether the vocal cords are
vibrating or not. It is also a matter of timing. We can see this if we place /p/
and /b/ in syllables: /ba/ and /pa/. When we produce /ba/, the lips are closed
initially, before bursting open. In most English speakers, voicing begins just a
fraction of a second after the lips have opened. In /pa/, meanwhile, voicing is
also switched on when the vowel is produced (all vowels are voiced), but
compared to /ba/, the start of voicing is delayed some time after the lips have
opened. There is, then, a difference in voice onset time (VOT). We perceive a
/b/ if voicing starts within 25 msec from when the lips open. And we perceive
a /p/ if the VOT is greater than 25 msec. This fine difference in timing is
critical for allowing us to differentiate /b/ from /p/, which in turn allows us to
distinguish different words in otherwise identical pairs, like pack-back, pull-
bull and pin-bin.
Eimas et al. (1971) tested infant sensitivity to voicing by synthesizing a range
of syllables that differed incrementally in VOT. Infants were tested on pairs
of syllables to see if they could discriminate between them, for example ba-3
and pa-1, in Figure 5.2 below. These two syllables differ in VOT by 20 msec
and, to adult ears, cross the boundary between /b/ and /p/. The stimulus pair
pa-1 versus pa-2, meanwhile, also differ by 20 msec in VOT, but this time
they both lie within the same phoneme category and would sound identical to
an adult.
Figure 5.2 Stimuli used by Eimas et al. (1971) to test infant sensitivity to
voice onset time (a negative VOT indicates the start of voicing before the
release of the lips)

Eimas et al. (1971) used a high-amplitude sucking (HAS) procedure to

habituate infants (see Box 5.1). The infant is played one syllable, say ba-3,
repeatedly. In this version of the habituation method, the infant has control
over how often they hear the syllable. The faster the infant sucks on a
specially rigged dummy, the more often ba-3 is heard. At first, sucking rate is
high, reflecting the high level of infant interest, but eventually the allure of
ba-3 fades and the sucking rate drops off. When habituation is achieved, a
new syllable with a different voice onset time is introduced, say, pa-1. If the
infant can detect a difference between ba-3 and pa-1, then the sucking rate
should revive to reflect renewed interest. And this is what Eimas et al. (1971)
found: infants hear the difference between /p/ and /b/. Critically, they also
discovered that infants were far less sensitive to the contrast within a
phonemic category. For example, infants who were habituated to ba-3
showed no revival in sucking rate when ba-2 was introduced. The two /b/
sounds were not distinguished. As noted above, ba-2 and pa-1 both differ
from ba-3 by 20 msec. But only pa-1 crosses the adult phoneme boundary.
Like adults, when VOT is greater than 25 msec, one-month-old infants hear
/pa/, when it is lower than 25 msec, they hear /ba/.
Box 5.1 Habituation Method
The habituation method provides psychologists with a window on
the mental life of infants. Patently, we cannot ask an infant about
what they perceive or know and we cannot subject them to
experiments where they have to say or do anything in response to
instructions. What we can do is bore them. In the habituation
method, a stimulus is presented to the infant repeatedly. When
interest in this first stimulus wanes, a new stimulus is introduced.
The question then is, can the infant detect the change of stimulus?
Infant attention can be measured in several different ways by
exploiting responses that are already in their behavioural
repertoire. These include looking time, head turning, heart rate
and the rate of sucking on a dummy. Looking time, in particular,
has long been a popular measure of infant interest (e.g., Fantz,
1961).
How can we tell when an infant has become habituated to a
stimulus? Csibra (2008) provides a typical example of how this is
decided. In Csibra’s study, the infant’s attention is first drawn to a
computer screen by a cartoon. Next, the habituation stimulus is
presented (in this case a video showing a box in motion), and the
computer begins timing infant looking. When the infant looks
away, an experimenter presses a key and the time of that first look
is recorded by the computer. Infants who play along with this
procedure will then alternate between looking back at the screen
for a while and then looking away again. On each occasion,
looking time can be measured (each ‘look’ is termed a trial). The
average looking time for the first three trials (T1) is calculated,
this being the point when infant interest is at its peak. As interest
declines, the infant spends less and less time looking at the
habituation stimulus. When the average looking time for three
subsequent, consecutive trials falls below 50 per cent of T1, then
the infant is said to be habituated. At this point, a test stimulus is
introduced that differs from the habituated stimulus. Is the infant
aware of the switch in stimuli? If looking time shoots up
significantly, then the infant is said to have dishabituated. The
 experimenter thus infers that the infant can distinguish the
habituated stimulus from the test stimulus.
If old habits die hard, then new ones die with ease. For this
reason, the habituation paradigm has proved to be enormously
productive in developmental research, having being used in
hundreds of studies. Recently, though, questions have been raised
about what prompts infants to dishabituate. One important factor
that is often overlooked is that infants vary in terms of the number
of trials it takes before they are habituated. For example, in Csibra
(2008), infants needed between 6 and 13 trials to reach
habituation. This is important because ‘fast’ and ‘slow’
habituators differ in their behaviour. For example, in Baillargeon
(1987), for infants aged 3–4 months, fast habituators showed
‘surprise’ at physically impossible events, but slow habituators
did not. Evidently, the basis for infant responding is not
straightforward (Schoner & Thelen, 2006). Differences in
susceptibility to habituation are compounded with the effects of
variation in stimulus display and experimental procedure. The
power of habituation as an experimental paradigm must therefore
be tempered with caution: we still need to explain why infants
dishabituate on a given occasion.
Since that early study, several other phonemic contrasts have been tested and
in almost every case, infant perception seems to match adult perception in
being categorical (Werker & Lalonde, 1988; Jusczyk, 1997). A sample of the
contrasts that have been tested is listed in Figure 5.3 below.
Figure 5.3 Studies demonstrating categorical perception in infants

For unfamiliar phoneme symbols, see the Pronunciation Guide at the

 end of the book.
Exercise 5.1: Phonological contrasts
You can discover the phonemes in a language by thinking about how
different sounds contrast to produce a difference in meaning. For
example, the two sequences, cat – pat are identical apart from the
initial sounds, /k/ and /p/. The contrast between /k/ and /p/ is sufficient
to change the meaning, so they are said to constitute a minimal pair.
Find the minimal pairs in the following set of words. Then consult the
chart of English phonemes at the back of the book and ‘translate’ the
English letters into phonemes (for example, the letter c in cat is written
phonemically as /k/). Remember to concentrate on how the words
sound, not on how they are spelled.
dog catch funny ceiling male
lake cheese ship bog chip
whale laze sham peas shack
been feeling cap honey dean
Another remarkable finding from this research is that very young infants are
sensitive to contrasts that do not feature in their own native language. For
example, the contrast between /ba/ and /pa/ is not exploited in Kikuyu, a
language spoken in Kenya. Nevertheless, one-month-old Kikuyu infants can
detect the voicing contrast between these two sounds (Streeter, 1976). In a
similar way, English–Canadian infants can distinguish contrasts found in
Polish, French and Czech (Trehub, 1976) as well as in Hindi and
Nthlakapmx, a North American language (Werker & Tees, 1984). And yes,
that is how you spell it. Just don’t ask me how to pronounce it (you’ll find out
why my Nthlakapmx pronunciation is so poor below). Overall, the large body
of research on infant speech perception yields two main conclusions: (1)
infant phoneme perception is categorical from the earliest time of testing,
suggesting that it is an inborn capacity; and (2) infants can discriminate
contrasts that appear in any of the world’s languages. For this reason, Kuhl
(2004: 833) suggests that infants are born ‘citizens of the world’, able to
acquire the phonemic contrasts of any language. Hence, the task of cutting up
the sea of sound into separate phoneme droplets is less bewildering than it
seems at first, because, to a large extent, our inborn perceptual mechanisms
are geared to do the job for us. The many different kinds of /p/ are grouped,
in perception, into a single category from the very start of life.
Specialization towards the native language
Why I don’t speak Nthlakapmx
We do not remain citizens of the world for very long. Between the ages of 6–
12 months there is a gradual specialization for the native language driven by
two complementary processes: both enhancement and decline in perceptual
sensitivity. On the one hand, infants tune in to their native language with
increasing precision and several months of exposure enhances sensitivity to
the particular features of the ambient language. On the other hand, there is a
decline in ability to detect contrasts in non-native languages. An early
demonstration of this decline is provided by Werker & Tees (1984), who
tested infants from an English-speaking background on contrasts found in
three languages: (1) English; (2) Hindi; and (3) Nthlakapmx, as shown below.
(Incidentally, you may sometimes see Nthlakapmx referred to as the more
pronounceable Salish, but this latter is actually a group of North American
Languages, of which Nthlakapmx is just one.) (See Table 5.1.)
Werker & Tees (1984) found that at 6–8 months, Canadian–English infants
could discriminate all three contrasts. From 8–10 months, though,
discrimination of non-native contrasts deteriorated sharply. By 12 months,
hardly any infants could distinguish the Hindi or Nthlakapmx contrasts. And
this, of course, is why my own Nthlakapmx pronunciation is so dismal. A
parallel case is provided in Japanese, in which /ɹ/ and /l/ are not contrastive.
This means, that to Japanese speakers, right and light sound like the same
word. Japanese infants seem to lose their ability to detect the /l/ – /ɹ/ contrast
by the age of 10–12 months (Tsushima, Takizawa, Sasaki, Nishi, Kohno,
Menyuk & Best, 1994; Kuhl, Stevens, Hayashi, Deguchi, Kiritani & Iverson,
2006). Recent evidence from neuroscience confirms these behavioural data.
Infant brain responses can be obtained in a non-invasive manner through the
use of event related potentials (ERP) (see Figure 10.2). In this technique,
electrodes placed on the scalp record electrical activity in the brain. ERP data
show changes in activity over time (they do not provide data on where
activity is located in the brain). It has been found that ERP responses to non-
native contrasts are present at 7 months, but have disappeared by 11 months
(Rivera-Gaxiola, Silva-Pereyra & Kuhl, 2005). In general, much research
now demonstrates that infant sensitivity to non-native contrasts declines from
about the age of six months onwards (for example, Werker & Lalonde, 1988;
Best & McRoberts, 1989, 2003).
Loss or decline?
Reading the literature, one can sometimes get the impression that infant
sensitivity to non-native contrasts ceases altogether by the end of the first
year. For example, Maye, Werker & Gerken (2002: B102) assert that ‘in
many cases … infants stop discriminating previously discriminable
contrasts’. More likely, though, we are talking about a decline, not a loss.
First, a small minority of 12-month-olds still perform well with non-native
contrasts (Werker & Tees, 1984). And some of those who show the more
usual decline still perform above chance (McMurray & Aslin, 2005; Kuhl et
al., 2006). We can also mention here children who move from their birth
culture at an early age to be adopted in a different country with a different
language – in one study, children from India aged 6–60 months (Singh,
Liederman, Mierzejewski & Barnes, 2011). Initial testing suggested that
these infants had lost their ability to distinguish contrasts from the birth
language. However, after training, performance improved significantly; a
group of control children, meanwhile, were immune to the training
intervention. In effect, native contrasts can be resuscitated, even if contact
with the birth language has been lost. Adults, too, can be trained to perceive
neglected contrasts. For example, Japanese adults can recover the ability to
distinguish /l/ and /ɹ/ (Flege, Takagi & Mann, 1995, 1996). And, depending
on the task they are asked to perform, adults can override categorical
perception (Schouten, Gerrits & van Hessen, 2003). Finally, in at least one
case, infant ability neither declines nor disappears, but is maintained over the
first year. English-learning infants can still distinguish between two click
sounds in Zulu, a South African language (Best, McRoberts & Sithole, 1988;
Best, McRoberts, Lafleur & Silverisenstadt, 1995). If you’ve ever heard click
sounds in languages, you will realize how distinctive they are and it is
perhaps this lack of similarity to other language sounds that helps explain
continued facility in discriminating among them.
Enhancement of native contrasts
Infant perception of unfamiliar, non-native languages may decline, but this is
compensated by an increase in facility with the native language over the first
year. Perception of native language is actively enhanced. For example,
Mandarin-learning infants get better at discriminating a contrast in their
language between so-called affricate and fricative sounds (Tsao, Liu &
Kuhl, 2006). (In English, the sound made by CH in chip is an affricate, while
SH in ship is a fricative). Enhancement of native contrasts is also necessary,
as another look at the /p – b/ contrast will reveal. Some languages do not
simply divide the phoneme pie into two neat slices, cutting along the /p – b/
border provided by our auditory systems. Thai, for example, divides up the
pie into three phoneme categories: /p/, /b/ and an aspirated /p/. We have
already seen that some /p/ sounds are aspirated, while others are not. Native
English speakers do not notice the difference, because aspiration is not
critical for distinguishing one word from another. But in Thai, aspiration is
contrastive. We can see how this works in Thai, by representing aspiration
with an /h/ (see Table 5.2).

The moral here is, if you’re visiting Thai friends and you want to borrow
their car, be careful what you ask for. You could end up driving off in their
dog. A further lesson is that the boundaries provided by our auditory system
are not set in stone. The infant must create categories that correspond
precisely with those in the native language (Maye, Weiss & Aslin, 2008).
Language-specific differences can be illustrated in a different way, with our
favourite /p/ – /b/ contrast. Once again, voice onset time is the critical factor.
We noted above that the /p/ – /b/ VOT boundary is 25 msec for some
varieties of English, but this is subverted in other cases, as shown in Table
5.3. It is through subtle acoustic distinctions of this kind that we identify
dialectal differences and speakers with foreign accents. In effect, our innate
auditory mechanisms do the really hard job of hacking through the /b/ – /p/
phonetic spectrum for us, dividing it into two manageable parts that
correspond roughly to what must be acquired. Then, through experience, this
initial boundary is finessed until it corresponds more closely with the precise
phonemic contrasts in the language being acquired. Even then, there is work
to do. Sounds differ in their salience for the child both with respect to their
voicing status and their position within a word. Word-final voiced stops like
/b, g/ are acquired with relative ease, while their voiceless counterparts, /p, k/,
are not acquired, either word-initially or word-finally, until the age of 20
months (Archer, Zamunerb, Engel, Fais & Curtin, 2016).
We need to consider why the infant’s perceptual abilities change over time.
Unfortunately, explanations are fairly thin on the ground. One possibility is
offered by Kuhl (2004) with her native language neural commitment (NLNC)
hypothesis. She suggests that ‘language learning produces dedicated neural
networks that code the patterns of native-language speech’ (ibid.: 838). Once
this neural commitment has been made, the potential of the brain to learn
language is directed towards the regularities of the native language and,
consequently, also interferes with the processing of non-native sound
patterns. Certainly, by the time we are adults, larger regions of the brain are
activated in response to the native language than to non-native input (Zhang,
Kuhl, Imada, Kotani & Tohkura, 2005). But we need equivalent evidence
from infants to provide strong confirmation of Kuhl’s hypothesis. Rivera-
Gaxiola et al. (2005) do show that infant ERP responses are stronger in some
infants for native versus non-native contrasts. The strength of ERP response
provides a measure of neural commitment, but, as mentioned, ERP data do
not provide information on precisely where something is happening in the
brain. So there is support for the idea that some neural resources somewhere
are committed to the native language, and also that some neural resources
somewhere are much less committed to non-native contrasts. But are they
one and the same neural resources? The obvious reading of the NLNC
hypothesis is that neural resources are used for the native language at the
expense of the non-native language. But some kind of imaging data would be
required to establish if this happens, since imaging data can tell us where
neural activity is located. We can see that the NLNC hypothesis is somewhat
vague about precisely what ‘neural commitment’ means. It may simply be
another way of saying that learning takes place.
 Sources: Caramazza & Yeni-Komshian, 1974; Christensen, 1984
In a similar way, Werker & Yeung (2005) talk about functional
reorganization of perceptual capacities as a result of increasingly selective
attention to the properties of the native language. The notion of functional
reorganization is reasonable enough as a label we can attach to describe what
happens in very broad terms. But it does not provide a satisfying explanation,
nor does it generate precise hypotheses that we might test about infant speech
perception. Broadly speaking, we can see that increasing commitment to the
native language is linked to decreasing ability with non-native speech
perception. These two developments are not independent. The fact that
infants become more attuned to the subtleties of their native language is not
surprising. The native language is pretty much all they hear, after all. But
why do we lose our capacity with non-native contrasts? The answer to this
question is much more open.
Individual differences in infant speech perception
Intriguingly, evidence is beginning to emerge that infants differ in their
perceptual abilities. By seven months, there is a negative correlation between
performance on native versus non-native contrasts. This means that infants
who do well on native contrasts simultaneously do less well on non-native
contrasts. It seems, then, that some children tune in to their native language
with relative ease, while at the same time being less fazed by non-native
distractions. Of great interest, research is beginning to show that these ‘easy
tuners’ have an advantage when it comes to later language learning (Tsao,
Liu & Kuhl, 2004; Newman, Ratner, Jusczyk, Jusczyk & Dow, 2006; Kuhl,
Conboy, Coffey-Corina, Padden, Rivera-Gaxiola & Nelson, 2008). Better
native phonetic perception is associated with accelerated growth on both
vocabulary and grammar more than two years later, at the age of 30 months
(Kuhl, Conboy, Padden, Nelson & Pruitt, 2005). It seems somewhat ironic
that those children who remain open to non-native contrasts are at a
disadvantage for learning their own language. But this ‘openness’ might
better be seen as a deficit that prevents maximal attunement to the ambient
language. One further source of individual variation has been identified
which stems, not from the infant, but from the linguistic environment. It has
been found that infants tune in to their native language more quickly (at six
months) if their mothers are especially responsive during interactions
(Elsabbagh, Hohenberger, Campos, Van Herwegen, Serres, De Schonen,
Aschersleben, & Karmiloff-Smith, 2013). In particular, these mothers
produced higher levels of mutual gaze, turn-taking and mutual affect and
altered their behaviour in direct response to the infants’ behaviour. It would
seem that the infant is encouraged by this kind of high quality input to tune in
more quickly to their native language.
Summary: Breaking the speech sound barrier
So far, we have seen that the daunting task of identifying individual
phonemes is made manageable by an innate capacity to perceive speech
sounds categorically. We saw that any given phoneme has many different
physical realizations: changes in gender, age and mood of speaker are
compounded by changes according to where a phoneme occurs in a syllable
and also by the sounds it is surrounded by. But we are born able to ignore
these differences and perceive different phonetic realizations of a sound as a
single phoneme. We also saw that over the first year of life, two related
processes take place. On the one hand, the child tunes into the native
language, becoming increasingly adept at identifying the precise phoneme
categories of the language being learned. On the other hand, an initial ability
to identify phoneme contrasts from any of the world’s languages fades away
from about the sixth month onwards. In the next section, we will see that the
infant is not content to identify the sounds of the ambient language. In this
regard, Jusczyk comments that:
it seems pretty clear that the basic speech perception capacities that
infants possess should prove useful in learning how the sound structure
of the native language is organized. What is less obvious is that these
same perceptual capacities could play a role in acquiring information
about other levels of language structure. (Jusczyk, 1997: 137)
We will turn our attention now to these other levels of language structure and
see, first, how infants come to identify individual words in the stream of
speech and, second, how they can even identify information about grammar.
All before their first birthday, when they break another sound barrier by
producing a word of their own.
Word segmentation
As you read this sentence, the white spaces at the beginning and end of each
word perform a great service: they segment the words for you. To see what a
great help this is, try reading the following (taken from Kuhl, 2004):
theredonateakettleoftenchips
This exercise is not that easy, even though you have two massive advantages
over the infant: (1) you know what a word is already; and (2) you know many
thousands of specific words that you can look out for. Even so, this particular
example is rather tricky, because it can be parsed (cut up) in two different
(admittedly rather nonsensical) ways:
the red on a tea kettle often chips
there, Don ate a kettle of ten chips
The spaces between printed words make life easy for the reader. But listeners
to spoken language have no such advantage, as we noted earlier, because
there are no clear gaps demarcating the end of one word and the start of
another. The sounds run into each other. We might also note that the
segmentation problem is compounded by the fact that human speech is very
fast. An average speaker produces something like 200 syllables per minute
(Aitchison, 1998). Even allowing for the fact that adults speak relatively
slowly to young children (Broen, 1972), the child must therefore process
speech at a phenomenal rate.
Despite the complexity of this task, there is evidence that infants can segment
words from fluent speech by the age of seven months (Jusczyk & Aslin,
1995). How do they do this? Researchers have examined two broad
categories of cue in the speech signal that might help the infant:
transitional probabilities: statistical information on the probability that
one syllable will follow another
prosodic cues: information from the ‘music’ of speech
In the following sections, we will look at each one of these potential cues to
word boundaries. Before we do so, though, have a go at segmentation
yourself with the following exercise.
Exercise 5.2: Where are the words?
At the start of this chapter, I mentioned the bamboozling experience of
eavesdropping idly on a foreign language speaker. But what if we
 listen carefully? Some research shows that adults can, in fact, detect
words in an unfamiliar language (Saffran, Newport & Aslin, 1996),
although the first language seems to interfere (Finn & Kam, 2008). Try
this out for yourself. If you have a bilingual speaker in your group,
record them speaking for no more than two minutes, uninterrupted,
about their friends, hobbies, or anything that takes their fancy. Listen
very carefully to the recording. Try to write down any words you hear.
Afterwards, ask your bilingual informant to check how well you did.
You can get some sense of the task facing the infant from this exercise,
but don’t forget that you started out with one major advantage: you
know what a word is, so you know what you’re looking for.
The baby statistician
We will first introduce the notion of uncertainty in the input. There is a lot of
information in the input that could guide infants in the task of segmentation.
But none of this information is 100 per cent reliable (Jusczyk, 1997). This
lack of certainty introduces the idea that language learning may involve
probabilistic inferences on the part of the child. Box 5.2 discusses probability
theory and hints at the massive impact it is beginning to have on theories of
child language development. For one thing, probabilistic inference presents a
theory of learning that stands in direct conflict with the nativist idea of inborn
knowledge. In fact, the statistical properties of language were dismissed by
Chomsky (1957) and have taken almost half a century to emerge once more
as an important influence on the child. Chomsky (1957: 15) based his
argument on a now famous example:
Colourless green ideas sleep furiously
Before Chomsky constructed this sentence, the probability that these
particular words had ever been placed in this particular order was (let’s lay
odds) zero. And yet the sentence is perfectly grammatical, even if it is
somewhat difficult to make sense of. This suggests that our analysis of the
sentence cannot depend on the probabilities with which words group together
(Seidenberg, MacDonald & Saffran, 2002: 553). The low chance of this
sentence occurring does not impede our recognition of a grammatical
sentence.
Box 5.2 What are the Chances? The rise of probability theory
The fundamental assumption in probability theory is that learning
is an uncertain process. In cognitive science, generally, the brain
is viewed as an information processor. Learning takes place when
we infer something new, on the basis of what we already know.
Given information (something we already know) might come to
us from the senses or be drawn from memory. In both cases, we
can use this ‘old’ information to infer something new. We can
make either deductive or inductive inferences.
In a deductive inference, the outcome is certain and incontestably
true (because the premises are true).

Premises: A is bigger than B

 B is bigger than C
Deductive inference: A is bigger than C
In fact, we make relatively few deductive inferences in everyday
life. Far more common are inductive inferences, in which we go
beyond the premises to arrive at a conclusion that is not
absolutely certain:
Premise: All the tigers I have ever seen (or heard about) have
orange fur and dark stripes.
Inductive inference: The next tiger I see will have orange fur
and dark stripes.
The probability that my inference will match reality is high in this
case, but it is not 100 per cent. The next tiger I come across might
have white fur and dark stripes. Although rare, white tigers do
exist. So my inductive inference could be proven wrong.
Probability theory provides a method for capturing uncertain
inference. It is based on the idea that events have a given
probability, or chance, of happening. For example, there is a 50
per cent chance of a tossed coin landing heads up (usually written
as a probability of 0.5). Numerous domains of learning and
knowledge can be modelled by the laws of probability, including
vision, motor control, causal learning, language at every level
(phonological, lexical, morphological, syntactic), in addition to
both speech perception and speech production. In the field of
 language acquisition, probability theory provides a serious (albeit
recent) alternative to Chomskyan theory. For Chomsky, a system
of rules specifies all and only the ‘legal’ sentences in a language.
These rules are held to determine what is learned about language
and how that knowledge is represented in the mind. Chomsky’s
black-and-white approach to grammaticality stands in sharp
contrast with probability theory. Grammaticality can be viewed as
a shaded concept, a matter of degree rather than absolute right or
wrong. Two people may differ on whether they accept a particular
form as grammatical (for example, shined as the past tense of
shine: Tom shined his torch in my face). The amount of
experience different people have with particular language forms
may affect the chances of them being accepted as grammatical.
The rise of probability theory follows in the wake of huge
advances in computational technology. Speech recognition,
automatic text analysis and machine translation have been made
possible through the development of powerful computational
systems. Quite simply, probability theory has been harnessed in
this research because it works in refining machine capabilities.
Almost incidentally, the fallout for cognitive science has been
profound. Psychologists interested in how the mind works now
have a powerful set of computational and theoretical tools for
modelling its functional capacity (Chater & Manning, 2006;
Chater, Tenenbaum & Yuille, 2006).
In fact, it has emerged that there are several probabilistic (or statistical)
properties of language that the child might make use of in the segmentation
of words. For example, in harmonic languages like Turkish and Hungarian,
the vowels within a word tend to be either all front vowels or all back vowels.
This means that the narrowest point in the airway, where the tongue almost
blocks off the air flow, is either at the front of the mouth or the back.
Statistically, two vowels of the same kind (front or back) will more likely
occur within a word, rather than across a word boundary (Ketrez, 2014).
There is a probabilistic cue, therefore, to identify where one word stops and
the next one begins. But can the child actually use cues of this kind to
segment words? Let’s take another example, offered by Saffran, Newport &
Aslin (1996): the simple phrase pretty baby. What are the chances of one
syllable following another? Taking samples of Infant Directed Speech, we
can calculate that, within the word pretty, the chance of pre- being followed
by -tty is very high, something like 80 per cent. This is why it is difficult to
think of another syllable that can follow pre- (pronounced as PRI-). Try it for
yourself. If we now consider the probability of -tty being followed by ba-, as
it does in pretty baby, the chance is much lower, something like 0.03 per
cent. This is because there are many other possible syllables that can follow -
tty: pretty hat, pretty flower, pretty perambulator, and so on. The odds that
one particular syllable will be followed by another are often referred to as
transitional probabilities. In our example, we can see that within the word
pretty, the transitional probability from one syllable to the next is high, but
across words (from -tty to ba-), the transitional probability is low. In essence,
syllable sequences that go together frequently have a relatively high chance
of being words. Can the infant compute transitional probabilities? If so, they
might use low transitional probabilities as a good (if not perfect) cue for
identifying the boundary between one word and the next.
Saffran, Newport & Aslin (1996) tested this possibility by teaching eight-
month-old infants a miniature artificial language. They generated four three-
syllable words, using a speech synthesizer. In one condition these words
were:
tupiro
golabu
bidaku
padoti
These words have been carefully designed so that transitional probabilities
within words are 1.0 (that is 100 per cent or totally predictable), while
between words, transitional probabilities were only 0.33. Thus, within words
of this language, pi is always followed by ro (in the word tupiro), that’s to
say, a transitional probability of 1.0. Across words, ro can be followed by go,
bi or pa (the first part of the other words in the language). Since these three
words follow ro an equal number of times, the transitional probability is 0.33
(a one in three chance). The words in this language were produced in a
monotone female voice with no pauses between words and no other cues to
word boundaries like stress or prosody (see below). Try saying the following
eight words at a normal speed in a flat voice with no hesitations or pauses:
padotigolabupadotibidakutupirotupirogolabubidaku
There isn’t much to learn from this little exercise, though we do gain some
idea of what life is like as a speech synthesizer. And it passes the time.
Infants listened to an unbroken string of these four novel words, played in
random order for two minutes, at the rate of 270 syllables per minute.
Following this familiarization phase, infants were exposed to four words
repeated continuously, two from the original set (tupiro and golabu), and also
two new words, composed from the same original set of syllables, but in a
scrambled order (dapiku and tilado). It was found that infants listened
significantly longer to the new words. There is evidence, then, that ‘infants
succeeded in learning particular groupings of three-syllable strings – those
strings containing higher transitional probabilities surrounded by lower
transitional probabilities’ (Saffran, Aslin & Newport, 1996: 1928). They
succeeded in finding words after only two minutes of listening.
One hiccup is that infants in these experiments come to rely on the fact that
all the words they hear have the same number of syllables. In Saffran, Aslin
& Newport (1996) all of the words have three syllables and, as we have seen,
learning runs smoothly. But if infants are trained on trisyllabic words, but
then switch to hear words of just two syllables (or vice versa), they are
thrown (Lew-Williams & Saffran, 2012). Word segmentation by the infant is
much less impressive. In real life, of course, children are exposed to Syllable
Salad, where the words vary much more haphazardly in terms of syllable
number. There is a disjunction, then, between the laboratory and the real-life
conditions faced by the child. We pursue the issue of ecological validity
further in the next section. In the meantime, we can consider the relationship
between segmentation ability and later word learning. It has been found that
infants vary in how well they do on Saffran, Newport & Aslin’s task.
Moreover, those infants who are proficient at segmentation when they are 7.5
months old turn out to have larger productive vocabularies at 24 months
(Singh, Reznick & Xuehua, 2012; Newman, Rowe & Ratner, 2015).
The power of the infant to exploit statistical properties in the input is
beginning to transform our view of child language learning (Romberg &
Saffran, 2010). Research is by no means confined to the problem of word
segmentation. The research on categorical perception has also been
interpreted in this way, since the child is mapping the probabilities that
different phonetic units belong within the same phonemic category, on the
basis of experience (Werker & Yeung, 2005). And we will see in Chapter 6
that statistical learning is beginning to be applied in the domain of word
learning. Another example is provided in the domain of grammatical rule
learning (Marcus, Vijayan, Rao & Vishton, 1999). Using similar methods to
Saffran, Newport & Aslin (1996), Marcus et al. created two artificial
languages, each one with a very simple rule for putting three syllables
together in a particular order:
ABAga ti ga
ABBga ti ti
Infants were familiarized with one of these two grammars and then, in the test
phase, they were exposed to three-syllable sequences containing entirely new
syllables. Thus, the infant who has been familiarized on the ABA sequence
with ga ti ga, would then hear novel syllable strings which were either
consistent with the ABA pattern (wo fe wo) or inconsistent (wo fe fe). The
idea is to see if the child can detect the pattern they have been trained on
(ABA versus ABB). This is why new syllables were used in the test phase
(wo and fe, rather than ga and ti). Marcus et al. (1999) found that seven-
month-old infants had a novelty preference for the inconsistent sequences:
they preferred to listen to wo fe fe (ABB) if they had been trained on ga ti ga
(ABA) (see also Gervain & Werker, 2013). These researchers argue that
infants have learned rules, not just particular strings of syllables, because the
test stimuli were entirely new. However, there is some doubt about whether a
grammatical rule (however simple) has been learned (Seidenberg et al.,
2002). There are statistical regularities in the stimuli used by Marcus that are
similar to the kind found in Saffran, Newport & Aslin’s (1996) word
segmentation study. And the ability to generalize beyond the input to new
stimuli is possible both with statistical learning and rule learning.
Studies on statistical learning demonstrate that infants possess very powerful
mechanisms for processing and analysing rapid auditory information.
Subsequent studies suggest that this ability is not confined to speech, but
applies equally well to rapid sequences of auditory tones and even to visual
stimuli (Saffran, Johnson, Aslin & Newport, 1999; Fiser & Aslin, 2001). This
brings us back, momentarily, to the debate on whether or not ‘speech is
special’. Cross-domain segmentation abilities suggest that speech does not
occupy a privileged position in human cognition. And work on cotton-top
tamarins (a kind of monkey) supports this position, since they too can learn
from the statistical regularities in speech (Saffran et al., 2008). The ability to
perform these kinds of analyses both across species and across different kinds
of stimuli suggests a domain-general learning capacity for language (Saffran
& Thiessen, 2007).
Learning in the real world
Before we move on from our baby statistician, it is worth pointing out that, in
Saffran et al.’s (1996) study, the transitional probabilities within words were
fixed in advance at 1.0. But the vast majority of real words in real languages
do not enjoy this perfect predictability. Take, for example, the syllable per- in
words like perceive, perhaps, personality, persuade, and so on. The syllables
after per- are all different, so the chance of correctly predicting what follows
plummets accordingly. When transitional probabilities are high, infants can
not only segment speech, they can also begin to learn about the meanings of
the words identified. In contrast, word learning is relatively poor when the
infant has to cope with low transitional probabilities (Hay, Pelucchi, Graf
Estes & Saffran, 2011). One other point to note is that, in real life, infants are
not exposed to the same four words repeated over and over again. It is
impressive that learning can take place after just two minutes of this kind of
exposure. But the intensity of the experience lowers its ecological validity.
One recent study tackled this problem by using words that the child had
already learned in isolation (Depaolis, Vihman & Keren-Portnoy, 2014). It
was found that infants only became adept at segmenting these words from
fluent speech at the age of 12 months. It would seem, then, that the real world
places greater demands on the infant’s ability to process and remember words
than the laboratory.
Ecological validity was improved in a study that looked at the influence of
Infant Directed Speech on word segmentation. But instead of using
synthesized stimuli, real speakers were recorded producing either Infant
Directed Speech (IDS) or Adult Directed Speech (ADS) (Thiessen, Hill &
Saffran, 2005). You will recall from Chapter 4 that IDS has a very distinctive
quality, especially notable for its high pitch and exaggerated, swooping pitch
contours. Thiessen et al. (2005) found that infants could find the word
boundaries when listening to Infant Directed Speech, but not from Adult
Directed Speech. This finding has been replicated more recently (Floccia,
Keren-Portnoy, Depaolis, Duffy, Delle Luche, Durrant, White, Goslin &
Vihman, 2016). Out of 13 studies, word segmentation was witnessed in just
one study, and this was distinguished by the use of an exaggerated form of
Infant Directed Speech. These findings might be explained by reference to
the power of IDS to grab the infant’s attention. Infants may learn more when
IDS is used because they are focusing especially closely on the experimental
stimuli. Not surprisingly, there is evidence that enhanced attention improves
learning and memory (e.g., Rensink, O’Regan & Clark, 2000). Another study
can be mentioned which is also grounded more firmly in the real world (Lew-
Williams, Pelucchi & Saffran, 2011). Eight-month-old children learning
English listened to Italian words, either in running speech or in isolation.
These infants segmented the speech and identified Italian word boundaries
more effectively when they heard the words in both contexts, as in normal
Child Directed Speech. Exposure to fluent speech alone was not sufficient to
help these children identify its statistical properties. Learning was therefore
most effective when the situation more closely reflected real life. In a similar
vein, it has been found that infants are not fazed by a switch from one
speaker to another when segmenting words from running speech (Graf Estes,
2012). They can therefore go beyond the artificial, synthesized speech of the
laboratory and generalize their learning in ecologically valid settings.
Transitional probabilities are clearly a powerful cue that help the infant find
word boundaries. But there are other sources of information in the input that
the infant might exploit. We next turn our attention to prosodic factors in the
input, to consider if they, too, are helpful.
Prosodic cues to speech segmentation
In running speech, some syllables seem to stand out, because they are
produced with more energy. In the linguistic parlance, they are stressed. The
perception of stress is caused by three factors: amplitude (louder syllables are
perceived as stressed); duration (longer syllables sound stressed); and an
increase in pitch. Of these three cues, an increase in pitch provides the
strongest cue that a given syllable is stressed (Fry, 1955). Changes in pitch
contribute to the prosody (or ‘music’) of speech. In normal running speech,
the pitch changes continuously, rising and falling, to create a kind of melody.
In the process, they contribute to our impression that some syllables are
stronger than others. In continuous speech, a pattern of alternating strong and
weak syllables can be discerned, as in the following lines from poems (strong
syllables have been highlighted).
And did those feet in ancient time … (Blake)
Stop all the clocks, cut off the telephone … (Auden)
I sing the body electric … (Whitman)
Do not go gentle into that good night … (Thomas)
The words in a given language have a characteristic stress pattern. English
words tend to have a strong–weak pattern (known as trochaic), while other
languages, like Polish, prefer an iambic, weak–strong pattern. But all
languages have exceptions to their general patterns. For example, in English,
about 90 per cent of all multisyllabic words do begin with a strong syllable:
water bicycle envelope
taxi pineapple garden
Nevertheless, a significant minority of words buck this trend, as in the
following:
indigestion guitar religion
compare resolve explore
The prosodic qualities of language are therefore predictable, but not perfectly
predictable. They are predictable in a probabilistic fashion. There is a
parallel, then, with the transitional probabilities of syllables discussed above.
The question then becomes: Are infants sensitive to the prosodic qualities of
their native language? The short answer is yes. We saw in Chapter 1 that the
foetus can recognize a piece of poetry from its prosodic properties. Beyond
The Cat in the Hat studies, we also know that, prior to birth, the foetus has a
preference for listening to their own mother’s voice (Kisilevsky, Hains, Lee,
Xie, Huang, Ye, Zhang & Wang, 2003). This means that the foetus can learn
and recognize (at least some of) the unique qualities in a particular speaker’s
voice. And shortly after birth, newborns can distinguish their native language
from a foreign language (Moon, Cooper & Fifer, 1993; Nazzi, Bertoncini &
Mehler, 1998). By five months, they can even distinguish one dialect of
English from another (Nazzi, Jusczyk & Johnson, 2000). In each case, the
infant makes use of prosodic qualities to distinguish one form of speech from
another.
But what about the task of segmenting words from the speech stream? Word
stress provides a useful cue to the boundaries of words. In English, it would
be a good (if not perfect) strategy to assume that each stressed syllable marks
the beginning of a word. Between six and nine months, infants do show
sensitivity to the stress pattern of their native language (Jusczyk, Houston &
Newsome, 1999). In fact, the stress pattern of the language biases the infant
in its segmentation behaviour. They can segment words like kingdom and
hamlet, that begin with a stressed syllable, but not words like beret and
device (Jusczyk, 2002). (If you’re wondering about beret, this was an
American experiment and they follow the French pronunciation (beret) –
personally, I emphasize the first syllable.) These studies show that infants
who hear a phrase like ‘guitar is’ assume that taris is a word: they assume
that the stressed syllable –tar marks the start of a word. The infant is thus
guided by the predominant trochaic stress pattern of English. Other studies
have also demonstrated infant ability to exploit stress cues. For example,
Saffran & Thiessen (2003) show that infants can segment words from an
iambic language, even when only 80 per cent of the words presented
followed the iambic stress pattern. Generally speaking, it is a good strategy to
cut up the speech stream according to the position of stressed syllables. The
chunks of phonetic information this process yields very often correspond to
words and may help focus attention on other cues that can help refine the
segmentation process.
One extra segmentation cue is the edge of an utterance. Utterances have a
characteristic intonation contour, something like a musical envelope that
contains the phonetic information. In particular, there is a coherent pitch
contour laid over the whole utterance, an overall pattern of rising and falling.
And utterance boundaries generally occur at the point where we take a breath,
so pauses are more likely. Not surprisingly, therefore, words placed at the
beginning or end of an utterance are more salient (stand out more) than words
buried somewhere in the middle (Fougeron & Keating, 1997). Given that
Infant Directed Speech is replete with short utterances, infants are exposed to
especially high numbers of utterance edges. It turns out that utterance edges
are especially good places to find words: eight-month-olds do better at
segmenting words from both the beginning and end of utterances than they
do from the centre (Seidl & Johnson, 2006).
Most of the word segmentation research discussed here deals with words that
begin with a consonant. Other research has shown that infants find vowel-
initial words especially difficult to segment (Nazzi, Dilley, Jusczyk,
Shattuck-Hufnagel & Jusczyk, 2005). In English, they also find verbs
difficult compared to nouns, even when these start with a consonant (Nazzi
et al., 2005). One reason for this may be that verbs tend to occur in the centre
of utterances, while nouns often begin and end utterances (for example, Mary
had a little lamb: Mary and lamb are nouns, while the verb had is tucked
away in the middle). It is important to remember, therefore, that word
segmentation is not a monolithic process. Certain kinds of words are easier to
locate in the speech stream than others. A 7.5-month-old can detect, with
relative ease, nouns at the edge of an utterance that begin with a consonant
(Jusczyk & Aslin, 1995). But it may not be until 16.5 months that verbs
beginning with a vowel are segmented successfully (Nazzi et al., 2005).
Unless, that is, the vowel-initial word (e.g., eye, apple) is placed at the edge
of an utterance. In that case, infants look more precocious, since 11-month-
olds can manage the task (Seidl & Johnson, 2008). Overall, we can see that
several factors enter into the task of word segmentation, including transitional
probabilities, phonological structure, prosodic structure, word class (noun or
verb) and utterance position (edge or centre).
Relative cue strength
Do segmentation cues vary in strength? Johnson & Jusczyk (2001) addressed
this question by pitting prosodic cues against statistical ones in an artificial
language taught to eight-month-olds. Syllables that ended words by statistical
rules were given word-initial stress cues (they were high-pitched, louder and
longer). If the infant did statistics, they would have found word endings. If
they were more musical (tuned into prosody) they would find word
beginnings. At eight months, it turns out that prosody wins. However,
Thiessen & Saffran (2003) replicated this study, using infants aged seven and
nine months. They found that, whereas nine-month-olds preferred stress cues,
seven-month-olds did better with statistical cues. There is, then, a
developmental progression. Happily, of course, these cues do not conflict
with one another in real languages. Evidence from adult artificial language
learning confirms this view, since the addition of prosodic cues enhances the
learning adults can do from transitional probabilities alone (Saffran, Newport
& Aslin, 1996). This confirms what any psychology student already knows:
statistics and stress go together.
Another test of relative cue strength is provided by Thiessen & Saffran
(2004). You will recall that our perception of stress is caused by increases in
three factors: pitch; duration; and loudness. Usually, these three factors co-
occur, but Thiessen & Saffran teased them apart and taught infants a
language where the only cue to stress was a change in loudness (or
amplitude). In fact, the change is even more subtle than that, since only the
amplitude of sounds with a relatively high pitch were altered (changes to the
so-called spectral tilt). For adults, spectral tilt is the weakest index of stress,
but nine-month-olds performed well at segmenting the speech stream with
only these subtle changes in loudness to help them.
By 12 months, however, infants perform much more like adults, and changes
in spectral tilt are much less powerful as a cue to word boundaries. We have
another demonstration, therefore, that word boundary cues do differ in
strength, but importantly, their strength for the infant varies across the course
of development. And the changes take place rapidly. From 9 to 12 months,
infant perception shifts so that the speech signal is processed in an integrated,
adult-like fashion, being less susceptible to subtle changes in just one factor.
We have seen that the infant’s segmentation abilities are impressive, but it is
worth asking what kind of words the infant fishes out of the speech stream.
Are they just strings of sound with no meaning? If so, the infant would have
discovered words in only the baldest sense. In fact, though, research is
beginning to show that information on transitional probabilities can enhance
the naming of words. In one study, 17-month-olds took part in a statistical
word segmentation task of the kind pioneered by Saffran (Estes, Evans,
Alibali & Saffran, 2007). In a subsequent task, these words were used to
name objects. Exposure to the words in the statistical learning task enhanced
performance in the object labelling task that followed. Hence, the statistical
task provides the child with information on what constitutes a ‘good word’ in
the language being learned. And this information makes it easier to map from
sound to meaning in the labelling task.
Discussion point 5.1: Infants and other animals
In Chapter 2 we reviewed evidence on animal abilities to perceive
and process human speech. Refer back to this evidence (especially
the section, Is speech special?). Then compare infant and animal
learning:
What are the similarities between animal perception and
infant perception? Consider in particular the perception of
phonemes and the statistical learning of word-like units.
In what ways do infants and non-humans differ? Consider
not only the outcomes of learning, but also the way in which
learning takes place.
If tamarin monkeys can process speech in a similar way to
human infants, what do you think prevents them from going
on to acquire the full complexities of language?
Grammar from the babble
By this stage, we might ask if there are any limits to the child’s remarkable
segmentation abilities. Phonemes and words come tumbling out of the river
of sound with apparent ease. But grammar might be more elusive. We
reviewed claims by Marcus et al. (1999; 2007) that infants can use input
regularities to learn rules. The rules were very simple, namely, that a sentence
must comprise just two kinds of element in simple strings of three (ABB or
ABA). You will recall that the jury is still out on whether infants are learning
grammatical rules or simply learning statistical regularities. But there is
stronger support for a less ambitious claim about grammar learning: infants
can use prosodic information to mark off one clause from another (Hirsh-
Pasek, Kemler Nelson, Jusczyk, Cassidy, Druss & Kennedy, 1987). You can
think of a clause as a ‘mini-sentence’, with each sentence comprising one or
more clauses (see the Glossary for a slightly more refined description). Of
course, infant ability to demarcate clauses depends on being able to detect
clause boundaries. But there are no perfectly predictable prosodic markers of
this kind. Instead, there are at least three different cues, all of which are
associated with the edges of clauses. Clause boundaries in English are often
(if not always) marked by pauses, syllable lengthening and a change in pitch
(Price, Ostendorf, Shattuck-Hufnagel & Fong, 1991). Of interest, there is a
parallel here between speech and music. These same three cues often mark
the end of musical phrases also (Jusczyk, 1997).
Hirsh-Pasek et al. (1987) tested infant sensitivity to clause boundaries. They
prepared two sets of stimuli from a single sample of Infant Directed Speech
(IDS). In one version, pauses were inserted at clause boundaries, happily
coinciding with the other two cues: pitch change and a drawn out final
syllable. This coincident version was pitted against a non-coincident version,
in which the pauses were inserted between words within the clause. It was
found that infants as young as seven months preferred to listen to the
coincident versions, that is, natural sounding clauses. No such preference was
found in a subsequent study that used Adult Directed Speech (Kemler
Nelson, Hirsh-Pasek, Jusczyk & Cassidy, 1989). By seven months, therefore,
infants are tuned in to the unique, exaggerated prosody of IDS. And they use
this special input to help locate groups of words (clauses) critical in the
acquisition of grammar. American nine-month-olds still prefer coincident
speech samples when exposed to an unfamiliar dialect from Britain (Polka,
Jusczyk & Ravchev, 1995). And younger American infants (4.5 months)
prefer coincident versions not only in English, but in Polish also (Jusczyk,
1989). However, the penchant for Polish clauses has gone by the age of six
months, a finding that is in accord with the research described above showing
a loss of non-native sensitivity in the latter half of the first year.
If your ultimate goal is to learn a system of grammatical rules, it is a good
start to be able to fish out a chunk of language from the input that
corresponds to a chunk of grammar. But the chunk of grammar that we have
considered (the clause) is quite a mouthful. Clauses can be any length and
vary considerably in their syntactic complexity. Clearly, there is much
analysis to do within the unit of the clause. The ability, discussed above, to
isolate individual words must be very helpful in this respect. And, one might
also predict, a sensitivity to word order would also be very useful. The rules
of grammar determine the order in which words are placed in a sentence.
Compare the grammatical When I am an old woman, I shall wear purple
(from the poem by Jenny Joseph), with the ungrammatical I I purple when
woman shall am an wear old. Word order matters, especially in languages
like English. Remarkably, infants are sensitive to changes of word order,
provided that the changes take place within the prosodic envelope provided
by a clause (Mandel, Kemler Nelson & Jusczyk, 1996; Soderstrom, Blossom,
Foygel & Morgan, 2008). Compare the following sentences, bearing in mind
that the words would and wood are pronounced identically, and are therefore
indistinguishable to an infant.
1. Cats would jump benches. → Cats jump wood benches.
2. Cats would. Jump benches. → Cats jump. Wood benches.
The examples in (a) comprise single, whole clauses, pronounced naturally as
a single prosodic unit. Observe how would/wood and jump have been
switched around to the right of the arrow. The examples in (b), meanwhile,
have been constructed from two fragments, with each fragment sounding
natural by itself. Placed together, though, the two fragments do not sound like
a single unit of speech. Try them for yourself, paying attention to rises and
falls in pitch. Mandel et al. (1996) found that two-month-old infants could
detect the change in word order, but only when presented within the single
prosodic unit, as in (a). The change in word order in (b) takes place across a
prosodic boundary, and was not detected by two-month-olds. It seems that
their analyses take place within the prosodic unit. And as we saw above, the
prosodic unit very often corresponds to the grammatical unit of the clause.
Phonemes, words and grammar: Summary
We have seen that the infant does a remarkable job in perceiving words, and
even clauses, in the first year of life. And of course, they achieve this at the
same time that they are hunting down the phonemes of the native language.
In essence, the infant transforms the flood of sound – with its lack of obvious
beginnings and endings – into sequences of language units. Phonemes, words
and phrases all exist as distinct units of linguistic analysis, so we can
reinvoke the image, cast aside at the start, of beads on a string. The infant has
to identify the separate beads and begin to learn how they are strung together.
We have reviewed evidence that the infant is well equipped for this herculean
task. The (probably innate) ability to perceive phonetic units categorically
provides a massive kick-start to the process of phoneme acquisition. And
infants are also well equipped to detect regularities in the music of speech –
its prosody – that correspond, to a significant degree, with linguistic units, in
particular words and clauses. Infants also possess a very impressive ability to
process strings of syllables. They can identify, learn and recall the order in
which syllables occur in sequences they hear. And they can compute
regularities in syllable order across many different strings of syllables. In
consequence, the boundary between one word and the next can be predicted,
in a probabilistic manner, by working out the likelihood that one syllable
follows another. If the chance of this transition is very low, it is likely that
one has reached a cut-off point between one word and the next. In the next
chapter, we will examine the child’s progress in word learning from the
second year of life onwards. It is one thing to identify that a certain string of
syllables constitutes a distinct unit. It is another thing to know that this unit is
a word with a meaning and a particular place in the grammar of the language.
In a Nutshell
Speech cannot be divided easily into separate units of sound
or words. Phonemes overlap and merge into one another in a
continuous fashion. And there are often no obvious breaks
between one word and the next.
Each language makes use of a small subset of the 800 or so
phonetic units that are found in the world’s languages.
The great task for the infant is to penetrate this babble and
identify phonemes, words and even units of grammar in the
ambient language.
Infants perceive sounds categorically: physically different
kinds of /p/ are perceived as a single sound.
Commitment to the sounds of the native language increases
during the first year, while the ability to detect non-native
contrasts declines.
Infants of seven months can observe, learn and remember
statistical regularities in sequences of syllables. They are
sensitive to the chances that one syllable will follow another
(transitional probabilities). In consequence, infants can make
a good ‘guess’ about where one word stops and the next one
 begins in running speech.
Infants can also use prosodic information (especially word
stress) to identify where one word stops and the next one
begins. Like statistical learning, prosodic information is
probabilistic: a stressed syllable in English has a good, but
not perfect, chance of signifying the start of a word.
Infants use multiple cues to identify word boundaries in a
probabilistic fashion. None of these cues is reliable by itself.
But working in concert, the information in the input is
enhanced.
Infants can even use prosodic information to identify large
chunks of grammar (the clause) and are sensitive to changes
in word order taking place within a prosodic unit.
Further Reading
Jusczyk, P.W. (1997). The discovery of spoken language.
Cambridge, MA: MIT Press.
This book is not bang up to date, but it still provides a very clear
survey of research on early infant perception, with a wealth of
material that I could not include here.
Vihman, M.M. (2014). Phonological development: The first two
years (2nd ed.). Malden, MA: Wiley-Blackwell.
The present chapter has been concerned with infant speech
perception, rather than production. This reference fills the gap
with a clear review of how children learn to get their tongues
round the sounds of their native language.
Websites
International Phonetic Association (IPA):
www.internationalphoneticassociation.org/
The chart at the end of the book shows you the
phonemes used in English, but the IPA pages host much
more information about phonetics, including a link to
the International Phonetic Alphabet that shows you all
the symbols used to represent the sounds of the world’s
languages. There is also information on the special fonts
needed to produce the weird squiggles and curlicues for
yourself. Google: International Phonetic Association.
 Jenny Saffran: www.waisman.wisc.edu/pi-Saffran-
Jenny.htm
Google: Jenny Saffran to find her academic homepage
with a link to useful publications, including Saffran et
al. (1996), the study that got the child language world
excited about statistics (strange, but true).
Patricia Kuhl: TED Talk:
www.ted.com/talks/patricia_kuhl_the_linguistic_genius_of_babies
Google: Patricia Kuhl – The linguistic genius of babies,
for a TED talk on the infant’s remarkable language
segmentation skills and how they change over the first
year of life. Observe how Kuhl describes the loss of
non-native discriminations in terms of a critical period
in development.
Still want more? For links to online resources relevant to this
chapter and a quiz to test your understanding, visit the companion
website at https://study.sagepub.com/saxton2e
6 The Developing Lexicon: What’s in a
Name?
Contents
Approaches to word learning 148
First words 149
Comprehension versus production 149
What do one-year-olds talk about? 149
Overextension 151
Categorically wrong 152
Lexical plugs: Pragmatic errors 154
Losing it: Retrieval failures 154
Lexical processing 154
Up, up and away: The vocabulary spurt 156
Why so fast? 157
Spurt? What spurt? 159
The rate of word learning 160
Ten words a day? 160
Fast mapping 161
Slow mapping: The gradual accretion of meaning 161
Biases 163
The return of the gavagai problem 163
A noun bias in the child and in research 163
Nouns are easy 164
Verbs are hard 165
The shape bias 167
The rise and fall of word learning biases 167
Associative learning: The origin of biases? 167
Where do biases go? 169
Some lexical gaps 170
Computational modelling based on probability theory 171
Overview
By the end of this chapter, you should be familiar with the kinds of
words that children first learn. We will also consider a common kind of
error – overextension – in which words are used with too wide a scope.
We will consider three possible causes of overextensions:
 a category error (misclassification of an object)
a pragmatic error (expediency determines the word chosen)
a memory failure (the wrong word is retrieved)
We will also investigate the so-called vocabulary spurt and consider
explanations for the speed with which children learn new words. We
then consider solutions to the gavagai problem, introduced in Chapter
1: How do children work out what words refer to? Three approaches
are considered:
conceptual biases
associative learning
probabilistic learning
Conceptual biases, including the whole-object bias and the shape bias,
determine how the child perceives and interprets objects and actions.
The child is thus equipped with expectations that help them interpret
what words refer to. Meanwhile, associative learning, long studied in
psychology, may explain how biases are acquired in the first place.
And finally, we will examine the recent use of probability theory to
model word learning in the child. On this approach, the probability of
the child converging on the correct meaning of a word increases when
the child is equipped with both prior knowledge about the world and
increasing experience of the objects in it.
Approaches to word learning
If there is one linguistic concept that almost everyone recognizes, it is the
word. Words are everywhere. If I don’t know the meaning of a word, I can
ask someone or reach for a dictionary. If I feel like relaxing I might play
Scrabble online (oh dear, the number of hours gone forever). Others do
crosswords. Children play I Spy. People have an awareness of what a word is,
in and of itself. And arguably, word is pretty much the only linguistic
category that enjoys widespread recognition and metalinguistic
understanding. Our sense of ease with the concept of word might cause us to
take words for granted. In a similar way, learning new words might not seem
like such a big deal. Even a word like axolotl does not take much effort to
learn. Until recently, I had never encountered this word, less yet the species
of Mexican salamander it denotes. But my friend Diana is now
unaccountably smitten by these strange creatures. In consequence, I have
learned the word axolotl, and much else besides, concerning Diana’s new
hobby (no, she doesn’t get out much). I was shown the actual creature (an
object in the world) and developed an impression (or concept) of the slithery
beast (no, not Diana) and, to help matters, was supplied with the word form
axolotl to name my new concept. Some kind of connections must presumably
be made between the object in the world, the concept it inspires, and the word
form that labels the concept. In this chapter, we will consider the nature of
these connections and the kinds of learning that might be involved to make
them. But first, let’s sketch out some of the major achievements in early word
learning.
First words
Comprehension versus production
Most children produce their first word at around the time of their first
birthday, as you will recall from Chapter 1. But the comprehension of words
may emerge sooner, as early as seven months in some cases (Harris, Yeeles,
Chasin & Oakley, 1995). This disjunction between comprehension and
production is a prevalent theme in child language. Children know more than
they can say. This is not surprising when you consider that to utter a word,
the child must retrieve it from memory, and then formulate and execute a
suitable motor-articulatory programme to pronounce it. During speech
production, the activation of voicing and the movements of the articulators,
including lips, tongue and velum (soft palate) are very finely co-ordinated in
their movements and timing. It is not easy getting your tongue round words.
The most obvious case of the comprehension–production disjunction was
provided in the last chapter. It takes a year to produce the first word, while
speech perception is remarkably sophisticated during this period. As we can
see in Figure 6.1, the number of words that children understand grows
quickly, reaching about 150 words by 16 months. But only a fraction of that
total is produced at the same age, something in the region of 30 words. The
lag in production is still in evidence at two years, and can even be witnessed
in adults (Gershkoff-Stowe & Hahn, 2013).
What do one-year-olds talk about?
The thing about toddlers is that they don’t have a terribly wide range of
interests. In conversation, the same topics crop up over and over again,
irrespective of who their parents are and regardless of where or how they’ve
been brought up.
Given the list of interests in Table 6.1, it is not surprising to find that parental
speech to young children is dominated by similar topics (Ferguson, 1977). As
we noted in Chapter 4, the parent tends to follow the child’s lead as a way of
initiating and maintaining the conversation. Another observation that brings
no surprises is that early words are easy to pronounce. A number of factors
make life easy for the child. For example, pill is easier than spill, because the
latter starts with a consonant cluster, that is, two consonants, /sp/, rather than
one, /p/. What’s more, spill begins with a so-called fricative, /s/, a category
of sound which is more difficult to produce than others (Stoel-Gammon &
Sosa, 2007). Beyond individual sounds, words with one or two syllables, like
hat or water, are easier than polysyllabic horrors like tyrannosaurus. Observe
that tyrannosaurus might well be on the lexical wish list of your average two-
year-old. But they are likely to avoid difficult words like this, or at best
mangle them, by dropping one or more syllables, and by reducing consonant
clusters to something more manageable. For example, bread becomes bed,
blue becomes boo, and school becomes cool (Menn & Stoel-Gammon, 2005).
Meanwhile, spaghetti reduces to ghetti, banana to nana, and crocodile to
crodile (Stoel-Gammon & Sosa, 2007). We see here a good example of how
different levels of language interact in development. The child’s lexical
production is directly affected by their phonological skills. The number of
words produced is limited, while those that do appear are often subject to
reductions of various kinds.
Figure 6.1 Median number of words produced and understood by children
aged 8 to 16 months
Source: Adapted from Fenson, L., Dale, P.S., Reznick, J.S., Bates, E.,
Thal, D.J. & Pethick, S.J. (1994). Variability in early communicative
development. Monographs of the Society for Research in Child
Development, 59/5, Figure 1, p. 35 and Figure 2, p. 38.
Overextension
Learning the word coat to refer to the garment worn when you go out is
useful. But much more useful is an understanding that the word coat can be
extended beyond that particular item of clothing to include your parents’
coats, the coats of people you see in the street, coats worn by TV characters,
and so on. The extension of meaning to include members of an entire
category is an essential part of word learning. In the early stages, though, a
limited grasp of word meaning can lead to category errors. In particular,
children overextend word meanings: they use words beyond their
conventional range of meaning. For example, Hildegard, aged 1;0 used the
word Papa to refer not only to her father, but to her grandfather and even her
mother (Leopold, 1939). By 1;2, she was calling any man she met papa
(history does not record how many paternity suits ensued). Between the ages
of 1;6 and 2;6, overextensions are very common, accounting for as many as
40 per cent of child word uses (Rescorla, 1980).
Child overextensions (Leopold, 1939; 1949; Clark, 1973):
1. ticktock
square-faced watch, clock, all clocks and watches, gas meter, fire hose
on spool, bath scale with round dial, round eraser
2. fly (insect)
fly, specks of dirt, dust, all small insects, child’s own toes, crumbs of
bread, toad
 3. ass
toy goat on wheels with rough hide, sister, wagon, all things that move,
all things with a rough surface
4. baby
self, other children, pictures of children, any child
5. sch!
music, cars and trains, pictures of cars, a toy wheelbarrow, old-
fashioned carriages, a riding motion
6. choo-choo
trains, Bradyscope (an early piece of psychological apparatus), planes,
wheelbarrow, street-car, trunk
Children overextend word meanings in both production and comprehension.
It is not so easy to observe overextension in comprehension, but McDonough
(2002) shows how it can be done. She presented children (mean age 1;11)
with pictures of three objects, for example, two birds and a hat. McDonough
first asked children to place a sticker on a bird and, when successful, they
were then asked to place a second sticker on another bird. Observe that the
child is not expected to say anything during this experiment; only
comprehension is being tested. The test items were sometimes from different
domains (e.g., animals and clothing, as in our example) or from the same
domain (e.g., all animals: two birds and a pig). Overextension errors were
especially high (29 per cent) when the items came from the same domain.
This means that, having correctly identified a bird, from a choice of two birds
and a pig, the child will, on occasion, place a sticker on the pig when asked to
find another bird. Hence, overextension is more likely when objects share
certain properties like animacy or shape. McDonough (2002) also reports that
overextensions are more frequent for unfamiliar items. Again, this makes
sense. The less familiar a given word is, the less likely that the proper scope
of its extension will be known. Hence, more errors will be made. But what is
the basis of these errors? Three accounts have been offered, which we turn to
below: (1) category error; (2) pragmatic error; and (3) retrieval error.
Categorically wrong
A category error occurs when the child’s emerging categories diverge from
adult norms. For example, the child who mislabels a hippo as cow may have
mistakenly placed hippos and cows together in the same category (Mervis,
1987). Why would the child do this? In this case, one might conjecture that
similarity in shape is driving the child’s desire to form a unified herd of
hippos and cows. We will look at the influence of shape in more detail below.
But shape is not the only basis for overextension errors. If you look again at
the examples above, you will see that size, sound, characteristic motion, taste
and texture have all been observed as the basis for overextensions. So
although hippos might fall (rather heavily) into the cow category, shape
similarity is not the only possible explanation for the child’s error. Cows and
hippos also share the same taxonomic category – they are both mammals (see
Box 6.1). Perhaps cow means mammal to this child, in which case, it is quite
legitimate to include hippos, too. As Clark (2003) points out, shape similarity
and taxonomic level are often confounded, making it difficult to determine
the basis for the child’s overextension. In both cases, though, we can say that
the child’s overextension is caused by an immature semantic system leading
to one or other kind of category error (Naigles & Gelman, 1995; Gelman,
Croft, Fu, Clausner & Gottfried, 1998).
Box 6.1 Prototype Theory and Category Membership
Prototype theory is based on the idea that category membership is
graded. Concepts may be more or less representative of a
particular category (Rosch, 1973). Consider, for example, an
ostrich, a penguin and a robin, all members of the bird category.
Most people will agree that the robin represents the most birdlike
bird among these three (Kintsch, 1981). In other words, robins are
prototypical birds, possessing all of the necessary, defining
features that define the concept of bird (e.g., has feathers, has
wings, can fly). Different birds can therefore be more or less
typical members of the bird category. Ostrich and penguin are
less typical, in part because neither of them fly.
Experiments have shown that prototypes possess psychological
reality. For example, reaction times are quicker and people are
more accurate when asked if a robin is a bird, compared with an
ostrich (Kintsch, 1981; Rogers & Patterson, 2007). Also, typical
category members, like robin, come to mind first when people are
asked to list all the birds they can think of (Mervis, Catlin &
Rosch, 1976).
Concepts often figure in hierarchical relationships with one
another. At the basic level, we find concepts that are most readily
learned and retrieved from memory, concepts like bird. There is
some evidence that children acquire basic-level concepts first
 (Spaepen & Spelke, 2007; though see Widen & Russell, 2008).
Superordinate to bird is the category of animal. Superordinate
categories tend to be abstract and difficult to visualize (what does
animal look like?). At the subordinate level we find more specific
instances of bird, like our friends the robin, the ostrich and the
penguin.
Problems have been raised with prototype theory, based, for
example, on the inherent vagueness of language terms (Hampton,
2007; see also Eysenck & Keane, 2005). Overall, though,
prototype theory has stood the test of time remarkably well and
remains useful in helping to explain the conceptual development
of children.
Lexical plugs: Pragmatic errors
A second cause of overextensions is the existence of a lexical gap – a
concept, but no word to name it. One way to plug this gap is by resorting to a
plausible word that has already been acquired. Take the child who possesses
the word dog, but not the word horse. What does she do when presented for
the first time with a horse? One solution is to reach for dog and use that – a
pragmatic solution in which ‘it is almost as if the child were reasoning: “I
know about dogs, that thing is not a dog, I don’t know what to call it, but it is
like a dog!”’ (Bloom, 1973: 79). In this way, all four-legged creatures can
end up being labelled dogs for a time. Pragmatic errors reveal that the
boundaries of lexical categories can be rather fluid. Gaps can also be filled by
the invention of a new word altogether, something that young children do on
occasion. Examples include the use of pourer for cup, bee-house instead of
beehive and plant-man for gardener (Clark, 1993; Becker, 1994). The origins
of these words are readily apparent in the child’s existing vocabulary and
demonstrate the child’s willingness and ability to be creative with the
linguistic resources at their disposal.
Losing it: Retrieval failures
The third possible cause of overextensions is a simple failure to retrieve the
correct word from memory. The word that is selected usually bears some
resemblance to the target word. The resemblance is usually based on a
confluence of perceptual and taxonomic similarities (like our cow/hippo
example, above), but can sometimes be based on perceptual similarity alone,
as in ball for apple. More rarely, the child might select a word that sounds
like the target, for example, moon instead of spoon (Gershkoff-Stowe, 2002).
In each case, the child knows the target word, but simply slips up in
production, producing a related alternative. Chomsky (1965) would describe
this phenomenon in terms of the difference between competence and
performance. The child’s performance (the word that is produced) does not
reflect the knowledge they possess. And children do sometimes produce
errors, even when they have demonstrated their comprehension of an object
name (Huttenlocher, 1974). In particular, children approaching their second
birthday start making errors when naming objects, despite previous correct
performance (Dapretto & Bjork, 2000). This deterioration in performance
coincides with the early stages of rapid vocabulary growth and indicates that
performance factors (unreliable retrieval from memory) can mask the child’s
true level of understanding. Failures in memory retrieval have been
implicated in other kinds of language error, most notable morphological
errors (see Chapter 7).
Lexical processing
We have seen that there are three plausible causes for children’s
overextension errors: (1) an immature category; (2) a lexical gap (pragmatic);
and (3) a faulty retrieval from memory. These explanations are not mutually
exclusive; they might all have some validity. Gershkoff-Stowe, Connell &
Smith (2006) bring all three together under the broad umbrella of lexical
processing. Research with adults suggests that when we name an object, there
are three levels of processing, namely, object, concept and word (Johnson,
Paivio & Clark, 1996; see also Figure 6.2 below). Naming starts at the object
level, in the act of perceiving a particular object. This perception then
activates our concept (or memory) of what we have seen. Catching sight of a
chair will thus activate the concept of chair in the mind. At the same time,
several related concepts will also be activated (for example, sofa, table, stool,
bookshelf). Each activated concept, in turn, causes the activation of the word
that labels the concept. Several words are thus activated in parallel. The word
which receives the highest level of activation is the one that wins out and is
actually produced. Factors affecting level of activation include perceptual
similarity, practice and context (Brown & Watson, 1987; Martin, Weisberg &
Saffran, 1989; Gershkoff-Stowe, 2001). First, activation is stronger the more
closely the object resembles an existing concept (e.g., the sight of a robin will
activate the concept bird more strongly than the sight of a penguin). Practice
also increases activation, as seen when high frequency words are recalled
relatively quickly. And finally, activation is increased if a concept, or a
related one, has been activated recently. This recency effect is an example of
priming and has been the subject of scores of studies in cognitive psychology
(Anderson, 1983).
So how does this model of lexical retrieval help account for our three
different kinds of overextension? First, category errors arise when a novel
object – for which the child has no name – activates an existing, similar
concept (and its associated name). In (a) above, we know that Hildegard
possesses the concept wristwatch and a name for it, ticktock. Imagine that she
then sees a gas meter for the first time. The perceptual similarity (a dial with
moving hands behind a glass face) might prompt the activation of the
wristwatch concept and hence the associated name, ticktock. To be a genuine
category error we would have to be confident that Hildegard truly considered
the gas meter to be an instantiation of her concept wristwatch. This is
perfectly possible, but it would be very difficult to establish empirically. How
could we be sure that Hildegard was not making the second kind of error,
based on pragmatic necessity? In that case, the word ticktock would have
been activated by the sight of the gas meter (by perceptual similarity) and
would have been used by Hildegard as a handy means of talking about the
gas meter in the absence of a suitable name. Hildegard would be perfectly
aware that gas meters and wristwatches are different things, but would be
plugging a lexical gap by labelling the gas meter ticktock.
The third type of overextension is caused by a retrieval error. Gershkoff-
Stowe et al. (2006) demonstrate that recently activated concepts can lead to
overextensions, because, on occasion, their continued high activation level
leads them to be selected, in place of the correct, similar concept. Gershkoff-
Stowe (2001) suggests that young children are especially vulnerable to this
kind of error in the early stages of word learning. The child’s lack of
experience, compared to an adult, means they have had less chance to form
strong links from object to concept, and from concept to word. In
consequence, competition from a recently activated concept is more likely to
interfere in the word retrieval process, leading to an incorrect selection.
Overall, the lexical retrieval model helps explain three seemingly disparate
causes for overextension errors. At the same time, it is apparent that actually
teasing apart which particular cause is responsible for which particular error
brings on an empirical migraine.
Figure 6.2 Levels of processing in object naming
 Source: From Gershkoff-Stowe, L., Connell, B. & Smith, L. (2006).
Priming overgeneralizations in two- and four-year-old children. Journal
of Child Language, 33/3, 461–486, Figure 1, p. 464: Levels of
processing and lexical competition involved in naming a perceived
object.
Up, up and away: The vocabulary spurt
One of the most repeated facts about word learning is that children start out
slowly, but then, suddenly, go into orbit with a step change in their rate of
learning (Goldfield & Reznick, 1992). Following production of the first word
at about the age of 12 months, further words are added rather slowly. By
about 18 months of age, the child has clawed their way to a productive
vocabulary of about 50 words. But from this point onwards, there is a sudden
acceleration in the rate of learning. One of the first studies to identify a
vocabulary spurt was Dromi (1987). Figure 6.3 shows data from Dromi’s
daughter, Keren, who was followed longitudinally between the ages of 10
and 17 months. Ages are shown in months, with days in parentheses (for
example, 13(11) = 13 months and 11 days).
As you can see, Keren’s spurt takes place at about the age of 15 months.
Others have suggested 18 months (for example, Bloom, 1973) or even 21
months as the age at which the naming explosion takes place (Plunkett &
Wood, 2004). Recent evidence points to a shift in brain activation patterns
between the ages of 20 and 24 months which are associated with a sudden
growth in vocabulary (Borgström, Torkildsen & Lindgren, 2015). Perhaps the
critical factor for the start of the spurt is not age, but the size of the child’s
productive vocabulary. But even then, the issue is not that straightforward.
Many studies suggest 50 words as the critical mass required to trigger a spurt,
but others argue that 100 words is more appropriate (Bates, Marchman, Thal,
Fenson, Dale, Reznick, Reilly & Hartung, 1994). Clearly, there are
disagreements about when the spurt begins. But the idea that a spurt does
take place enjoys wide acceptance (Kauschke & Hofmeister, 2002; though
see below).
Figure 6.3 The vocabulary spurt in a child at the one-word stage

Source: From Dromi, E. (1987). Early lexical development. Cambridge:

Cambridge University Press. Figure 1, p. 111: Keren’s cumulative
lexicon at the one-word stage.
Why so fast?
Diesendruck (2007) suggests that prodigious word learning is related to a
generally impressive capacity for learning. Other authors, too, ascribe the
child’s lexical prowess to general cognitive abilities (Bloom, 2000; Poulin-
Dubois & Graham, 2007). These include an extensive knowledge about the
properties of objects and the relations they can enter into with one another.
Children are also equipped with sophisticated social–cognitive skills that
allow them to interpret what adults are referring to. These latter skills include
theory of mind, that is, the ability to attribute mental states (including beliefs,
desires and intentions) to oneself and others (Premack & Woodruff, 1978).
The child’s increasing memory capacity and general learning abilities
probably do contribute to an explanation of rapid word learning. But that is
not the same as a spurt. Perhaps a further candidate, the onset of
combinatorial speech, holds the key. Bates & Goodman (1997) observe a
close interrelationship between words and grammar and certainly, the point at
which word learning takes off in earnest (roughly 18 months) is also the point
at which the child begins to put words together in multi-word sequences. As
Figure 6.4 shows, there is a very strong relationship between the size of the
child’s vocabulary and the grammatical complexity of their utterances.
Moreover, this relationship is non-linear. This means that growth (in both
vocabulary and grammar) is not steady, but increases rapidly after a phase
where growth has been more sedate. This tight relationship between
vocabulary and grammar persists until the child is at least 30 months old
(Fenson et al., 1994). In a similar vein, early vocabulary learning also
predicts the child’s facility with literacy at the age of ten years (Song, Su,
Kang, Liu, Zhang, McBride-Chang, Tardif, Li, Liang, Zhang & Shu, 2015).
Figure 6.4 The relationship between grammar and vocabulary

Source: From Bates, E. & Goodman, J.C. (1997). On the inseparability

of grammar and the lexicon: Evidence from acquisition, aphasia and
 real-time processing. Language and Cognitive Processes, 12(5–6), 507–
584, p. 517, Figure 2.
Spurt? What spurt?
The sudden acceleration in rate of word learning has variously been described
as a burst, a spurt or an explosion. All very exciting, but not necessarily true.
Bloom (2004) suggests that the vocabulary spurt is actually a myth. First, no-
one seems to agree on what counts as a spurt. We have already seen that
disagreement exists about both the age and vocabulary size marking the onset
of the spurt. There is no consensus, either, on what rate of learning might
identify a spurt. On one view, the spurt has started when the child has
accrued ten or more new object names in a single three-week period (Gopnik
& Meltzoff, 1986). Mervis & Bertrand (1995), meanwhile, look for ten or
more words in a two-week period, at least half of which are object names.
But perhaps this is nit-picking. Most authors do use roughly similar criteria,
however arbitrarily chosen they might seem.
Figure 6.5 Different ways of modelling vocabulary growth
 Source: Ganger, J. & Brent, M.R. (2004). Reexamining the vocabulary
spurt. Developmental Psychology, 40/4, 621–632. Figure 1: A spurtlike
function (logistic) superimposed on slightly modified data from Child
041B (p. 623), and Figure 2: A nonspurtlike curve (quadratic)
superimposed on the same data shown in Figure 1 (p. 623).
Perhaps more seriously, there is evidence that the vocabulary spurt is a
minority taste. Ganger & Brent (2004) examined the growth curves for 38
children, looking for a kink, or inflection point, in the graph that would
indicate a dramatic shift in rate of learning. As shown in Figure 6.5, different
kinds of growth curve can be fitted to the same set of data. Statistical
methods can help choose which curve function models the data most closely.
Ganger & Brent were looking for a logistic function, essentially a growth
curve with a kink in it that would denote a spurt. But this kind of curve was
rarely the best model of the data, being most suitable for only five out of the
38 children they examined. Unless all children show a spurt, its significance
in theories of language development remains unclear. If we accept the
existence of a spurt, then we must thereby accept a discontinuity in
development – a qualitative shift from one stage to another. But shifts of this
kind are difficult to explain and, wherever possible, are avoided in theorizing,
if only for the sake of parsimony (Pinker, 1984). Nevertheless, the fact that
some children undergo a spurt requires explanation. Perhaps individual
differences, in the constitution of either the child or the input (or both), hold
the key.
The rate of word learning
Ten words a day?
Spurt or no spurt, the rate at which the child learns words appears prodigious.
The child shifts from knowing just a handful of words at 18 months to several
thousand by five years. But how fast is that? A figure that is often bandied
about is a rate of 10 words per day. But this figure, which finds its origins in
Carey (1978), overstates the case. And it also masks considerable variation
over time. Children do not actually hit their peak until they are about 10 years
old. The estimates shown in Table 6.2 assume an average vocabulary size of
60,000 words by the time of (American) High School graduation (Anglin,
1993). Word learning begins quite slowly but then picks up speed throughout
the school years.
The figures in Table 6.2 represent our current best estimates of vocabulary
growth, but, as we pointed out in Chapter 1, estimating vocabulary size is a
notoriously difficult venture. In addition, quoting average figures can be
deceptive, because it breeds the idea that the child is moving along at a
constant rate. In fact, it is very difficult to determine, in absolute terms,
whether word learning is fast or slow (see Chapter 8 where this issue is
considered in relation to grammar). What we can say is that, within each age
band, there will be times of accelerated growth, balanced by times when
things ease off (Bloom, 2004).
Fast mapping
We have established that children are capable of learning words quickly. But
how do they manage it? One answer is that children often need very little
experience to learn and retain (at least something about) individual words. In
fact, a single exposure to a new word can be sufficient to lodge it in the
child’s head. The first study to show this fast mapping effect was conducted
by Carey & Bartlett (1978). They taught three- and four-year-old children a
new colour term, chromium, which they used for a shade of olive green. (Yes,
chromium is a shiny metal, but these children were not to know that.) Given a
choice of two trays – one blue, one chromium – children were asked to
‘Bring me the chromium tray, not the blue one, the chromium one’. If you
had never before heard the word chromium, what would you do? These
children fetched the olive-green coloured tray and, moreover, they retained
some understanding of chromium as a colour name even after six weeks
(Carey, 1978). Fast mapping ability improves during development, with four-
year-olds out-performing their 18-month-old peers (Schmidt, Oliveira, Santos
Lotério & Gomes, 2016). But retention from fast mapping is not always good
(Horst & Samuelson, 2008). While children can learn something from very
limited experience, the benefits of practice, through repeated exposure,
should not be underestimated. Practice with one set of words can even lead
children to cope better in learning an entirely different set of words
(Gershkoff-Stowe & Hahn, 2007).
Slow mapping: The gradual accretion of meaning
At the start of this chapter, I noted that axolotl is a new word for me. Maybe
for you, too. If so, you may at least recall that axolotls are a kind of
salamander from Mexico. This means that without ever meeting a real-life
axolotl, you can begin to use the word successfully. But there is more to
know about the meaning of axolotl. They can regenerate entire limbs, should
they happen to lose one. They remain in the larval stage of development
throughout their lives, failing to undergo metamorphosis. And so on. The
concept of axolotl is not an all-or-nothing affair. As adults, then, we often
have only partial knowledge of a word’s meaning. Malapropisms provide
another example of partial lexical knowledge:
He is the very pine-apple of politeness. (Mrs Malaprop, from Sheridan’s
play The Rivals)
We heard the ocean is infatuated with sharks. (Stan Laurel)
Republicans understand the importance of bondage between a mother
and child. (Dan Quayle)
For children, too, it can take some time to acquire the full meaning of a word,
especially abstract terms like truth or purity, or complex verbs like pour and
fill (Gropen, Pinker, Hollander & Goldberg, 1991). Try ‘pouring the glass
with milk’ or ‘filling milk into the glass’, if you think the meanings of pour
and fill are straightforward. Not surprisingly, therefore, children do not learn
all there is to know about a word straightaway. Notice how, in this case at
least, we have turned the comprehension–production disjunction on its head.
Full comprehension of a word often lags behind the ability to produce it.
Generally speaking, we can see that an adult-like ability to understand and
use words may take several years to achieve.
Exercise 6.1
Give yourself three days to learn 30 new words, but do it in a language
that is completely foreign to you. You should be able to find reference
materials in your University library for Mandarin or Swedish or
whatever language takes your fancy. Compile a list containing concrete
objects and actions, that is, things you can see or touch, like cup, table,
eat, jump. Avoid abstract words like constitution, beauty, decide. On
Day 1 make a set of drawings depicting each word. Then start
practising. On Day 3, test yourself by naming each picture. Then
consider the following points:
How well did you do? 100 per cent correct?
 How many times did you have to repeat the words to make them
stick (if they did stick, that is)?
What kind of exposure did you have to each word?
Would a different method of learning have been more effective?
How does the way you learned these words differ from the way a
two-year-old would have learned them? See also the Website
recommendations at end of this chapter.
Biases
The return of the gavagai problem
Remember what gavagai means? If your fast mapping is not so fast any
more, slip back to Chapter 1, or see if the following brief reminder works.
Let’s take a two-year-old and show her a corkscrew. For the sake of both the
argument and her liver, let’s assume that corkscrews are a novel item for this
child. What does the word corkscrew refer to? Following Quine (1960), we
cannot assume a priori that the word denotes the whole object. In principle,
there is nothing to stop the child assuming that the word refers to the handle
of the corkscrew alone, or the corkscrew plus the bottle it is opening, or the
corkscrew as it appears at 7am on Sunday morning (quite evil). And so on.
This infinity of referents is logically possible. Patently, though, children are
not misled in this way. They do not lurch all over the place, mislabelling
concepts and acquiring bizarre word meanings. But what stops the child from
doing this? The answer that has dominated research on word learning for the
past 20 years or so is the notion of a learning bias. If children are biased to
pay attention to certain aspects of a scene, they are more likely to make the
same connections between the world and its words as adults do. We can
begin to explore the topic of word-learning biases through a more general
bias: the dominance of nouns over other kinds of word in the child’s early
vocabulary.
A noun bias in the child and in research
Nouns, in the form of object names, have a special place in the child’s early
vocabulary. In every language studied so far, object names occur with an
unusually high frequency when compared with the vocabularies of older
children and adults. This pattern holds for Dutch, English, French, Hebrew,
Italian, Korean and Spanish (Bornstein, Cote, Maital, Painter, Park, Pascual,
Venuti & Vyt, 2004). It also holds for less well studied languages like Tzeltal
(a Mayan language spoken in Mexico), Kaluli (spoken in Western Samoa)
and Navajo (a native American language) (Gentner & Boroditsky, 2001).
There may be some exceptions, though. Children learning Chinese, Korean or
Japanese may learn verbs earlier than nouns, or possibly at the same time,
with no privileged place for nouns (Tardif, 1996; Choi, 1998). But the picture
is far from clear. Other studies report the opposite, more conventional pattern
of noun dominance, even if verbs are learned earlier than in English (for
example, Au, Dapretto & Song, 1994; Rescorla, Lee, Oh & Kim, 2013; Hao,
Liu, Shu, Xing, Jiang & Li, 2015). The confused picture may stem from the
methods used in these studies. For example, the method of data collection
influences the kind of language produced by the child. Both Chinese- and
English-speaking children produce more nouns when reading books with
their parents than when playing with toys (Tardif, Gelman & Xu, 1999). In
addition, many studies rely on parental checklists. In this approach, parents
are given a list of words and asked to check those which, in their opinion,
their child is already producing. But as noted above, child usage does not
always give the best indicator of child comprehension. Experimental
approaches provide much clearer support for the dominance of nouns over
verbs. In a novel word learning study, three-year-olds learning English,
Japanese and Chinese all succeeded in acquiring new nouns but not new
verbs (Imai, Li, Haryu, Okada, Hirsh-Pasek, Golinkoff & Shigematsu, 2008).
Nouns are easy
The research on nouns versus verbs leads us to ask two questions. First, why
are nouns so easy? And second, why are verbs so difficult? If we look at
nouns first, it is important to note that we are not talking about just any kind
of noun. The nouns children first acquire are typically object names that refer
to physically manifest things I can see and touch, like jelly or mud. But many
nouns are abstract, like happiness, purity or antidisestablishmentarianism
(second appearance for this word – look it up: it’s the longest conventional
word in English). Other nouns are concrete, but refer to things or events that
are difficult to observe directly, or point out in a simple way, for example,
symbiosis or sunshine or fission. It is only certain kinds of noun – object
names – that children learn with particular ease. Having said that, not all
nouns in the child’s early vocabulary are object names: some of the nouns in
the child’s first 50 words refer to parts, actions, locations and substances
(Bloom, Tinker & Margulis, 1993). Object names might be easy, but other
kinds of noun are not impossible.
We have now narrowed our question about the noun bias down to the
following: what is it about object names that children find relatively easy?
The answer lies in the nature of objects and how we perceive them. It seems
that we are born viewing the world in terms of whole objects (Spelke, 1994).
We expect objects to adhere to certain principles, namely: cohesion,
continuity, solidity and contact. Cohesion refers to the fact that objects
comprise a connected and bounded region of matter. Continuity decrees that
objects follow a continuous path through space. Only in Doctor Who, and the
like, do objects disappear in one place and pop up in another. Solidity,
meanwhile, suggests we have a (reasonable) belief that one object cannot
pass through another (again, Doctor Who fans notwithstanding). And finally,
the principle of contact determines that inanimate objects move only when
they are touched by another object or person. There is good evidence from
developmental psychology that we are born with these expectations and that
we view the world as a place comprising separate, whole objects (Spelke &
Kinzler, 2007).
Some authors talk in terms of a universal whole-object bias (for example,
Imai et al., 2008) that compels the child to assume that words refer to whole
objects. Indeed, if a 12-month-old hears a new word, their initial assumption
is that it refers to an object category (Waxman & Markow, 1995). So
powerful is this bias, the perception of angles in geometric figures can be
skewed (Gibson, Congdon & Levine, 2015). Erroneously, four-year-olds are
swayed by perception of an object – specifically, the length of the lines in a
figure – when making judgements about the size of its angles. There is
evidence, also, that the whole object bias is more powerful than other word-
learning biases. Three-year-olds retain object words learned from fast
mapping after one week, but failed to recall words describing shape, texture
or colour (Holland, Simpson & Riggs, 2015). Observe how the whole-object
bias greatly reduces the complexity of Quine’s gavagai problem. It would
simply never occur to a child that a word might refer to just part of an object
or include entities beyond an object’s boundaries.
The whole-object bias is part of human cognition generally (Patterson, Bly,
Porcelli & Rypma, 2007). It is not part of language and does not compel us to
use language to name objects. Many animals, including rats and monkeys,
have been shown to possess a whole-object bias (Winters, Saksida & Bussey,
2008). But humans are unique in exploiting their whole-object bias in the
naming of objects. What prompts us to do this? The answer may lie in the
property of shared intentionality. Humans, unlike chimpanzees, seem
compelled to share their attention and interest in what they are doing,
including picking out objects of interest to talk about (Tomasello &
Carpenter, 2007; see also Chapter 9). Shared intentionality, combined with
the whole-object bias, may therefore explain the dominance of object names
in early child vocabularies. Object names not only make possible the learning
of other kinds of noun, they provide a framework for learning verbs. For
example, if one knows the object names ball and girl, it will be easier to
grasp the meaning of the verb kick when viewing a scene described by The
girl kicked the ball.
Verbs are hard
Now let’s tackle the second question: why are verbs so difficult? As with
object names, the answer seems to lie partly in the way we perceive and think
about the world. Actions are inherently more difficult than objects to pick out
as discrete entities from the environment (Gentner, 1982, 2006). With
objects, the cohesion principle ensures that the boundaries of an object are
easy to detect. In contrast, deciding where an action begins and ends is not
always straightforward. Also, many actions change as they unfold over time
(think of dancing or swimming). There is some evidence that children
perform better when the visual salience of an action is improved (Abbot-
Smith, Imai, Durrant & Nurmsoo, 2017). Generally speaking, though,
children seem to have more difficulty encoding (and later recalling) actions
than they do objects (Imai, Haryu & Okada, 2005). Verbs are also difficult
for distributional reasons. This point relates to our discussion in Chapter 5,
where we discovered the statistical exploits of the preverbal infant. It turns
out that nouns are more predictable than verbs with regard to the linguistic
contexts in which they occur. For example, the probability of the being
followed by a noun is 0.19 (almost 1 in 5). Verbs are significantly less
predictable. But in one experiment, the predictability of verbs was increased
by increasing the number of times infants heard them with the morpheme –
ing (jumping, diving, running) (Willits, Seidenberg & Saffran, 2015). Given
this leg-up, infants as young as 7.5 months found verbs less taxing to
identify. The distributional properties of nouns and verbs therefore contribute
to the relative ease with which they are learned.
Children also have difficulty in extending verb meanings to new instances.
For example, I may first encounter the verb jump in the sentence Simon is
jumping on the bed. Generalizing beyond this context, I might change the
agent (Martine is jumping on the bed). I might also change the object, the
thing being jumped on (Simon is jumping on the sofa). In one study, children
were provided with unusual (novel) objects that were paired with novel
actions. For example, a rugby ball-shaped object with fins was paired with
the action of lightly tossing and catching it with both hands (Imai et al.,
2005). It was found that three-year-olds can generalize verb use to a new
agent (as with the switch from Simon to Martine), but only when the original
object was used (the bed). Of course, actions can often be applied to more
than one object. The action of cutting can be applied to a potato, a cabbage or
my finger. But object and action seem to be fused together conceptually for
three-year-olds. When the action was applied to a different object (as in the
sofa), children failed to generalize their knowledge of the new verb. By five
years, in contrast, children could cope with changes in either actor or object.
Why are verb meanings less easily extended than noun meanings? The
answer to this question is not so clear. The child may sometimes need to
witness specific actor–object combinations several times before they can be
encoded and identified as nameable actions. That is, the child needs more
experience with verbs than with nouns. And the reason for this may lie in the
relative difficulty we have in perceiving actions as discrete events in the
world. This problem arises for the simplest kind of verbs – concrete actions
depicting events that we can readily perceive and participate in. Once the
hurdle of concrete actions has been crossed, the child still faces other kinds of
verbs, a situation paralleling the case for nouns discussed above. For
example, some verbs have concrete referents, but are not so easily observed,
like putrefy or indulge. Yet others are abstract, like sanctify or excel. The
naming of concrete objects is therefore only the first, most tractable, step in
learning the full range of word types that must be acquired.
Figure 6.6 Stimuli used in a study demonstrating the shape bias in early word
learning (from Landau, Smith & Jones, 1988)
The shape bias
As we have seen, the whole-object bias provides a good launching pad in the
task of object naming. But it will only get the child so far. Object names must
also be extended appropriately to new examples from the same category. We
have already hinted that the child might make use of an object’s shape in this
respect. In addition to the whole-object bias, then, a shape bias has also been
proposed. An early demonstration of the shape bias is provided by Landau,
Smith & Jones (1988). In their experiment, they created a real object from
wood and named it a dax (as shown at the top of Figure 6.6). Children were
told ‘this is a dax’ and then asked about each of the remaining objects in
Figure 6.6 with the question, ‘Is this a dax?’ For children aged two and three
years, a new object was accepted as a dax if it had the same shape as the
original dax. Other factors, meanwhile, were overridden, including size,
colour, pattern, material and texture. The shape bias is a developmental
phenomenon, with children becoming progressively more adept at using
shape to guide word learning between the ages of 12 and 24 months (Hupp,
2015). In some cases, as in Spanish-speaking children, the shape bias
develops more slowly, if at all (Hahn & Cantrell, 2012). But generally
speaking, the shape bias is a pervasive factor in early word learning
(Samuelson, Horst, Schutte & Dobbertin, 2008).
The rise and fall of word learning biases
Associative learning: The origin of biases?
The shape bias has come to dominate research on children’s early extension
of word meaning. But other biases have also been documented in the
literature. The most prominent of these are summarized in Table 6.3. There is
a general consensus that these special biases exist and greatly facilitate
children’s learning of new categories and words. But two important questions
remain unanswered. First, where do these biases come from? And second,
where do they go? You will not be surprised that the first question – the
question of origin – reduces quickly to our old friend nature–nurture. It is
possible that children are born with particular ways of perceiving and
conceptualizing the world. These cognitive dispositions are then later
harnessed in the acquisition of word meaning. A strong candidate in this
regard is the whole-object bias, which many researchers consider to be a
biologically determined system quite independent of language (Spelke &
Kinzler, 2007). Other researchers sometimes point to the universality of word
learning biases, but do not equate that universality with a genetic origin (for
example, Imai & Haryu, 2004). This caution applies even though the focus is
on the child’s conceptual, rather than linguistic, knowledge. On the whole,
word learning researchers have been more hesitant when it comes to grasping
the nativist nettle than in other areas of child language, like grammar or
morphology (see Chapters 7 and 8).
The alternative to nativist accounts rests on the child’s ability to learn. And in
research on the shape bias, we find the most fully developed account of how
the child might acquire conceptual knowledge of shape through perceiving
and interacting with objects in the world (Smith, 2001; Colunga & Smith,
2008). This approach is known as the Attentional Learning Account (ALA)
and relies on the well-established phenomenon of associative learning.
Associative learning takes place when two perceptual cues co-occur in a
predictable way. The presence of one cue can be taken as a reliable (if not
perfect) indication that the second cue will also be present. Once an
association has been formed, then the presence of one cue will automatically
direct the learner’s attention to the second cue. Attention thus becomes
selective. In the word learning context, attention becomes selectively focused
on object shape as a reliable cue to an object’s category, as indicated by its
name. In this regard, it is telling that more than 90 per cent of the count
nouns first acquired by children are most easily categorized according to
their shape (Smith, 2000). Shape is therefore a reliable cue to the name that
an object will be labelled with in episodes where parents say things like ‘This
is a ’. In this way, the linguistic cue becomes associated with the perceptual
cue of shape. The shape bias ends up being specific to the domain of
language, but it ‘emerges from very general learning processes, processes
that in and of themselves have no domain-specific content’ (Smith, 1999:
182).

The importance of associative learning for word learning is not contested.

But not all researchers agree that the ALA can explain the origins of
conceptual knowledge about shape (or indeed, other aspects of conceptual
knowledge). A major sticking point has been the age at which the child
demonstrates their knowledge of shape in categorization tasks. In early
versions of the ALA, it was assumed that the child required considerable
experience of object–word pairings for the shape bias to emerge. It was
suggested that this point might not be reached until about age 2;6 when the
child’s vocabulary is something like 50–100 words. However, recent studies
have shown that conceptual categories are in evidence much earlier – as
young as 12 months – and, moreover, they are not necessarily bound to the
act of naming (Samuelson & Smith, 2005; Booth & Waxman, 2008; Hupp,
2008). For example, Waxman & Markow (1995) showed that 12-month-old
children can form different animal categories (e.g., duck, lion, bear, dog),
much earlier than suggested by the ALA account. Moreover, it is not clear
that these younger children are using shape as the basis for categorization.
Other factors, like the particular form and position of head, eyes and legs may
explain children’s success (Gershkoff-Stowe & Smith, 2004). It does seem
that children have already developed certain key aspects of conceptual
knowledge before they begin the task of word learning. The further
development of conceptual knowledge then becomes intertwined with further
advances in vocabulary acquisition. The way in which associative learning
interacts with conceptual knowledge has not yet been determined (Elman,
2008). We must watch this space (with attention to its shape) for further
clarification.
Where do biases go?
Perhaps more mysterious than the issue of where biases come from is the
question of where they go, or rather, the question of how children override
their biases as learning progresses. Biases are helpful in the early stages of
word learning, but they must be overcome at some point. A quick look at the
lexicon beyond the realm of concrete count nouns reveals why this must be
so. Take, for example, the whole-object bias. Many words do not refer to
whole-objects at all, but, instead, are used to refer to parts of objects (like the
leg of a chair) or properties (like the colour of the chair). And many (many)
other words have nothing to do with objects at all, for example, those words
used for actions, events and relations. Another bias, mutual exclusivity, is
overridden very early on (Markman, 1990). This bias decrees that there
should be one, and only one, label for each object (see Table 6.3). But at the
age of two years, my son Alex called our pet both by its name (Michelangelo)
as well as by the basic-level term, cat. (Incidentally, Michelangelo was
named after a ninja turtle, not a Renaissance artist.) Mutual exclusivity can be
overridden, but, to complicate matters, it does not entirely disappear (Jaswal,
2010). Children can be seen adhering to mutual exclusivity years later, at the
age of six (Davidson & Tell, 2005). The example of different names for our
pet could also be seen as a case of overriding the basic-level assumption (see
Box 6.1). While cat is the basic-level term, the child also acquires both
superordinate (animal) and subordinate terms (Michelangelo). As learning
progresses, many other terms could be recruited for our cat, including feline,
pet, mammal and moggy. In a similar vein, the child can go beyond the
taxonomic bias by the age of four years (Srinivasan & Snedeker, 2014).
Even the mighty shape bias disappears when children are presented with
complex-shaped objects or when they are not forced to choose – as in many
experiments – one object from an array of objects (Cimpian & Markman,
2005). Another way to overcome the shape bias is to provide a strong hint of
animacy by adding eyes to the test object and by manipulating the texture
(e.g., making it furry or feathered) (Jones, Smith & Landau, 1991). In other
words, children’s attention can be drawn beyond the perceptual attributes of
shape during the process of categorization and naming. As children get older,
they become increasingly adept in this regard. For example, children aged
three and four years can focus on an artefact’s function – what it is used for –
something that is not always obvious from appearance alone (Kemler Nelson,
Frankenfield, Morris & Blair, 2000). Children can also attend to other
conceptual distinctions, including real versus toy animals (Carey, 1985), rigid
objects versus non-solid substances (Soja, Carey & Spelke, 1991), and
animate versus inanimate (Jones et al., 1991). As children’s conceptual
knowledge about objects increases, it can be deployed with increasing
sophistication in the acquisition of new words.
Biases do have an important part to play in vocabulary acquisition and the
fact that they can, indeed must, be overridden is important also. We might
think of these biases as a series of springboards that help get the child started,
both in the early stages of word learning, but also later on, when particular
words are new to the child. They help narrow down the range of possible
hypotheses about what words mean. In this way, they go a long way to
solving Quine’s gavagai problem. What we do not yet know is where these
biases come from. As noted, most researchers commit happily to the idea that
the whole-object bias is innate, but there is much more evasion for other
biases (e.g., Imai & Haryu, 2004). If the question of origins has not yet been
answered satisfactorily, even less is known about the disappearance of word
learning biases. Numerous studies demonstrate that the child can be flexible
in what they attend to when categorizing and naming. But as yet, it is not
clear how or why word learning biases operate at some times, but not at
others. What we are left with at present is simply the fact that these biases
exist, that they have a universal currency in children learning very different
languages, and that they are used to guide the expansion of children’s
vocabularies.
Some lexical gaps
The origin and demise of word learning biases are but two mysteries that
remain unresolved in research on word learning. Before we move on, it is
worth pointing out some other rather large gaps in our knowledge about word
learning. As we have seen, a large amount of research effort has been devoted
to just one very specific kind of word, namely, concrete count nouns (like
chair, dog, tree). In recent years, the palette of enquiry has been expanded
somewhat to encompass verbs (e.g., Hirsh-Pasek & Golinkoff, 2006;
Golinkoff & Hirsh-Pasek, 2008) and adjectives (Ninio, 2004; Mintz, 2005;
Booth & Waxman, 2009; Hall, Williams & Bélanger, 2015), but we still
know relatively little about other types of word, including articles (a, the,
some), prepositions (by, through, from) and pronouns (he, she, they) among
others. Evidently, there is still a lot to learn about how children acquire their
knowledge of words.
Computational modelling based on probability
theory
Innate biases and associative learning have recently joined forces under the
flag of probability theory. To remind yourself about the basic precepts of this
approach, slide back to Box 5.2. The aim is to devise a computer program
that can model child word learning. To do this successfully, modellers aim to
simulate the conditions of learning in the following ways: (1) equip the
model at the outset with the same knowledge that the child brings to word
learning; (2) provide the model with input that equates to that experienced by
the child; and (3) show that the outcome of learning is the same as the
child’s. With regard to (1), the learner’s initial beliefs are sometimes known
as the learner’s hypothesis space. The hypotheses that a learner can entertain
about word meaning are constrained within this space by the knowledge and
learning mechanisms that are brought to the acquisition process. With regard
to (2), the input can be simulated in various ways, but emphasis is often
placed on frequency (see Chapter 9). Ideally, the frequency of exposure to the
words being learned should equate to that experienced by the child. The
context in which words are presented should also be similar. Finally, the
success of a simulation can be judged quite straightforwardly by comparing
how well the model does with how well the child does. The model should
achieve a similar degree of success to the child, not only in terms of accuracy,
but also in terms of errors. We want our model to behave like the child, so
similar kinds of errors should occur (for example, overextensions) at similar
frequencies.
To date, Xu & Tenenbaum (2007) provide the most serious attempt to model
word learning computationally. Their approach is based on Bayes’ theorem,
named after an eighteenth-century mathematician, Thomas Bayes (1702–
1761). Bayes’ theorem provides a formula for calculating the probability that
a belief is true. In this case, we are concerned with learners’ beliefs (or
hypotheses) about word meanings. Support for a given hypothesis
accumulates with experience, to the point where the chance that it is correct
becomes very high. Nevertheless, the learner must always make some leap of
faith beyond the evidence (an inductive inference). We can see this more
clearly if we resurrect Quine’s gavagai problem. First, let’s take the issue of
prior knowledge (or beliefs). We have seen that there are good grounds for
equipping the learner with a whole-object bias at the outset of learning. In so
doing, we greatly narrow down the hypothesis space (the kinds of hypotheses
that will be entertained). The learner is much more likely to predict that
gavagai means rabbit (the whole object) rather than the rabbit’s left ear or
the rabbit plus the ground it is touching. Even so, the learner cannot be
absolutely certain that gavagai means rabbit. But the chances of being
correct have increased. In addition to prior knowledge, we can think about
the effects of the input. As experience is accumulated, the learner will
probably encounter gavagai in different times and places. But the pairing of
the word gavagai with the actual rabbit will remain constant. In consequence,
increasing experience will strengthen the association between gavagai and
the rabbit. Taken together, we can see how prior knowledge and experience
constrain probabilistic reasoning, in the process of honing in on the meaning
of gavagai.
Xu & Tenenbaum’s model was presented with the task of learning 24 object
names. Pairs of objects drawn from this set were rated by 22 adults according
to their similarity with one another (on a scale from 1 to 9). These ratings
were then used to inform the hypothesis space of the model. The objects were
organized into a hierarchy of clusters, based on how closely the objects were
judged to resemble one another. Similarity judgements can be based on
shape, size, colour, texture, animacy, and many other properties (both
perceptual and non-perceptual). Some or all of these relevant distinguishing
properties might enter into a given judgement, plus information gleaned from
prior experience with the objects. The model’s hypothesis space, therefore, is
very rich in terms of the prior knowledge it brings to the learning task, going
well beyond the whole-object bias. One limitation is thus that a whole raft of
word learning biases are conflated together. As noted, the prior knowledge of
the model was based on adult ratings of the objects. But the basis for those
ratings was not established. Future work might tease apart the differential
effects of different kinds of prior knowledge. Even if we cannot be certain
which aspects of prior knowledge were critical, the model was nevertheless
very successful at categorizing the objects in the same way as three- and four-
year-old children.
Another successful feature of the model is that it took into account the fact
that our experience with different words will vary greatly. Hence, the model
used probabilistic inference to arrive at different degrees of knowledge about
a word’s meaning, based on the particular level of input experienced. In
consequence, the model ended up with a more extensive knowledge of some
words than others. This attempt to capture different degrees of knowledge
about particular words is especially valuable, since it chimes well with the
observation, made above, that understanding of word meaning accumulates
gradually. Xu & Tenenbaum (2007: 270) conclude that ‘only a combination
of sophisticated mental representations and sophisticated statistical inference
machinery will be able to explain how adults and children can learn so many
words, so fast and so accurately’. But where do the ‘sophisticated mental
representations’ (the learner’s prior knowledge) come from? Are they innate?
Or are they learned, as a prelude to word learning proper? Modelling
approaches do not yet provide an answer. But given the way that probabilistic
machine learning closely models young children’s behaviour, we must take
seriously the idea that word learning is, in part at least, a statistical process,
based on the laws of probability.
In a Nutshell
Comprehension of words precedes the child’s ability to
produce them: children know more than they can say.
Overextension of a word’s meaning is a common error in the
early stages of vocabulary learning.
 Three possible causes of overextensions are: (1) a category
error (the child misclassifies objects); (2) a pragmatic error
(plugging a lexical gap); and (3) a memory retrieval error
(the wrong word is selected).
Some, but possibly not all, children undergo a vocabulary
spurt – a sudden acceleration in the rate of word learning, at
about 18 months.
Whether or not there is a spurt, lexical acquisition appears to
be fast, peaking during the late primary school years.
Knowledge of individual word meanings is acquired
gradually.
Conceptual biases, including the whole-object bias and the
shape bias, provide solutions to Quine’s gavagai problem.
They explain how children are constrained in the meanings
they ascribe to words.
Concrete count nouns dominate children’s early
vocabularies. Other kinds of noun, and verbs, are more
challenging.
Conceptual biases may be innate, though few researchers
adopt an avowedly nativist stance.
Associative learning has been used to explain the origins of
word learning biases, but biases appear much earlier than this
account might predict.
Children must override conceptual biases in order to expand
their vocabularies.
We still require satisfactory explanations both for the origin
of biases and for how they are overcome in development.
Probability theory provides a method of modelling child
word learning computationally. It combines a rich prior
knowledge, based on conceptual biases, with statistical
learning.
Further Reading
Hall, D.G. & Waxman, S.R. (Eds.) (2004). Weaving a lexicon.
Cambridge, MA: MIT Press.
This book presents a wide-ranging collection of papers from
specialists in lexical acquisition. It covers the period of early word
learning and claims also to look at later acquisition. Be warned,
though: ‘later’ means ‘post-first year, but still pre-school’ in a
language acquisition context. Nevertheless, this is a weighty tome
that covers all the major bases of research in the field.
Hirsh-Pasek, K. & Golinkoff, R.M. (Eds.) (2006). Action meets
word: How children learn verbs. Oxford: Oxford University
Press.
Another sizeable collection of papers, this book considers verb
learning and, in the process, makes a concerted effort to break
away from the massive research focus on early concrete nouns.
These papers set the research ball rolling in a new and potentially
exciting direction, by addressing a question we touched on here:
why are verbs relatively difficult to acquire?
Websites
Online Scrabble: www.isc.ro
Play to your heart’s content when you should be writing an
essay and learn two things about language acquisition: (1)
we continue to acquire new words throughout the lifespan;
and (2) we can learn and use words without any idea what
they mean or even what grammatical category they belong
to. Try vly, euoi and zati for size – all English words.
Tips on vocabulary learning: www.sheppardsoftware.com/
vocabulary_tips.htm
A commercial site, but here they offer you ten free tips on
how to improve your vocabulary. Does the young child use
any of these methods? Do you? What are the differences
between child and adult word learning?
Wordbank: http://wordbank.stanford.edu
An open database on children’s vocabulary growth which
contains data on more than 60,000 children across 21
languages. You can explore a wealth of data profiles and
analyses or even use free software to do your own
investigations (see also Frank, Braginsky, Yurovsky &
Marchman, 2016).
Still want more? For links to online resources relevant to this
chapter and a quiz to test your understanding, visit the companion
website at https://study.sagepub.com/saxton2e
7 The Acquisition of Morphology:
Linguistic Lego
Contents
Inflection 176
The acquisition of inflection 178
Whole word learning 178
The past tense debate: Rules or connections? 180
A dual-route account: Words and Rules theory 180
The acquisition of words and rules 181
The blocking hypothesis 182
Words and Rules: The story so far 184
Connectionism and a single-route account 185
Problems with connectionist models 187
Crosslinguistic evidence 188
Summary: One route or two? 189
Compounding and derivation 190
Derivation 190
Compounding 191
Derivation: The influence of productivity 193
Early compounds 194
Relating the parts to the whole in compounds 196
Complex compounds: Three processes combined 197
Morphology in the school years 199
Morphological awareness 199
Connections with vocabulary, reading and spelling 200
Morphological awareness in teachers 201
Overview
Morphology is concerned with the internal structure of words, with
morpheme being defined as the smallest unit of meaning in a language.
By the end of this chapter, you should be able to describe three
different ways of constructing complex words from morphemes:
inflection (jump + -ed → jumped)
derivation (inspect + -ion → inspection)
compounding (lady + bird → ladybird)
We shall examine how the child acquires these three processes. It will
 become apparent that a great deal of attention has focused on just one
aspect of inflectional morphology: the English past tense. Researchers
have investigated the way we learn and represent regular past tense
forms (moved, pushed) versus irregular forms (fallen, held). You
should develop an understanding of the regular–irregular distinction, in
relation to two competing accounts of how we represent past tense
morphology:
single-route
dual-route
The first approach, advocated by computer modellers in the
connectionist tradition, argues that all past tense forms are learned in
the same way. The second argues that two systems are required:
grammar (for regular forms) and the lexicon (for irregular forms). You
should come to appreciate arguments on both sides of this debate and
gain some insight into why it remains unresolved after 30 years of
enquiry.
We will also look at the acquisition of both derivation and
compounding. You should become familiar with the way children
acquire complex compounds like bus drivers and dog catchers, which
combine all three morphological processes. We will examine the
nativist suggestion that children may be born knowing how to assemble
words of this kind. We will conclude by looking at the development of
morphology in the school years. The acquisition of explicit knowledge
about morphology facilitates acquisition in other domains, including
vocabulary, reading and spelling.
Inflection
Begin this chapter with a deep breath. Hold it for a few seconds and then
release. Repeat this cycle until you achieve a state of complete tranquillity.
Then get ready to analyse the internal structure of words in terms of
inflectional, derivational and compound morphology. Deep breathing may or
may not help, but we do need to learn something about morphology in order
to appreciate the task which faces the language-learning child. The
complexities of complex words can overwhelm, especially if we look beyond
English. In Finnish, for example, it has been estimated that the morphological
system yields more than 2,000 possible word forms for nouns (Karlsson &
Koskenniemi, 1985). Despite this huge complexity, every typically
developing child born to Finnish-speaking parents acquires Finnish
morphology. If they can do it, so can you. Or rather, you can discover,
without too much pain, the essentials of morphology. Let’s start with a
definition. Morphemes can be defined as the smallest unit of meaning in a
language. These units can be smaller than a word, which means that every
word comprises one or more morphemes. The study of morphology is
therefore concerned with the internal structure of words. Inflection,
derivation and compounding provide three different ways of constructing
complex words, each of which pose different challenges for the language-
learning child. We will start with inflection, where our first observation is
that this aspect of morphology has devoured the lion’s share of research
effort. In fact, just one aspect of inflection in a single language – the English
past tense – has been (almost) all-consuming. But we shall try to throw the
net wider, once we have worked out what all the fuss is about with inflection.
Inflection is a process that changes the grammatical function of a lexeme
through the expression of properties such as number, gender, case and tense.
We can take inflection for number as an example. In English, there is no
inflection to mark a singular noun (one beetle, one atom). But we do add -s to
mark plurality (two beetles, twelve atoms). Inflection does not create a new
lexeme, that is, a new word with a distinct meaning. Instead, inflection yields
a range of different word forms that fulfil particular grammatical functions.
As we can see, one such function is the expression of plurality. English is
rather feeble when it comes to inflection, having only eight different
inflectional forms (see inflectional morpheme in the Glossary). Other
languages, including Finnish and Russian, have vastly more complex systems
of inflection. One complication in English (and other languages) is that
inflectional paradigms can have exceptions, which leads to a traditional
distinction between regular and irregular forms. The regular inflection for
plural is -s (trees, planes, ideas). But there are exceptions. I can roast two
geese, not *gooses, for three women, not *womans (I hope they’re hungry).
Goose and woman are irregular nouns. Similarly, the past tense in English
has a regular pattern, which (roughly speaking) involves adding -ed. I can say
that yesterday, I washed, shaved and talked. But again, there are irregular
exceptions. These include ate instead of *eated, drove instead of *drived, and
bought, not *buyed. Pinker and Prince (1988) suggest there are 181 irregular
verbs in English, but the list is not immutable. New irregular forms do
emerge from time to time. For example, it is common (in the UK) to hear text
as the past tense form for text (e.g., He text me last night), rather than the
regular form texted. The verb text is similar, therefore to verbs like cut and
shut, which also show no change from present to past forms. Pinker (1999:
15) suggests that, ‘whereas regulars are orderly and predictable, irregulars are
chaotic and idiosyncratic’. Thus, the past tense of sing is sang, but for bring,
we say brought (not *brang). The past tense of drink is drank, but for think it
is thought (not *thank). Viewed in this way, it is difficult to discern any
pattern among different irregular forms. But there is some order in the chaos
Pinker perceives, as we shall see, and this has a direct bearing on how the
child might acquire categories like the past tense.
A moment ago we were rather rough in our handling of the regular inflection,
-ed. Let’s smooth things out now, to help with the discussion that follows. If
we focus on the way -ed is pronounced, it becomes apparent that it varies.
Consider phoned, where -ed is pronounced /d/. The final phoneme of the root
form, phone, is /n/ (think sounds, not spelling). /n/ is a voiced sound and so is
/d/. Next up is danced, which ends in /t/, a voiceless sound. The final
phoneme of the root form, dance, is /s/, which is also voiceless. Hence, -ed
can be pronounced as either /d/ or /t/. The choice is determined by the voicing
of the final sound in the stem: the inflection takes on the same voicing state
as the root. In grown up linguistic talk, which you can drop into your next
conversation with Chomsky, we say that the past tense inflection is
phonologically conditioned. One further wrinkle. Some regular verbs insert a
vowel before /d/ (e.g., wanted, batted, patted and carted). Again, this is
phonologically conditioned: observe that the root form ends in /t/ in each
case.
The acquisition of inflection
Whole word learning
In principle, a child could get by in life without any knowledge of
morphology, nor any of the skills required to dissect words into their
constituent morphemes. All we would need is a powerful long-term memory
storage system, something which humans do, in fact, possess. Using a
sufficiently extensive memory, we could learn every word form as it stands,
with no reference to or acknowledgement of any internal morphological
structure. On this model of learning, the child would treat words like parrots,
believing and wanted as indivisible wholes. The child would never need to
analyse words into their constituent parts (e.g., parrot + -s). But none of the
analyses in Table 7.1 would be done at any point in the course of acquisition.

There is, in fact, some evidence that children do start out as ‘whole word’
learners, with no regard for the internal structure of words. For example,
children acquiring Hungarian produce correctly inflected words, even during
the one-word stage (MacWhinney, 1985). However, it is unlikely that the
child has assembled these words from separate morphemes, or has any
knowledge that these words can be broken down to yield distinct units of
meaning. This is because, initially, there are no signs of child productivity.
Children do not produce inflected words of their own, but confine their
output to words they have been exposed to in parental input. This
phenomenon has been observed in languages with rich inflectional systems,
including Hungarian and Turkish (MacWhinney, 1985; Aksu-Koç & Slobin,
1985; Aksu-Koç, 1997). In the absence of productivity, these early words are
best regarded as unanalysed wholes, even if they seem morphologically
complex from an adult perspective. It takes some time for children to work
out the meanings of inflections. This becomes apparent if we give children a
choice of two pictures to look at, one depicting a single novel object, the
other depicting two novel objects (see Figure 7.1).
Figure 7.1 Testing the comprehension of the plural marker -s (after Jolly &
Plunkett, 2008)

If the child hears Look at the jeels!, they should look at the two-object
picture, if, that is, they understand the meaning of the plural inflection -s. By
the same token, children should focus on the single-object picture if they hear
Look at the jeel! However, Jolly & Plunkett (2008) found that 24-month-olds
are insensitive to the plural marker. It is not until 30 months that children’s
viewing is guided by the presence or absence of the plural marker. The task
in this experiment taps into children’s comprehension of language. But we
know that they correctly produce plural markers of their own well before 30
months. In a similar fashion, Dispaldro & Benelli (2012) report that Italian
children only begin to understand that singular forms refer to one object at
about age four, while they are at least six years old before they comprehend
that plural forms refer to ‘more than one’. There is increasing evidence,
therefore, that correct production of morphologically complex words does not
necessarily mean that the child grasps either the form or meaning of a word’s
internal structure.
In fact, we know that, eventually, the child does begin to analyse the internal
structure of words. A simple demonstration of this fact was provided in
Chapter 1 with Berko’s (1958) wug test. You may recall that four-year-olds
can pluralize a nonsense word like wug – to produce wugs – even though we
can be sure they have never heard the plural form. The child is therefore
being productive in applying the plural inflectional morpheme -s to the noun
wug. A number of factors seem to nudge the child into realizing the ways in
which language encodes the distinction between single versus multiple
entities (Lanter & Basche, 2014). One such factor is the complexity of
singular–plural morphology in a given language. Russian, with its complex
morphology, presents the child with multiple cues for learning the singular–
plural distinction, whereas Japanese and English are much simpler.
Accordingly, Russian children show signs of understanding the singular–
plural distinction earlier than their Japanese and English contemporaries
(Sarnecka, Kamenskaya, Yamana & Yudovina, 2004). Parallel findings have
been reported for the acquisition of plural morphology within a single
language, the dialectal variants of Spanish found in Chile and Mexico (Miller
& Schmitt, 2012). The richness of the input has a direct influence on
learning. Chilean children hear fewer full-form plural endings and,
consequently, take longer to acquire the plural marker than their Mexican
counterparts. This finding is confirmed in a large study across six different
languages: Croatian, Dutch, French, German, Greek and Russian (Xanthos,
Laaha, Gillis et al., 2011). The complexities of the particular language being
learned are reflected directly in the morphological richness of the input
available to children, which in turn has a direct impact on the speed at which
morphological features are acquired.
The past tense debate: Rules or connections?
Back in the 1980s, when telephones were stupid (or at least, not yet
smartphones), we did at least have some form of computers, however large
and unwieldy. Thanks to computational power, a debate was launched by
Rumelhart & McClelland, in 1986, which has rumbled on for many years.
The questions thrown up by these researchers strike at the core of how
fundamental aspects of language are acquired and then represented in the
human mind. The frenzy of controversy is largely in abatement these days,
but not because the issues have been resolved, nor because the questions
raised have become any less interesting. Rather, as is tantalisingly often the
case, in both child language and psychology more widely, we lack the
wherewithal to resolve the matter one way or the other. In addition, we shall
see that concessions have been made, perforce, on both sides of the debate.
The result has been a blurring of theoretical boundaries which makes
adjudicating between the two sides more difficult. The two competing
approaches are known as: (1) single-route; and (2) dual-route theories.
Connectionist theorists provide the best-known single-route accounts, while
so-called Words and Rules theory is the most prominent dual-route model. A
good deal of the battle has been fought on the terrain of cognitive
psychology, with the focus on how inflections are represented in the adult
mind, rather than acquired by the child. But we shall try to keep the spotlight
trained on the issue of acquisition.
A dual-route account: Words and Rules theory
As the name suggests, Words and Rules (WR) theory claims that language
offers two ways of representing grammatical categories like past tense:
regular past tense forms are generated by a rule, while irregular forms are
stored in the lexicon. The mental lexicon is conceived as part of declarative
memory, our long-term repository of facts. The ‘facts’ that are stored about
each word include information about meaning, pronunciation and
grammatical category (e.g., verb). Together, this information comprises the
lexical entry for a word. Regular forms are produced by a mental production
line that puts words together at the time of speaking. The regular rule can
apply to any word with the symbol V (verb) as part of its lexical entry. For
example, the word talk is a candidate for undergoing the regular rule, because
it will be marked as V in its lexical entry. The past tense rule is enacted by
retrieving the base form, talk (from lexical memory), together with the past
tense morpheme -ed. An operation that merges the two elements is also
required to allow V + -ed to be concatenated into a single word online. This
unification rule, V + -ed, is a grammatical process that pays no heed to any
non-grammatical aspects of the base form (semantic or phonological). The
grammatical rule ‘add -ed’ thus applies to any regular verb with equal
facility. The second system produces irregular past tense forms. They are
simply acquired and stored as whole forms in the mental lexicon. They are
also retrieved from memory, as and when required, like any other word. In
consequence, an irregular form like bought is treated as an unanalysed whole,
even though we could argue that it bears some kind of inflection for past
tense – a vowel change plus word-final /t/ in this case. In summary, WR
theory depends on two systems – lexicon and grammar – for producing past
tense (and other) inflected forms.
Over the past 30 years or so, a wealth of evidence has been brought to bear in
favour of Words and Rules theory (e.g., Jaeger, Lockwood, Kemmerer, Van
Valin, Murphy & Khalak, 1996; Clahsen, Hadler & Weyerts, 2004; Ullman,
Pancheva, Love, Yee, Swinney & Hickok 2005; Shankweiler, Palumbo,
Fulbright, Mencl, Van Dyke, Kollia, et al., 2010; Magen, 2014; Smith-Lock,
2015). In its first blush of youth, Words and Rules theory provided a very
straightforward account of inflection. But life (and child language) is rarely
that simple. Take, for example, the ‘blind’ nature of the regular rule, ‘add -
ed’. The rule is blind in the sense that it should apply to any word marked V
(verb), with no regard for any other factors like meaning or sound. But this is
not the case. It has been discovered that the rule applies in a graded way,
being influenced by the phonological properties of particular verbs (Albright
& Hayes, 2003). The phonology of most verbs does not allow us to predict
whether they will be regular or irregular. But verbs that end with an unvoiced
fricative /s, f, ʃ, ʧ/ as in place, laugh, wish and itch are different. They are
always regular. This is probably why novel verbs ending with an unvoiced
fricative are more likely to be treated as regular than other verbs (Albright &
Hayes, 2003). This finding suggests that the regular past tense rule is not
blind to phonological factors in the way that Words and Rules theory
assumes (see also Harris & Humphreys, 2015). In consequence, the bedrock
of Words and Rules, which is the operation of rules over linguistic symbols,
is called into question. Albright & Hayes (2003) demonstrated that regular
verbs are not all equal. Another source of inequality is in the frequency of
regular verbs. Many regulars occur only rarely in conversation, but there are
some high frequency regular verbs (for example, looked, robbed, dropped)
(Clahsen et al., 2004). Reaction time data from children aged eight to 12
years suggest that high frequency regulars often behave like irregular forms,
being produced as pre-packaged whole forms, rather than being put together
online (Dye, Walenski, Prado, Mostofsky & Ullman, 2013). Hence, Words
and Rules theory has been amended: regular forms are not inevitably
produced online by a grammatical process.
The acquisition of words and rules
Let’s now consider the acquisition of words and rules. We can quickly
dispatch the ‘words’ part of this double act to models of lexical acquisition
(see Chapter 6). The way in which irregular plurals and past tense forms are
learned is not assumed to differ from the way in which any other word is
acquired. Having said that, the meanings ‘more than one’ for plurals and
‘before now’ for past tense have to be acquired as part of the lexical entries
for irregulars. With respect to the rules part of WR theory, it is clear that
particular rules must be learned from experience. They could not possibly be
innate, because languages vary in how they mark tense: ‘add -ed’ is fine for
English, but it does not work for Japanese. Some languages, including
Chinese, do not even mark tense on verbs.
We know that children begin to segment words into roots and affixes some
time in their second year (Slobin, 1978). What we do not yet know is how
they do it. Pinker (1999: 193) suggests that it may not be such a difficult task:
‘by subtracting walk from walked, push from pushed, and so on, the child can
isolate -ed’. But if it is so straightforward, why do children not perform this
analysis much earlier in development? After all, they solve the seemingly
harder task of segmenting phonemes and words from continuous speech in
the first year of life (Chapter 5). But whereas words are isolated before the
child’s first birthday, inflections are not produced until the child is about two
years old, and are not fully comprehended until about the age of three. One
more paradox for the pot, it would seem. What we do know is that there is a
close relationship between the frequency of inflections in Child Directed
Speech and the order they are acquired. The more frequent a given suffix in
CDS, the sooner it is acquired (Warlaumont & Jarmulowicz, 2012).
Discussion point 7.1: The segmentation problem
in morphology
As we have seen, little attention has been paid to the issue of how
children identify bound morphemes as units of language in their own
right within words. Go back to Chapter 5 and review the evidence on
the segmentation of words from continuous speech by infants.
Compare the evidence on segmentation for both words and
morphemes, respectively. The following recent article will be helpful:
Marchetto, E. & Bonatti, L.L. (2015). Finding words and word
 structure in artificial speech: The development of infants’ sensitivity to
morphosyntactic regularities. Journal of Child Language, 42(4), 873–
902.
Seven-month-old infants can distinguish word-like units, but what
do they actually know about such words?
At seven months, what else is there to learn about words? A
stopover in Chapter 6 might help with this question.
Why do you think there seems to be a disjunction between the age
at which the child segments words versus bound morphemes?
Consider the kinds of knowledge on display and the difference
between comprehension and production when you discuss this
issue. Consider also the different kinds of data that are brought to
bear on this issue in the different domains of morphology and
word learning.
The blocking hypothesis
One of the hallmarks of child language, observed with delight by hordes of
parents, is the overregularization error (OR). Even into the school years,
children misapply a regular inflection to irregular forms. For example, at one
point or another, many children add the regular -ed’ to the irregular verb
break. The result is breaked, instead of broke. ‘How cute!’ you doubtless cry.
And now you can also cry: ‘That’s an overregularization!’ (not so cute). WR
theory needs to account for the emergence of overregularization errors and
also for their eventual disappearance. It does this by proposing a blocking
mechanism that connects lexicon and grammar. On this account, when an
irregular form like drank is retrieved from memory, an inhibitory signal is
sent from the lexicon to the grammar, thereby suppressing the regular process
(drink + -ed) (see Figure 7.2).
Figure 7.2 The mechanism for blocking overregularized forms

On this view, both lexicon and grammar are activated in parallel every time
we produce a past tense form. In the case of an irregular verb like drink, the
activation of drank simultaneously blocks the production of drinked (Marcus,
Pinker, Ullman, Hollander, Rosen & Xu, 1992). It is assumed that blocking is
an innate mechanism, ‘built in to the circuitry that drives language
acquisition’ (Pinker, 1999: 197). We should be clear, though, that there is no
evidence to support this claim. Blocking prevents the occurrence of past tense
and plural errors. Which is all very well, apart from one small matter. As
noted above, errors do occur (see below).
Alex (aged 4;1–4;9)
Daniel’s broked my mast
And he sticked his tongue out at Snow White.
In the old days, they eated bad things.
Well, Tippy bited me.
I drawed a picture of a pirate today.
He shooted the fish.
I wish you buyed some for me.
If blocking exists, there must be some spanner in the works, preventing its
smooth operation. Marcus et al. (1992) suggest that blocking is vulnerable in
young children, because their memory retrieval system is immature and
sometimes lets them down. If the child fails to retrieve drove, the regular rule
proceeds unhindered, to give drived. Errors like drived are therefore caused
by lapses in memory retrieval. This explanation seems to solve the problem
of past tense errors, but it does so by sweeping the detritus of child language
under the carpet of cognitive psychology. However, very little is known
about the development of lexical retrieval. Generally speaking, in fact, the
blocking hypothesis has attracted remarkably little empirical attention.
One source of evidence that has been considered centres on the frequency of
overregularization errors. Originally, Marcus et al. (1992) suggested that
overregularization rates are low (4.2 per cent), because the regular process is
overwhelmingly blocked by the presence of irregular forms in the child’s
repertoire. But Maslen, Theakston, Lieven & Tomasello (2004) report OR
rates as high as 43 per cent in one case. It turns out that OR rates can be very
high for brief periods of development, especially for high frequency verbs
(Maratsos, 2000). In fact, even adherents of Words and Rules theory now
acknowledge that past tense errors are common (for example, Pinker &
Ullman, 2002: 456). A further problem is that overregularization errors seem
to start at different points in development for different inflections. Thus plural
ORs (foots, mouses) appear in child speech three months before their past
tense counterparts (Maslen et al., 2004). This finding is problematic for the
blocking hypothesis. Once the process of inflection is available to children, it
should apply to nouns and verbs equally well (Marcus, 1995). One more
point of discomfort for the blocking account is the occurrence of blended past
tense forms. These are forms like droved, where the irregular form (drove)
has been retrieved, but where the suffixation of the regular -ed ending seems
to have taken place, too. Such forms are embarrassing to the blocking
hypothesis because they suggest that retrieval of the irregular form has failed
to block the default regular process. In summary, what little evidence we
currently have does not favour the blocking hypothesis. The concept of
blocking (or inhibition) is very well established in cognitive psychology (e.g.,
Sumner & Samuel, 2007). But within Words and Rules theory, blocking
stands on shaky ground. An alternative account is required for the emergence,
and eventual disappearance, of overregularization errors.
Words and Rules: The story so far
Let’s have a quick recap on where we’ve got to with the dual-route model,
before we look at the alternative, single-route approach. We have seen that
WR theory provides a very simple model of the English past tense (and the
plural), based on the distinction between regular and irregular forms. Regular
past tense forms are assembled online by a grammatical process, while
irregulars are retrieved as whole forms from the lexicon. Meanwhile,
overregularization errors are prevented by a blocking mechanism, inhibiting
the regular process when an irregular verb is activated. Occasional child
errors are then attributed to an immature memory retrieval system. We have
already seen that the simplicity of this model is belied by the complexities of
language. Three problems have arisen so far. First, the past tense rule (‘add -
ed’) should apply equally well to any regular verb, but it does not. Second,
high frequency regular verbs seem to behave like irregular verbs, stored as
whole forms in the lexicon. And third, the blocking hypothesis is not well
supported. As we proceed into connectionist waters, one further fundamental
challenge for WR theory will emerge: the validity of the basic distinction
between regular versus irregular. Seemingly so clear-cut in English, this
distinction begins to look illusory when other languages (like Polish) enter
the fray. But place that thought on hold for the time being, while we head
down a single route to the past tense.
Connectionism and a single-route account
Connectionist models are computer simulations that aim to emulate the
language learning behaviours of the child. The more closely the computer
model echoes child language data, the more plausible it becomes that the way
in which the computer learns reflects the way in which the child learns. The
startling discovery by connectionist modellers was that a single route for
learning the past tense may suffice. The single system is a pattern associator,
which detects and exploits the numerous regularities and quasi-regularities
that exist across the board among both regular and irregular verbs. The fallout
from the single-route model has been dramatic. It calls into question the
distinction between regular and irregular in inflectional morphology. It
further suggests that morphemes are not explicitly represented as discrete
entities. And the need for linguistic symbols like V, together with the rules
for combining symbols, are dispensed with. These are all foundational
assumptions of modern (and ancient) linguistic theory, so it is no wonder that
the advent of connectionism, about 30 years ago, was heralded as the dawn of
a new age in linguistics and cognitive science.
Irregular past tense forms are neither chaotic nor idiosyncratic, in the way
characterized by Pinker (1999). Arguably, only two verbs in English are
genuinely irregular, the so-called suppletive forms be (past tense was or
were) and go (past tense went). For the remaining irregulars we can detect
nine different clusters of verbs, each with its own pattern for the formation of
past tense forms (McClelland & Patterson, 2002). For example, 28 verbs
which end in /d/ or /t/ have identical forms in both present and past tense,
including cut, hit and bid. Another group, also ending in either /d/ or /t/
simply change the vowel in the past tense, for example hide/hid, slide/slid
and fight/fought. Yet another group replaces word-final /d/ with /t/ to produce
the past tense form: send/sent, bend/bent and build/built. By this point, your
cunning eye will have detected a pattern, based on the popularity of /d/ and /t/
for past tense endings. Fully, 59 per cent of the 181 irregular verbs listed by
Pinker & Prince (1988) take word-final /d/ or /t/. And you will recall that
these two phonemes constitute the realizations of the regular past tense
morpheme -ed. Connectionist models exploit the systematic patterns that
exist among sub-groups of irregulars, including the frequent appearance of /d/
and /t/. They treat the learning problem for these patterns as identical to the
problem of finding the pattern in regular verbs. Hence, all verbs are learned
by the same model in the same way at the same time, by establishing the
various patterns of association that exist across present and past tense forms
in different sub-groups of verbs.
In the connectionist approach, learning is achieved by a network of units that
correspond, metaphorically, to neurons in the brain. Each unit is connected to
many other units, in a way that mirrors the synaptic connections between
neurons. The level of activation for a given unit is computed from the
activation levels of all the units feeding into it. If the overall activation level
is sufficiently high, the unit will be switched on, causing it to send a signal
via all of its connections to the output layer. If, on the other hand, the
activation level of a unit does not exceed a critical threshold, it remains
switched off.
The sample network in Figure 7.3 looks complicated enough with its
numerous connections. But the real thing is even more bamboozling. In
Rumelhart & McClelland (1986), the input and output layers each contain
460 separate units, each of which is switched either on or off for a given
learning trial. The task of the network is to match the root form of a verb with
the correct past tense form. It does this by gradually forming patterns of
association between the features of words. The association is not between the
whole words, drive and drove, but between parts of words: dr- and dr-, dri-
and dro-, -ive and -ove, and so on. At the same time, other associations are
inhibited, for example, between -rive and -rived, to discourage drived as the
output for drive. The focus on features allows the network to detect the
patterns that exist among different families of verbs. For example, most of
the units that are turned on when shrink is entered will also be activated when
stink is entered. The output from the network gradually converges on the
correct form.
In Rumelhart & McClelland’s (1986) original model, 420 verbs were each
entered into the model 200 times, yielding a total of 84,000 trials. The end
result of all this activity was very impressive. The model converted most of
the input verbs into their correct past tense counterparts, both regular
(seem/seemed, melt/melted) and irregular (make/made, sing/sang). The next
challenge was to see if the model could generalize beyond the verbs it had
been trained on and generate appropriate past tense forms for 86 unfamiliar
verbs. The model managed to add -ed to three quarters of the new regular
verbs. It did so by analogy with the verbs in its training set. The model also
generated some overregularization errors, similar to those made by children,
like gived and digged. Intriguingly, these errors emerged after an initial
period when the correct irregular form was produced, somewhat like children
(though see below). The model also did well in recognizing clusters of
irregular verbs. It could therefore cope with a new verb like cling/clung by
analogy with ring/rung and sing/sung. The model also produced blended
forms, like gaved and stepted, which we have seen are a feature of child
language, but which are not easily accommodated by Words and Rules
theory. Overall, then, the achievements of connectionist models provide a
serious challenge to the idea that we need two systems for producing past
tense forms. Maybe just one system – a simple pattern associator – is all that
is required.
Figure 7.3 Connectionist model of past tense inflection

Problems with connectionist models

All is not entirely rosy in the connectionist garden. Early models suffered
from a number of deficiencies (see Pinker & Prince, 1988, for a full account).
One problem was the presence of a ‘teacher’ in the model, essentially a way
to compare its output against the target and adjust learning where the two did
not match. This ‘teacher’ can be seen as a form of corrective input, a concept
which, you will recall, is anathema to many theorists, both nativist and non-
nativist (but see Chapter 4 for an alternative view). Another problem was
with homophones like brake and break. Connectionist models that input only
information about phonology will treat these two verbs as identical, because
they sound the same. Yet one is regular, while the other is irregular
(brake/braked versus break/broke). A third problem is with novel and very
low frequency regular verbs. Connectionist models have difficulty in
generalizing to novel words that differ from those in their training set. They
either produce odd blends (like trilb/treelilt) or no output at all. The failure of
connectionist models in this regard may be inherent in their design (Marcus,
2001). Children and adults, in contrast, can cope with the extremes of novelty
and inflect whatever verb comes their way. Even odd concoctions that violate
the phonological constraints of English, like ploamph, can be accommodated
by people (ploamphed), but not by connectionist models.
A fourth problem for connectionism, which we can mention briefly, is the
modelling of U-shaped development, a phrase that captures the dip in
performance we mentioned earlier: errors with irregulars creep in after a
period of correct performance, before being phased out again. The U-shape
appears if we make a graph with percentage correct use on the vertical axis,
and time on the horizontal axis. Unhappily, Rumelhart & McClelland based
their model on inaccurate descriptions of this U-shape. They assumed a
simple developmental sequence for irregulars: correct → error → correct.
But correct uses never disappear. In fact, even when overregularization errors
emerge, they remain in the minority, with correct forms predominating.
Hence, the kind of U-shaped behaviour generated by Rumelhart &
McClelland’s model is not the kind actually exhibited by children. We have
made brief mention of just four problems with early connectionist models:
the presence of a teacher; homophones; low frequency and/or unusual verbs;
and the course of U-shaped development. Most of these problems have, to
some extent, been addressed in subsequent versions of connectionist models.
But no model has accommodated all of these problems in a single, supersize
solution to all connectionist headaches. Instead, individual models have
chipped away at individual problems.
Crosslinguistic evidence
Despite the massive amount of research on inflection, the scope of that
research is surprisingly narrow. The English past tense has devoured the
lion’s share of attention, with a few scraps thrown to English plurals. But we
should recall that inflection in English is a paltry affair. Inflection in other
languages, like Finnish, Hungarian and Russian, is vastly more complex, but
relatively little is yet known about the acquisition of these inflection systems
in their own right. Even where crosslinguistic research has been undertaken,
much of it is designed to contribute to the debate inspired by the English past
tense. Notable exceptions include research on the acquisition of plurals in
German (Szagun, 2001) and Arabic (Albirini, 2015). These languages each
have complex systems for pluralisation and, not surprisingly, children are
well into their school years before adult-like control is achieved. For
example, in Jordanian Arabic, children start out with a single default (or
regular) plural form – so-called ‘feminine sound’ plurals – but, later on, they
come to utilise two default forms, as witnessed in the adult language
(Albirini, 2015). The notion of two regular plural forms is, of course, alien to
English. What’s more, the plot thickens. Jordanian Arabic actually presents
children with four separate plural forms which are not mastered until about
the age of eight years. Critically, the acquisition of these forms is not driven
solely by considerations of default forms. In the early years, frequency and
productivity of plural forms have a strong influence on acquisition, with
predictability emerging as a further influence in the later years. Frequency
also figures large in a recent study of Finnish past tense acquisition
(Kirjavainen, Nikolaev & Kidd, 2012). Recall that, according to Words and
Rules theory, the regular past tense rule (‘add -ed’) applies to any and all
regular verbs with equal facility. There should be no differences among
verbs. But Kirjavainen et al. (2012) show that both frequency and
phonological factors affect children’s production of past tense forms. They
conclude that the single associative mechanism envisaged by connectionists
can account for the data. If we look beyond English, therefore, regularity may
not be the main factor driving acquisition of inflected forms. Even more
startling, crosslinguistic research now indicates that the distinction between
regular and irregular, seemingly so clear in English, is all but illusory in other
languages. Dąbrowska (2005) offers the genitive and dative systems in Polish
as a case in point (see case). There is no default regular marker in Polish and
most forms have irregular versions. In addition, both semantic and
phonological properties influence how nouns are inflected, contra Words and
Rules theory (Ramscar, Dye & Hübner, 2013). And regularity is a poor
predictor of productivity in Polish adults, again contra Words and Rules
theory. These findings suggest that the regular–irregular distinction may be
an artefact, something we impose on the data of English, without it
possessing any genuine psychological significance. But there is a fly in this
ointment. While Polish tells us to forget about the regular–irregular
distinction, up pops Hungarian. Like Polish, Hungarian has its complications
when viewed through the simplistic lens of English. For example, it is an
agglutinative language, which means that words can contain several
morphemes, essentially combined in (sometimes quite long) chains to express
features such as number, tense and gender. A recent reaction time study on
both noun and verb morphology in Hungarian found evidence in favour of
the regular–irregular distinction (Nemeth, Janacsek, Turi, Lukacs, Peckham,
Szanka, Gazso, Lovassy & Ullman, 2015). On the whole, though, it is
perhaps quite telling that the weight of crosslinguistic evidence tends to
favour the single route approach (in this vein, see also Nicoladis & Paradis,
2012).
Summary: One route or two?
The debate between single-route versus dual-route theories has generated a
huge amount of research. A wide range of methodologies has been applied,
including computer modelling and reaction time studies. More recently,
evidence has been forthcoming from neuroscience, throwing data on brain
responses into the pot. But not even this multiple-method approach has been
able to provide definitive evidence in favour of one theory over the other. In
this vein, a number of recent studies have recruited evidence on brain
responses to support Words and Rules theory (e.g., Bakker, MacGregor,
Pulvermüller & Shtyrov, 2013; Regel, Opitz, Müller & Friederici, 2015). But
similar methods also yield evidence for a single route, by demonstrating that
morphologically complex words are retrieved as whole words, rather than
being the product of online rules (Hanna & Pulvermüller, 2014; see also
Danko, Boytsova, Solovjeva, Chernigovskaya & Medvedev, 2014). And so
the controversy rumbles on, with no killer blow (or even peaceful resolution)
in sight. A major difficulty is that most of the findings can be accounted for
by both single-route and dual-route theories, while evidence in favour of one
theory over the other tends to be fairly weak (for recent examples, see
Ambridge, 2010; Magen, 2014). To accommodate contrary findings,
substantial modification of the original theories has proven necessary. For
example, we noted that Words and Rules theory now accepts the need for a
pattern associator in lexical memory. This acknowledges the way irregular
verbs are organized in distinct sub-groups. They are not just a list of
unconnected exceptions to the regular pattern. WR theory also accepts that
high frequency regulars may be stored in memory, not assembled online.
What remains, though, is the insistence that we have two systems for
producing inflected forms. Inflection is conceived as a mental process,
available to us for producing complex words. This idea is not accepted by
many connectionists. But it is less controversial than the notion that inflection
is a default process that applies unfailingly to regular forms. This is because
the notion of regular versus irregular is very hard to maintain when we look
beyond the confines of English. Pattern associators, too, have their problems.
In particular, they are nowhere near as proficient as the young child in
generalizing acquired patterns to new forms, especially if those forms are rare
or unusual in some way. The debate, therefore, continues. To date, there is no
incontrovertible proof that linguistic symbols and rules are redundant
concepts in psychology. But connectionism has demonstrated that a world
without symbols and rules is at least conceivable. As Seidenberg (1992: 93)
puts it, ‘any system can be described by a set of rules if the rules do not have
to apply to all cases’. In the absence of agreement, we will let Groucho Marx
have the final word:
Pardon me. I was using the subjunctive instead of the past tense. We’re
way past tents. We’re living in bungalows now.
(Animal Crackers, 1930)
We are way past the past tense now and heading for the bungalow where
compounding and derivation live.
Compounding and derivation
Beyond inflection, you will recall that language offers two further ways of
constructing complex words: compounding and derivation. In both cases, the
outcome is a new lexeme, with its own meaning. Derived words are formed
by the addition of an inflection to a word, as in confusion (confuse + -ion),
while compounds are made by joining two words together, as in apple juice.
We shall consider the mechanics of each process first and then consider how
the child tackles the acquisition of these morphological processes.
Derivation
Derivation creates new lexemes, through the addition of morphemes to a
base form (see Box 7.1). This process can be fairly straightforward, as in the
derivation of the noun driver, from the verb drive plus -er. But it can also be
more complex, with more than one bound morpheme involved in the creation
of a word. The word unbelievable, for example, can be analysed as un- +
believe + -able. Anglin (1993) identifies in excess of 100 different
derivational affixes in English. Try your hand with a few of them in the
Exercise below.
Exercise 7.1: Derived forms
What is the root in each of the following derived forms? What
grammatical category does the root belong to? What category does the
derived form belong to? Look up noun, verb and adjective in the
Glossary if you need a kick start.
obesity selection realize playful
unholy swimmer successful speciality
repellent curable reorganize membership
Londoner blacken clearance accidental
Box 7.1 Derivational Morphology
Derivational morphology creates new lexemes, words with
different meanings from the roots they are derived from. A
derived word may therefore have its own separate dictionary
entry. Sometimes the grammatical category also changes. For
example, the adjective desirable is derived from the verb desire,
as shown in the table below. The table also shows how, in
English, derivational morphemes can be prefixes or suffixes,
attaching to the left or right of a word, respectively.
Root form    Derived form
  Verb Noun Adjective
Verb → repaint driver desirable
  undo flattery resentful
Noun → demonize childhood selfish
  vaccinate lioness wholesome
Adjective → weaken craziness unhappy
  purify desperation yellowish
Most derivational processes are productive, that is, they can apply to a range
of different words. For example, the suffix -able (meaning ‘able to be’) can
be added to pretty much any verb in English: doable, driveable, changeable.
Admittedly, some such forms sound a bit odd (smellable, feelable, takeable),
but -able still stands as a highly productive morpheme, used in the creation of
new words. Another highly productive morpheme in English is -er, used to
denote ‘someone who performs an action’, as in lover, walker and baker.
Other morphemes are less productive, being restricted to a narrower range of
root forms. For example, the suffix -ity can only be added to a fairly limited
range of adjectives to produce nominal forms (curiosity, sincerity, sanity, but
not *warmity, *sadity, *bitterity). As we shall see, variations in productivity
have an impact on the child’s acquisition of morphology.
Compounding
Like derivation, compounding also results in a new lexeme, but this time, it is
formed by bringing two or more lexemes together, most commonly some
combination of nouns and adjectives in English. For example, wind and mill
(two nouns) can be compounded to produce the noun, windmill. Notice how
the meaning of the compound is not necessarily just the sum of its parts. It
typically takes on a meaning of its own. Another point to note is that
compounds can be written as a single word (runway), with a hyphen (part-
time) or as two separate words (table tennis). Nowadays, we even have the
spectre of capital letters word-internally (PowerPoint, SlideShare). The
choice of one written form over another does not seem to be systematic and
does not affect the linguistic status of compounds.
The internal structure of compounds can be analysed into a head and a
modifier. In English, the head appears on the right. Broadly speaking, the
head determines the meaning of the compound and also dictates which
grammatical category it belongs to. For example, in watertight, the head
element is an adjective, tight, so the compound is also an adjective. In
daredevil, the head element, devil, is a noun, and the compound is a noun.
The head element is also the part of the compound that can be modified by
inflection. For example, the plural of landlord is landlords. The plural marker
-s is added to the head, lord, not to the modifier, land (*landslord). How do
we know that two landlords is correct while *two landslord is wrong? More
to the point, how does the child acquire this knowledge? We shall return to
this question below. Our final observation is that inflection, derivation and
compounding do not exist in isolation from one another. We have just seen
that compounds can be inflected (landlords). But they can also include
derived forms, yielding so-called synthetic compounds. For example, train
driver contains the derived form driver (drive + -er), while washing machine
contains washing (wash + -ing). We thus reach the pinnacle of morphological
complexity, where inflection, derivation and compounding are combined. We
shall consider how the child copes with this morphological mayhem below
(see Table 7.2).
 Exercise 7.2: Morphological processes
In the following compounds, identify which morphological processes
have been applied: inflection, derivation and/or compounding. Be
careful. Sometimes more than one morphological process is involved
in the creation of a word form. Sometimes, none at all (a root form by
itself).
airheads sampling insider elephant
disgraceful pleased coffee table recycles
hat maker indefatigable reasserted sweetcorn
monster washing machine dressmakers unhappily
We will cover both kinds of process in this section, if only because research
studies have often brought them together. You might wonder why inflection
has been left out of this party, and you would be right to wonder. In fact,
studies are beginning to suggest that different morphological processes may
be quite intimately related. In this vein, Rabin & Deacon (2008) asked
children aged 7 to 10 to complete word fragments like ne _ _. Possible
answers include nest, next, neat and need. Before doing this task, children
were primed in different ways, by giving them a list of words to read. In
amongst this list, some children saw the inflected word needs, others saw the
derived word needy, while others saw a control word needle. The initial
sound sequence of needle (need) is shared by needs and needy, but it has a
different root from need. It was found that needs and needy, but not needle,
were effective prompts. Moreover, children were equally likely to complete
ne _ _ as need when primed with inflected and derived forms (needs or
needy). In consequence, Rabin & Deacon (2008) argue that inflected and
derived forms may be mentally represented by children in a similar way,
according to morphological principles.
Derivation: The influence of productivity
Derivation creates new words through the addition of affixes to a stem.
Affixes vary in terms of their productivity, which is a measure of how many
different words they can attach to. Try Exercise 7.3 below to get a feeling for
the way in which productivity varies across different affixes.
Exercise 7.3: Productivity of derivational
suffixes
Try deriving new words by adding the following suffixes to nouns,
verbs and adjectives chosen at random. Identify the category (noun,
verb, adjective) of both the original word (the root) and the derived
form. You will find that you cannot just add any affix to any word.
There are constraints. Also, some affixes apply to a wider range of
words than others. That is, they are more or less productive. Which
affix do you think is the most productive in the following list?
pre- de- -ity
in- -cy -less
under- -able -er
Children acquire the most productive suffixes first. For example, there are
three different suffixes in English that are used to mark an agent, the doer of
an action: -ian; -er; and -ist. Picture a small band of musicians with
trumpeter, drummer, pianist and violinist. If you’re feeling strong, repeat
Exercise 7.3 and think of more examples for each suffix (physician, golfer,
chemist …). You will probably discover that -er is far more productive than
the other two, and this is because -ist and -ian are largely confined to words
with Greek and Latin roots. Not surprisingly, children can use -er to mark
novel agents from the age of two or three, whereas creative uses of -ist and -
ian lag behind by a year or two (Clark & Hecht, 1982). Examples of
productivity with -er are provided by Clark (2003) in a child aged 2;2 to 2;9:
Come here, brusher.
Herb a bad guy, because I a gunner.
I’m a big reacher.
I’m going to shut that door hard because I’m a shutter.
I have a sip. I am a sipper.
In fact, -er is one of the earliest suffixes witnessed, together with the
diminutive suffix -ie, as in doggie and horsie. The influence of affix
productivity has also been observed in other languages. For example,
children learning Hebrew acquire the most productive suffixes early and use
them more readily in their own complex word production (Berman, 1987).
Productivity can also be gauged across different morphological processes. At
the start of this chapter we observed that languages exploit inflection,
derivation and compounding to different degrees. That is, each process is
more or less productive, and these differences are reflected in the rate of
acquisition by the child. In French and Hebrew, for example, derivation is
much more productive than compounding. Accordingly, children acquire
derived forms before compounds in these two languages. In German,
meanwhile, the reverse is true. Children acquire compounds, which are
especially productive in German, before derived forms (Clark, 1993). Now
that compounds have cropped up again, let’s explore further.
Early compounds
Compounds are in evidence very early on in children’s speech output. In one
recent study of Finnish, Estonian and Sami (the latter being native to northern
Europe), children as young as 1;10 showed evidence of compounds (Argus,
Laalo & Johansen Ijäs, 2014). Initially, however, children do not treat
compounds as complex words (Berman, 1987). Instead, they seem to follow
the ‘whole-word’ pattern of learning which we encountered above with
inflected words. The internal structure of a compound like football is not
analysed into its component parts (foot + ball), even if the child has acquired
each component as an individual word in its own right. Instead, football is
understood as an indivisible linguistic unit, in much the same way as other
polysyllabic words, like ballet or tennis. The treatment of compounds as
whole words, or holophrases, can persist into the school years for particular
compounds. For example, Berko (1958) found that 98 per cent of four- to
seven-year-olds were not aware that birthday referred to a particular day.
Nevertheless, from three years on, children begin to recognize the distinct
components in some compounds. In one test of this development, Nicoladis
(2003) asked children to pick out the correct referent for novel compounds
from a choice of four pictures, as in Figure 7.4.
Figure 7.4 Comprehension of a novel compound (‘sun bag’) by three- and
four-year-olds (Nicoladis, 2003)
Three-year-olds are more prone than four-year-olds to select a single object
from this array, rather than two interacting objects. It takes time, therefore, to
realize that interacting objects are most readily labelled by a compound noun.
Errors in the analysis of compounds can still be discerned in primary school-
aged children, as the following definitions reveal (Berko, 1958: 170):
Breakfast is called breakfast because you have to eat it fast when you
rush to school.
Friday is a day when you have fried fish.
An airplane is called an airplane because it is a plain thing that goes in
the air.
Although the child may sometimes miss the mark in their analyses, the
critical point to note is that they can perform such analyses at all. In fact,
from the age of two, English-speaking children begin to construct compounds
for themselves, as in house-smoke, for smoke from a chimney, and cup-egg,
for a boiled egg (Clark, 1981).
Do you remember that compounds comprise a head and a modifier? Of
course you do. In English, the order of these elements is modifier + head. The
head element determines the essential meaning of the compound. In
blackboard the head is board, modified by black. Children’s knowledge of
the modifier + head order was tested by asking them to interpret novel
compounds (Clark, Gelman & Lane, 1985). As you may imagine, a pencil-
tree will look very different from a tree-pencil, but even children as young as
2;4 succeed on this task 49 per cent of the time, rising to 96 per cent by the
age of 4;0. In Hebrew, the order is reversed, but Berman & Clark (1989)
demonstrated that children begin to discern the correct order from 2;5. The
four-year-old whose comprehension of compounds is almost perfect, may
nevertheless produce ordering errors. For example, when asked ‘What could
you call someone who pulls wagons’, observed responses included A pull-
wagon and A puller-wagon (Clark, Hecht & Mulford, 1986). But the head
element, pull(er) should appear to the right (wagon puller). Hence,
comprehension of the correct ordering for head and modifier precedes the
child’s production ability.
Relating the parts to the whole in compounds
In compounds, the relationship between head and modifier can take on many
different forms. The so-called thematic relation between head and modifier
determines how a given compound must be interpreted. Krott, Gagné &
Nicoladis (2009) provide the following examples:
chocolate muffin a muffin that HAS chocolate in it
cardboard box a box that is MADE OF cardboard
mountain bird a bird that is LOCATED on mountains
mountain magazine a magazine ABOUT mountains
If the meaning of a compound is unfamiliar, then the appropriate thematic
relation between the component parts must be inferred. In some cases, there
are morphological markers available to help out. For example, in bus driver,
the -er suffix can be used to deduce the relationship between bus and driver: -
er is agentive, so driver is a ‘person who drives’. It then follows that bus is
what the driver drives, yielding the thematic relation OF (‘a driver OF
buses’). In other cases, the use of a new compound in context will help the
child interpret its meaning and the relations between the components. A third
way to solve this problem is to rely on prior learning to work out the meaning
of new compounds. They could do so by drawing an analogy between the
new compound and ones they already know. If chocolate muffin is new, then
the child might rely on knowledge of other chocolaty compounds to deduce
its meaning. In this vein, Krott & Nicoladis (2005) report that four- and five-
year-olds do indeed infer the thematic relations of new compounds by
analogy with familiar compounds.
Learning by analogy is a useful strategy, but it will not always work by itself.
The problem is that some compounds have a number of competing
interpretations, as we have seen with mountain bird versus mountain
magazine. When there is competition of this kind, Krott et al. (2009) suggest
that the child will prefer the most frequently encountered interpretation for
any new compound. For example, for mountain, the most frequent thematic
relation is LOCATED (mountain bird, mountain goat, mountain lion, and so
on). And it turns out that, by the age of four years, children choose the most
frequent thematic relation for novel compounds about 40 per cent of the time
(Krott et al., 2009). If this figure does not impress you, remember that
children manage this with no contextual or linguistic support. Their responses
in this study are based solely on the ability to recognize analogies between
new compounds and previously learned ones.
Complex compounds: Three processes combined
Near the start of this chapter we ascended to the summit of Morphology
Mountain without oxygen, and planted a flag in words that combine all three
morphological processes. Words that incorporate inflection, derivation and
compounding include train drivers, flower sellers and alto saxophonists (see
also Table 7.1). In this section, we shall see that three-year-old children cope
well with this complexity, since they demonstrate surprisingly adult-like
intuitions about complex compounds. Young children do so well, in fact, that
some authors assume that the rules underlying the construction of such
compounds must be innate (Gordon, 1985). So what are these rules?
Kiparsky (1983) provides the most popular analysis, known as level ordering,
which predicts that production of complex compounds passes through three
hierarchically sequenced levels (see Table 7.3).

Looking at the compounds that emerge from this sequence, you might
possibly wince at men hunters, with its irregular plural modifier, men. Men
hunters might give you pause for thought (see below), but that’s nothing
compared to heads hunter. That one really jars. According to Kiparsky, such
forms are prevented because the plural inflection -s can only be added after
compounding has taken place, not before. This is why regular plural forms,
like heads, cannot occur within compounds. If you agree with Kiparsky, and
give the green light to men hunter, then we need to explain why irregular
plurals can occur within compounds. According to level ordering, the reason
is that irregular plurals like men are retrieved like any other word from the
lexicon at Level 1. Compounding then takes place at Level 2 (men + hunter).
If we wanted to, we could then even inflect the whole compound at Level 3
to yield men hunters (with the same meaning as man hunters: ‘several people
who hunt men’).
Gordon (1985: 79) laid his hands on a Cookie Monster puppet and tested the
level ordering hypothesis with children aged 3;2 to 5;10. The following
introduction was made:
Do you know who this is? … It’s the Cookie Monster. Do you know
what he likes to eat? (Answer: Cookies). Yes – and do you know what
else he likes to eat? – He likes to eat all sorts of things …
After a period of training, children were asked to name other objects that
Cookie Monster might eat. These included a rat, a collection of four rats, a
mouse and four mice. Gordon does not say if these were real rats and mice.
The child was then asked, for example, ‘What do you call someone who eats
rats?’ Rat and mouse have similar meanings, but you will have picked up that
rat is regular, while mouse is irregular. Therefore, level ordering predicts the
following possible responses:
rat eater YES
rats eater NO
mouse eater YES
mice eater YES
When faced with a rat or a mouse, most people scream. But these children
blithely produced compounds like rat eater and mouse eater on 98 per cent of
occasions. When asked simply to name a group of mice, some of the younger
children overregularized and said mouses. But these same children reduced
mouses to the singular form when the compound was elicited, giving mouse
eater. Children who used the correct irregular mice, also used it in the
compound, mice eater, 90 per cent of the time. Only rats eater was
conspicuous by its absence. It seems, therefore, that Kiparsky’s level ordering
model correctly predicts child speech. Irregular plurals can appear within
compounds, regular plurals cannot. Gordon (1985) asserts that knowledge of
level ordering is innate. He suggests that children could not have learned
these constraints because they are not exposed to relevant information in the
input which could assist learning. Thus, children avoid compounds that are
prohibited by level ordering (rats eater), while producing compounds that are
permitted (mice eater), despite having heard neither of them. If three-year-
olds possess this knowledge in the absence of help from the input, it is
assumed that the knowledge is innate. This line of reasoning is often used to
support nativist claims and we will examine it in more detail in the next
chapter. For the moment, it is worth pointing out that Gordon (1985) did not
actually examine the input to children. He relied instead on an early database
of adult written English (Kučera & Francis, 1967). Laying that aside, though,
other problems have arisen with the level ordering account.
In the first place, there are exceptions. In English, it is possible to have a
regular plural in modifier position: drinks cabinet, weapons inspector, Parks
Commission, awards ceremony, sports announcer. It is also possible in other
languages, including French, where there is a head + modifier order in
compounds. The word for nutcracker in French – casse-noisettes – glosses as
‘break hazelnuts’ with a regular plural modifier, noisettes. If level ordering
exists, neither French nor English should allow regular plurals in compounds.
A second problem is that studies since Gordon (1985) have found that
children do produce regular plurals inside compounds sometimes. In this
respect, Nicoladis (2003) discovered an interesting crosslinguistic difference
between monolingual English-speaking children and bilingual children
acquiring French and English. The bilingual children produced English rats-
eater compounds on 15 per cent of occasions, compared with just 2 per cent
for mono-lingual English speakers. French is more lenient than English in
allowing such compounds, which helps explain this pattern of findings. In
contrast, level ordering hits the buffers as an explanation. If level ordering is
an innate constraint on children, it should apply to speakers of all languages
equally. It turns out that children in the early stages of learning about
compounds are very likely to produce regular plural compounds. Nicoladis &
Murphy (2004) report that 50 per cent of children who make other kinds of
compound error (for example, ringer-bell instead of bell-ringer) also produce
rats-eater errors. In contrast, those children who are error-free on the bell-
ringer front cannot be induced to produce rats-eater. There is a
developmental progression, therefore, in which children learn to avoid
regular plurals in compounds.
Perhaps the most interesting finding from Gordon (1985), which still requires
explanation, is the propensity of children to say mice eater. The viability of
level ordering depends crucially on the acceptability of irregular plurals in
non-head position. However, I suggested, above, that you might balk at mice
eater. If so, you are joined by the adults studied by Haskell, MacDonald &
Seidenberg (2003). They found a hierarchy of acceptability for different
modifiers:
regular singular > irregular plural > regular plural
rat eater  mice eater  rats eater
Haskell et al. (2003) argue that phonological and semantic factors influence
the acceptability of different compounds (though see Berent & Pinker, 2008,
for a counterblast). In any event, according to level ordering, we should be
equally happy with both rat eater and mice eater. But this is not the case. Not
only do adults avoid producing compounds like mice eater (Murphy, 2000),
they are not that keen when they hear them either (Haskell et al., 2003). So
why do children produce compounds like mice eater? Buck-Gengler, Menn
& Healy (2004) focus on the preference even young children have for
singular forms within compounds. Gordon (1985) showed that, even when
the child has just named a group of rats in the plural, their subsequent
compound will be rat eater, with a singular rat modifier. Access to the
singular form (rat) is straightforward, even when the plural form (rats) has
been produced. But access to the singular form may be less straightforward
for irregulars. Having named a group of mice, it is more likely that a
subsequent compound will be mice eater, because the singular form (mouse)
is relatively difficult to access when first primed with mice. Haskell et al.
(2003) report reaction time data from adults in support of this idea. What we
need now, therefore, are some equivalent data from young children. In any
event, we have seen that children’s understanding of how compounds are
assembled and interpreted develops gradually over the pre-school years. In
the next section, we will see how morphological knowledge continues to
grow past the age of five and influence development in other domains of
language and learning.
Morphology in the school years
Morphological awareness
Once children have acquired some knowledge of morphology, they will find
themselves in possession of a very useful tool for learning in other domains,
including the lexicon, literacy and spelling. But children differ in their level
of morphological prowess and the rate at which they learn. And these
individual differences have a lasting impact on later linguistic development.
This was demonstrated in a study of Finnish inflection, which, you will
remember, allows for more than 2,000 possible nominal forms. Lyytinen &
Lyytinen (2004) showed that children with a more advanced knowledge of
inflection at two years have a higher vocabulary level at five years. You can
also see the impact of complex morphology in any Helsinki supermarket. The
aisles are littered with screaming toddlers who’ve just been told they have to
learn Finnish morphology. Individual differences can also be seen in children
acquiring Cantonese, Mandarin and Korean (McBride-Chang, Tardif, Cho,
Shu, Fletcher, Stokes, Wong & Leung, 2008). In these languages,
compounding is enormously productive as a way of creating new words. In
Mandarin, for example, baseball (‘bang-qiu’) emerged from the fusion of the
words for stick (‘bang’) and ball (‘qiu’). And there are thousands more where
that came from. In fact, McBride-Chang et al. (2008) argue that
compounding is so prevalent in these languages that this aspect of
morphological knowledge is essential for the comprehension and production
of even minimally complex words. This study demonstrated that knowledge
of morphology and lexical knowledge are intimately linked. Four- and five-
year-olds with a large vocabulary do well on tests of morphological
awareness when tested one year later. Causation runs in the opposite
direction also. A high level of morphological awareness at four years predicts
a more extensive vocabulary by age five. Hence, these two abilities feed off
each other to encourage mutual growth. The critical factor seems to be
metalinguistic awareness, specifically, morphological awareness. McBride-
Chang et al. tested children’s awareness of and ability to manipulate
morphological structures. They did not simply measure how accurate
children were in their use of morphemes. Metalinguistic skills clearly provide
a significant boost to the child’s learning in both lexical and morphological
domains.
Connections with vocabulary, reading and spelling
Metalinguistic skills are sometimes exploited explicitly in classroom settings.
The child’s ability to reflect consciously on language structure provides
teachers with an opportunity to intervene systematically and supply useful
information. Nippold (2016) provides examples from American textbooks
which show how morphology is sometimes incorporated into the school
curriculum. For example, one student workbook provides a table of
derivational suffixes, with definitions and examples of the following kind:
-able ‘capable of being, worthy of,’ as in lovable
-er ‘one who performs an action,’ as in baker
-ful ‘full of, characterized by,’ as in painful
Another text described by Nippold (2016) provides step-by-step instructions
on how to determine the meaning of an unfamiliar compound word, starting
out with the injunction to segment the word into its constituent roots (for
example, rowboat = row/boat, snowstorm = snow/storm). There is, then, a
clear expectation that explicit instruction will enhance both literacy and word
learning. As we have seen, evidence from child language research is
beginning to confirm the validity of these assumptions.
One recent study has demonstrated that morphological awareness can be
enhanced through direct intervention in the classroom. Ravid & Geiger
(2009) implemented a three-month programme with children aged 9–10 years
in Israeli schools, with the aim of increasing awareness of Hebrew
morphology. This study is distinguished by the decision to incorporate
linguistic humour into the teaching materials. For example, a particular noun
pattern was introduced, CaCéCet, which is used in Hebrew to designate
different diseases (where C stands for consonant). Children were encouraged
to play around with this format and come up with novel diseases of their own,
one such being dabéret, which glosses as ‘talkativeness’. You may not split
your sides at this sample of merry-making, but I guess you had to be there.
To nine-year-old speakers of Hebrew, this is hilarious. More to the point, it is
instructive. Post-intervention, child scores on tests of inflection and
derivation, and on their awareness of these categories, had increased
significantly. Morphological intervention has recently proven effective with
children learning Arabic, also in Israeli schools (Taha & Saiegh-Haddad,
2016).
Morphological awareness in teachers
In contrast to explicit teaching, there are settings where morphology is largely
absent from teachers’ training and understanding. One study in the UK found
that, in a sample of 51 London teachers, not one used the word morpheme,
either in the classroom or in interviews about their practice (Hurry, Nunes,
Bryant, Pretzlik, Parker, Curno & Midgley, 2005). Many of these teachers
did have some explicit awareness of words like prefix and suffix, but the
fundamental importance of morphemes, including their crucial role in
generating complex words, was neither remarked upon nor observed in
classroom practice. Curiously, this lack of awareness existed, despite explicit
reference to morphology in the UK National Curriculum. The story has a
happy ending, though. Hurry et al. (2005) incorporated an intervention
programme for these teachers into their study, which proved to be very
successful. In particular, increased morphological awareness in teachers was
associated with subsequent gains in the spelling prowess of the children they
taught (aged 7–11 years).
The impact of morphological knowledge on spelling is not confined to
English. Similar effects have been witnessed for Hebrew (Levin, Ravid &
Rapaport, 2001; Gillis & Ravid, 2006). Moreover, there is evidence that
children as young as five years recruit their knowledge of inflection and
derivation in the service of spelling (see Pacton & Deacon, 2008). At the
same time, children with depressed morphological skills, including those with
dyslexia, experience problems in reading and spelling (Siegel, 2008). In a
similar vein, Freyd & Baron (1982) report quite dramatic differences in levels
of morphological awareness among different groups of children. They
showed that advanced 10-year-olds were significantly better at defining
complex derived words than average 13-year-olds. Tellingly, the younger,
precocious children actively analysed derived words into their constituent
morphemes. Many of the older children, meanwhile, showed no evidence of
using morphological knowledge to help them define words like numberless,
endurance, invaluable and solidify. These older children did improve after a
classroom intervention, though. It is possible, therefore, that for some
children, morphological awareness develops spontaneously, while for others,
explicit teaching may be required. Overall, we see that significant individual
differences exist in levels of morphological awareness among children. And
these differences have a direct impact, not only on the ability to analyse
words, but also on the development of related skills, including reading and
spelling.
In a Nutshell
Children acquire three different processes for producing and
understanding complex words: inflection; derivation; and
compounding.
In the early stages, children are ‘whole-word’ learners. They
produce morphologically complex words without any
appreciation of their internal structure.
English past tense inflection has dominated research over the
past 30 years, with debate centred on the regular–irregular
distinction:
The single-route model argues that both regular and
irregular past tense forms can be learned by a single
process, using a pattern associator that learns all new
forms by analogy with known verbs. Models that
achieve this kind of learning nevertheless have
problems learning the past tense of new or unusual
verbs that differ from anything they have been trained
on previously.
The dual-route model argues that regular forms are
assembled online through a grammatical process (‘add -
ed’), while irregular forms are stored as whole-word
forms in the lexicon. But the regular–irregular
distinction may be illusory. It may look defensible for
English, but far less so for other languages, like Polish.
Derivational morphemes differ in their degree of
productivity, with more productive forms being acquired
early.
The interpretation of new compounds is facilitated in three
ways: (1) use of contextual information; (2) use of linguistic
context; and (3) drawing analogies between new compounds
and previously learned ones.
For complex compounds, level ordering theory predicts that
compounds must be assembled before inflection takes place
(rat + eater → rat eater; then rat eater + -s → rat eaters).
Inflection should not be possible before compounding (rat +
s → rats; then rats + eater →*rats eater).
Even three-year-old children avoid *rats eater
compounds.
The argument that level-ordering is innate has been
challenged. Recent evidence suggests that children
gradually learn the constraints governing the formation
of complex words.
Children can develop an explicit awareness of morphology in
the school years, sometimes through explicit teaching.
 Knowledge of morphology can help children develop in
other domains, including vocabulary, reading and spelling.
further reading
Pinker, S. (1999). Words and rules: The ingredients of language.
New York: Perennial.
One thing you are guaranteed in a book by Steven Pinker is a tidal
wave of fascinating detail. No argument is left to its own devices.
All are supported by evidence. At the same time you will have no
trouble picking out the broad themes in Pinker’s defence of
Words and Rules theory.
Seidenberg, M.S. (1992). Connectionism without tears. In S.
Davis (Ed.), Connectionism: Theory and practice (pp. 84–122).
New York: Oxford University Press.
I found this chapter too late, so my own tears flowed freely. A bit
like when I first saw Bambi. Save money on Kleenex and read
this chapter. It may be getting on in years, but it still provides a
good starting point for the connectionist take on language
learning.
websites
Jean Berko Gleason’s academic homepage:
www.bu.edu/psych/faculty/gleason/
A pioneer in the field of child language research, Jean Berko
Gleason’s wugs and nizzes remain relevant more than half a
century after their first appearance. There is a link on this site
to Berko (1958), the original article in Word, which is still
well worth reading. You can also access the original pictures
of her novel creatures. Don’t expect great art, but do expect
great ingenuity and a lasting influence.
Still want more? For links to online resources relevant to this
chapter and a quiz to test your understanding, visit the companion
website at https://study.sagepub.com/saxton2e
8 Linguistic Nativism: To the Grammar
Born
Contents
Universal Grammar 206
The problem of linguistic diversity 209
Core versus periphery 210
Parameters of variation 211
Setting parameters: Triggers 212
Arguments for linguistic nativism 213
Some initial observations 213
Limited exposure to linguistic input 214
No direct instruction 214
Ease and speed of language acquisition 215
The poverty of stimulus argument 216
Plato’s problem 216
Degenerate input 217
Negative evidence: Corrective input for grammatical errors 217
Knowledge in the absence of experience: The case of structure
dependence 218
The origins of structure dependence 221
The imitation of grammatical structures 222
Evidence from children 223
Poverty of the stimulus: Summary 226
The contents of UG: What precisely is innate? 227
Conclusion 230
Overview
By the end of this chapter you should be familiar with the concept of
Universal Grammar (UG) and the main arguments in favour of the idea
that it is innate. You should be able to describe the problem of
linguistic diversity: if we are born with the same knowledge of
grammar, why do the grammars of the world’s languages seem so
different? You should be able to describe two approaches to this
problem:
core UG versus peripheral, language-specific aspects of grammar
parameters: UG does allow for some variation among languages
 You should also develop an awareness that arguments in favour of UG
rest largely on the assumption that the child’s linguistic environment is
impoverished: the information it supplies is too meagre to explain the
rich knowledge of grammar that every typical child attains. In this
regard, you should be able to describe one of Chomsky’s best-known
examples of a property of grammar (structure dependence) that we all
seem to acquire despite limited experience. You should also be able to
summarize other arguments in favour of linguistic nativism: language
acquisition seems quick and effortless, even though children are not
explicitly taught anything about language, and even though it is
assumed they receive no help in the form of corrections for
grammatical errors. Finally, you should gain some idea about how well
supported, empirically, Chomsky’s theory of Universal Grammar is.
Universal Grammar
In this chapter, we will focus on the ‘nature side’ of the nature–nurture
argument (see Box 8.1). In particular, we will examine the theory that
knowledge of grammar is genetically determined. If your eyebrows shoot up
at this idea, you are not alone. Many people find this notion deeply
implausible. Nevertheless, the concept of innate grammar has proven
remarkably tenacious over the past half century, due in large part to the
efforts of Noam Chomsky (see Box 1.4). He argues that children are born
with a Universal Grammar, or UG. The idea is that ‘certain aspects of our
knowledge and understanding are innate, part of our biological endowment,
genetically determined, on a par with the elements of our common nature that
causes us to grow arms and legs rather than wings’ (Chomsky, 1988: 4). On
this view, language is an organ of the body, albeit a mental organ, and like
other organs it grows according to a genetic blueprint. Notice, therefore, that
innate does not necessarily mean ‘present at birth’. Babies are not born
talking. But nor are they born with adult arms and legs. The development of
arms and legs and (for Chomsky) language, depends on maturation: growth
that unfolds over time according to a biological timetable. On this view, we
are all born with the same potential for language (Universal Grammar), which
then develops into knowledge of particular languages (Swahili, Greek,
Gujarati, and so on) according to our individual experience. UG thus
corresponds to the initial state of the language faculty at birth. Language
acquisition is the process whereby ‘a person proceeds from a genetically
determined initial state S0 through a sequence of states S1, S2, …, finally
reaching a ‘steady state’ SS which then seems to change only marginally (say,
by the addition of new vocabulary)’ (Chomsky, 1980b: 37). The steady state
(SS) corresponds to knowledge of a recognizable language like Polish or
Vietnamese. Chomsky summarizes the difference between UG and the
particular languages we end up acquiring in the following way:
The grammar of a particular language is an account of the state of the
language faculty after it has been presented with the data of experience;
universal grammar is an account of the initial state of the language
faculty before any experience. (Chomsky, 1988: 61)
Box 8.1 The Origins of ‘Nature–Nurture’
‘Nature–nurture’ is an elegant phrase that has entered the
language. It encapsulates an important issue in the life sciences
and provides an easy reference point for anyone with an interest
in the origins of human behaviour. To students of psychology, it
might sometimes seem that this phrase is quite modern,
coinciding with the massive increase in research on genetics over
the past 50 years or so. But in fact, the phrase is very old. Francis
Galton, the nineteenth century eugenicist, is often credited as the
first to popularize this phrase (Fancher, 1979). Others (including
Conley, 1984) point out that Shakespeare juxtaposes nature and
nurture in about 1610 in his play The Tempest. In Act IV, Scene I
of this play, the central character, Prospero, fulminates against
Caliban:
A devil, a born devil, on whose nature
Nurture can never stick.
But it turns out that Shakespeare was not the originator of this
famous phrase. In 1582, about 30 years before The Tempest, an
Elizabethan head teacher called Richard Mulcaster argued that
‘Nature makes the boy toward; nurture sees him forward’. As
Teigen (1984: 363) observes, Mulcaster stressed the ‘harmony
and mutual dependence’ of nature and nurture. These two
concepts were not, originally, set up in opposition to one another
in the manner so prevalent these days.
How plausible is the notion of a universal grammar? If we sign up for this
idea, we must tackle three complementary questions:
 1. How does UG theory cope with linguistic diversity – the fact that the
grammars of the world’s languages seem to differ greatly?
2. Why suppose that UG is present at birth?
3. Precisely which aspects of grammar are universal?
In the following sections, we will consider each of these questions in turn.
Before we do, though, it is worth pointing out that the concept of innate is
not straightforward (Maratsos, 1999; Scholz & Pullum, 2006). Mameli &
Bateson (2006) offer 26 different definitions of ‘innate’, none of which is
entirely satisfactory. They identify the following definition as the one
adhered to by Chomsky:
A trait is innate if and only if its development doesn’t involve the
extraction of information from the environment. (Mameli & Bateson,
2006: 159)
On this approach, the proof that language is innate comes from a
demonstration that the child’s linguistic environment is not responsible for
language acquisition. If the information required for learning does not come
from the environment, then it must derive from the genome. Of note,
Chomsky’s argument does not rely on any simple equation of ‘innate’ with
the concept of ‘genetic determination’. Rather, ‘innate’ is equated with a lack
of environmental influence. One problem with this definition is that it does
not cover all traits that one finds in nature. Mameli & Bateson (2006) point
out that scars or calluses are not readily considered innate, but neither do they
stem from the extraction of information from the environment. With regard to
language, this definition faces a further profound empirical problem.
Currently, we have no principled way of distinguishing experience that
provides information for learning from experience that merely provides
developmental support (Lehrman, 1970).
We will explore Chomsky’s ideas about lack of information in the
environment below. In the meantime, Jackendoff (1997: 6) encapsulates the
nativist position for us: ‘the human genome specifies the growth of the brain
in such a way that UG is an emergent functional property of the neural
wiring’. On this view, therefore, the form that language takes is inevitable,
because it is predetermined in the human genome. In his later writings,
though, Jackendoff (2002: 71–72) blurs the edges of Chomsky’s position, by
dropping the equation of UG with innate knowledge:
Universal Grammar is not the grammar of any single language: it is the
prespecification in the brain that permits the learning of language to take
 place. So the grammar acquiring capacity is what Chomsky claims is
innate.
There is a fundamental difference between innate knowledge and an innate
grammar acquiring capacity. Inheriting knowledge of the category of noun is
not the same as being born with the capacity to acquire this knowledge. It is
worth repeating Chomsky’s own words on this issue:
certain aspects of our knowledge and understanding are innate, part of
our biological endowment, genetically determined. (Chomsky, 1988: 4)
If we compare these two definitions, we see that Jackendoff presents us with
a far less radical version of the nativist position (cf., Pearl, 2014). It is
patently obvious that we are endowed with the mental machinery that permits
the acquisition of grammar. Where researchers still disagree is on the issue of
how specialized that mental machinery is. Nativists argue that the grammar
acquiring capacity is dedicated exclusively to language. Non-nativists, on the
other hand, consider grammar to be simply one of many mental achievements
acquired by general-purpose cognitive learning mechanisms. We will
examine the non-nativist approach to grammar in the next chapter. In the
meantime, we can note that Chomsky promulgates a somewhat startling
belief: some of our knowledge of grammar is innate. And lest we think that
Chomsky is out on a limb here, swaying in the breeze all by himself, rest
assured that he is not alone:
The interesting claim that nativism makes is that there are innate ideas,
not simply that there are innate learning mechanisms. (Valian, 2014: 90)
Discussion point 8.1: Nature–nurture beyond
language
The nature–nurture question extends far beyond the confines of child
language. Think, for example, of the controversy that rages over race
and IQ. Herrnstein & Murray (1994), in particular, are famous
(infamous, even) for suggesting that African Americans not only do
less well on IQ tests than white Americans, but that genetic factors play
a significant role in explaining the observed differences (see
Mackintosh, 1998, for a survey of this issue that is both balanced and
thorough).
List different domains of human behaviour. Start your list with
language and IQ, but add as many others as you can: different
facets of personality, perhaps, musical ability, and …?
 Review your list and consider your intuitions about the extent to
which genetic factors are responsible both for similarities and
differences among people.
Do you have different intuitions about the role of genes for
different domains of behaviour? If so, why?
In what ways might the nature–nurture issue, as it applies to
language, be more than just a scientific issue? What political and
social implications might there be?
The problem of linguistic diversity
How can we all be born with the same knowledge of grammar, when people
around the world grow up speaking such a massive array of different
languages? One recent estimate puts the figure at 6,912 separate languages
(Gordon, 2005). To the casual observer, the grammatical systems of the
world’s languages seem quite different from one another. For example,
Japanese treat adjectives as though they were verbs. In Russian, verbs of
motion are marked to show whether you go on foot or by transport; they also
indicate whether you are going one-way or making a round trip. Yucatec
Maya, meanwhile, a central American language spoken in Belize and
Guatemala, places the verb at the beginning of the sentence, usually
followed by the object, then subject. Yucatec Maya, then, is a VOS
language, while English is SVO (My friend dances like a lunatic could be
glossed something like: dances – like a lunatic – my friend in Yucatec Maya).
A quick world tour would reveal many grammatical idiosyncrasies that seem
exotic from the standpoint of English and which differentiate one language
from another. But why do languages differ in these ways? Linguists have
grappled with this question for centuries. One explanation, offered by von
Humboldt, suggested that geopolitical factors underlay linguistic diversity:
‘the structure of languages differs among mankind, because and insofar as the
mental individuality of nations is itself different’ (1836/1988: 47). Social and
political factors have also been emphasized by anthropologists who paint a
picture of each human language as a repository of linguistic and cultural
uniqueness. UG theorists adopt a radically different perspective, offering
three different observations on the issue of linguistic diversity:
1. Fundamentally, human languages share many grammatical properties
that may not always be obvious at first glance.
2. There are some differences between the grammars of different
 languages, but these are not part of UG, and are therefore peripheral.
3. UG offers a menu of options for certain properties of grammar
(parameters of variation). Different languages have ‘chosen’ different
options, but they all fall within UG.
To take the first point, there are many ways in which languages resemble one
another. To take a simple example, all languages seem to have grammatical
categories like noun and verb. Other categories are also widespread,
including, preposition, determiner and syntactic features like gender, tense
and number. Valian (2014) argues that it is not necessary for every language
to express every feature of UG – the absence of tense in Chinese provides a
case in point. Of course, this creates a headache. How do we establish that a
given near-universal feature is actually part of UG? And if it is part of UG,
why is it not expressed in every human language? Absolute universals are
rare (Evans & Levinson, 2009), but the hunt for them is a longstanding
project within linguistics (for example, Greenberg, 1966). In Chomsky’s
approach, when a universal property of language is discovered, it becomes a
likely candidate for inclusion as part of UG. It should also be borne in mind
that properties of UG may not be obvious from the surface properties of
different languages. We may need to dig beneath the surface to demonstrate
the fundamental similarities that unite different languages. We will look in
more detail below at one aspect of UG, known as structure dependence. In
the meantime, let’s consider the second point from our list above.
Core versus periphery
For Chomsky, Universal Grammar comprises a set of core linguistic
properties, common to all human languages. But this leaves a considerable
number of other, peripheral properties of grammar that are not part of UG.
These latter aspects of grammar can be unique to particular languages (or
groups of languages). For example, languages often have idiomatic
expressions whose meaning is quite different from their literal interpretation:
Tim’s got ants in his pants or Naomi only made it by the skin of her teeth. The
periphery also contains examples that violate the principles of UG. For
instance, the English sentence The more the merrier should not be
grammatical because it lacks a verb (Cook, 1988). Similarly, a sentence like
Myself, I wouldn’t go there should not be possible (Salkie, 1990). According
to UG, reflexive elements like myself should refer back to a person or thing
that has already been introduced, as in Patricia made the cheese soufflé
herself. In this example, the antecedent of herself is Patricia. But in Myself, I
wouldn’t go there, myself does not have an antecedent. This latter example
(and others like it) should set alarm bells ringing. If Universal Grammar
constrains the form of human languages, how can we violate its principles
with such alacrity? Moreover, the core–periphery distinction begins to look
like a convenient way of disposing of awkward examples that do not fit the
theory. Quirky aspects of grammar that are not easily explained can be
assigned to the periphery.
The concept of peripheral grammar goes some way to explaining the
diversity we see, both within and across different languages. It is worth
pointing out that anything on the periphery presents a learning problem. No
helping hand is available from Universal Grammar, which suggests that the
child must be equipped with general learning mechanisms that are not
dedicated specially to language, but which are capable of acquiring at least
some aspects of grammar. How these general learning mechanisms might
operate is an issue that has never been broached by UG theorists. The
existence of a periphery, though, demonstrates that Universal Grammar
cannot account for the acquisition of all aspects of grammar in human
languages.
Parameters of variation
Some of the variety between languages is encompassed within UG via the
notion of parameters of variation (Chomsky, 1981). The idea is that UG
presents a limited range of options or parameters. On this view, the ‘“menu
choices” languages opt for can vary, but what’s on the menu does not’
(Berwick & Chomsky, 2016: 55). Each individual language (like English or
Japanese) has ‘chosen’ one of the parameter settings available. An example
will help illustrate this idea (see also Box 1.2 on Syntactic Categories). In
English, verb phrases can be formed by a verb followed by a noun phrase:
Verb Phrase → Verb + Noun Phrase
 eat + a biscuit
 take + your time
 open + the door
The head of the verb phrase (VP) in these examples is the verb (eat, take,
open) and in each case it is followed by a complement noun phrase. Our
simple rule captures this regularity of English: VP → V + NP. But this rule
misses out on the chance to capture a much broader generalization. The
interesting property of grammar here is the order of heads and complements
within phrases. In our Verb Phrase, the order is ‘head followed by
complement’. And it turns out that, in English, this head + complement
ordering applies to every kind of phrase (including Noun Phrase, Adjective
Phrase and Prepositional Phrase). In other languages, like Japanese, the order
is reversed: complement + head. The so-called Head Parameter can then be
invoked to capture this property of human languages. Universal Grammar
presents two choices: head-first or head-last. For the language-learning child,
life is made much simpler than having to learn the rules for each kind of noun
phrase, each kind of verb phrase, and so on, separately. Instead, the child
simply needs to discover the order of heads and complements in the language
they are exposed to. Once the Head Parameter is set, the child will know the
order of constituents in every kind of grammatical phrase. Thus, parameters
can be seen as a kind of switch that needs to be set one way or the other. In
this way, parameters ‘reduce the difficulty of the learning problem’ for the
child (Gibson & Wexler, 1994: 407).
Setting parameters: Triggers
We must next explain how parameters are set, that is, we must explain how
the child comes to know which kind of language they are learning (e.g., head-
first or head-last). The answer provided by UG theory is through the action of
a trigger. ‘Parameters are alleged to be unacquired. What is acquired is a
particular setting of a parameter by the process of being triggered by an
environmental stimulus’ (Scholz & Pullum, 2006: 63). The trigger is
generally taken to be exposure to a relevant phrase in the language being
learned (see the section on imitating grammar, below). The child exposed to
Japanese will hear phrases where the head comes last, while the child
learning English will hear head-first phrases. In principle, exposure to a
single relevant example could trigger the setting of the parameter, bringing
with it a wide range of syntactic knowledge. At this point, though, we run
into a problem. The notion of ‘simple exposure’ as a trigger for this kind of
learning is deceptive. In order for the head parameter to be triggered, the
child would need to have learnt a substantial amount about language. They
have to recognize grammatical phrases for what they are, which, in turn,
requires the child to understand in advance the notions of head and
complement. Once these concepts are in place, then exposure to a phrase (the
trigger) will allow the child to recognize the order of these critical elements
(head first or last) and hence set the parameter accordingly. But how does the
child come to know what heads and complements are in the first place? UG
theory currently offers no answer to this question, which is unfortunate,
because it goes to the heart of the problem facing the language-learning child.
Universal Grammar may furnish the child with knowledge about grammar,
but somehow the child has to recognize the expression of that knowledge in
the language they experience. This conundrum has been described recently as
the linking problem (Ambridge, Pine & Lieven, 2014) and strikes at the heart
of the nativist enterprise. Providing the child with innate knowledge of
grammar seems to cut through the learning problem thrown up by a
supposedly impoverished input. But it throws up the equally vexing problem
of how that innate knowledge makes contact with the linguistic input children
hear. On this view, Universal Grammar does not simplify the learning
problem facing the child.
The concept of trigger being considered here assumes that some information
in the environment, the trigger, is directly relevant to what is acquired (in this
case, an actual example of head–complement ordering). But the term trigger
has been used with different meanings over the years (Stich, 1975). An
alternative view is to look at triggers as a kind of catalyst, a concept
borrowed from chemistry to denote something that causes a chemical
reaction, but which is itself unaffected by it. Either way, though, the child is
faced with the prior problem of identifying both heads and complements for
what they are. As Smith (2004: 80) admits, this leads to a ‘vacuous claim –
the child only needs to learn what it needs to learn’. The unanswered question
remains: how does the child link their innate knowledge to the input they
hear?
The concept of parameters seems to present an elegant solution to the
problem of linguistic diversity. A wide range of seemingly disparate facts
about different languages can be captured by a single parameter with a very
limited range of settings (usually just two). At the same time, it is interesting
to note that after almost 30 years of effort, disappointingly few parameters
have actually been established within linguistic theory. Fewer still have been
investigated empirically using data from child language development (though
see Hyams, 1992; Snyder, 2007). Critically, there is currently no convincing
explanation for how the child might set parameters.
Arguments for linguistic nativism
Some initial observations
We can start with the observation that, to some eyes, language development
looks very fast. As noted in Chapter 1, every typically developing infant
undergoes a dramatic transformation. Capable of little more than belching
and wailing at birth, it is not so long before the child can hold their own in
conversation, expressing their opinions, desires and interests. It has been
suggested that children ‘display virtual adult competence by age 3 or 4’
(McGilvray, 2006: 107). By ‘competence’, McGilvray has in mind the mental
representation of a mature system of adult grammar. While many would
regard McGilvray’s assertion as extreme, most would nevertheless agree that
children do seem to soak up language like a sponge. Moreover, nativists
emphasize that children achieve the feat of language acquisition with very
little help from parents and others. Contrary to the picture of the input
presented in Chapter 4, it is argued that the input for language learning is
seriously underdetermined. This means that the input lacks critical
information needed to permit successful language acquisition.
To summarize, then, nativists present us with a paradox. On the one hand,
grammar is portrayed as a highly abstract, complex system of knowledge. On
the other hand, the input is too poor in quality to explain how language could
be learned. Despite this contradiction, children nevertheless seem to acquire
language, both rapidly and with ease. We are thus skewered by a simple point
of logic: ‘if the child’s linguistic experience does not provide the basis for
establishing a particular aspect of linguistic knowledge, there must be another
source for that knowledge’ (Lightfoot, 2005: 50). As we have seen, that
alternative source of knowledge is held to be Universal Grammar. Of course,
this argument is only as sound as the assumptions on which it rests:
1. limited exposure to linguistic input
2. no teaching: children acquire language with no help from parents or
others
3. speed: children acquire language quickly
4. ease: language is acquired with little ostensible effort
5. poverty of stimulus: the quality of information available in the input is
too meagre to account for the rich system of language that children
acquire.
We shall consider each one of these assumptions in more detail.
Limited exposure to linguistic input
The first assumption is encapsulated by Chomsky (1975: 4) with his
suggestion that ‘a normal child acquires … [language] knowledge on
relatively slight exposure’. The amount of input available appears very
limited if we compare it to the full range of sentences the child might
possibly be exposed to. In fact, there is literally no end to the number of
different possible sentences we might produce or hear. The rules of grammar
provide a toolkit for producing an infinite number of sentences. It is always
possible to produce a novel sentence, one that has never before been uttered,
by simply adding another phrase. For example, The cat in the hat can become
The cat in the hat on the moon, which can, in turn, become The cat in the hat
on the moon ate a sandwich, and so on. This property of language was
encapsulated by the German philosopher, Wilhem von Humboldt
(1836/1988: 91), who observed that language can ‘make infinite employment
of finite means’. We see, then, that the language-learning child is exposed to
a fraction only of the possible sentences. Moreover, each child is exposed to a
unique, and therefore idiosyncratic, sample. To compensate for the child’s
lack of input, Universal Grammar is invoked by nativist theorists.
Before we subscribe to Chomsky’s position we should consider how
appropriate it is to present the child’s learning task in this way. The child is
not attempting the impossible, by trying to acquire every sentence of the
language. Instead, the goal is to acquire the system of rules that will allow
any sentence to be interpreted. And as von Humboldt points out, the system
of rules is finite. The child therefore uses a finite sample of sentences to
acquire a finite system of rules. Put in these terms, the task facing the child
seems more tractable. Moreover, one might argue that children are exposed to
a very large amount of language. On a standard estimate, children are
exposed to language for about 10 hours per day (Clark, 2003; Tomasello &
Stahl, 2004) which must amount to millions of words within the first five
years (I say ‘must’ because no-one has actually counted). As we saw in
Chapter 4, the amount of input varies according to the child’s individual
circumstances. And the differences do impact on the pace of language
development. But even in cases of relative deprivation, the number of
sentences heard in the early years must still run into the millions. Viewed in
this way, the child’s input sample does not seem all that ‘slight’.
No direct instruction
The second assumption underpinning UG concerns the role of instruction.
Chomsky argues that no-one teaches pre-school children how to talk. Parents
and teachers often object at this point, being very clear in their own minds
that they do teach their children language. For older children, there is some
justification for this view. At school, children are explicitly taught hundreds,
if not thousands, of new words. Indeed, intervention programmes have been
devised to increase children’s vocabularies, on the basis that explicit
instruction is both possible and beneficial (for example, Nash & Snowling,
2006). But grammar, not vocabulary, is the focus here. And more to the
point, most researchers do not focus on school-aged children. As noted
above, by the age of four or five the child’s knowledge of grammar rivals that
of any adult. So any explicit teaching at school could only tinker round the
edges of an already mature system. In any event, direct instruction is
probably impossible. How would we give a grammar lesson to a one-year-
old? At this age, the child does not possess the language to understand
explicit instruction in language about language. That is, Eve lacks the
requisite metalinguistic abilities. Typical estimates suggest that the ability to
think about and analyse language structure (for example, work out the
number of syllables in a word) does not emerge in any serious way until
about the fourth or fifth year (Bowey, 2005; though see Clark, 2003). But the
child seems to get by just fine without sophisticated metalinguistic skill or
direct instruction, as we can see from Eve’s linguistic prowess.
But Chomsky may be leading us down a blind alley. His assumption is that
direct instruction would be the best source of input for a language learner.
But thousands of adult second language learners, struggling in evening
classes, can attest that overt teaching about grammar does not bring them to
the point of fluency, however hard they try. We have even less reason to
suppose that direct instruction would be the best resource for the toddler. If
the metaphor of teaching is at all helpful, then we might cast Child Directed
Speech as a better form of instruction. As we saw in Chapter 4, parents
modify and simplify their speech in myriad ways to support the acquisition
needs of the child (Clark, 2003). CDS is closely geared to the language-
learning needs of a cognitively and linguistically naive learner, in a way that
explicit teaching could never be.
Ease and speed of language acquisition
We can take the next two assumptions together, namely, that language is
acquired with ease and at high speed. Moreover, the facility with language
learning is witnessed in all typically developing children, irrespective of their
general intelligence or memory capacity. Von Humboldt (1836/1988: 58)
made this observation in the nineteenth century:
all children, under the most diverse conditions, speak and understand at
about the same age, varying only within a brief time-span.
In the modern era, Chomsky presents a twin-pronged approach to the
argument for innate language universals. On the one hand, he stresses what
he takes to be the Herculean labour that language learning presents in its
complexity, extent and level of abstractness. And on the other, he suggests
that children saunter through the task, barely breaking sweat. The problem,
though, is that we cannot judge, in any straightforward manner, whether
learning is quick or slow. One way might be to compare language acquisition
with learning in other cognitive domains during the same developmental
period. Thus, we might observe that many four-year-olds lack the fine motor
skills and physical co-ordination required to tie their own shoe laces. They
also typically lack conservation, the logical ability to judge that, when a ball
of clay is rolled out into a sausage, no clay is either lost or gained in the
process (Piaget & Inhelder, 1974). And many four-year-olds cannot yet count
to 20 (Fluck & Henderson, 1996). At the same time, those same children can
produce sentences that approach or surpass 20 words in length with no
discernible difficulty.
Alex aged four years:
If any princesses are catched they have to save her and climb up the pole
and the king will say ‘Thank you’ and ‘You can marry her’.
Are you looking for the green one who goes at the top of the tree?
It’s very good because it’s got those things who I’ve never had before.
Some frogs are like chameleons because they turn to green on leaves
and turn to brown on branches.
The contrast between counting and sentence construction is dramatic. The
ability to recall 20 number words in the correct sequence seems relatively
straightforward. It requires no complex algorithm or rules to place each
number name in the correct order. It is more like placing beads on a string,
one after the other. Each number name occurs in the same place in the
sequence every time. Moreover, children receive a massive amount of
practice with this particular sequence of words. Yet many three- and four-
year-olds cannot manage this task. In contrast, Alex’s sentences (including
the first, with 27 words) require a sophisticated understanding of grammatical
principles. One cannot just throw the words together in any old order.
Of course, counting is not the only cognitive capacity acquired by children.
Other aspects of cognition make language learning look comparatively slow.
For example, visual acuity – the ability to see objects clearly – develops
rapidly in the first few months of life and has reached near adult levels by the
age of two years (Courage & Adams, 1990). Similarly, the child’s
understanding of object permanence is adult-like within the first two years of
life (Ruffman, Slade & Redman, 2005). Thus, toddlers appreciate that an
object continues to exist even when it is concealed, and that it remains the
same object regardless of which angle it is viewed from or how it is
illuminated. Many even argue that a mature knowledge of object permanence
is evident within the first few weeks of life (Baillargeon, Li, Luo & Wang,
2006). Examples from general cognition demonstrate that there is no
straightforward way of judging how quickly a particular mental capacity is
acquired. One is left, then, with the rather unsatisfactory conclusion, that
‘children appear to acquire language quickly’ (McGilvray, 2006: 107, my
emphasis). To summarize so far, there is something intuitively appealing
about Chomsky’s observations about language learning. At first blush, it does
seem that children acquire language quickly, effortlessly, on limited exposure
and with no direct teaching. But on closer examination, none of these
assumptions is well supported. We must turn instead to the heavy artillery in
the nativist armoury: the argument from the poverty of the stimulus.
The poverty of stimulus argument
Plato’s problem
The concept of innate knowledge stems back to Ancient Greece and the
Socratic Dialogues of Plato (427–347 BC). Socrates leads a slave boy
through a series of questions, and demonstrates to the slave’s owner, Meno,
that the boy understands the basic principles of geometry. Through his
questioning, Socrates guides the boy to produce a square with an area of 82
by extending one that is 42. Remarkably, the boy achieves this, even though
he has never been taught geometry. In consequence, Plato concludes that the
boy knows more than can be explained on the basis of experience; the boy’s
knowledge must be inherent within his soul. Chomsky’s central contribution
to twentieth-century thought was to argue that the child faces ‘Plato’s
problem’ (Chomsky, 1988) within the domain of language acquisition. The
argument is that every typical child knows far more about language than
could possibly be learned from experience (Chomsky, 1980b; Marcus, 1993;
Pinker, 1994; Uriagereka, 1998; Lasnik, 2000). This is because the input (or
‘stimulus’) is held to be inherently impoverished. In this section, we will
consider three aspects of the poverty of stimulus argument:
1. Degenerate input: parents provide a poor model for language learning.
2. Negative evidence: parents do not correct their children’s errors.
 3. Structure dependence: children have knowledge of certain aspects of
grammar, despite a lack of evidence for them in the input.
It is not looking good for the input. First, it’s degenerate. Then it is poverty-
stricken. Next thing we know, it’ll be out on the streets, begging. We have
covered the first two points in Chapter 4, so we will provide only a brief
summary here. The third point, though, on the grammatical property of
structure dependence is worth expanding in some detail. With this example,
Chomsky provides a modern version of the Socratic dialogues, in which the
reader is cast as the slave boy, guided towards a revelation about innate ideas.
Degenerate input
If I wanted to learn the art of soufflé making, I would probably not ask my
husband, who did once manage to boil an egg, but then dropped it on the
floor. This is true. So I would more naturally turn to a Michelin-starred chef,
like Heston Blumenthal, the guiding principle being that if you want to learn
something, it makes sense to acquire the best possible model as an example
to learn from. What kind of model do children have for language acquisition?
You may recall that Chomsky (1965: 31) described what children hear as
‘fairly degenerate in quality’, further remarking that ‘much of the actual
speech observed consists of fragments and deviant expressions of a variety of
sorts’ (ibid.: 201). But who was Chomsky observing? As we saw in Chapter
4, it turns out that these remarks are appropriate when applied to adults
interacting with each other. But when adults (and others) address young
children the picture is radically different. The description of Child Directed
Speech reveals that children are exposed to a remarkably good model of
language from which to learn. The input is simplified and clarified in myriad
ways, to meet the communicative needs of a linguistically naive
conversational partner. The linguistic environment of the child is far from
being degenerate.
Negative evidence: Corrective input for
grammatical errors
The next plank in the poverty of stimulus argument is the so-called ‘no
negative evidence’ assumption. The idea is that parents do not correct their
children’s grammatical errors. That is, they receive no information about
what is, and is not, grammatical. But children must get their knowledge about
the bounds of grammar from somewhere. If the environment is lacking, then
one can conclude that this knowledge is innate. We will not dwell on the ‘no
negative evidence’ assumption here because we examined it in some depth in
Chapter 4. Suffice to say, many authors are reluctant to abandon this
assumption (for example, Cowie, 1997; Chomsky, 1999; Crain & Lillo-
Martin, 1999; Maratsos, 1999; Marcus, 1999; Rohde & Plaut, 1999; Smith,
2004; Lust, 2006). At the same time, we saw in Chapter 4 that numerous
studies now suggest that negative input is available, for every grammatical
category studied so far, for every individual child on whom data are
available, and in every culture that has been studied. So far, then, we have
seen that the input is neither degenerate nor lacking in negative input. What,
then, of the third part of this argument, the idea that children acquire complex
grammatical knowledge without ever being exposed to critical input data. We
turn to this idea in the next section.
Knowledge in the absence of experience: The case of
structure dependence
Chomsky asserts that children acquire knowledge of grammar in the absence
of experience (Berwick, Pietroski, Yankama & Chomsky, 2011). A dramatic
example is provided by deaf children who create their own system of manual
communication (known as homesign). They do this in the absence of
exposure to conventional sign language (or any other form of language)
(Hunsicker & Goldin-Meadow, 2012). Recall that 90 per cent of parents with
deaf children are not deaf themselves, so they are ill-equipped to sign
proficiently with their children (Meadow, 1980). It is argued that such
children can combine homesigns into recognisable linguistic phrases with
hierarchical structure. The absence of experience in this case seems clear. But
Chomsky would argue that all children lack the necessary experience to
acquire certain aspects of grammar. We can illustrate this idea by considering
a property of grammar known as structure dependence. Let’s start by asking
what you know about structure dependence and the rules governing question
formation in English. Being asked what you know about grammar might
make you squirm, especially if you belong to the lost generation who were
not taught anything about grammar at school. But breathe again. We are not
concerned with your explicit knowledge of grammatical rules. Rather, the
concern is with your implicit knowledge, your intuitions about what is
grammatical versus ungrammatical. My hunch is that all native English
speakers will have the same intuitions about the examples that follow. We
will pick up the notion of shared intuitions later on because it bears directly
on the poverty of stimulus argument. In the meantime, let’s take an
innocuous statement and produce the corresponding question:

What is the rule for question formation? Perhaps the simplest rule would be:
‘move the 3rd word to the front of the sentence’. But this breaks down very
quickly, even for a very similar sentence like: This newborn baby is called
Saskia. We get linguistic garbage: *Baby this newborn is called Saskia? Our
first conclusion, therefore, is that the linear position of words (1st, 2nd, 3rd,
and so on) has no bearing on the issue of question formation. An alternative
rule might be: ‘Move is to the front of the sentence’. Happily, this rule works
for both of our examples so far:
This baby is called Saskia → Is this baby called Saskia?
This newborn baby is called Saskia → Is this newborn baby called
Saskia?
We seem to be up and running, so let’s try our rule out on something more
complex. Sometimes we find two instances of is in a single sentence. Which
one do we move? In the next example, we get a well-formed question by
moving the first is.

If we try moving the second is in this example, things go badly awry:

It looks like we’re onto something. Our question rule now is: ‘Move the first
is in a sentence to the front’. But this rule breaks down rather quickly:

When we move the first is in this sentence, the result is ungrammatical. More
bothersome still, moving the second is produces a well-formed question:

So in some cases we move the first instance of is, and in others, we move the
second. Again, we see that the question formation rule cannot be based on the
linear order of elements (1st is versus 2nd is). Instead, we need to consider
the grammatical structures involved. To construct a grammatical question for
our examples, we need the following rule: ‘Move the instance of is directly
following the subject of the sentence to the front’. The subject in each of our
examples is marked in bold below.
Subject of the sentence:
1. This baby is called Saskia.
2. This newborn baby is called Saskia.
3. Saskia is the only baby who is named after a cat.
4. The baby who is named after a cat is growing whiskers.
We can see now where the phrase structure dependent comes from. Our rule
is based, or rather depends, on the grammatical structure of the subject. The
number of words in the subject phrase is irrelevant. The subject might be a
single word, as in (3), or a lengthy phrase, as in (4). The critical factor is that
the rule applies to the grammatical structure. The following point is also
worth emphasizing:
There is no logical reason why languages should use structure-
dependent rather than linear rules. Languages can easily be constructed
that use computationally simpler linear rules. (Chomsky, 1988: 46)
Exercise 8.1: BE as copula or auxiliary
So far, we have avoided talking about which grammatical category is
belongs to. In fact, it belongs to two different categories: copula and
auxiliary verb. The exercise below helps clarify the difference
between them. The verb BE has different forms (be, am, is, are, was,
were, been). Compare:
We can be happy / I am happy / You are happy / He was happy
She is happy / They were happy / They have been happy.
In addition to different forms, BE has two different grammatical
functions in English, appearing either as auxiliary verb or copula.
Auxiliary verbs are sometimes called ‘helping verbs’, since they play a
supporting role to so-called main verbs, like talk, dance or eat.
Compare I dance like Fred Astaire with I can dance like Fred Astaire
or I could dance like Fred Astaire. Other auxiliaries include do, have,
must, might, may, shall, will and would. The copula, on the other hand,
is used to link a subject to the rest of the sentence, in the absence of
any other verb. It is somewhat like the equals sign in X = Y, as in
Saskia is a baby. In the following examples, identify which forms of
BE are an auxiliary verb and which are the copula:
1. Louise is a cat lover.
 2. Ian must be mad to buy so much cat food.
3. Patricia and Colin are having kittens about the fleas in their
house.
4. Saskia is wondering what all the interest in cats is about.
5. I am wondering what cat would taste like if it were roasted.
6. Cats were worshipped in Ancient Egypt.
7. The cat that is stealing my milk has been known to steal chicken,
too.
8. Is there any reason why cat owners do not look like their pets?
The answers are at the back of the book. Do not be too downhearted if
you don’t get all of them right first time. Even child language experts
sometimes slip up on the copula versus auxiliary BE contrast (e.g.,
Pinker, 1994: 41; MacWhinney, 2004: 890).
Returning to structure dependence, we can finesse our rule. Questions are
formed by moving the copula or auxiliary verb that occurs directly after the
subject to the front of the sentence. The auxiliary does not have to be is (or
other forms of be). It could be any auxiliary verb (for example, shall, might,
will), as in the example below:

In fact, we can generalize even further, going beyond the confines of question
formation to draw the following broad conclusion: ‘all rules in all languages
are structure dependent’ (Smith, 2004: 50). For this reason, structure
dependence is held to be part of Universal Grammar.
Before we move on, it is worth inserting one note of caution. The discussion
of question formation is based on the assumption that we start with a
statement and convert it into a question by moving an auxiliary or copula.
But this kind of movement is not essential and numerous theories of grammar
get by quite happily without it (e.g., see Clark & Lappin, 2011: 36). In
addition, it has never been shown that this kind of movement corresponds to
any kind of mental operation. Many years ago, when Chomsky’s ideas first
came to prominence, experiments were conducted to see if the number of
grammatical operations – transformations – needed to convert one structure
into another was reflected in actual processing time. The results were
disappointing (Fodor & Garrett, 1967; Watt, 1970). Evidently, there is no
straightforward relationship between linguistic theory and psycholinguistic
processing. Questions might well be formed ab initio with no reference to
statements.
The origins of structure dependence
We can now get back to the poverty of the stimulus. The first thing to note is
how quick and strong our intuitions are about the sentences in the examples
above. We know, without prompting, which sentences are grammatical and
which ones are ungrammatical. I assume here that all native speakers of
English will agree with these judgements. If that is the case, we all have the
same tacit knowledge about the grammar of complex questions, and
moreover, that knowledge is structure dependent. But where did this
knowledge come from? If learning is involved, then Chomsky (1980b) argues
that we would need to be exposed to complex questions like Is Saskia the
only baby who is named after a cat? Simple questions, like Is this baby called
Saskia?, do not allow us to determine which kind of rule the speaker is
following. But Chomsky assumes that the critical evidence ‘is not available
to the child in terms of direct linguistic experience’ (Chomsky, 1980b: 115).
That is, children are supposed never to hear complex questions of the type
described above. And this time, one of Chomsky’s armchair speculations
turns out to be accurate. MacWhinney (2004) combed through the CHILDES
database looking for evidence of this structure in the speech addressed by
adults to children aged up to 5;0. But he uncovered just one single example in
something like three million utterances. It looks as though children never (or
almost never) hear such sentences. A similar search was conducted on the
archives of the Wall Street Journal by Pullum & Scholz (2002). The results
were slightly more encouraging, since about 1 per cent of sentences
conformed to the complex question structure under consideration. But
exemplars are still rare. And, more to the point, young children do not read
the Wall Street Journal. There is, though, another form of input that might
help children, namely questions of the following kind (from MacWhinney,
2004: 890):
Where is the dog that you like?
There is only one instance of the copula here (and no auxiliaries). But like
our original examples, this question has a complex structure, with one clause
(that you like) embedded inside a main clause (where is the dog…). To form
the question, we must also move the copula from the main clause, as in our
previous examples. This kind of question might therefore provide the
evidence that children need to learn the structure-dependent question rule.
MacWhinney (2004) reports that this kind of structure crops up hundreds of
times in the CHILDES corpus. This sounds like a healthy figure, until we do
a simple calculation. Even 1,000 such cases over a total of 3,000,000
amounts to just 0.03 per cent of the input, spread over many different
children. To all intents and purposes, therefore, the relevant examples of
auxiliary or copula movement are seldom, if ever, witnessed by the child.
The imitation of grammatical structures
The preceding discussion highlights an important assumption underpinning
the nativist approach to language learning, namely, that the child must hear
examples of the ‘acquirendum’ (the structure to be learned) in order to learn
it themselves. As mentioned, non-nativists have extended this idea by looking
for structures that are not identical to the acquirendum, but which share
critical properties (Where is the dog that you like?) (Pullum & Scholz, 2002;
MacWhinney, 2004). But there is a confusion here between the acquirendum,
which is a grammatical rule, and the output when that rule is applied in a
particular instance (a sentence). When we listen to someone speak, all we
hear are the products of grammatical rules, not the rules themselves. The
child cannot imitate grammar. Imagine how helpful it would be if every word
the child heard was tagged in some way, to indicate what grammatical
category each word belonged to:

With this kind of tagging, the task of working out the rules of grammar would
be greatly ameliorated. Observe, though, that even this system is
underdetermined in some ways. That is, it lacks useful information. For
example, it would also help to know that the verb (sat) appears in the past
tense, and that article and noun combine into a grammatical unit (noun
phrase, or determiner phrase, depending on one’s theoretical persuasion).
And so on. Of course, this kind of help is not available. So how does the child
recognize grammatical categories in the input? This is a problem that still
taxes language acquisition researchers from both nativist and non-nativist
traditions. Either grammatical categories like noun phrase must be
constructed by the child, or they are given as part of UG. In either case, these
categories must be acquired from a stream of speech that is not tagged in
advance. Innate knowledge, by itself, cannot explain language acquisition
(cf., Ambridge et al., 2014). This important point is generally glossed over, as
revealed in the following remark by Chomsky (1972: 30): ‘the idea of
“structure-dependent operations” is part of the innate schematism applied by
the mind to the data of experience’. As we saw with triggers, the issue that
has not been resolved is just how the mind ‘applies’ its innate knowledge to
the input. A bidirectional learning problem persists, therefore: ‘how does the
child map the words she hears onto the categories she has; how does she map
the categories she has onto the words she hears?’ (Valian, 2014: 81).
Evidence from children
We have seen that complex questions, of the kind discussed above, are
surpassingly rare. This applies not only to Child Directed Speech, but also to
sources of formal written English aimed at adults. We might ask, therefore, if
such questions are a minority taste. Perhaps only some people acquire this
kind of complex structure. But do not forget your own intuitions about the
sample sentences above. They should tell you that we all have the same
knowledge, regardless of our individual experience. We can tap into an
adult’s knowledge of grammar by asking directly about their intuitions. But
we cannot do this with a very young child. Instead, researchers have to be
more ingenious. One approach has been to elicit questions from children by
rigging the conversation so that a complex question is called for. If Universal
Grammar has endowed children with knowledge of structure dependence,
then any questions we can squeeze out of them should be grammatical.
Moreover, none of their questions should violate structure dependence.
Unfortunately, very little research has addressed this question directly. In one
series of experiments, children aged 3–5 years were presented with a familiar
figure, Jabba the Hut from Star Wars, and were encouraged to ask him a
series of questions (Crain & Nakayama, 1987; Nakayama, 1987). An
experimenter showed children various pictures and for each one elicited a
question with the frame: Ask Jabba if ____________, for example, Ask Jabba
if the boy who is being kissed by his mother is happy. This is not an easy task
and children make frequent errors (62 per cent of the time for the younger
children). A frequent kind of error was the production of questions with three
auxiliary verbs: *Is the boy who is being kissed by his mother is happy?
Nakayama (1987) argues that this task can easily overload the child’s
processing capacity. But the child’s grammatical abilities are not overloaded,
because children never produced questions that violated the structure-
dependent question formation rule (for example, *Is the boy who being kissed
by his mother is happy?). In consequence, Crain & Nakayama (1987)
conclude that the initial state of the language faculty (Universal Grammar)
embodies structure dependence.
There have been some acute criticisms of Crain & Nakayama’s methodology
(for example, Drozd, 2004; MacWhinney, 2004). But more seriously, the
empirical findings of these early studies have been challenged. Ambridge,
Rowland & Pine (2008) presented children with a picture that depicted a
simple scene, for example, of two boys, one running fast and the other
walking. The experimenter then encouraged children to ask a toy dog a
question about this scene: Ask the dog if the boy who is running fast can jump
high. This set-up is rather contrived, but at least children were being asked to
pose a real question, that is, one they did not know the answer to. The answer
was depicted on the back of the card, ‘visible’ only to the toy dog. The dog
was not only happy to do experiments, but could talk into the bargain,
supplying the child with appropriate responses to their questions. See what I
mean by contrived? Fortunately, children are very tolerant of adults and do
their best to keep them amused. The method adopted by Ambridge et al.
(2008) differed from Crain & Nakayama (1987) in two further important
ways. First, there were more trials, that is, opportunities to produce a
question: 16 per child, compared with a maximum of six. And second, a
range of different auxiliary verbs was used. Crain & Nakayama (1987)
confined their test questions to copula and auxiliary is. But it turns out that
children make relatively few errors with these two forms (e.g., Ambridge,
Rowland, Theakston & Tomasello, 2006; Rowland, 2007; but see
Santelmann, Berk, Austin, Somashekar & Lust, 2002). The low error rates
are a sign that children become competent with is relatively quickly, and are
therefore less likely to produce errors, even with complex questions. To see if
children are willing to violate structure dependence, more error-prone
auxiliaries are suitable, like can, should and do, or even different forms of be
(are instead of is).
With these changes in methodology, Ambridge et al. (2008) report that six-
year-old children produce structure dependence errors about 5 per cent of the
time:
*Can the boys who run fast can jump high?
*Can the girl who dance can sing?
*Can the girl who make cakes can make biscuits?
*Can the men that drive cars can drive tractors?
It seems, then, that UG is not acting like a rigid straitjacket, compelling
children to observe structure dependence at all times. But why are children
willing to produce corrupt sentences for some, but not all structures?
Ambridge et al. (2008) suggest that close attention to the input may provide
the key. Of special interest are sequences of words that appear side by side in
the input, so called bi-grams. For example, sequences like who unhappy
almost never occur. Try to construct a sentence for yourself that contains this
sequence of words. But the child would have to produce this sequence in
order to violate structure dependence, as in the following example:

Perhaps children are unwilling to produce bi-grams that they have not
actually heard themselves. This so-called conservative approach to learning
has often been remarked upon in young children (for example, Snyder, 2007;
see also Chapter 9). It would also help explain why some structure
dependence errors do occur. For example, the sequence who dance is far
more likely to occur in a range of sentence types (People who dance live
longer). This may explain why children are more willing to produce an error
like *Can the girl who dance can sing? This explanation is intriguing, but
may suffer from ‘Little Englisher’ syndrome. As we observed in Chapter 7,
the majority of language acquisition research is conducted on (and in)
English (see also Wierzbicka, 2011). And this can distort our view, not only
of what the child learns, but how. In this vein, it turns out that the solution
pursued by Ambridge et al. (2008) may not fare so well beyond the confines
of English (Berwick, Pietroski, Yankama & Chomsky, 2011).
If we maintain the focus beyond English, recent evidence from Japanese
children provides support for Chomsky’s position (Suzuki & Yoshinaga,
2013). These authors examined quantifiers, which modify nouns to express
amount (e.g., several fish, many fish, a few fish). Japanese has so-called
floating quantifiers, a phrase which captures the relatively free word order
observed in Japanese. Take, for example, the quantifier nin, which is used to
count people:
San-nin-no gakusee-ga koronda.
three students fell down
‘Three students fell down.’
Gakusee-ga san-nin koronda.
students three fell down
‘Three students fell down.’
These examples have been simplified to allow us to focus on the essential
point, which is that the quantifier (and the number it attaches to – in this case
three) can ‘float’ or appear in different positions without affecting the
meaning. Critically, though, quantifiers cannot just float off anywhere. Their
movement is structure dependent and not based on linear order. The input
available to children provides almost no information on the syntactic
constraint which governs floating quantifiers (Suzuki & Yoshinaga, 2013). At
the same time, though, Japanese five- and six-year-olds interpret floating
quantifiers correctly, even when they do not occur directly next to the noun
they modify. Their interpretation is based on linguistic structure, not on the
linear order of words, and there seems to be little help available in the input.
We have at least one case, therefore, where learning seems to take place in
the absence of experience.
Discussion point 8.2
Let’s have a heated debate. But first you will need to do some
preparatory reading. Organize yourself into groups of three. Everyone
should read articles (1) and (5), below. Each person should also take
one of the articles (2) to (4). Not as much reading as it might appear
(roughly 20 pages each). And worth the effort. In your groups consider
the evidence and arguments in favour of each position. How
convincing is the empirical evidence offered in the first article (Lidz,
Waxman & Freedman, 2003)? How convincing are their arguments, in
light of the criticisms which follow?
1. Lidz, J., Waxman, S. & Freedman, J. (2003). What infants know
about syntax but couldn’t have learned: Experimental evidence for
syntactic structure at 18 months. Cognition, 89(3), B65–B73.
2. Tomasello, M. (2004). Syntax or semantics? Response to Lidz et
al. Cognition, 93(2), 139–140.
3. Akhtar, N., Callanan, M., Pullum, G.K. & Scholz, B.C. (2004).
Learning antecedents for anaphoric one. Cognition, 93(2), 141–
145.
4. Regier, T. & Gahl, S. (2004). Learning the unlearnable: The role
of missing evidence. Cognition, 93(2), 147–155.
5. Lidz, J. & Waxman, S. (2004). Reaffirming the poverty of the
stimulus argument: A reply to the replies. Cognition, 93(2), 157–
165.
Poverty of the stimulus: Summary
The poverty of the stimulus argument does not look well supported. We have
seen that a number of longstanding assumptions about the input to the child
have been refuted. Principal among these are the idea that parents supply
degenerate input and also that they do not provide corrective information for
grammatical errors. A further observation was that young children receive no
direct language teaching. But this is something of a red herring: overt
teaching could never be the most efficacious kind of input for a naive
language learner. The kind of input they actually receive is more promising in
its simplicity, clarity and responsiveness to the child’s learning needs. We
also considered the idea that language acquisition is quick and effortless. But
defining what these two concepts mean with regard to language learning was
problematic. And a survey of other feats in cognitive development revealed
that it is not always easy to characterize language as the most rapid
achievement of mental development.
We then turned our attention to a specific property of grammar: structure
dependence. Chomsky poses a valuable question: Where does our knowledge
of structure dependence come from, given that we never hear complex
questions of the critical type as we grow up? We observed that despite a lack
of input exposure, we, as adults, do seem to share a knowledge of what
makes complex questions either grammatical or ungrammatical. In
consequence, Chomsky concludes that structure dependence must be part of
Universal Grammar. But we also noted that one does not necessarily need to
hear a structure in order to acquire it. In this regard, research on machine
learning has begun to show that complex questions can be acquired without
them appearing in the input to the model (Lewis & Elman, 2001; Reali &
Christiansen, 2005; see also Regier & Gahl, 2004). The success of models in
this regard suggests that there is no poverty of stimulus problem for the child.
At the same time, it remains to explain how the child makes the leap from
simple input to complex output.
We also examined the occurrence of structure dependence errors by children.
We should first recall that such errors are very rare, even when children are
placed under a hot spotlight and offered unusual inducements like talking
dogs. We should also note that their violations of UG do not force us to
abandon UG as a model of child grammar. We all make mistakes.
Unfortunately, though, the mistakes that occur create an empirical headache
for UG theory. They make it very difficult to demonstrate, unequivocally,
that UG is the guiding force behind child utterances. Errors could arise
because child grammars bear no relation to the theoretical description offered
by UG theory. Or they could arise because the child does possess UG, but
their knowledge of structure dependence is momentarily derailed in the
production of a particular question. Distinguishing between these two
possibilities may be well nigh impossible. What does remain possible, in
principle, is that structure dependence is a property of language that children
are born with. But it remains for UG theorists to demonstrate precisely how
innate knowledge might assist the child in the acquisition of language
(Ambridge et al., 2014).
Let’s take our discussion of the poverty of the stimulus back to where we
started: the slave boy, whose knowledge of geometry is inherent in his soul.
Close examination of this story shows that Socrates suggests most of the
answers to the boy and even provides ‘negative evidence’ in the form of
corrections for his errors (Clark & Lappin, 2011). Experience conducive to
learning was, in fact, provided. In the case of language acquisition, also, it
seems likely that the input available is far richer in information than nativists
allow (see Chapter 4).
The contents of UG: What precisely is innate?
We have considered one aspect of grammar – structure dependence – in some
detail, because of Chomsky’s explicit claim that it constitutes part of
Universal Grammar. But what else belongs in UG? Frustratingly, there is no
simple answer to this question. One answer seems to be ‘anything and
everything a UG theorist studies’. In this vein, several book-length treatments
have been published which present the theory of Universal Grammar (e.g.,
Cowper, 1992; Radford, Atkinson, Britain, Clahsen & Spencer, 1999; Adger,
2003). These texts are based on the assumption that the numerous aspects of
grammar they describe are all part of UG. For example, Napoli (1993: 4–5)
declares that ‘the principles developed in [her] book are taken to be
universals of human language’. But no attempt is made to justify this
assumption. Instead, what we get, if anything, is an appeal to the poverty of
stimulus argument: linguistic Principle X is taken to be part of UG because
the child’s linguistic environment lacks the information required to help them
learn this Principle. But this blanket assumption does not get us very far.
Linguists (being academics) are likely to disagree on how best to capture the
description and explanation of Principle X (for a case in point, see competing
accounts of the so-called null-subject phenomenon: Hyams, 1992; Huang,
1995). When competing accounts are advanced, which one should we prefer?
Poverty of stimulus arguments cannot help us decide. The reason is that each
and every account of Principle X can rely on the poverty of stimulus
argument to support its claim to UG membership. In other words, poverty of
stimulus can never function as an evaluation metric: it cannot be used for
choosing among competing explanations to determine the precise constitution
of Universal Grammar.
In contrast to the ‘anything and everything’ approach, the last 15 years has
witnessed a much more streamlined approach. First, Hauser et al. (2002)
proposed that UG contains just one property of grammar: recursion.
Recursion occurs where the input to a rule also features as part of its output
(see Box 8.2). But is recursion so uniquely privileged? As we pointed out in
Chapter 2, it has been argued that songbirds show hierarchical structure and
recursive syntactic pattern learning in their songs (Gentner, Fenn, Margoliash
& Nusbaum, 2006; Abe & Watanabe, 2011). Meanwhile, some human
languages do not appear to show recursion, most notably, the Amazonian
language, Pirahã (Everett, 2008). Moreover, recursion is a complex property,
and one might argue, for example, that one cannot have recursion without
both structure dependence and grammatical categories to apply recursion to.
Chomsky’s assertion about recursion is regarded by many as a major
recantation of his former position on Universal Grammar (Pinker &
Jackendoff, 2005). But Chomsky doesn’t let the grass grow. While his
detractors languish in an overgrown meadow, he has moved on. Currently,
Chomsky argues that a much simpler property of language is both
biologically endowed and uniquely human (part of the Faculty of Language –
Narrow, as discussed in Chapter 2). The property in question is known as
Merge, and allows us to combine two linguistic units to create a new kind of
unit (Berwick & Chomsky, 2016). We can see these two features at work in
the simplest of noun phrases: electric piano. The adjective electric and the
noun piano combine to produce a linguistic unit, a noun phrase, which is
distinct from both of its constituents. Merge allows for the creation of
hierarchical linguistic structures, referred to by Chomsky as the Basic
Property of language (ibid., 2016). Once a language is organised in terms of
hierarchical phrases, the notion of structure dependence falls out naturally.
Undoubtedly, the plausibility of Universal Grammar is enhanced by reducing
it to the single, simple property of Merge. As ever, though, there is debate:
see Matthews (2007) for a sceptical look at the notion of hierarchy in
linguistic phrases.
Box 8.2 Recursion
To understand recursion, we first need a few basics concerning
the rules of grammar. Words combine to make phrases and
phrases combine into sentences in a rule-governed way. In
English, I can combine two words like the and hobbit to produce
an acceptable phrase: the hobbit. If I switch the order round, the
phrase is no longer grammatical: hobbit the. Our grammatical
phrase (the hobbit) can function as a unit: we can move it around
as though the two words were glued together and drop it into
different positions within a sentence (though not just in any old
position): The hobbit in my garden likes toast. Or I just met the
hobbit of my dreams. The important thing about our phrase is that
the two words, the and hobbit, are particular kinds of words. The
belongs to the category of determiner (which also includes a and
some), while hobbit belongs to the category of noun (to which we
can add many thousands of other nouns, like computer, water and
truth).
Traditionally, our phrase would be described as a noun phrase,
and the rule for making noun phrases of this kind would be:
Noun Phrase → Determiner + Noun
More recent treatments argue that our phrase, the hobbit, is better
described as a determiner phrase (Radford et al., 1999; although
see Matthews, 2007). But fortunately, our demonstration of
syntactic rules is not affected by this twist.
Recursion is a special kind of procedure in which a category (like
noun phrase) is embedded within itself. For example, in English,
we can use recursion to embed one relative clause inside another.
A relative clause is like a mini-sentence, tucked inside another
sentence and is often introduced by that, which or who, as in: I
like the hobbit who won the race. This relative clause is part of a
larger noun phrase (the hobbit who won the race). We thus have
two different kinds of noun phrase. One is fairly simple (the
hobbit), while the other is more complex (the hobbit who won the
race). Our complex noun phrase can be captured by the following
rule:
Noun Phrase → Noun Phrase + Relative Clause
   the hobbit + who won the race
 Now we get to the nub of recursion. Observe that we have a
simple noun phrase, the race (determiner + noun), embedded
within the relative clause. If we choose, we can take this latter
noun phrase and expand it with our second rule (noun phrase +
relative clause). For example: I like the hobbit who won the race
that started in the park. In principle, we can repeat this process ad
nauseum, or more properly, ad infinitum. Recursion thus allows
us to take our limited palette of rules and produce an infinite
number of sentences, in a potentially endless cycle. This is
because the input to the rule (in our case, noun phrase) also
features in the output. Thus, one can always add a new clause: I
like the hobbit who won the race that started in the park that once
had a bandstand that … and so on.
From the outside, the pace of change in UG theory can seem bewildering.
The so-called Chomskyan revolution, is, in fact, a series of ongoing
revolutions, as the ideas and theories develop over time. We do not need to
tussle further with the arguments for and against the idea that UG is restricted
to a very limited set of properties. The point here is to highlight the fact that,
as yet, there is very little agreement about the precise contents of Universal
Grammar. Critically, if we do not know what is, and is not, hypothesized to
be innate, we have no way of testing nativist assumptions.
Conclusion
This chapter has introduced the idea of innate knowledge of grammar,
characterized by Chomsky as Universal Grammar. We saw that the main
source of evidence in support of this idea comes from an analysis of the
child’s linguistic environment. Chomsky argues that the input is
impoverished, in the sense that it lacks the information necessary to permit
the acquisition of grammar by a domain-general learning mechanism. We
reviewed the so-called argument from the poverty of the stimulus in some
detail and found that none of its tenets are especially well supported by
available evidence. At the same time, they are not decisively refuted either.
The idea that knowledge of grammar might be innate may have seemed rather
bizarre at the beginning of this chapter. Perhaps now, though, you may accept
that Chomsky’s position has an intuitive appeal. All typical children grapple
with the obscure complexities of grammar with a speed and facility that can
seem quite astonishing. It is therefore natural to predict that genetic factors
are implicated. But as we have seen, the arguments for linguistic nativism do
not stand up to close empirical scrutiny. The central problem is that Chomsky
relies heavily on the argument from the poverty of the stimulus to support his
case. And we should emphasize that it is largely a matter of argument.
Empirical evidence in support of the poverty of the stimulus is thin on the
ground. Chomsky (following Koryé, 1968) has often argued that theories are
not refuted by data, only by analyses. This situation is unsatisfactory for
many psychologists because they are used to testing their hypotheses against
substantive data, supported by statistical analyses. Linguistics (still the
world’s major supplier of Chomskyan nativists) is less steeped in this
tradition. Empirical studies by nativists tend to present evidence from very
small samples of children (sometimes only one) and adopt a research strategy
that, again, psychologists can find frustrating. Typically, these studies report
that children of a particular age, say two or three years, demonstrate
knowledge of a particular Principle of UG (for example, Hyams, 1986;
Radford, 1990; Snyder, 2007). The conclusion that the Principle is innate
depends largely on the youth of the children. Two-year-olds, it is argued,
cannot have been exposed to the relevant experience to explain their
linguistic prowess. But two years is a long time in language acquisition, as
we saw in Chapters 5 and 6. And it is very difficult to demonstrate that
children lack a particular learning experience with regard to language. In
many cases, therefore, the empirical evidence we might require is simply not
available. Chomsky himself has recently drawn attention to the limits of our
knowledge in this respect:
I’ve been working on Universal Grammar for all these years; can anyone
tell you precisely how it works [how it develops into a specific
language]? It’s hopelessly complicated. (Chomsky, 2012: 54)
When we have the wherewithal to understand the workings of the human
mind better, we might come closer to discovering what, if any, aspects of
linguistic knowledge are genetically determined.
In a Nutshell
Chomsky asserts that knowledge of grammar is innate, in the
form of Universal Grammar.
UG is the initial state of the language faculty which then
develops into knowledge of a particular language, like
English or Korean.
The grammars of the world’s languages share many
 properties; according to nativists, the diversity one sees is
either superficial or encompassed within UG as a limited
menu of options (parameters).
Chomsky argues that language acquisition is fast, effortless
and achieved on the basis of limited exposure to a poor
model of language. None of these claims is well supported.
The poverty of the stimulus argument is based on the belief
that the input for learning lacks the information necessary to
explain how the child comes to acquire what they know
about grammar.
Structure dependence is a property of grammar used by
Chomsky to demonstrate that even very young children
possess an understanding of complex, abstract aspects of
grammar despite never having heard anyone use these
grammatical structures.
Children can be induced to produce structure dependence
errors in an experimental setting, even though this violates
the precepts of Universal Grammar.
No-one has ever produced a definitive list (or even a working
list) of the particular aspects of grammar that are held to be
part of UG. This makes it difficult to determine precisely
what is, and is not, innate.
Chomsky’s theory of UG is not well supported by empirical
evidence. But neither has it been decisively refuted.
Further Reading
Botha, R.P. (1989). Challenging Chomsky: The generative garden
game. Oxford: Basil Blackwell.
In this book, Botha draws crucial distinctions for the reader that
help one appreciate the subtlety and precision of Chomsky’s
work. In the process, Botha provides a very witty analysis of how
Chomsky has fended off his critics over the years, written partly
in an eighteenth-century style. Not as odd as it sounds. Highly
entertaining, in fact.
McGilvray, J. (Ed.) (2005). The Cambridge companion to
Chomsky. Cambridge: Cambridge University Press.
There have been many books published about Chomsky. This
collection covers three Chomskyan themes in depth: language,
mind and politics. Several academic heavyweights slug it out in
the ring on Chomsky’s behalf. This collection is authoritative, up
to date and wide-ranging, but do bear in mind the partisan nature
of the authors.
Sampson, G. (2005). The ‘language instinct’ debate (revised ed.).
London: Continuum.
Also not balanced in its approach, but this time the stance is
avowedly anti-nativist. This book overflows with provocative
arguments and provides an entertaining, sometimes overheated,
attack on the nativist enterprise in child language. Again, read
with caution: how well supported, empirically, are Sampson’s
ideas?
Websites
Languages of the World: www.omniglot.com/
This site is a treasure trove of basic information on many of
the world’s languages. Rummage around in this site and
marvel at the number and diversity of different languages
currently spoken throughout the world. Then consider
Chomsky’s fundamental claim: the observed differences
among languages are, in critical respects, superficial. What
matters are the common features that all languages share
(Universal Grammar).
Chomsky’s influence beyond linguistics:
https://chomsky.info/
You will find sufficient material here to satisfy your
curiosity about one of the world’s leading intellectuals. The
focus is on his political views, but there are also references
and links to his academic writings in linguistics, philosophy
and (latterly) evolutionary theory.
Still want more? For links to online resources relevant to this
chapter and a quiz to test your understanding, visit the companion
website at https://study.sagepub.com/saxton2e
9 The Usage-based Approach: Making it Up
as You Go Along
Contents
Language knowledge from language use 236
Social cognition 237
Dyadic and triadic interaction 237
Collaborative engagement and intention-reading 238
Collaborative engagement as a basis for language development 239
Early constructions: A route into grammar 240
In the beginning was the utterance 240
From single-unit to multi-unit speech 242
Does the child go from fully concrete to fully abstract? 244
The productivity puzzle 245
The transitivity bias 245
Pattern finding 248
Type frequency: A route to productivity 249
Sounds familiar: The role of frequency 251
Early productivity: Syntactic bootstrapping 252
Constraining productivity 255
Conservative learning 256
Entrenchment 256
Pre-emption 259
Summary: Reining back on productivity 261
Overview
This chapter deals with the most prominent and most comprehensive
non-nativist account of syntax acquisition: usage-based theory.
By the end of this chapter, you should be able to describe some of the
critical factors in infant social communication that are said to underpin
later language development. These include the child’s use of pointing
and the emergence of collaborative engagement in pursuit of a shared
goal.
You should also be able to discuss and evaluate usage-based research
on the child’s earliest speech output, with its focus on entire utterances,
rather than individual words. The syntactic categories witnessed in
adult language are said to emerge only gradually, as the result of the
 child’s experience with language. You should be able to describe the
progression towards syntax from early, lexically specific, structures
through to broad syntactic generalizations which permit the child to be
linguistically productive.
You should gain some appreciation of the challenges facing the usage-
based approach. In particular, we will focus on the problems raised in
explaining child productivity: (1) how do children acquire general
rules, based on mature syntactic categories, that allow them to create
their own sentences? And (2) how do children constrain their
productivity and avoid producing errors? You will gain some
appreciation of current usage-based answers to these questions. In the
process, it will become apparent that the issue of child productivity
continues to present a serious challenge for all theories of child syntax
acquisition.
Language knowledge from language use
As the name suggests, the usage-based approach rests on the assumption that
our knowledge of language is obtained from the use of language (Langacker,
1987). There is a natural sympathy, therefore, with functionalism, the belief
that ‘the forms of natural languages are created, governed, constrained,
acquired and used in the service of communicative functions’ (Bates &
MacWhinney, 1989: 3). In considering how language is used, the overriding
question is What is language for?, with the answer being some function like
requesting, directing or informing. Every utterance is produced with the
intention of getting things done or understood. The task of the child is to
make the necessary connections between communicative functions and the
linguistic forms which are used to express those functions. Take, for
example, the transitive construction in English:
noun phrase + verb + noun phrase
Max  kissed  Anna
The function of transitives is to denote how one participant (Max) acts in
some way on another (Anna). This function maps onto a particular default
syntactic form, which in English includes placing the agent (Max) first,
followed by the action (kiss) and ending with the patient (Anna). In the
usage-based approach the child must understand the function of this sentence
first, as a basis for acquisition of the syntactic structure. Generally speaking,
‘children acquire language first and foremost by understanding how others
use language’ (Tomasello, 2009: 86). We will organize our discussion of the
usage-based approach around three main themes:
1. Social cognition as a foundation for language learning.
2. Pattern-finding: how the child creates grammatical structures and uses
them productively.
3. Constraining productivity: how the child avoids overgeneralization.
In sharp contrast with the nativist approach, usage-based theory does not
assume innate knowledge of grammar. Instead, grammar is learned. Hence,
in the initial stages of language acquisition, the child’s knowledge of
language is held to be qualitatively different from that of an adult. The child
converges gradually on the adult state. Moreover, language is not regarded as
a unique cognitive phenomenon that requires specialized (domain-specific)
learning mechanisms. Instead, the learning mechanisms required by the child
are said to be domain-general. In other words, the learning mechanisms used
in the acquisition of language are applied to other, non-linguistic forms of
input, including, for example, sequences of auditory tones (Saffran et al.,
1999). On this view, language is not considered to be uniquely different, or
difficult, as an object of learning. But before we consider how the child
tackles the complexities of grammar, let’s first examine the social world of
the infant and consider how the infant’s communication skills might pave the
way into language.
Social cognition
Dyadic and triadic interaction
In usage-based theory, the child’s understanding of language functions is
rooted firmly in the social act of communication. It is argued that certain
aspects of social interaction are unique to humans, in particular, the ability to
read and share the intentions of others (Tomasello, Carpenter, Call, Behne &
Moll, 2005). On this view, the complexities of language, including grammar,
derive from this more basic aspect of human interaction. The communication
skills of the infant develop over the first year. Very early on, at about 5–6
weeks, infants demonstrate an interest in attending to the mother’s face and
voice, together with an increase in smiling and cooing. This social awakening
in the infant is reciprocated by mothers, with increases in their own
vocalizations and displays of interest and affection. Arguably, the adult’s
ability to capture and then hold the child’s attention is critical for language
development. By the age of 11–14 months, infants have developed a
sophisticated ability to follow changes in the direction of an adult’s gaze
(Scaife & Bruner, 1975). Critically for language development, the level of
shared attention witnessed in adult–child discourse is positively correlated
with later vocabulary and syntax level, for both typical children and for those
with autistic spectrum disorder (Tomasello, Mannle & Kruger, 1986; Akhtar,
Dunham & Dunham, 1991; Rosenthal Rollins & Snow, 1998).
The ability to follow the mother’s direction of gaze is just one feature of
shared attention. In Chapter 4 we remarked on the adult’s use of exaggerated
intonation and high pitch to gain an infant’s attention. It has also been found
that adults move differently when interacting with infants, adopting a closer
proximity, increased enthusiasm and exaggerated actions. Moreover, infants
prefer to look at this special form of infant-directed motion (Brand &
Shallcross, 2008). Generally, it has been found that two-year-olds whose
mothers indulge in high levels of attention-holding devices score highly on
general measures of verbal learning one year later (Schmidt & Lawson,
2002).
Early communication is dyadic, in the sense that it involves two participants.
Over time, though, the infant becomes increasingly interested in the world of
objects and, by three months, is less willing to gaze solely at the mother. In
response, mothers adopt a more vigorous, playful approach to interaction,
typified by a more excited, arousing mode of speech (Trevarthen, 1983). A
triadic form of engagement emerges towards the end of the first year, in
which adult and infant jointly direct their attention towards a third entity
which constitutes a shared goal (Tomasello et al., 2005). This shared goal
might manifest in turn-taking games, like rolling a ball back and forth,
building a tower from blocks, or book sharing activities. From engagement in
triadic activities of this kind, the infant comes to understand that other people
have a particular goal in mind which they are trying to achieve through a plan
of action. By five months, infants can interpret the goal-directed nature of
simple behaviours, such as the act of reaching for an object and grasping it
(Woodward, 2009). If an adult reaches for one of two objects, a seven-month-
old will typically reach for the same object. But the infant will not do so if the
adult directs an ambiguous action towards the object, such as making contact
with the back of their hand (Hamlin, Hallinan & Woodward, 2008). Hence,
the infant’s response seems to be driven by their interpretation of adult
reaching as a form of goal-directed behaviour.
Collaborative engagement and intention-reading
At about 10 months, the uniquely human behaviour of pointing emerges,
typically before the child has produced their first word. The act of pointing
shares with many words the power to function as an act of reference: they can
both pick out entities in the world. But the meaning of pointing is especially
underdetermined. To work out what someone is pointing at, I must do more
than simply follow the direction in which their index finger is extended. The
target is narrowed down in this way, but not precisely specified. Am I
pointing at a book? Or the image on the cover? Or at the name of the author?
And why am I pointing in the first place? Pointing is underdetermined, both
in terms of what is being pointed at, and why. If this seems surprising, then it
is because we can take for granted people’s intentions and goals. The same
pointing behaviour can be interpreted in many different ways according to the
shared understanding of the participants.
In a recent study, children aged 14 months were engaged in a game in which
the object was to tidy up various toys (Liebal, Behne, Carpenter &
Tomasello, 2009). By 14 years, as any parent will tell you, this tidying
behaviour has ceased altogether. But these tender infants were very co-
operative. They engaged in this activity on a one-to-one basis with an adult,
before being joined by a second adult. In one experimental condition, the
familiar adult pointed at one of the remaining toys, while in a second
condition, the unfamiliar adult did the pointing. Infants were far more likely
to retrieve the object being pointed at, and tidy it away, if the familiar adult
was doing the pointing. Hence, the infant only interpreted the pointing as a
request to tidy up if it was done by the adult who had been sharing this
tidying behaviour previously. Thus, infants use their shared understanding to
interpret other people’s intentions (cf., Sommerville & Crane (2009) for the
influence of prior information on infant behaviour). Joint attention provides a
means for picking out objects in the world and labelling them, thus assisting
word learning. That said, it should be pointed out (!), that, in some
circumstances, even two-year-olds can learn words in the absence of joint
attention (Scofield & Behrend, 2011).
Tomasello et al. (2005) argue that the ability to engage in a collaborative
activity, with a shared goal, is a uniquely human ability. Thus, it is difficult to
imagine two chimpanzees co-operating on even simple activities like carrying
something together. Chimpanzees are simply not motivated to share
intentions, produce a plan of action and divide up the task so that each
participant has their own role in achieving a joint goal. Tomasello (2009: 72)
underscores the unique nature of human social cognition and argues that it
‘paves the way for the acquisition of the “arbitrary” linguistic conventions
that infants use’.
At the one-word stage, children sometimes combine a spoken word with a
gesture (especially pointing). These gesture–speech combinations are of
value because they augment the expressive power of the child considerably.
For example, the word teddy, by itself, can convey any number of
communicative intentions (I like my teddy, My teddy has been out in the rain,
This teddy isn’t mine, and so on). But if the child points towards a teddy that
is out of reach on a shelf, and says ‘teddy’, then a more specific meaning is
conveyed by the gesture–speech combination (Give me my teddy, perhaps).
The age at which gesture–speech combinations emerge predicts the age at
which the first entirely spoken two-word combinations emerge (Iverson &
Goldin-Meadow, 2005). Rowe & Goldin-Meadow (2009a) report that parents
depict, on average, 26 different meanings per hour through gesture with their
14-month-old children. But, as we saw in Chapter 4, parents differ in their
rates of gesturing, which, in turn, affects how often children gesture
themselves. These differences are important because ‘high-gesture’ 18-
month-olds develop relatively large vocabularies by 42 months (Rowe &
Goldin-Meadow, 2009b). Of interest, these ‘high-gesture’ children seem to
get it from their ‘high-gesture’ parents. What we see, then, is that progress in
language development is associated with an extensive use of gesture.
Collaborative engagement as a basis for language
development
The importance of social cognition is emphasized by demonstrating close
overlaps between early communication and language. In this vein, we are
offered the following argument:
conversation is an inherently collaborative activity in which the joint
goal is to reorient the listener’s intentions and attention so that they align
with those of the speaker. (Tomasello et al., 2005: 683)
This might seem plausible at first blush, but row back and read that quotation
again. You may discover that it rests on some odd assumptions. First, it
assumes that the listener has acquiesced somehow in a ‘joint goal’, the aim of
which is to change their own point of view. More problematic still: who is the
speaker and who is the listener? In any conversation, these roles alternate
with each turn. And of course, we do not always collaborate in conversation.
In an argument, each person has an independent and conflicting goal to
realign their opponent’s views (or intentions). There is no joint goal.
Other parallels between collaborative engagement and language seem more
straightforward. For example, the intention-reading skill witnessed in non-
verbal communication may well be useful in verbal exchanges. Hence, ‘if we
assume that children are good at working out the intention underlying the
verbal utterances of others, they have a head start into the linguistic system’
(Behrens, 2009: 396). A component of intention-reading is the ability to take
into account prior shared experience. As we have seen, children respond
differentially to adult pointing, according to the level of joint prior
knowledge shared between adult and child (Liebal et al., 2009). An
equivalent within language is our use of pronouns. To interpret a pronoun
like she, as in She opened another window, both speaker and hearer must
share previous knowledge concerning the identity of she. In fact, it takes
some time for children to appreciate the anaphoric (‘referring back’) function
of pronouns (Wales, 1986). Pronouns are just one example of so-called
deictic items that rely on a shared context between speaker and listener for
their interpretation. Other examples include demonstratives like here and
there, verbs of motion like come and go, and articles like the. For young
children especially, deictic items are crucial, since 73 per cent of all mother–
child utterances involve reference to objects and events in the immediate
environment (Cross, 1977).
In the usage-based theory, it is predicted that the infant’s impressive
communication skills function as a precursor to language development. Of
particular importance are the emergence of intention-reading skills and an
ability to pursue joint goals in a collaborative way. Some authors suggest that
intention-reading skills not only precede, but constitute a ‘crucial prerequisite
for language acquisition’ (Diessel, 2004: 38). But arguments in favour of this
idea are somewhat speculative. As we have seen, the key social-cognitive
skills emerge at about the same time as spoken language. Children start
pointing only just in advance of producing their first spoken word. And
typically, the first word, at about 12 months, precedes signs of truly
collaborative social engagement at about 14 months. And we should not
forget that both social-cognitive and linguistic skills develop throughout the
first year, long before production of the first word. It is difficult, therefore, to
construe the ability to engage collaboratively as a necessary precursor for
language. Rather, it seems that language and early communication share
many functional characteristics and develop, to some extent, in tandem.
Early constructions: A route into grammar
In the beginning was the utterance
From the moment of the first word, it seems that children follow an orderly
progression from utterances that comprise first one word, then two, three and
four words, and more, in progressively longer utterances. The examples
below are from Eve (Brown, 1973):
one word → two words → three words
pencil dollie shoe man taste it
cup that pencil lie down stool
hat finger stuck drop a cheese
This view of development seduces us into believing that the word is the
critical unit in language development. The child simply needs to acquire the
capacity to stick more and more words together in the right form and in the
right order. But this view may be mistaken. Tomasello (2009) suggests that it
is the utterance, not the word, that should grab the headlines. On this view,
the child is driven to understand the communicative function of whole
utterances. To do so, the child must determine the adult’s intentions in using
a piece of language with them, be it (from an adult perspective) a single word
or several. In consequence, the child understands (or tries to understand) a
full communicative act, comprising the intentions of the speaker and what
they are referring to. In consequence, the child’s own early linguistic
productions, even when they seem to comprise more than one word, are often
best analysed as holophrases, that is, indivisible units of language that
convey a single, complete communicative intention (Barrett, 1995; Pine,
1995; Tomasello, 2003):
you do it
all gone
here you are
go away
I wanna do it
lemme see it
where the bottle
These phrases may seem like strings of individual words, but close analysis
reveals that the child does not use the words independently. By the same
token, a single-word utterance might be classified as a holophrase, an idea
with a long history in child language research. Thus, Sully (1895: 171)
remarks that ‘the single-worded utterance of the child is … [ a ] “sentence-
word”’.
Two pieces of evidence support the holophrase hypothesis: (1) the words in a
holophrase are pronounced with the intonation pattern of whole utterances,
be they comprised of one, two or more words; and (2) the words always
appear together in the same order. With regard to intonation, three different
patterns emerge early on which the child uses for distinct communicative
functions: (1) declarative, for conveying information; (2) imperative, for
orders; and (3) interrogative, for questions. Each kind of utterance has a
characteristic intonation pattern which can be found overlaid on children’s
holophrases. Utterance-level intonation patterns have been observed in early
single-word speech, so it is reasonable to analyse such cases as holophrases
also (Dore, 1975). On the second point, consider the phrase Daddy gone. If
the two words Daddy and gone do not each appear separately in the child’s
speech, but do appear consistently together in the same order, then one might
reasonably assume that Daddy gone is a holophrase. Pine & Lieven (1993)
examined the patterns in multi-word speech in children’s first 100 words.
They found that, on average, 11 per cent of child utterances could be
described as holophrases.
On Pine & Lieven’s (1993) evidence, holophrases occur only rarely.
Admittedly, their figure does not include single-word holophrases, while their
sample does include children who are relatively advanced (100 words). But
one might still expect a higher incidence rate if the whole utterance were
truly dominant in the child’s initial speech output. Children who are
dominated by the utterance ‘do not try to learn words directly; they try to
comprehend utterances and in doing so they often must comprehend a word
in the sense of determining the functional role it is playing in the utterance’
(Tomasello, 2009: 74). If that is the case, we must ask how the child becomes
aware of different functional roles within a single utterance (e.g., in the act of
giving there is a giver, the thing given, and a recipient). What prompts this
segmentation of a communicative act, and, also, a linguistic utterance, into
constituent functional parts? Of course, long before any such functional
analysis is undertaken, infants are already highly adept at segmenting
utterances. They segment the speech stream into individual words using
statistical cues (see Chapter 5). Hence, a challenge for the usage-based
approach is to integrate what we know about early speech segmentation with
the view that the utterance, not the word, is the starting point for the child
several months later. Generally speaking, we still need to explain how the
child makes the transition from single-unit speech, be it a single word or a
single utterance, to multi-unit speech.
From single-unit to multi-unit speech
At about 18 months the child starts going stir crazy and finally breaks out of
Holophrase Prison. The child begins to construct utterances from multiple,
separate components. Within the usage-based approach, three types of early
multi-word construction are recognized:
1. word combinations
2. pivot schemas
3. item-based constructions.
Word combinations comprise two words that each have roughly the same
status and which ‘partition the experiential scene into multiple symbolizable
units’ (Tomasello, 2009: 76). A simple example, provided by Tomasello, is
the child who sees a ball on a table and says Ball table. To do this, the child
brings two separate words together. Hence, mastery of individual words, qua
words, is a necessary precursor to word combinations. The second
construction type, pivot schemas, were first identified by Braine (1963). In a
pivot schema, the so-called pivot word is dominant, in the sense that it
determines the function of the utterance as a whole. Each schema comprises a
pivot and a variable slot that can be filled by a range of different items. The
order of pivot and slot is fixed, with the order reflecting that which the child
hears in the input. The following three examples are drawn from Braine
(1963).
Pivot schemas in children aged 19 and 20 months:
more + ________more cookie, more fish, more high, more hot, more
juice, more read
no + ________ no bed, no fix, no home, no mama, no more, no down
other + ________ other bib, other bread, other milk, other pocket, other
shirt, other shoe
The ‘no’ row is especially fascinating: how did a one-year-old write such
great blues lyrics? Pivot schemas constitute the first example of linguistic
abstraction in the child’s progress towards syntax. In each case, the slot is
filled by a word that matches the function of the particular pivot word. For
example, in the more + schema, the pivot word (more) seems to function as a
way of requesting that an action be repeated or continued. Recall that in the
usage-based approach, the function takes priority. What is the child using
language for? The child’s task is to acquire the mapping between particular
functions and the particular forms (constructions) used to express those
functions. In a pivot schema, the words that fill the slot will therefore
conform to the function determined by the pivot word. There is no
suggestion, in the usage-based approach, that the words in a pivot schema
belong to adult syntactic categories, like noun or noun phrase. Bannard &
Lieven (2012) report that for one two-year-old, almost 60 per cent of his
utterances involved the insertion of material in a slot, in the manner of a pivot
schema.
One step beyond the pivot schema comes the item-based construction
(MacWhinney, 2014a). At this point, we detect the stirrings of syntax within
the child. Syntactic devices are now used in the construction of multi-word
utterances. The systematic use of word order and the emergence of
morphological marking provide two examples of this ‘syntactic marking’
(Tomasello, 2009: 77). We can take word order as an example, having
encountered it already in Chapter 1. Compare the following two sentences:
Richard introduced Chrissy at the poetry reading.
Chrissy introduced Richard at the poetry reading.
This simple example shows that, in English, the meaning of a sentence can be
fundamentally altered if the words appear in a different order. The ordering
of Richard and Chrissy determines who is the agent (‘the doer’) and who is
the patient (‘the done to’). As it happens, children learning English seem to
use word order correctly from the very start (Radford, 1990). Even children
as young as two years can act out sentences like Make the doggie bite the cat,
a sentence which could be ambiguous to a naive child, because both the dog
and the cat are capable of being the agent that bites (DeVilliers & DeVilliers,
1973). But why are these early syntactic structures referred to as item-based?
The answer is fundamental to the usage-based approach: it is assumed that
the child’s early constructions are tied very closely to specific lexical items.
Item-based constructions are like islands, quite separate from one another,
even though, in the adult grammar, they might be governed by the same rules
(or schemas). This idea was captured in the so-called verb island hypothesis,
which predicts that children learn about word order (and other grammatical
relations) on a verb-by-verb basis (Tomasello, 1992). The emphasis on verbs
arises from their central role in dictating the format of whole sentences. As
Tomasello (2003: 122) observes, ‘the most abstract constructions
characteristic of adult linguistic competence typically revolve around verbs in
one way or another’. We shall see this more clearly, below, in our discussion
of verb transitivity. The emphasis on verb islands stems, in part, from the
assumption that children’s own initial use of verbs is predicted to mirror what
they hear, with no awareness of the wider world of general syntactic patterns.
For example, in a transitive sentence with take – Angela takes aspirin – the
child may know that Angela is the ‘taker/doer’ and aspirin is the ‘taken/done
to’. But on a usage-based approach, the child will not know, initially, that the
‘doer’ and ‘done to’ occupy these same positions for all transitive verbs. The
child needs to recognize the pattern across multiple verbs to extract the
general meaning and the general construction of the transitive. In the early
stages, therefore, the child should show no signs of the broad syntactic
generalizations that we recognize in the adult grammar. The idea of starting
small with lexically specific constructions has considerable mileage. In this
vein, one finds that children use low-scope slots in pivot schemas, that is,
slots that admit only a small range of lexical items (Cameron-Faulkner,
Lieven & Tomasello, 2003). Over development, children extend the range of
slots in schemas. They begin to build bridges between verb islands. More
broadly, the child begins to form linguistic generalizations as their use of
language becomes increasingly schematic (Boyd & Goldberg, 2012).
Does the child go from fully concrete to fully
abstract?
An abiding theme in the usage-based approach is that ‘constructions vary
along a scale of … abstractness’ (Diessel, 2004: 18; see also Lieven, 2016).
The child’s early constructions are described as concrete because they
comprise ‘concrete pieces of language, not categories’ (Tomasello, 2009: 76).
In a similar vein, some aspects of adult linguistic knowledge are also
described as concrete, for example, idioms like kick the bucket. In contrast,
constructions are deemed to be abstract if they consist of grammatical
categories like noun phrase or subject. But the use of the terms concrete and
abstract is problematic. There is nothing concrete about the representation of
language in the mind. Even if child utterances have a concrete reality in the
physical production of sound, they stem from mental representations, which
are abstract. Linguistic representations do not have to be identical to those of
the adult grammar in order to be abstract. And Lieven & Tomasello (2008:
170) allow that ‘children are capable of abstraction from the beginning’. We
need to tread carefully here, because abstraction is not the same as abstract.
Abstraction is a process that can be used in the formation of abstract
categories. For example, presented with a set of white objects (a lily, some
milk and a wedding dress) we can abstract the quality they have in common,
which is their whiteness. Abstraction is different from abstract, but its
presence from the very start indicates that abstract (not concrete)
representations are created by the child. In addition, as mentioned above,
Tomasello (2009: 76) suggests that the child’s very earliest two-word
combinations ‘partition the experiential scene into multiple symbolizable
units’. Symbols, of course, are the prime example of an abstract object.
Another problem is the projection of a sliding scale from ‘totally concrete’
(Tomasello, 2009: 76) to ‘fully abstract’ (Lieven & Tomasello, 2008: 175).
The notion of increasingly abstract language is nowhere explained or defined,
and is, in fact, difficult to conceive of. What could it mean for something to
be more or less abstract? Equally, how can one thing be more concrete than
another? If there is a sliding scale from highly concrete through to highly
abstract, it would be useful to see examples, with justification, at each point
on the scale. In philosophy, phenomena are described as either concrete or
abstract (Dummett, 1981). Abstract objects are not accessible to the senses
(they cannot be seen, heard or felt); they cannot be causes of change; and
they occupy neither space nor time (try explaining the whereabouts of the
number 6). While there are problems with each of these three criteria (see
Hale, 1987), the distinction between concrete and abstract is clear. Arguably,
therefore, usage-based researchers use these terms inappropriately. When
they talk about a shift from highly concrete to highly abstract, they are more
plausibly talking about a shift in knowledge from linguistically specific to
linguistically general. As Lieven & Tomasello (2008: 170) allow, ‘what
changes over development is the scope of the abstraction’ made by the child.
Framed in usage-based terms, this process includes the ability to insert an
increasingly wide range of items into a particular slot in an item-based
construction. It also includes the ability to connect separate verb islands into
a more general schema. A shift from specific to general is not only more
plausible, but has a much firmer foundation in the history of child language
research (e.g., Kuczaj, 1982).
The productivity puzzle
So far, usage-based theory has taken us on a journey from the child’s earliest
social skills through to a use of language that includes rudimentary syntactic
marking. We now continue on this theoretical path into more complex
aspects of syntax. As mentioned, the child must begin to generalize their
linguistic knowledge, abstracting patterns which allow for syntactic
productivity. A major issue of contention is the point at which children begin
to make such generalizations. As we shall see, usage-based theory predicts
that it takes some time before we see any productivity that is based on adult-
like syntactic categories. By contrast, a substantial body of research, in the
so-called syntactic bootstrapping framework, suggests that children do, in
fact, show signs of productivity at 24 months, or even younger. Efforts have
been made to accommodate these findings within the usage-based approach,
but it will become apparent that the dust has far from settled on the puzzle of
how and when children become syntactically productive.
The transitivity bias
As noted above, usage-based researchers argue that early constructions are
lexically specific. It takes some considerable time before the child manifests
any understanding of general syntactic categories. In consequence, child
speech should lack not only linguistic generalizations, but, logically, there
should be no overgeneralizations either. If you recall, overgeneralization is
the application of a linguistic rule (generalization) beyond its normal
confines, with the result that errors are produced. Tomasello (2003: 176)
asserts that ‘we should expect few overgeneralizations early; such errors
should begin only after a certain age – in the domain of syntax perhaps age
3’. But this position has proven to be very difficult to maintain. Others argue
that there are numerous signs of early productivity, with concomitant
overgeneralizations. In fact, we have only just noted that usage-based
researchers themselves now confirm that ‘children are capable of abstraction
from the beginning’ (Lieven & Tomasello, 2008: 170). A basic example is
the child’s ability to take a linguistic label like tree and apply it to trees of
different shapes and sizes (see Chapter 6). But we can also find examples in
the domain of syntax. Although children as young as 2;1 can successfully
interpret transitive sentences, their knowledge is fragile and errors are
produced (Brooks & Tomasello, 1999; Goldberg, 2006; Dittmar et al., 2011).
To appreciate this research, we need first to examine the notion of
transitivity in a bit more detail. As will become apparent, a good deal of
empirical work has been done on transitivity, both within and beyond the
usage-based approach. We can start by observing that the acquisition of each
new verb entails working out its particular transitivity status.
 transitive hit Ashley hit his friend.
 push Amanda pushed her luck.
intransitive fall Olga fell.
 sleep Richard slept.
bitransitive roll Kate rolled the ball. / The ball rolled.
 eat Clare ate a bhaji. / Clare ate.
As you can see, English verbs fall into one of three categories with respect to
transitivity: (1) transitive verbs, which take an obligatory direct object (the
noun phrase after the verb); (2) intransitive verbs, which never take a direct
object; and (3) bitransitive verbs, which swing both ways, with an optional
direct object, which means that they can occur in both transitive and
intransitive forms. The majority of verbs are bitransitive. That is, the majority
of verbs that can take an object can also appear without one (Huddleston,
1984).
Exercise 9.1: Verb transitivity
Categorize the following verbs on the basis of their transitivity
(transitive, intransitive or bitransitive). The best way of doing this is to
make up some sentences for yourself using the verb and some suitable,
simple noun phrases. Let’s try this with the verb cook:
transitive: The man cooked the steak. (YES)
intransitive: The man cooked. (YES)
Conclusion: cook works both ways, so it is bitransitive. Now try
these:
want wait feel
jump sleep disappear
leave take read
As mentioned, we witness child productivity in the domain of verb
transitivity (Kline & Demuth, 2014). In particular, children sometimes deploy
the transitive frame too broadly, applying it to intransitive verbs also, with
the result that errors occur: *Don’t giggle me or *Mommy I disappeared my
orange juice or *It won’t cough me no more (Bowerman, 1982; Brooks &
Zizak, 2002). These utterances would sound more natural if they were
restored to their normal intransitive status, with no direct object after the
verb, possibly something like: Don’t giggle; My orange juice disappeared;
and It won’t make me cough no more. Errors in the other direction are much
less common, that is, errors in which a transitive verb is mistakenly used in
an intransitive frame (e.g., *Ashley hit). Children are therefore said to have a
transitivity bias (Maratsos, Gudeman, Gerard-Ngo & DeHart, 1987). This
bias has been observed in experiments that introduce novel verbs to children
(Brooks & Tomasello, 1999; Brooks & Zizak, 2002). For example, children
were introduced to the novel verb tam, as shown in Figure 9.1.
Figure 9.1 The novel verb ‘tam’, depicting a particular manner of motion
(adapted from Brooks & Tomasello, 1999)

Tam is a so-called ‘manner of motion’ verb (like swing) and can appear in
both transitive and intransitive frames (as below). Children were exposed to
tam in just one of these two frames, in training sessions.
Transitive: The girl is tamming the house.
Intransitive: The house is tamming.
The aim was to determine how willing children are to use a verb with a
transitivity status different from the one they have experienced in training. In
this regard, Brooks & Zizak (2002) confirmed the transitivity bias in children
aged four to six years. If tam was modelled as transitive, children used it in a
transitive frame themselves on 88 per cent of occasions. But if tam was
modelled as intransitive, then children followed the intransitive pattern only
46 per cent of the time. In this latter condition, children were prone to use
tam in a transitive frame (the transitivity bias). The transitivity bias has also
been observed in children as young as 30 months (Brooks & Tomasello,
1999).
Pattern finding
The transitivity bias suggests that two-year-old children may possess a
general schema, or pattern, that guides them in their use of new verbs. The
child may no longer be tied to isolated utterances which they have heard
other people using. We need to consider, therefore, how the child achieves
this linguistic productivity. To this end, a number of possible mechanisms
have been suggested in usage-based theory, including: (1) distributional
analysis; (2) analogy; and (3) categorization. We will describe the first two
briefly here, and explore the third, categorization, in more detail in the next
section, in relation to the concept of type frequency.
Distributional analysis has been used by linguists for centuries as a method
for identifying the members of a syntactic category. It involves finding
patterns distributed across a corpus of language. For example, the category of
verb phrase minimally contains a verb, but can take on many different
forms. The distribution of verb phrases is such that they often appear after a
noun phrase. To see how this works, let’s first take a simple noun phrase,
our neighbour. Now consider the phrases below. Only those which can
follow our neighbour will qualify as verb phrases:
Our neighbour …
ate two profiteroles and a cabbage
a dog
should make less noise late at night
finally lost his marbles
died
friendly
Eliminating a dog and friendly, we see that all the other examples qualify as
verb phrases because they meet our distributional criterion. It has long been
suggested that even young children may perform distributional analyses (for
example, Maratsos, 1982). At the same time, it has not proven possible, yet,
to demonstrate empirically how children might actually manage this feat. The
same empirical problem faces the concept of analogy formation. The idea is
that the child can make analogies across isolated constructions, exploiting the
commonalities in evidence to extract a productive generalization. For
example, consider the following four sentences:
Maria dyes her hair.
Cathrine drinks vodka.
Lisa drives a motorbike.
André cures the insane.
As you will recognize, these are all simple transitive sentences, united both
by a similar linguistic structure and by a similar meaning, in which an agent
(e.g., Maria) causes something to happen to a patient (her hair). By analogy,
we can perceive the structural similarity that unites these four sentences. The
question then is, can the young child detect such analogies and exploit them
in the formation of broad linguistic categories? It is assumed that analogizing
is a domain-general cognitive skill, not one that is tied exclusively to
language. Analogies are formed by mapping information from a source entity
to a target entity. This mapping is made possible by recognizing some kind of
similarity between the source and the target. Following Gentner (1983), two
kinds of similarity are recognized: ‘substantial (or object) similarity, which
involves the recognition of shared attributes, and structural similarity, which
involves the recognition of shared structures or relationships’ (Diessel, 2009:
251). When it comes to syntax, it is this latter kind of analogy that the child
must make. Observe that we do not yet know how the child achieves the
necessary ‘recognition of similarity’. But Diessel & Tomasello (2005) do
show that, for children aged four and five years, structural similarity is a key
factor in the acquisition of German relative clauses (see also Brandt, Diessel
& Tomasello, 2008). Analogy formation provides one route to productivity.
Another possibility, entertained in usage-based theory, is the exploitation of
frequency information in the input. This ability in the child exemplifies the
third of our three ‘productivity mechanisms’ listed above: categorization. We
consider the role of frequency in the categorization process next.
Type frequency: A route to productivity
You will be familiar by now with the usage-based credo that knowledge of
language emerges from the use, or experience, of language. The amount of
experience is therefore critical, and is typically measured by the frequency of
language forms in the input. Two different kinds of frequency are identified:
token frequency and type frequency. Token frequency is simply the number
of times that a given linguistic item appears in a corpus of adult–child speech.
The token does not have to be a single word. It could be an entire phrase like
I dunno (Bybee & Scheibmann, 1999; Tomasello, 2003). Type frequency,
meanwhile, refers to the number of times that a linguistic form occurs in a
particular slot. In a particular sample of language, we might take a simple
frame, That’s [Noun Phrase], and count how many different noun phrases fill
the slot (That’s a teddy, That’s my book, and so on). The two kinds of
frequency are predicted to have different effects on child productivity. High
token frequency protects the child from error (see below), while high type
frequency allows the child to generalize their learning (Lieven & Tomasello,
2008). As we shall see, type frequency provides the child with an opportunity
to exercise their categorization skills.
The basic idea underlying type frequency is both simple and plausible. When
children develop a slot that can be filled by different items, they get an
opportunity to form generalizations about grammatical classes of items. A
simple example might be the slot in a simple noun phrase that can be filled by
different nouns, as in the +. This slot can be filled by a range of different
words (e.g., tomato, squirrel, tarmac, hydrangea). The appearance of these
words in the same slot may encourage the child to abstract the features that
these words have in common and, eventually, formulate a single category to
which they all belong (noun). By increasing type frequency, that is, the
number of different words that appear in the slot, the child will have more
opportunities to note the commonalities among different items and thus form
a broad grammatical category with a range of different members. In this way,
we can regard the use of type frequency as an example of the child’s
categorization abilities.
One study manipulated type frequency by varying the number of different
verbs that appeared in the following slot: It was the cup that the
frog___________, as in, It was the cup that the frog took (Ambridge,
Theakston, Lieven & Tomasello, 2006). In one condition, children heard ten
different verbs in this slot (including pull, touch, grab), compared with just
two in a second condition. It was predicted that the greater type frequency in
the first condition would help children become more proficient with this
complex structure. But this variation in type frequency made no difference
for children aged four and five years. The problem might lie in the
conception of what type frequency is and how it functions. There seems to be
an unwarranted expectation that the child can learn something about syntax –
how to put words together correctly into sentences – rather than something
much more modest, which is to identify the members of a syntactic category,
like noun. To put it another way, the process of filling slots is paradigmatic
rather than syntagmatic. To appreciate the difference, imagine paradigms on a
vertical axis, with syntagms on the horizontal. The members of a paradigm
occupy the same slot, and therefore belong to the same grammatical category
(noun in this case):
the tomato
 squirrel
 tarmac
 hydrangea
Syntagms, meanwhile, are ‘horizontal’ in the sense that different categories
are brought together, one after the other, to form larger groupings according
to the rules of grammar. These larger groupings of categories combine to
yield superordinate categories, organized hierarchically, as in the noun phrase
shown in Figure 9.2.
As we can see, one could learn something about the members of a syntactic
category from the items that fill a particular slot. And once the child is
equipped with a category like noun, then further progress could be made in
learning how this category combines with others (see Lieven, Behrens,
Speares & Tomasello, 2003). But even shifting to the syntax of a simple noun
phrase like the fat squirrel would represent a substantial advance. In this
light, it is perhaps no wonder that the children in Ambridge et al. (2006) did
not leap straight into the complexities of the structures underlying It was the
cup that the frog took. At best, therefore, type frequency could provide only a
very partial account of how the child makes linguistic generalizations.
Ambridge et al. (2006) stick closely to the standard definition of type
frequency: ‘the frequency with which different actual [sic] forms occur in the
same slot’ (Lieven & Tomasello, 2008: 174). In other cases, though, things
get rather muddled. Lieven & Tomasello (2008) exemplify the concept of
type frequency in two ways: (1) by comparing two forms of the German
present perfect; and (2) by comparing the regular past and regular plural in
English. In both cases, therefore, comparisons are made across different
constructions, which takes type frequency at least one step beyond what
happens in a single slot in a single construction. Bybee (2008) attempts to
salvage this position by broadening out the definition of type frequency to
include not only the different contents of a slot, but also ‘the number of items
that exemplify a pattern’ (ibid.: 221). With two kinds of type frequency at
large, though, the concept begins to lose its coherence. What, if anything,
underpins both kinds of type frequency? At present, we only have a vague
formulation to address this issue: ‘type frequency measures items that are
similar in some way’ (Lieven & Tomasello, 2008: 179). But the notion of
‘similarity’ is not yet well specified. It is perhaps not surprising, therefore,
that, as yet, ‘we are a long way from understanding all the ways in which
similarity is identified and changes with development’ (ibid.: 179). Evidently,
it has not yet been explained how, in the formation of linguistic categories,
the child might recruit some notion of ‘similarity’ or ‘analogy’ or
‘distribution’. In fact, this is an abiding problem which challenges all child
language researchers, not just those within the usage-based approach. The
following observation, made in 1982, remains just as true today:
Figure 9.2 Syntagmatic structure of a noun phrase

exactly how the child’s diverse analytic abilities and biases combine to
result in adult formal categorical organizations remains a central
problem. (Maratsos, 1982: 265)
Sounds familiar: The role of frequency
Usage-based theory could not get along without frequency effects. To be fair,
no theory could, but frequency holds a special place in the usage-based
pantheon. Numerous empirical studies in the usage-based tradition
underscore the effects of frequency on development (e.g., Rowland, 2007;
Abbot-Smith & Tomasello, 2009; Räsänen, Ambridge & Pine, 2014).
Intuitively, we can appreciate the importance of frequency. We all know that
practice makes perfect. Compare the first soufflé you ever made with your
latest effort (what? – you’ve never made a soufflé?). Exposure to the same
input conditions on repeated occasions leads to better learning. But as we
have seen, the concept of frequency is not that straightforward. There are
different kinds of frequency which at present are not well defined. And the
influence of frequency is bound to change if only because the child develops
and will be processing input in different ways to different ends over time. In
addition, frequency effects are not always observed (MacWhinney, 2014b).
Recall the fast mapping children are capable of in learning new words
(Chapter 6). A single exposure to a new word can suffice, on occasion, for
learning to take place. We should also be clear that – as usage-based theorists
allow – frequency by itself explains nothing. Frequency is not a learning
mechanism (Ambridge, Kidd, Rowland & Theakston, 2015). We still need an
explanation, therefore, for what the child does with input information and
why learning is rendered more effective by repeated exposure. That said, it is
worth highlighting those areas of development where frequency effects are
apparent (Ambridge et al.,ibid.):
1. Words. Words which occur frequently in the input are learned sooner
than less frequent forms (Fenson et al., 1994).
 2. Inflectional morphology. High frequency is associated with higher levels
of correct usage and, coincidentally, lower levels of error (see the
section on Constraining productivity, below). For example, Polish
children perform better marking dative, accusative and genitive case on
novel nouns in cases where input frequency is high (Dąbrowska &
Szczerbinski, 2006).
3. Simple grammar. Cues which assist in the learning of simple structures
– including the transitive – are more abundant in high-frequency
structures and are associated with more efficient learning (e.g., Matsuo,
Kita, Shinya, Wood & Naigles, 2012).
4. Complex grammar. The passive provides an example here. Passive
constructions are extremely rare in the input to children (and also rare in
Adult Directed Speech) (Gordon & Chafetz, 1991). Accordingly,
passives are rare in the speech of young children (Israel, Johnson &
Brooks, 2001).
Early productivity: Syntactic bootstrapping
The transitivity bias exposes a fundamental problem for usage-based theory.
Early syntactic productivity should not be in evidence, but it seems that child
creativity bursts out at an early age, regardless. Early syntactic prowess has
been remarked upon most pointedly within an alternative theoretical
framework, known as syntactic bootstrapping (for example, Naigles, 1990,
1996; Naigles & Kako, 1993; Fisher, Klingler & Song, 2006; Yuan & Fisher,
2009).
Syntactic bootstrapping is a procedure by which children use the syntax
in which a word is placed to narrow down or constrain the meaning of
the word. (Naigles & Swenson, 2007: 213)
Before we see how this works, observe that this prediction requires a
relatively sophisticated knowledge of syntactic categories in two-year-old
children. You may wonder, with some reason, where this knowledge comes
from, but since this is not our main focus, see Naigles & Swenson (2007) for
further discussion. We can, at least, note a sharp contrast with the usage-
based approach.
To illustrate, let’s return to the first example in this chapter: Max kissed
Anna. We first grant the child an understanding of causative verbs in
transitive sentence structures. Our task is to consider how the child might use
this knowledge of verb transitivity to work out the meaning of a new verb.
Imagine, for example, that kiss is a new word for the child. Since this word
appears in a familiar sentence frame, the child can use this structure to work
out something about the meaning and syntactic privileges of kiss. The typical
meaning conveyed by this structure, as a whole, is causative: an agent (Max)
causes something to happen to an object or person (Anna). In consequence,
the child who has never before encountered kiss will not be stranded. Its
appearance after Max, in this particular sentence frame, reveals that kiss is a
verb. Moreover, it is a causative verb, the precise meaning of which can be
inferred from observation of Max and Anna.
We want to know if two-year-olds understand the causative meaning
associated with the transitive sentence structure. Several studies suggest that
they do. In one study, children were shown videos of two Sesame Street
characters (Big Bird and Cookie Monster) interacting with each other (Hirsh-
Pasek, Golinkoff & Naigles, 1996). Children aged 19–28 months were shown
two different scenes simultaneously on a split screen, as shown in Figure 9.3
below.
Figure 9.3 24-month-old children prefer to look at a causative event (on the
left) when they hear a transitive sentence: ‘Look at Cookie Monster squatting
Big Bird’ (from Hirsh-Pasek et al., 1996)

For children as young as 24 months, when they hear Look at Cookie Monster
squatting Big Bird, their attention is drawn to the left-hand video. Of course,
this is appropriate, because Cookie Monster is actively doing something to
Big Bird in this video. In the right-hand video, by contrast, Big Bird and
Cookie Monster are just squatting independently, side by side. Neither figure
is making the other one do anything. Hirsh-Pasek et al. (1996) conclude that
children aged 24 months use syntax (the transitive frame) to guide their
interpretation of a verb (squat) which was unknown to them. You may think
that squat sounds odd when used transitively in this way. But if you have no
idea what squat means, nor how it is used, then you will see that the two-
year-old’s responses were appropriate. Subsequent work has argued that 21-
month-olds appreciate the significance of transitive word order in an abstract,
verb-general way (see the brief summary of Gertner et al., 2006, in Chapter
1). By 24 months, children can sometimes repair sentences which are missing
a direct object (e.g., The zebra brings something rather than The zebra
brings) (Naigles & Maltempo, 2011). This level of sophistication is not
predicted by usage-based theory at such a young age.
Generally, there is a close correspondence between syntax and meaning
(semantics) which the child can make use of. Other theories, including usage-
based theory, also rest on the assumption that the child can exploit the close
links between syntax and semantics. Usage-based theory and syntactic
bootstrapping share other basic assumptions, too. For example, they both give
prominence to learning about the meanings and constraints on individual
lexical items (Fisher, 2002). And they both give priority to the utterance,
acknowledging that most of the words children hear are used in sentences,
not in isolation. But the two theories part company with respect to the stage at
which children are capable of making genuinely syntactic generalizations.
Usage-based researchers continue to maintain that the child’s early multi-
word speech is not based on genuinely syntactic categories (Peter, Chang,
Pine, Blything & Rowland, 2015). They shore up their position by
highlighting differences in methodology across the two camps. One critical
difference is that syntactic bootstrapping studies tend to rely on child
comprehension, while usage-based work examines the child’s speech output.
We have already seen that the level of comprehension shown by a child is
typically out of kilter with their productive expertise (see Chapter 6). In
essence, children look like syntactic wizards when all we measure is their
viewing preferences, but far less competent when they open their mouths to
talk.
Two recent studies suggest that two-year-olds may have some grasp of verb
transitivity, but that their knowledge is only partial (Dittmar, Abbot-Smith,
Lieven & Tomasello, 2008a, 2008b). In English, word order provides an
overwhelmingly important cue concerning transitivity. In German, the
situation is more complex. Both word order and case marking provide cues to
the identity of the agent. The agent noun bears a special morphological
marker (for nominative case) to indicate its status as agent. And, as in
English, it often appears first in the sentence. Thus, the two cues of word
order and case marking can support each other. But it is possible for them to
conflict: an agent with nominative marking does not have to come first in a
German sentence. In such cases, case marking takes precedence as the means
for identifying the agent. Dittmar et al. (2008a) found that German two-year-
olds showed some understanding of verb transitivity, but only when both
cues coincided. Five-year-olds, meanwhile, could cope with word order by
itself, but not case marking. Only seven-year-olds could cope with both cues,
and rely on case marking alone when the two cues were placed in conflict
with one another. Dittmar et al. (2008b) also show that two-year-olds require
extensive prior training with the specific verbs used in any test of transitivity
in order to succeed at all. In consequence, Dittmar et al. (2008b) argue that
two-year-olds possess a weak representation of transitivity only. What we
need now, of course, is some principled way of measuring the strength of
mental representations in young children: what does it mean to have a weak
representation?
Discussion point 9.1: Syntactic productivity
Assemble in groups of four and then split into two pairs. Each pair
should take one of the articles below, and then read it independently:
Dittmar, M., Abbot-Smith, K., Lieven, E. & Tomasello, M. (2008b).
Young German children’s early syntactic competence: A preferential
looking study. Developmental Science, 11(4), 575–582.
Yuan, S. & Fisher, C. (2009). ‘Really? She blicked the baby?’: Two-
year-olds learn combinatorial facts about verbs by listening.
Psychological Science, 20(5), 619–626.
Compare notes with your partner and try to resolve any difficulties you
may have encountered with the text. Then regroup as a foursome and
discuss the different perspectives which the two papers bring to bear on
the issue of syntactic productivity.
Gather together the arguments and evidence both for and against
each position. Make notes. The four of you will need to pool
resources to get a complete picture.
At what age do children exploit knowledge of syntax in a
 productive way?
What are the limitations of the preferential looking method?
Which account do you find more convincing? Why?
Generally speaking, child productivity with syntax is a major headache for
theories of child language acquisition. Recent research has done much to
shed light on the productivity puzzle, but there is still more to learn. The
simple usage-based prediction that children are not productive in syntax
before the age of three cannot be sustained. In comprehension, at least,
children display some grasp of verb transitivity at 24 months. Usage-based
researchers have tried to accommodate this early productivity with the idea
that comprehension studies tap into ‘weak’ syntactic representations. This
idea lacks refinement at present, but recent usage-based studies have shown
that, however impressive the two-year-old’s syntactic prowess might appear,
their knowledge of transitivity is not adult-like until the school years, some
time after the child’s fifth birthday. It will be interesting to see how different
theoretical approaches converge on the solution to the productivity problem
in the coming years.
Constraining productivity
As Berko (1958: 150) pointed out: ‘the acquisition of language is more than
the storing up of rehearsed utterances, since we are all able to say what we
have not practiced and what we have never before heard’. But linguistic
creativity, encountered in Chapters 2 and 4, presents a problem that taxes
every theory of syntax acquisition: how does the child constrain
productivity? What stops the child from going too wild and producing all
manner of sentences that are not, in fact, permissible in the adult grammar?
The problem is that the child must become productive, but not too
productive. In this section, we review two mechanisms for reining in
linguistic generalizations, advanced within the usage-based approach:
1. Entrenchment: increasing experience with a particular structure makes it
less likely that errors will occur.
2. Pre-emption: hearing a verb used in a rare structure may prevent
erroneous use of that verb in a more common structure.
Both of these mechanisms rest on the assumption that children are inherently
cautious, or conservative, in their learning. This assumption underpins pretty
much every theoretical approach to syntax acquisition in young children
(including the nativist approach considered in Chapter 8). It would be useful,
therefore, to set the scene with a brief overview of the conservative child,
before we see how this general idea manifests in entrenchment and pre-
emption.
Conservative learning
In acquiring a new verb, the child must work out its transitivity status. One
way of doing this would be to simply observe what adults do and follow their
lead. This approach is known as conservative learning, a phrase which
denotes a cautious approach in the use of new structures. An entirely
conservative learner would never use a verb in a structure that they had not
actually witnessed for themselves. This seems like a sensible approach
because learning will be error-free. To some extent at least, children must be
conservative learners, because they are not rampantly unconstrained in their
use of language. We do not hear bizarre sequences like *His hit Ashley friend
or *Bhaji a ate Clare. For this reason, all language acquisition theories, both
nativist and non-nativist, assume a degree of conservatism in the child
(Pinker, 1989; Radford, 1990; Tomasello, 2003). In the nativist approach, the
child is held in check by Universal Grammar (UG). The child’s mind is
configured to acquire only certain kinds of syntactic knowledge, those
described in UG theory (see Chapter 8). In the usage-based theory, the child’s
conservatism is ascribed to the influence of the input. In essence, the child
sticks closely to what they hear in their own speech output, a theme we
elaborate on below. A fundamental challenge for every theory is that children
are not confined in a linguistic straitjacket. We have seen numerous examples
throughout this book of child errors, cases where children overgeneralize the
use of linguistic structures beyond their conventional scope. Children are not
unfailingly conservative. The abiding challenge is to explain why the child is
conservative, but only partially conservative.
Entrenchment
In the usage-based approach, frequency is paramount. The frequency of
linguistic forms in the input provides a key stimulus for language
development. We have already considered type frequency, above, and now
turn our attention to token frequency. For a given corpus, token frequency is
the number of times that a linguistic form occurs in the input to the child. A
relationship is assumed between exposure to a linguistic form and the mental
representation of that form in the child. In short, the more often a linguistic
form occurs in the input, the more often it is experienced by the child, and the
stronger the child’s representation of it becomes. On this view, every time a
linguistic structure occurs, its mental representation in the child is reinforced
or increasingly entrenched. An important consequence for the child is that
this structure will be activated more easily when using it themselves on
subsequent occasions (Diessel, 2009). The actual use of language therefore
affects the learning that takes place. On this view, it is predicted that
linguistic forms with high token frequency will be learned early and lead to
more strongly entrenched linguistic representations (Brooks, Tomasello,
Dodson & Lewis, 1999; Theakston & Lieven, 2008). And the more deeply
entrenched a structure is, the more likely it becomes that this will form the
basis of the child’s own speech output. In consequence, the likelihood of the
child being diverted into producing an error is diminished. We see, therefore,
that high token frequency – as the basis of entrenchment – is predicted to
protect the child from error. Confusingly, Lieven contradicts herself on this
point, arguing that ‘heavily entrenched word sequences compete with the
emerging generalizations in production, and this leads to the production of
errors’ (Bannard & Lieven, 2012: 9). However, the standard usage-based
orthodoxy regards entrenchment as a protection against error.
If token frequency has a strong influence on child learning, we should expect
to see a close relationship between adult input and child output. And, in
particular cases, this is exactly what we do see, especially in the early stages.
One study of children aged 1;9 to 2;6 found that 45 per cent of all maternal
utterances began with one of the following 17 lexemes: what (most frequent
at 8.6 per cent), that; it; you; are/aren’t; I; do/does/did/don’t; is; shall; a;
can/can’t; where; there; who; come; look; let’s (these last three being least
frequent at 1 per cent each) (Cameron-Faulkner et al., 2003). In other words,
children hear the same few words over and over again. Moreover, these
words occur in very specific, item-based frames. For example, why occurred
in four different frames:
why don’t __________
 do __________
 ’s __________
 not __________
We can assess the influence of input frequency by examining which
structures children use themselves most often. As predicted, those structures
children hear most often are, in certain cases, the ones that they, in turn, use
most often themselves. This is true of the copula(s) in the following three
frames: There’s [NP]; That’s [NP]; and It’s [NP]. As ever, NP stands for
noun phrase (a dog, my hat). The frequency of maternal usage correlates
strongly with child usage in all three frames (Cameron-Faulkner et al., 2003).
Similar relationships have been found for other structures, too, including uses
of the verb go and, more generally, the child’s first uses of particular verbs
(Theakston, Lieven, Pine & Rowland, 2002, 2004; Ambridge et al., 2008).
The evidence so far suggests that high token frequency leads to greater
entrenchment. We now need to consider how entrenchment might protect the
child from error. Rowland (2007) addresses this question in a study that
looked at children’s production of different question types. Errors of the
following kind were observed in a corpus of ten children aged 2;1 to 5;1:
where does he goes?
where does he does go?
where does you go?
where he does go?
does donuts also have TV?
does her like it?
does he going to the shops?
Don’t get the wrong impression from this list. Overall, errors with questions
are fairly infrequent, occurring only 7 per cent of the time. But certain kinds
of question were more problematic, most notably, WH-questions, those
questions that start with a WH- word like what, why or where. One particular
kind of WH-question starts with a WH-word and is followed by a modal
auxiliary verb: Where will you go? What can you see? Who shall I invite
(do/does is an auxiliary verb, but not a modal auxiliary). For this kind of
question, the child error rate was relatively high at 18 per cent. Of interest
here, the frequency of such questions in the input was relatively low. Hence,
errors are more frequent in cases where the child lacks experience. In
contrast, high token frequency seems to protect the child from error. Even so,
we might point out that the level of protection afforded is not that dramatic.
Entrenchment is also in evidence in a study where children were asked to
judge the acceptability of structures (e.g., Daddy giggled the baby). Children
aged five to ten years were more likely to reject ungrammatical sentences
which deployed high frequency verbs (Ambridge, Bidgood, Twomey, Pine,
Rowland & Freudenthal, 2015). Again, then high frequency protected the
children from error.
The concept of entrenchment conjures up an image of doughty toddlers,
valiantly digging their way down into a linguistic trench. The deeper they dig
themselves in, the more protected they are from enemy fire, in the form of
grammatical errors. But this notion of entrenchment raises three problems.
First, child grammar develops. In usage-based terminology, islands are
connected, analogies are drawn, connections between related constructions
are made, and schemas become increasingly generalized. But what, in this
shifting array, is entrenched? The most obvious answer is that the child’s
earliest linguistic constructions become entrenched (holophrases, early word
combinations, pivot schemas, item-based constructions). More complex
structures cannot become entrenched early on, because the child has not yet
acquired them. And input frequency cannot be put on hold until some
convenient moment, later in development, before it starts exerting its effects.
But if early constructions are entrenched, we need to explain how the child
climbs out of the trench and moves on to the next level of syntactic
complexity. But how can this be achieved, if initial representations are
entrenched? The more entrenched a structure is, the less susceptible it should
be to change.
The second problem is related to the first, and also stems from the fact that
grammar develops. As the child moves through successive points in
development, it seems unlikely that the child’s knowledge at each point
becomes separately entrenched, in successive turns. But intermediate states of
knowledge should each be subject to the same forces of entrenchment. One
might seek an exemption for intermediate states and determine that only the
endpoint, the equivalent of adult grammatical knowledge, becomes
entrenched. But logically, that cannot work. From the child’s point of view, a
particular state of linguistic knowledge is not intermediate. It does not
constitute a staging post on the way to somewhere else (more complex, more
adult, more mature). It only looks intermediate from an adult point of view.
The general problem, then, is to explain what, in development, becomes
entrenched and why. This issue has not yet been broached by usage-based
researchers.
The third problem with entrenchment is that, in fact, it does not apply in
many cases. Instead, entrenchment is selective. This follows, in part, because
child speech is never going to be the mirror image of parental input. For one
thing, children younger than two years will not be using any of the more
complex structures found in parental speech. And some structures will be
avoided for pragmatic reasons. For example, imperatives like Sit down! or
Eat up! are frequent in maternal speech but rare in child speech. Two-year-
olds have their own ways of bossing their parents about. Dąbrowska (2005:
201) also notes a lack of entrenchment with her observation that, even at the
age of four years, Polish children continue to make case marking errors with
the genitive, despite having heard ‘the genitive form of a masculine noun one
every two or three minutes’ for several years. And Maslen et al. (2004) report
that some verbs, like come, continue to be overregularized as comed (rather
than came) even after many thousands of exposures to the correct form.
Without doubt, frequency is an important factor in language acquisition. The
case of WH-questions, plus several other examples raised in Chapter 4,
confirm the significance of input frequency effects. In a handful of cases,
there are signs that high input frequency may serve to protect the child
somewhat against grammatical errors. But we need to establish why
frequency effects are in evidence in some cases, but not in others. Evidently,
the frequency of linguistic forms in the input is just one factor at work.
Hopefully, therefore, researchers will broaden their palette of enquiry to
consider a wider range of factors. As it stands, the child’s linguistic
environment is characterized in a rather simplistic manner. For example, it
neglects any consideration of adult–child interaction (Chapter 4). And
research on input frequency does not yet account for how the child’s learning
mechanisms respond to different kinds of input. For example, it is plausible
that, in some instances, rare input events could have a major impact on the
child. A relatively large step forward would be made on the basis of relatively
little experience. Successful acquisition of different linguistic structures may
require different amounts and kinds of both input and interaction, and these
will likely exert their effects in complex ways (Goodman, Dale & Li, 2008).
We can conclude this section by drawing a careful distinction between
frequency effects and entrenchment. In usage-based theory, it is predicted
that entrenchment helps explain how input frequency exerts its influence. But
there is reason to believe that the concept of entrenchment is flawed.
Fundamentally, it does not sit well with the idea that grammar develops,
because it suggests that early, immature constructions will become
entrenched. In consequence, one is left with the difficult problem of
explaining how the child climbs out of the trench and moves on to the next
stage. Happily, usage-based theory offers a further mechanism that may help
protect the child from error, known as pre-emption. We consider pre-emption
in what follows.
Pre-emption
We start this section with three verbs, swing, wash and play, that can occur in
transitive sentences with an agent as subject:
Wilkie swung his golf club.
Lorraine washed her towels.
Phuong played her favourite DVD.
As noted above, verbs of this kind are causative (the agent causes something
to happen). Other verbs, like disappear, can also express a causative
meaning, but not in this kind of sentence pattern. We cannot say *The
magician disappeared the rabbit. Instead, we have to resort to a rarer
construction known as the periphrastic causative and say The magician made
the rabbit disappear.
The idea behind pre-emption is that when children hear a verb being used in a
rare construction, they will be put off using it in the more common
construction available (Goldberg, 1995; 2011). Thus, exposure to disappear
in a periphrastic causative sentence will pre-empt its use by the child in a
simple transitive construction. Potentially, therefore, pre-emption could
counteract the effects of input frequency. It allows rarer constructions to exert
an influence where necessary. Also, there is no need to panic if errors do crop
up from time to time. The child must simply be patient. Eventually, a
periphrastic causative will turn up and pre-empt further errors. Of course,
periphrastic causatives may well be like the 37 bus. You wait for hours in
vain and then two turn up at once. Or worse, none turn up at all.
There have been very few attempts to test the idea that pre-emption may
work as an error protection mechanism. With the focus on the periphrastic
causative, Brooks & Zizak (2002) introduced children aged four and six years
to two novel verbs: our old friend tam (see Figure 9.1) and dack. The
meaning of both verbs involved a particular manner of motion, but one was
designated as transitive (The mouse dacked the block), with the other being
intransitive (The tree is tamming). In one condition, children were exposed to
pre-emptive uses of tam and dack by the experimenter, in periphrastic
causative sentences like The rabbit made the car dack. If pre-emption works,
then this experience should discourage the child from using dack in simple
causative sentences (*The rabbit dacked the car). As described above, in
children’s own use of these verbs, the transitivity bias was found. This means
that, generally speaking, errors were more likely with the intransitive verb.
However, such errors were rarer for children in the pre-emption condition. It
seems, therefore, that both four- and six-year-olds were protected from error.
Remarkably, it did not take much pre-emption to produce this effect.
Children only heard 12 pre-emptive sentences with each verb (see also
Brooks & Tomasello, 1999).
Less encouraging results have been reported by Ambridge et al. (2015). No
pre-emptive effect was observed for children aged five and nine years when
exposed to the rare versions of a verb argument structure (e.g., Lisa made
the cup fall). Children rated the acceptability of sentences on a five-point
scale. This provided a measure of their willingness to reject pre-empted
sentence constructions (*Lisa fell the cup). It turned out that children of both
ages accepted ungrammatical overgeneralized sentences more readily than
adults and that, overall, there was no evidence of pre-emption. We have a
conflict, therefore, with the findings of Brooks & Zizak (2002). Ambridge et
al. (2015) suggest that pre-emption may only be observed when the pre-
empting construction is much more frequent than the erroneous version. The
problem is, though, that pre-empting constructions are very rare in Child
Directed Speech (0.4 per cent in the case of the periphrastic causative). But
recall that Brooks & Zizak not only used novel verbs, they found a pre-
emptive effect after only 12 exposures to each verb. What’s more, these
structures are meant to be rare by definition. It is their very unusualness
which is supposed to ring alarm bells in the child, prompting them to reject
more conventional forms for particular verbs. Incidentally, this point
highlights an intrinsic tension between entrenchment and pre-emption. While
pre-emption trades on the rarity of structures, entrenchment depends on
frequency to account for conservative learning. Frequency is at work with
pre-emption, to the extent that the rarity of a pre-emptive structure is
established in comparison with its frequent counterpart. But how does the
child manage to override the effects of entrenchment, in special cases only,
and suddenly be so impressed by a structure precisely because of its scarcity?
The outlook for pre-emption as a mechanism in child language does not look
entirely secure.
Summary: Reining back on productivity
This section has considered how the child might constrain productivity within
the usage-based approach. The two mechanisms we examined were
entrenchment and pre-emption. Increasing experience with a structure is
supposed to cause it to become entrenched in the child’s system. This then
protects the child from error, because the structures they have heard most
often will be accessed most quickly and easily, reducing the chances that
erroneous alternatives will be selected. We saw that frequency effects are
present in some, but by no means all, cases. More problematic, though, is the
concept of entrenchment. How does the child climb out of the trench and
move on to acquire more complex, adult-like structures? Frequency effects
are legion in psychology, both within and beyond the confines of child
language. What we see, therefore, is that the concept of entrenchment may
not be the best way of accounting for them. We have also examined pre-
emption, the idea that exposure to a rare structure can stop the child from
producing a more obvious, more frequent alternative. Pre-emption is an anti-
frequency device that has been demonstrated for at least one structure, the
periphrastic causative, in older children aged four years and upwards.
Overall, usage-based theory still has some considerable way to go in tackling
how the child constrains productivity. We should not be too surprised at this,
though. It remains a perennial challenge for all theories of syntax acquisition.
In a Nutshell
Usage-based theory argues that knowledge of language
develops from the use of language. It is a non-nativist theory.
The social skills of communication which develop in the first
year are assumed to provide the precursor for language
acquisition. However, language and social cognition seem to
develop in tandem.
In usage-based theory, early child speech lacks adult
syntactic categories. The child moves towards syntax via
four utterance types:
holophrases: express a single communicative intention,
even if they appear to comprise more than one word.
word combinations: two-word utterances, where each
word has roughly the same status.
pivot schemas: a pivot word and an empty slot filled by
words with the same communicative function.
item-based schemas: word combinations which are
lexically specific, but which bear the early signs of
syntactic marking (for example, via morphological
marking or word order).
Heavy emphasis is placed on input frequency, which means
that other aspects of input and interaction are neglected. Two
 kinds of input frequency are identified:
type frequency: the number of different items that
appear in a particular item-based slot. The concept of
type frequency is not yet well specified and, to date, its
predicted effects on language learning are not supported
empirically.
token frequency: the number of times that a linguistic
item appears in a corpus of adult–child speech (see
entrenchment below).
The child develops increasingly general syntactic schemas,
by identifying the similarities and patterns across different
item-based schemas. Distributional analysis, analogy and
categorization are assumed to contribute to this end.
Two mechanisms are predicted to help children constrain
their developing productivity appropriately and avoid
overgeneralizations:
entrenchment: high token frequency is predicted to
cause increasingly strong, or entrenched,
representations of linguistic structures. But the concept
of entrenchment is problematic. It is not clear how the
child moves on from an early entrenched structure to
acquire more adult-like forms of syntactic knowledge.
pre-emption: an anti-frequency mechanism. Experience
of a verb in a rare construction will cause the child to
avoid using that verb in a more common structure.
There is conflicting evidence about how sensitive
children are to pre-emptive input experience.
Further Reading
Tomasello, M. (2009). The usage-based theory of language
acquisition. In E.L. Bavin (Ed.), The Cambridge handbook of
child language (pp. 69–87). Cambridge: Cambridge University
Press.
In this chapter, Tomasello sets out the usage-based stall with a
broad overview of the theory. You could read this chapter as if it
were a manifesto. You end up with a very clear statement of the
usage-based position, but you will not get much feel for the
substantive challenges that the theory still faces.
Gleitman, L.R. (2014). Syntactic bootstrapping. In P.J. Brooks &
V. Kempe (Eds.), Encyclopedia of language development (pp.
616–618). London: Sage.
Gleitman provides a succinct overview of syntactic bootstrapping,
the alternative to the usage-based approach mentioned here. She
argues that children acquire syntactic categories early and exploit
this knowledge to infer the meanings of new verbs. See also
Naigles & Swenson (2007) for a more comprehensive review.
Websites
Michael Tomasello’s academic homepage:
www.eva.mpg.de/psycho/staff/tomas/
You will have gathered by now that the driving force behind
the usage-based approach to language development is
Michael Tomasello. His homepage provides an excellent
source of recent, downloadable research reports on child
language. Tomasello has wide research interests, so you will
also come across interesting comparative research on
communication in chimpanzees (relevant to Chapter 2).
Cynthia Fisher’s academic homepage:
http://www.psychology.illinois.edu/people/clfishe
Cynthia Fisher is a leading exponent of syntactic
bootstrapping, a theory which ascribes a relatively
sophisticated level of syntactic knowledge to children by the
age of 24 months. This work therefore provides a direct
challenge to the usage-based belief that early child speech is
not based on productive, adult-like syntactic generalizations.
If the URL defeats you, Google Cynthia Fisher and you’ll
get there quite soon.
Still want more? For links to online resources relevant to this
chapter and a quiz to test your understanding, visit the companion
website at https://study.sagepub.com/saxton2e
10 You Say Nature, I Say Nurture: Better
Call the Calling Off Off
Contents
Nature and nurture in the study of child language 266
The genetic basis of language development 266
Integrating ‘nurture’ into theories of syntax acquisition 267
Some basic facts 269
Dietrich Tiedemann (1787) 269
Child language: A timeline 270
The ‘nature’ in nature–nurture: Something must be innate 271
Learning mechanisms 272
Domain-general learning: How general? 272
Linguistic nativism: The need for learning mechanisms 274
Methodology: Limitations and possibilities 275
Language acquisition: The state of the art 278
Child language: Acquisition and development 281
Overview
This chapter provides a review of facts and ideas presented in the
preceding nine chapters. In so doing, the abiding theme of nature and
nurture is adopted as a framework for discussion.
At the end of this chapter, you should have a good grasp of some of the
basic facts of child language acquisition, having organized them on a
timeline. You should also be able to describe some of the fundamental
challenges facing both nativist and non-nativist theories of child
language. With regard to the non-nativist position, you should be aware
of the problems raised by the notion of a domain-general learning
mechanism. It is unlikely that any learning mechanism is truly domain
general and, moreover, it is likely that several learning mechanisms
may be required to help account for language acquisition. In the
nativist case, you should be aware that language-learning mechanisms
are required, irrespective of whether the child is genetically endowed
with Universal Grammar.
A broad overview of methodology is also provided, with a reminder of
the variety and ingenuity on display in child language research.
Mention is also made of new methods, especially in neuroscience, that
 promise much for the future. Finally, I ask 40 prominent child language
researchers to gaze into my crystal ball and predict the future. No, not
the winning lottery numbers (I missed a trick there). But how they view
recent achievements in the field and where they envisage the next
major advances will come in child language research.
Nature and nurture in the study of child language
The genetic basis of language development
We saw in Chapter 8 that Chomsky’s theory of Universal Grammar has not
fared well empirically. It is one thing to demonstrate that a particular aspect
of syntax, like structure dependence, is a universal property of human
language. It is quite another to demonstrate that it is genetically determined.
The method Chomsky offers does not seem viable empirically, even though,
logically, it makes perfect sense: any aspect of our knowledge of syntax
which could not have been attained through experience must be inborn. The
empirical problem is to demonstrate the poverty of our experience, relative to
the richness of what we come to know. This problem, if we accept it as a
genuine problem, has not yet been solved. As we saw in Chapter 4, it is no
straightforward matter to demonstrate that the input is impoverished.
Meanwhile, an increasing number of authors have mounted demonstrations
that syntax can be acquired from the information available to the child (see
Clark & Lappin, 2011). Chomsky’s approach to the ‘nature of nature’
question is not the only one available. Two prominent alternatives are: (1)
twin studies in which the relative contribution of genes to a particular
behaviour is estimated; and (2) molecular genetic research which aims to
identify specific genes that are implicated in language development. I will
make only brief mention of them here, but you can follow them up from the
Further Reading section, below.
Twin studies have a long history in the study of intelligence and IQ
(Mackintosh, 1998). The two members of any pair of twins share the same
environment as they grow up (to a significant degree). But whereas identical
twins share 100 per cent of their genes, non-identical twins share only 50 per
cent of their genes. Therefore, if a given behaviour is due, even in part, to
genes, we would expect identical twins to resemble each other more closely
than non-identical twins. This reasoning allows one to calculate a so-called
heritability estimate, a figure which indicates what proportion of the variance
in a behaviour can be attributed to genetic factors. Thus far, twin studies of
this kind suggest that two aspects of language-related behaviour may be
partly determined by genetic factors: (1) morphology and syntax; and (2)
phonological short-term memory (Newbury, Bishop & Monaco, 2005).
Molecular genetic research, meanwhile, provides a second method for
investigating the genetic contribution to language acquisition. Individual
genes which may be important can now be identified. At the same time, the
links between cognitive functions (like language) and the action of genes is
both indirect and complex. Researchers are only just beginning to think about
how to unravel the causal chains between genes and behaviour, but the
prospect of being able to do so is at least in sight, somewhere on the horizon.
To conclude, Chomsky’s method for identifying inherited aspects of
language – based on his poverty of stimulus argument – has not proven
satisfactory empirically. Molecular genetics promises more in this direction,
but the following observation provides a cautionary tale:
Complex traits are not controlled by single genes (Risch, 2000). ‘At
least 40 loci have been associated with human height, a classic complex
trait with an estimated heritability of about 80%, yet they explain only
about 5% of phenotypic variance despite studies of tens of thousands of
people’ (Manolio et al., 2009). (Valian, 2014: 84)
Identifying 40 separate gene loci is one thing. Puzzling out how they actually
operate in development is quite another. If human height is so complex –
including significant effects of environment – then one can only guess at the
task facing researchers when it comes to language acquisition.
Integrating ‘nurture’ into theories of syntax
acquisition
The phrase ‘nature–nurture,’ with its catchphrase quality and popular appeal,
is deceptive. ‘Nature–nurture’ implies an opposition between the learner
(nature) and the input to learning (nurture). But this is a false opposition and
you will not find any serious researcher willing to subscribe to it. Instead,
debate centres on the characteristics of both nature and nurture and the
complex ways in which they interact. With respect to nature, we have just
seen that non-nativist approaches must appeal to innate learning mechanisms.
They are not the sole preserve of linguistic nativists. With respect to nurture,
all theories should take into account the linguistic environment and explain
how the child makes use of it in the course of language acquisition. You may
have gathered from Chapters 8 and 9 that the child’s linguistic environment
does not, in fact, feature very prominently in current theories of syntax
acquisition. In both nativist and non-nativist theories, the influence of the
linguistic environment is downplayed in several ways. First, the emphasis is
almost exclusively on input, rather than interaction, as the source of
information for learning. And second, within the domain of input, the focus
has been on just one feature: frequency of occurrence. On this approach, the
critical factor is the sheer number of times the child hears a particular
linguistic form. In the nativist approach, very low frequencies (potentially a
single exposure) might be all that is required to trigger successful acquisition
(Chapter 8). In the non-nativist, usage-based approach, huge input
frequencies are implicated as a key factor in explaining how the child
achieves deeply ‘entrenched’ learning. We observed problems with the
concept of entrenchment in Chapter 9. And Valian (2014: 82) points out that
‘frequency effects are … inconsistent and limited. Children produce
structures they have never heard’. Low frequencies or high, in both cases,
reference to the child’s linguistic environment reduces largely to a matter of
input frequency (Saxton, 2009). The richness of interaction and what it can
yield for learning are neglected. As we saw in Chapter 4, there are numerous
examples of how the quality of both input and interaction impact on language
acquisition. In the 1970s, there was an explosion of interest in the linguistic
‘nurture’ of the child, but the numerous discoveries made at that time, and
since then, have not yet been fully integrated within a theory of the child’s
linguistic nature, either nativist or non-nativist. Nurture has not been that well
nurtured.
Discussion point 10.1: Nature and nurture in
the study of child language
A good deal of research has been motivated by fascination with the
ways in which genes and environment influence language acquisition.
This is perhaps not surprising, given the massive influence of
Chomsky’s ideas over the past half century. As another good review
exercise, skim each chapter for topics of interest and investigate what
is known and/or believed about the genetic basis of the behaviour in
question. For instance, you might take categorical perception of
phonemes (Chapter 5), or the shape bias in word learning (Chapter 6),
or the blocking mechanism in morphology (Chapter 7). Or something
else. Try to be specific in your chosen topics (not just grammar or
 words, say). If you work in a group, you could split this task and each
bring something different to the discussion. In your review, address the
following questions, bearing in mind that you’ll get a fuller picture if
you track down some of the original sources. For each aspect of
language:
What assumptions, if any, do different researchers make about the
role of genes? Dig deep: sometimes these assumptions are
implicit.
What empirical evidence do we have concerning the role of
genes? How convincing is this evidence?
Is the chosen language feature assumed to be unique to human
development?
Is it assumed to be both unique to humans and to language? The
distinction between Faculty of Language Broad versus Narrow –
FLB versus FLN – may help out here (Chapter 2).
What is known about the role of the environment? How
convincing is this evidence?
You may well conclude from your discussion that, in many cases, claims
concerning the genetic basis of language development are not that well
supported empirically. Put another way, almost every assertion concerning
the genetic basis of language remains contested, especially when claims are
made about the human-specific and/or language-specific nature of genetic
influence. As noted, explanations of language acquisition are unfailingly
controversial. This can be quite frustrating. Does anyone agree on anything?
Well, on the basic facts, the answer is, largely, yes. On theory and
explanation, no. But don’t be frustrated. Exult, instead, at the richness and
diversity of ideas and argument that we get to play with. There is plenty of
room for you, Dear Reader, to jump in as a researcher of the future and find
answers to the questions: How? and Why?
Before I shuffle off for a cup of tea, let’s give brief consideration to one
reason why we do not yet have all the answers. Put simply, psychology is
hard. We ask big questions about big topics, like mind, consciousness and
development. Research on language acquisition, as a branch of psychology, is
not exempt. In fact, it faces the special problems encapsulated by W.C.
Fields, a movie star comic from the first half of the twentieth century: ‘Never
work with children or animals’. Child language researchers do both. To
encapsulate the problem: how do we get inside the mind of a child (or
animal) and find out what is going on? It is difficult enough with adults, but
as we noted in Chapter 1, we cannot simply ask children (or animals) to tell
us what they know. You need language – the very thing we are interested in –
in order to make this communication. And of course, very young children
lack sufficient language to provide us with any workable answers. What can
we do instead? Plenty, in fact. Hopefully, you will have been struck already
by the ingenuity in the range of different research methods deployed by
researchers.
Some basic facts
Dietrich Tiedemann (1787)
The first systematic study we have of child language acquisition was
published by a German biologist, Dietrich Tiedemann, in 1787. Tiedemann
kept a record of his son’s development – including language development –
from birth through to the age of almost 30 months (anglophones can refer to
the translation by Murchison & Langer, 1927). What we discover in
Tiedemann’s work is a rich set of observations that are both accurate and a
source of perennial research interest. For example, there are notes on
imitation, object naming, inflectional morphology, gesture–speech
combinations, shared intentionality, grammatical errors and effects of input
frequency. Tiedemann was ahead of his time. As you will have noticed, the
topics that caught his attention excite the interest of researchers currently,
more than two centuries later. Extracts from Tiedemann’s work are gathered
together in Appendix 1. You will also find reference there to chapters in this
book where the various topics are discussed from a twenty-first-century
perspective. Tiedemann’s observations are impressive, but, of course, they do
not add up to a complete record of child language. Nothing like. But spooling
forward all the way to the 1980s, it was still possible for one commentator to
assert that ‘what is both desirable and possible in the study of language
development at the present time is more facts, more flower-picking natural
history’ (Campbell, 1986: 30). In other words, at that time it seemed as if
many of the basic facts had still not been established. This, at least, is far less
true today. We now have a substantial body of evidence to tussle over,
following a huge increase in the volume of research over the past 30 years or
more. Research on child language has been so prolific, in fact, that no single
volume could now contain it all. I have pointed the way to this wider world
by throwing the spotlight on some (just some) of the key topics and points of
dispute. Hopefully, you now feel better equipped to explore further for
yourself.
Child language: A timeline
Exercise 10.1: Review of some basic facts
Inspired by Tiedemann, let’s try to get some basic facts together. Copy
the timeline below onto the biggest piece of paper you can find and
then insert key features of child language according to the age at which
they first appear. The space above the line is for comprehension – the
age at which the child shows initial understanding of some key
linguistic feature. And the space below the line is for production – the
age at which the child actually first produces a particular aspect of
language in their own speech. To get started, you might look at the
chapter-end summaries and then skim through each chapter in turn,
extracting key features as you go. Include features from every aspect of
child language: phonology; vocabulary; morphology; grammar; and
pragmatics (Chapter 1). I have taken an easy one for myself, by adding
the child’s first word for you: above the line (comprehension), there are
claims that some words can be understood as early as seven months;
while below the line (production) 12 months is nearer the mark.
This exercise will help you create your own survey of child language.
But be aware: there are limitations. First, marking an average age of
appearance for each feature may create an illusion of specificity. Age
of acquisition estimates, when based on spontaneous child speech, are
always constrained by the particular sample of language they come
from. A child might know and use a particular structure, but simply not
display that knowledge in the particular sample taken by the researcher.
Second, there are considerable individual differences among children,
so a range of ages for each feature would present a more complete (if
messier) picture.
Figure 10.1 Exercise: A child language timeline
 And now, a minor revelation: I’ve had a go at this exercise myself (see
Figure 10.3, p. 282). Don’t peek until you’ve tried it yourself. You will
find, along with me, that you end up with a rough sketch only of the
landmarks in child language. But it does provide a handy, at-a-glance
point of reference. You might see this exercise as a work in progress.
Add to it as you read beyond the confines of this book and as new
discoveries are made. And bear in mind that language acquisition does
not cease at 60 months. Vocabulary, morphology, syntax and
pragmatics all develop during the school years. What we currently
know about development in the school years is probably just a fraction
of what remains to be discovered. The relative paucity of research
effort focused on older children can be attributed, in part, to the long-
held and widespread belief that, for language acquisition, it’s all over,
bar the shouting, by the age of four or five years (Nippold, 2016,
provides a rare exception). As we saw in Chapter 8, some authors even
suggest that there is not much left for the child to acquire, apart from
vocabulary, past the age of three years (McGilvray, 2006). There is,
however, no sound empirical basis for this conclusion. If you want to
extend your chart into the school years, buy yourself a roll of cheap
lining paper (the sort used beneath wallpaper) and make yourself a
child language scroll. Then dress up as Psamtik I of Ancient Egypt
(Chapter 3) and parade around with your scroll, like a language-mad
Pharaoh.
No? Just me then.
The ‘nature’ in nature–nurture: Something must be
innate
Our timeline provides a broad overview of some basic facts of child
language. Incidentally, so too does the table summarizing Tiedemann’s
observations (see Appendix 1). From Tiedemann to the present day,
agreement on the basic facts has been comparatively easy to establish. But
coming to grips with the facts is one thing. Explaining them is quite another.
It will be apparent by now that research on child language is nothing if not
controversial. Pretty much every fact in the canon of child language research
remains open to interpretation. The seeds of controversy were already
sprouting in Ancient times (Marx, 1964). Epicurus (341–271 BCE) provided
the first rumination we have on the origins of language, with his argument
that language is a biological function, like hearing or vision. Epicurus
believed that neither God nor the power to reason could explain our ability to
form words and make ourselves understood. For Epicurus, the answer lay,
instead, with Nature. Of course, a contradictory view soon surfaced. Zeno
(333–262 BCE) argued that the human power to reason could explain the
origins of language. For Zeno, language is the product of learning. The ideas
of Epicurus and Zeno are not framed in terms of the origins of language in
the child. But the questions they considered and the theoretical positions they
adopted, are relevant to child language. In particular, we see an emphasis on
either nature or nurture in explanations of how we come to know what we do
about language. Of course, we know full well that the issue is not that simple.
Both nature and nurture have a role to play in any account of language
acquisition. If we start with nature, all theoretical perspectives must
accommodate the fact that the child brings a genetically determined
mind/brain to the task of language development:
The question is not whether learning presupposes innate structure – of
course it does; that has never been in doubt – but rather what those
innate structures are in particular domains. (Chomsky, 1975: 13)
This observation entails that non-nativist theories of child language depend
on ‘innate structure’ just as surely as nativist theories do. But what is the
character of this innate structure? According to the non-nativist view,
language does not emerge from the application of domain-specific learning
mechanisms, dedicated to processing exclusively linguistic input (Ibbotson &
Tomasello, 2016). Nor, of course, do non-nativists allow for innate linguistic
knowledge. But the non-nativist must subscribe to some kind of innately
specified learning mechanism. As noted in Chapter 9, this mechanism is
typically described as a ‘domain-general’ learning mechanism.
Learning mechanisms
Domain-general learning: How general?
A domain-general learning mechanism can learn with equal facility from
different kinds of input. We encountered one example of domain-general
learning – associative learning – in Chapters 6 and 7. This form of learning
has been studied in numerous domains within psychology for many decades.
We considered its role both in early word learning (Chapter 6) and in the
learning of irregular word forms (Chapter 7). Very simply, associative
learning takes place when the presence of one cue provides a reliable
indication that a second cue is also present. Reliability need not be perfect, a
fact which allows for associative learning to be described in probabilistic, or
statistical, terms (Chapter 5). Whatever the precise nature of the learning
mechanism, the notion of domain-general learning has certainly taken hold
within child language research. In Chapters 2 and 5, we considered the
argument that infants and tamarin monkeys possess a domain-general
capacity to process rapid sequences of discrete acoustic events. These include
both the strings of phonemes found in continuous speech and also sequences
of pure, non-linguistic tones (Saffran et al., 1999). But just how general is the
‘general’ in ‘domain-general’? Apologies – that’s a lot of generals for a book
that comes in peace. In the case of tone- and phoneme-strings, the relevant
stimuli both reside in the auditory domain. Moreover, they share certain
formal characteristics, being: (1) discrete events; (2) organized sequentially;
and (3) presented at a consistently high speed. Initial efforts have been made
to extend research interest across sensory domains, to include vision (Fiser &
Aslin, 2001; Chapter 5). But a truly domain-general learning mechanism
would be adapted to process inputs from any and all sensory modalities in the
course of learning. And the senses we should take into account are numerous.
They are not confined to the band of five, familiar to most people: vision,
hearing, taste, smell and touch. The following are also considered to be
distinct senses, following the differentiation of touch into separate
components: vestibular sense (balance and movement); proprioception (the
relative position of different body parts); pain; and temperature. We also have
a number of internal senses, including, for example, pulmonary stretch
receptors for the control of breathing. It seems highly unlikely that a single
learning mechanism could take as its input information received from all the
human senses. This means that a literal reading of the term ‘domain-general
learning’ appears untenable. Indeed, the term domain-general might have to
be abandoned for something that is, ironically, more specific.
What is far more plausible is that we possess learning mechanisms which can
operate on more than one kind of input, either within, or even across, sensory
modalities. The work of Saffran and colleagues supports this position, but as
they rightly suggest, even with this scaled down version of ‘domain-general’
there are serious empirical challenges:
How does one ‘prove’ that a learning mechanism is domain-general?
Even the clearest cases – where learners show equivalent performance
when acquiring materials from two different domains, given the same
patterns in the input – could equally well represent two parallel learning
mechanisms in lieu of a single domain-general mechanism. (Saffran &
Thiessen, 2007: 77)
Evidence already suggests that different kinds of learning mechanism may
indeed be required for language learning. In the early stages, we witness the
infant segmenting word-like units from a rapid flow of acoustic events. In
this case, the input for learning is the uninterrupted strings of phonemes in
the speech the infant hears. Later on, these words become, in turn, the input
to a different kind of pattern finding – one that involves detecting the
principles that bind words together into sentences – the acquisition of syntax.
Hence, the input that forms the basis for learning is different in each case.
Rapid sequences of phonemes are not the same as individual words, and it is
entirely possible that the ‘pattern-finding’ mechanism, which the infant
applies to each kind of input is also different in each case (Saffran &
Thiessen, 2007). Moreover, the actual kinds of pattern they are finding –
word forms versus syntactic rules – also differ in obvious ways. In any event,
there may well be more than one learning mechanism at work in the
acquisition of language (cf. Lieven, 2016).
The possibility of multiple mechanisms also presents itself in the case of
second language learning late in life. In our discussion of critical periods
(Chapter 3), we raised the possibility that the way young children acquire
their first language differs, qualitatively, from the way adults acquire a
second language. While all typical children attain fluent mastery of their
native tongue, few adults who embark on a second language attain the child’s
level of competence. Hence, there may be different learning mechanisms at
work at different times of life. While we noted that there are vocal critics of
this position (Hakuta et al., 2003), there remains support for the idea that later
language development does not rely on the same learning mechanisms
exploited by the child. In that case, we should observe that the learning
mechanisms brought to bear later in life must be impressive indeed. Adult
powers of working memory and long-term memory, together with
sophisticated metalinguistic skills, are some of the learning tools which
undoubtedly support later second language learning. Although the vast
majority of adults do not attain native speaker levels of competence, many
such learners nevertheless do very well. The learning mechanisms at their
disposal must therefore be relatively powerful. This follows because these
adult learners cope fairly well with the complexities of language learning,
without the advantage of learning mechanisms specially adapted for the
purpose. Of course, these conclusions only follow if one supports the notion
of critical periods.
To conclude, it is apparent that non-nativist theories rely on innate learning
mechanisms to the same extent as nativist theories do. In the non-nativist
case, there remains a lot of work to do in teasing apart the different
possibilities and working out how the innate learning machinery, applied by
the child in the course of language acquisition, actually works. The concept
of domain-general learning undoubtedly needs refinement. It is unlikely that
any learning mechanism is truly domain-general, even though some may well
apply to more than one form of input, and even to inputs from different
sensory modalities. In addition, it is probable that more than one learning
mechanism is exploited in the course of language acquisition. It will be
fascinating to see how our understanding of domain-general learning
develops in the years ahead.
Linguistic nativism: The need for learning
mechanisms
Innate mechanisms of learning aside, we should briefly review the concept of
innate ideas, advanced by Plato and revived by Chomsky. We discovered in
Chapter 8 that this concept is surprisingly ethereal. Despite some popular
conceptions, no-one seems to suggest that we are born with a knowledge of
grammar in our heads. Linguistic nativists do suggest that we possess a
grammar-acquiring capacity, referred to as Universal Grammar (Jackendoff,
2002; Chapter 8). Moreover, this capacity is domain-specific. It can only
acquire human grammar. Hence, our knowledge of grammar is acquired
because, when presented with linguistic input, that is all our minds can
acquire. It seems, therefore, that, even in the nativist approach, we have not
quite put learning mechanisms aside. After all, the concept of a ‘grammar
acquiring capacity’ sounds much more like a learning mechanism than a set
of innate ideas. This conclusion might not find ready agreement from
linguistic nativists. You will recall from Chapters 1 and 8 that the role of
learning is downplayed to the point where the concept of learning is actually
discarded (Box 1.5). But nativists still need to explain how the categories of
Universal Grammar are realized through experience with a particular
language. A notable attempt to accommodate learning in a nativist framework
is provided by Yang (2004). His reliance on probabilistic learning (Box 5.2)
allows for the possibility of domain-general learning without sacrificing a
nativist perspective. The point to be made here is that language acquisition
mechanisms are required in a nativist approach, and these may or may not be
domain-general. This holds true regardless of one’s beliefs about the genetic
underpinnings of grammatical knowledge (Fodor, 2001).
Methodology: Limitations and possibilities
With regard to methodology, we can start by confirming that some of the
oldest tools in the box continue to be invaluable. For example, observational
data, in the form of video and audio recordings of adult–child interactions,
have always been a mainstay of child language research and show no signs of
losing their influence as a vital source of information (e.g., Behrens, 2008;
see also Box 1.3). Another technique that still earns its keep is the elicited
production method, where particular language forms are coaxed out of
children (see the studies on structure dependence described in Chapter 8:
Crain & Nakayama, 1987; Ambridge, Rowland & Pine, 2008). Add to these
both novel word studies (Box 4.2) and the use of grammaticality judgements
(Chapter 3) and we get a flavour of the wide range of methods, both old and
new, that continue to illuminate the field.
Over the past few decades, there have also been significant steps forward in
methodology (Hoff, 2012). For example, in the 1970s, the habituation
method first allowed researchers to tap into the way infants perceive human
speech. As described in Box 5.1, in this method, a stimulus is presented
repeatedly to infants while their attention is gauged in some way (for
example, rate of sucking on a dummy). When interest in this initial stimulus
wanes (sucking rate declines), a new stimulus is introduced. If sucking rate
then revives, it is taken as a sign that the infant has detected that the original
stimulus and the new one are different. A range of infant behaviours, not just
rate of sucking, are recruited as measures of infant attention, including heart
rate and head turning. The habituation method has revealed the power of two-
month-old infants to perceive phonemic contrasts categorically, for example,
the /p/ – /b/ distinction (Chapter 5). An extension of this method also
permitted study of the segmentation problem in the 1990s: the way in which
the infant detects the boundaries between individual words in continuous
speech (Chapter 5). In this case, the measure taken was the so-called head-
turn preference. Given a choice of linguistic stimuli, the infant displays a
preference by turning its head towards one of them. In common with other
successful methods, a behaviour which is already in the child’s repertoire
(e.g., head turning or sucking rate) is exploited as a measure of infant
attention to language. Incidentally, our friend Tiedemann (1787) observed the
reliability of head turning as a measure of infant attention, when his son was
four months old: ‘the boy, when hearing a sound, always turned his face in
the direction whence it came; so he had already learned to tell what he heard
through the right ear and what through the left’ (Murchison & Langer, 1927:
217). Of course, it took a while to exploit the potential of head turning as a
viable research method. But we got there in the end.
Another recent method, also based on infant perceptual preferences, is the so-
called intermodal preferential looking paradigm (mentioned in Chapters 1
and 9). The young child sits on their mother’s lap in front of two screens,
placed side by side. After a familiarization period, two video events are
played at the same time, together with a linguistic stimulus that matches just
one of the videos. The ‘intermodal’ in the title, therefore, refers to the
conjunction of two perceptual modes: vision (the video) and sound (the
language heard by the infant). The behavioural measure taken is the amount
of time the child spends looking at the events on one screen compared with
the other. As we saw, inferences can then be made about the child’s
understanding of language, including verb meanings and grammatical
relations. Both preferential listening in infants and preferential looking in
young children are ingenious methods that have enriched the repertoire of
methods in child language research.
Advances are also being made in brain research, an area touched on in
Chapter 3, in our review of critical periods. The ways in which the child’s
brain responds to linguistic stimuli can be monitored in both time and space.
With respect to time, we can chart the course of brain activity as it unfolds
over time, using the method of Event Related Potentials (ERP) (e.g., Kaduk,
Bakker, Juvrud, Gredebäck, Westermann, Lunn & Reid, 2016). And with
regard to space, we can locate which regions of the brain experience peak
activity, following stimulus presentation, using a number of methods,
including: (1) functional Magnetic Resonance Imaging (fMRI) (Dehaene-
Lambertz, Hertz-Pannier, Dubois, Meriaux, Roche, Sigman & Dehaene,
2006; Dubois, Hertz-Pannier, Cachia, Mangin, Le Bihan & Dehaene-
Lambertz, 2009; Xiao, Brauer, Lauckner, Zhai, Jia, Margulies & Friederici,
2016); and (2) near-infrared spectroscopy (Minagawa-Kawai, Mori, Hebden
& Dupoux, 2008). In both cases, measures of brain activity are based on
oxygen levels in the blood: active brain regions are characterized by
relatively high levels of oxygen use. The scope of this book allows only the
briefest mention of neuroscientific methods, but see the Further Reading
section if you want to delve further.
Figure 10.2 An ERP study with a six-month-old child

Source: Benasich, A.A., Choudhury, N., Friedman, J.T., Realpe-Bonilla,

T., Chojnowska, C. & Gou, Z.K. (2006). The infant as prelinguistic
model for language learning impairments: Predicting from event-related
potentials to behavior. Neuropsychologia, 44(3), 396–411, Figure 1, p.
399: Photograph of a 6-month-old child seated on his mother’s lap
during an ERP testing session using a dense array Geodesic Sensor Net
system (Electric Geodesic, Inc., Eugene Oregon, USA).
There is space to make two points about brain research here. First, the ERP
method has considerable promise for use in child language research. It is a
non-invasive technique that monitors the time course of brain activity via
numerous electrodes placed on the scalp (Männel, 2008). As you can see
from Figure 10.2, the child looks like some crazy kind of multiple socket,
with all the plugs pushed in, and yet, somehow, seems entirely oblivious
(maybe it’s an outtake from The Matrix). ERP research has been valuable in
helping confirm findings from other sources, including research on early
speech perception (Kuhl & Rivera-Gaxiola, 2008; Chapter 5). ERP has also
been used in studies on children who experience significant delays in
language acquisition. In one such study, 12-month-olds from families with a
history of language problems were monitored, using ERP, until the age of 36
months (Benasich, Choudhury, Friedman, Realpe-Bonilla, Chojnowska &
Gou, 2006). Benasich et al. were interested in the infants’ ability to process
rapid sequences of auditory tones (as found in fluent speech). Infants differ in
this respect and these differences can be detected via ERP. Impairment in
rapid auditory processing at 12 months is associated with depressed
performance on measures of spoken language at 36 months. Put another way,
infants who are proficient at processing rapid sound sequences acquire
language more quickly and more successfully than less proficient children.
We see how measures of brain activity, early in life, can function as
predictors of future language learning.
A second point we can make about brain research concerns the use of fMRI
scanners to locate regions of peak brain activity. You may know that fMRI
scanners require the participant to lie very still in a solid, enclosed tube for an
extended period of time. You will not be surprised to learn, therefore, that
this method is not yet in use with toddlers. Infants, just a few months old,
have been placed in scanners, but they have to be asleep at the time (without
any form of sedation). Linguistic stimuli are played to the sleeping infant and
brain responses can be recorded. Research is in the very early stages in this
field (see Further Reading) and has focused on the issue of brain asymmetry,
a potential sign of specialization for language: in most people, language is
processed in the left brain hemisphere (mentioned in Chapter 3). Observe
that, currently, we can only scan infants when they are asleep. Hence, we
have no control comparison with the infant brain in the waking state which is,
after all, what we are most interested in. While fMRI research on infants is
now underway, it is clear that serious empirical challenges remain to be
tackled.
Advances in neuroscience, based on the fMRI method, have been
phenomenal over the last 20 years or so (e.g., Aslin, Shukla & Emberson,
2015). But the squirming toddler, reluctant (or rather, unable) to lie still in a
scanner, presents a significant challenge. Child language researchers cannot
always take immediate advantage of advances in technology, a point which
brings us back to where we started with respect to methodology: it ain’t easy.
Methodological advances have been made, and will continue to be made. But
it is worth reflecting that much of the theoretical disagreements witnessed
between child language researchers find their origin in the limitations of their
methods. To give just one example, we have seen how the preferential
looking method has been used to investigate understanding of transitive
sentences in 24-month-olds (Chapter 9). Do infants really understand the role
of agent of a causative event? Moreover, do they expect the agent to come
first in a simple transitive sentence like Big Bird is squatting Cookie
Monster? As we discovered, for one group of researchers, the answer is yes
(Gertner et al., 2006). Others are much more cautious (Dittmar et al., 2008a).
These latter, more sceptical researchers, do not want to attribute an extensive
degree of syntactic knowledge to very young children on the basis of their
looking behaviour. But as Valian (2014: 80) points out:
Since all experiments impose cognitive and executive function demands
on children (e.g. attending to the experimenter, following directions,
understanding the task, keeping on task, inhibiting extraneous stimuli),
we run the risk of underestimating or mischaracterizing children’s
syntactic knowledge.
Language acquisition: The state of the art
I first encountered child language at the age of 20, when I went to study
psychology at Edinburgh University. As luck would have it, it turned out that
I was expected to study more than just straight psychology, so I plumped for
linguistics as well (and Japanese and social anthropology, too, as it happens).
Hitherto, linguistics was unknown to me as an academic discipline, but it
captured my imagination, especially, perhaps, because I studied it alongside
psychology. Psycholinguistics combines the heady complexities of linguistics
with the even greater complexities of the human mind. Fundamentally,
psycholinguistics poses two related questions: how is language acquired and
how is language represented in the mind? These are big questions. As
someone new to academic life, child language presented itself as an
especially fascinating topic within psycholinguistics. My guiding light at the
time was a newly published collection, edited by Eric Wanner and Leila
Gleitman, entitled Language acquisition: The state of the art. The papers in
Wanner & Gleitman (1982) collectively asserted the right of child language
to be considered as the focus of serious scientific enquiry. The field had
already come far since the early days of Jean Berko Gleason’s wugs and
Roger Brown’s longitudinal studies of Adam, Eve and Sarah (Berko, 1958;
Brown & Bellugi, 1964; see also Chapter 1). By the 1980s, a wealth of
imaginative theorising, constrained by a scientific approach, had created a
substantial, independent field of enquiry.
But where are we now, more than one third of a century down the line from
Wanner & Gleitman (1982)? I thought it would be worthwhile to elicit the
views of leaders in the field who are active in research today. This involved
the email equivalent of cold-calling a large number of individuals, to canvas
their views. My initial cohort was assembled from the editorial boards of
three child language journals (Journal of Child Language, First Language
and Language Acquisition), plus a number of other likely candidates known
to me from their work and/or personal contact. Gratifyingly, a significant
number on my initial list either failed to dump me into their Junk folders or
were too slow in hitting the Block button. In fact, something like 40
prominent researchers were generous enough to give their attention to my
questions and provide their perspectives on the field. I asked respondents to
consider the achievements of the recent past and then cast their minds into the
future and predict the next big breakthroughs, in theory, methodology and
empirical discovery. The answers were various and fascinating. In what
follows, I have identified authors who associated themselves with particular
themes (full names can be found in the Acknowledgements at the beginning
of the book). Of necessity, I am painting with a broad brush. You should
therefore follow up on any particular interests through the published work of
authors who catch your eye.
In theoretical terms, the last ten years or so have seen the usage-based theory
establish itself as a viable alternative to Chomsky’s theory of Universal
Grammar (Chapter 8). Numerous alternatives to UG theory have been
advanced over the years, but usage-based theory has undoubtedly made the
largest impact, both theoretically and empirically (Chapter 9; Tomasello,
2003). Several commentators pointed this out and observed a concomitant
expansion of interest in domain-general cognition, as noted above (Berman,
Bybee, Guasti, Guerts, Lieven, Pearl, Ravid, Saffran, Shirai, Tomasello,
Ullman). On this view, memory (declarative and procedural), theory of mind
and even musical cognition (think of intonation) all contribute to language
acquisition. The issue of how children process input is relevant here, in
particular, the influence of input frequency (Chiat, Clark, Phillips, Rowland).
In the pursuit of these issues, growing connections to allied disciplines are
becoming increasingly important, in particular, experimental psychology and
neuroscience (Guasti, Li, Ota, Saffran, Ullman). The rise of usage-based
theory has also given rise to the hope that research in future will be less
polarised between usage-based and UG-based approaches (Chiat, Marshall,
Morgan).
With regard to methodology, two broad themes emerged: (1) the use of
modelling; and (2) the rise of methods which tap into online processing. A
number of respondents emphasized the importance of computational
modelling as a means of examining the interactions between input and
learning (Chapter 7 and Becker, Chiat, Clark, Kidd, Li, Lieven, Ota, Pearl,
Phillips, van Hout). The value of mathematical models was also raised
(Yang). In both cases, the successes of the past 30 years have paved the way
for yet more sophisticated use of modelling in the future. The success of
models depends in large part on the robustness of the information fed into
them. In this respect, a number of authors point to advances in the use of
large and/or dense input databases (Berman, Becker, Christophe, Lieven,
Pearl). The task of the model is to acquire syntactic categories from the same
kinds of input which are available to the child. Hitherto, no model has
managed to generate the full complexities of syntax from scratch
(Soderstrom, 2014). While this may be possible in the future, one is left with
the tantalising possibility that some language-specific, innate machinery will
be required to get the learning started. Did anyone say UG? Try as some non-
nativists might, they cannot quite shake off the possibility that some form of
language-related, innate constraints are needed (cf., Pérez-Leroux &
Kahnemuyipour, 2014: e115).
The second aspect of methodology which captures the attention of the child
language community is the recent advent of techniques for investigating
cognitive processing as it happens in real time (Conti-Ramsden, Kidd,
Naigles, Rowland, Saffran). Online processing methods include those
mentioned above, such as EEG and MRI, in addition to eye-tracking
technology, which provides a sophisticated indication of child attention on a
moment-by-moment basis (Oakes, 2012; Odean, Nazareth & Pruden, 2015).
These methods allow for the integration of what we know about language
processing with what we know about language development, an endeavour
which is in its infancy (Chiat, Clark, Pearl). More pointedly, they render the
issue of how children process linguistic input tractable for the first time. The
difficulties of working with very young children notwithstanding, these
methods promise much for the future.
At least one (anonymous) contributor voiced disquiet about the empirical
clout of child language studies, citing tiny sample sizes, lax statistics and a
pervasive failure to replicate key findings. Statistical analyses have
undoubtedly become more refined (Ravid), but there is always room for
improvement. And the failure to replicate is not confined to child language
but is, arguably, the scourge of psychology more widely. What can one do?
Thankfully, it is not all doom and gloom. Projects like Stanford University’s
Metalab tackle this problem, to some extent, by combining the data from
numerous related studies and subjecting them to sophisticated statistical
techniques (see Websites below). Arguably, meta-analyses should become a
standard tool in future child language research (Marshall, Ota). In addition,
new tools for automated analyses of large databases are emerging, which
complement those already available via CHILDES (Chapter 1). These include
LENA (Language Environment Analysis), a set of tools for analysing the
language input of young children (e.g., Ko, Seidl, Cristia, Reimchen &
Soderstrom, 2016).
In recent years, the empirical storehouse of data on child language has been
filling up fast. We now have a more extensive knowledge of the milestones in
both initial and more complex syntax (Becker, Clark), semantics (van Hout)
and pragmatics (Clark, Naigles) (Chapters 4, 6, 8 and 9). Efforts to tie it all
together have also moved forward, in order to understand how syntax,
semantics and pragmatics influence each other in development (Lieven,
Tomasello). Again, though, there is a lot more to discover on this complex
topic.
A final common thread which ran through the responses of many researchers
in this ad hoc survey was the problem of English (Berman, Becker, Clark,
Golinkoff, Guasti, Kidd, Li, Lieven, Pearl, Ravid, van Hout). The English
language has been the elephant in the room for decades now, in the sense that
the vast majority of research has been conducted on children acquiring
English. Yet we live in a multilingual world. Different languages may well
present the child with particular challenges, which might easily be neglected
if the focus is on just one language. We saw this in Chapter 7, where the
notion of regularity, as witnessed in English morphology, looked less
universal as soon as one took a short boat trip across the Baltic to nearby
Poland. In fact, English can look like a rare beast when looked at from the
perspective of certain linguistic features, such as tones. Some 90 per cent of
the world’s languages – including Mandarin, Cantonese and Thai – are tonal
(Golinkoff). This means that the meaning of a word depends in part on its
associated intonation contour (e.g., for Mandarin: rise, fall, fall-rise or high
level). In Mandarin, the sequence ma can mean mother, hemp, horse or scold,
depending on the tone deployed. Imagine saying My mother scolded the
horse for eating the hemp in Mandarin. Of course, English does not use tones
– just one example where our view of the learning problem facing the young
child might be impoverished by an exclusive focus on English. There is also
the fact that many (probably most) children are bilingual or multilingual. Yet
a large proportion of early child language research was conducted on
monolingual English speakers. A standard argument in UG theory was that
one could study any human language in the pursuit of those properties of
syntax which are universal (e.g., Chomsky, 1965: 36). This follows from the
assertion that all languages express the universals embodied in UG. From this
it also follows that one can comfortably zoom in on one’s own native
language, which just happened to be English in the case of many North
American (and British) researchers. And in the early days of creating a
scientific discipline out of child language studies – the 1960s – the field was
dominated by native English speakers. Undoubtedly things have changed.
Databases like CHILDES now contain numerous sets of non-English data
(Berman, Pearl) and research on bilingual (and multilingual) acquisition has
flourished (Behrens, Clark, Guasti, Howe). At the same time, there is still
great potential for expanding the horizons of research beyond English.
Finally, several researchers expressed individual hopes and beliefs about the
topics which they consider should attract more research attention in the
future. These include: language variability, including delay and disorder
(Behrens, Naigles); multi-modal learning, as witnessed in speech-gesture
interactions (Guasti, Marshall, Naigles and Chapter 4); the need for more
longitudinal research (Conti-Ramsden, Chiat) and more individual case
studies (Clark); a greater focus on input and interaction (Berman, Behrens,
Clark); less focus on input frequency (MacWhinney); and greater attention to
practical applications of child language research (Chiat, Golinkoff). This
latter point is of particular importance. There are many lessons to be learned
from the study of typical language development which can be applied in
cases of atypical development, for example, in the development of
therapeutic interventions for children with delays and disorders of various
kinds (e.g., Paul & Norbury, 2012).
Child language: Acquisition and development
Way back in Chapter 1 (Box 1.5), I suggested that the terms acquisition and
development are somewhat loaded terms. They bear subtle connotations
which can reveal which side of the nativist fence a researcher stands on.
While acquisition is favoured by nativists, development is more the preserve
of non-nativists. I would be the first to admit that this distinction is rather
subtle and by no means a foolproof index of an author’s theoretical
persuasion. At the same time, preferred usage of one term over the other is
prevalent and quite often reveals one’s bias. The concept of acquisition, very
broadly, suggests that possession is taken of a finished product.
Development, meanwhile, sits more comfortably with the non-nativist belief
that change takes place over time, through learning. But neither acquisition
nor development is so restrictive in scope (check the dictionary to confirm
this point). These terms can be used interchangeably without violence, either
to their inherent meanings or to the user’s theoretical stance. Since the first
edition of this book appeared there are signs that the child language
community is becoming less sectarian in its use of acquisition versus
development (e.g., Rowland, 2014). This can only be a good thing. Both
acquisition and development can help describe both nature and nurture at
work in the child. Child language is, undoubtedly, a complex topic. I can only
urge you to revel in its complexities. There is always something to excite the
imagination, always the inspiration of finding oneself at the point where
philosophy meets science. Always the simple wonder in observing the
acquisition and development of language in the child.
Figure 10.3 A child language timeline
In a Nutshell
The earliest systematic study of child language was reported
by Tiedemann in 1787. Many of his observations remain
relevant for current research.
You can get a good overview of some basic facts in the field
by creating your own timeline.
Both nativist and non-nativist theories depend on innate
learning mechanisms to help explain how the child acquires
language:
Non-nativist theories rely on the idea of a domain-general
learning mechanism, one that is not specifically adapted for
language. But no learning mechanism can, plausibly, be
entirely domain-general, and several learning mechanisms
are probably required to explain language acquisition.
Nativist theories also require some kind of learning
mechanisms, irrespective of the view that the child is
innately endowed with Universal Grammar. These learning
mechanisms may or may not be domain-general.
Child language studies make use of a wide range of different
methods and much progress has been made in the past 30
years or so. At the same time, the difficulties inherent in
studying both infants and young children mean that the
interpretation of child language behaviours often remains
 open. In consequence, the study of child language continues
to be both controversial and richly stimulating.
Further Reading
Ambridge, B. & Rowland, C.F. (2013). Experimental methods in
studying child language acquisition. WIREs Cognitive Science.
doi: 10.1002/wcs.1215.
This article provides a thorough review of methods in child
language research, including the latest methods for exploring the
workings of the infant brain. There is also a very useful checklist
at the end for researchers planning experimental work with
children. Use it when planning your own research project.
Dehaene-Lambertz, G. & Spelke, E.S. (2015). The infancy of the
human brain. Neuron, 88(1), 93–109.
A taste of the future. This paper provides a survey of recent
research on the infant brain, including discussion of research on
language development.
Mueller, K.L., Murray, J.C., Michaelson, J.J., Christiansen, M.H.,
Reilly, S. & Tomblin, J.B. (2016). Common genetic variants in
FOXP2 are not associated with individual differences in language
development. PLoS One, 11(4), pp.e0152576.
Discovery of the gene known as FOXP2 was presented in some
media reports as discovery of ‘the gene for language’. But as this
paper reveals, the relationship between FOXP2 and language is
much more complex (see also Wohlgemuth, Adam & Scharff,
2014, for a songbird twist).
Newbury, D.F., Bishop, D.V.M. & Monaco, A.P. (2005). Genetic
influences on language impairment and phonological short-term
memory. Trends in Cognitive Sciences, 9(11), 528–534.
A summary of research on genetic causes of language impairment
in children. No specific genes are identified in this kind of
research. Instead, different aspects of language behaviour are
identified as having a partly genetic basis.
Lidz, J. & Gagliardi, A. (2015). How nature meets nurture:
Statistical learning and Universal Grammar. Annual Review of
Linguistics, 11(1), 333–353.
Where does research on Universal Grammar go next? This paper
attempts to point the way. While arguing that innate language
learning mechanisms are necessary, these authors also concede
the need for more general cognitive learning mechanisms.
Websites
Metalab at Stanford University:
http://metalab.stanford.edu/
A research tool for enhancing the statistical power of child
language studies. Meta-analyses aggregate the data across
numerous related studies to examine how strong the reported
effects are. In addition, researchers can plan more rigorously
how many participants they will need to produce findings
with any real statistical authority. Metalab points the way to
a statistically much more sophisticated future for child
language research.
W.C. Fields in Pool Sharks (1915)
Search for this in YouTube and enjoy one of the comedy
greats from more than 100 years ago. W.C. Fields was the
man who first advised: ‘never work with children or
animals’. Child language researchers do both, of course. In
any event, W.C. Fields broke his own rule on several
occasions, including in the 1933 version of Alice in
Wonderland (a poor quality version is available on
YouTube). About 45 minutes into the movie, W.C. Fields
appears as a querulous Humpty Dumpty, espousing some of
Lewis Carroll’s observations on word meaning.
Fred Astaire and Ginger Rogers in Let’s Call the Whole
Thing Off (1937)
Search YouTube for Fred Astaire and Ginger Rogers in
Let’s Call the Whole Thing Off. You’ll find them tap dancing
on roller skates to an old song by George and Ira Gershwin
in the movie Shall We Dance (see Gershwin, 1959). If you
know this song at all, you probably know the ‘punch line’:
Let’s call the whole thing off. But actually, the real point of
the song is expressed in this part of the lyric:
For we know we
Need each other, so we
Better call the calling off off.
When it comes to nature and nurture, this part of the lyric is
 much more appropriate. Often referred to misleadingly as the
nature–nurture controversy, we all know that, in fact, nature
and nurture were meant for each other.
Still want more? For links to online resources relevant to this
chapter and a quiz to test your understanding, visit the companion
website at https://study.sagepub.com/saxton2e
Answers to Exercises
Chapter 1: Prelude: Landmarks in the Landscape
of Child Language
Exercise 1.2: Grammatical categories

A couple of points to note:

1. Divorce can be both a noun (her divorce came through) and a verb
(Hillary should’ve divorced Bill). Complete can be both a verb (I
completed the puzzle) and an adjective (This is a complete mess). Many
words can function in more than one grammatical category.
2. Even if you didn’t know that pusillanimous means weak-willed or
cowardly, you can probably work out that it is an adjective by testing
out which sentence slots it fits into most naturally:
*He is a pusillanimous (noun)
*He pusillanimoused my car yesterday (verb)
He is a pusillanimous man (adjective)
Chapter 3: The Critical Period Hypothesis: Now or
Never?
Exercise 3.1: Subject, verb and object
1. The subject of the sentence is marked in bold:
1. Gordon’s hair looks like nylon.
2. I couldn’t help noticing the baby elephant in your fridge.
3. The man who lives next door to me has a fondness for figs.
4. The end of the world is nigh.(e) Have you ever seen a Buster
Keaton movie?
2. The verb is marked in bold:
1. Nigella licked her spoon in a rather suggestive fashion.
2. Now is the winter of our discontent.
3. Henrietta scoffed a big bag of toffees on the bus.
4. Fly me to the moon.
5. Play it again, Sam. (I know, I know: Ingrid Bergman doesn’t
actually say this).
3. The object is marked in bold:
1. That really takes the biscuit.
2. Frankly, my dear, I don’t give a damn.
3. I simply adore champagne.
4. How many cream buns can you eat at one sitting? (This is a
question, which is why the standard S-V-O order has been switched
round).
5. I’ll make him an offer he can’t refuse.
4. Subject, verb and object in Genie’s sentences
The sentences below, all uttered by Genie, appear in the text of Chapter
3. Check your analyses against my own, below, to see if Genie gets the
correct basic English SVO order. Of course, this is only possible in
cases where she produces at least two of the three elements, S, V and O.
Missing elements have been marked with Ø. Some sentences have
material additional to the basic S, V and O (at hospital in like good
Harry at hospital; no more in no more have).
Only one element here, an odd noun phrase that could appear as either
subject or object, but here it is produced by itself.
Dentist say drink water is complicated. It seems to have a main clause
(dentist say) and a sub-clause (drink water), as in The dentist says ‘drink
water’.
Also complex, but much more difficult to interpret.†††† This looks like
two separate noun phrases, but my analysis of one as subject and the
other as object could well be fanciful.
Chapter 5: Language in the First Year: Breaking
the Sound Barrier
Exercise 5.1: Phonological contrasts
The key to this exercise is to find word pairs which differ by one sound only,
and then find the phonemic symbol for each contrasting sound from the
Pronunciation Guide at the back of the book.
Chapter 7: The Acquisition of Morphology:
Linguistic Lego
Exercise 7.1: Derived forms
The following symbols are used for grammatical categories: noun (N); verb
(V); adjective (A). Answers are given in the form: root → derived form.
obese → obesity N → N
holy → unholy A → A
repel → repellent V → A
London → Londoner N → N
select → selection V → N
swim → swimmer V → N
cure → curable V → A
black → blacken A → V
real → realize A → V
success → successful N → A
organize → reorganize V → V
clear → clearance V → N
play → playful V → A
special → speciality A → N
member → membership N → N
accident → accidental N → A
Exercise 7.2: Morphological processes
airheads compound (air + head) and inflection (plural -s)
disgraceful derivation (dis- + grace + -ful)
hat maker derivation (make + -er) and compound (hat + maker)
monster simple root form
sampling inflection (sample + -ing)
pleased inflection (please + -ed)
indefatigable derivation (in- + de- + fatigue + -able)
washing machine inflection (wash + -ing) and compounding (washing
+ machine)
insider derivation (in- + side + -er)
coffee table compound (coffee + table)
reasserted derivation (re- + assert) and inflection (reassert + -ed)
dressmakers derivation (make + -er) and compound (dress + maker)
 and inflection (dressmakers)
elephant simple root form
recycles derivation (re- + cycle) and inflection (recycle + -s)
sweetcorn compound (sweet + corn)
unhappily derivation (un- + happy + -ily)
Exercise 7.3: Productivity of derivational suffixes
I provide a few examples below, with the following category abbreviations:
Noun (N); verb (V); and Adjective (A). Observe again how the categories of
root and derived forms can differ. You will probably have found it especially
easy to come up with examples for -able and -er, a sign (however informal)
that they are highly productive affixes in English. Notice how you sometimes
have to clear away more than one affix to get at the root form.
Affix Example Category of
  Derived Form Root Form
pre- predetermined A determine (V)
 premeditated A meditate (V)
in- inedible A eat (V)
 inoperable A operate (V)
under- undernourished A nourish (V)
 underpaid A pay (V)
de- depopulate V populate (V)
 deform V form (V)
-cy frequency N frequent (A)
 delinquency N delinquent (A)
-able passable N pass (V)
 enjoyable A enjoy (V)
-ity density N dense (A)
 acidity N acid (A)
-less speechless A speech (N)
 witless N wit (N)
-er farmer N farm (N)
 mover N move (V)
Chapter 8: Linguistic Nativism: To the Grammar
Born
Exercise 8.1: BE as copula or auxiliary
1. Louise is a cat lover.copula
2. Ian must be mad to buy so much cat food.copula
3. Patricia and Colin are having kittens about the fleas in their
house.auxiliary
4. Saskia is wondering what all the interest in cats is about.
 1st is = auxiliary; 2nd is = copula
5. I am wondering what cat would taste like if it were roasted.
 am = auxiliary; were = copula
6. Cats were worshipped in Ancient Egypt.copula
7. The cat that is stealing my milk has been known to steal chicken, too.
 is = auxiliary; been = auxiliary
8. Is there any reason why cat owners do not look like their pets?copula
Chapter 9: The Usage-Based Approach: Making it
Up as You Go Along
Exercise 9.1: Verb transitivity
Most verbs in English are bitransitive, because they can appear with or
without a direct object (Huddleston, 1984).
transitiveintransitivebitransitive
want sleep leave
feel disappear wait
take  jump
  read
This exercise is not quite as cut and dried as it might seem because people
can vary in their use of particular verbs according to the dialect or level of
formality being used. For example, I know someone (a native English
speaker) who uses want in intransitive sentences:
Me: Do you want a cup of tea?
Him: I don’t want.
Appendix 1: Observations on Language
Acquisition Made by Dietrich Tiedemann
(1787)
Page references are from the translation by Murchison & Langer (1927).
Appendix 2: Pronunciation Guide: English
Phonemes
The chart below provides a pronunciation guide for the phonemes of English
(for more on phonemes, see Box 1.2, Chapter 1). A special set of symbols is
used to represent phonemes, written within slash brackets / /. While the
symbols for phonemes sometimes correspond to English letters, at other
times they do not. You should also bear in mind that this chart holds true only
for one variety of English (known as received pronunciation). There is some
variation for other dialects, especially among the vowel sounds.
Glossary of Linguistic Terms
accusative
See case.
adjective
A syntactic category that often denotes states (happy, dull, quick,
disgusting). Adjectives are used to describe or qualify a noun (a happy
banker, a disgusting mess). They often have comparative (-er) and
superlative (-est) forms (happier, dullest), though there are exceptions
(*disgustinger). Adjectives also often have adverb counterparts ending
in -ly (dully, quickly) and can be turned into nouns with -ness
(happiness, quickness). Again, there are exceptions (*disgustingness).
adverb
A syntactic category typically used to denote the manner in which
something occurs. English adverbs often end in -ly (sweetly,
wonderfully, badly, stupendously), as in: Albert is stupendously well
informed or She sings wonderfully for a hippo.
affix
A morpheme that is added to a word either on the left hand side
(prefix) or on the right hand side (suffix), when reading. In resealing,
re- is a prefix, while -ing is a suffix, added to left and right of seal.
affricate
A sequence of two consonants, stop + fricative, that often behave as a
single phoneme in languages (including English). In English there are
two affricates: /t∫/ (the first sound in cheap), /ʤ/ (first in jeep).
agent
Typically, a person who intentionally causes something to happen (Paul
in Paul divorced Heather; Ziggy in Ziggy played guitar). The agent is
one kind of thematic role, describing the relationship between a noun
phrase and a verb, in this case, the ‘doer of the action’.
article
In English, a and the are articles, words that combine with a noun to
form a noun phrase (a reptile, the cartoon). Articles constitute a sub-
class of determiners.
auxiliary verb
A verb that supports the main verb of a sentence, hence the traditional
description as a ‘helping’ verb. In English, auxiliaries include so-called
 modals: can, could, shall, should, will, may, might, must (Sara will enjoy
the Red Arrows air show). Modal auxiliaries express modality, that is,
possibility, necessity or futurity. Other auxiliaries include do, have and
be (The au pair has left again). It is possible to have more than one
auxiliary in a sentence: Daniel might have eaten another ice-cream.
Further, auxiliaries, but not main verbs, can be inverted when forming
questions:
Chris can have a third helping → Can Chris have a third helping?
bitransitive
A bitransitive verb can appear in either transitive or intransitive sentence
frames (see also transitivity). For example, remember can be transitive,
as in Lawrence finally remembered my new locker key, or intransitive, as
in Lawrence finally remembered.
case
The case of a noun or pronoun marks its grammatical function within a
sentence (e.g., subject, object or possessor). Typically, the form of a
noun or pronoun will change according to its case. English has a simple
case system with special forms mostly confined to pronouns, in order to
mark nominative (subject), accusative (object) or genitive (possessor)
case.

English has no dative case (for indirect objects), though it is familiar to

generations of school children tortured by Latin declensions. Other
languages, like Latin, Russian or German, have much more elaborate case
systems, marking many more aspects of meaning. Many languages also apply
case marking beyond pronouns and nouns, to include adjectives and articles
also.
Child Directed Speech (CDS)
The special mode of speech, or register, adopted by adults and older
children when speaking with young children. Parents, in particular,
simplify and adapt their speech to meet the needs of very young children
in a wide variety of ways. Phonology, vocabulary and grammar, and
features of interaction, are all simplified and clarified in various ways.
clause
Sentences can comprise a single clause or several clauses, where each
 clause is a kind of ‘mini-sentence’. Each clause contains a subject and a
verb. Simple sentences comprise a single clause, as in Gary likes
chocolate cake (where Gary is the subject and like is the verb). You can
identify how many clauses there are in a sentence by counting how
many main verbs there are. For example, there are two clauses in: Gary
likes chocolate cake, but he hates anything with cream. Another
example of a multi-clause sentence, with main verbs highlighted, is: The
house that Jack built last year has fallen down already. In this example,
one clause is embedded within another: The house [that Jack built last
year] has fallen down. See also relative clause.
common versus proper nouns
Common nouns refer to a class of entities. The common noun chair does
not pick out a particular piece of furniture, but can be used for any and
all chairs (a class of entities). Proper nouns are used for unique entities,
like London or Madonna or Batman. See also noun.
competence versus performance
The knowledge we have of language (our competence) can be obscured
by performance factors, including hesitations, slips of the tongue and
faulty retrieval from memory (Chomsky, 1965). Hence, the errors we
make in the act of producing language may conceal the fact that our
underlying knowledge is perfect.
complement
A complement is an expression that combines with a head word to form
a syntactic phrase. In the noun phrase, the man, the head of the phrase
is the noun, man, and the complement is provided by the article, the. In
the very tall man, the complement is the very tall. A sentential
complement, meanwhile, is like a mini-sentence or clause embedded
within another main clause. Consider Shiho didn’t expect that her
mother would marry a woman. The sentential complement is: that her
mother would marry a woman.
consonants and vowels
Consonants are produced with some restriction of the airflow through
the mouth. There might be a complete blockage of the air, as with /m, p,
b/, where the lips are closed at the start (see stop). Or there might be a
partial restriction, as with /f, v, s, z/, where the articulators are so close
that friction can be heard (see fricative). Vowels, meanwhile, are
produced without any restriction of the air flow through the vocal tract
 (see also Appendix 2, Pronunciation Guide).
copula
A verb that connects a subject with the rest of the sentence, in the
absence of a main verb. In English the verb BE can be used as a copula,
in its various forms (e.g., am, is, are, were). The copula can be thought
of as the equals sign in the formula X = Y, as in: Jill is fond of red wine
or Jill and Wendy are sisters.
count versus mass nouns
Some languages, including English, distinguish these two kinds of noun.
Count nouns are used for objects or entities that can be individuated and
counted, like apples, ideas or books. Mass nouns are used for substances
that cannot be so readily perceived as anything but an unbounded mass,
like wine, or sand, or tarmac. The difference is reflected in the grammar.
For example, we can pluralize count nouns and say three ideas, but this
cannot be done with mass nouns: *three water.
demonstrative
Terms used to describe a location either relatively close to a speaker
(this, these, here) or relatively far from them (that, those, there). Their
interpretation is context-dependent, since their meaning is partly
determined by the particular person who is speaking.
derivational morpheme
A morpheme added to a stem to produce a new word, which may
belong to a different category from the stem. The morpheme -er added
to the verb believe produces a noun, believer. Adding -ment to the verb
excite, produces the noun excitement. See also inflectional morpheme.
determiner
A syntactic category of words used to modify nouns, but with no
semantic content (meaning) of their own, including articles (a, the),
demonstratives (this, that) and quantifiers (each, every). Determiners
are often used to pick out the particular referent one is talking about (a
prince among men; this apple is rotten; every time we say goodbye).
determiner phrase
A syntactic phrase comprising (minimally) a determiner and a noun: a
pickle; this day. The determiner is described as the head of the phrase.
Confusingly, phrases like this are also traditionally described as noun
phrases (with the noun as head). The choice is a matter of dispute
among linguists (we’d better leave them to it).
direct object
See object.
ecological validity
The extent to which a study reflects conditions in the real world.
Laboratory experiments are often accused of having low ecological
validity because the things people are asked to do in test cubicles can
bear little relation to real-life situations and behaviours. High ecological
validity is desirable, because it allows one to generalize the findings of a
study more readily and draw conclusions about people in general.
fricative
A class of consonants characterized by a hissing quality, caused by air
pushing through a very narrow gap in the vocal tract. Consider the word-
initial fricatives in the following: /f/ in fricative, /θ/ in think and /t∫/ in
champ. Now have some fricative fun with She sells sea shells on the
seashore.
genitive
See case.
grammar
Traditionally, the grammar (or syntax) of a language comprises the rules
that determine how words are put together to make sentences. In
English, The cat sat on the mat is a grammatical sentence because the
six words have been placed in a particular order. If I combine the same
six words in a different way, the result can be ungrammatical: *Mat the
the on sat cat. In this latter case, the rules (or principles) of grammar
have been violated. This traditional interpretation of grammar still has
some currency. But you should be aware that grammar, as a term, is
quite slippery. In this book I use this word in two ways: (1) to cover
both morphology and syntax; and (2) as a substitute for syntax. Others
use grammar in a very wide sense to embrace syntax, phonology and the
lexicon, that is, the whole language system.
head
A head is an expression that can combine with modifier words to form a
syntactic phrase. This determines the nature of the overall phrase. A
particular head can have one or more modifiers or none at all. In the
noun phrase, Queen of Sheba, the head of the phrase is the noun,
Queen, and the modifier is provided by of Sheba. In the verb phrase,
meeting Shirley MacLaine, the head is the verb, meeting, while the
 modifier is Shirley MacLaine (remember me, Shirley?).
heritability
A measure of the contribution of genetic factors to the differences
observed between people. One way of estimating heritability is to
compare identical twins (who share 100 per cent of their genes) with
non-identical twins (who share 50 per cent of their genes). If genes are
important for explaining variation in a particular behaviour, then
identical twins will be more similar to one another than non-identical
twins.
homophone
Two words which are pronounced in the same way, but which have
different meanings (and often different spellings), for example, bear and
bare, or meat and meet.
imperative
In English, the verb form that appears in sentences for giving orders. In
sit down!, the verb appears as sit (not sits or sitting). In some languages,
including Latin and Hungarian, the imperative is marked with its own
set of affixes.
inflectional morpheme
Inflections are added (in most languages) to the ends of words to denote
a particular grammatical function. For example, -ed added to the end of
a verb indicates that the action took place in the past (compare walk and
walked). Inflections do not change the grammatical category of the word
they are attached to (walk and walked are both verbs). English has only
eight different inflectional morphemes. Other languages have much
richer inflectional morphology (e.g., Finnish and Russian). See also
morphology and derivational morpheme.
innateness (of language)
The property of being inborn or genetically determined. To what extent
is language acquisition determined by the action of genes? It is not
controversial to suggest that something about language development is
innate. All typically developing human beings acquire language and so
must be endowed with the capacity to do so. The problem is to
determine what, precisely, is innate about language acquisition. The
degree to which different behaviours are pre-programmed by genes
constitutes a central concern within the life sciences generally. When
applied to language, a particular interest has been to determine how far
knowledge of grammar is innate. Those, like Chomsky, who emphasize
innate properties are nativists.
instrumental
A grammatical case, used to mark the instrument or means by which an
action is accomplished. English does not have instrumental case, but
many other languages do (including Hungarian and Russian).
intonation
The melody of speech. Each utterance has a pitch contour, created by
successive rises and falls in pitch. Very often, the end of an utterance is
marked by a significant rise or fall in pitch level. Intonation contours can
convey semantic or syntactic distinctions. Consider Paula sings loudly
when chopping onions. Produced with a falling intonation at the end, we
interpret this utterance as a statement. But produced with rising
intonation at the end, it becomes a question. Incidentally, Paula believes
that this prevents onions bringing tears to her eyes. I kid you not.
intransitivity
See transitivity.
irregular past tense verbs
See regular and irregular past tense verbs.
lexeme
Words can appear in different inflected forms: dance, dances, dancing,
danced. These four word forms have, essentially, the same meaning, so
they are different versions of a single lexeme, often written in capitals
(DANCE). Derivational morphology produces new lexemes: DANCE
→ DANCER. Inflectional morphology produces new word forms, not
new lexemes: dance → danced.
lexicon
 The mental dictionary comprising a speaker’s knowledge of words and
morphemes. Grammatical information about words is also included in
the lexicon, such as part of speech (noun, verb) as well as information
on verb argument structure.
linguistics
The study of language per se, rather than the study of a language. An
adult who succeeds in learning English, French, Warlpiri and Welsh can,
legitimately, call themselves a linguist. But in an academic context, a
linguist is someone who studies language. An academic linguist is
interested in explaining what it means to know a language, and what
knowledge of language might comprise. Critically, the study of
linguistics can encompass all languages, in the quest to determine what
aspects of language might be universal versus what might be
idiosyncratic or particular to individual languages. The major branches
of linguistics include: syntax (often used as a synonym for grammar);
semantics (to do with patterns of meaning); and phonology (the sound
system of a language).
main clause
A sentence can comprise a main clause – the indispensible core of the
sentence –and one or more subordinate clauses. In On her wedding
day, Kaoru needed help with her kimono, the main clause is Kaoru
needed help with her kimono.
mass nouns
See count versus mass nouns.
mean length of utterance / MLU
The average length of the utterances in a corpus. The MLU of child
speech provides a very rough measure of syntactic development, on the
assumption that longer utterances are more complex grammatically.
However, the relationship between length and complexity breaks down
beyond an MLU of about 4.0. An MLU greater than 1.0 indicates that
the child has moved from single-word to multi-word speech. Brown
(1973) provides a standard method for calculating MLU, based on the
average number of morphemes (not words) per utterance.
metalinguistic awareness / knowledge / skill
Knowledge about language and the ability to reflect on it. Metalinguistic
awareness does not emerge in any serious way until the child reaches
school age (about four or five years). Perhaps the best-known example is
 the development of so-called phonological awareness in children who
are learning to read. One aspect of phonological awareness is the ability
to recognize explicitly that, for example, the word cat has three
constituent speech sounds, or phonemes: / k æ t /. The child needs
metalinguistic awareness in order to analyse the sound structure of
words in this way.
modifier
See head.
morpheme
See morphology and inflectional morpheme.
morphology
The study of how words are constituted. Morphemes are the smallest
units of meaning in a language. Words like garden or swim each
constitute a single morpheme, because they correspond to a single unit
of meaning that cannot be decomposed any further: gar- in garden does
not have its own meaning independent of the whole word. Garden is a
free morpheme, because it can stand alone as a word in its own right.
Bound morphemes, on the other hand, cannot stand alone but are
attached to other morphemes within a word. For example, singing
comprises two morphemes: sing + ing, where -ing is a bound morpheme,
used to denote an action that is ongoing or continuous; foolishness has
three morphemes: fool + ish + ness. See also inflectional morpheme
and derivational morpheme.
nativism
See innateness.
negation
A type of grammatical construction used to indicate that a proposition is
false. In English, negation can be marked by not or n’t, as in Lill didn’t
wear a kimono at the wedding. Negation can also be marked with never,
nobody and nothing.
negative evidence
Information which signals that an utterance is ungrammatical. Parental
corrections of child grammatical errors constitute a form of negative
evidence.
nominative
See case.
noun
 A syntactic category which typically denotes some kind of entity
(haggis, mountain, wig). But many nouns do not refer to an easily
identified object in the world (invasion, sickness, frenzy). Nouns are
therefore defined in grammatical terms. For example, in English, nouns
are identified as those words that can co-occur with articles in noun
phrases (a haggis, a frenzy). See also count versus mass nouns and
common versus proper nouns.
noun phrase
A syntactic phrase with a noun as its head (man in a polite young man;
excesses in the excesses of youth). Noun phrases feature as the subject
or the objects of a sentence. See also determiner phrase.
object
A grammatical category which is the complement of a transitive verb
(the biscuit in You ate my last biscuit). We can distinguish direct and
indirect objects. Where verbs take only one object, it is a direct object
(my last biscuit in our example). Some verbs, like give, also have an
indirect object:

In English, the object comes after the verb and is marked with objective
case. Tim volleyed it into the net, is possible, because it has objective
case, whereas *Tim volleyed its into the net is disallowed, because its
does not have objective case.
orthography
The written form of a language as codified in a set of symbols
(graphemes). In many languages, graphemes are alphabetic, with each
letter in the alphabet corresponding to particular phonemes. The
grapheme–phoneme correspondence is highly predictable in some
languages (like Italian), which means that if you see a particular
sequence of letters, you will know how to pronounce it. In other
languages, like English, the grapheme–phoneme correspondence is far
less regular. For example, the letters -ough can be pronounced in several
different ways. Compare tough, through, plough, cough, hiccough,
though. Alphabets represent only one form of writing system. In other
systems, graphemes represent a whole syllable (hiragana in Japanese) or
even whole words (kanji in Chinese).
overextension
 The use of a word beyond its conventional range. For example, one
child used the word papa to mean his father, but also used papa to
denote his grandfather and mother. For the child, the extent of the
word’s meaning was broader than normal.
overgeneralization
An error in which a grammatical rule is applied beyond its normal range.
Verbs like stole are described as transitive, because they require an
object, as in Lesley stole our bank book. Sometimes children
overgeneralize the transitive pattern, by applying it to intransitive verbs
like giggle that do not require an object: me in Don’t giggle me.
overregularization
The misapplication of a grammatical rule to irregular forms. For
example, regular plural nouns are formed by adding -s to the singular
(one rocket, two rockets). If -s is added to an irregular noun like sheep,
an overregularized form is produced: sheeps. Overregularization is a
particular kind of overgeneralization.
part of speech
Traditionally, the grammatical category of a word. Traditional grammars
identify eight parts of speech in English, as below, though modern
linguists offer different analyses (e.g., they generally include
determiners as a separate part of speech).
parse
The analysis of language, to establish the structure of words, phrases or
sentences. Most frequently used to describe the analysis of sentences to
determine their linguistic structure, but one could also parse a word into
its constituent morphemes.
patient
An object noun phrase in a sentence with a thematic role, or meaning,
that denotes the recipient of an action. In Olga swallowed an aspirin, the
patient is an aspirin. The patient can also be described as one of a
verb’s arguments.
performance
See competence.
person
Traditional grammars of English recognize three persons: first, second
and third. First person (I, we) includes the speaker. Second person (you)
excludes the speaker, but includes the addressee. Third person (he, she,
it, they) refers to neither the speaker nor the addressee, but rather,
someone or something else.
phoneme
See phonology and Box 2.2.
phonology
The sound system of a language. Each language deploys a repertoire of
sounds that are used to convey meaning. The smallest unit of sound is
known as the phoneme (see Box 2.2) and is written between slash
brackets (e.g., /b/). Note that /b/ denotes a meaningful unit of sound, it
does not – despite appearances – denote the English letter ‘b’ (that is a
matter of orthography, not phonology). If two sounds are contrastive,
then they will constitute separate phonemes in a language. Consider the
words bat and mat. These two words are identical phonologically, that
is, they sound identical, except for the initial sound. The physical
difference between the two sounds /b/ and /m/ is sufficient to denote a
difference of meaning. Hence, /b/and /m/ are distinct phonemes in
English.
pitch
See intonation.
possessive
A structure that indicates possession. In English, possession can be
 marked with a possessive suffix which appears in written English as an
apostrophe followed by S: Yvonne’s government report. Having said
that, the days of the apostrophe seem to be numbered. If it appears at all,
it is often in the wrong place. My favourite is Mr Cheap Potatoe’s, a
shop I used to frequent. Farewell, Apostrophe! Rest in peace.
poverty of the stimulus
The idea that the linguistic environment of the child is impoverished. On
this nativist view, the input does not contain sufficient information to
allow for the acquisition of grammar. In consequence, innate knowledge
must be invoked to explain how the child nevertheless manages to
acquire language.
pragmatics
The study of how people use language. Pragmatic factors include the
context in which an utterance is produced and affect our interpretation of
sentences. If I say ‘It’s cold in here’ you are unlikely to believe that I’m
making a factual statement about the ambient temperature. You are more
likely to interpret it as a request to close the window or turn up the
heating.
prefix
A morpheme that is added to the left of a word form (as written). In
undo, the prefix un- is added to do. In devalue, the prefix de- is added to
value.
preposition
A grammatical category typically used to denote location or manner
(e.g., in, of, by, for, under, with, from, at, on). In English, prepositions
can usually be modified by straight or right (right under, straight from).
A prepositional phrase has a preposition as its head, as in over the top,
under the wire or through the middle.
pronoun
A word that can be used as a substitute for a noun. English pronouns
include I, you, he, she, it, we, they. In the sentence, Jane trapped a thief
in her car, we can substitute the pronoun she for Jane: She trapped a
thief in her car (well done, Jane). Notice that pronouns do not have an
intrinsic meaning. We need to refer to the context (linguistic or social) in
order to work out which person she refers to.
proper nouns
See common versus proper nouns.
regular and irregular past tense verb
Regular verbs follow a default pattern. In English, if I want to talk about
past events, I can add -ed to the end of a regular verb (danced, called,
wanted). This allows me to say things about the past, like: I could’ve
danced all night. If you consider how -ed is pronounced, you will
discover that, phonologically, there are three possibilities: /t/ as in
danced; /d/ as in called; or /ıd/ as in wanted. Most verbs in English are
regular, but some verbs are irregular: they do not follow this simple
default pattern: buy changes to bought (not *buyed); go changes to went
(not *goed); and eat changes to ate (not *eated).
relative clause
A clause that ‘relates to’ or modifies an expression. In The neighbours
(who thought we were out), the relative clause is bracketed and modifies
the noun phrase the neighbours. Relative clauses in English can be
introduced with a relative pronoun (e.g., who, that, which).
segmentation problem
The problem of identifying individual linguistic units, like phoneme or
word, from continuous speech. This is problematic because, physically,
there are often no clear boundaries to mark the beginning or end of
individual phonemes or words. The child must nevertheless identify the
boundaries between linguistic units.
sentence
A clause which can stand by itself and which comprises a subject noun
phrase followed by a verb phrase. In written English, the independence
of sentences is marked by starting a new sentence with a capital letter
and terminating it with a full stop.
sentential complement
A kind of subordinate clause. In I think that Mo wants to be a
burlesque queen, the sentential complement is that Mo wants to be a
burlesque queen. See also complement.
stop
A consonant sound in which the flow of air leaving the mouth is
completely blocked for a very brief period. The stop sounds of English
include /b, p, m, t, d, n, g, k/ (see also the Pronunciation Guide).
subject
A grammatical category which takes so-called nominative case marking,
which in English can be seen in the choice of pronouns: He cooks really
 well. *Him cooks really well. *His cooks really well. Only he is allowed
in subject position, because it carries the correct nominative case. In
English, the subject is positioned at the start of a sentence. We can also
identify the subject from the fact that it agrees with the verb. That is, the
ending on the verb is dictated by the properties of the subject. In Gary
reads Private Eye, we add -s to the verb to mark agreement with the
singular third person subject Gary. In They read Private Eye, the
uninflected verb form read now agrees with the plural subject they.
subordinate clause
A clause which appears as part of another main clause. In Shiho wanted
to go shopping in London, the subordinate clause is to go shopping in
London.
suffix
A morpheme that is added to the right of a word form (when written).
In walked, the suffix -ed is added to walk (see inflectional morpheme).
In madness, the suffix -ness is added to mad (see derivational
morpheme).
syllable
A phonological unit of speech, whose nucleus is usually a vowel. There
can be optional consonants both before (onset) and after (coda) the
nucleus. The word eye comprises a single syllable with just a vowel
nucleus (no onset or coda). Pie has both an onset, /p/ and a nucleus,
while ape has both a vowel and coda, /p/ (ignore spellings of eye, pie
and ape and focus on how they sound). Cat meanwhile has it all: onset–
nucleus–coda: /kæt/. There are typically ten syllables in each line of a
Shakespeare sonnet. Try counting them in the first line of Sonnet 18:
Shall I compare thee to a summer’s day?
syntax
The rules that determine how words are put together to make sentences.
If you think this sounds like grammar, you’d be right. But linguists
sometimes use grammar as an umbrella term for other aspects of
language. See the entry under grammar to discover how these two
terms differ.
transitivity
Verbs like take, eat or kick, which require a direct object, are transitive.
Compare Diana watered her garden with the odd *Diana watered. The
verb water implies that something is watered (the object). Other verbs
 describe states or activities which are not directed towards anything, and
which therefore do not need an object. These verbs are intransitive.
Compare Sally slept with *Sally slept the bed. The verb sleep is
intransitive and so does not require an object (the bed).
verb
Verbs are a syntactic category, often described as ‘doing words’ or
‘actions’, as in: Sue cooks really well or Jim cleaned the brass door
plate. But verbs are not confined to observable actions. For example,
they can also be used to denote states (believe) and sensations ( feel).
What unites all verbs is that they can take inflections, like -ed, -s or -ing
(governed, governs, governing). In essence, verbs constitute the heart of
a sentence. Subject and object are complements of the verb in a
sentence.
verb argument structure
The arguments in a sentence describe the different roles played by the
participants. The sentence, André lost some weight at last, comprises a
predicate (the verb lose) and two arguments (André and some weight at
last). The argument structure of a verb dictates how many arguments are
required, and what roles they fulfil. For example, the verb sleep requires
only one argument (the person sleeping), as in Richard slept (*Richard
slept the bed is not possible). Other verbs require three arguments, for
example, give, as in Martha’s voice gives me the creeps. We need three
arguments: the giver (Martha), the given (the creeps) and a recipient
(me).
verb phrase
A syntactic phrase which is headed by a verb. The verb phrase is
bracketed in the following: Dante (has an unusual name) or Isaac (plays
computer games all day). The verb phrase can comprise a verb plus its
arguments (see verb argument structure). For example, the verb hold
requires one argument: hold (the fort). Deliver requires two arguments,
as in deliver (the pizza) (to Gabriel).
voiced and voiceless sounds
A voiced sound is produced when the vocal cords (located in your
Adam’s apple) are vibrating. Voiceless sounds are produced without
vocal cord vibration. Place your fingers on your Adam’s apple and try
alternating between the voiceless /f/ and the voiced /v/.
vowel
 See consonants and vowels.
word form
See lexeme.
References
Abbot-Smith, K., Imai, M., Durrant, S. & Nurmsoo, E. (2017). The role of
timing and prototypical causality on how preschoolers fast-map novel verb
meanings. First Language, 37(2), 186–204.
Abbot-Smith, K. & Tomasello, M. (2009). The influence of frequency and
semantic similarity on how children learn grammar. First Language, 30(1),
1–23.
Abe, K. & Watanabe, D. (2011). Songbirds possess the spontaneous ability to
discriminate syntactic rules. Nature Neuroscience, 14(8), 1067–1074.
Adams, A.M. & Gathercole, S.E. (2000). Limitations in working memory:
Implications for language development. International Journal of Language
and Communication Disorders, 35(1), 95–116.
Adger, D. (2003). Core syntax: A minimalist approach. Oxford: Oxford
University Press.
Aitchison, J. (1998). The articulate mammal: An introduction to
psycholinguistics (4th ed.). London: Routledge.
Aitchison, J. (2010). Teach yourself linguistics (7th ed.). London: Hodder
Headline.
Akhtar, N. (2005). The robustness of learning through overhearing.
Developmental Science, 8(2), 199–209.
Akhtar, N., Callanan, M., Pullum, G.K. & Scholz, B.C. (2004). Learning
antecedents for anaphoric one. Cognition, 93(2), 141–145.
Akhtar, N., Dunham, F. & Dunham, P.J. (1991). Directive interactions and
early vocabulary development: The role of joint attentional focus. Journal
of Child Language, 18, 41–49.
Aksu-Koç, A. (1997). Verb inflection in Turkish: A preliminary analysis of
the early stage. In W.U. Dressler (Ed.), Studies in pre- and proto
morphology (pp. 127–139). Vienna: Verlag Der Österreichischen
Akademie der Wissenschaften.
Aksu-Koç, A. & Slobin, D.I. (1985). The acquisition of Turkish. In D.I.
Slobin (Ed.), The crosslinguistic study of language acquisition (Vol. 1)
(pp. 839–878). Hillsdale, NJ: Erlbaum.
Albin, D.D. & Echols, C.H. (1996). Stressed and word-final syllables in
infant-directed speech. Infant Behavior & Development, 19(4), 401–418.
Albirini, A. (2015). Factors affecting the acquisition of plural morphology in
Jordanian Arabic. Journal of Child Language, 42(4), 734–762.
Albright, A. & Hayes, B. (2003). Rules vs. analogy in English past tenses: A
computational/experimental study. Cognition, 90(2), 119–161.
Ambridge, B. (2010). Children’s judgments of regular and irregular novel
past-tense forms: New data on the English past-tense debate.
Developmental Psychology, 46(6), 1497–1504.
Ambridge, B. (2013). How do children restrict their linguistic
generalizations?: An (un-) grammaticality judgment study. Cognitive
Science, 37(3), 508–543.
Ambridge, B., Bidgood, A., Twomey, K.E., Pine, J.M., Rowland, C.F. &
Freudenthal, D. (2015). Preemption versus entrenchment: Towards a
construction-general solution to the problem of the retreat from verb
argument structure overgeneralization. PLoS One, 10(4), e0123723.
Ambridge, B., Kidd, E., Rowland, C.F. & Theakston, A.L. (2015). The
ubiquity of frequency effects in first language acquisition. Journal of Child
Language, 42(2), 239–273.
Ambridge, B., Pine, J.M. & Lieven, E.V.M. (2014). Child language
acquisition: Why universal grammar doesn’t help. Language, 90(3),
e53–e90.
Ambridge, B., Pine, J.M., Rowland, C.F. & Young, C.R. (2008). The effect
of verb semantic class and verb frequency (entrenchment) on children’s
and adults’ graded judgements of argument-structure overgeneralization
errors. Cognition, 106(1), 87–129.
Ambridge, B. & Rowland, C.F. (2013). Experimental methods in studying
child language acquisition. WIREs Cognitive Science. doi:
10.1002/wcs.1215.
Ambridge, B., Rowland C.F. & Pine, J.M. (2008). Is structure dependence an
innate constraint? New experimental evidence from children’s complex-
question production. Cognitive Science, 32(1), 222–255.
Ambridge, B., Rowland, C.F., Theakston, A. & Tomasello, M. (2006).
Comparing different accounts of children’s non-inversion errors in object
wh-questions: What experimental data can tell us? Journal of Child
Language, 30(3), 519–557.
Ambridge, B., Theakston, A.L., Lieven, E.V.M & Tomasello, M. (2006). The
distributed learning effect for children’s acquisition of an abstract syntactic
construction. Cognitive Development, 21(2), 174–193.
Anderson, D.R. & Pempek, T.A. (2005). Television and very young children.
American Behavioral Scientist, 48(5), 505–522.
Anderson, J. (1983). The architecture of cognition. Cambridge, MA: Harvard
University Press.
Anglin, J.M. (1993). Knowing versus learning words. Monographs of the
Society for Research in Child Development, 58, 176–186.
Archer, S.L., Zamunerb, T., Engel, K., Fais, L. & Curtin, S. (2016). Infants’
discrimination of consonants: Interplay between word position and
acoustic saliency. Language Learning and Development, 12(1), 60–78.
Archibald, L.M.D. & Gathercole, S.E. (2007). Nonword repetition and serial
recall: Equivalent measures of verbal short-term memory? Applied
Psycholinguistics, 28(4), 587–606.
Argus, R., Laalo, J. & Johansen Ijäs, K. (2014). Acquisition of compound
nouns in Estonian, Finnish and Sami: Similarities and differences. The
Central and Eastern European Online Library, 08–09, 648–669.
Arunachalam, S. (2013). Two-year-olds can begin to acquire verb meanings
in socially impoverished contexts. Cognition, 129, 569–573.
Asano, T., Kojima, T., Matsuzawa, T., Kubota, K. & Murofushi, K. (1982).
Object and colour naming in chimpanzees (Pan troglodytes). Proceedings
of the Japan Academy Series B – Physical and Biological Sciences, 58(5),
118–122.
Aslin, R.N., Shukla, M. & Emberson, L.L. (2015). Hemodynamic correlates
of cognition in human infants. Annual Review of Psychology, 66,
349–379.
Au, T.K.F., Dapretto, M. & Song, Y.K. (1994). Input versus constraints:
Early word acquisition in Korean and English. Journal of Memory and
Language, 33(5), 567–582.
Bachelard, G. (1960/1971). The poetics of reverie: Childhood, language, and
the cosmos (translated by D. Russell). Boston, MA: Beacon Press.
Baillargeon, R. (1987). Object permanence in 3½-month-old and 4½-month-
old infants. Developmental Psychology, 23(5), 655–664.
Baillargeon, R., Li, J., Luo, Y.Y. & Wang, S.H. (2006). Under what
conditions do infants detect continuity violations? Attention and
Performance, 20, 163–188.
Baker, P. & Eversley, J. (Eds.) (2000). Multilingual capital: The languages of
London’s schoolchildren and their relevance to economic, social and
educational policies. London: Battlebridge.
Bakker, I., MacGregor, L.J., Pulvermüller, F. & Shtyrov, Y. (2013). Past
tense in the brain’s time: Neurophysiological evidence for dual-route
 processing of past-tense verbs. NeuroImage, 71, 187–195.
Bannard, C. & Lieven, E. (2012). Formulaic language in L1 acquisition.
Annual Review of Applied Linguistics, 32, 3–16.
Barnes, S., Gutfreund, M., Satterly, D. & Wells, G. (1983). Characteristics of
adult speech which predict children’s language development. Journal of
Child Language, 10(1), 65–84.
Barr, R. & Wyss, N. (2008). Reenactment of televised content by 2-year olds:
Toddlers use language learned from television to solve a difficult imitation
problem. Infant Behavior and Development, 31(4), 696–703.
Barrett, M. (1995). Early lexical development. In P. Fletcher & B.
MacWhinney (Eds.), The handbook of child language (pp. 362–392).
Oxford: Blackwell.
Bates, E., Dale, P.S. & Thal, D. (1995). Individual differences and their
implications for theories of language development. In P. Fletcher & B.
MacWhinney (Eds.), The emergence of language (pp. 29–79). Mahwah,
NJ: Lawrence Erlbaum.
Bates, E. & Goodman, J.C. (1997). On the inseparability of grammar and the
lexicon: Evidence from acquisition, aphasia and real-time processing.
Language and Cognitive Processes, 12(5–6), 507–584.
Bates, E. & Goodman, J.C. (1999). On the emergence of grammar from the
lexicon. In B. MacWhinney (Ed.), The emergence of language. Mahwah,
NJ: Lawrence Erlbaum.
Bates, E. & MacWhinney, B. (1989). Functionalism and the competition
model. In B. MacWhinney & E. Bates (Eds.), The crosslinguistic study of
sentence processing (pp. 3–76). Cambridge: Cambridge University Press.
Bates, E., Marchman, V.A., Thal, D., Fenson, L., Dale, P., Reznick, J.S.,
Reilly, J. & Hartung, J. (1994). Developmental and stylistic variation in the
composition of early vocabulary. Journal of Child Language, 21(1),
85–123.
Bavin, E.L. (1992). The acquisition of Warlpiri. In D.I. Slobin (Ed.), The
crosslinguistic study of language acquisition (Vol. 3). Hillsdale, NJ:
Lawrence Erlbaum.
Becker, J.A. (1994). ‘Sneak-shoes’, ‘sworders’, and ‘nose-beards’: A case
study of lexical innovation. First Language, 14(2), 195–211.
Beckers, G.J.L. (2011). Bird speech perception and vocal production: A
comparison with humans. Human Biology, 83(2), 191–212.
Behrens, H. (Ed.) (2008). Corpora in language acquisition research: History,
 methods, perspectives. Amsterdam: John Benjamins.
Behrens, H. (2009). Usage-based and emergentist approaches to language
acquisition. Linguistics, 47(2), 383–411.
Bekoff, M. (2013). We are animals and therein lies hope for a better future.
Blog post. Psychology Today. www.psychologytoday.com/blog/animal–
emotions/201312/we–are–animals–and–therein–lies–hope–better–future.
Accessed December 1, 2015.
Benasich, A.A., Choudhury, N., Friedman, J.T., Realpe-Bonilla, T.,
Chojnowska, C. & Gou, Z. (2006). The infant as a prelinguistic model for
language learning impairments: Predicting from event-related potentials to
behavior. Neuropsychologia, 44(3), 396–411.
Berent, I. & Pinker, S. (2008). Compound formation is constrained by
morphology: A reply to Seidenberg, MacDonald & Haskell. The Mental
Lexicon, 3(2), 176–187.
Berko, J. (1958). The child’s learning of English morphology. Word, 14,
150–177.
Berman, R. (1985). The acquisition of Hebrew. In D.I. Slobin (Ed.), The
crosslinguistic study of language acquisition (Vol. 1). Hillsdale, NJ:
Lawrence Erlbaum.
Berman, R. (1987). A developmental route: Learning about the form and use
of complex nominals in Hebrew. Linguistics, 25, 1057–1085.
Berman, R. & Clark, E.V. (1989). Learning to use compounds for contrast:
Data from Hebrew. First Language, 9(3), 247–270.
Berwick, R.C. & Chomsky, N. (2016). Why only us: Language and
evolution. Cambridge, MA: MIT Press.
Berwick, R.C., Okanoya, K., Beckers, G.J.L., Bolhuis, J.J. (2011). Songs to
syntax: The linguistics of birdsong. Trends in Cognitive Science, 15,
113–121.
Berwick, R.C., Pietroski, P., Yankama, B. & Chomsky, N. (2011). Poverty of
the stimulus revisited. Cognitive Science, 35(7), 1207–1242.
Best, C.T. & McRoberts, G.W. (1989). Phonological influences on the
perception of native and non-native speech contrasts. Paper presented at
the biennial meeting of the Society for Research in Child Development.
Best, C.T. & McRoberts, G.W. (2003). Infant perception of non-native
consonant contrasts that adults assimilate in different ways. Language and
Speech, 46, 183–216.
Best, C.T., McRoberts, G.W., Lafleur, R. & Silverisenstadt, J. (1995).
 Divergent developmental patterns for infants’ perception of two nonnative
consonant contrasts. Infant Behavior & Development, 18(3), 339–350.
Best, C.T., McRoberts, G.W. & Sithole, N.M. (1988). Examination of
perceptual reorganization for nonnative speech contrasts: Zulu click
discrimination by English-speaking adults and infants. Journal of
Experimental Psychology, 14(3), 345–360.
Bialystok, E. (1997). The structure of age: In search of barriers to second
language acquisition. Second Language Research, 13, 116–137.
Bialystok, E. & Hakuta, K. (1994). In other words: The science and
psychology of second language acquisition. New York: Basic Books.
Bialystok, E. & Hakuta, K. (1999). Confounded age: Linguistic and cognitive
factors in age differences for second language acquisition. In D. Birdsong
(Ed.), Second language acquisition and the critical period hypothesis.
Mahwah, NJ: Lawrence Erlbaum.
Bialystok, E. & Miller, B. (1999). The problem of age in second language
acquisition: Influences from language, task, and structure. Bilingualism:
Language and Cognition, 2(2), 127–145.
Birdsong, D. (1992). Ultimate attainment in second language acquisition.
Language, 68, 706–755.
Birjandi, P. & Nasrolahi, A. (2012). Negative evidence in Iranian children:
How do Iranian parents correct their children’s errors? Journal of
Language Teaching and Research, 3(4), 700–706.
Bishop, D.V.M., Adams, C.V. & Norbury, C.F. (2006). Distinct genetic
influences on grammar and phonological short-term memory deficits:
Evidence from 6-year-old twins. Genes Brain and Behavior, 5(2),
158–169.
Blom, E. & Paradis, J. (2016). Introduction: Special issue on age effects in
child language acquisition. Journal of Child Language, 43(3), 473–478.
Bloom, L. (1973). One word at a time: The use of single-word utterances
before syntax. The Hague: Mouton.
Bloom, L., Hood, L. & Lightbown, P.M. (1974). Imitation in language
development: If, when, and why. Cognitive Psychology, 6(3), 380–420.
Bloom, L., Tinker, E. & Margulis, C. (1993). The words children learn:
Evidence against a noun bias in early vocabularies. Cognitive
Development, 8, 431–450.
Bloom, P. (2000). How children learn the meanings of words. Cambridge,
MA: MIT Press.
Bloom, P. (2004). Can a dog learn a word? Science, 304(5677), 1605–1606.
Bloom, P. & Markson, L. (1998). Capacities underlying word learning.
Trends in Cognitive Sciences, 2, 67–73.
Bohannon, J.N., Padgett, R.J., Nelson, K.E. & Mark, M. (1996). Useful
evidence on negative evidence. Developmental Psychology, 32(3),
551–555.
Bohannon, J.N. & Stanowicz, L. (1988). The issue of negative evidence:
Adult responses to children’s language errors. Developmental Psychology,
24, 684–689.
Bohannon, J.N. & Warren-Leubecker, A. (1985). Theoretical approaches to
language acquisition. In J. Berko Gleason (Ed.), The development of
language (pp. 173–226). Columbus, OH: C.E. Merrill.
Bongaerts, T., Planken, B. & Schils, E. (1995). Can late starters attain a
native accent in a foreign language? A test of the critical period hypothesis.
In D. Singleton & Z. Lengyel (Eds.), The age factor in second language
acquisition: A critical look at the critical period hypothesis. Clevedon:
Multilingual Matters.
Booth, A.E. & Waxman, S.R. (2008). Taking stock as theories of word
learning take shape. Developmental Science, 11(2), 186–194.
Booth, A.E. & Waxman, S.R. (2009). A horse of a different color: Specifying
with precision infants’ mappings of novel nouns and adjectives. Child
Development, 80(1), 15–22.
Borgström, K., Torkildsen, J. von Koss & Lindgren, M. (2015). Substantial
gains in word learning ability between 20 and 24 months: A longitudinal
ERP study. Brain and Language, 149, 33–45.
Bornstein, M.H., Cote, L.R., Maital, S., Painter, K., Park, S.Y., Pascual, L.,
Venuti, P. & Vyt, A. (2004). Cross-linguistic analysis of vocabulary in
young children: Spanish, Dutch, French, Hebrew, Italian, Korean, and
American English. Child Development, 75(4), 1115–1139.
Botha, R.P. (1989). Challenging Chomsky: The generative garden game.
Oxford: Basil Blackwell.
Bowerman, M. (1982). Reorganizational processes in lexical and syntactic
development. In E. Wanner & L.R. Gleitman (Eds.), Language acquisition:
The state of the art. Cambridge: Cambridge University Press.
Bowey, J.A. (2005). Grammatical sensitivity: Its origins and potential
contribution to early word reading skill. Journal of Experimental Child
Psychology, 90(4), 318–343.
Boyd, J.K. & Goldberg, A.E. (2012). Young children fail to fully generalize a
novel argument structure construction when exposed to the same input as
older learners. Journal of Child Language, 39(3), 457–481.
Braine, M.D.S. (1963). The ontogeny of English phrase structure: The first
phase. Language, 39(1), 1–13.
Braine, M.D.S., Brooks, P.J., Cowan, N., Samuels, M.C. & Tamis-LeMonda,
C.S. (1993). The development of categories at the semantics syntax
interface. Cognitive Development, 8(4), 465–494.
Brand, R.J. & Shallcross, W.L. (2008). Infants prefer motionese to adult-
directed action. Developmental Science, 11(6), 853–861.
Brandt, S., Diessel, H. & Tomasello, M. (2008). The acquisition of German
relative clauses: A case study. Journal of Child Language, 35(2), 325–348.
Broca, P. (1865). Sur le siège de la faculté du langage articulé suivies d’une
observation d’aphémie, Bulletin de la Société Anatomique de Paris, 6, 330.
Broen, P.A. (1972). The verbal environment of the language-learning child.
Monograph of the American Speech and Hearing Association, 17.
Brooks, P.J. & Kempe, V. (2012). Language development. Chichester: BPS
Blackwell.
Brooks, P.J. & Tomasello, M. (1999). How children constrain their argument
structure constructions. Language, 75(4), 720–738.
Brooks, P.J., Tomasello, M., Dodson, K. & Lewis, L.B. (1999). Young
children’s overgeneralizations with fixed transitivity verbs. Child
Development, 70(6), 1325–1337.
Brooks, P.J. & Zizak, O. (2002). Does pre-emption help children learn verb
transitivity? Journal of Child Language, 29(4), 759–781.
Brown, G.D.A. & Watson, F.L. (1987). First in, first out: Word learning age
and spoken word frequency as predictors of word familiarity and word
naming latency. Memory and Cognition, 15(3), 208–216.
Brown, R. (1958). Words and things: An introduction to language. New
York: Free Press.
Brown, R. (1973). A first language. London: George Allen & Unwin.
Brown, R. & Bellugi, U. (1964). Three processes in the child’s acquisition of
syntax. Harvard Educational Review, 34, 133–151.
Brown, R. & Hanlon, C. (1970). Derivational complexity and order of
acquisition in child speech. In R. Brown (Ed.), Psycholinguistics. New
York: Free Press.
Bruer, J.T. (2001). A critical and sensitive period primer. In D.B. Bailey, J.T.
 Bruer, F.J. Symons & J.W. Lichtman (Eds.), Critical thinking about critical
periods. Baltimore, MD: Paul Brookes.
Buchan, H. & Jones, C. (2014). Phonological reduction in maternal speech in
northern Australian English: Change over time. Journal of Child Language,
41(4), 725–755.
Buck-Gengler, C.J., Menn, L. & Healy, A.F. (2004). What ‘Mice Trap’ tells
us about the mental lexicon. Brain and Language, 90(1–3), 453–464.
Buttelmann, D., Carpenter, M., Call, J. & Tomasello, M. (2007). Enculturated
chimpanzees imitate rationally. Developmental Science, 10(4), F31–F38.
Bybee, J. (2008). Usage-based grammar and second language acquisition. In
P. Robinson & N.C. Ellis (Eds.), Handbook of cognitive linguistics and
second language acquisition (pp. 216–236). New York: Routledge.
Bybee, J.L. & Scheibmann, J. (1999). The effect of usage on degrees of
constituency: The reduction of don’t in English. Linguistics, 37(4),
575–596.
Čadková, L. (2015). Do they speak language? Biosemiotics, 8(1), 9–27.
Cameron-Faulkner, T. (2012). A functional account of verb use in the early
stages of English multiword development. Journal of Child Language,
39(4), 885–897.
Cameron-Faulkner, T., Lieven, E. & Tomasello, M. (2003). A construction
based analysis of child directed speech. Cognitive Science, 27(6),
843–873.
Cameron-Faulkner, T. & Noble, C. (2013). A comparison of book text and
Child Directed Speech. First Language, 33(3), 268–279.
Campbell, R.N. (1986). Language acquisition and cognition. In P. Fletcher &
M. Garman (Eds.), Language acquisition (2nd ed.). (pp. 30–48).
Cambridge: Cambridge University Press.
Campbell, R.N. & Grieve, R. (1982). Royal investigations of the origin of
language. Historiographia Linguistica IX, 1(2), 43–74.
Caramazza, A. & Yeni-Komshian, G.H. (1974). Voice onset time in two
French dialects. Journal of Phonetics, 2, 239–245.
Carey, S. (1978). The child as word learner. In M. Halle, G. Miller & J.
Bresnan (Eds.), Linguistic theory and psychological reality. Cambridge,
MA: MIT Press.
Carey, S. (1985). Conceptual change in childhood. Cambridge, MA: MIT
Press.
Carey, S. & Bartlett, E. (1978). Acquiring a single new word. Papers and
 Reports on Child Language Development, 15, 17–29.
Cattell, R. (2000). Children’s language: Consensus and controversy. London:
Cassell.
Cazden, C.B. (1965). Environmental assistance to the child’s acquisition of
grammar. Unpublished PhD thesis, Harvard University, Cambridge, MA.
Cerri, G., Cabinio, M., Blasi, V., Borroni, P., Iadanza, A., Fava, E., Fornia,
L., Ferpozzi, V., Riva, M., Casarotti, A., Boneschi, F.M., Falini, A. &
Bello, L. (2015). The Mirror Neuron System and the strange case of
Broca’s Area. Human Brain Mapping, 36, 1010–1027.
Chater, N. & Manning, C.D. (2006). Probabilistic models of language
processing and acquisition. Trends in Cognitive Sciences, 10(7), 335–344.
Chater, N., Tenenbaum, J.B. & Yuille, A. (2006). Probabilistic models of
cognition: Conceptual foundations. Trends in Cognitive Sciences, 10(7),
287–291.
Chen, J., Rossum, D. & ten Cate, C. (2015). Artificial grammar learning in
zebra finches and human adults: XYX versus XXY. Animal Cognition,
18(1), 151–164.
Cheshire, J. (1982). Variation in an English dialect: A sociolinguistic study.
Cambridge: Cambridge University Press.
Choi, S. (1998). Verbs in early lexical and syntactic development in Korean.
Linguistics, 36(4), 755–780.
Cholewiak, D.M., Sousa-Lima, R.S. & Cerchio, S. (2013). Humpback whale
song hierarchical structure: Historical context and discussion of current
classification issues. Marine Mammal Science, 29(3), E312–E332.
Chomsky, N. (1957). Syntactic structures. The Hague: Mouton.
Chomsky, N. (1959). Review of Verbal behavior by B.F. Skinner, Language,
35(1), 26–58.
Chomsky, N. (1965). Aspects of the theory of syntax. Cambridge, MA: MIT
Press.
Chomsky, N. (1966). Cartesian linguistics: A chapter in the history of
rationalist thought. New York: Harper & Row.
Chomsky, N. (1972). Problems of knowledge and freedom. London: Fontana.
Chomsky, N. (1975). Reflections on language. London: Temple Smith.
Chomsky, N. (1980a). Rules and representations. Oxford: Blackwell.
Chomsky, N. (1980b). Initial states and steady states. In M. Piattelli-
Palmarini (Ed.), Language and learning: The debate between Jean Piaget
and Noam Chomsky. Cambridge, MA: Harvard University Press.
Chomsky, N. (1981). Lectures on government and binding. Dordrecht: Foris.
Chomsky, N. (1985). Turning the tide: U.S. intervention in Central America
and the struggle for peace. London: Pluto Press.
Chomsky, N. (1988). Language and problems of knowledge: The Managua
lectures. Cambridge, MA: MIT Press.
Chomsky, N. (1995). The minimalist program. Cambridge, MA: MIT Press.
Chomsky, N. (1999). On the nature, use, and acquisition of language. In
W.C. Ritchie & T.K. Bhatia (Eds.), Handbook of child language
acquisition (pp. 33–54). San Diego, CA: Academic Press.
Chomsky, N. (2005). Imperial ambitions: Conversations with Noam
Chomsky on the post 9/11 world: Interviews with David Barsamian.
London: Hamish Hamilton.
Chomsky, N. (2011). Language and other cognitive systems: What is special
about language? Language Learning and Development, 7(4), 263–278.
Chomsky, N. (2012). The science of language. Cambridge: Cambridge
University Press.
Chouinard, M.M. & Clark, E.V. (2003). Adult reformulations of child errors
as negative evidence. Journal of Child Language, 30(3), 637–669.
Chow, K.L. & Stewart, D.L. (1972). Reversal of structural and functional
effects of long-term visual deprivation in cats. Experimental Neurobiology,
34, 409–433.
Christensen, J.B. (1984). The perception of voice-onset-time: A cross-
language study of American English and Danish. Annual Report of the
Institute of Phonetics University of Copenhagen, 18, 163–184.
Cimpian, A. & Markman, E.M. (2005). The absence of a shape bias in
children’s word learning. Developmental Psychology, 41(6), 1003–1019.
Clahsen, H., Hadler, M. & Weyerts, H. (2004). Speeded production of
inflected words in children and adults. Journal of Child Language, 31(3),
683–712.
Clancy, P. (1985). The acquisition of Japanese. In D.I. Slobin (Ed.), The
crosslinguistic study of language acquisition (Vol. 1). Hillsdale, NJ:
Lawrence Erlbaum.
Clancy, P. (1989). A case study in language socialisation: Korean wh-
questions. Discourse Processes, 12, 169–191.
Clark, A. & Lappin, S. (2011). Linguistic nativism and the poverty of the
stimulus. Oxford: Wiley Blackwell.
Clark, E.V. (1973). What’s in a word? On the child’s acquisition of semantics
 in his first language. In T.E. Moore (Ed.), Cognitive development and the
acquisition of language (pp. 65–110). New York: Academic Press.
Clark, E.V. (1979). Building a vocabulary: Words for objects, actions, and
relations. In P. Fletcher & M. Garman (Eds.), Language acquisition (pp.
149–160). Cambridge: Cambridge University Press.
Clark, E.V. (1981). Lexical innovations: How children learn to create new
words. In W. Deutsch (Ed.), The child’s construction of language (pp.
299–328). London: Academic Press.
Clark, E.V. (1993). The lexicon in acquisition. Cambridge: Cambridge
University Press.
Clark, E.V. (2003). First language acquisition. Cambridge: Cambridge
University Press.
Clark, E.V. & Bernicot, J. (2008). Repetition as ratification: How parents and
children place information in common ground. Journal of Child Language,
35(2), 349–371.
Clark, E.V., Gelman, S.A. & Lane, N.M. (1985). Compound nouns and
category structure in young children. Child Development, 56(1), 84–94.
Clark, E.V. & Hecht, B.F. (1982). Learning to coin agent and instrument
nouns. Cognition, 12(1), 1–24.
Clark, E.V., Hecht, B.F. & Mulford, R.C. (1986). Acquiring complex
compounds: Affixes and word order in English. Linguistics, 24, 7–29.
Clark, R. (1977). What’s the use of imitation? Journal of Child Language,
4(3), 341–359.
Clarke, E., Reichard, U.H. & Zuberbühler, K. (2006). The syntax and
meaning of wild gibbon songs. PLoS One, 1, e73.
Cleave, P., Becker, S., Curran, M., Van Horne, A. & Fey, M.E. (2015). The
efficacy of recasts in language intervention: A systematic review and meta-
analysis. American Journal of Speech-Language Pathology, 24(2),
237–255.
Cole, E.B. & St. Clair-Stokes, J. (1984). Caregiver-child interactive
behaviours – a videotape analysis procedure. Volta Review, 86(4),
200–216.
Colunga, E. & Smith, L.B. (2008). Knowledge embedded in process: The
self-organization of skilled noun learning. Developmental Science, 11(2),
195–203.
Conley, J.J. (1984). Not Galton, but Shakespeare: A note on the origin of the
term ‘nature and nurture’. Journal of the History of the Behavioral
 Sciences, 20, 184–185.
Cook, R., Bird, G., Catmur, C. Press, C. & Heyes, C. (2014). Mirror neurons:
From origin to function. Behavioral and Brain Sciences, 37, 177–241.
Cook, V.J. (1988). Chomsky’s Universal Grammar: An introduction. Oxford:
Basil Blackwell.
Cooper, R.P. & Aslin, R.N. (1990). Preference for infant-directed speech in
the first month after birth. Child Development, 61(5), 1584–1595.
Coppieters, R. (1987). Competence differences between native and near-
native speakers. Language, 63, 544–573.
Courage, M.L. & Adams, R.J. (1990). Visual acuity assessment from birth to
three years using the acuity card procedures: Cross-sectional and
longitudinal samples. Optometry and Vision Science, 67, 713–718.
Cowie, F. (1997). The logical problem of language acquisition. Synthese,
111, 17–51.
Cowper, E.A. (1992). A concise introduction to syntactic theory: The
Government-binding approach. Chicago, IL: University of Chicago Press.
Crain, S. & Lillo-Martin, D. (1999). An introduction to linguistic theory and
language acquisition. Oxford: Blackwell.
Crain, S. & Nakayama, M. (1987). Structure dependence in grammar
formation. Language, 63(3), 522–543.
Cross, T.G. (1977). Mothers’ speech adjustments: The contribution of
selected child listener variables. In C.E. Snow & C.A. Ferguson (Eds.),
Talking to children: Language input and acquisition (pp. 151–188).
Cambridge: Cambridge University Press.
Csibra, G. (2008). Goal attribution to inanimate agents by 6.5-month-old
infants. Cognition, 107(2), 705–717.
Curtiss, S. (1977). Genie: A psycholinguistic study of a modern-day ‘wild
child’. New York: Academic Press.
Curtiss, S. (1988). Abnormal language acquisition and the modularity of
language. In F.J. Newmeyer (Ed.), Linguistics: The Cambridge survey
(Vol. 2). Cambridge: Cambridge University Press.
Curtiss, S., Fromkin, V., Krashen, S., Rigler, D. & Rigler, M. (1974). The
linguistic development of Genie. Language, 50(3), 528–554.
Dąbrowska, E. (2004). Language, mind and brain: Some psychological and
neurological constraints on theories of grammar. Edinburgh: Edinburgh
University Press.
Dąbrowska, E. (2005). Productivity and beyond: Mastering the Polish
 genitive inflection. Journal of Child Language, 32(1), 191–205.
Dąbrowska, E. & Szczerbinski, M. (2006). Polish children’s productivity
with case marking: The role of regularity, type frequency, and
phonological diversity. Journal of Child Language, 33(3), 559–597.
Danko, S., Boytsova, J., Solovjeva, M., Chernigovskaya, T. & Medvedev, S.
(2014). Event-related brain potentials when conjugating Russian verbs:
The modularity of language procedures. Human Physiology, 40(3),
237–243.
Dapretto, M. & Bjork, E.L. (2000). The development of word retrieval
abilities in the second year and its relation to early vocabulary growth.
Child Development, 71(3), 635–648.
Davidson, D. & Tell, D. (2005). Monolingual and bilingual children’s use of
mutual exclusivity in the naming of whole objects. Journal of Experimental
Child Psychology, 92(1), 25–45.
de la Mora, D.M. & Toro, J.M. (2013). Rule learning over consonants and
vowels in a non-human animal. Cognition, 126(2), 307–312.
de Saussure, F. (1916/1974). Course in general linguistics. London:
Fontana/Collins.
de Waal, F.B.M. (1995). Bonobo, sex and society. Scientific American,
March, 58–64.
DeCasper, A.J. & Spence, M.J. (1986). Prenatal maternal speech influences
newborns’ perception of speech sounds. Infant Behavior and Development,
9(2), 133–150.
Dehaene-Lambertz, G., Hertz-Pannier, L., Dubois, J., Meriaux, S., Roche, A.,
Sigman, M. & Dehaene, S. (2006). Functional organization of perisylvian
activation during presentation of sentences in preverbal infants.
Proceedings of the National Academy of Sciences of the United States of
America, 103(38), 14240–14245.
Dehaene-Lambertz, G. & Spelke, E.S. (2015). The infancy of the human
brain. Neuron, 88(1), 93–109.
DeHart, G.B., Sroufe, L.A. & Cooper, R.G. (2000). Child development: Its
nature and course (4th ed.). Boston, MA: McGraw-Hill.
DeKeyser, R.M. (2000). The robustness of critical period effects in second
language acquisition. Studies in Second Language Acquisition, 22,
499–533.
Demetras, M.J., Post, K.N. & Snow, C.E. (1986). Feedback to first language
learners: The role of repetitions and clarification questions. Journal of
 Child Language, 13, 275–292.
Depaolis, R.A., Vihman, M.M. & Keren-Portnoy, T. (2014). When do infants
begin recognizing familiar words in sentences? Journal of Child Language,
41(1), 226–239.
DeVilliers, J. & DeVilliers, P. (1973). Development of the use of word order
in comprehension. Journal of Psycholinguistic Research, 2(4), 331–341.
Diesendruck, G. (2007). Mechanisms of word learning. In E. Hoff & M.
Shatz (Eds.), Blackwell handbook of language development (pp. 257–276).
Malden, MA: Blackwell.
Diessel, H. (2004). The acquisition of complex sentences. Cambridge:
Cambridge University Press.
Diessel, H. (2009). On the role of frequency and similarity in the acquisition
of subject and non-subject relative clauses. In T. Givón & M. Shibatani
(Eds.), Syntactic complexity (pp. 251–276). Amsterdam: John Benjamins.
Diessel, H. & Tomasello, M. (2005). A new look at the acquisition of relative
clauses. Language, 81(4), 882–906.
Dilley, L.C., Millett, A.L., Devin McAuley, J. & Bergeson, T.R. (2014).
Phonetic variation in consonants in infant-directed and adult-directed
speech: The case of regressive place assimilation in word-final alveolar
stops. Journal of Child Language, 41(1), 155–175.
Dispaldro, M. & Benelli, B. (2012). Putting singular and plural morphology
in context. Journal of Child Language, 39(4), 863–884.
Dittmar, M., Abbot-Smith, K., Lieven, E. & Tomasello, M. (2008a). German
children’s comprehension of word order and case marking in causative
sentences. Child Development, 79(4), 1152–1167.
Dittmar, M., Abbot-Smith, K., Lieven, E. & Tomasello, M. (2008b). Young
German children’s early syntactic competence: A preferential looking
study. Developmental Science, 11(4), 575–582.
Dittmar, M., Abbot-Smith, K., Lieven, E. & Tomasello, M. (2011). Children
aged 2;1 use transitive syntax to make a semantic-role interpretation in a
pointing task. Journal of Child Language, 38(5), 1109–1123.
Dooling, R.J. & Brown, S.D. (1990). Speech perception by budgerigars
(Melopsittacus undulatus): Spoken vowels. Perception and Psychophysics,
47, 568–574.
Dore, J. (1975). Holophrases, speech acts and language universals. Journal of
Child Language, 2(1), 21–40.
Dromi, E. (1987). Early lexical development. Cambridge: Cambridge
 University Press.
Drozd, K.F. (2004). Learnability and linguistic performance. Journal of Child
Language, 31(2), 431–458.
Dubois, J., Hertz-Pannier, L., Cachia, A., Mangin, J.F., Le Bihan, D. &
Dehaene-Lambertz, G. (2009). Structural asymmetries in the infant
language and sensori-motor networks. Cerebral Cortex, 19(2), 414–423.
Dummett, M. (1981). Frege: Philosophy of language (2nd ed.). London:
Duckworth.
Dye, C.D., Walenski, M., Prado, E.L., Mostofsky, S. & Ullman, M.T. (2013).
Children’s computation of complex linguistic forms: A study of frequency
and imageability effects. PLoS One, 8(9), e74683.
Eadie, P.A., Fey, M.E., Douglas, J.M. & Parsons, C.L. (2002). Profiles of
grammatical morphology and sentence imitation in children with Specific
Language Impairment and Down syndrome. Journal of Speech, Language,
and Hearing Research, 45, 720–732.
Eimas, P.D. (1974). Auditory and linguistic processing of cues for place of
articulation by infants. Perception and Psychophysics, 16, 513–521.
Eimas, P.D. (1975). Auditory and phonetic coding of cues for speech:
Discrimination of [r-l] distinction by young infants. Perception &
Psychophysics, 18(5), 341–347.
Eimas, P.D. & Miller, J.L. (1980a). Contextual effects in infant speech
perception. Science, 209, 1140–1141.
Eimas, P.D. & Miller, J.L. (1980b). Discrimination of the information for
manner of articulation. Infant Behavior and Development, 3, 367–375.
Eimas, P.D., Siqueland, E.R., Jusczyk, P.W. & Vigorito, J. (1971). Speech
perception in infants. Science, 209, 1140–1141.
Elman, J.L. (2008). The shape bias: An important piece in a bigger puzzle.
Developmental Science, 11(2), 219–222.
Elsabbagh, M., Hohenberger, A., Campos, R., Van Herwegen, J., Serres, J.,
De Schonen, S., Aschersleben, G. & Karmiloff-Smith, A. (2013).
Narrowing perceptual sensitivity to the native language in infancy:
Exogenous influences on developmental timing. Behavioral Science, 3(1),
120–132.
Emmorey, K., Bellugi, U., Friederici, A. & Horn, P. (1995). Effects of age of
acquisition on grammatical sensitivity: Evidence from on-line & off-line
tasks. Applied Psycholinguistics, 16, 1–23.
Erbaugh, M.S. (1992). The acquisition of Mandarin. In D.I. Slobin (Ed.), The
 crosslinguistic study of language acquisition (Vol. 3). Hillsdale, NJ:
Lawrence Erlbaum.
Estes, K.G., Evans, J.L., Alibali, M.W. & Saffran, J.R. (2007). Can infants
map meaning to newly segmented words? Statistical segmentation and
word learning. Psychological Science, 18(3), 254–260.
Evans, N. & Levinson, S.C. (2009). The myth of language universals:
Language diversity and its importance for cognitive science. Behavioral
and Brain Sciences, 32(5), 429–448.
Everett, D. (2008). Don’t sleep, there are snakes: Life and language in the
Amazonian jungle. London: Profile.
Eysenck, M.W. & Keane, M.T. (2005). Cognitive psychology: A student’s
handbook (5th ed.). Hove: Psychology Press.
Fancher, R.E. (1979). A note on the origin of the term ‘nature and nurture’.
Journal of the History of the Behavioral Sciences, 15, 321–322.
Fantz, R.L. (1961). The origin of form perception. Scientific American,
204(5), 66–72.
Farrant, B.M. & Zubrick, S.R. (2013). Parent–child book reading across early
childhood and child vocabulary in the early school years: Findings from
the Longitudinal Study of Australian Children. First Language, 33(3),
280–293.
Farrar, M.J. (1992). Negative evidence and grammatical morpheme
acquisition. Developmental Psychology, 28(1), 90–98.
Fenson, L., Dale, P.S., Reznick, J.S., Bates, E., Thal, D.J. & Pethick, S.J.
(1994). Variability in early communicative development. Monographs of
the Society for Research in Child Development, 59(5).
Ferguson, C.A. (1977). Baby talk as a simplified register. In C.E. Snow &
C.A. Ferguson (Eds.), Talking to children. Cambridge: Cambridge
University Press.
Ferguson, C.A. & Debose, C.E. (1977). Simplified registers, broken
language, and pidginization. In A. Valdman (Ed.), Pidgin and creole
linguistics. Bloomington, IN: Indiana University Press.
Fernald, A. (1989). Intonation and communicative intent in mothers’ speech
to infants: Is the melody the message? Child Development, 60, 1497–1510.
Fernald, A. & Mazzie, C. (1991). Prosody and focus in speech to infants and
adults. Developmental Psychology, 27, 209–221.
Fernald, A. & Simon, T. (1984). Expanded intonation contours in mothers’
speech to newborns. Developmental Psychology, 20(1), 104–113.
Fernald, A., Taeschner, T., Dunn, J., Papousek, M., Deboysson-Bardies, B. &
Fukui, I. (1989). A cross-language study of prosodic modifications in
mothers’ and fathers’ speech to preverbal infants. Journal of Child
Language, 16(3), 477–501.
Fernald, A. & Weisleder, A. (2015). Twenty years after ‘meaningful
differences,’ it’s time to reframe the ‘deficit’ debate about the importance
of children’s early language experience. Human Development, 58(1), 1–4.
Finn, A.S. & Kam, C.L.H. (2008). The curse of knowledge: First language
knowledge impairs adult learners’ use of novel statistics for word
segmentation. Cognition, 108(2), 477–499.
Fiser, J. & Aslin, R.N. (2001). Unsupervised statistical learning of higher-
order spatial structures from visual scenes. Psychological Science, 12,
499–504.
Fisher, C. (2002). The role of abstract syntactic knowledge in language
acquisition: A reply to Tomasello (2000). Cognition, 82(3), 259–278.
Fisher, C., Klingler, S.L. & Song, H.J. (2006). What does syntax say about
space? 2-year-olds use sentence structure to learn new prepositions.
Cognition, 101(1), B19–B29.
Flege, J.E., Takagi, N. & Mann, V. (1995). Japanese adults can learn to
produce English /r/ and /l/ accurately. Language and Speech, 38(1), 25–55.
Flege, J.E., Takagi, N. & Mann, V. (1996). Lexical familiarity and English-
language experience affect Japanese adults’ perception of j and l. Journal
of the Acoustical Society of America, 99(2), 1161–1173.
Floccia, C., Keren-Portnoy, T., Depaolis, R., Duffy, H., Delle Luche, C.,
Durrant, S., White, L., Goslin, J. & Vihman, M. (2016). British English
infants segment words only with exaggerated infant-directed speech
stimuli. Cognition, 148, 1–9.
Floor, P. & Akhtar, N. (2006). Can 18-month-old infants learn words by
listening in on conversations? Infancy, 9(3), 327–339.
Fluck, M. & Henderson, L. (1996). Counting and cardinality in English
nursery pupils. British Journal of Educational Psychology, 66, 501–517.
Fodor, J.A. (2001). Doing without What’s within: Fiona Cowie’s criticism of
nativism. Mind, 110, 99–148.
Fodor, J.A. & Garrett, M. (1967). Some syntactic determinants of sentential
complexity. Perception and Psychophysics, 2, 289–296.
Forrest, K. & Elbert, M. (2001). Treatment for phonologically disordered
children with variable substitution patterns. Clinical Linguistics and
 Phonetics, 15(1), 41–45.
Fougeron, C. & Keating, P. (1997). Articulatory strengthening at edges of
prosodic domains. Journal of the Acoustical Society of America, 101,
3728–3740.
Fouts, R.S. & Fouts, D.H. (2004). Primate language. In R.L. Gregory (Ed.),
The Oxford companion to the mind (pp. 631–633). Oxford: Oxford
University Press.
Frank, M.C., Braginsky, M., Yurovsky, D. & Marchman, V.A. (2016).
Wordbank: An open repository for developmental vocabulary data. Journal
of Child Language. Published online: 18 May 2016, pp. 1–18.
Freedman, R.D. (1979). Developmental neurobiology of vision. New York:
Plenum Press.
Freyd, P. & Baron, J. (1982). Individual differences in acquisition of
derivational morphology. Journal of Verbal Learning and Verbal Behavior,
21(3), 282–295.
Fry, D.B. (1955). Duration and intensity as physical correlates of linguistic
stress. Journal of the Acoustical Society of America, 27, 765–768.
Furrow, D., Baillie, C., McLaren, J. & Moore, C. (1993). Differential
responding to two- and three-year-olds’ utterances: The role of
grammaticality and ambiguity. Journal of Child Language, 20, 363–375.
Gallese, V., Fadiga, L., Fogassi, L. & Rizzolatti, G. (1996). Action
recognition in the premotor cortex. Brain, 119(2), 593–609.
Gampe, A., Brauer, J. & Daum, M.M. (2016). Imitation is beneficial for verb
learning in toddlers. European Journal of Developmental Psychology.
Published online: 3 February 2016, pp. 1–20.
Gampe, A., Liebal, K. & Tomasello, M. (2012). Eighteen-month-olds learn
novel words through overhearing. First Language, 32(3), 385–397.
Ganger, J. & Brent, M.R. (2004). Reexamining the vocabulary spurt.
Developmental Psychology, 40(4), 621–632.
Gardner, B.T. & Gardner, R.A. (1991). Chimp-language wars. Science,
252(5009), 1046.
Gardner, R.A. & Gardner, B.T. (1969). Teaching sign language to a
chimpanzee. Science, 165, 664–673.
Garnica, O.K. (1977). Some prosodic and paralinguistic features of speech to
young children. In C.E. Snow & C.A. Ferguson (Eds.), Talking to children
(pp. 63–88). Cambridge: Cambridge University Press.
Gathercole, S.E. (2006). Nonword repetition and word learning: The nature
 of the relationship. Applied Psycholinguistics, 27(4), 513–543.
Gathercole, S.E., Willis, C., Emslie, H. & Baddeley, A.D. (1992).
Phonological memory and vocabulary development during the early school
years: A longitudinal study. Developmental Psychology, 28(5), 887–898.
Gelman, S., Croft, W., Fu, P., Clausner, T.C. & Gottfried, G. (1998). Why is
a pomegranate an apple? The role of shape, taxonomic relatedness, and
prior lexical knowledge in children’s overextensions. Journal of Child
Language, 25(2), 267–291.
Gentner, D. (1982). Why nouns are learned before verbs: Linguistic relativity
versus natural partitioning. In S.A. Kuczaj (Ed.), Language development,
Vol. 2: Language, thought, and culture (pp. 301–334). Hillsdale, NJ:
Lawrence Erlbaum.
Gentner, D. (1983). Structure-mapping: A theoretical framework for analogy.
Cognitive Science, 7(2), 155–170.
Gentner, D. (2006). Why verbs are hard to learn. In K. Hirsh-Pasek & R.M.
Golinkoff (Eds.), Action meets word: How children learn verbs (pp.
544–564). Oxford: Oxford University Press.
Gentner, D. & Boroditsky, L. (2001). Individuation, relativity, and early word
learning. In M. Bowerman & S.C. Levinson (Eds.), Language acquisition
and conceptual development (pp. 215–256). Cambridge: Cambridge
University Press.
Gentner, T.Q., Fenn, K.M., Margoliash, D. & Nusbaum, H.C. (2006).
Recursive syntactic pattern learning by songbirds. Nature, 440(7088),
1204–1207.
Gershkoff-Stowe, L. (2001). The course of children’s naming errors in early
word learning. Journal of Cognition and Development, 2(2), 131–155.
Gershkoff-Stowe, L. (2002). Object naming, vocabulary growth, and the
development of word retrieval abilities. Journal of Memory and Language,
46(4), 665–687.
Gershkoff-Stowe, L., Connell, B. & Smith, L. (2006). Priming
overgeneralizations in two- and four-year-old children. Journal of Child
Language, 33(3), 461–486.
Gershkoff-Stowe, L. & Hahn, E.R. (2007). Fast mapping skills in the
developing lexicon. Journal of Speech Language and Hearing Research,
50(3), 682–697.
Gershkoff-Stowe, L. & Hahn, E.R. (2013). Word comprehension and
production asymmetries in children and adults. Journal of Experimental
 Child Psychology, 114(4), 489–509.
Gershkoff-Stowe, L. & Smith, L. (2004). Shape and the first hundred nouns.
Child Development, 75(4), 1098–1114.
Gershwin, I. (1959). Lyrics on several occasions. New York: Knopf.
Gertner, Y., Fisher, C. & Eisengart, J. (2006). Learning words and rules:
Abstract knowledge of word order in early sentence comprehension.
Psychological Science, 17(8), 684–691.
Gervain, J. & Werker, J.F. (2013). Learning non-adjacent regularities at age
0;7. Journal of Child Language, 40(4), 860–872.
Gibson, D.J., Congdon, E.L. & Levine, S.C. (2015). The effects of word-
learning biases on children’s concept of angle. Child Development, 86(1),
319–326.
Gibson, E. & Wexler, K. (1994). Triggers. Linguistic Inquiry, 25, 407–454.
Gillette, J., Gleitman, H., Gleitman, L.R. & Lederer, A. (1999). Human
simulations of vocabulary learning. Cognition, 73(2), 135–176.
Gillis, S. & Ravid, D. (2006). Typological effects on spelling development: A
crosslinguistic study of Hebrew and Dutch. Journal of Child Language,
33(3), 621–659.
Gleitman, L.R. (2014). Syntactic bootstrapping. In P.J. Brooks & V. Kempe
(Eds.), Encyclopedia of language development (pp. 616–618). London:
Sage.
Goldberg, A.E. (1995). Constructions: A construction grammar approach to
argument structure. Chicago, IL: University of Chicago Press.
Goldberg, A.E. (2006). Constructions at work: The nature of generalization
in language. Oxford: Oxford University Press.
Goldberg, A.E. (2011). Corpus evidence of the viability of statistical
preemption. Cognitive Linguistics, 22(1), 131–153.
Goldfield, B.A. & Reznick, J.S. (1992). Rapid change in lexical development
in comprehension and production. Developmental Psychology, 28,
406–413.
Goldman, H.I. (2001). Parental reports of ‘MAMA’ sounds in infants: An
exploratory study. Journal of Child Language, 28, 497–506.
Golinkoff, R. (1986). ‘I beg your pardon?’: The preverbal negotiation of
failed messages. Journal of Child Language, 13, 455–476.
Golinkoff, R.M., Can, D.D., Soderstrom, M. & Hirsh-Pasek, K. (2015).
(Baby) talk to me: The social context of infant-directed speech and its
effects on early language acquisition. Current Directions in Psychological
 Science, 24(5), 339–344.
Golinkoff, R.M. & Hirsh-Pasek, K. (2008). How toddlers begin to learn
verbs. Trends in Cognitive Sciences, 12(10), 397–403.
Gombert, J.E. (1992). Metalinguistic development (translated by T. Pownall).
New York: Harvester Wheatsheaf.
Goodman, J.C., Dale, P.S. & Li, P. (2008). Does frequency count? Parental
input and the acquisition of vocabulary. Journal of Child Language, 35(3),
515–531.
Gopnik, A. & Meltzoff, A.N. (1986). Words, plans, things, and locations:
Interactions between semantic and cognitive development in the one-word
stage. In S. Kuczaj & M. Barrett (Eds.), The development of word
meaning. New York: Springer-Verlag.
Gordon, P. (1985). Level-ordering in lexical development. Cognition, 21(2),
73–93.
Gordon, P. & Chafetz, J. (1991). Verb-based vs. class-based accounts of
actionality effects in children’s comprehension of the passive. Cognition,
36(3), 227–254.
Gordon, R.G. (Ed.) (2005). Ethnologue: Languages of the world (15th ed.).
Dallas, TX: SIL International.
Graf Estes, K. (2012). Infants generalize representations of statistically
segmented words. Frontiers in Psychology, 3. Published online: 29
October 2012.
Gravetter, F.J. & Walnau, L.B. (1992). Statistics for the behavioral sciences:
A first course for students of psychology and education (3rd ed.). St Paul,
MN: West.
Greenberg, J.H. (1966). Language universals: With special reference to
feature hierarchies. The Hague: Mouton.
Greer, R. & Du, L. (2015). Experience and the onset of the ability to name
objects incidentally by exclusion. The Psychological Record, 65(2),
355–373.
Griffiths, P. (1986). Early vocabulary. In P. Fletcher & M. Garman (Eds.),
Language acquisition: Studies in first language development (2nd ed.) (pp.
279–306). Cambridge: Cambridge University Press.
Grimshaw, G.M., Adelstein, A., Bryden, M.P. & MacKinnon, G.E. (1998).
First-language acquisition in adolescence: Evidence for a critical period for
verbal language development. Brain & Language, 63, 237–255.
Gropen, J., Pinker, S., Hollander, M. & Goldberg, R. (1991). Syntax and
 semantics in the acquisition of locative verbs. Journal of Child Language,
18(1), 115–151.
Haggan, M. (2002). Self-reports and self-delusion regarding the use of
Motherese: Implications from Kuwaiti adults. Language Sciences, 24(1),
17–28.
Hahn, E.R. & Cantrell, L. (2012). The shape-bias in Spanish-speaking
children and its relationship to vocabulary. Journal of Child Language,
39(2), 443–455.
Hakuta, K. (2001). A critical period for second language acquisition? In D.B.
Bailey, J.T. Bruer, F.J. Symons & J.W. Lichtman (Eds.), Critical thinking
about critical periods. Baltimore, MD: Paul Brookes.
Hakuta, K., Bialystok, E. & Wiley, E. (2003). Critical evidence: A test of the
critical-period hypothesis for second-language acquisition. Psychological
Science, 14(1), 31–38.
Hale, B. (1987). Abstract objects. Oxford: Basil Blackwell.
Hall, D.G. & Waxman, S.R. (Eds.) (2004). Weaving a lexicon. Cambridge,
MA: MIT Press.
Hall, D.G., Williams, S. & Bélanger, J. (2015). Learning count nouns and
adjectives: Understanding the contributions of lexical form class and
social-pragmatic cues. Journal of Cognition and Development, 11(1),
86–120.
Hall, S., Rumney, L., Holler, J. & Kidd, E. (2013). Associations among play,
gesture and early spoken language acquisition. First Language, 33(3),
294–312.
Hamilton, M. (1999). Ethnography for classrooms: Constructing a reflective
curriculum for literacy. Pedagogy, Culture and Society, 7(3), 429–444.
Hamlin, J.K., Hallinan, E.V. & Woodward, A.L. (2008). Do as I do: 7-
month-old infants selectively reproduce others’ goals. Developmental
Science, 11(4), 487–494.
Hampton, J.A. (2007). Typicality, graded membership, and vagueness.
Cognitive Science, 31(3), 355–384.
Hanna, J. & Pulvermüller, F. (2014). Neurophysiological evidence for whole
form retrieval of complex derived words: A mismatch negativity study.
Frontiers in Human Neuroscience, 8.
Hao, M., Liu, Y., Shu, H., Xing, A., Jiang, Y. & Li, P. (2015).
Developmental changes in the early child lexicon in Mandarin Chinese.
Journal of Child Language, 42(3), 505–537.
Happé, F. (1994). Autism: An introduction to psychological theory. London:
UCL Press.
Hare, B., Rosati, A., Kaminski, J., Bräuer, J., Call, J. & Tomasello, M.
(2010). The domestication hypothesis for dogs’ skills with human
communication: A response to Udell et al. (2008) and Wynne et al. (2008).
Animal Behavior, 79(2), e1–e6.
Harley, B. & Hart, D. (1997). Language aptitude and second language
proficiency in classroom learning of different starting ages. Studies in
Second Language Acquisition, 19, 379–400.
Harris, L. & Humphreys, G. (2015). Semantic impairment and past tense verb
production: Neighbourhood and frequency modulation of irregular past
tense production. Aphasiology, 29(7), 799–825.
Harris, M., Yeeles, C., Chasin, J. & Oakley, Y. (1995). Symmetries and
asymmetries in early lexical comprehension and production. Journal of
Child Language, 22(1), 1–18.
Hart, B. & Risley, T.R. (1995). Meaningful differences in the everyday
experiences of young American children. Baltimore, MD: Paul Brookes.
Harweth, R.S., Smith III, E.L., Crawford, M.J.L. & van Noorden, G.K.
(1989). The effects of reverse monocular deprivation in monkeys. I.
Psychophysical experiments. Experimental Brain Research, 74, 327–337.
Haskell, T.R., MacDonald, M.C. & Seidenberg, M.S. (2003). Language
learning and innateness: Some implications of Compounds Research.
Cognitive Psychology, 47(2), 119–163.
Hauser, M., Chomsky, N. & Fitch, T. (2002). The faculty of language: What
is it, who has it, and how did it evolve? Science, 298(5598), 1569–1579.
Hauser, M.D., Newport, E.L. & Aslin, R.N. (2001). Segmentation of the
speech stream in a non-human primate: Statistical learning in cotton-top
tamarins. Cognition, 78(3), B53–B64.
Hay, J.F., Pelucchi, B., Graf Estes, K. & Saffran, J.R. (2011). Linking sounds
to meanings: Infant statistical learning in a natural language. Cognitive
Psychology, 63(2), 93–106.
Hayes, K.J. & Hayes, C. (1951). The intellectual development of a home-
raised chimpanzee. Proceedings of the American Philosophical Society,
95(2), 105–109.
Heath, S.B. (1983). Ways with words. Cambridge: Cambridge University
Press.
Heibeck, T.H. & Markman, E.M. (1987). Word learning in children: An
 examination of fast mapping. Child Development, 58, 1021–1034.
Herman, L.M., Kuczaj, S.A. & Holder, M.D. (1993). Responses to
anomalous gestural sequences by a language-trained dolphin: Evidence for
processing of semantic relations and syntactic information. Journal of
Experimental Psychology: General, 122(2), 184–194.
Herman, L.M., Richards, D.G. & Wolz, J.P. (1984). Comprehension of
sentences by bottlenosed dolphins. Cognition, 16, 129–219.
Hernandez, A.E. & Li, P. (2007). Age of acquisition: Its neural and
computational mechanisms. Psychological Bulletin, 133(4), 638–650.
Herrnstein, R.J. & Murray, C. (1994). The bell curve: Intelligence and class
structure in American life. New York: Free Press.
Herschensohn, J.R. (2007). Language development and age. Cambridge:
Cambridge University Press.
Hillenbrand, J.M., Minifie, F.D. & Edwards, T.J. (1979). Tempo of spectrum
change as a cue in speech sound discrimination by infants. Journal of
Speech and Hearing Research, 22, 147–165.
Hills, T. (2013). The company that words keep: Comparing the statistical
structure of child- versus adult-directed language. Journal of Child
Language, 40(3), 586–604.
Hirsh-Pasek, K. & Golinkoff, R.M. (Eds.) (1996). The origins of grammar:
Evidence from early language comprehension. Cambridge, MA: MIT
Press.
Hirsh-Pasek, K. & Golinkoff, R.M. (Eds.) (2006). Action meets word: How
children learn verbs. Oxford: Oxford University Press.
Hirsh-Pasek, K., Golinkoff, R.M. & Naigles, L. (1996). Young children’s use
of syntactic frames to derive meaning. In K. Hirsh-Pasek & R.M.
Golinkoff (Eds.), The origins of grammar: Evidence from early language
comprehension (pp. 123–158). Cambridge, MA: MIT Press.
Hirsh-Pasek, K., Kemler Nelson, D.G., Jusczyk, P.W., Cassidy, K.W., Druss,
B. & Kennedy, L. (1987). Clauses are perceptual units for young infants.
Cognition, 26(3), 269–286.
Hirsh-Pasek, K., Treiman, R. & Schneiderman, M. (1984). Brown & Hanlon
revisited: Mothers’ sensitivity to ungrammatical forms. Journal of Child
Language, 11, 81–88.
Hockett, C.F. (1963). The problem of universals in language. In J.H.
Greenberg (Ed.), Universals of language (pp. 1–29). Cambridge, MA: MIT
Press.
Hockett, C.F. & Altmann, S. (1968). A note on design features. In T. Sebeok
(Ed.), Animal communication: Techniques of study and results of research.
(pp. 61–72). Bloomington, IN: Indiana University Press.
Hoff, E. (2001). Language development (2nd ed.). Belmont, CA: Wadsworth.
Hoff, E. (2004). Progress, but not a full solution to the logical problem of
language acquisition. Journal of Child Language, 31(4), 923–926.
Hoff, E. (2006). How social contexts support and shape language
development. Developmental Review, 26(1), 55–88.
Hoff, E. (2012). Guide to research methods in child language. London:
Blackwell-Wiley.
Hoff, E., Core, C. & Bridges, K. (2008). Non-word repetition assesses
phonological memory and is related to vocabulary development in 20- to
24-month-olds. Journal of Child Language, 35(4), 903–916.
Hoff, E. & Naigles, L. (2002). How children use input to acquire a lexicon.
Child Development, 73(2), 418–433.
Hoff, E. & Shatz, M. (Eds.) (2007). Blackwell handbook of language
development. Oxford: Blackwell.
Hoff-Ginsberg, E. (1985). Some contributions of mothers’ speech to their
children’s syntactic growth. Journal of Child Language, 12(2), 367–385.
Holland, A., Simpson, A. & Riggs, K.J. (2015). Young children retain fast
mapped object labels better than shape, color, and texture words. Journal
of Experimental Child Psychology, 134, 1–11.
Hollich, G. & Golinkoff, R.M. (2007). Young children associate novel words
with complex objects rather than salient parts. Developmental Psychology,
43(5), 1051–1061.
Holtheuer, C. & Rendle-Short, J. (2013). Ser and estar: Corrective input to
children’s errors of the Spanish copula verbs. First Language, 33(2),
155–167.
Horst, J.S. & Samuelson, L.K. (2008). Fast mapping but poor retention by
24-month-old infants. Infancy, 13(2), 128–157.
Huang, B.H. (2014). The effects of age on second language grammar and
speech production. Journal of Psycholinguist Research, 43, 397–420.
Huang, B.H. (2015). A synthesis of empirical research on the linguistic
outcomes of early foreign language instruction. International Journal of
Multilingualism, doi: 10.1080/14790718.2015.1066792.
Huang, Y. (1995). On null subjects and null objects in generative grammar.
Linguistics, 33(6), 1081–1123.
Hubel, D.H. & Wiesel, T.N. (1963). Receptive fields of cells in striate cortex
of very young, visually inexperienced kittens. Journal of Neurophysiology,
26(6), 994–1002.
Hubel, D.H. & Wiesel, T.N. (1970). The period of susceptibility to the
physiological effects of unilateral eye closure in kittens. Journal of
Physiology, 206(2), 419–436.
Huddleston, R. (1984). Introduction to the grammar of English. Cambridge:
Cambridge University Press.
Hunsicker, D. & Goldin-Meadow, S. (2012). Hierarchical structure in a self-
created communication system: Building nominal constituents in
homesign. Language, 88(4), 732–763.
Hupp, J.M. (2008). Demonstration of the shape bias without label extension.
Infant Behavior and Development, 31(3), 511–517.
Hupp, J.M. (2015). Development of the shape bias during the second year.
The Journal of Genetic Psychology, 176(2), 82–92.
Hurry, J., Nunes, T., Bryant, P., Pretzlik, U., Parker, M., Curno, T. &
Midgley, L. (2005). Transforming research on morphology into teacher
practice. Research Papers in Education, 20(2), 187–206.
Hurtado, N., Marchman, V.A. & Fernald, A. (2008). Does input influence
uptake? Links between maternal talk, processing speed and vocabulary
size in Spanish-learning children. Developmental Science, 11(6), F31–F39.
Huttenlocher, J. (1974). The origin of language comprehension. In R.L. Solso
(Ed.), Theories in cognitive psychology: The Loyola Symposium.
Potomac, MD: Lawrence Erlbaum.
Hyams, N. (1986). Language acquisition and the theory of parameters.
Dordrecht: Reidel.
Hyams, N. (1992). A reanalysis of null subjects in child language. In J.
Weissenborn, H. Goodluck & T. Roeper (Eds.), Theoretical issues in
language acquisition: Continuity and change in development. (pp.
249–267). Hillsdale, NJ: Lawrence Erlbaum.
Hyltenstam, K. & Abrahamsson, N. (2000). Who can become native-like in a
second language? All, some, or none? Studia Linguistica, 54(2), 150–166.
Ibbotson, P. & Tomasello, M. (2016). What’s universal grammar? Evidence
rebuts Chomsky’s theory of language learning. [Online]. Available at:
www.salon.com/2016/09/10/what-will-universal-grammar-evidence-
rebuts-chomskys-theory-of-language-learning_partner/. Accessed 17
October 2016.
Imai, M. & Haryu, E. (2004). Word-learning biases and lexical development.
In D.G. Hall & S. Waxman (Eds.), Weaving a lexicon (pp. 411–444).
Cambridge, MA: MIT Press.
Imai, M., Haryu, E. & Okada, H. (2005). Mapping novel nouns and verbs
onto dynamic action events: Are verb meanings easier to learn than noun
meanings for Japanese children. Child Development, 76(2), 340–355.
Imai, M., Li, L.J., Haryu, E., Okada, H., Hirsh-Pasek, K., Golinkoff, R.M. &
Shigematsu, J. (2008). Novel noun and verb learning in Chinese-, English-,
and Japanese-speaking children. Child Development, 79(4), 979–1000.
Immelmann, K. & Suomi, S.J. (1981). Sensitive phases in development. In K.
Immelmann, G.W. Barlow, L. Petrinovich & M. Main (Eds.), Behavioral
development: The Bielefeld interdisciplinary project. (pp. 395–431).
Cambridge: Cambridge University Press.
Ioup, G., Boustagui, E., El Tigi, M. & Moselle, M. (1994). Reexamining the
Critical Period Hypothesis: A case study of successful adult SLA in a
naturalistic environment. Studies in Second Language Acquisition, 16(1),
73–98.
Israel, M., Johnson, C. & Brooks, P.J. (2001). From states to events: The
acquisition of English passive participles. Cognitive Linguistics, 11(1–2),
103–129.
Iverson, J.M. & Goldin-Meadow, S. (2005). Gesture paves the way for
language development. Psychological Science, 16(5), 367–371.
Jackendoff, R. (1993). Patterns in the mind: Language and human nature.
New York: Harvester Wheatsheaf.
Jackendoff, R. (1997). The architecture of the language faculty. Cambridge,
MA: MIT Press.
Jackendoff, R. (2002). Foundations of language: Brain, meaning, grammar,
evolution. Oxford: Oxford University Press.
Jaeger, J.J., Lockwood, A.H., Kemmerer, D.L., Van Valin, R.D., Murphy,
B.W. & Khalak, H.G. (1996). A positron emission tomographic study of
regular and irregular verb morphology in English. Language, 72, 451–497.
Jakobson, R. (1941/1968). Child language, aphasia and phonological
universals. The Hague: Mouton.
Jaswal, V.K. (2010). Explaining the disambiguation effect: Don’t exclude
mutual exclusivity. Journal of Child Language, 37(1), 95–113.
Jespersen, O. (1922). Language: Its nature, development and origin. London:
George Allen and Unwin.
Johnson, C.J., Paivio, A. & Clark, J.M. (1996). Cognitive components of
picture naming. Psychological Bulletin, 120(1), 113–139.
Johnson, E.K. & Jusczyk, P.W. (2001). Word segmentation by 8-month-olds:
When speech cues count more than statistics. Journal of Memory and
Language, 44(4), 548–567.
Johnson, J.S. & Newport, E.L. (1989). Critical period effects in second
language learning: The influence of maturational state on the acquisition of
English as a second language. Cognitive Psychology, 21(1), 60–99.
Johnson, J.S. & Newport, E.L. (1991). Critical period effects on universal
properties of language: The status of subjacency in the acquisition of a
second language. Cognition, 39, 215–258.
Jolly, H.R. & Plunkett, K. (2008). Inflectional bootstrapping in 2-year-olds.
Language and Speech, 51(1&2), 45–59.
Jones, S.S., Smith, L.B. & Landau, B. (1991). Object properties and
knowledge in early lexical learning. Child Development, 62(3), 499–516.
Joyce, J. (1922/1960). Ulysses. Harmondsworth: Penguin.
Juffs, A. & Harrington, M. (1995). Parsing effects in L2 sentence processing:
Subject and object asymmetries in WH-extraction. Studies in Second
Language Acquisition, 17(4), 483–516.
Jusczyk, P.W. (1989). Perception of cues to clausal units in native and non-
native languages. Paper presented at the biennial meeting of the Society for
Research in Child Development, Kansas City.
Jusczyk, P.W. (1997). The discovery of spoken language. Cambridge, MA:
MIT Press.
Jusczyk, P.W. (2002). How infants adapt speech-processing capacities to
native language structure. Current Directions in Psychological Science,
11(1), 15–18.
Jusczyk, P.W. & Aslin, R.N. (1995). Infants’ detection of sound patterns of
words in fluent speech. Cognitive Psychology, 29(1), 1–23.
Jusczyk, P.W., Copan, H. & Thompson, E. (1978). Perception by two-month-
olds of glide contrasts in multisyllabic utterances. Perception and
Psychophysics, 24(6), 515–520.
Jusczyk, P.W., Houston, D. & Newsome, M. (1999). The beginnings of word
segmentation in English-learning infants. Cognitive Psychology, 39(3–4),
159–207.
Kaduk, K., Bakker, M., Juvrud, J., Gredebäck, G., Westermann, G., Lunn, J.
& Reid, V.M. (2016). Semantic processing of actions at 9 months is linked
 to language proficiency at 9 and 18 months. Journal of Experimental Child
Psychology, 151, 96–108.
Kail, R. (1990). The development of memory in children (3rd ed.). New
York: W.H. Freeman.
Kaminski, J., Call, J. & Fischer, J. (2004). Word learning in a domestic dog:
Evidence for ‘fast mapping’. Science, 304(5677), 1682–1683.
Kaminski, J., Shulz, L. & Tomasello, M. (2012). How dogs know when
communication is intended for them. Developmental Science, 15(2),
222–232.
Karlsson, F. & Koskenniemi, K. (1985). A process model of morphology and
lexicon. Folia Linguistica, 29, 207–231.
Karmiloff, K. & Karmiloff-Smith, A. (2001). Pathways to language: From
fetus to adolescent. Cambridge, MA: Harvard University Press.
Kaschak, M.P., Loney, R.A. & Borreggine, K.L. (2006). Recent experience
affects the strength of structural priming. Cognition, 99(3), B73–B82.
Kauschke, C. & Hofmeister, C. (2002). Early lexical development in
German: A study on vocabulary growth and vocabulary composition
during the second and third year of life. Journal of Child Language, 29(4),
735–757.
Kellog, W.N. & Kellog, L.A. (1933). The ape and the child: A study of
environmental influence upon early behavior. New York: McGraw-Hill.
Kemler Nelson, D.G., Frankenfield, A., Morris, C. & Blair, E. (2000). Young
children’s use of functional information to categorize artifacts: Three
factors that matter. Cognition, 77(2), 133–168.
Kemler Nelson, D.G., Hirsh-Pasek, K., Jusczyk, P.W. & Cassidy, K.W.
(1989). How the prosodic cues in motherese might assist language
learning. Journal of Child Language, 16(1), 55–68.
Kempe, V., Brooks, P.J., Mironova, N., Pershukova, A. & Fedorova, O.
(2007). Playing with word endings: Morphological variation in the
learning of Russian noun inflections. British Journal of Developmental
Psychology, 25(1), 55–77.
Kessel, F.S. (Ed.) (1988). The development of language and language
researchers: Essays in honor of Roger Brown. Hillsdale, NJ: Lawrence
Erlbaum.
Ketrez, F.N. (2014). Harmonic cues for speech segmentation: A cross-
linguistic corpus study on child-directed speech. Journal of Child
Language, 41(2), 439–461.
Key, A.P.F., Lambert, E.W., Aschner, J.L. & Maitre, N.L. (2012). Influence
of gestational age and postnatal age on speech sound processing in NICU
infants. Psychophysiology, 49(5), 720–731.
King, S.L. & Janik, V.M. (2013). Bottlenose dolphins can use learned vocal
labels to address each other. Proceedings of the National Academy of
Sciences of the United States of America, 110(32), 13216–13221.
Kintsch, W. (1981). Semantic memory: A tutorial. In R.S. Nickerson (Ed.),
Attention and Performance (Vol. VIII). Hillsdale, NJ: Lawrence Erlbaum.
Kiparsky, P. (1983). Word-formation and the lexicon. In F. Ingemann (Ed.),
Proceedings of the 1981 Mid-America linguistics conference (pp. 3–29).
Lawrence, KS: University of Kansas Press.
Kirjavainen, M., Nikolaev, A. & Kidd, E. (2012). The effect of frequency and
phonological neighbourhood density on the acquisition of past tense verbs
by Finnish children. Cognitive Linguistics, 23(2), 273–315.
Kisilevsky, B.S., Hains, S.M.J., Lee, K., Xie, X., Huang, H.F., Ye, H.H.,
Zhang, K. & Wang, Z.P. (2003). Effects of experience on fetal voice
recognition. Psychological Science, 14(3), 220–224.
Kitamura, C. & Lam, C. (2009). Age-specific preferences for infant-directed
affective intent. Infancy, 14(1), 77–100.
Klatt, D.H. & Stefanski, R.A. (1974). How does a mynah bird imitate human
speech? The Journal of the Acoustical Society of America, 55(4), 822–832.
Kline, M. & Demuth, K. (2014). Syntactic generalization with novel
intransitive verbs. Journal of Child Language, 41(3), 543–574.
Kluender, K. (1991). Effects of first formant onset properties on voicing
judgments result from processes not specific to humans. Journal of the
Acoustical Society of America, 90(1), 83–96.
Ko, E.-S., Seidl, A., Cristia, A., Reimchen, M. & Soderstrom, M. (2016).
Entrainment of prosody in the interaction of mothers with their young
children. Journal of Child Language, 43(2), 284–309.
Koryé, A. (1968). Metaphysics and measurement: Essays in scientific
revolution. London: Chapman and Hall.
Krcmar, M., Grela, B. & Lin, K. (2007). Can toddlers learn vocabulary from
television? An experimental approach. Media Psychology, 10(1), 41–63.
Krott, A., Gagné, C.L. & Nicoladis, E. (2009). How the parts relate to the
whole: Frequency effects on children’s interpretations of novel
compounds. Journal of Child Language, 36(1), 85–112.
Krott, A. & Nicoladis, E. (2005). Large constituent families help children
 parse compounds. Journal of Child Language, 32(1), 139–158.
Kučera, H. (1992). The mathematics of language. In The American heritage
dictionary of the English language (3rd ed.) (pp. xxxi–xxxiii). Boston,
MA: Houghton Mifflin.
Kučera, H. & Francis, W.M. (1967). Computational analysis of present day
American English. Providence, RI: Brown University Press.
Kuczaj, S.A. (1982). On the nature of syntactic development. In S.A. Kuczaj
(Ed.), Language development. Volume 1: Syntax and semantics (pp.
37–71). Hillsdale, NJ: Lawrence Erlbaum.
Kuhl, P.K. (2004). Early language acquisition: Cracking the speech code.
Nature Reviews Neuroscience, 5(11), 831–843.
Kuhl, P.K., Conboy, B.T., Coffey-Corina, S., Padden, D., Rivera-Gaxiola, M.
& Nelson, T. (2008). Phonetic learning as a pathway to language: New
data and native language magnet theory expanded (NLM-e). Philosophical
Transactions of the Royal Society B – Biological Sciences, 363(1493),
979–1000.
Kuhl, P.K., Conboy, B.T., Padden, D., Nelson, T. & Pruitt, J. (2005). Early
speech perception and later language development: Implications for the
‘critical period’. Language Learning and Development, 1, 237–264.
Kuhl, P.K. & Miller, J.D. (1975). Speech perception by the chinchilla:
Voiced–voiceless distinction in alveolar plosive consonants. Science, 190,
69–72.
Kuhl, P.K. & Padden, D.M. (1983). Enhanced discriminability at the phonetic
boundaries for the place feature in macaques. Journal of the Acoustical
Society of America, 73, 1003–1010.
Kuhl, P.K. & Rivera-Gaxiola, M. (2008). Neural substrates of language
acquisition. Annual Review of Neuroscience, 31(1), 511–534.
Kuhl, P.K., Stevens, E., Hayashi, A., Deguchi, T., Kiritani, S. & Iverson, P.
(2006). Infants show a facilitation effect for native language phonetic
perception between 6 and 12 months. Developmental Science, 9(2),
F13–F21.
Kuhl, P.K., Tsao, F. & Liu, H. (2003). Foreign language experience in
infancy: Effects of short term exposure and interaction on phonetic
learning. Proceedings of the National Academy of Sciences, 100,
9096–9101.
Laakso, M. & Soininen, M. (2010). Mother-initiated repair sequences in
interactions of 3-year-old children. First Language, 30(3–4), 329–353.
Ladefoged, P. (2004). Vowels and consonants: An introduction to the sounds
of language (2nd ed.). Oxford: Blackwell.
Ladefoged, P. (2006). A course in phonetics (5th ed.). Boston, MA: Thomas
Wadsworth.
Landau, B., Smith, L.B. & Jones, S.S. (1988). The importance of shape in
early lexical learning. Cognitive Development, 3(3), 299–321.
Lane, H. (1976). The wild boy of Aveyron. Cambridge, MA: Harvard
University Press.
Langacker, R.W. (1987). Foundations of cognitive grammar (Vol. 1):
Theoretical prerequisites. Stanford, CA: Stanford University Press.
Lanter, J.A. & Basche, R.A. (2014). Effect of number and similarity on
children’s plural comprehension. First Language, 34(6), 519–536.
Larsen-Freeman, D. (2015). Saying what we mean: Making a case for
‘language acquisition’ to become ‘language development’. Language
Teaching, 48(4), 491–505.
Lasnik, H. (2000). Syntactic Structures revisited: Contemporary lectures on
transformational theory. Cambridge, MA: MIT Press.
Lee, D. (1992). Universal Grammar, learnability, and the acquisition of
English reflexive binding by L1 Korean speakers. Unpublished doctoral
dissertation, University of Southern California, Los Angeles.
Lehrman, D.S. (1970). Semantic and conceptual issues in the nature–nurture
problem. In L. Aronson, E. Tobech, D.S. Lehrman & J. Rosenblatt (Eds.),
Development and evolution of behavior: Essays in memory of T.C.
Schneirla. San Francisco, CA: W.H. Freeman.
Lenneberg, E.H. (1967). Biological foundations of language. New York:
Wiley.
Leopold, W.F. (1939). Speech development of a bilingual child: A linguist’s
record. Vol. I: Vocabulary growth in the first two years. Evanston, IL:
Northwestern University Press.
Leopold, W.F. (1949). Speech development of a bilingual child: A linguist’s
record. Vol. III: Grammar and general problems in the first two years.
Evanston, IL: Northwestern University Press.
Leopold, W.F. (1952). Bibliography of child language. Evanston, IL:
Northwestern University Press.
Levin, I., Ravid, D. & Rapaport, S. (2001). Morphology and spelling among
Hebrew-speaking children. Journal of Child Language, 28(3), 741–769.
Levitt, A., Jusczyk, P.W., Murray, J. & Carden, G. (1988). The perception of
 place of articulation contrasts in voiced and voiceless fricatives by two-
month-old infants. Journal of Experimental Psychology: Human
Perception and Performance, 14, 361–368.
Lew-Williams, C., Pelucchi, B. & Saffran, J.R. (2011). Isolated words
enhance statistical language learning in infancy. Developmental Science,
14(6), 1323–1329.
Lew-Williams, C. & Saffran, J.R. (2012). All words are not created equal:
Expectations about word length guide infant statistical learning. Cognition,
122(2), 241–246.
Lewis, J.D. & Elman, J.L. (2001). Learnability and the statistical structure of
language: Poverty of stimulus arguments revisited. In Proceedings of the
26th annual Boston University Conference on language development (pp.
359–370). Somerville, MA: Cascadilla.
Lidz, J. & Gagliardi, A. (2015). How nature meets nurture: Statistical
learning and Universal Grammar. Annual Review of Linguistics, 11(1),
333–353.
Lidz, J. & Waxman, S. (2004). Reaffirming the poverty of the stimulus
argument: A reply to the replies. Cognition, 93(2), 157–165.
Lidz, J., Waxman, S. & Freedman, J. (2003). What infants know about syntax
but couldn’t have learned: Experimental evidence for syntactic structure at
18 months. Cognition, 89(3), B65–B73.
Liebal, K., Behne, T., Carpenter, M. & Tomasello, M. (2009). Infants use
shared experience to interpret pointing gestures. Developmental Science,
12(2), 264–271.
Lieberman, A.M., Borovsky, A., Hatrak, M. & Mayberry, R.I. (2015). Real-
time processing of ASL signs: Delayed first language acquisition affects
organization of the mental lexicon. Journal of Experimental Psychology:
Learning, Memory, and Cognition, 41(4), 1130–1139.
Lieberman, P. (1993). An untrustworthy guide to ape-language research.
Current Anthropology, 34(3), 327–328.
Lieven, E.V.M. (1994). Crosslinguistic and crosscultural aspects of language
addressed to children. In C. Gallaway & B. Richards (Eds.), Input and
interaction in language acquisition (pp. 56–73). Cambridge: Cambridge
University Press.
Lieven, E.V.M. (2016). Usage-based approaches to language development:
Where do we go from here? Language and Cognition, 8(3), 346–368.
Lieven, E.V.M., Behrens, H., Speares, J. & Tomasello, M. (2003). Early
 syntactic creativity: A usage-based approach. Journal of Child Language,
30(2), 333–370.
Lieven, E.V.M. & Tomasello, M. (2008). Children’s first language
acquisition from a usage-based perspective. In P. Robinson & N.C. Ellis
(Eds.), Handbook of cognitive linguistics and second language acquisition
(pp. 168–196). New York: Routledge.
Lightfoot, D. (1989). The child’s trigger experience: Degree-0 learnability.
Behavioral and Brain Sciences, 12, 321–375.
Lightfoot, D. (2005). Plato’s problem, UG, and the language organ. In J.
McGilvray (Ed.), The Cambridge companion to Chomsky (pp. 42–59).
Cambridge: Cambridge University Press.
Linebarger, D.L. & Walker, D. (2005). Infants’ and toddlers’ television
viewing and language outcomes. American Behavioral Scientist, 48(5),
624–645.
Long, M.H. (2005). Problems with supposed counter-evidence to the critical
period hypothesis. International Review of Applied Linguistics, 43(4),
287–317.
Lukens, H.T. (1894). Preliminary report on the learning of language.
Pedagogical Seminary, 3, 424–460.
Lust, B. (2006). Child language: Acquisition and growth. Cambridge:
Cambridge University Press.
Lyn, H. & Savage-Rumbaugh, E.S. (2000). Observational word learning in
two bonobos (Pan paniscus): Ostensive and non-ostensive contexts.
Language & Communication, 20(3), 255–273.
Lyytinen, P. & Lyytinen, H. (2004). Growth and predictive relations of
vocabulary and inflectional morphology in children with and without
familial risk for dyslexia. Applied Psycholinguistics, 25(3), 397–411.
Mackintosh, N.J. (1998). IQ and human intelligence. Oxford: Oxford
University Press.
MacWhinney, B. (1985). Hungarian language acquisition as an
exemplification of a general model of grammatical development. In D.I.
Slobin (Ed.), The crosslinguistic study of language acquisition (Vol. II)
(pp. 1069–1155). Hillsdale, NJ: Lawrence Erlbaum.
MacWhinney, B. (2000). The CHILDES project: Tools for analyzing talk.
Vol. 2: The database (3rd ed.). Mahwah, NJ: Lawrence Erlbaum.
MacWhinney, B. (2004). A multiple process solution to the logical problem
of language acquisition. Journal of Child Language, 31(4), 883–914.
MacWhinney, B. (2014a). Item-based patterns in early syntactic
development. In T. Herbst, H.-J. Schmid & S. Faulhaber (Eds.), Valency
relations (pp. 33–70). Berlin: Springer.
MacWhinney, B. (2014b). What we have learned. Journal of Child Language,
41(S1), 124–131.
MacWhinney, B. & O’Grady, W. (Eds.) (2015). The handbook of language
emergence. New York: Wiley.
Magen, H.S. (2014). A behavioral study of regularity, irregularity and rules
in the English past tense. Journal of Psycholinguistic Research, 43(6),
791–814.
Mameli, M. & Bateson, P. (2006). Innateness and the sciences. Biology &
Philosophy, 21(2), 155–188.
Mandalaywala, T., Fleener, C. & Maestripieri, D. (2015). Intelligence in
nonhuman primates. In S. Goldstein, D. Princiotta & J.A. Naglieri (Eds.),
Handbook of intelligence: Evolutionary theory, historical perspective, and
current concepts (pp. 27–46). Berlin: Springer.
Mandel, D.R., Kemler Nelson, D.G. & Jusczyk, P.W. (1996). Infants
remember the order of words in a spoken sentence. Cognitive
Development, 11(2), 181–196.
Männel, C. (2008). The method of event-related brain potentials in the study
of cognitive processes: A tutorial. In A.D. Friederici & G. Thierry (Eds.),
Early language development: Bridging brain and behaviour (pp. 1–22).
Amsterdam: John Benjamins.
Manolio, T.A., Collins, F.S., Cox, N.J., Goldstein, D.B., Hindorff, L.A.,
Hunter, D.J. & Visscher, P. M. (2009). Finding the missing heritability of
complex diseases. Nature, 461(7265), 747–753.
Maratsos, M. (1982). The child’s construction of grammatical categories. In
E. Wanner & L.R. Gleitman (Eds.), Language acquisition: The state of the
art (pp. 240–265). Cambridge: Cambridge University Press.
Maratsos, M. (1999). Some aspects of innateness and complexity in
grammatical acquisition. In M. Barrett (Ed.), The development of language
(pp. 191–228). Hove: Psychology Press.
Maratsos, M. (2000). More overregularisations after all: New data and
discussion on Marcus, Pinker, Ullman, Hollander, Rosen & Xu. Journal of
Child Language, 27(1), 183–212.
Maratsos, M., Gudeman, R., Gerard-Ngo, P. & DeHart, G. (1987). A study in
novel word learning: The productivity of the causative. In B. MacWhinney
 (Ed.), Mechanisms of language acquisition (pp. 89–113). Hillsdale, NJ:
Lawrence Erlbaum.
Marchetto, E. & Bonatti, L.L. (2015). Finding words and word structure in
artificial speech: The development of infants’ sensitivity to
morphosyntactic regularities. Journal of Child Language, 42(4), 873–902.
Marcus, G.F. (1993). Negative evidence in language acquisition. Cognition,
46, 53–85.
Marcus, G.F. (1995). Children’s overregularization of English plurals: A
quantitative analysis. Journal of Child Language, 22(2), 447–460.
Marcus, G.F. (1999). Language acquisition in the absence of explicit negative
evidence: Can simple recurrent networks obviate the need for domain-
specific learning devices? Cognition, 73(3), 293–296.
Marcus, G.F. (2001). The algebraic mind: Integrating connectionism and
cognitive science. Cambridge, MA: MIT Press.
Marcus, G.F., Fernandes, K.J. & Johnson, S.P. (2007). Infant rule learning
facilitated by speech. Psychological Science, 18(5), 387–391.
Marcus, G.F., Pinker, S., Ullman, M., Hollander, M., Rosen, T.J. & Xu, F.
(1992). Overregularization in language acquisition. Monographs of the
Society for Research in Child Development, Serial no. 228.
Marcus, G.F., Vijayan, S., Rao, S.B. & Vishton, P.M. (1999). Rule learning
by seven-month-old infants. Science, 283(5398), 77–80.
Markman, E.M. (1990). Constraints children place on word meanings.
Cognitive Science, 14(1), 57–77.
Markman, E.M. & Abelev, M. (2004). Word learning in dogs? Trends in
Cognitive Sciences, 8(11), 479–481.
Markman, E.M. & Hutchinson, J.E. (1984). Children’s sensitivity to
constraints on word meaning: Taxonomic versus thematic relations.
Cognitive Psychology, 16(1), 1–27.
Markman, E.M. & Wachtel, G.F. (1988). Children’s use of mutual
exclusivity to constrain the meanings of words. Cognitive Psychology,
20(2), 121–157.
Marsden, R.E. (1902). Record from a parent note-book. Paidologist, 4,
97–100.
Marsh, H.L., Vining, A.Q., Levendoski, E.K., Judge, P.G. & Call, J. (2015).
Inference by exclusion in lion-tailed macaques (Macaca silenus), a
hamadryas baboon (Papio hamadryas), capuchins (Sapajus apella), and
squirrel monkeys (Saimiri sciureus). Journal of Comparative Psychology,
 129(3), 256–267.
Marshall, P.J. & Meltzoff, A.N. (2014). Neural mirroring mechanisms and
imitation in human infants. Philosophical Transactions of the Royal
Society of London. Series B, Biological Sciences, 369(1644), 20130620.
Martin, N., Weisberg, R.W. & Saffran, E.M. (1989). Variables influencing
the occurrence of naming errors: Implications for models of lexical
retrieval. Journal of Memory and Language, 28(4), 462–485.
Marx, O. (1964). The history of the biological basis of language. In E.H.
Lenneberg (Ed.), New directions in the study of language (pp. 443–469).
Cambridge, MA: MIT Press.
Maslen, R.J.C., Theakston, A.L., Lieven, E.V.M. & Tomasello, M. (2004). A
dense corpus study of past tense and plural overregularization in English.
Journal of Speech, Language, and Hearing Research, 47(6), 1319–1333.
Masur, E.F. & Flynn, V. (2008). Infant and mother-infant play and the
presence of the television. Journal of Applied Developmental Psychology,
29(1), 76–83.
Masur, E.F., Flynn, V. & Olson, J. (2016). Infants’ background television
exposure during play: Negative relations to the quantity and quality of
mothers’ speech and infants’ vocabulary acquisition. First Language,
36(2), 109–123.
Masur, E.F. & Olson, J. (2008). Mothers’ and infants’ responses to their
partners’ spontaneous action and vocal/verbal imitation. Infant Behavior &
Development, 31(4), 704–715.
Matsuo, A., Kita, S., Shinya, Y., Wood, G.C. & Naigles, L. (2012). Japanese
two-year-olds use morphosyntax to learn novel verb meanings. Journal of
Child Language, 39(3), 637–663.
Matthews, D. (Ed.) (2014). Pragmatic development in first language
acquisition. Amsterdam: John Benjamins.
Matthews, P.H. (2007). Syntactic relations: A critical survey. Cambridge:
Cambridge University Press.
May, L., Byers-Heinlein, K., Gervain, J. & Werker, J.F. (2011). Language
and the newborn brain: Does prenatal language experience shape the
neonate neural response to speech? Frontiers in Psychology, 2. Published
online: 21 September 2011.
Mayberry, R.I. & Eichen, E.B. (1991). The long-lasting advantage of learning
sign language in childhood: Another look at the critical period for language
acquisition. Journal of Memory & Language, 30, 486–512.
Maye, J., Weiss, D.J. & Aslin, R.N. (2008). Statistical phonetic learning in
infants: Facilitation and feature generalization. Developmental Science,
11(1), 122–134.
Maye, J., Werker, J.F. & Gerken, L. (2002). Infant sensitivity to
distributional information can affect phonetic discrimination. Cognition,
82(3), B101–B111.
McBride-Chang, C., Tardif, T., Cho, J.R., Shu, H., Fletcher, P., Stokes, S.F.,
Wong, A. & Leung, K. (2008). What’s in a word? Morphological
awareness and vocabulary knowledge in three languages. Applied
Psycholinguistics, 29(3), 437–462.
McClelland, J.L. & Patterson, K. (2002). Rules or connections in past-tense
inflections: What does the evidence rule out? Trends in Cognitive
Sciences, 6(11), 465–472.
McComb, K., Shannon, G., Sayialel, K.N. & Moss, C. (2014). Elephants can
determine ethnicity, gender and age from acoustic cues in human voices.
Proceedings of the National Academy of Sciences of the United States,
111(114), 5433–5438.
McDonough, L. (2002). Basic-level nouns: First learned but misunderstood.
Journal of Child Language, 29(2), 357–377.
McEwen, F., Happé, F., Bolton, P., Rijsdijk, F., Ronald, A., Dworzynski, K.
& Plomin, R. (2007). Origins of individual differences in imitation: Links
with language, pretend play, and socially insightful behavior in two-year-
old twins. Child Development, 78(2), 474–492.
McGilvray, J. (Ed.) (2005). The Cambridge companion to Chomsky.
Cambridge: Cambridge University Press.
McGilvray, J. (2006). On the innateness of language. In R. Stainton (Ed.),
Contemporary debates in cognitive science (pp. 97–112). Oxford:
Blackwell.
McMurray, B. & Aslin, R.N. (2005). Infants are sensitive to within-category
variation in speech perception. Cognition, 95(2), B15–B26.
Mead, M. (1930). Growing up in New Guinea: A study of adolescence and
sex in primitive societies. Harmondsworth: Penguin.
Meadow, K.P. (1980). Deafness and child development. London: Edward
Arnold.
Medawar, P.B. (1967). The art of the soluble. London: Methuen.
Mehler, J., Bertoncini, J., Barrière, M. & Jassik-Gerschenfeld, D. (1978).
Infant recognition of mother’s voice. Perception, 7, 491–497.
Meltzoff, A.N. (1988). Infant imitation and memory: Nine-month-olds in
immediate and deferred tests. Child Development, 59, 217–225.
Meltzoff, A.N. (2002). Elements of a developmental theory of imitation. In
A.N. Meltzoff & W. Prinz (Eds.), The imitative mind: Development,
evolution, and brain bases (pp. 19–41). Cambridge: Cambridge University
Press.
Meltzoff, A.N. & Moore, M.K. (1983). Newborn infants imitate adult facial
gestures. Child Development, 54, 702–709.
Meltzoff, A.N. & Moore, M.K. (1997). Explaining facial imitation: A
theoretical model. Early Development and Parenting, 6(3–4), 179–192.
Menn, L. & Stoel-Gammon, C. (2005). Phonological development: Learning
sounds and sound patterns. In J. Berko Gleason (Ed.), The development of
language (6th ed.) (pp. 62–111). Boston, MA: Pearson.
Mervis, C.B. (1987). Child-basic object categories and early lexical
development. In U. Neisser (Ed.), Concepts and conceptual development:
Ecological and intellectual factors in categorization (pp. 201–233). New
York: Cambridge University Press.
Mervis, C.B. & Bertrand, J. (1995). Acquisition of the novel name-nameless
category. Child Development, 65, 1646–1663.
Mervis, C.B., Catlin, J. & Rosch, E. (1976). Relationships amongst goodness-
of-example, category norms, and word frequency. Bulletin of the
Psychonomic Society, 7(3), 283–284.
Messer, D.J. (1981). The identification of names in maternal speech to
infants. Journal of Psycholinguistic Research, 10(1), 69–77.
Miles, L.H. (1983). Apes and language: The search for communicative
competence. In J. de Luce & H.T. Wilder (Eds.), Language in primates:
Perspectives and implications. New York: Springer.
Miller, K.L. & Schmitt, C. (2012). Variable input and the acquisition of
plural morphology, Language Acquisition, 19(3), 223–261.
Minagawa-Kawai, Y., Mori, K., Hebden, J.C. & Dupoux, E. (2008). Optical
imaging of infants’ neurocognitive development: Recent advances and
perspectives. Developmental Neurobiology, 68(6), 712–728.
Mintz, T.H. (2005). Linguistic and conceptual influences on adjective
acquisition in 24- and 36-month-olds. Developmental Psychology, 41(1),
17–29.
Moerk, E.L. (1991). Positive evidence for negative evidence. First Language,
11(2), 219–251.
Moerk, E.L. & Moerk, C. (1979). Quotations, imitations and generalizations:
Factual and methodological analyses. International Journal of Behavioral
Development, 2(1), 43–72.
Moon, C., Cooper, R.P. & Fifer, W.P. (1993). Two-day-olds prefer their
native language. Infant Behavior and Development, 16, 495–500.
Morgan, J.L., Bonamo, K.M. & Travis, L.L. (1995). Negative evidence on
negative evidence. Developmental Psychology, 31(2), 180–197.
Morgan, J.L. & Travis, L.L. (1989). Limits on negative information in
language input. Journal of Child Language, 16, 531–552.
Morrison, C.M. & Ellis, A.W. (2000). Real age of acquisition effects in word
naming and lexical decision. British Journal of Psychology, 91, 167–180.
Mueller, K.L., Murray, J.C., Michaelson, J.J., Christiansen, M.H., Reilly, S.
& Tomblin, J.B. (2016). Common genetic variants in FOXP2 are not
associated with individual differences in language development. PLoS
One, 11(4), pp.e0152576.
Mulcaster, R. (1582). The first part of the Elementarie, which entreateth
chefelie of the right writing of our English tung. London: T. Vautraillier.
Mumme, D.L. & Fernald, A. (2003). The infant as onlooker: Learning from
emotional reactions observed in a television scenario. Child Development,
74, 221–237.
Muncer, S.J. & Ettlinger, G. (1981). Communication by a chimpanzee: First-
trial mastery of word order that is critical for meaning, but failure to negate
conjunctions. Neuropsychologia, 19(1), 73–78.
Murchison, C. & Langer, S. (1927). Tiedemann’s observations on the
development of the mental faculties of children. Pedagogical Seminary and
Journal of Genetic Psychology, 34, 205–230.
Murphy, R.A., Mondragon, E. & Murphy, V.A. (2008). Rule learning by rats.
Science, 319(5871), 1849–1851.
Murphy, V.A. (2000). Compounding and the representation of L2 inflectional
morphology. Language Learning, 50(1), 153–197.
Murphy, V.A. (2010). The relationship between age of learning and exposure
to the second language in L2 children. In E. Macaro (Ed.), Continuum
companion to second language acquisition. London: Continuum.
Nagy, E., Pal, A. & Orvos, H. (2014). Learning to imitate individual finger
movements by the human neonate. Developmental Science, 17(6),
841–857.
Nagy, W.E. & Anderson, R.C. (1984). How many words are there in printed
 school English? Reading Research Quarterly, 19(3), 304–330.
Naigles, L. (1990). Children use syntax to learn verb meanings. Journal of
Child Language, 17(2), 357–374.
Naigles, L.R. (1996). The use of multiple frames in verb learning via
syntactic bootstrapping. Cognition, 58(2), 221–251.
Naigles, L.G. & Gelman, S. (1995). Overextensions in comprehension and
production revisited: Preferential-looking in a study of dog, cat, and cow.
Journal of Child Language, 22(1), 19–46.
Naigles, L.G. & Kako, E.T. (1993). First contact in verb acquisition:
Defining a role for syntax. Child Development, 64(6), 1665–1687.
Naigles, L.R. & Maltempo, A. (2011). Verb argument structure acquisition in
young children: Defining a role for discourse. Journal of Child Language,
38(3), 662–674.
Naigles, L.R. & Mayeux, L. (2000). Television as an incidental teacher. In D.
Singer & J. Singer (Eds.), Handbook of children and the media (pp.
135–153). New York: Sage.
Naigles, L.R. & Swensen, L.D. (2007). Syntactic supports for word learning.
In E. Hoff & M. Shatz (Eds.), Blackwell handbook of language
development (pp. 212–231). Malden, MA: Blackwell.
Nakayama, M. (1987). Performance factors in subject-auxiliary inversion by
children. Journal of Child Language, 14(1), 113–125.
Napoli, D.J. (1993). Syntax: Theory and problems. Oxford: Oxford
University Press.
Nash, H. & Snowling, M. (2006). Teaching new words to children with poor
existing vocabulary knowledge: A controlled evaluation of the definition
and context methods. International Journal of Language and
Communication Disorders, 41(3), 335–354.
Nazzi, T., Bertoncini, J. & Mehler, J. (1998). Language discrimination by
newborns: Towards an understanding of the role of rhythm. Journal of
Experimental Psychology: Human Perception and Performance, 24,
756–766.
Nazzi, T., Dilley, L.C., Jusczyk, A.M., Shattuck-Hufnagel, S. & Jusczyk,
P.W. (2005). English-learning infants’ segmentation of verbs from fluent
speech. Language and Speech, 48, 279–298.
Nazzi, T., Jusczyk, P.W. & Johnson, E.K. (2000). Language discrimination
by English-learning 5-month-olds: Effects of rhythm and familiarity.
Journal of Memory and Language, 43, 1–19.
Neiworth, J.J. (2013). Chasing sounds. Behavioural Processes, 93, 111–115.
Nelson, K.E., Denninger, M.M., Bonvillian, J.D., Kaplan, B.J. & Baker, N.
(1984). Maternal input adjustments and non-adjustments as related to
children’s linguistic advances and to language acquisition theories. In A.D.
Pellegrini & T.D. Yawkey (Eds.), The development of oral and written
language in social contexts. Norwood, NJ: Ablex.
Nemeth, D., Janacsek, K., Turi, Z., Lukacs, A., Peckham, D., Szanka, S.,
Gazso, D., Lovassy, N. & Ullman, M.T. (2015). The production of
nominal and verbal inflection in an agglutinative language: Evidence from
Hungarian. PLoS One, 10(3), e0119003.
Neville, H.J. & Bruer, J.T. (2001). Language processing: How experience
affects brain organization. In D.B. Bailey, J.T. Bruer, F.J. Symons & J.W.
Lichtman (Eds.), Critical thinking about critical periods (pp. 173–192).
Baltimore, MD: Paul Brookes.
Neville, H.J., Coffey, S.A., Lawson, D.S., Fischer, A., Emmorey, K. &
Bellugi, U. (1997). Neural systems mediating American Sign Language:
Effects of sensory experience & age of acquisition. Brain & Language, 57,
285–308.
Newbury, D.F., Bishop, D.V.M. & Monaco, A.P. (2005). Genetic influences
on language impairment and phonological short-term memory. Trends in
Cognitive Sciences, 9(11), 528–534.
Newman, A.J., Bavelier, D., Corina, D., Jezzard, P. & Neville, H.J. (2002). A
critical period for right hemisphere recruitment in American Sign
Language processing. Nature Neuroscience, 5(1), 76–80.
Newman, R., Ratner, N.B., Jusczyk, A.M., Jusczyk, P.W. & Dow, K.A.
(2006). Infants’ early ability to segment the conversational speech signal
predicts later language development: A retrospective analysis.
Developmental Psychology, 42(4), 643–655.
Newman, R., Rowe, M.L. & Ratner, N.B. (2015). Input and uptake at 7
months predicts toddler vocabulary: The role of child-directed-speech and
infant processing skills in language development. Journal of Child
Language, 43(5), 1158–1173.
Newport, E.L. (1975) Motherese: The speech of mothers to young children.
Unpublished doctoral dissertation, University of Pennsylvania,
Philadelphia, PA.
Newport, E.L. (1988). Constraints on learning and their role in language
acquisition: Studies of the acquisition of American Sign Language.
 Language Sciences, 10, 147–172.
Newport, E.L. (1990). Maturational constraints on language learning.
Cognitive Science, 14, 11–28.
Newport, E.L., Bavelier, D. & Neville, H.J. (2001). Critical thinking about
critical periods: Perspectives on a critical period for language acquisition.
In E. Dupoux (Ed.), Language, brain and cognitive development: Essays in
honor of Jacques Mehler (pp. 481–502). Cambridge, MA: MIT Press.
Newport, E.L., Gleitman, H. & Gleitman, L.R. (1977). Mother, I’d rather do
it myself: Some effects and non-effects of maternal speech style. In C.E.
Snow & C.A. Ferguson (Eds.), Talking to children (pp. 101–149).
Cambridge: Cambridge University Press.
Newport, E.L., Hauser, M.D., Spaepen, G. & Aslin, R.N. (2004). Learning at
a distance II: Statistical learning of non-adjacent dependencies in a non-
human primate. Cognitive Psychology, 49(2), 85–117.
Nicoladis, E. (2003). What compound nouns mean to preschool children.
Brain and Language, 84(1), 38–49.
Nicoladis, E. & Murphy, V.A. (2004). Level-ordering does not constrain
children’s ungrammatical compounds. Brain and Language, 90(1–3),
487–494.
Nicoladis, E. & Paradis, J. (2012). Acquiring regular and irregular past tense
morphemes in English and French: Evidence from bilingual children.
Language Learning, 62(1), 170–197.
Ninio, A. (1986). The direct mapping of function to form in children’s early
language. Journal of Psycholinguistic Research, 15, 559 (Abstract).
Ninio, A. (2004). Young children’s difficulty with adjectives modifying
nouns. Journal of Child Language, 31(2), 255–285.
Nippold, M.A. (2016). Later language development: The school-age and
adolescent years (4th ed.). Austin, TX: Pro-Ed.
Nishimura, T. (2005). Developmental changes in the shape of the
supralaryngeal vocal tract in chimpanzees. American Journal of Physical
Anthropology, 126(2), 193–204.
Oakes, L.M. (2012). Advances in eye tracking in infancy research. Infancy,
17, 1–8.
Obler, L.K. & Gjerlow, K. (1999). Language and the brain. Cambridge:
Cambridge University Press.
Ochs, E. (1982). Talking to children in Western Samoa. Language in Society,
11, 77–104.
Odean, R., Nazareth, A. & Pruden, S.M. (2015). Novel methodology to
examine cognitive and experiential factors in language development:
Combining eye-tracking and LENA technology. Frontiers in Psychology,
6, 1266.
O’Doherty, K., Troseth, G., Shimpi, P.M., Goldberg, E., Akhtar, N. & Saylor,
M.M. (2011). Third-party social interaction and word learning from video.
Child Development, 82(3), 902–915.
Ohms, V.R., Beckers, G.J.L., ten Cate, C. & Suthers, R.A. (2012). Vocal tract
articulation revisited: The case of the monk parakeet. The Journal of
Experimental Biology, 215(1), 85–92.
Olson, J. & Masur, E.F. (2012). Mothers respond differently to infants’
familiar versus non-familiar verbal imitations. Journal of Child Language,
39(4), 731–752.
Olson, J. & Masur, E.F. (2015). Mothers’ labeling responses to infants’
gestures predict vocabulary outcomes. Journal of Child Language, 42(6),
1289–1311.
Owens, R.E. (2008). Language development: An introduction (7th ed.).
Boston, MA: Pearson.
Pacton, S. & Deacon, S.H. (2008). The timing and mechanism of children’s
use of morphological information in spelling: A review of evidence from
English and French. Cognitive Development, 23(3), 339–359.
Pallier, C., Bosch, L. & Sebastián-Gallés, N. (1997). A limit on behavioral
plasticity in speech perception. Cognition, 64(3), B9–B17.
Pallier, C., Dehaene, S., Poline, J.-B., Le Bihan, D., Argenti, A.M., Dupoux,
E. & Mehler, J. (2003). Brain imaging of language plasticity in adopted
adults: Can a second language replace the first? Cerebral Cortex, 13(2),
155–161.
Pan, B.A. (2005). Semantic development: Learning the meanings of words.
In J. Berko Gleason (Ed.), The development of language (6th ed.) (pp.
112–147). Boston, MA: Pearson.
Patterson, F.G. (1978). The gestures of a gorilla: Language acquisition in
another pongid. Brain and Language, 5, 72–97.
Patterson, F.G., Tanner, J. & Mayer, N. (1988). Pragmatic analysis of gorilla
utterances: Early communicative development in the gorilla Koko. Journal
of Pragmatics, 12, 35–54.
Patterson, J.L. (2002). Relationships of expressive vocabulary to frequency of
reading and television experience among bilingual toddlers. Applied
 Psycholinguistics, 23(4), 493–508.
Patterson, M.D., Bly, B.M., Porcelli, A.J. & Rypma, B. (2007). Visual
working memory for global, object, and part-based information. Memory
& Cognition, 35(4), 738–751.
Paul, R. & Norbury, C.F. (2012). Language disorders from infancy through
adolescence: Listening, speaking, reading, writing, and communicating
(4th ed.). St Louis, MO: Elsevier.
Pearl, L. (2014). Evaluating learning-strategy components: Being fair.
(Commentary on Ambridge, Pine, & Lieven). Language, 90(3),
e107–e114.
Peña, M., Maki, A., Kovačič, D., Dehaene-Lambertz, G., Koizumi, H.,
Bouquet, F. & Mehler, J. (2003). Sounds and silence: An optical
topography study of language recognition at birth. Proceedings of the
National Academy of Sciences, 100(20), 11702–11705.
Penner, S.G. (1987). Parental responses to grammatical and ungrammatical
child utterances. Child Development, 58, 376–384.
Pepperberg, I.M. (2006). Cognitive and communicative abilities of Grey
Parrots. Applied Animal Behavior Science, 100, 77–86.
Pepperberg, I.M. (2010). Vocal learning in Grey Parrots: A brief review of
perception, production, and cross-species comparisons. Brain and
Language, 115, 81–91.
Pepperberg, I.M. & Carey, S. (2012). Grey Parrot number acquisition: The
inference of cardinal value from ordinal position on the numeral list.
Cognition, 125(2), 219–232.
Pepperberg, I.M., Koepke, A., Livingston, P., Girard, M. & Leigh, A.H.
(2013). Reasoning by inference: Further studies on exclusion in Grey
Parrots (Psittacus erithacus). Journal of Comparative Psychology, 127(3),
272–281.
Perani, D., Paulesu, E., Galles, N.S., Dupoux, E., Dehaene, S., Bettinardi, V.,
Cappa, S.F., Fazio, F. & Mehler, J. (1998). The bilingual brain:
Proficiency and age of acquisition of the second language. Brain, 121,
1841–1852.
Pérez-Leroux, A.T. & Kahnemuyipour, A. (2014). News, somewhat
exaggerated: Commentary on Ambridge, Pine, & Lieven. Language, 90(3),
e115–e125.
Peter, M., Chang, F., Pine, J.M., Blything, R. & Rowland, C.F. (2015). When
and how do children develop knowledge of verb argument structure?
 Evidence from verb bias effects in a structural priming task. Journal of
Memory and Language, 81, 1–15.
Philippon, A.C., Cherryman, J., Bull, R. & Vrij, A. (2007). Earwitness
identification performance: The effect of language, target, deliberate
strategies and indirect measures. Applied Cognitive Psychology, 21(4),
539–550.
Phillips, J.R. (1973). Syntax and vocabulary of mothers’ speech to young
children: Age and sex comparisons. Child Development, 44, 182–185.
Piaget, J. & Inhelder, B. (1974). The child’s construction of quantities:
Conservation and atomism. London: Routledge & Kegan Paul.
Piattelli-Palmarini, M. (Ed.) (1980). Language and learning: The debate
between Jean Piaget and Noam Chomsky. Cambridge, MA: Harvard
University Press.
Pierotti, R. & Annett, C. (1994). Patterns of aggression in gulls: Asymmetries
and tactics in different social categories. The Condor, 96(3), 590–599.
Pilley, J.W. (2013). Border collie comprehends sentences containing a
prepositional object, verb, and direct object. Learning and Motivation,
44(4), 229–240.
Pilley, J.W. & Reid, A.K. (2011). Border collie comprehends object names as
verbal referents. Behavioural Processes, 86(2), 184–195.
Pine, J.M. (1995). Variation in vocabulary development as a function of birth
order. Child Development, 66(1), 272–281.
Pine, J.M. & Lieven, E.V.M. (1993). Reanalyzing rote-learned phrases:
Individual differences in the transition to multi-word speech. Journal of
Child Language, 20(3), 551–571.
Pinker, S. (1984). Language learnability and language development.
Cambridge, MA: Harvard University Press.
Pinker, S. (1989). Learnability and cognition: The acquisition of argument
structure. Cambridge, MA: MIT Press.
Pinker, S. (1994). The language instinct: The new science of language and
mind. London: Penguin.
Pinker, S. (1999). Words and rules: The ingredients of language. New York:
Perennial.
Pinker, S. & Jackendoff, R. (2005). The faculty of language: What’s special
about it? Cognition, 95(2), 201–236.
Pinker, S. & Prince, A. (1988). On language and connectionism: Analysis of
a parallel distributed processing model of language acquisition. Cognition,
 28(1–2), 73–193.
Pinker, S. & Ullman, M.T. (2002). The past and future of the past tense.
Trends in Cognitive Sciences, 6(11), 456–463.
Plomin, R. (1990). Nature and nurture: An introduction to human behavioral
genetics. Pacific Grove, CA: Brooks/Cole.
Plunkett, K. & Wood, C. (2004). The development of children’s
understanding of grammar. In J. Oates & A. Grayson (Eds.), Cognitive and
language development in children (pp. 163–204). Milton Keynes: Open
University Press.
Polka, L., Jusczyk, P.W. & Ravchev, S. (1995). Methods for studying speech
perception in infants and children. In W. Strange (Ed.), Speech perception
and linguistic experience: Theoretical and methodological issues in cross-
language speech research. Timonium, MD: York Press.
Poremba, A., Malloy, M., Saunders, R.C., Carson, R.E., Herscovitch, P. &
Mishkin, M. (2004). Species-specific calls evoke asymmetric activity in
the monkey’s temporal poles. Nature, 427, 448–451.
Post, K. (1994). Negative evidence in the language learning environment of
laterborns in a rural Florida community. In J.L. Sokolov & C.E. Snow
(Eds.), Handbook of research in language development using CHILDES
(pp. 132–173). Hillsdale, NJ: Lawrence Erlbaum.
Poulin-Dubois, D. & Graham, S.A. (2007). Cognitive processes in early word
learning. In E. Hoff & M. Shatz (Eds.), Blackwell handbook of language
development (pp. 191–211). Malden, MA: Blackwell.
Premack, D. & Woodruff, G. (1978). Does the chimpanzee have a theory of
mind? Behavioral and Brain Sciences, 1(4), 515–526.
Price, P.J., Ostendorf, M., Shattuck-Hufnagel, S. & Fong, C. (1991). The use
of prosody in syntactic disambiguation. Journal of the Acoustical Society
of America, 90(6), 2956–2970.
Pullum, G.K. & Scholz, B.C. (2002). Empirical assessment of stimulus
poverty arguments. Linguistic Review, 19, 9–50.
Purver, M., Ginzburg, J. & Healey, P. (2001). On the means for clarification
in dialogue. In J. van Kuppevelt & R. Smith (Eds.), Current and new
directions in discourse and dialogue (pp. 235–256). Dordrecht: Kluwer.
Pye, C. (1986). Quiché Mayan speech to children. Journal of Child
Language, 13(1), 85–100.
Quine, W.V.O. (1960). Word and object. Cambridge, MA: MIT Press.
Rabin, J. & Deacon, H. (2008). The representation of morphologically
 complex words in the developing lexicon. Journal of Child Language,
35(2), 453–465.
Radford, A. (1990). Syntactic theory and the acquisition of English syntax:
The nature of early child grammars of English. Oxford: Basil Blackwell.
Radford, A., Atkinson, M., Britain, D., Clahsen, H. & Spencer, A. (1999).
Linguistics: An introduction. Cambridge: Cambridge University Press.
Raffaele, P. (2006). Speaking bonobo. Smithsonian, 37(8), 74.
Ramirez, N.F., Leonard, M.K., Torres, C., Hatrak, M., Halgren, E. &
Mayberry, R.I. (2014). Neural language processing in adolescent first-
language learners. Cerebral Cortex, 24(10), 2772–2783.
Ramirez, N.F., Liebermann, A.M. & Mayberry, R.I. (2013). The initial stages
of first-language acquisition begun in adolescence: When late looks early.
Journal of Child Language, 40(2), 391–414.
Ramírez-Esparza, N., García-Sierra, A. & Kuhl, P.K. (2014). Look who’s
talking: Speech style and social context in language input to infants are
linked to concurrent and future speech development. Developmental
Science, 17(6), 880–891.
Ramscar, M., Dye, M. & Hübner, M. (2013). When the fly flied and when the
fly flew: How semantics affect the processing of inflected verbs. Language
and Cognitive Processes, 28(4), 468–497.
Räsänen, S.H.M., Ambridge, B. & Pine, J.M. (2014). Infinitives or bare
stems? Are English-speaking children defaulting to the highest-frequency
form? Journal of Child Language, 41(4), 756–779.
Ravid, D. & Geiger, V. (2009). Promoting morphological awareness in
Hebrew-speaking grade-schoolers: An intervention using linguistic humor.
First Language, 29(1), 81–112.
Reali, F. & Christiansen, M.H. (2005). Uncovering the richness of the
stimulus: Structure dependence and indirect statistical evidence. Cognitive
Science, 29, 1007–1028.
Regel, S., Opitz, A., Müller, G. & Friederici, A.D. (2015). The past tense
debate revisited: Electrophysiological evidence for subregularities of
irregular verb inflection. Journal of Cognitive Neuroscience, 27(9),
1870–1885.
Regier, T. & Gahl, S. (2004). Learning the unlearnable: The role of missing
evidence. Cognition, 93, 147–155.
Rensink, R.A., O’Regan, J.K. & Clark, J.J. (2000). On the failure to detect
changes in scenes across brief interruptions. Visual Cognition, 7(1–3),
 127–145.
Rescorla, L.A. (1980). Overextension in early language development. Journal
of Child Language, 7, 321–335.
Rescorla, L.A., Lee, Y., Oh, K. & Kim, Y. (2013). Lexical development in
Korean: Vocabulary size, lexical composition, and late talking. Journal of
Speech, Language, and Hearing Research, 56(2), 735–747.
Ribeiro, L.A., Zachrisson, H.D., Gustavson, K. & Schj⊘lberg, S. (2016).
Maternal distress during pregnancy and language development in
preschool age: A population-based cohort study. European Journal of
Developmental Psychology, 13(1), 20–39.
Rice, M.L., Huston, A.C., Truglio, R. & Wright, J. (1990). Words from
Sesame Street: Learning vocabulary while viewing. Developmental
Psychology, 26(3), 421–428.
Rice, M.L. & Woodsmall, L. (1988). Lessons from television: Children’s
word learning when viewing. Child Development, 59(2), 420–429.
Risch, N.J. (2000). Searching for genetic determinants in the new
millennium. Nature, 405(6788), 847–856.
Rivas, E. (2005). Recent use of signs by chimpanzees (Pan Troglodytes) in
interactions with humans. Journal of Comparative Psychology, 119(4),
401–417.
Rivera-Gaxiola, M., Silva-Pereyra, J. & Kuhl, P.K. (2005). Brain potentials
to native and non-native speech contrasts in 7- and 11-month-old
American infants. Developmental Science, 8(2), 162–172.
Rizzolatti, G. & Arbib, M.A. (1998). Language within our grasp. Trends in
Neurosciences, 21(5), 188–194.
Rochat, P. (2001). The infant’s world. Cambridge, MA: Harvard University
Press.
Rogers, T.T. & Patterson, K. (2007). Object categorization: Reversals and
explanations of the basic-level advantage. Journal of Experimental
Psychology: General, 136(3), 451–469.
Rohde, D.L.T. & Plaut, D.C. (1999). Language acquisition in the absence of
explicit negative evidence: How important is starting small? Cognition,
72(1), 67–109.
Romberg, A. & Saffran, J.R. (2010). Statistical learning and language
acquisition. Advanced Review WIREs Cognitive Science. Published
online: 26 July 2010.
Rondal, J. & Cession, A. (1990). Input evidence regarding the semantic
 bootstrapping hypothesis. Journal of Child Language, 17, 711–717.
Rosch, E.H. (1973). Natural categories. Cognitive Psychology, 4(3),
328–350.
Rosen, S. & Iverson, P. (2007). Constructing adequate non-speech analogues:
What is special about speech anyway? Developmental Science, 10(2),
165–168.
Rosenthal Rollins, P. & Snow, C.E. (1998). Shared attention and grammatical
development in typical children and children with autism. Journal of Child
Language, 25(3), 653–673.
Rowe, M.L. (2008). Child-directed speech: Relation to socioeconomic status,
knowledge of child development and child vocabulary skill. Journal of
Child Language, 35(1), 185–205.
Rowe, M.L. (2012). A longitudinal investigation of the role of quantity and
quality of child-directed speech in vocabulary development. Child
Development, 83(5), 1762–1774.
Rowe, M.L. & Goldin-Meadow, S. (2009a). Early gesture selectively predicts
later language learning. Developmental Science, 12(1), 182–187.
Rowe, M.L. & Goldin-Meadow, S. (2009b). Differences in early gesture
explain SES disparities in child vocabulary size at school entry. Science,
323(5916), 951–953.
Rowland, C.F. (2007). Explaining errors in children’s questions. Cognition,
104(1), 106–134.
Rowland, C.F. (2014). Understanding language acquisition. Abingdon:
Routledge.
Ruffman, T., Slade, L. & Redman, J. (2005). Young infants’ expectations
about hidden objects. Cognition, 97(2), B35–B43.
Rumelhart, D.E. & McClelland, J.L. (1986). On learning the past tenses of
English verbs. In J.L. McClelland, D.E. Rumelhart & the PDP Research
Group (Eds.), Parallel distributed processing: Explorations in the
microstructure of cognition. Vol. 2: Psychological and biological models.
Cambridge, MA: Bradford Books/MIT Press.
Rymer, R. (1993). Genie: A scientific tragedy. Harmondsworth: Penguin.
Sachs, J., Bard, B. & Johnson, M.L. (1981). Language learning with
restricted input: Case studies of two hearing children of deaf parents.
Applied Psycholinguistics, 2, 33–54.
Sachs, J., Brown, R. & Salerno, R. (1976). Adult’s speech to children. In W.
von Raffler Engel & Y. Lebrun (Eds.), Baby talk and infant speech (pp.
 240–245). Lisse: Peter de Ridder Press.
Saffran, J.R. (2003). Statistical language learning: Mechanisms and
constraints. Current Directions in Psychological Science, 12(4), 110–114.
Saffran, J.R., Aslin, R.N. & Newport, E.L. (1996). Statistical learning by 8-
month-old infants. Science, 274(5294), 1926–1928.
Saffran, J.R., Hauser, M., Seibel, R.L., Kapfhamer, J., Tsao, F. & Cushman,
F. (2008). Grammatical pattern learning by infants and cotton-top tamarin
monkeys. Cognition, 107, 479–500.
Saffran, J.R., Johnson, E.K., Aslin, R.N. & Newport, E.L. (1999). Statistical
learning of tone sequences by human infants and adults. Cognition, 70,
27–52.
Saffran, J.R., Newport, E.L. & Aslin, R.N. (1996). Word segmentation: The
role of distributional cues. Journal of Memory and Language, 35, 606–621.
Saffran, J.R., Pollak, S.D., Seibel, R.L. & Shkolnik, A. (2007). Dog is a dog
is a dog: Infant rule learning is not specific to language. Cognition, 105(3),
669–680.
Saffran, J.R. & Thiessen, E.D. (2003). Pattern induction by infant language
learners. Developmental Psychology, 39, 484–494.
Saffran, J.R. & Thiessen, E.D. (2007). Domain-general learning capacities. In
E. Hoff & M. Shatz (Eds.), Blackwell handbook of language development
(pp. 68–86). Malden, MA: Blackwell.
Saito, A. & Shinozuka, K. (2013). Vocal recognition of owners by domestic
cats (Felis catus). Animal Cognition, 16(4), 685–690.
Salkie, R. (1990). The Chomsky update: Linguistics and politics. London:
Unwin Hyman.
Sampson, G. (2005). The ‘language instinct’ debate (revised ed.). London:
Continuum.
Samuelson, L.K., Horst, J.S., Schutte, A.R. & Dobbertin, B.N. (2008). Rigid
thinking about deformables: Do children sometimes overgeneralize the
shape bias? Journal of Child Language, 35(3), 559–589.
Samuelson, L.K. & Smith, L.B. (2005). They call it like they see it:
Spontaneous naming and attention to shape. Developmental Science, 8,
182–198.
Santelmann, L., Berk, S., Austin, J., Somashekar, S. & Lust, B. (2002).
Continuity and development in the acquisition of inversion in yes/no
questions: Dissociating movement and inflection. Journal of Child
Language, 29(4), 813–842.
Sarnecka, B.W., Kamenskaya, T.O., Yamana, Y. & Yudovina, J.B. (2004).
Language as lens: Plural marking and numeral learning in English,
Japanese, and Russian. Somerville, MA: Cascadilla Press.
Savage-Rumbaugh, E.S., Murphy, J., Sevcik, R.A., Brakke, K.E., Williams,
S.L. & Rumbaugh, D.M. (1993). Language comprehension in ape and
child. Monographs of the Society for Research in Child Development.
Serial 233, 58, 3–4.
Saxton, M. (1995). Negative evidence versus negative feedback: A study of
corrective input in child language acquisition. Unpublished doctoral
dissertation, University of Oxford.
Saxton, M. (1997). The Contrast theory of negative input. Journal of Child
Language, 24, 139–161.
Saxton, M. (2000). Negative evidence and negative feedback: Immediate
effects on the grammaticality of child speech. First Language, 20(2),
221–252.
Saxton, M. (2008). What’s in a name? Coming to terms with the child’s
linguistic environment. Journal of Child Language, 35(3), 677–686.
Saxton, M. (2009). The inevitability of Child Directed Speech. In S. Foster-
Cohen (Ed.), Advances in language acquisition (pp. 62–86). London:
Palgrave Macmillan.
Saxton, M., Backley, P. & Gallaway, C. (2005). Negative input for
grammatical errors: Effects after a lag of 12 weeks. Journal of Child
Language, 32, 643–672.
Saxton, M., Houston-Price, C. & Dawson, N. (2005). The Prompt hypothesis:
Clarification requests as corrective input for grammatical errors. Applied
Psycholinguistics, 26(3), 393–414.
Saxton, M., Kulcsar, B., Marshall, G. & Rupra, M. (1998). Longer-term
effects of corrective input: An experimental approach. Journal of Child
Language, 25, 701–721.
Scaife, M. & Bruner, J. (1975). The capacity for joint visual attention in
infants. Nature, 253, 265–266.
Schachter, F.F., Fosha, D., Stemp, S., Brotman, N. & Ganger, S. (1976).
Everyday caretaker talk to toddlers vs. 3s and 4s. Journal of Child
Language, 3(2), 221–245.
Schachter, J. (1990). On the issue of completeness in second language
acquisition. Second Language Research, 6, 93–124.
Schlesinger, I.M. (1977). Role of cognitive development and linguistic input
 in language acquisition. Journal of Child Language, 4(2), 153–169.
Schmidt, A., Oliveira, F.M., Santos Lotério, L. & Gomes, G. (2016).
Learning name–object relations after a single exclusion trial in 18- to 48-
month-old children. The Psychological Record, 66(1), 53–63.
Schmidt, C.L. (1996). Scrutinizing reference: How gesture and speech are
coordinated in mother–child interaction. Journal of Child Language, 23(2),
279–305.
Schmidt, C.L. & Lawson, K.R. (2002). Caregiver attention-focusing and
children’s attention-sharing behaviours as predictors of later verbal IQ in
very low birthweight children. Journal of Child Language, 29(1), 3–22.
Scholz, B.C. & Pullum, G.K. (2006). Irrational nativist exuberance. In R.J.
Stainton (Ed.), Contemporary debates in cognitive science (pp. 59–80).
Oxford: Basil Blackwell.
Schoner, G. & Thelen, E. (2006). Using dynamic field theory to rethink
infant habituation. Psychological Review, 113(2), 273–299.
Schouten, B., Gerrits, E. & van Hessen, A. (2003). The end of categorical
perception as we know it. Speech Communication, 41, 71–80.
Scofield, J. & Behrend, D.A. (2011). Clarifying the role of joint attention in
early word learning. First Language, 31(3), 326–341.
Scovel, T. (1988). A time to speak: A psycholinguistic inquiry into the
critical period for human speech. Cambridge, MA: Newbury House.
Seidenberg, M.S. (1992). Connectionism without tears. In S. Davis (Ed.),
Connectionism: Theory and practice (pp. 84–122). New York: Oxford
University Press.
Seidenberg, M.S., MacDonald, M.C. & Saffran, J.R. (2002). Does grammar
start where statistics stop? Science, 298(5593), 553–554.
Seidenberg, M.S. & Petitto, L.A. (1979). Signing behavior in apes: A critical
review. Cognition, 7, 177–215.
Seidl, A. & Johnson, E.K. (2006). Infant word segmentation revisited: Edge
alignment facilitates target extraction. Developmental Science, 9(6),
565–573.
Seidl, A. & Johnson, E.K. (2008). Boundary alignment enables 11-month-
olds to segment vowel initial words from speech. Journal of Child
Language, 35(1), 1–24.
Seifert, K.L., Hoffnung, R.J. & Hoffnung, M. (2000). Lifespan development
(2nd ed.). Boston, MA: Houghton Mifflin.
Seitz, S. & Stewart, C. (1975). Imitations and expansions: Some
 developmental aspects of mother–child communication. Developmental
Psychology, 11(6), 763–768.
Seuss, Dr (1958). The cat in the hat. London: Collins.
Seyfarth, R.M. & Cheney, D.L. (2003). Signalers and receivers in animal
communication. Annual Review of Psychology, 54, 145–173.
Shady, M. & Gerken, L. (1999). Grammatical and caregiver cues in early
sentence comprehension. Journal of Child Language, 26(1), 163–175.
Shankweiler, D., Palumbo, L.C., Fulbright, R.K., Mencl, W.E., Van Dyke, J.,
Kollia, B., et al. (2010). Testing the limits of language production in long-
term survivors of major stroke: A psycholinguistic and anatomic study.
Aphasiology, 24(11), 1455–1485.
Shatz, M. & Gelman, R. (1973). The development of communication skills:
Modifications in the speech of young children as a function of the listener.
Monographs of the Society for Research in Child Development, serial no.
5, 38.
Shaw, G.B. (1911). The doctor’s dilemma, getting married, and the shewing-
up of Blanco Posnet. London: Constable.
Shneidman, L.A., Arroyo, M.E., Levine, S.C. & Goldin-Meadow, S. (2013).
What counts as effective input for word learning? Journal of Child
Language, 40(3), 672–686.
Shultz, S. & Vouloumanos, A. (2010). Three-month-olds prefer speech to
other naturally occurring signals. Language Learning and Development,
6(4), 241–257.
Siegel, L.S. (2008). Morphological awareness skills of English language
learners and children with dyslexia. Topics in Language Disorders, 28(1),
15–27.
Singh, L., Liederman, J., Mierzejewski, R. & Barnes, J. (2011). Rapid
reacquisition of native phoneme contrasts after disuse: You do not always
lose what you do not use. Developmental Science, 14(5), 949–959.
Singh, L., Reznick, J.S. & Xuehua, L. (2012). Infant word segmentation and
childhood vocabulary development: A longitudinal analysis.
Developmental Science, 15(4), 482–495.
Sinnott, J.M. (1998). Comparative phoneme boundaries. Current Topics in
Acoustical Research, 2, 135–138.
Skinner, B.F. (1957). Verbal behavior. New York: Appleton-Century-Crofts.
Slobin, D.I. (1973). Cognitive prerequisites for the development of grammar.
In C.A. Ferguson & D.I. Slobin (Eds.), Studies of child language
 development. New York: Holt, Rinehart & Winston.
Slobin, D.I. (1978). A case study of early language awareness. In A. Sinclair,
R.J. Jarvella & W.J.M. Levelt (Eds.), The child’s conception of language
(pp. 45–54). New York: Springer.
Smith, L.B. (1999). Children’s noun learning: How general learning
processes make specialized learning mechanisms. In B. MacWhinney
(Ed.), The emergence of language (pp. 277–303). Mahwah, NJ: Lawrence
Erlbaum.
Smith, L.B. (2000). Learning how to learn words: An associative crane. In
R.M. Golinkoff, K. Hirsh-Pasek, L. Bloom, L.B. Smith, A.L. Woodward,
N. Akhtar, M. Tomasello & G. Hollich (Eds.), Becoming a word learner: A
debate on lexical acquisition (pp. 51–80). Oxford: Oxford University
Press.
Smith, L.B. (2001). How domain-general processes may create domain-
specific biases. In M. Bowerman & S.C. Levinson (Eds.), Language
acquisition and conceptual development (pp. 101–131). Cambridge:
Cambridge University Press.
Smith, N. (2004). Chomsky: Ideas and ideals (2nd ed.). Cambridge:
Cambridge University Press.
Smith, N.A. & Trainor, L. (2008). Infant-directed speech is modulated by
infant feedback. Infancy, 13(4), 410–420.
Smith-Lock, K.M. (2015). Rule-based learning of regular past tense in
children with specific language impairment. Cognitive Neuropsychology,
32(3–4), 221–242.
Snow, C.E. (1972). Mothers’ speech to children learning language. Child
Development, 43(2), 549–566.
Snow, C.E. (1981). The uses of imitation. Journal of Child Language, 8(1),
205–212.
Snow, C.E. (1987). Relevance of a critical period to language acquisition. In
M.H. Bornstein (Ed.), Sensitive periods in development: Interdisciplinary
perspectives (pp. 183–210). Hillsdale, NJ: Lawrence Erlbaum.
Snow, C.E. (1995). Issues in the study of input: Finetuning, universality,
individual and developmental differences and necessary causes. In P.
Fletcher & B. MacWhinney (Eds.), The handbook of child language (pp.
180–193). Oxford: Basil Blackwell.
Snow, C.E., Arlman-Rupp, A., Hassing, Y., Jobse, J., Joosten, J. & Vorster, J.
(1976). Mothers’ speech in three social classes. Journal of Psycholinguistic
 Research, 5, 1–20.
Snyder, W. (2007). Child language: The parametric approach. Oxford:
Oxford University Press.
Soderstrom, M. (2007). Beyond babytalk: Re-evaluating the nature and
content of speech input to preverbal infants. Developmental Review, 27(4),
501–532.
Soderstrom, M. (2014). All hands on deck: In defense of the prosodic
bootstrapping hypothesis and multiple theoretical approaches.
(Commentary on Ambridge, Pine, & Lieven.) Language, 90(3),
e126–e130.
Soderstrom, M., Blossom, M., Foygel, R. & Morgan, J.L. (2008). Acoustical
cues and grammatical units in speech to two preverbal infants. Journal of
Child Language, 35(4), 869–902.
Soja, N.N., Carey, S. & Spelke, E.S. (1991). Ontological categories guide
young children’s inductions of word meaning – object terms and substance
terms. Cognition, 38(2), 179–211.
Sommerville, J.A. & Crane, C.C. (2009). Ten-month-old infants use prior
information to identify an actor’s goal. Developmental Science, 12(2),
314–325.
Song, L., Spier, E.T. & Tamis-LeMonda, C.S. (2014). Reciprocal influences
between maternal language and children’s language and cognitive
development in low-income families. Journal of Child Language, 41(2),
305–326.
Song, S., Su, M., Kang, C., Liu, H., Zhang, Y., McBride-Chang, C., Tardif,
T., Li, H., Liang, W., Zhang, Z. & Shu, H. (2015). Tracing children’s
vocabulary development from preschool through the school-age years: An
8-year longitudinal study. Developmental Science, 18(1), 119–131.
Sontag, L.W. & Wallace, R.I. (1936). Changes in the heart rate of the human
fetal heart in response to vibratory stimuli. American Journal of Diseases,
51, 583–589.
Sorace, A. (1993). Incomplete vs. divergent representations of unaccusativity
in non-native grammars of Italian. Second Language Research, 9, 22–47.
Spaai, G.W.G., Derksen, E.S., Hermes, D.J. & Kaufholz, P.A.P. (1996).
Teaching intonation to young deaf children with the intonation meter. Folia
Phoniatrica et Logopaedica, 48, 22–34.
Spaepen, E. & Spelke, E. (2007). Will any doll do? 12-month-olds’ reasoning
about goal objects. Cognitive Psychology, 54(2), 133–154.
Spelke, E.S. (1994). Initial knowledge: Six suggestions. Cognition, 50,
443–447.
Spelke, E.S. & Kinzler, K.D. (2007). Core knowledge. Developmental
Science, 10(1), 89–96.
Spence, M.J. & DeCasper, A.J. (1987). Prenatal experience with low-
frequency maternal-voice sounds influences neonatal perception of
maternal voice samples. Infant Behavior and Development, 10, 133–142.
Srinivasan, M. & Snedeker, J. (2014). Polysemy and the taxonomic
constraint: Children’s representation of words that label multiple kinds.
Language Learning and Development, 10(2), 97–128.
Stanfield, C., Williamson, R. & Özçalişkan, Ş. (2014). How early do children
understand gesture–speech combinations with iconic gestures? Journal of
Child Language, 41(2), 462–471.
Stark, R.E. (1986). Prespeech segmental feature development. In P. Fletcher
& M. Garman (Eds.), Language acquisition: Studies in first language
development (2nd ed.) (pp. 149–173). Cambridge: Cambridge University
Press.
Stern, D.N., Spieker, S., Barnett, R.K. & MacKain, K. (1983). The prosody
of maternal speech: Infant age and context related changes. Journal of
Child Language, 10(1), 1–15.
Stern, W. (1924). Psychology of early childhood up to the sixth year of age.
New York: Holt.
Stich, S. (1975). Innate ideas. Berkeley, CA: University of California Press.
Stilwell-Peccei, J. (2006). Child language: A resource book for students.
Abingdon: Routledge.
Stoel-Gammon, C. & Sosa, A.V. (2007). Phonological development. In E.
Hoff & M. Shatz (Eds.), Blackwell handbook of language development
(pp. 238–256). Malden, MA: Blackwell.
Stoeger, A., Mietchen, D., Oh, S., De Silva, S., Herbst, C., Kwon, S. & Fitch,
W.T. (2012). An Asian elephant imitates human speech. Current Biology,
22(22), 2144–2148.
Strapp, C.M. (1999). Mothers’, fathers’, and siblings’ responses to children’s
language errors: Comparing sources of negative evidence. Journal of Child
Language, 26, 373–391.
Strapp, C.M., Bleakney, D.M., Helmick, A.L. & Tonkovich, H.M. (2008).
Developmental differences in the effects of negative and positive evidence.
First Language, 28(1), 35–53.
Strapp, C.M. & Federico, A. (2000). Imitations and repetitions: What do
children say following recasts? First Language, 20(3), 273–290.
Streeter, L.A. (1976). Language perception of 2-month-old infants shows
effects of both innate mechanisms and experience. Nature, 259, 39–41.
Sully, J. (1895). Studies of childhood. London: Longmans Green.
Sumner, M. & Samuel, A.G. (2007). Lexical inhibition and sublexical
facilitation are surprisingly long lasting. Journal of Experimental
Psychology – Learning Memory and Cognition, 33(4), 769–790.
Suzuki, T. & Yoshinaga, N. (2013). Children’s knowledge of hierarchical
phrase structure: Quantifier floating in Japanese. Journal of Child
Language, 40(3), 628–655.
Swensen, L.D., Naigles, L.R. & Fein, D. (2007). Does maternal input affect
the language of children with autism? In H. Caunt-Nolan, S. Kulatilake &
I. Woo (Eds.), Proceedings of the 30th Annual Boston University
conference on language development. Somerville, MA: Cascadilla Press.
Szagun, G. (2001). Learning different regularities: The acquisition of noun
plurals by German-speaking children. First Language, 21(2), 109–141.
Taha, H. & Saiegh-Haddad, E. (2016). The role of phonological versus
morphological skills in the development of Arabic spelling: An
intervention study. Journal of Psycholinguist Research, 45(3), 507–535.
Taine, H. (1877). On the acquisition of language by children. Mind, 2,
252–259.
Tamis-LeMonda, C.S., Bornstein, M., Kahana-Kalman, R., Baumwell, L. &
Cyphers, L. (1998). Predicting variation in the timing of language
milestones in the second year: An events history approach. Journal of
Child Language, 25, 675–700.
Tamis-LeMonda, C.S., Kuchirko, Y. & Song, L. (2014). Why is infant
language learning facilitated by parental responsiveness? Current
Directions in Psychological Science, 23(2), 121–126.
Tardif, T. (1996). Nouns are not always learned before verbs: Evidence from
Mandarin speakers’ early vocabularies. Developmental Psychology, 32(3),
492–504.
Tardif, T., Gelman, S.A. & Xu, F. (1999). Putting the ‘noun bias’ in context:
A comparison of English and Mandarin. Child Development, 70(3),
620–635.
Taylor, N., Donovan, W., Miles, S. & Leavitt, L. (2009). Maternal control
strategies, maternal language usage and children’s language usage at two
 years. Journal of Child Language, 36(2), 381–404.
Teigen, K.H. (1984). Note on the origin of the term nature and nurture – not
Shakespeare and Galton, but Mulcaster. Journal of the History of the
Behavioral Sciences, 20(4), 363–364.
ten Cate, C. (2014). On the phonetic and syntactic processing abilities of
birds: From songs to speech and artificial grammars. Current Opinion in
Neurobiology, 28, 157–164.
Tenenbaum, E.J., Sobel, D.M., Sheinkopf, S.J., Malle, B.F. & Morgan, J.L.
(2015). Attention to the mouth and gaze following in infancy predict
language development. Journal of Child Language, 42(6), 1173–1190.
Terrace, H. (1979). Nim. New York: Knopf.
Theakston, A.L. & Lieven, E.V.M. (2008). The influence of discourse
context on children’s provision of auxiliary BE. Journal of Child
Language, 35(1), 129–158.
Theakston, A.L., Lieven, E.V.M., Pine, J.M. & Rowland, C.F. (2002). Going,
going, gone: The acquisition of the verb ‘go’. Journal of Child Language,
29(4), 783–811.
Theakston, A.L., Lieven, E.V.M., Pine, J.M. & Rowland, C.F. (2004).
Semantic generality, input frequency and the acquisition of syntax. Journal
of Child Language, 31(1), 61–99.
Thiessen, E.D., Hill, E.A. & Saffran, J.R. (2005). Infant-directed speech
facilitates word segmentation. Infancy, 7(1), 53–71.
Thiessen, E.D. & Saffran, J.R. (2003). When cues collide: Use of stress and
statistical cues to word boundaries by 7- to 9-month-old infants.
Developmental Psychology, 39(4), 706–716.
Thiessen, E.D. & Saffran, J.R. (2004). Spectral tilt as a cue to word
segmentation in infancy and adulthood. Perception and Psychophysics,
66(5), 779–791.
Thurm, A., Lord, C., Lee, L.C. & Newschaffer, C. (2007). Predictors of
language acquisition in preschool children with autism spectrum disorders.
Journal of Autism and Developmental Disorders, 37(9), 1721–1734.
Tomasello, M. (1992). First verbs: A case study of early grammatical
development. New York: Cambridge University Press.
Tomasello, M. (2003). Constructing a language: A usage-based theory of
language acquisition. Cambridge, MA: Harvard University Press.
Tomasello, M. (2004). Syntax or semantics? Response to Lidz et al.
Cognition, 93(2), 139–140.
Tomasello, M. (2005). Beyond formalities: The case of language acquisition.
The Linguistic Review, 22(2–4), 183–197.
Tomasello, M. (2009). The usage-based theory of language acquisition. In
E.L. Bavin (Ed.), The Cambridge handbook of child language (pp. 69–87).
Cambridge: Cambridge University Press.
Tomasello, M. & Carpenter, M. (2007). Shared intentionality. Developmental
Science, 10(1), 121–125.
Tomasello, M., Carpenter, M., Call, J., Behne, T. & Moll, H. (2005).
Understanding and sharing intentions: The origins of cultural cognition.
Behavioral and Brain Sciences, 28(5), 675–691.
Tomasello, M. & Kaminski, J. (2009). Like infant, like dog. Science,
325(5945), 1213–1214.
Tomasello, M., Mannle, S. & Kruger, A.C. (1986). Linguistic environment of
1- to 2-year-old twins. Developmental Psychology, 22, 169–176.
Tomasello, M. & Stahl, D. (2004). Sampling children’s spontaneous speech:
How much is enough? Journal of Child Language, 31(1), 101–121.
Trehub, S.E. (1976). The discrimination of foreign speech contrasts by
infants and adults. Child Development, 47, 466–472.
Trevarthen, C. (1983). Interpersonal abilities of infants as generators for
transmission of language and culture. In A. Oliverio & M. Zapella (Eds.),
The behaviour of human infants. London: Plenum.
Tsao, F.M., Liu, H.M. & Kuhl, P.K. (2004). Speech perception in infancy
predicts language development in the second year of life: A longitudinal
study. Child Development, 75(4), 1067–1084.
Tsao, F.M., Liu, H.M. & Kuhl, P.K. (2006). Perception of native and non-
native affricate-fricative contrasts: Cross-language tests on adults and
infants. Journal of the Acoustical Society of America, 120(4), 2285–2294.
Tsushima, T., Takizawa, O., Sasaki, M., Nishi, K., Kohno, M., Menyuk, P. &
Best, C.T. (1994). Discrimination of English /r-l/ and /w-y/ by Japanese
infants at 6–12 months: Language specific developmental changes in
speech perception abilities. Paper presented at International Conference on
Spoken Language Processing, 4. Yokohama, Japan.
Uchikoshi, Y. (2005). Narrative development in bilingual kindergartners: Can
Arthur help? Developmental Psychology, 41(3), 464–478.
Ullman, M.T., Pancheva, R., Love, T., Yee, E., Swinney, D. & Hickok, G.
(2005). Neural correlates of lexicon and grammar: Evidence from the
production, reading and judgment of inflection in aphasia. Brain and
 Language, 93(2), 185–238.
Uriagereka, J. (1998). Rhyme and reason: An introduction to minimalist
syntax. Cambridge, MA: MIT Press.
Valian, V. (2014). Arguing about innateness. Journal of Child Language,
41(S1), 78–92.
Vasilyeva, M., Waterfall, H. & Huttenlocher, J. (2008). Emergence of syntax:
Commonalities and differences across children. Developmental Science,
11(1), 84–97.
Vihman, M.M. (2014). Phonological development: The first two years (2nd
ed.). Malden, MA: Wiley-Blackwell.
Vivanti, G., Nadig, A., Ozonoff, S. & Rogers, S.J. (2008). What do children
with autism attend to during imitation tasks? Journal of Experimental
Child Psychology, 101(3), 186–205.
von Frisch, K. (1954). The dancing bees: An account of the life and senses of
the honey bee (translated by D. Ilse). London: Methuen.
von Humboldt, W. (1836/1988). Über die Verschiedenheit des menschlichen
Sprachbaues und ihren Einfluss auf die geistige Entwickelung des
Menschengeschlechts. English translation by P. Heath. Cambridge:
Cambridge University Press.
Vouloumanos, A. & Curtin, S. (2014). Foundational tuning: How infants’
attention to speech predicts language development. Cognitive Science,
38(8), 1675–1686.
Vouloumanos, A., Hauser, M.D., Werker, J.F. & Martin, A. (2010). The
tuning of human neonates’ preference for speech. Child Development,
81(2), 517–527.
Vouloumanos, A., Kiehl, K.A., Werker, J.F. & Liddle, P.F. (2001). Detection
of sounds in the auditory stream: Event-related fMRI evidence for
differential activation to speech and nonspeech. Journal of Cognitive
Neuroscience, 13(7), 994–1005.
Vouloumanos, A. & Werker, J.F. (2007). Listening to language at birth:
Evidence for a bias for speech in neonates. Developmental Science, 10(2),
159–171.
Vygotsky, L.S. (1934/1962). Thought and language. Cambridge, MA: MIT
Press.
Wade, T. (1992). A comprehensive Russian grammar. Oxford: Blackwell.
Wales, R. (1986). Deixis. In P. Fletcher & M. Garman (Eds.), Language
acquisition (2nd ed.). (pp. 401–428). Cambridge: Cambridge University
 Press.
Walle, E.A. & Campos, J.J. (2014). Infant language development is related to
the acquisition of walking. Developmental Psychology, 50(2), 336–348.
Wallman, J. (1992). Aping language. Cambridge: Cambridge University
Press.
Wallman, J. (1993). On Aping language: Reply to Lieberman. Current
Anthropology, 34(4), 431–432.
Wanner, E. & Gleitman, L.R. (Eds.) (1982). Language acquisition: The state
of the art. Cambridge: Cambridge University Press.
Warlaumont, A.S. & Jarmulowicz, L. (2012). Caregivers’ suffix frequencies
and suffix acquisition by language impaired, late talking, and typically
developing children. Journal of Child Language, 39(5), 1017–1042.
Warren-Leubecker, A. & Bohannon, J.N. (1984). Intonation patterns in child-
directed speech – mother–father differences. Child Development, 55(4),
1379–1385.
Wartenburger, I., Heekeren, H.R., Abutalebi, J., Cappa, S.F., Villringer, A. &
Perani, D. (2003). Early setting of grammatical processing in the bilingual
brain. Neuron, 37(1), 159–170.
Wasserman, E.A., Brooks, D.I. & McMurray, B. (2015). Pigeons acquire
multiple categories in parallel via associative learning: A parallel to human
word learning? Cognition, 136, 99–122.
Watt, W.C. (1970). On two hypotheses concerning psycholinguistics. In J.R.
Hayes (Ed.), Cognition and the development of language. New York:
Wiley.
Waxman, S.R. & Markow, D.B. (1995). Words as invitations to form
categories: Evidence from twelve- to thirteen-month-old infants. Cognitive
Psychology, 29, 257–302.
Weisleder, A. & Fernald, A. (2013). Talking to children matters: Early
language experience strengthens processing and builds vocabulary.
Psychological Science, 24(11), 2143–2152.
Weisleder, A., Otero, N., Marchman, V.A. & Fernald, A. (2015). Child-
directed speech mediates SES differences in language-processing skill and
vocabulary in Spanish-learning children. Paper presented at the Biennial
Meeting of the Society for Research in Child Development, Philadelphia,
PA.
Weiss, D.J. & Newport, E.L. (2006). Mechanisms underlying language
acquisition: Benefits from a comparative approach. Infancy, 9(2),
 241–257.
Weissenborn, J., Goodluck, H. & Roeper, T. (1992). Introduction: Old and
new problems in the study of language acquisition. In J. Weissenborn, H.
Goodluck & T. Roeper (Eds.), Theoretical issues in language acquisition:
Continuity and change in development. (pp. 1–23). Hillsdale, NJ:
Lawrence Erlbaum.
Weppelman, T.L., Bostow, A., Schiffer, R., Elbert-Perez, E. & Newman, R.S.
(2003). Children’s use of the prosodic characteristics of infant-directed
speech. Language and Communication, 23(1), 63–80.
Werker, J.F. & Hensch, T.K. (2015). Critical periods in speech perception:
New directions. Annual Review of Psychology, 66, 173–196.
Werker, J.F. & Lalonde, C. (1988). Cross-language speech perception: Initial
capabilities and developmental change. Developmental Psychology, 24,
672–683.
Werker, J.F. & McLeod, P.J. (1989). Infant preference for both male and
female infant-directed talk: A developmental study of attentional and
affective responsiveness. Canadian Journal of Psychology-Revue
Canadienne De Psychologie, 43(2), 230–246.
Werker, J.F. & Tees, R.C. (1984). Cross-language speech perception:
Evidence for perceptual reorganization during the first year of life. Infant
Behavior and Development, 7, 49–63.
Werker, J.F. & Yeung, H.H. (2005). Infant speech perception bootstraps
word learning. Trends in Cognitive Sciences, 9(11), 519–527.
White, L. & Genesee, F. (1996). How native is near-native? The issue of
ultimate attainment in adult second language acquisition. Second Language
Research, 12, 233–365.
Widen, S.C. & Russell, J.A. (2008). Children acquire emotion categories
gradually. Cognitive Development, 23(2), 291–312.
Wierzbicka, A. (2011). Bilingualism and cognition: The perspective from
semantics. In V. Cook & B. Bassetti (Eds.), Language and bilingual
cognition (pp. 191–218). New York: Psychology Press.
Willits, J.A., Seidenberg, M.S. & Saffran, J.R. (2015). Distributional
structure in language: Contributions to noun-verb difficulty differences in
infant word recognition. Cognition, 132(3), 429–436.
Wilson, B., Smith, K. & Petkov, C.I. (2015). Mixed-complexity artificial
grammar learning in humans and macaque monkeys: Evaluating learning
strategies. European Journal of Neuroscience, 41(5), 568–578.
Winters, B.D., Saksida, L.M. & Bussey, T.J. (2008). Object recognition
memory: Neurobiological mechanisms of encoding, consolidation and
retrieval. Neuroscience and Biobehavioral Reviews, 32(5), 1055–1070.
Wohlgemuth, S., Adam, I. & Scharff, C. (2014). FoxP2 in songbirds. Current
Opinion in Neurobiology, 28, 86–93.
Woodward, A.L. (2009). Infants’ grasp of others’ intentions. Current
Directions in Psychological Science, 18(1), 53–57.
Wuillemin, D., Richardson, B. & Lynch, J. (1994). Right hemisphere
involvement in processing later-learned languages in multilinguals. Brain
& Language, 46, 620–636.
Xanthos, A., Laaha, S., Gillis, S., et al. (2011). On the role of morphological
richness in the early development of noun and verb inflection. First
Language, 31(4), 461–479.
Xiao, Y., Brauer, J., Lauckner, M., Zhai, H., Jia, F., Margulies, D.S. &
Friederici, A.D. (2016). Development of the intrinsic language network in
preschool children from ages 3 to 5 years. PLoS One, 11(11), e0165802.
Xu, F. & Tenenbaum, J.B. (2007). Word learning as Bayesian inference.
Psychological Review, 114(2), 245–272.
Yang, C.D. (2004). Universal grammar, statistics or both? Trends in
Cognitive Sciences, 8(10), 451–456.
Yang, C.D. (2013). Ontogeny and phylogeny of language. Proceedings of the
National Academy of Sciences of the United States of America, 110(16),
6324–6327.
Yeni-Komshian, G.H., Flege, J.E. & Liu, S. (2000). Pronunciation
proficiency in the first and second languages of Korean-English.
Bilingualism: Language and Cognition, 3(2), 131–149.
Yoshinaga-Itano, C. (2003). From screening to early identification &
intervention: Discovering predictors to successful outcomes for children
with significant hearing loss. Journal of Deaf Studies and Deaf Education,
8(1), 11–30.
Yu, A.C. & Margoliash, D. (1996). Temporal hierarchical control of singing
in birds. Science, 273(5283), 1871–1875.
Yuan, S. & Fisher, C. (2009). ‘Really? She blicked the baby?’: Two-year-
olds learn combinatorial facts about verbs by listening. Psychological
Science, 20(5), 619–626.
Zhang, Y., Kuhl, P.K., Imada, T., Kotani, M. & Tohkura, Y. (2005). Effects
of language experience: Neural commitment to language-specific auditory
 patterns. Neuroimage, 26(3), 703–720.
Zhang, Y.W., Jin, X.M., Shen, X.M., Zhang, J.M. & Hoff, E. (2008).
Correlates of early language development in Chinese children.
International Journal of Behavioral Development, 32(2), 145–151.
Author Index
Abbot-Smith, K., 15–16, 165, 246, 251, 254–255, 278
Abe, K., 228
Abelev, M., 39
Abrahamsson, N., 75
Abutalebi, J., 81
Adam, I., 284
Adams, A.M., 101
Adams, R.J., 216
Adelstein, A., 80
Adger, D., 228
Aitchison, J., 24, 131
Akhtar, N., 96, 226, 237
Aksu-Koç, A., 178
Albin, D.D., 88
Albirini, A., 188
Albright, A., 181
Alibali, M.W., 141
Altmann, S., 31–32
Ambridge, B., 104, 111, 114, 189, 212, 223, 224–225, 227, 250, 251,
257, 260, 275, 283
Anderson, D.R., 95
Anderson, J., 155
Anderson, R.C., 9
Anglin, J.M., 160, 161tab, 190
Annett, C., 33
Arbib, M.A., 99
Archer, S.L., 128
Archibald, L.M.D., 101
Argenti, A.M., 79
Argus, R., 194
Arlman-Rupp, A., 95
Arroyo, M.E., 96
Arunachalam, S., 96
Asano, T., 41
Aschersleben, G., 130
Aschner, J.L., 67
Aslin, R.N., 47, 48, 51tab, 87, 127, 128, 131, 132, 134–135, 136, 137,
140, 273, 277
Atkinson, M., 228, 229
Au, T.K.F., 163
Austin, J., 224
Bachelard, G., 13
Backley, P., 106
Baddeley, A.D., 101
Baillargeon, R., 124, 216
Baillie, C., 106
Baker, N., 103
Baker, P., 120
Bakker, I., 189
Bakker, M., 277
Bannard, C., 243, 257
Bard, B., 95
Barnes, J., 127
Barnes, S., 90
Barnett, R.K., 88
Baron, J., 201
Barr, R., 95
Barrett, M., 241
Barrière, M., 6
Bartlett, E., 161
Basche, R.A., 179
Bates, E., 5, 67, 96, 150fig, 156, 158, 236, 252
Bateson, P., 208
Baumwell, L., 8
Bavelier, D., 75, 76, 78, 81, 82
Bavin, E.L., 114
Bayes, Thomas, 171
Becker, J.A., 154
Becker, S., 111, 279, 280
Beckers, G.J.L., 50tab, 51tab
Behne, T., 237, 238–239, 240
Behrend, D.A., 239
Behrens, H., 240, 250, 281
Bekoff, M., 53
Bélanger, J., 171
Bellugi, U., 81, 102, 278
Benasich, A.A., 276fig, 277
Benelli, B., 179
Berent, I., 199
Bergeson, T.R., 89
Berk, S., 224
Berko, J., 12, 107, 179, 194, 195, 203, 255, 278
Berman, R., 114, 194, 196, 278, 279, 280, 281
Bernicot, J., 97, 100, 101
Bertoncini, J., 6, 139
Bertrand, J., 160
Berwick, R.C., 51tab, 52, 211, 218, 225, 228
Best, C.T., 126, 127
Bettinardi, V., 81
Bialystok, E., 75, 76, 77, 78, 81–82, 274
Bidgood, A., 258
Bird, G., 99
Birdsong, D., 76
Birjandi, P., 113, 114
Bishop, D.V.M., 101, 267, 284
Bjork, E.L., 154
Blair, E., 170
Bleakney, D.M., 107
Blom, E., 84
Bloom, L., 101, 156, 164
Bloom, P., 8, 10, 39, 101, 154, 157, 159, 160
Blossom, M., 142
Bly, B.M., 165
Blything, R., 254
Bohannon, J.N., 87, 94, 105, 106
Bolhuis, J.J., 51tab
Bolton, P., 101
Bonamo, K.M., 106
Bonatti, L.L., 182
Bongaerts, T., 66
Booth, A.E., 169, 171
Borgström, K., 156
Bornstein, M.H., 8, 163
Boroditsky, L., 163
Borovsky, A., 81
Bosch, L., 66
Bostow, A., 113
Botha, R.P., 21, 232
Bouquet, F., 46
Boustagui, E., 76
Bowerman, M., 246
Bowey, J.A., 215
Boytsova, J., 189
Braginsky, M., 174
Braine, M.D.S., 90, 242–243
Brakke, K.E., 37, 43, 44, 45, 48–49, 50tab, 51tab
Brand, R.J., 238
Brandt, S., 249
Brauer, J., 102, 277
Bräuer, J., 40
Brent, M.R., 159, 160
Bridges, K., 101
Britain, D., 228, 229
Broca, P., 71
Broen, P.A., 88, 100, 131
Brooks, D.I., 39, 51tab
Brooks, P.J., 89, 90, 112, 246, 247, 252, 257, 260, 263
Brotman, N., 87
Brown, Roger, 2, 3, 18–19, 50tab, 72, 90, 102–103, 104–105, 111, 155,
240–241, 278
Brown, S.D., 45
Bruer, J.T., 61, 62, 79
Bruner, J., 237
Bryant, P., 201
Bryden, M.P., 80
Buchan, H., 91
Buck-Gengler, C.J., 199
Bull, R., 6
Bussey, T.J., 165
Buttelmann, D., 49
Bybee, J.L., 249, 251, 278
Byers-Heinlein, K., 5
Cachia, A., 277
Čadková, L., 44
Call, J., 38, 39, 40, 49, 237, 238, 239
Callanan, M., 226
Cameron-Faulkner, T., 96, 114, 244, 257
Campbell, R.N., 63, 270
Campos, J.J., 18
Campos, R., 130
Can, D.D., 89
Cantrell, L., 167
Cappa, S.F., 81
Caramazza, A., 129tab
Carden, G., 125fig
Carey, S., 38, 40, 51tab, 160, 161, 170
Carpenter, M., 49, 165, 237, 238–239, 240
Carson, R.E., 45, 50tab
Cassidy, K.W., 141, 142
Catlin, J., 153
Catmur, C., 99
Cattell, R., 97, 98
Cazden, C.B., 102
Cerchio, S., 51tab
Cerri, G., 99
Cession, A., 90
Chafetz, J., 252
Chang, F., 254
Chasin, J., 149
Chater, N., 134
Chen, J., 49
Cheney, D.L., 33
Chernigovskaya, T., 189
Cherryman, J., 6
Cheshire, J., 30
Chiat, Shula, 279, 280, 281
Cho, J.R., 199–200
Choi, S., 163
Chojnowska, C., 276fig, 277
Cholewiak, D.M., 51tab
Chomsky, Noam, 5, 18–22, 23, 24–25, 30, 33, 49, 52, 53–54, 66, 77, 87,
96, 97–98, 111–112, 132, 134, 154, 178, 206–207, 208, 209, 210, 214,
215, 216, 217, 218, 220, 221, 222, 223, 225, 227, 228, 230–233, 266,
267, 272, 281
Choudhury, N., 276fig, 277
Chouinard, M.M., 105, 113
Chow, K.L., 62
Christensen, J.B., 129tab
Christiansen, M.H., 227, 284
Cimpian, A., 170
Clahsen, H., 181, 228, 229
Clancy, P., 114
Clark, A., 221
Clark, E.V., 8, 97, 100, 101, 105, 113, 151tab, 152, 153, 154, 194, 195,
196, 214, 215, 227, 266, 279, 280, 281
Clark, J.M., 155
Clark, R., 100
Clarke, E., 33
Clausner, T.C., 153
Cleave, P., 111
Coffey, S.A., 81
Coffey-Corina, S., 130
Cole, E.B., 87
Collins, F.S., 267
Colunga, E., 167
Conboy, B.T., 130
Congdon, E.L., 164
Conley, J.J., 207
Connell, B., 154, 156fig
Conti-Ramsden, Gina, 279, 281
Cook, R., 99
Cook, V.J., 211
Cooper, R.G., 98
Cooper, R.P., 87, 139
Copan, H., 125fig
Coppieters, R., 75
Core, C., 101
Corina, D., 81
Cote, L.R., 163
Courage, M.L., 216
Cowan, N., 90
Cowie, F., 218
Cowper, E.A., 228
Cox, N.J., 267
Crain, S., 218, 224, 275
Crane, C.C., 239
Crawford, M.J.L., 62
Croft, W., 153
Cross, T.G., 240
Csibra, G., 124
Curno, T., 201
Curran, M., 111
Curtin, S., 89, 128
Curtiss, S., 66, 80, 83
Cushman, F., 48, 137
Cyphers, L., 8
Dąbrowska, E., 21, 188, 252, 259
Dale, P.S., 96, 150fig, 156, 158, 252, 259
Danko, S., 189
Dapretto, M., 154, 163
Daum, M.M., 102
Davidson, D., 170
Dawson, N., 107, 111
de la Mora, D.M., 51tab
de Saussure, F., 9, 33
De Schonen, S., 130
De Silva, S., 37, 50tab
de Waal, F.B.M., 37
Deacon, H., 193
Deacon, S.H., 201
Debose, C.E., 89
Deboysson-Bardies, B., 88
DeCasper, A.J., 120
Deguchi, T., 126
Dehaene, S., 79, 81, 277
Dehaene-Lambertz, G., 46, 277, 283
DeHart, G.B., 98, 247
DeKeyser, R.M., 75, 76
Delle Luche, C., 137
Demetras, M.J., 106
Demuth, K., 246
Denninger, M.M., 103
Depaolis, R.A., 137
Derksen, E.S., 66
DeVilliers, J., 243
DeVilliers, P., 243
Devin McAuley, J., 89
Diesendruck, G., 157
Diessel, H., 240, 244, 249, 257
Dilley, L.C., 89, 139, 140
Dispaldro, M., 179
Dittmar, M., 15–16, 246, 254–255, 278
Dobbertin, B.N., 167
Dodson, K., 257
Donovan, W., 92
Dooling, R.J., 45, 50tab
Dore, J., 241
Douglas, J.M., 103
Dow, K.A., 130
Dromi, E., 156, 157fig
Drozd, K.F., 224
Druss, B., 141, 142
Du, L., 39
Dubois, J., 277
Duffy, H., 137
Dummett, M., 244
Dunham, F., 237
Dunham, P.J., 237
Dunn, J., 88
Dupoux, E., 79, 81, 277
Durrant, S., 137, 165
Dworzynski, K., 101
Dye, C.D., 181
Dye, M., 188
Eadie, P.A., 103
Echols, C.H., 88
Eichen, E.B., 81
Eimas, P.D., 45, 122–123, 125fig
Eisengart, J., 15, 253, 278
El Tigi, M., 76
Elbert, M., 103
Elbert-Perez, E., 113
Ellis, A.W., 67
Elman, J.L., 169, 227
Elsabbagh, M., 130
Emberson, L.L., 277
Emmorey, K., 81
Emslie, H., 101
Engel, K., 128
Epicurus, 272
Erbaugh, M.S., 114
Estes, K.G., 141
Etlinger, G., 51tab
Evans, J.L., 141
Evans, N., 210
Everett, D., 228
Eversley, J., 120
Eysenck, M.W., 153
Fadiga, L., 99
Fais, L., 128
Fancher, R.E., 207
Fantz, R.L., 123
Farrant, B.M., 93
Farrar, M.J., 106
Fazio, F., 81
Federico, A., 106
Fedorova, O., 89
Fein, D., 103
Fenn, K.M., 228
Fenson, L., 150fig, 156, 158, 252
Ferguson, C.A., 89, 149
Fernald, A., 88, 89, 93, 95, 96, 100
Fernandes, K.J., 47, 141
Fey, M.E., 103, 111
Fields, W.C., 284–285
Fifer, W.P., 139
Finn, A.S., 132
Fischer, A., 81
Fischer, J., 38
Fiser, J., 136, 273
Fisher, C., 15, 252, 253, 254, 255, 263, 278
Fitch, T., 49
Fitch, W.T., 37, 49, 50tab, 53–54, 228
Fleener, C., 40
Flege, J.E., 66, 127
Fletcher, P., 199–200
Floccia, C., 137
Floor, P., 96
Fluck, M., 215
Flynn, V., 96
Fodor, J.A., 221, 275
Fogassi, L., 99
Fong, C., 142
Forrest, K., 103
Fosha, D., 87
Fougeron, C., 139
Fouts, D.H., 37
Fouts, R.S., 37
Foygel, R., 142
Francis, W.M., 198
Frank, M.C., 174
Frankenfield, A., 170
Freedman, J., 226
Freedman, R.D., 62
Freudenthal, D., 258
Freyd, P., 201
Friederici, A.D., 81, 189, 277
Friedman, J.T., 276fig, 277
Fromkin, V., 66, 83
Fry, D.B., 138
Fu, P., 153
Fukui, I., 88
Fulbright, R.K., 181
Furrow, D., 106
Gagliardi, A., 284
Gagné, C.L, 196
Gahl, S., 226, 227
Gallaway, C., 106
Galles, N.S., 81
Gallese, V., 99
Galton, Francis, 207
Gampe, A., 96, 102
Ganger, S., 87, 159, 160
García-Sierra, A., 87
Gardner, B.T., 37, 41, 44, 50tab, 51tab
Gardner, R.A., 37, 41, 44, 50tab, 51tab
Garnica, O.K., 88
Garrett, M., 221
Gathercole, S.E., 97, 101
Gazso, D., 189
Geiger, V., 200
Gelman, S., 113, 153, 163, 196
Genesee, F., 76
Gentner, D., 163, 165, 249
Gentner, T.Q., 228
Gerard-Ngo, P., 247
Gerken, L., 100, 127
Gerrits, E., 127
Gershkoff-Stowe, L., 93, 149, 154, 155, 156fig, 161, 169
Gershwin, I., 285
Gertner, Y., 15, 253, 278
Gervain, J., 5, 136
Gibson, D.J., 164
Gibson, E., 212
Gillette, J., 87
Gillis, S., 179–180, 201
Ginzburg, J., 111
Girard, M., 39
Gjerlow, K., 71, 81, 99
Gleitman, H., 30, 87, 90, 263
Gleitman, L.R., 30, 87, 90, 278
Goldberg, A.E., 260
Goldberg, E., 96
Goldberg, R., 162, 246
Goldfield, B.A., 156
Goldin-Meadow, S., 94, 96, 218, 239
Goldman, H.I., 8
Goldstein, D.B., 267
Golinkoff, R.M., 15, 89, 111, 112, 164, 168tab, 171, 253, 280, 281
Gombert, J.E., 7
Gomes, G., 161
Goodluck, H., 104
Goodman, J.C., 5, 67, 158, 259
Gopnik, A., 160
Gordon, P., 197–199, 252
Gordon, R.G., 21, 209
Goslin, J., 137
Gottfried, G., 153
Gou, Z., 276fig, 277
Graf Estes, K., 137, 138
Graham, S.A., 157
Gredebäck, G., 277
Greenberg, J.H., 210
Greer, R., 39
Grela, B., 95
Grieve, R., 63
Griffiths, P., 41
Grimshaw, G.M., 80
Gropen, J., 162
Guasti, Marisa Teresa, 279, 280, 281
Gudeman, R., 247
Guerts, Bart, 279
Gustavson, K., 7
Gutfreund, M., 90
Hadler, M., 181
Haggan, M., 113
Hahn, E.R., 93, 149, 161, 167
Hakuta, K., 75, 77, 78, 81–82, 274
Hale, B., 245
Halgren, E., 81
Hall, D.G., 171, 173
Hall, S., 94
Hallinan, E.V., 238
Hamilton, M., 114
Hamlin, J.K., 238
Hampton, J.A., 153
Hanlon, C., 104–105, 111
Hanna, J., 189
Hao, M., 163
Happé, F., 18, 101
Hare, B., 40
Harley, B., 76
Harrington, M., 75
Harris, M., 149
Hart, B., 92–93
Hart, D., 76, 93
Hartung, J., 156
Harweth, R.S., 62
Haryu, E., 164, 165, 167, 170
Haskell, T.R., 199
Hatrak, M., 81
Hauser, M., 47, 48, 49, 51tab, 53–54, 137, 228
Hay, J.F., 137
Hayashi, A., 126
Hayes, B., 181
Hayes, C., 36–37, 44
Hayes, K.J., 36–37, 44
Healey, P., 111
Healy, A.F., 199
Heath, S.B., 112–113
Hebden, J.C., 277
Hecht, B.F., 194, 196
Heekeren, H.R., 81
Heibeck, T.H., 38
Helmick, A.L., 107
Henderson, L., 215
Hensch, T.K., 66, 83
Herman, L.M., 36, 51tab
Hermes, D.J., 66
Hernandez, A.E., 67
Hernstein, R.J., 209
Herschensohn, J.R., 74, 79, 82
Herscovitch, P., 45, 50tab
Hertz-Pannier, L., 277
Heyes, C., 99
Hickok, G., 181
Hill, E.A., 137
Hills, T., 88
Hindorff, L.A., 267
Hirsh-Pasek, K., 15, 89, 105, 106, 112, 141, 142, 164, 171, 174, 253
Hockett, Charles, 30, 31–32, 33, 35, 36, 44
Hoff, E., 66, 91, 93, 97, 101, 275
Hoff-Ginsberg, E., 103
Hoffnung, M., 77
Hoffnung, R.J., 77
Hofmeister, C., 156
Hohenberger, A., 130
Holder, M.D., 51tab
Holland, A., 164
Hollander, M., 162, 183–184
Holler, J., 94
Hollich, G., 168tab
Holt, E., 116
Holtheuer, C., 107
Hood, L., 101
Horn, P., 81
Horst, J.S., 161, 167
Houston, D., 139
Houston-Price, C., 107, 111
Huang, H.F., 139
Huang, Y., 67, 77–78, 228
Hubel, D.H., 59–60, 61–62
Hübner, M., 188
Huddleston, R., 246
Hunsicker, D., 218
Hunter, D.J., 267
Hupp, J.M., 167, 169
Hurry, J., 201
Hurtado, N., 93
Huston, A.C., 95
Hutchinson, J.E., 168tab
Huttenlocher, J., 92, 154
Hyams, N., 213, 228, 231
Hyltenstam, K., 75
Ibbotson, P., 272
Imada, T., 128
Imai, M., 164, 165, 167, 170
Immelmann, K., 62
Inhelder, B., 215
Ioup, G., 76
Israel, M., 252
Iverson, P., 46, 126, 239
Jackendoff, R., 21, 47, 49, 52, 54, 208, 228, 274
Jaeger, J.J., 181
Jakobson, R., 87
Janik, V.M., 50tab
Jarmulowicz, L., 89, 182
Jassik-Gerschenfeld, D., 6
Jaswal, V.K., 170
Jespersen, O., 113
Jezzard, P., 81
Jia, F., 277
Jiang, Y., 163
Jin, X.M., 93
Jobse, J., 95
Johansen Ijäs, K., 194
Johnson, C., 252
Johnson, C.J., 155
Johnson, E.K., 136, 139, 140
Johnson, J.S., 71, 74–75, 76fig, 77, 78
Johnson, M.L., 95
Johnson, S.P., 47, 136, 141, 273
Jolly, H.R., 179
Jones, C., 91
Jones, S.S., 166fig, 167, 168tab, 170
Joosten, J., 95
Joyce, J., 12
Judge, P.G., 39
Juffs, A., 75
Jusczyk, A.M., 130, 139, 140
Jusczyk, P.W., 45, 124, 125fig, 130–131, 132, 139, 140, 141, 142, 143,
144
Juvrud, J., 277
Kaduk, K., 276
Kahana-Kalman, R., 8
Kahnemuyipour, A., 279
Kail, R., 18
Kako, E.T., 252
Kam, C.L.H., 132
Kamenskaya, T.O., 179
Kaminski, J., 38, 40
Kang, C., 158
Kapfhamer, J., 48, 137
Kaplan, B.J., 103
Karlsson, F., 177
Karmiloff, K., 97
Karmiloff-Smith, A., 97, 130
Kaufholz, P.A.P., 66
Kauschke, C., 156
Keane, M.T., 153
Keating, P., 139
Kellog, Luella, 36
Kellog, Winthrop, 36
Kemler Nelson, D.G., 141, 142, 143, 170
Kemmerer, D.L., 181
Kempe, V., 89, 112, 263
Kennedy, L., 141, 142
Keren-Portnoy, T., 137
Kessel, F.S., 19
Ketrez, F.N., 134
Key, A.P.F., 67
Khalak, H.G., 181
Kidd, E., 94, 114, 188, 251, 279, 280
Kiehl, K.A., 45
Kim, Y., 163
King, S.L., 50tab
Kintsch, W., 153
Kinzler, K.D., 164, 167
Kiparsky, P., 197
Kiritani, S., 126
Kirjavainen, M., 188
Kisilevsky, B.S., 139
Kita, S., 252
Kitamura, C., 91
Klatt, D.H., 50tab, 51tab
Kline, M., 246
Klingler, S.L., 252
Kluender, K., 45, 50tab
Koepke, A., 39
Kohno, M., 126
Koizumi, H., 46
Kojima, T., 41
Kollia, B., 181
Koryé, A., 230
Koskenniemi, K., 177
Kotani, M., 128
Kovačič, D., 46
Krashen, S., 66, 83
Krcmar, M., 95
Krott, A., 196
Kruger, A.C., 237
Kubota, K., 41
Kuchirko, Y., 103
Kuczaj, S.A., 51tab, 245
Kučera, H., 9, 198
Kuhl, P.K., 45, 50tab, 87, 95, 126, 127, 128, 130, 131, 145, 277
Kulcsar, B., 107
Kwon, S., 37, 50tab
Laaha, S., 174, 179–180
Laakso, M., 107
Laalo, J., 194
Ladefoged, P., 121
Lafleur, R., 127
Lalonde, C., 124, 126
Lam, C., 91
Lambert, E.W., 67
Landau, B., 166fig5, 167, 168tab, 170
Lane, H., 64
Lane, N.M., 196
Langacker, R.W., 236
Langer, S., 18, 97, 269, 275, 295
Lanter, J.A., 179
Lappin, S., 221, 227, 266
Larsen-Freeman, D., 22
Lasnik, H., 217
Lauckner, M., 277
Lawson, D.S., 81
Lawson, K.R., 238
Le Bihan, D., 79, 277
Leavitt, L., 92
Lederer, A., 87
Lee, D., 75
Lee, K., 139
Lee, L.C., 102
Lee, Y., 163
Lehrman, D.S., 208
Leigh, A.H., 39
Lenneberg, Eric, 59, 66, 71, 72, 73, 77, 79, 82
Leonard, M.K., 81
Leopold, W.F., 18, 151–152
Leung, K., 199–200
Levendoski, E.K., 39
Levin, I., 201
Levine, S.C., 96, 164
Levinson, S.C., 210
Levitt, A., 125fig
Lewis, J.D., 227
Lewis, L.B., 257
Lew-Williams, C., 135, 137
Li, H., 158
Li, J., 216
Li, L.J., 164
Li, P., 67, 163, 259
Liang, W., 158
Liddle, P.F., 45
Lidz, J., 226, 284
Liebal, K., 96, 238–239, 240
Lieberman, A.M., 80, 81
Lieberman, P., 44
Liederman, J., 127
Lieven, E.V.M., 15–16, 113, 184, 212, 223, 227, 241–242, 244, 245,
246, 249, 250–251, 254–255, 257, 259, 273, 278, 279, 280
Lightbown, P.M., 101
Lightfoot, D., 96, 213
Lillo-Martin, D., 218
Lin, K., 95
Lindgren, M., 156
Linebarger, D.L., 96
Linnaeus, Carl, 63
Liu, H., 95, 128, 130, 158
Liu, S., 66
Liu, Y., 163
Livingston, P., 39
Lockwood, A.H., 181
Long, M.H., 66
Lord, C., 102
Lovassy, N., 189
Love, T., 181
Lukacs, A., 188, 189
Lukens, H.T., 18, 87
Lunn, J., 277
Luo, Y.Y., 216
Lust, B., 218, 224
Lyn, H., 38
Lynch, J., 81
Lyytinen, H., 199
Lyytinen, P., 199
MacDonald, M.C., 132, 199
MacGregor, L.J., 189
MacKain, K., 88
MacKinnon, G.E., 80
Mackintosh, N.J., 209, 267
MacWhinney, Brian, 3, 19, 178, 221, 222, 224, 236, 243, 251, 281
Maestripieri, D., 40
Magen, H.S., 181, 189
Maital, S., 163
Maitre, N.L., 67
Maki, A., 46
Malle, B.F., 97
Malloy, M.R.C., 45, 50tab
Maltempo, A., 253
Mameli, M., 208
Mandalaywala, T., 40
Mandel, D.R., 142, 143
Mangin, J.F., 277
Mann, V., 127
Männel, C., 277
Manning, C.D., 134
Mannle, S., 237
Manolio, T.A., 267
Maratsos, M., 184, 208, 218, 247, 248, 251
Marchetto, E., 182
Marchman, V.A., 93, 156, 174
Marcus, G.F., 47, 136, 141, 183–184, 187, 217, 218
Margoliash, D., 51tab, 228
Margulies, D.S., 277
Margulis, C., 164
Mark, M., 105
Markman, E.M., 37, 38, 39, 50tab, 168tab, 170
Markow, D.B., 164, 169
Markson, L., 8
Marsden, R.E., 18
Marsh, H.L., 39
Marshall, G., 107, 279, 281
Marshall, P.J., 99
Martin, A., 47
Martin, N., 155
Marx, O., 272
Maslen, R.J.C., 184, 259
Masur, E.F., 96, 99, 101–102
Matsuo, A., 252
Matsuzawa, T., 41
Matthews, D., 116
Matthews, P.H., 228
May, L., 5
Mayberry, R.I., 80, 81
Maye, J., 128
Mayer, N., 41, 127
Mayeux, L., 96
Mazzie, C., 100
McBride-Chang, C., 158, 199–200
McClelland, J.L., 180, 185–187
McComb, K., 50tab
McDonough, L., 152
McEwen, F., 101
McGilvray, J., 213, 216, 232, 271fig
McLaren, J., 106
McLeod, P.J., 88
McMurray, B., 39, 51tab, 127
McRoberts, G.W., 126, 127
Mead, M., 113
Meadow, K.P., 79, 218
Medawar, P.B., 71
Medvedev, S., 189
Mehler, J., 6, 46, 79, 81, 139
Meltzoff, A.N., 99, 100, 160
Mencl, W.E., 181
Menn, L., 151, 199
Menyuk, P., 126
Meriaux, S., 277
Mervis, C.B., 152, 153, 160, 168tab
Messer, D.J., 89
Michaelson, J.J., 284
Midgley, L., 201
Mierzejewski, R., 127
Mietchen, D., 37, 50tab
Miles, L.H., 41
Miles, S., 92
Miller, B., 75, 78
Miller, J.D., 45, 50tab
Miller, J.L., 125fig
Miller, K.L., 179
Millett, A.L., 89
Minagawa-Kawai, Y., 277
Mintz, T.H., 171
Mironova, N., 89
Mishkin, M., 45, 50tab
Moerk, C., 100
Moerk, E.L., 100, 106
Moll, H., 237, 238, 239
Monaco, A.P., 267, 284
Mondragon, E., 49
Moon, C., 139
Moore, C., 106
Moore, M.K., 99
Morgan, J.L., 97, 106, 142, 279
Mori, K., 277
Morris, C., 170
Morrison, C.M., 67
Moselle, M., 76
Moss, C., 50tab
Mostofsky, S., 181
Mueller, K.L., 284
Mulcaster, Richard, 207
Mulford, R.C., 196
Müller, G., 189
Mumme, D.L., 95
Muncer, S.J., 51tab
Murchison, C., 18, 97, 269, 275, 295
Murofushi, K., 41
Murphy, B.W., 181
Murphy, J., 37, 43, 44, 45, 48–49, 50tab
Murphy, R.A., 49
Murphy, V.A., 49, 51tab, 75, 76, 198
Murray, C., 209
Murray, J., 125fig
Murray, J.C., 284
Nadig, A., 102
Nagy, W.E., 9, 99
Naigles, L., 93, 96, 103, 153, 252, 253, 279, 280, 281
Nakayama, M., 224, 275
Napoli, D.J., 228
Nash, H., 214
Nasrolahi, A., 113, 114
Nazareth, A., 280
Nazzi, T., 139, 140
Neiworth, J.J., 49
Nelson, K.E., 103, 105
Nelson, T., 130
Nemeth, D., 188
Neville, H.J., 75, 76, 78, 79, 81, 82
Newbury, D.F., 267, 284
Newman, A.J., 81
Newman, R.S., 113, 130, 136
Newport, E.L., 30, 45, 47, 48, 51tab, 71, 74–75, 76, 77, 78, 81, 82, 87,
90, 132, 134–135, 136, 137, 140, 273
Newschaffer, C., 102
Newsome, M., 139
Nicoladis, E., 189, 194–195, 196, 198
Nikolaev, A., 188
Ninio, A., 87, 171
Nippold, M.A., 200, 271fig
Nishi, K., 126
Nishimura, T., 36
Noble, C., 96
Norbury, C.F., 101, 281
Nunes, T., 201
Nurmsoo, E., 165
Nusbaum, H.C., 228
Oakes, L.M., 280
Oakley, Y., 149
Obler, L.K., 71, 81, 99
Ochs, E., 87, 114
Odean, R., 280
O’Doherty, K., 96
Oh, K., 163
Oh, S., 37, 50tab
Ohms, V.R., 50tab
Okada, H., 164, 165
Okanoya, K., 51tab
Olson, J., 96, 101–102
Opitz, A., 189
Orvos, H., 99
Ostendorf, M., 142
Otero, N., 93
Owens, R.E., 98
Özçalişkan, Ş., 94
Ozonoff, S., 102
Pacton, S., 201
Padden, D.M., 45, 50tab, 130
Padgett, R.J., 105
Painter, K., 163
Paivio, A., 155
Pal, A., 99
Pallier, C., 66, 79
Palumbo, L.C., 181
Pan, B.A., 151tab
Pancheva, R., 181
Papousek, M., 88
Paradis, J., 84, 189
Park, S.Y., 163
Parker, M., 201
Parsons, C.L., 103
Pascual, L., 163
Patterson, F.G., 41, 51tab, 95, 153, 185
Patterson, M.D., 165
Paul, R., 281
Paulesu, E., 81
Pearl, L., 208, 279, 280, 281
Peckham, D., 189
Pelucchi, B., 137
Pempek, T.A., 95
Peña, M., 46
Penner, S.G., 106
Pepperberg, I.M., 39, 40–41, 50tab, 51tab
Perani, D., 81
Pérez-Leroux, A.T., 279
Pershukova, A., 89
Peter, M., 254
Pethick, S.J., 150fig, 156, 158, 252
Petkov, C.I., 51tab
Philippon, A.C., 6
Phillips, J.R., 87, 279
Piaget, J., 21–22, 215
Piattelli-Palmarini, M., 22
Pierotti, R., 33
Pietroski, P., 218
Pilley, J.W., 39, 50tab, 51tab
Pine, J.M., 104, 212, 223, 224–225, 227, 241–242, 251, 254, 257, 258,
275
Pinker, Steven, 44, 47, 49, 52, 54, 90, 112, 114, 160, 162, 177, 182,
183–184, 185, 187, 199, 202, 217, 221, 228, 256
Planken, B., 66
Plato, 216–217, 274
Plaut, D.C., 218
Plomin, R., 101
Plunkett, K., 156, 179
Poline, J.-B., 79
Polka, L., 142
Pollak, S.D., 47
Porcelli, A.J., 165
Poremba, A., 45, 50tab
Post, K., 106
Post, K.N., 106
Poulin-Dubois, D., 157
Prado, E.L., 181
Premack, D., 157
Press, C., 99
Pretzlik, U., 201
Price, P.J., 142
Prince, A., 177, 185, 187
Pruden, S.M., 280
Pruitt, J., 130
Pullum, G.K., 208, 212, 222, 226
Pulvermüller, F., 189
Purver, M., 110
Pye, C., 113, 114
Quine, W.V.O., 9–10, 163, 164, 170, 171
Rabin, J., 193
Radford, A., 228, 229, 231, 243, 256
Raffaele, P., 38
Ramirez, N.F., 80, 81
Ramírez-Esparza, N., 87
Ramscar, M., 188
Rao, S.B., 136
Rapaport, S., 201
Räsänen, S.H.M., 251
Ratner, N.B., 130, 136
Ravchev, S., 142
Ravid, D., 200, 201, 279, 280
Reali, F., 227
Realpe-Bonilla, T., 276fig, 277
Redman, J., 216
Regel, S., 189
Regier, T., 226, 227
Reichard, U.H., 33
Reid, A.K., 39, 50tab, 51tab
Reid, V.M., 277
Reilly, J., 156
Reilly, S., 284
Rendle-Short, J., 107
Rescorla, L.A., 152, 163
Reznick, J.S., 136, 150fig, 156, 158, 252
Ribeiro, L.A., 7
Rice, M.L., 95
Richards, D.G., 36
Richardson, B., 81
Riggs, K.J., 164
Rigler, D., 66, 83
Rigler, M., 66, 83
Rijsdijk, F., 101
Risch, N.J., 267
Risley, T.R., 92–93
Rivas, E., 41–42
Rivera-Gaxiola, M., 126, 128, 130, 277
Rizzolatti, G., 99
Rochat, P., 18
Roche, A., 277
Roeper, T., 104
Rogers, S.J., 102, 153
Rohde, D.L.T., 218
Romberg, A., 136
Ronald, A., 101
Rondal, J., 90
Rosati, A., 40
Rosch, E.H., 153
Rosen, S., 46, 183–184
Rosenthal Rollins, P., 237
Rossum, D., 49
Rowe, M.L., 92, 94, 136, 239
Rowland, C.F., 21, 104, 114, 224–225, 251, 254, 257–258, 275, 279,
281, 283
Ruffman, T., 216
Rumbaugh, D.M., 37, 43, 44, 45, 48–49, 50tab, 51tab
Rumelhart, D.E., 180, 185–187
Rumney, L., 94
Rupra, M., 107
Russell, J.A., 153
Rymer, R., 66, 83
Rypma, B., 165
Sachs, J., 90, 95
Saffran, E.M., 155
Saffran, J.R., 47, 48–49, 121fig, 132, 134–135, 136, 137, 139, 140, 141,
145, 165, 237, 273, 279
Saiegh-Haddad, E., 201
Saito, A., 50tab
Saksida, L.M., 165
Salerno, R., 90
Salkie, R., 211
Sampson, G., 232
Samuel, A.G., 184
Samuels, M.C., 90
Samuelson, L.K., 161, 167, 169
Santelmann, L., 224
Santos Lotério, L., 161
Sarnecka, B.W., 179
Sasaki, M., 126
Satterly, D., 90
Saunders, R.C., 45, 50tab
Savage-Rumbaugh, E.S., 37, 38, 43, 44, 45, 48–49, 50tab, 51tab
Saxton, M., 87, 88, 105, 106, 107, 111, 115, 268
Sayialel, K.N., 50tab
Saylor, M.M., 96
Scaife, M., 237
Schachter, F.F., 87
Schachter, J., 75
Scharff, C., 284
Scheibmann, J., 249
Schiffer, R., 113
Schils, E., 66
Schj⊘lberg, S., 7
Schlesinger, I.M., 87
Schmidt, A., 161
Schmidt, C.L., 94, 238
Schmitt, C., 179
Schneiderman, M., 105
Scholz, B.C., 208, 212, 222, 226
Schoner, G., 124
Schouten, B., 127
Schutte, A.R., 167
Scofield, J., 239
Scovel, T., 66
Sebastián-Gallés, N., 66
Seibel, R.L., 47, 48, 137
Seidenberg, M.S., 38, 132, 136, 165, 189, 199, 203
Seidl, A., 139, 140, 280
Seifert, K.L., 77
Seitz, S., 103
Serres, J., 130
Sevcik, R.A., 37, 43, 44, 45, 48–49, 50tab, 51tab
Seyfarth, R.M., 33
Shady, M., 100
Shallcross, W.L., 238
Shankweiler, D., 181
Shannon, G., 50tab
Shattuck-Hufnagel, S., 139, 140, 142
Shatz, M., 97, 113
Shaw, G.B., 59
Sheinkopf, S.J., 97
Shen, X.M., 93
Shigematsu, J., 164
Shimpi, P.M., 96
Shinozuka, K., 50tab
Shinya, Y., 252
Shkolnik, A., 47
Shneidman, L.A., 96
Shtyrov, Y., 189
Shu, H., 158, 163, 199–200
Shukla, M., 277
Shultz, S., 47
Shulz, L., 40
Siegel, L.S., 201
Sigman, M., 277
Silva-Pereyra, J., 126, 128
Silverisenstadt, J., 127
Simon, T., 88
Simpson, A., 164
Singh, L., 127, 136
Sinnott, J.M., 46
Siqueland, E.R., 45
Sithole, N.M., 127
Skinner, B.F., 87, 98, 99
Slade, L., 216
Slobin, D.I., 100, 178, 182
Smith, K., 51tab
Smith, L., 154, 156fig
Smith, L.B., 166fig, 167, 168tab, 169, 170
Smith, N., 20, 21, 212–213, 218
Smith, N.A., 88, 91
Smith III, E.L., 62
Smith-Lock, K.M., 181
Snedeker, J., 170
Snow, C.E., 59, 87, 88, 91, 95, 100, 106, 237
Snowling, M., 214
Snyder, W., 213, 225, 231
Sobel, D.M., 97
Soderstrom, M., 89, 142, 279, 280
Soininen, M., 107
Soja, N.N., 170
Solovjeva, M., 189
Somashekar, S., 224
Sommerville, J.A., 239
Song, H.J., 252
Song, L., 93, 103
Song, S., 158
Song, Y.K., 163
Sontag, L.W., 5
Sorace, A., 71, 75
Sosa, A.V., 149, 151
Sousa-Lima, R.S., 51tab
Spaai, G.W.G., 66
Spaepen, E., 48, 153
Speares, J., 250
Spelke, E., 153, 164, 167, 170, 283
Spence, M.J., 5–7, 120
Spencer, A., 228, 229
Spieker, S., 88
Spier, E.T., 93
Srinivasan, M., 170
Sroufe, L.A., 98
St. Clair-Stokes, J., 87
Stahl, D., 214
Stanfield, C., 94
Stanowicz, L., 106
Stark, R.E., 2
Stefanski, R.A., 50tab, 51tab
Stemp, S., 87
Stern, D.N., 88
Stern, W., 99
Stevens, E., 126
Stewart, C., 103
Stewart, D.L., 62
Stich, S., 212
Stilwell-Peccei, J., 98
Stoeger, A.S., 37, 50tab
Stoel-Gammon, C., 149, 151
Stokes, S.F., 199–200
Strapp, C.M., 106, 107
Streeter, L.A., 125
Su, M., 158
Sully, J., 241
Sumner, M., 184
Suomi, S.J., 62
Suthers, R.A., 50tab
Suzuki, T., 225–226
Swensen, L.D., 103, 252
Swinney, D., 181
Szagun, G., 188
Szanka, S., 189
Szczerbinski, M., 252
Taeschner, T., 88
Taha, H., 201
Taine, H., 18, 39
Takagi, N., 127
Takizawa, O., 126
Tamis-LeMonda, C.S., 90, 93, 103
Tamis-LeMonda, M., 8
Tanner, J., 41
Tardif, T., 158, 163, 199–200
Taylor, N., 92
Tees, R.C., 126, 127tab
Teigen, K.H., 207
Tell, D., 170
ten Cate, C., 49, 50tab
Tenenbaum, E.J., 97
Tenenbaum, J.B., 134, 171, 172
Terrace, H., 41, 44
Thal, D.J., 96, 150fig, 156, 158, 252
Theakston, A.L., 114, 184, 224, 250, 251, 257, 259
Thelen, E., 124
Thiessen, E.D., 137, 139, 140, 273
Thompson, E., 125fig
Thurm, A., 102
Tiedemann, Dietrich., 18, 97, 269–270, 271, 275, 283, 295–297
Tinker, E., 164
Tohkura, Y., 128
Tomasello, M., 15–16, 40, 49, 96, 105, 111, 165, 184, 214, 224, 226,
237, 238–239, 240, 241, 242, 243, 244, 245–246, 247, 249, 250–251,
254–255, 256, 257, 259, 260, 262–263, 272, 278, 279, 280
Tomblin, J.B., 284
Tonkovich, H.M., 107
Torkildsen, J., 156
Toro, J.M., 51tab
Torres, C., 81
Trainor, L., 88, 91
Travis, L.L., 106
Treiman, R., 105
Trevarthen, C., 238
Troseth, G., 96
Truglio, R., 95
Tsao, F., 48, 128, 130, 137
Tsushima, T., 126
Turi, Z., 189
Twomey, K.E., 258
Uchikoshi, Y., 96
Ullman, M.T., 181, 183–184, 189, 279
Uriagereka, J., 217
Valian, V., 209, 210, 223, 267, 268, 278
Van Herwegen, J., 130
van Hessen, A., 127
Van Horne, A., 111
van Hout, Angelika, 279, 280
van Noorden, G.K., 62
Van Valin, R.D., 181
Vasilyeva, M., 92
Venuti, P., 163
Vigorito, J., 45
Vihman, M., 137, 144
Vijayan, S., 136
Villringer, A., 81
Vining, A.Q., 39
Vishton, P.M., 136
Visscher, P.M., 267
Vivanti, G., 102
Von Frisch, K., 34fig, 35, 36
von Humboldt, W., 22, 30, 97, 210, 214, 215
Vorster, J., 95
Vouloumanos, A., 45, 46, 47, 89
Vrij, A., 6
Vygotsky, L.S., 96
Vyt, A., 163
Wachtel, G.F., 168tab
Wade, T., 12
Walenski, M., 181
Wales, R., 240
Walker, D., 96
Wallace, R.I., 5
Walle, E.A., 18
Wallman, J., 44
Wang, S.H., 216
Wang, Z.P., 139
Wanner, Eric, 278
Warlaumont, A.S., 89, 182
Warren-Leubecker, A., 87, 94
Wartenburger, I., 81
Wasserman, E.A., 39, 51tab
Watanabe, D., 228
Waterfall, H., 92
Watt, W.C., 221
Waxman, S.R., 164, 169, 171, 173, 226
Weisberg, R.W., 155
Weisleder, A., 93, 96
Weiss, D.J., 45, 128
Weissenborn, J., 104
Wells, G., 90
Weppelman, T.L., 113
Werker, J.F., 5, 45, 46, 47, 66, 83, 88, 124, 126, 127, 127tab, 129, 136
Westermann, G., 277
Wexler, K., 212
Weyerts, H., 181
White, L., 76, 137
Widen, S.C., 153
Wierzbicka, A., 225
Wiesel, T.N., 59–60, 61–62
Wiley, E., 77, 78, 81–82, 274
Williams, S.L., 37, 43, 44, 45, 48–49, 50tab, 51tab, 171
Williamson, R., 94
Willis, C., 101
Willits, J.A., 165
Wilson, B., 51tab
Winters, B.D., 165
Wohlgemuth, S., 284
Wolz, J.P., 36
Wong, A., 199–200
Wood, C., 156
Wood, G.C., 252
Woodruff, G., 157
Woodsmall, L., 95
Woodward, A.L., 238
Wright, J., 95
Wuillemin, D., 81
Wyss, N., 95
Xanthos, A., 179–180
Xiao, Y., 277
Xie, X., 139
Xing, A., 163
Xu, F., 163, 171, 172, 183–184
Xuehua, L., 136
Yamana, Y., 179
Yang, C.D., 41, 275
Yankama, B., 218, 225
Ye, H.H., 139
Yee, E., 181
Yeeles, C., 149
Yeni-Komshian, G.H., 66, 129tab
Yeung, H.H., 129, 136
Yoshinaga, N., 225–226
Yoshinaga-Itano, C., 79
Young, C.R., 104, 257
Yu, A.C., 51tab
Yuan, S., 252, 255
Yudovina, J.B., 179
Yuille, A., 134
Yurovsky, D., 174
Zachrisson, H.D., 7
Zamunerb, T., 128
Zeno, 272
Zhai, H., 277
Zhang, J.M., 93, 128
Zhang, K., 139
Zhang, Y.W., 93, 158
Zhang, Z., 158
Zizak, O., 247, 260
Zuberbühler, K., 33
Zubrick, S.R., 93
Subject Index
abstract utterances, 244–245
abstraction, 244, 245
accusative case, 12, 252, 261, 302
acquirendum, 222
acquisition, language, 17–18, 21–22, 73–76, 114–115, 213, 215–216,
226–227, 278–281
see also milestones, language development; timelines, language
development
activation, word, 155, 156fig
adjectives, 171, 209, 301
Adult Directed Speech (ADS), 88–89, 91–92, 137, 142
see also Child Directed Speech (CDS); Infant Directed Speech
(IDS)
adverbs, 68, 301, 308
affixes, 190, 193–194, 291–292, 301
affricate sounds, 128, 301
age
acquisition of grammar, 59
first words, 7–8
first year see first year
milestones in language development, 8, 16–18, 67, 280
newborn infants, 5–7, 120
notation, 3
school years, 75, 77, 173, 182, 188, 195, 199–201, 214–215, 271,
306
timeline of language development, 282fig
vocabulary at five years old, 17tab
vocabulary at one year old, 149, 151
agents, 90, 243, 254, 278, 301
Akbar the Great of India, 63
Alex (parrot), 40–41, 50tab, 51tab, 54
American Sign Language, 80, 81–82
amplitude, 138, 140
Anal and Kamala, 65tab
analogy formation, 248–249
animals
apes, 36, 37, 38, 43, 44
bees, 34fig, 35, 36
birds, 31, 33–34, 39, 40–41, 49, 50tab, 51tab, 54, 228
bonobos, 37–38, 39, 40, 41, 42, 43, 44, 50tab, 51tab, 55
categorical perception, 45–47
cats, 50tab, 59–62
chimpanzees, 36–40, 41–45, 50tab, 51tab, 54, 239
chinchillas, 45
communication, 33–36
comprehension of spoken English, 43
dogs, 38–40, 50tab, 51tab
elephants, 37, 50tab
faculty of language – broad (FLB), 49, 52fig
faculty of language – narrow (FLN), 49, 52fig, 228
gibbons, 33, 34–35
gorillas, 54
grammar, 41–45, 48–49
honey bees, 34fig, 35, 36
language, 23, 28–29
lexigrams, 37–38
linguistic limitations, 44–45
monkeys, 28, 39, 45, 47, 48–49, 50tab, 51tab, 53, 99, 136–137,
165, 273
parrots, 39, 40–41, 50tab, 51tab, 54
pigeons, 39, 51tab
primates, 36–40, 41–45, 48–49, 50–51tab
rats, 49
songbirds, 51tab, 228
statistical learning, 47–48
symbolic expression, 297
talking versus sign language, 36–37
tamarin monkeys, 48–49, 51tab, 136–137, 273
teaching words to, 36–41, 45
western gulls, 33–34
zebra finches, 49, 51tab
Anna (feral child), 65tab
apes, 36, 37, 38, 43, 44
see also bonobos; chimpanzees
Arabic, 188
arbitrariness, 30, 31, 33, 36, 38, 44
articles, 44, 171, 240, 301
associative learning, 167, 169, 171, 272
Attentional Learning Account (ALA), 167, 169
autism, 102, 237
auxiliary verbs, 220–221, 222, 224, 292, 301–302
babbling, 17tab, 295
basic level bias, 168tab, 170
Basic Property of language, 228
Bayes’ theorem, 171
bees, 34fig, 35, 36
behaviour in phrases, 13, 14
biases, word learning
associative learning, 167, 169, 171
basic level, 168tab, 170
gavagai problem, 163, 164, 170, 171–172
mutual exclusivity, 168tab, 169–170
nouns, 163–165
overriding of, 169–170
shape, 153, 166fig, 167–169, 170
taxonomic, 153, 168tab, 170
verbs, 163, 164, 165–166
whole object, 164–165, 167, 168tab, 169, 170, 171
bi-grams, 225
bilingualism, 280
birds, 31, 33–34, 39, 40–41, 49, 50tab, 51tab, 54, 228
bitransitive verbs, 246, 293, 302
blocking hypothesis, 182–184
bonobos, 37–38, 39, 40, 41, 42, 43, 44, 50tab, 51tab, 55
brain
Broca’s area, 81, 99
critical periods, 79, 81, 82
event-related potentials (ERPs), 126, 128, 276–277
functional Magnetic Resonance Imaging (fMRI), 277
 further reading on, 283–284
hemispheres, 45, 71, 79, 81, 99, 277
mirror neurons, 99, 102
near-infrared spectroscopy, 277
single versus dual-route theories, 189
broadcast transmission, 31, 44
Broca’s area, 81, 99
Cantonese language, 199, 280
cases, 12, 254, 302
categorical perception, 45–47, 50tab, 122–126, 136
categorization, 17tab, 169, 170, 248, 249–251, 262
category errors, 151, 152–154, 155
cats, 50tab, 59–62
Chelsea, 80
Child Directed Speech (CDS)
characteristics, 88–94
definition, 302
dynamic register, 91
frequency of inflections, 182
individual differences, 92–94
morphology, 89–90
phonology, 88
responsiveness to, 295
socio-economic status, 92–93, 94
syntax, 90
the term, 87–88
universal grammar (UG), 215
universality, 112–114
vocabulary, 89
Child Language Data Exchange System (CHILDES), 19, 25, 222, 280,
281
chimpanzees, 36–40, 41–45, 50tab, 51tab, 54, 239
chinchillas, 45
Chinese language, 163, 164, 181, 210
Chomsky, Noam see author index
CLAN (Computerized Language Analysis), 19
clarification questions (CQs), 111
clauses, 141–143, 222, 302
coarticulation, 121
cognitive development, 17tab, 18, 98–99
cohesion principle, 164
collaborative engagement, 238–240
common nouns, 39, 302
communication, 29
competence, 154, 213, 303
complement noun phrases, 211, 212
complements, 302
complex compounds, 197–199
compounding
complex compounds, 197–199
early compounds, 194–196
nature of, 190, 191–192
relating parts to whole, 196
comprehension, 50tab, 149, 150fig, 152, 162, 282fig, 296
computational modelling, 171–172, 279
Computerized Language Analysis (CLAN), 19
concepts, 155, 156fig
concrete utterances, 244–245
connectionism, 180, 185–187, 189, 203
conservative learning, 256
consonants, 48, 121, 139, 299, 303
content words, 68
contrastive discourse, 105–110
conversation, 239–240
copulas, 220–221, 222, 224, 257, 292, 303
core-periphery, linguistic diversity, 210–211
corrective input
contrastive discourse, 105–110
imitation, 115
negative feedback, 110–111
no negative evidence assumption, 103–105, 111, 112
recasts, 102–103
cotton-top tamarin monkeys see tamarin monkeys
count nouns, 169, 171, 303
counting, 215–216
creativity, 30–31, 32, 36
critical periods
brain, 79, 81, 82
cat’s eyes, 59–61
cut-off, 62, 71, 75, 77, 78, 79, 83
deaf people, 73, 79–81
for different aspects of language, 66–71
Eric Lenneberg’s hypothesis, 59
identification, 61
linguistic deprivation, 62–73, 80–81
nature of, 58–59, 61–62
offset, 59, 72, 78, 79, 82, 83
second language learning, 62, 66, 73–79, 274
cut-off, critical periods, 62, 71, 75, 77, 78, 79, 83
dative case, 12, 188, 252, 302
deafness, 73, 79–81, 218
declarative, 241
degenerate input, 217, 226
deictic, 240
demonstratives, 240, 303
derivation, 178tab, 190–191, 192, 192tab, 193–194, 197tab, 201, 291
derivational morphemes, 191, 303, 305
derivational morphology, 190–191
derivational suffixes, 193–194, 200, 291–292
derived forms, 192, 193, 194, 290–292
determiner phrases, 223, 229, 303
diary studies, 18
digit span, 17tab, 18
direct contrast hypothesis, 105
direct instruction, 214–215, 226
direct objects, 246
Disapprovals, 104, 111
displacement, 30, 32, 34–35, 36, 44
distributional analysis, 248
dogs, 38–40, 50tab, 51tab
domain-general learning, 39, 137, 237, 272–274, 275, 279
duality of patterning, 32, 44
dual-route theories, 180–184, 189, 202
duration, 138, 140
dyadic interaction, 237–238
dynamic register, 91
early compounds, 194–196
early constructions, 295, 296
echoic behaviour, 98
ecological validity, 136, 137–138, 303–304
edges, utterances, 139, 140
EEG (electroencephalography), 280
elephants, 37, 50tab
elicited production method, 275
E.M., 80–81
emotions, 17tab
English language
discriminating phonemic contrasts, 127tab
inflection, 177
nouns, 164
object names, 163
past tense, 188
phonemes, 299
phrases, 211
plurals, 179, 188, 198
research focus on, 280–281
specialization towards native language, 126
verb transitivity, 254
entrenchment, 256–259, 260–261, 268
Epicurus, 272
error-contingent clarification questions (CQs), 111
errors, 103–105, 106, 107, 111, 113, 257–258, 259, 260, 261, 297
establishment, 279
Estonian language, 194
event-related potentials (ERPs), 126, 128, 276–277
exclusion learning, 38, 39
faculty of language – broad (FLB), 49, 52fig
faculty of language – narrow (FLN), 49, 52fig, 228
fast mapping, 38, 93, 161, 251
feral children, 63–64, 65tab
finetuning, 91
Finnish language, 177, 188, 194, 199
first words, 7–8
first year
categorical perception, 122–126
grammar learning, 141–143
individual differences in speech perception, 130
specialization towards native language, 126–129
vocabulary at five years old, 17tab
vocabulary at one year old, 149
word segmentation see word segmentation
FLB (faculty of language – broad), 49, 52fig
FLN (faculty of language – narrow), 49, 52fig, 228
floating quantifiers, 225–226
fMRI (functional Magnetic Resonance Imaging), 277, 280
foetuses, 5–7, 120, 138–139
FOXP2 gene, 284
French language, 194, 198
frequency see token frequency; type frequency
frequency effects, 251–252, 261, 268
frequency of occurence, words, 252
fricative sounds, 128, 149, 181, 296, 304
function words, 68, 69
functional Magnetic Resonance Imaging (fMRI), 277, 280
functional reorganization, perceptual capacities, 129
functionalism, 236
gavagai problem, 9–10, 163, 164, 170, 171–172
gaze, of mother, 237, 238
generalization, 10, 37, 38, 245, 250, 254, 255
see also overgeneralization
genes, 19–23, 62, 99, 101, 102, 206–207, 208, 209, 230, 231, 266–267,
268–269, 272, 275, 284, 304, 305
see also heritability; nature-nurture debate
Genie, 64, 66, 67–71, 80, 81, 83, 84
genitive case, 12, 188, 252, 259, 302
German language, 188, 194, 249, 251, 254
gestures, 37, 94, 101, 239, 281, 296
gibbons, 33, 34–35
gorillas, 54
grammar
age acquired by, 59
animals, 41–45, 48–49
definition, 304
early, 141–143
E.M., 80–81
first year, 141–143
grammar-acquiring capacity, 274
grammatical categories, 14, 44, 70, 180, 210, 223, 228, 244, 287
grammatical errors, 30–31
grammaticality, 75, 134, 275
infants versus monkeys, 48–49
Noam Chomsky see author index
parental input, 92
rule learning, 136
vocabulary, 158fig
see also universal grammar (UG)
Gua, 36
habituation method, 123–124, 275
Head Parameters, 211, 212
heads, 191–192, 195–196, 211–212, 304
head-turn preference, 275
Hebrew language, 194, 200–201
height, 267
hemispheres, brain, 45, 71, 79, 81, 99, 277
heritability, 101, 267, 304
Herodotus, 63
high-amplitude sucking (HAS), 6, 123
Hindi language, 125, 126, 127tab
historical figures, 271–272
holophrases, 194, 241–242, 258
homesign, 218
homophones, 187, 304
honey bees, 34fig, 35, 36
Hungarian language, 178, 188, 305
hypothesis space, 171, 172
iambic pattern, 138
imitation
actors, time lag and fidelity, 99–101
cognitive development, 98–99
corrective input, 115
early constructions, 295, 296
individual differences in, 101–102
in language development, 114–115
linguistic creativity, 97
nature of, 87
poverty of stimulus argument, 222–223
research focus on, 97
Skinner and Chomsky, 98
imperative, 241, 304
inductive inferences, 132–134
Infant Directed Speech (IDS), 87, 89, 91, 134, 137, 142
see also Adult Directed Speech (ADS); Child Directed Speech
(CDS)
infinite monkey theorem, 28–29
infinitive, 297
inflection, 14, 176–180, 181, 182–183, 184, 185, 186fig, 188, 189, 190,
192–193, 194, 197, 199–200, 201, 291, 304–305, 311
inflectional morphemes, 11–12, 67, 177, 179, 304–305
inflectional morphology, 11, 176, 185, 252, 269, 297, 304, 305
innateness, of language, 20–22, 206, 207–208, 227–230, 305
see also nativism; nature-nurture debate
input, 87, 88, 95, 114–115
input databases, 279
input frequency, 114, 252, 257, 258, 259, 260, 261, 268, 269, 279, 281,
297
instrumental, 12, 305
intelligence/IQ, 267
intention-reading, 238–240, 241, 296
interaction, 87, 94, 95–96, 114–115, 268
intermodal preferential looking paradigm, 255, 275–276, 278
International Phonetic Alphabet (IPA), 46, 145
interrogative, 241
intonation, 6, 241, 279, 280, 305
intransitive verbs, 246–247, 293
IQ, 72, 209
irregular nouns, 177
irregular past tense verbs, 107–110, 180–181, 185, 202, 309–310
irregular verbs, 107, 177–178, 183, 184, 185, 187, 189
Isabelle, 72
Italian language, 179
item-based constructions, 242, 243, 245, 258
Japanese language, 126, 127, 163, 164, 179, 181, 209, 211, 212,
225–226
jargoning, 17tab
journals, 278
Kaluki language, 163
Kanzi, 37–38, 39, 40, 41, 42, 43, 44, 50tab, 51tab, 55
Kasper Hauser, 65tab
Koko, 54
Korean language, 163, 199
language
acquisition and development, 17–18, 21–22, 73–76, 114–115, 213,
215–216, 226–227, 278–281
design of, 30–36
levels of, 4–5
statistical properties, 131, 140, 141
talk and communication, 29
Language Environment Analysis (LENA), 280
learning, 21–22, 137–138
see also word learning
learning mechanisms
domain-general learning, 272–274
linguistic nativism, 274–275
LENA (Language Environment Analysis), 280
level ordering, 197–199
lexemes, 177, 190, 191, 257, 305
lexical entries, 180, 181
lexical plugs, 154
lexical processing, 154–156
lexicons, 36, 67, 180, 181, 306
lexigrams, 37–38
limited exposure to linguistic input, 213, 214
linguistic competence, 154, 213, 303
linguistic creativity, 97
linguistic deprivation, 62–73, 80–81
linguistic development see language, acquisition and development
linguistic diversity
core versus periphery, 210–211
nature of, 209–210
parameters of variation, 210, 211–212
triggers for setting parameters, 212–213
linguistic nativism see nativism
linguistics, 4, 20, 306
linking problem, 212–213
Little Englisher syndrome, 225
looking behaviour, 123–124, 168tab, 255, 275–276, 278
machine learning, 227
main clauses, 306
Majola children, 65tab
malapropisms, 162
mama, 7–8
Mandarin language, 128, 199–200, 280
mass nouns, 303
mathematical modelling, 279
mean length of utterance (MLU), 90, 92, 306
memory, 101, 178, 180, 181, 183, 184, 267, 274, 279, 284
Merge, 228
Metalab, 280, 284
metalinguistic skills, 7, 32, 148, 199–201, 215, 274, 306
methodologies, 275–278, 279–280, 283
milestones, language development, 8, 16–18, 67, 280
mirror neurons, 99, 102
modelling, 279
Model/Rival technique, 40
modifiers, 191, 195–196, 304
molecular genetics, 267
monkeys, 28, 39, 45, 47, 48–49, 50tab, 51tab, 53, 99, 136–137, 165, 273
see also infinite monkey theorem
morphemes, 10–12, 177, 185, 188, 201, 303, 304–305
morphological awareness, 199–201
morphological errors, 154
see also errors
morphological marking, 243, 254, 262
morphology
Child Directed Speech, 89–90
compounding, 178tab, 191–193, 194–199, 291
definition, 306–307
derivation, 178tab, 190–191, 192, 192tab, 193–194, 197tab, 201,
291
genetic basis of, 267
inflection, 176–180, 181, 182–183, 184, 185, 186fig, 188, 189, 190,
192–193, 197, 199–200, 201, 291
past tense debate see past tense debate
school years, 199–201
see also morphemes
Motherese, 87, 112
motion, verbs of, 240
multilingualism, 280
multi-unit speech, 242–244
mutual exclusion, 51tab
mutual exclusivity bias, 168tab, 169–170
native language, 126–129
native language neural commitment (NLNC), 128–129
nativism, 20, 21–22, 44, 87, 96, 104, 167, 198, 208–209, 213, 232, 256,
267, 268, 274–275, 281, 305, 306
nature-nurture debate
domain-general learning, 272–274, 275
genes, 268–269
genetic basis of language development, 266–267
historical figures, 271–272
 linguistic nativism, 274–275
nurture in syntax acquisition, 267–268
origin of term, 207
race and IQ, 209
see also nativism; non-nativism
Navajo language, 163
near-infrared spectroscopy, 277
negations, 90, 307
negative evidence, 105–107, 111, 113, 217–218, 226, 307
negative feedback, 110–111
neuroscience, 126, 189, 277, 279
newborn infants, 5–7, 120
Nim, 41, 42–43
no negative evidence assumption, 103–105, 111, 112, 217–218
nominative case, 254, 297, 302
non-nativism, 21, 22, 104, 111, 114, 187, 208–209, 222–223, 256,
267–268, 272, 274, 279, 281, 283
see also usage-based theory
non-verbal imitation, 101–102
nonword repetition tasks, 101
noun phrases, 14, 90, 211, 223, 228, 229–230, 244, 246, 248, 249, 250,
303, 307
nouns
biases in word learning, 163–165
common, 39, 302
count, 169, 171, 303
definition, 307
English language, 164
irregular, 177
mass, 303
proper, 39
Russian, 13
word segmentation, 139–140
novel utterances, 30–31
novel verbs, 181, 247, 260
novel words, 107–110, 275
Nthlakapmx language, 125–126, 127tab
Nullarbor Nymph, 64
nurture, 267–268
object names, 39, 41, 50tab, 51tab, 154, 160, 163, 164, 165, 296
object permanence, 18, 216
objects, 69–70, 155, 156fig, 157, 307
observation, 275
ocular dominance columns, 60, 61
offset, critical periods, 59, 72, 78, 79, 82, 83
operant conditioning, 98
orthography, 307
overextension
category errors, 152–154, 155
lexical plugs, 154
lexical processing, 154–156
nature of, 151–152, 308
pragmatic errors, 152, 154, 155
retrieval failures, 152, 154, 155
overgeneralization, 245, 256, 308
overregularization errors (ORs), 182–184, 187, 245, 308
paradigmatic process, 250
parameters of variation, 210, 211–212
parental input, grammar, 92
parrots, 39, 40–41, 50tab, 51tab, 54
parse, 308
parts of speech, 308
passives, 252
past tense debate
blocking hypothesis, 182–184
connectionism and single-route accounts, 185–187, 189, 203
crosslinguistic evidence, 188–189
English language, 188
Word and Rules theory, 180–184, 188, 189
patients, 237, 243, 308
pattern associators, 185, 187, 189
pattern finding, 248–249, 273
pauses, 142
peak plasticity, 62, 78–79, 83
perceptuo-motor development, 17tab
performance, 75, 154, 303
peripheral grammar, 210–211
periphrastic causative, 260
persons, 308
phonemes, 17tab, 32, 36, 37, 45, 46, 48, 53, 66–67, 88–89, 120–122,
143, 299
see also categorical perception; International Phonetic Alphabet
(IPA)
phonemic contrasts, 124–129, 275
phonological contrasts, 125, 289–290
phonological memory, 267, 284
phonologically conditioned, 178
phonology, 4, 5, 66–67, 71, 82, 83, 88, 267, 309
phrases, 211
pigeons, 39, 51tab
pitch, 138, 140, 142, 143
pivot schemas, 242–244, 258
plasticity, 78–79
Plato’s problem, 216–217, 274
play, 17tab
plurals, 177, 179, 181, 184, 188, 192, 197–199, 251, 303, 305, 308
pointing, 238–239, 240, 295
Polish language, 142, 188, 252, 259, 280
possessive, 309
poverty of stimulus argument
constitution of universal grammar, 228
degenerate input, 217, 226
empirical support for, 230
evidence from children, 223–226
imitation of grammatical structures, 222–223
machine learning, 227
nature of, 213, 309
negative evidence, 217–218, 226
Plato’s problem, 216–217
structure dependence, 218–227, 266
universal grammar see poverty of stimulus argument
pragmatic errors, 152, 154, 155
pragmatics, 4, 280, 309
predictive dependencies, 48–49
pre-emption, 256, 259–261
preferential looking paradigm, 255, 275–276, 278
prefixes, 191, 201, 309
pregnancy, 5–7
prepositions, 171, 309
primates, 36–40, 41–45, 48–49, 50–51tab
probability theory, 132–134, 171, 172
production, word, 50tab, 149–151, 152, 162
productivity
conservative learning, 256
derivational suffixes, 193–194
entrenchment, 256–259
frequency effects, 251–252
pattern finding, 248–249, 273
pre-emption, 259–261
syntactic bootstrapping, 252–255
timelines, 282fig
transitivity bias, 245–248
type frequency, 249–251
prompt hypothesis, 111
pronouns, 171, 240, 309
pronunciation, 296, 299
proper nouns, 39, 302
prosodic cues, 131, 138–140
prototype theory, 153
Psamtik I, 63
psycholinguistics, 278
quantifiers, 225–226
question formation, 218–222, 224–225, 227, 257–258
race, 209
rapid fading, 31, 44
rats, 49
recasts, 102–103, 104–106, 111, 113–114, 115, 116
recency effect, 155
recursion, 228, 229–230
reference, 39
regular past tense verbs, 177–178, 180, 181, 184, 185, 187, 188,
309–310
regular verbs, 107, 177, 178
relative clauses, 90, 229, 249, 310
relative cue strength, 140–141
repetition, 97
representations, 213, 244, 254, 256–257, 258
research, 18–19, 97, 170–171, 268–269, 272, 280–281, 283–284
retrieval failures, 152, 154, 155
rule learning, grammar, 136
Russian language, 12, 13, 89–90, 177, 179, 188, 209, 305
Sami language, 194
school years, 75, 77, 173, 182, 188, 195, 199–201, 214–215, 271, 306
second language learning, 62, 66, 73–79, 274
segmentation problem, 10, 131, 182, 275, 310
semantic bootstrapping, 90
semanticity, 30, 31, 33, 36, 38, 44
semantics, 254, 280
sensitive periods, 62
see also critical periods
sentences, 273, 296, 310
sentential complements, 90, 310
shape bias, 153, 166fig, 167–169, 170
shared intentionality, 165, 269, 295, 296
sign language, 36–37, 44, 79, 80, 81–82, 218
similarity, 249, 251
single unit speech, 242–244
single-route theories, 180, 185–187, 189, 203
slow mapping, 161–162
social cognition
collaborative engagement, 238–240
dyadic and triadic interaction, 237–238
intention-reading, 238–240, 241
usage-based theory, 237–240
social development, 17tab
socio-economic status (SES), 92–93, 94
songbirds, 51tab, 228
Spanish language, 179
specialization, 44
spectral tilt, 140
speech
notation for sounds, 46
perception, 130
processing, 45–48
spelling, 201
spurt, vocabulary, 8
statistical analysis, 132–137, 280, 284
statistical learning, 47–48
stop sounds, 296, 310
stresses, in speech see prosodic cues
structural similarity, 249
structure dependence, 218–227, 228, 266
subjects, 69–70, 90, 244, 310
subordinate clauses, 90, 310
substantial similarity, 249
suffixes, 191, 193–194, 201, 291–292, 310
SVO language, 69, 210
syllable patterns, 47–48, 51tab
syllables, 121, 131, 142, 310–311
symbolic expression, 297
syntactic bootstrapping, 245, 252–255, 263
syntactic marking, 243, 245
syntactic productivity see productivity
syntagmatic process, 250
syntax, 13–16, 18, 67–71, 82, 83, 90, 97, 267–268, 280, 311
synthetic compounds, 192
talking, 29, 50tab
tamarin monkeys, 28, 48–49, 51tab, 136–137, 273
taxonomic bias, 153, 168tab, 170
teachers, 200, 201
Teletubbies, 96
television, 95–96, 117
tenses, 177, 181, 210, 297
see also past tense debate
Thai language, 128, 129tab, 280
thematic relation, 196
theory of mind, 17tab, 18, 157, 279
Tiedemann, Dietrich see author index
timelines, language development, 282fig
see also milestones, language development
token frequency, 249, 256, 257, 258, 262
tones, 280
Tracton, South Carolina, 112–113, 114
transformations, 221
transitional probabilities, 131, 135, 137, 138, 140, 141
transitive verbs, 236–237, 243–244, 246–247, 252–253, 254, 255, 256,
278, 293, 308
transitivity, 243–244, 245–248, 254, 293, 311
transitivity bias, 245–248, 252
triadic interaction, 237–238
triggers, parameters of variation, 212–213
trochaic, 138
Turkish language, 178
twin studies, 266–267
Twins Early Development Study, 101
type frequency, 248, 249–251, 256, 262
Tzeltal language, 163
universal grammar (UG)
alternatives to, 279
challenges to, 207
Child Directed Speech, 215
conservative learning, 256
direct instruction, 214–215, 226
ease and speed of language acquisition, 213, 215–216, 226–227
empirical support for, 230–231, 266
innate elements, 227–230
limited exposure to linguistic input, 213, 214
linguistic diversity see linguistic diversity
linguistic nativism, 213, 267, 274
 see also nativism
nature of, 206–207
poverty of stimulus see poverty of stimulus argument
structure dependence, 227, 228
universals, 210
usage-based theory
collaborative engagement, 238–240
conservative learning, 256
early constructions see utterances
entrenchment, 256–259, 260–261, 268
establishment of, 279
frequency effects, 251–252, 261, 268
further reading on, 262–263
holophrases, 258
intention-reading, 238–240, 241, 296
nature of, 236–237
pattern finding, 248–249
pre-emption, 256, 259–261
productivity see productivity
social cognition, 237–240
syntactic bootstrapping, 252–255, 263
transitivity bias, 245–248, 252
type frequency, 249–251
word segmentation, 241
U-shaped development, 187
utterances
comprehension, 296
concrete and abstract, 244–245
edges, 139, 140
holophrases, 241–242
making generalizations, 245
from single unit to multi-unit speech, 242–244
verb argument structure, 260, 311
verb island hypothesis, 243–244, 245
verb phrases, 211, 248, 311
verbs
auxiliary, 220–221, 222, 224, 292, 301–302
 bias in word learning, 163, 164, 165–166
irregular, 107, 177–178, 183, 184, 185, 187, 189
irregular past tense, 107–110, 177, 180–181, 185, 202, 309–310
of motion, 240
nature of, 13, 311
novel, 181, 247, 260
regular, 107, 177, 178
regular past tense, 177–178, 180, 181, 184, 185, 187, 188, 309–310
Russian language, 209
transitive, 236–237, 243–244, 246–247, 252–253, 254, 255, 256,
278, 293, 308
word order, 69–70
word segmentation, 139–140
Victor, Wild Boy of Aveyron, 64
Viki, 36, 44
see also chimpanzees
vocabulary
Child Directed Speech, 89
deaf people, 81
at five years old, 17tab
grammar, 158fig
growth see word learning, rate of
at one year olds, 149, 151
size, 8–9, 18, 25, 96, 160
socio-economic status, 92
spurt, 8
vocal-auditory channel, 31, 36, 44
voice
recognition, 50tab
tone, 91
voice onset time (VOT), 122–123, 128, 129tab
voiced and voiceless sounds, 122, 128, 178, 311
voicing, 122–123, 149, 178
VOS language, 210
vowels, 48, 121, 122, 139, 299, 303
waggle dance, 34fig, 35
Wall Street Journal, 222
Washoe, 37, 41, 42, 44, 50tab, 54
see also chimpanzees
western gulls, 33–34
whole object bias, 164–165, 167, 168tab, 169, 170, 171
whole word learning, 178–180
Wild Peter of Hanover, 65tab
Word and Rules (WR) theory, 180–184, 188, 189
word learning
approaches to, 148–149
areas for research, 170–171
biases see biases, word learning
computational modelling, 171–172
in dogs, 38–40
first words, 149–151
further reading on, 172–173
gavagai problem, 163
intention-reading, 296
overextension see overextension
rate of, 160–162
vocabulary spurt, 156–160
word segmentation, 136
word segmentation
early grammar, 141–143
first year see word segmentation
learning in the real world, 137–138
nouns, 139–140
prosodic cues to, 131, 138–140
relative cue strength, 140–141
statistical properties of language, 131, 132–137, 140, 141
usage-based theory, 241
verbs, 139–140
word learning, 136
words
activation, 155, 156fig
combinations, 51tab, 242, 258
comprehension, 50tab, 149, 150fig, 152, 162
first, 7–8, 149–151
 forms, 13–14, 177
frequency of occurence, 252
learning see word learning
meaning ascription, 9–10
nature of, 8
novel, 107–110, 275
order, 15–16, 69–70, 142–143, 243, 254
production, 50tab, 149–151, 152, 162
segmentation see word segmentation
stresses see prosodic cues
working memory, 101, 274
WR (Word and Rules) theory, 180–184, 188, 189
Yucatec Maya language, 209–210
zebra finches, 49, 51tab
Zeno, 272
Zulu language, 127