Cognitive Collocations PDF

Download as pdf or txt
Download as pdf or txt
You are on page 1of 27

Going Beyond Patterns: Involving Cognitive

Analysis in the Learning of Collocations


DILIN LIU
University of Alabama
Tuscaloosa, Alabama, United States

Since the late 1980s, collocations have received increasing attention in


applied linguistics, especially language teaching, as is evidenced by the
many publications on the topic. These works fall roughly into two lines of
research (a) those focusing on the identification and use of collocations
(Benson, 1989; Hunston, 2002; Hunston & Francis, 2000; Smadja &
McKeown, 1991; Wouden, 1997) and (b) those focusing on the learning
and teaching of collocations, including the development of reference
books and textbooks (Bahns & Eldaw, 1993; Benson, Benson, & Ilson,
1997; Crowther, Dignen, & Lea, 2002; Hill & Lewis, 2002; Keshavarz &
Salimi, 2007; Lewis, 2000, 2002; McCarthy & ODell, 2005; Nesselhauf,
2003; ODell & McCarthy, 2008; Sun & Wang, 2003; Webb & Kagimoto,
2009). Although these publications have greatly enhanced our under-
standing of collocations and their role learning and teaching, a close look
at them indicates a lack of critical examination of the definition and the
nature of collocations and the way collocations are taught. This article
aims to address the issue through (a) a close critical examination of the
collocations in existing teaching or reference materials and the typical
way they are taught and (b) a corpus-based analysis of some representative
collocations. On the basis of the examination and analysis, the article
argues for a more effective pedagogical approach to collocations that
involves corpus-based cognitive analysis of collocations.
doi: 10.5054/tq.2010.214046

REVIEW OF THE LITERATURE


What Constitutes a Collocation? An Examination of Definitions
To begin with, it is necessary to note that the discussion of the
definition of collocation here is not meant as a thorough review but only a
brief overview of the issue from a language-teaching perspective.1 It is
1
For example, this overview does not cover recent psycholinguistic and natural language
processing research. Although such research has provided evidence for the existence of
certain combinations of words as lexical units, it does not have direct bearing on the issue I
address in the article: Whether collocations are generally arbitrary.

4 TESOL QUARTERLY Vol. 44, No. 1, March 2010


well known that what goes under the header of collocation is very
heterogeneous (Wouden, 1997, p. 53). The meaning of the term often
varies, depending on ones research purpose and theoretical orienta-
tion. Yet a closer look at the term shows that its various uses boil down
basically to two major meanings. The first is the uncountable use of the
term referring to a linguistic propertythe tendency for certain words
to occur together (Finch, 2000, p. 152), that is, to co-occur more often
than by chance. The second is the countable use of the term referring to
specific habitual combinations of words that are often treated as lexical
items or units. The first meaning is a much broader concept, because it
covers not only the habitual combinations of words but also the issue of
collocation in cases where the lexical items that often co-occur do not
form lexical units. For example, the word toy co-occurs with children
more frequently than with women and men [because] toys belong to
children, on the whole, rather than adults (Hunston, 2002, p. 68).
However, although toy and children often co-occur, they do not form a
habitual combination as a lexical unit. Their co-occurrence can only be
considered a collocation issue in the broad sense defined by Sinclair
(1991): Collocation is the occurrence of two or more words within a
short space of each other in a text (p. 170).
This broader use of the term seems to be of interest only to those
scholars working on lexicogrammatical patternings (e.g., Hunston, 2002;
Hunston & Francis, 2000; Sinclair, 1991). For those working in language
teaching and lexicography (e.g., Bahns & Eldaw, 1993; Benson, 1989;
Keshavarz & Salimi, 2007; Lewis, 2002; Nesselhauf, 2003; Willis, 2003),
collocation means primarily habitual combinations of words, such as do
(not make) laundry and make (not do) a case, although these scholars do
not always agree on what types of habitual combinations may be
considered collocations. Language educators primary interest in
habitual combination is best evidenced by their emphasis on the
collocatability or restrictions on how words can be combined. For
example, according to Richards, Platt, and Webber (1985, p. 46), the
study of collocations focuses on the restrictions of how words can be
used together, for example which prepositions are used with particular
verbs and which verbs and nouns are used together.
Obviously, it makes good sense for language educators to focus on
habitual collocations and exclude co-occurrences of words that are not
habitual combinations, because the latter type does not have much
pedagogical value. However, as already mentioned, language educators
currently do not have a consensus on what types of word combinations
qualify as collocations. They often cannot agree on how fixed in
structure and how transparent in meaning a word combination must be
in order to be considered a collocation (Fernando, 1996; Nesselhauf,
2003). For some scholars, a collocation is fairly fixed in structure, but for

GOING BEYOND PATTERNS 5


others it can have considerable variation. In terms of semantic
transparency, most scholars (e.g., Fernando, 1996; Moon, 1998;
Nesselhauf, 2003) believe that collocations are fairly transparent and
thus should not include idioms, although a few (e.g., Palmer, 1933;
Wouden, 1997) consider idioms to be collocations. Despite the
divergent views, most language educators appear to agree that
collocations are those word combinations that have restricted structural
variation and differ from entirely free word combinations on the one
hand and idioms on the other (Bahns & Eldaw, 1993; Crowther, Dignen,
& Lea, 2002; McCarthy & ODell, 2005 ; Nesselhauf, 2003; ODell &
McCarthy, 2008). Collocations so defined are, using Moons (1998)
terminology, restricted collocations (p. 27) or, in Fernandos (1996)
words, habitual collocations with restricted variance (p. 32). Yet
even with this narrow definition of collocation, it is often still difficult to
decide whether specific word combinations are collocations. This is
because the key element in the definition is restricted variation and,
when used to identify collocations, it is not a binary yes-or-no criterion
but rather one of scale (see Nesselhauf, 2003, p. 225). To help alleviate
the problem, some scholars have divided collocations into scaled
subcategories, for example, strong, medium strength, and weak
(Crowther, Dignen, & Lea, 2002, p. vii) or fixed, strong, and weak
(ODell & McCarthy, 2008, p. 8).

Are Collocations Arbitrary? Clarification of Key Terms


A major issue related to the definition of collocations that is of
particular importance for this article is whether collocations are
arbitrary. A quick look at the literature on the issue seems to suggest
that many linguists and language educators consider collocations
arbitrary. For example, Benson (1989) contends that collocations
should be defined not just as recurrent word combinations, but as
arbitrary recurrent word combinations (p. 3). Similarly, Smadja and
McKeown (1991) believe that [a] collocation is an arbitrary recurrent
word combination (p. 230). Lewis (2002) also sees collocations the
same way: Collocation, as I have shown, is arbitrary: high/tall building;
tall boy but not *high boy. Prices rise and fall; you can rise to the occasion but
not *fall to the occasion (p. 26). Likewise, Nesselhauf (2003) uses
arbitrary restriction on substitutability (p. 225) as the key feature in
defining collocations. In a similar vein, other scholars have used words
like unmotivated and unpredictable/unpredictability in defining
collocations (e.g., Chan & Liou, 2005, p. 231; Hunston, 2002, p. 86;
Nation, 2001, p. 317). Given the prominence of the issue in this article, it
is important that researchers accurately understand what the scholars

6 TESOL QUARTERLY
mean by the use of the word arbitrary (and its related terms unmotivated
and unpredictable) and to what extent they believe collocations are
arbitrary.
A scrutiny of the scholars discussions, including their explanations of
why they consider collocations arbitrary, suggests that there are at least
three different perspectives or meanings in the use of arbitrary. The first
one views collocations as arbitrary from a cross-linguistic perspective,
as shown in Bensons (1989) support for his argument about
collocations arbitrariness:

The arbitrary (as opposed to free) nature of collocations can be demon-


strated when they are juxtaposed with corresponding collocations in other
languages. . . For example, when referring to a police officer whose duty it is
to keep traffic moving, one says in French r gler la circulation, in German
den Verkehr regeln (p. 3)

The interlanguage comparisons given in the quote clearly illustrate


Bensons cross-linguistic perspective on the issue. The second perspec-
tive is an intralinguistic one, which sees collocations arbitrariness in
terms of their syntactic and semantic abnormality, as expressed clearly by
Smadja and McKeown (1991):

A collocation is arbitrary because it cannot be predicted by syntactic or


semantic rules. For example, strong and powerful are both adjectives and
are synonymous in meaning, but strong is used to modify tea and
powerful to modify car and not vice versa (p. 230)

A strong version of this intralinguistic view is that the meaning of a


collocation cannot be derived from the meanings of the words in it.
Instead, its meaning is a unit of meaning arising out of the collocation in
and of itself. In this view, strong tea or powerful car is a unique linguistic
unit whose meaning is noncomposite. Of course, not all of those who
hold the intralinguistic perspective on collocations arbitrariness share
this view, at least not completely. Some do not consider noncomposite
meaning to be a feature of collocations (Bahns & Eldaw, 1993; McCarthy
& ODell, 2005; Nesselhauf, 2003). To them, noncomposite meaning is a
distinctive feature of idioms only, that is, it distinguishes idioms from
collocations. There are still others who either think there is only a
certain degree of noncomposite meaning in collocations or believe only
some collocations are noncomposite in meaning (Crowther, Dignen, &
Lea, 2002; Nation, 2001).
The third perspective approaches the arbitrariness of collocations
both as a language-processing and a language-learning issue, as can be
seen in Lewis (2002) comment after he supported his argument about

GOING BEYOND PATTERNS 7


the arbitrariness of collocations with the examples of tall but not *high boy
and rise but not *fall to the occasion: This non-generalisability clearly
indicates that we meet and store words in the prefabricated chunks upon
which the Lexical Approach is based (p. 26). Of course, Lewis
perspective can be considered a combination of the first and the second,
because it encompasses both the cross-linguistic view (i.e., learning a
second language [L2] using his Lexical Approach) and the intralinguistic
view (i.e., collocations are prefabricated and unsuitable for normal
language processing). Like Lewis, most L2 educators appear to use the
term arbitrary in the combined sense (e.g., Keshavarz & Salimi, 2007;
Nesselhauf, 2003), because, in their discussion about collocation learning,
they all point out both the problem of semantic arbitrariness in the choice
of words in collocations and the special difficulty such arbitrariness
presents to L2 learners. Nations (2001) definition of collocations best
represents this two-sense view: From a learning point of view, it makes sense
to regard collocations as items which frequently occur together and have
some degree of semantic unpredictability (emphasis added, p. 317).
Next, the terms unmotivated and unpredictable are discussed. It seems
that these terms are synonymous with arbitrary, because they are often
used to define or portray the arbitrariness of collocations. An excellent
example is found in Smadja and McKeowns (1991) explanation quoted
earlier: A collocation is arbitrary because it cannot be predicted by
syntactic or semantic rules (underscoring added, p. 230). Compared with
unmotivated, unpredictable is used much more widely, perhaps because it
can be used in both the cross-linguistic and intralinguistic senses; that is,
collocations can be unpredictable either because of their difference from
other languages or because of their semantic and/or syntactic uncon-
ventionality. In contrast, unmotivated is generally used in the intralinguistic
sense because, when researchers say a collocation is semantically
unmotivated, they mean there is no clear reason for the selection of the
words in the combination based on the meanings. Of course, when L2
speakers consider a collocation unmotivated, it can become a cross-
linguistic issue, if they base their decision on the meanings of the words in
their first language (L1) rather than in the L2, but making a decision in
this manner is not really valid, a point I discuss in detail later.
Concerning the extent to which these scholars believe collocations are
arbitrary, my examination suggests a substantial variation, but their views
can be divided roughly into two major groups. The first holds the view that
collocations are all arbitrary, although their degree of arbitrariness may
vary (e.g., Benson, 1989; Nesselhauf, 2003; Smadja & McKeown, 1991).
The second group takes the position that not all collocations are arbitrary
(e.g., Crowther, Dignen, & Lea, 2002; McCarthy & ODell, 2005).
It is also important to note that many educators belief that
collocations are arbitrary can be seen in their discussion about how

8 TESOL QUARTERLY
collocations should be taught. For example, as early as the first part of the
last century, Palmer (1933) argued that each collocation must or should
be learnt, or is best or most conveniently learnt as an integral whole or
independent entity, rather than by the process of piecing together their
component parts (p. 4). The same view is still held by many today. In a
discussion about collocation teaching, Lewis (2000) writes,

Collocations are not words which we, in some sense, put together, they
occur naturally, and the first task of the language teacher is to ensure that
they are not unnecessarily taken apart in the classroom . . . the words [of a
collocation] should be recorded together. (p. 132)

In the same vein, Hill (2000) suggests that collocations should be learned
in such a way that we can retrieve them from our mental lexicon just as we
pull a phone number or address from our memory (p. 53).
However, recent cognitive and corpus-based research appears to
suggest that even those collocations that have been used as classic
examples to illustrate the arbitrariness of collocations might not be
arbitrary from an intralinguistic perspective. A case in point is the
high/tall building but no high boy example that some scholars have
used to show the arbitrariness of collocations (e.g., Lewis, 2002). Dirven
and Taylor (1988) did an excellent cognitive analysis of the usage
patterns of high and tall. They demonstrate that, although both
adjectives can express vertical extent measuring from the base of an
entity (often the ground or floor), for example, high and tall building or
mountain, only high can express positional high, a concept that does
not entail measuring from the ground but only in reference to ground
or floor level as in the cases of high ceiling and high clouds (p. 380). They
also show that tall is often used to describe living things such as people
and plants, whereas high is not. Tsuis (2004) interesting corpus study
reveals that tall is used almost exclusively to modify concrete things and
its overall frequency is low; in contrast, high boasts a much higher
frequency and wider semantic range, because it can be used to describe
many abstract concepts such as intensity, amount, and quality. Also,
although high is sometimes also used to modify concrete nouns, the
percentage is very low with a ratio of approximately 30% concrete versus
70% abstract. One weakness of Tsuis (2004) study is that it did not
explore the causes for the usage and semantic differences, a problem
found in most corpus-based studies on collocations. Specifically, she did
not explore why high is used mostly as a modifier of abstract nouns while
tall is mostly used to describe concrete nouns. A cognitive analysis similar
to Dirven and Taylors could have provided the answer.
Based on Dirven and Taylors (1988) study and also according to the
Oxford English Dictionary (OED; 2008), the key meaning of tall deals with

GOING BEYOND PATTERNS 9


the height of people or things, especially living things, measured from
the ground up; in contrast, high, while also having this meaning, may
refer to a positional high space not measured from ground up. This
nonground-based meaning of high may account for the extensive
abstract and figurative use of the adjective because things that are said to
be high in the abstract sense (e.g., high spirit and high motivation) do not
have any true base (ground) from which to measure their height. The
semantic and usage differences also help explain why one can say high/
tall building but only tall boy. In fact, because of its aforementioned
core meaning and usage, high is seldom used as a modifier of building,
according to my corpus examination.

PURPOSE OF THIS STUDY AND THE DEFINITION OF


ARBITRARY ADOPTED
The new research findings about the use of high/tall raise serious
questions about the arbitrariness of most collocations and the way
collocations should be taught. It is the purpose of this study to examine
what typical types of collocations are included in current collocation
dictionaries and textbooks, whether they are mostly arbitrary based on
close corpus-based cognitive analyses, and how collocations are actually
taught. Based on the research findings of the study, appropriate
pedagogical suggestions are proposed. Because there are different
interpretations or perspectives on the meaning of arbitrary as shown
earlier, spelling out the definition adopted for this study is first in order.
The meaning of arbitrary used in this study is limited to the
aforementioned intralinguistic one; that is, it means semantically unmoti-
vated. It excludes the cross-linguistic meaning, because although it is
certainly valid to use contrastive analysis to draw L2 learners and teachers
attention to the unique features of collocations, it does not seem valid to
use cross-linguistic differences as evidence or reason for viewing
collocations as arbitrary lexical units. Such a conclusion overlooks two
important facts. The first is that often a lexical item does not have a
single, fixed meaning but rather an array of senses related in principled
ways to its prototypical value (Langacker, 2008, p. 72). So a lexical item in
a collocation that may appear strange based on its core or prototypical
meaning may be entirely sensible when viewed in one of its related senses,
a point I return to later with examples. The second fact is that a vocabulary
item may be semantically gridded differently in different languages; that
is, its semantic mapping may vary from language to language. Thus the
collocations in the target language that appear arbitrary from the
perspective of a learners L1 can be completely sensible, based on the
semantic grids or networks of its lexical items in the target language.

10 TESOL QUARTERLY
It is a well known fact that, whereas a word may often have the same
core or prototypical meaning in two different languages, its extended or
peripheral meanings can differ substantially. For example, although the
word run has the same core meaning of fast motion in both English and
Chinese, it boasts a much more extended semantic mapping in English
than in Chinese. The word has, among others, the following extended
meanings in English that are absent in Chinese: (1) to function, for
example, The program or computer runs perfectly; ( 2) to manage,
for example, run a company; (3) to campaign, for example, run for
president. The opposite is true, however, with the semantic mappings
of the word open in Chinese and English. Its semantic mapping is much
broader in Chinese than in English. In Chinese, open can express, among
others, the following meanings that are not found in English: (1) to turn
something from an off status to an on (operating) status, for example,
open a light or computer; (2) to operate equipment, for example,
open a machine (not in the sense of taking it apart); (3) to conduct or
run, or example, open a meeting or lecture (not in the sense of
beginning a meeting or lecture, a use that is also found in English).
Thus, if the collocation patterns of a word based on its semantic
mapping in a given language are examined, it is often found that such
patterns are sensible, that is, not arbitrary or at least not entirely. One
more reason, a very important one, for employing this intralinguistic
perspective to examine whether a collocation is arbitrary is that it may
help language learners and teachers understand its semantic motiva-
tions, which in turn can help them better grasp it. It is very important to
note, however, that, in arguing for using intralinguistic semantic analysis
to understand and determine collocations motivations, I do not mean
that it is not valid to use contrastive analysis to help learners notice cross-
linguistic differences in collocation patterns. Such analysis is valid, but it
is not enough. Students need to know and will benefit from knowing why
the words in L2 collocations collocate the way they do.

DATA SELECTION
Selection of Collocation Dictionaries and Textbooks
The materials examined in this study are all the collocation dictionaries
and textbooks currently available: Benson, Benson, and Ilsons (1997) BBI
Dictionary of English Word Combinations, Crowther, Dignen, and Leas
(2002) Oxford Collocations Dictionary for Students of English, and Hill and
Lewis (2002) LTP Dictionary of Selected Collocations, and McCarthy and
ODells (2005; ODell and McCarthy, 2008) collocation textbook series.
The reason for including only two textbooks (the two in McCarthy and
ODells series) is that they are the only existing collocation textbooks

GOING BEYOND PATTERNS 11


very popular ones, thoughproduced by one of the most prestigious
publishers in language teaching and written by renowned authors. These
four publications (with the two textbooks in McCarthy & ODells series
considered as one) are examined to see not only what types of collocations
they include but also how they present collocations for teaching and
learning, information that helps researchers understand how collocations
are actually taught. Research articles and Web sites offering collocation
learning activities are also examined for how they present collocations for
teaching and learning.

Selection of Collocation Items for Examination


Because of space limitations, it is not possible to scrutinize all the
collocation items in all four publications but a review shows that most of
them are not arbitrary. Thus, selecting some examples is necessary. The
collocations in the noun entries of ability, work, and trip are chosen for
analysis for the following reasons. First, although the collocation items in
these publications are primarily noun and verb based, noun entries
appear to be much more predominant, outnumbering the verb entries,
especially in the LTP and the Oxford Collocation Dictionaries. In fact, the
Oxford Collocations Dictionary does not have entries for any of the common
verbs, such as do, have, make, and take. Second, ability and work are
selected because ability is the first and work is the last noun entry found in
all three dictionaries (they are also in McCarthy & ODells textbook
series but not the first or last noun entries in it). In other words, they
were not selected arbitrarily but for their representativeness (i.e.,
included in all four publications). The reason for selecting the collocates
of trip is as follows: Like the other two nouns, trip is an entry found in all
four publications; more importantly, as is shown in the analysis, unlike
ability and work, it has a few seemingly arbitrary collocationsmake/take/
have a tripand, furthermore, the verbs in these seemingly arbitrary
collocations are some of the most frequently used collocation-making
verbs. In this sense, they have special representative value for the issue of
arbitrariness being examined.
These seemingly arbitrary collocations (i.e., make/take/have a trip
collocations) and powerful car/strong tea are then used for corpus-based
cognitive analysis to determine their arbitrariness. The reason for
including make/take/have a trip for such a close analysis is fourfold. First,
verbnoun collocations are one of the most common type of
collocations and the focus of most studies on the teaching of
collocations (Bahns & Eldaw, 1993; Chan & Liou, 2005; Nesselhauf,
2003). Second, make/take/have a trip is one of the few verbnoun
collocations where multiple verbs appear to be used interchangeably; it

12 TESOL QUARTERLY
is very interesting to know whether the verbs are indeed interchangeable
in meaning, that is, whether the choice of them is unmotivated. Third, as
already mentioned, make, take and have are three of the most frequently
used verbs in English, especially as delexicalized verbs that form
numerous verbnoun collocations. Fourth, it is assumed that verb
noun collocations formed with delexicalized verbs tend to be
semantically unmotivated (Chan & Liou, 2005, pp. 235236); it is very
important to test this assumption in the case of make/take a trip. The
reason for including powerful car/strong tea is twofold: (1) they are
adjectivenoun collocations, and (2), more importantly, they have been
repeatedly used, even fairly recently, as examples to show the arbitrary
nature of collocations (e.g., Hunston, 2002, p. 68; Smadja & McKeown,
1991, p. 230). Testing their arbitrariness is thus especially meaningful. In
the analysis of these classic examples, I also briefly examine one other
pair of arbitrary collocations: strong wind versus heavy rain.

Corpora Used
The main corpus used for the study is the 360- million-word Corpus of
Contemporary American English (COCA) developed by Mark Davies
(2008) of Brigham Young University (the corpus has grown to 400
million words since the completion of my data analysis with the 2008 and
2009 data added). It is chosen for its free online access; its large size,
with comprehensive and representative data covering spoken English,
fiction, magazine and newspaper articles, and academic writing (no L2
data, though); the contemporariness in its data (19902008); and its
powerful, multifunction, and user-friendly search engine. The World
Wide Web and the 100-million-word British National Corpus (BNC; a
mega corpus with a data structure very similar to that of the COCA) have
also been used in a few instances to make sure the findings from the
COCA data are not merely idiosyncratic usages of American English
(when or how they are used are explained in the data analysis section).

DATA ANALYSIS AND DISCUSSION


Verb Collocations With Ability/Work/Trip Listed in the Four
Publications
First, I compiled all the verb collocates of each of the three nouns
listed in the four publications, and the results are reported in Table 1.
For each noun, a few of the verbs appeared in all four publications but
most in only one, with some appearing in two or three. An examination
of the collocates suggests that, in the case of the collocates of ability and

GOING BEYOND PATTERNS 13


TABLE 1.
Distribution of the Verb Collocates of Ability/Work/Trip in the Four Publications

Number of the publications


Verb collocates Noun they appeared in
Demonstrate, have Ability 4
Recognize 3
Appreciate, develop, show 2
Acquire, assess, display, encourage, exhibit, foster, lack, 1
lose, measure, nurse, overrate, stifle, underrate, use
Do Work 4
Carry out, complete, go to, have, lose, stop 3
Begin, find, finish, get, get to, seek, take on, undo 2
Arrive at, be engaged in, be off, commence, continue, 1
create, embark on, get down to, go about, halt, leave,
look for, make, need, neglect, oversee, produce, quit,
set about/to, start, supervise, take off from, undo
Take A trip 4
Arrange, cancel, go on, make, organize, plan, 3
Have 2
Be back from, be on, book, come back from, cut short, 1
embark on, extent, postpone

work, none appears actually arbitrary. The collocates of ability all fall
neatly into a few very sensible semantic groups, for example, with
demonstrate, display, exhibit, and show all expressing the meaning of
showing ability, whereas assess, measure, overrate, and underrate convey the
meaning of measuring ability. Similarly, the collocates of work also fall
into sensible semantic groups. The only items that may appear arbitrary
are those phrasal verbs like carry out, but this is a different issue because,
if a person understands that one of the meanings of carry out is doing or
performing, then its collocation with work is perfectly sensible. In the
case of the collocates of trip, most of them do not sound arbitrary either
(e.g., arrange, cancel, extend, plan, organize, and postpone a trip), but there
are a few (have, make, and take a trip) that may be considered somewhat
arbitrary, especially from a cross-linguistic perspective. To determine
whether they are truly semantically unmotivated, however, calls for a
close corpus-based cognitive analysis, which is done later. The most
important evidence that most of the verb collocates of the three nouns
are not arbitrary is not the finding that they all fall into sensible semantic
groups, however, but the fact that the collocates of a noun in a given
semantic group are largely interchangeable with no or little change in
meaning, as shown in the case of the showing ability group: Any one of
the verbs demonstrate, display, exhibit, or show conveys basically the same
meaning when collocated with ability. This fact suggests that they are not
arbitrary collocations, because one of the main arguments about the
arbitrariness of collocations (e.g., Lewis, 2002; Nesselhauf, 2003) is that a

14 TESOL QUARTERLY
(true) collocation does not allow its collocate(s) to vary freely, at least
not so extensively.

Take a Trip Versus Make a Trip Versus Have a Trip Versus Do a Trip
To search for all the tokens of the verb + trip structure in the COCA, I
entered for query [v*] a trip and [v*] a * trip. The search yielded
598 tokens of take a or an (adjective) trip (including all the tense or
aspect forms of the verb; of the 598 tokens, 199 contained an adjective,
e.g., take a fishing trip); 331 tokens of make a or an (adjective) trip (with
189 containing an adjective); 151 tokens of have a or an (adjective) trip
with 141 containing an adjective; and 13 tokens of do a or an (adjective)
trip with 10 containing an adjective. The results indicate that take a trip
and make a trip constitute the typical collocations for expressing the idea
of to go on a trip. However, have a trip also has a substantial number of
tokens, and it is included in two of the four publications, but a close look
at its token suggests that it has a uniquely different meaning. It has a very
low frequency when used without an adjective but boasts a fairly high
frequency when used with an adjective. A close reading of its tokens
indicates that 77 of the total 141 (a little over one-half) were used as an
expression of good wishes for a safe or good trip (e.g., have a wonderful
trip). In other words, have here means to experience or enjoy, a meaning that
is found in its semantic mapping based on the OED. In fact, the majority
(86%) of the have a trip tokens are used to convey this meaning. Thus, in
general, the meaning of have a trip differs from that of take/make a trip.
Before I discuss do a trip, the collocation with the lowest frequency in
the set, let me examine the use of take a trip and make a trip. Are the two
collocations really synonymous? A close reading of the tokens in
concordance lines suggests that the answer is no. Whereas take a trip is
used often for a trip of leisure, make a trip is typically used for a business
trip or a trip that appears particularly purposeful and effortful from the
speakers point of view. The following examples from the COCA help
illustrate the point.

Examples of take a trip


1. . . . she took a trip to England in 1877 for a much needed rest.
2. On one day, we took a trip to Rainbow Bridge National Monument, the
largest natural bridge in the world.
3. Two judges from Louisiana recently took a trip to Hong Kong and Bangkok
and charged it off to the taxpayers of Louisiana.

Examples of make a trip


4. Youll probably have to make a trip to the assessors office to get this much
detail.

GOING BEYOND PATTERNS 15


5. President Bush made a trip to Capitol Hill to meet with Senate Republicans
and try to change some minds.
6. This past fall when three of however [sic] U.S. Congressman made a trip to
Iraq they caught a lot of flak for it.

The difference between the two collocations can also be seen in their
tokens with an adjective. Of the 199 tokens of take a or an (adjective) trip,
the most frequent token was take a field trip (to a museum, show, etc.),
with 25 tokens (one eighth of the total number of tokens). Furthermore,
the adjectives used in many of the remaining tokens were leisure-related,
such as boat, train, holiday, sightseeing, and fishing. In contrast, none of the
189 tokens of make a or an (adjective) trip contains a leisure-related
adjective, and the token with the highest frequency in this group is make
a special trip, with 25 occurrences (more than one eighth of the total). By
the meaning of the word special, a special trip must be one made with
some unusual purpose or effort. A quick check of the BNC shows similar
patterns in British English: The most common adjectives for take a * trip
are day and boat, whereas the most common one for make a * trip is special
and for have a * trip is nice.
Based on this semantic difference between take a trip and make a trip,
one can say that the verb choice in each collocation is not arbitrary but is
motivated by its core meaning. Certainly, neither verb here is used in its
core sense; instead, they are both used in the sense of to accomplish/
perform (OED), one of the extended meanings found in the semantic
mapping of each verb. Yet, shaped by its core meaning, each verb in its
collocation conveys a distinct meaning of its own. It is known that the
core meaning of make is to create, a process that is more purposeful and/
or effortful than that of grip, the core meaning of take. This difference in
meaning between the two verbs in fact also appears in many other
collocations made up of the two verbs. It is salient in pairs such as make
versus take a phone call, make versus take an offer or deal, make versus take an
issue. In each pair, the make collocation suggests an action involving
more planning, effort, and/or initiation. One can also find the same
difference when comparing most of the other make + noun collocations
with most of the other take + noun collocations: make a or an decision/
effort/argument/plan versus take a break/vacation/walk/your time.
Now I discuss do a trip, a collocation in the set that is not listed in any
of the four collocation publications, perhaps because of its very low
frequency. It may have been considered as an idiosyncratic expression
used only by some individuals, but a scrutiny of its tokens seems to
suggest that do a trip often conveys a unique meaningthe meaning of
complete a trip, especially in the sense of completing a trip as an
achievement, as can be seen in the following COCA examples and as
evidenced by the fact that 8 of its total 13 tokens are in the past tense:

16 TESOL QUARTERLY
7. We just did a trip to Central America
8. Each summer, we do a mission trip.
9. My wife and I did a 9,200-mile trip around the country in 1991.

Apparently, replacing do in any one of the three sentences with make or take
would result in a loss of its semantic focus on the trip as an
accomplishment. This use of do in this collocation is not arbitrary,
however, because, according to the OED (online), to accomplish,
complete, finish, bring to a conclusion is one of the key meanings of the
verb, a meaning perhaps best shown in the popular utterance: We did it!
It is thus fairly safe to conclude from the earlier analysis that the make/take/
have/do a trip collocations are not arbitrary but semantically motivated.
Of course, as mentioned in the literature review section on the
arbitrariness of collocations, some linguists, especially those of the Firth
HallidaySinclair school, argue that the meaning of a collocation does not
come from its individual components but arises out of the collocation in
and of itself. In their view, a word in isolation does not really have a
meaning and, hence, there is no such thing as the core meaning of a word.
Yet, according to cognitive linguists and the results of the corpus-based
cognitive analysis of the collocations done in this article, the choice of
words in collocations is not random in general but is motivated by the
lexical items semantics, including their core meanings. Otherwise, why
does the combination of make and a trip result in a collocation referring to
a trip that is more purposeful and effortful than take a trip, not the other
way around? Similarly, as is shown later, the choice of the lexical items in
powerful tea/strong car, and so on, is also motivated by their semantics.

Strong Tea Versus Powerful Car


Regarding this well-known pair of collocations, I first tested whether
they are truly established habitual collocations by querying the COCA. I
searched not only for strong tea and powerful car but also for what were
considered impossible collocations: powerful tea and strong car. The
search produced 45 tokens of strong tea (17 of which were strong +
adjective + tea) with no tokens of powerful tea and 21 tokens of powerful
car or cars, with a surprising finding of four tokens of strong car or cars. To
make sure that powerful was indeed not used at all with tea and that the
four tokens of strong car were not some extreme cases of idiosyncratic
expressions, I conducted an Internet search of the BNC and the World
Wide Web. The results from the BNC are similar to those of the COCA,
in that although there is no token of powerful tea, there is one token of
strong cars. Interestingly, there are three false tokens of powerful tea and
strong car; they all appeared as examples of incorrect usages in an
academic article on vocabulary semantic analysis. The results from the

GOING BEYOND PATTERNS 17


World Wide Web include 18 tokens of powerful tea, with 13 being powerful
+ adjective + tea and 15 additional tokens of strong car. Yet a close
examination of the tokens in light of the meanings of strong and powerful
indicates that the collocations, including powerful + adjective + tea and
strong car, are not really arbitrary.
First, strong tea versus powerful tea is examined. A scrutiny of the tokens
shows that all the strong tea tokens (e.g., Examples 10, 11, and 12) refer to
the taste or smell of the tea, whereas those of powerful tea concern the force
or effect of some herbal tea (Examples 13, 14, and 15).

10. Kai . . . went into the main room to make herself a bracing cup of strong
sweet tea.
11. The heady odor of strong tea filled the kitchen.
12. For the first time, I noticed the smell of strong beef tea.
13. Poppy pod tea has been used as an old time powerful tea with many medical
purposes.
14. This is a very powerful immune boosting tea.
15. Wild Power Tea is a powerful nutrient-rich cleansing tea.

The distinctive meanings of strong versus powerful tea make very good
sense, based on core meanings of the two adjectives. According to the
OED, strong can express the meaning of a high degree in flavor, smell,
and taste, a meaning that is not in the semantic mapping of powerful. On
the other hand, one of the core meanings of powerful is producing great
effect or reaction, the very meaning used in the tokens of powerful tea
(although strong also has this meaning, it is not as intense as powerful in
expressing the meaning).
As for why both powerful car and strong car were found as results, the
question can also be answered in light of the semantic mappings of the
two adjectives. When the expression powerful car is used, the meaning
is that the car has a powerful engine that is capable of producing great
force, as can be seen in the following COCA example:

16. Only a powerful car can race uphill or overtake another without the gas
pedal being fully depressed and emissions at their worst.

This use of powerful in powerful car makes perfect sense because being
capable of exerting great physical force is one of the core meanings of
this adjective, a meaning that is not clearly present in the semantic
mapping of strong, where the closest meaning is exerting great muscular
force (OED, underline added). This adjective muscular suggests that
strong typically describes a human being or animal (not a machine) that
can produce great force. Then why do people sometimes say strong car as
shown in the corpus examples I found?

18 TESOL QUARTERLY
A reading of the 20 tokens of strong car or cars shows the following
usage and meaning distributions. Nine (a little less than one half) of the
tokens were used in writings about racing cars or car racing; six were
used in the captions of Internet video clips showing cars and trucks that
were either indestructible by very strong force or able to survive running
in extremely rough driving conditions; four tokens were used in articles
about the quality of some car models. The remaining token was in an
article talking about how it was now possible to use plastics material to
build strong cars. It is obvious that strong car used in this latter article and
in the captions of the six Internet video clips refers to cars that were
solidly made, capable of supporting strain or withstanding force; that is,
it refers to the superior structural strength of the cars involved. If this
meaning of strong car is examined in light of the semantic mapping of
the word strong, the use of strong in this sense is very logical because,
according to the OED, strong may be used to describe materials or
things that are capable of supporting strain or withstanding force;
solidly made.
Now the meaning of strong car is examined in the remaining 13
tokens. An examination suggests that, in some of the tokens, strong car
appears to refer to a powerful car, but in the others it is not clear
whether it means a powerful car, a solidly built car, or perhaps both, as
can be seen in the following examples (Examples 17 and 18 are from the
COCA, Example 19 is from the BNC, and Example 20 is from the World
Wide Web):

17. The 500-mile race will be more competitive, he said. Just one on one,
Im not sure you could pass the leader unless youve got an awful [sic]
strong car. I dont plan to be the one doing the passing, though.
18. A year into my seniority, indeed, the joys of Triumph driving had started
to wane. Fast car, strong car for its price and class, but after all, only a
common Triumph. Drivers of my sophistication hankered for a more
challenging machine.
19. The Volkswagen Polo has been reborn. No longer is it the slow, old
relative of a young family of quick and strong cars. It may have been
ignored since 1981 and left to make do with just two small engines.
20. JL Racings Saab 9-3 is ONE STRONG CAR. . . . The car spun around four
times before heading over a number of grass berms, and finally coming to
a halt. Luckily Jason managed to keep the car off the wall and get the car
back to pits. There was some minor damage with a bent tie-rod and the
brakes had some issues, but the team were able to get things going again
for Saturdays sessions.

In Example 17, because of the statement Im not sure you could pass
the leader unless youve got an awful strong car, it is almost certain that
strong car refers to the power of the car. Yet one cannot be sure what

GOING BEYOND PATTERNS 19


strong car really means in the remaining three examples. Example 20 is
particularly interesting. The description about the racing car being able
to finish the race after surviving the accident certainly suggests that
strong car here implies the solid structure of the car and its ability to
withstand destructive force, but it could also have two additional
meanings: One referring to the power the car had shown in finishing the
race, and the other, a metonymic use, meaning a car that is a strong
contestant or strong contender, that is, one with a strong chance of winning.
The fact that strong car has now been used in the sense of powerful car, as
shown in a few of the corpus examples, suggests that perhaps a semantic
conflation of strong and powerful in reference to the power of a car is
taking place, especially in the context of car racing. Some speakers seem
to have now expanded the meaning of strong from capable of exerting
muscular force to capable of exerting nonmuscular physical force. Such
a semantic shift is not entirely arbitrary, though, because strong used in
this sense still means capable of producing physical force, although no
longer force produced by muscles.
In fact, the semantically motivated use of powerful/strong in colloca-
tions can also be seen in the findings of a corpus study on the use of
powerful supporters versus strong supporters reported in Church and Mercer
(1993): Whereas strong supporters refers to those who are enthusiastic,
convinced, vociferous, etc., powerful supporters typically means those
who will bring others with them (p. 20). The reason for the different
use is not difficult to see, because to be a powerful supporter or to bring
many others along, one has to have power, be it political, economical, or
some other power. It is now clear that the use of powerful car and strong
tea, together with the newly discovered uses, is not really arbitrary but
semantically motivated.
Furthermore, a cognitive analysis of strong wind versus heavy rain,
another still widely used classic example for showing the arbitrariness of
collocations (e.g., Keshavarz & Salimi, 2007), shows that they are not
really arbitrary, either. As is well known, rain is made up of water and, as
such, it has weight. Therefore, the use of heavy to modify rain makes
perfect sense. In contrast, wind has basically no weight, but it has force.
Hence, the use of strong to describe wind is entirely logical. The two
collocations are obviously very semantically motivated based on human
experience. In short, all the earlier analyses indicate that most of the
collocations in the reference and teaching materials, including even
those often considered arbitrary, are in fact motivated. Such a finding is
in line with cognitive linguists argument that linguistic structure is a
direct reflex of cognition in the sense that a particular linguistic
expression is associated with a particular way of conceptualizing a given
situation (Lee, 2001, p. 1).

20 TESOL QUARTERLY
Dominant Practices in Collocation Teaching
The preceding data examination has demonstrated that most
collocations are not arbitrary but motivated. Therefore, an analysis
and understanding of their motivations should help students better
grasp them; that is, collocations should not be taught only as fixed
lexical units. However, as shown in the section on the arbitrariness of
collocations, many language educators (e.g., Hill, 2000; Lewis, 2000,
2002; Palmer, 1933) insist that collocations are arbitrary and that they
should be taught as lexical chunks without any compositional analysis. If
such a view is widely held and practiced, then changes should be made.
It is thus necessary first to determine the dominant collocation teaching
practices currently being used. To this end, I examined, in addition to
the four dictionary-textbook publications already mentioned, the articles
in Lewis (2000) edited book on collocation teaching, the section on
collocation teaching in Nations (2001, pp. 335343) and Willis (2003,
pp. 142167) books, three published studies on collocation teaching in
which the learning and teaching activities were described (Chan & Liou,
2005; Sun & Wang, 2003; Webb & Kagimoto, 2009), and three Web sites
that offer collocation learning and teaching activities (Kisito, 2010;
Macmillan English Campus, n.d.; Snashall, n.d.).
The results of the examination suggest that collocations are currently
taught primarily as fixed units, because most of the teaching activities,
including cross-linguistic comparison, are noticingmemorization in
nature, although a few experts (Nation, 2001; Willis, 2003) recommend
different learning techniques or methods for different types of colloca-
tions. For example, Nation (2001) suggests studying their parts and
history for learning idioms (for Nation, collocations include idioms),
finding patterns for collocations of some unpredictability (e.g., take
medicine), and paying attention to individual parts for very predictable
collocations like a clear day (pp. 335336). Willis (2003) recommends
organizing collocations by their meanings. Yet, overall, the most widely
used activities are the noticingmemorization type. They include, among
others, identifying or marking collocations in a passage or in collocation
dictionaries; reading passages with collocations highlighted or marked;
filling in the blanks with the right word in a collocation; choosing or
matching correct collocates; translating collocations from L2 back into L1
or vice versa; and memorization-type activities like repetition and
rehearsal (Hill, M. Lewis, & M. Lewis, 2000).
Recently, corpora, especially corpus concordancing, have been used
to help learners learn collocation patterns, but their use has been
limited to the function of exposing students to and helping them notice
and identify collocations (Chan & Liou, 2005; McCarthy & ODell, 2005;
Sun & Wang, 2003; Woolard, 2000). It does not involve any analysis of

GOING BEYOND PATTERNS 21


the reasons that words in collocations collocate the way they do. In fact,
my research failed to find any semantic-analysis collocation learning
activities that aim to find motivations of collocations, with the exception
of one activity in McCarthy and ODells (2005) book. In a unit titled
Everyday Verbs 1, which focuses on a comparison between the make
collocations and the do collocations, McCarthy and ODell (2005, p. 18)
write, If you remember that the basic meaning of make is about
producing something and the basic meaning of do is about performing
an action, then the collocations on this page may seem more logical.
Such an explanation should help students better understand and grasp
the collocations of the two verbs. Unfortunately, the earlier quote
appears to be all that the two authors offered about the motivations
behind the collocations in their entire series. For instance, in the next
two units after Everyday Verbs 1 (i.e., in Everyday Verbs 2 and
Everyday Verbs 3), in which collocations of some other common verbs
(e.g., get, go, have, etc.) are introduced, there is no discussion at all about
the issue of motivation behind any of the collocations of the verbs
involved. In other words, ODell and McCarthys (2008) discussion
about the motivations of collocations is indirect and inadequate. This
inadequacy might result from their belief that most of the collocations in
their series are not predictable, for they stated in the advanced book
(p. 8) that, in their series, they pay most attention to those
[collocations] that are not predictable. The phrase not predictable,
as I explained earlier, is largely synonymous with arbitrary in this context.
Yet, as I have shown in my data analysis and show again in the next
section, most collocations, especially those in the existing collocation
dictionaries, are not arbitrary, at least not completely.

INCORPORATING A COGNITIVE ANALYSIS IN THE


LEARNING AND TEACHING OF COLLOCATIONS
Rationale
The earlier examination has shown that, although collocations are
generally not arbitrary, they are currently taught mostly as prefabricated
chunks using primarily noticingmemorization strategies. This noticing-
and-memorization-only approach is problematic not only because it
ignores the motivated nature of most collocations but also, and more
importantly, because it takes away from the study of collocations any
cognitive and linguistic analysis, a very important and useful part of the
language-acquisition process. Wray (2000) has identified this problem in
the lexical approach to the teaching of formulae: It does not allow
students to generalize what they learn. Furthermore, such a learning

22 TESOL QUARTERLY
practice runs against new findings and theories in cognitive linguistics
about language use and learning. According to the cognitive linguist
Langacker (2008),

conventional usage almost always has conceptual motivation. Though it has


to be learned, it represents a particular way of construing the situation
described. With proper instruction, the learning of a usage is thus a matter of
grasping the semantic spin it imposes, a far more natural and enjoyable process
than sheer memorization. (Emphasis added, pp. 7273)

Based on such a belief about language usage and learning, cognitive


linguists have been very critical of the traditional practice of treating
language primarily as an arbitrary system:

Language has traditionally been understood as operating under its own set of
rules and properties, most of which have been assumed to be largely arbitrary,
idiosyncratic, and mysterious. . . . Lexical items with multiple meanings are
presented as homophones, with virtually no attempt to demonstrate any
motivated connections among the meanings. (Tyler, 2008, p. 458)

This criticism is especially valid for the current collocation teaching


practice. Changes are needed.
In fact, and more importantly, recent research has shown strong
positive effects of cognitive analysis on the learning of some fixed
expressions, such as figurative phrasal verbs and idioms (Boers, 2000;
Kovecses & Szabo, 1996). In an experimental study, Kovecses and Szabo
taught two groups of English as a foreign language students phrasal
verbs made up of the particles up and down (e.g., move up, go up, cut down,
die down). The students in the experimental group were made aware of
the conceptual metaphors that motivate the phrasal verbs, such as more is
up and less is down. The students in the control group learned the phrasal
verbs without exploring the motivating conceptual metaphors. The
results indicate that the experimental group significantly outperformed
the control group on the achievement test. Boers (2000) reports similar
findings in an experimental study involving phrasal verbs as well as other
figurative expressions. In fact, cognitive analysis can also enhance
students learning of other difficult aspects of language. In a study on
the learning of the preposition over, Tyler (2008) also reports results of
experimental studies that exhibit positive effects of cognitive analysis on
students learning of lexicogrammatical items. Furthermore, corpus-
based cognitive analyses have become an increasingly useful approach in
the research on lexicogrammar and have yielded many new insights into
lexical and syntactical usage patterns valuable for language learning (e.g.,
Gries & Stefanowitsch, 2006; Grondelaers, Geeaerts, & Speelman, 2007).

GOING BEYOND PATTERNS 23


Given these research findings and the fact that collocations are
generally motivated, researchers can expect that a cognitive analysis will
assist students in grasping collocations more effectively. Before I
continue, it is paramount to note that I am not suggesting that collocation
noticing, exposure, and memorization activities should be replaced with
cognitive analysis. I am only arguing for adding such analysis. Although I
see strong benefits of cognitive analysis in learning collocations, based on
the new research findings, I also understand the need for students to
notice, memorize, and repeatedly practice collocations to attain a good
grasp of them and a better English proficiency as a whole. Later, I use just
a few specific examples to show how incorporating such an analysis in the
corpus-based study of collocations may enhance students learning. The
corpus I use is again the COCA. There are, of course, other useful free
online corpora, including the BNC via the Brigham Young University
interface and the WebCorp (Birmingham City University, RDUES, 2009).
Teachers and learners can choose which one to use based on the learning
purpose, context, and setting.

Examples of Corpus-Based Cognitive Analysis and Useful


Strategies
I begin with the collocations of three everyday verbs listed in three
separate tables in McCarthy and ODells (2005) textbook. Table 2
combines, in a condensed fashion, the three tables of the noun
collocates of make, do, and have in the textbook (pp. 18 and 22). Because
their selection of the collocations was corpus-based, no corpus analysis is
needed. Only some cognitive analysis should be added. The only
cognitive semantic analysis included in the textbook is the statement
cited earlier: If you remember that the basic meaning of make is about
producing something and the basic meaning of do is about performing
an action (p. 18). Although the statement is helpful, more meaningful
analysis can be done to help students better differentiate the semantic
and usage patterns between the two verbs.
For example, one can ask students to discuss which verbs collocations
(the actions expressed) involve more initiation, planning, and effort. It
should not be difficult for them to figure out that the collocations of
make do; for example, making changes/choices/contributions/decisions/an
effort/friends typically entails more initiation, planning, and effort than
doing exercise/a favor/hair/homework/the ironing/shopping, because the latter
actions are much more routine, and often they are things that one is
asked or required to do. Then having students examine the core meaning
of make should help them understand the differences between the two
verbs usage patterns, because the core meaning of make is create, an

24 TESOL QUARTERLY
TABLE 2.
Typical Collocations of Everyday Verbs Introduced in McCarthy and ODell (2005)

Make Do Have
Make arrangement for Do your best Have an accident
Make a change or changes Do damage Have an argument or a row
Make a choice Do an experiment Have a break
Make a comment or comments Do exercises Have a conversation or a chat
Make a contribution to Do someone a good turn or Have difficulty
Make a decision Do someone a favor Have a dream or a nightmare
Make an effort Do harm Have an experience
Make an excuse Do your hair Have a feeling
Make a friend Do your homework Have fun or a good time
Make an improvement Do the ironing or shopping Have a look
Make a mistake or washing, etc. Have a party
Make a phone call Do some work Have a problem or problems
Make progress Have a try or a go

activity whose process involves a substantial amount of initiation,


planning, and effort, a point discussed earlier. Now the unique semantic
features of the have collocations are discussed. No semantic analysis about
them is included in the textbook, perhaps because of the authors belief
that many of the collocations are surprising, that is, unpredictable
(McCarthy & ODell, 2005, p. 22). Yet the fact is that the have collocations
are not surprising, and some interesting analysis can be effectively done to
help students understand why they are so used. For example, as my
analysis of the Have a good/wonderful trip collocation in the last section
indicates, one of the main meanings of have collocations is experience.
In fact, almost all the have collocations in McCarthy and ODells list
express this meaning, for have in them can be replaced by experience:
experience (instead of have) a/an accident/break/difficulty/dream/feeling/fun/
problem, etc. To draw students attention to this fact, the researcher can ask
the question: What does the verb have really mean in these collocations?
and they should not have much difficulty answering it. An understanding
of this point should in turn help them better distinguish the have
collocations from the make and the do collocations.
A cognitive analysis may help students better grasp the use of be/get on
(not at/in) the phone, a verb + preposition + noun collocation. The reason
for choosing this collocation is that many of my students have asked me
why one says on the phone, not *at/in the phone. It seems rather arbitrary to
them. To help students understand that this is really a motivated usage,
researchers can first have students conduct a COCA search of the
prepositions used in the phrase by entering [be] [i*] the phone and
[get] [i*] the phone. The results show that the only prepositions used
systematically in the given collocations are on and off with 1,118 tokens of
be on the phone, 262 get on the phone, 17 be off the phone, and 202 get off the
phone. Then students can be asked to compare the meanings of the four

GOING BEYOND PATTERNS 25


different collocations, especially the meaning of on the phone versus that of
off the phone. The students can fairly easily determine that be on the phone
means be in the process of having a phone conversation and get on the
phone means pick up the phone to start a conversation, whereas be/get off
the phone conveys the opposite meaning. Then, one should ask the
students why on and off help convey opposite meanings. By checking the
meanings of the two prepositions in the dictionary, they learn that one of
the extended meanings of on is being in operation/use (a meaning that
is not in the semantic mappings of upon, at, etc.), and one of the extended
meanings of off is not being in operation/use. This knowledge should
enable the students to see the motivations for the collocations involved.
Finally, there are some useful collocation learning and teaching
strategies that incorporate corpus-based cognitive analysis. Here are a few:
1. Comparing-contrasting-explaining differences between similar pairs of
collocations (e.g., make/take/have a trip and the typical make versus take
versus have versus do collocations).
2. Examining the motivations of collocations in comparison or contrast with
their counterparts in learners L1 (when learners have a homogeneous
L1), for example, take medicine in English versus eat medicine in Chinese.
The English use of take here can be attributed to the broad semantic
mapping of the verb and its extended meaning of to receive into ones
body by eating, drinking, swallowing, for example, take food, nutrition, etc.
(OED online), whereas the Chinese use of eat is quite understandable,
thanks to its extremely broad semantic mapping in Chinese, for example,
eat a loss (suffer a loss) and eat bitterness (endure hardship).
3. Making good use of corpora and dictionaries (both general learner
dictionaries and collocation dictionaries) in identifying collocations and
learning their motivations.
4. Organizing collocations by meaning, based on semantic motivations
rather than in an undifferentiated way.

Benefits
Besides helping learners better understand collocations, corpus-based
cognitive analysis can also help learners use collocations more
productively than the noticingmemorization approach can. This is
because there are simply too many collocations to memorize; for
example, learners cannot memorize all the collocations with the verbs
make and have mentioned earlier. Knowing the different motivations for
the typical collocations of the two verbs may help students understand
and use the verbs more accurately. Of course, because not all the
collocations are motivated, at least not clearly motivated, sometimes a
search for the motivation(s) for a collocation may fail. Yet, despite the
failed attempt, the cognitive exploration process that the students went

26 TESOL QUARTERLY
through should still be beneficial, because of the additional opportunity
for processing the collocation and the extra attention the students paid
to its composition. The exploration process should raise students
consciousness of the collocation, which should in turn result in better
retention. Another likely positive value of a cognitive analysis in the
learning of collocations is that it may not only enhance students grasp
of the collocations being learned but also enable students to gain a
better understanding of the key words in the collocations. For example,
in the process of gaining an understanding of the semantic motivations
of make a trip versus take a trip, the students should simultaneously learn,
explicitly or implicitly via corpus examples, the core meanings of the two
verbs involved, which can in turn lead them to a better understanding
of the make + noun collocations versus the take + noun ones in general. By
the same token, a corpus analysis of chief + noun and main + noun
collocations may enhance students grasp of the core meanings of the
two adjectives and their overall usage patterns.

Caveats and Challenges


It is paramount to note, however, that there are many caveats and
challenges in the use of corpus-based cognitive analyses in collocation
teaching. First, the approach will not work well for young children because
of their limited cognitive ability. Second, it may not work for students who
do not benefit from cognitive analysis because of their learning styles.
Third, the approach should be used cautiously with students of low
language proficiency. For these students, raw corpus data should be
avoided. Instead, the teacher can select, from the raw data, examples that
suit the students level and ability. Fourth, the level of complexity of
cognitive analysis in collocation learning can vary substantially from one
collocation to another, thus presenting different levels of challenge. For
example, the reason for the use of the preposition between in the
collocation a gulf between a and b may be somewhat easier to understand
than the reason for the use of on in the collocation be on the phone. Although
the core meaning of the preposition between is used in the gulf between
collocation, it is not the case with the use of on in on the phone, as explained
earlier. Because of the difference in the complexity of cognitive analyses
involved in collocation learning, a teacher should be prepared to provide
appropriate assistance to the students, based on the level of the difficulty of
the collocations being learned. Visual aids including diagrams are an
especially useful tool. For example, to help students understand the use of
between instead of in or any other prepositions in the gulf between
collocation, one can draw a picture with a gulf (lake/river) in the middle
and the A person/thing on one side of the gulf and the B person/thing on

GOING BEYOND PATTERNS 27


the other side. Viewing this picture, students can see clearly the reason for
using between, rather than one of the other prepositions. Similarly, a
diagram can be drawn to help illustrate the semantic difference between
the high and tall collocations.

CONCLUSION
With the use of a cognitive analysis of some collocations in corpus data,
this article has attempted to demonstrate that most collocations in the
existing collocation dictionaries and textbooks, including even many of
those traditionally considered arbitrary, are motivated, if viewed in light of
the semantic mappings of the key lexical items involved. Based on this
finding about the nature of collocations, new research findings on the
positive effects of cognitive analysis on the learning of lexical items, and
cognitive linguistic theories about language use and learning, the article
argued for the need to include a cognitive analysis in the learning of
collocations, in addition to noticing, memorization, and other appropriate
learning activities. Also discussed with examples was how such an analysis
may not only help students grasp collocations more effectively but may also
enable them to gain a better understanding of the meanings and usages of
the key words in the collocations. The article addressed the caveats and
challenges in using the proposed teaching practice. Empirical research
testing the effectiveness of including cognitive analysis in collocation
teaching is the logical next step. Such research is necessary in helping us
gain a better understanding not only of the teaching of collocations but
also of the value of a cognitive approach to language teaching in general.

ACKNOWLEDGMENTS

I would like to thank the three anonymous reviewers and TESOL Quarterly editor
Diane Belcher for their extremely valuable comments and suggestions. They have
helped me significantly enhance the quality of this article.

THE AUTHOR

Dilin Liu is an associate professor and coordinator of applied linguistics/TESOL in


the English Department at the University of Alabama, Tuscaloosa, Alabama, United
States. His main research interests are grammar and vocabulary learning and
teaching and corpus-based language description and learning.

REFERENCES

Bahns, J., & Eldaw, E. (1993). Should we teach EFL students collocations? System, 21,
101114.

28 TESOL QUARTERLY
Benson, M. (1989). The structure of the collocational dictionary. International Journal
of Lexicography, 2, 114.
Benson, M., Benson, E., & Ilson, R. (1997). The BBI dictionary of English word
combinations. Amsterdam, The Netherlands: John Benjamins.
Birmingham City University, Research and Development Unit for English Studies.
(2009). Webcorp [Web site]. Birmingham, England: Author. Retrieved from
http://www.webcorp.org.uk.
Boers, F. (2000). Metaphor awareness and vocabulary retention. Applied Linguistics,
21, 553571.
Chan, T., & Liou, H. (2005). Effects of Web-based concordancing instruction on EFL
students learning of verbnoun collocations. Computer Assisted Language Learning,
18, 231250.
Church, K. W., & Mercer, R. L. (1993). Introduction. In S. Armstrong (Ed.), Using
large corpora (pp.124). Cambridge, MA: MIT Press.
Crowther, J., Dignen, S., & Lea, D. (Eds.). (2002). Oxford collocations dictionary for
students of English. Oxford, England: Oxford University Press.
Davies, M. (2008). Corpus of contemporary American English online. Retrieved
from http://www.americancorpus.org/.
Dirven, R., & Taylor, J. R. (1988). The conceptualisation of vertical space in English:
The case of Tall. In B. Rudzka-Ostyn (Ed.), Topics in cognitive linguistics (pp. 379
402). Amsterdam, The Netherlands: John Benjamins.
Fernando, C. (1996). Idioms and idiomaticity. Oxford, England: Oxford University Press.
Finch, G. (2000). Linguistic terms and concepts. New York, NY: St. Martins.
Gries, S. T., & Stefanowitsch, A. (2006). Corpora in cognitive linguistics: Corpus-based
approaches to syntax and lexis. New York, NY: Mouton De Gruyter.
Grondelaers, S., Geeraerts, D., & Speelman, D. (2007). A case for a cognitive corpus
linguistics. In M. Gonzales-Marquez, I. Mittelberg, S. Coulson, & M. Spivey (Eds.),
Methods in cognitive linguistics (pp. 149169). Amsterdam, The Netherlands: John
Benjamins.
Hill, J. (2000). Revising priorities: From grammatical failure to collocational success.
In M. Lewis (Ed.), Teaching collocation: Further development in the lexical approach
(pp. 4769). Hove, England: Heinle & Heinle.
Hill, J., & Lewis, M. (2002). LTP dictionary of selected collocations. Boston, MA: Heinle &
Heinle.
Hill, J., Lewis, M., & Lewis, M. (2000). Classroom strategies, activities and exercises.
In M. Lewis (Ed.), Teaching collocation: Further development in the lexical approach
(pp. 86117). Hove, England: Heinle & Heinle.
Hunston, S. (2002). Corpora in applied linguistics. Cambridge, England: Cambridge
University Press.
Hunston, S., & Francis, G. (2000). Pattern grammar: A corpus-driven approach to the
lexical grammar of English. Amsterdam, The Netherlands: John Benjamins.
Keshavarz, M. H., & Salimi, H. (2007). Collocational competence and cloze test
performance: A study of Iranian EFL learners. International Journal of Applied
Linguistics, 17, 8192.
Kisito, F. (2010). ESL galaxy [Web site]. Retrieved from http://esl-galaxy.com.
Kovecses, Z., & Szabo, P. (1996). Idioms: A view from cognitive semantics. Applied
Linguistics, 17, 326355.
Langacker, R. W. (2008). Cognitive grammar as basis for language instruction. In P.
Robinson & N. Ellis (Eds.), Handbook of cognitive linguistics and second language
acquisition (pp. 6688). New York, NY: Routledge.
Lee, D. (2001). Cognitive linguistics: An introduction. Oxford, England: Oxford
University Press.

GOING BEYOND PATTERNS 29


Lewis, M. (2000). Teaching collocation: Further development in the lexical approach. Hove,
England: Heinle & Heinle.
Lewis, M. (2002). Implementing the lexical approach. Boston, MA: Thomson/Heinle.
Macmillan English Campus. (n.d.). One stop English [Web site]. Basingstoke,
England: Author. Retrieved from http://www.onestopEnglish.com.
McCarthy, M., & ODell, F. (2005). English collocations in use: Intermediate. Cambridge,
England: Cambridge University Press.
Moon, R. (1998). Fixed expressions and idioms in English: A corpus-based approach.
Oxford, England: Clarendon Press.
Nation, P. (2001). Learning vocabulary in another language. Cambridge, England:
Cambridge University Press.
Nesselhauf, N. (2003). The use of collocations by advanced learners of English and
some implications for teaching. Applied Linguistics, 24, 223242.
ODell, F., & McCarthy, M. (2008). English collocations in use: Advanced. Cambridge,
England: Cambridge University Press.
Oxford English Dictionary (OED) Online. (2008). Retrieved from http://dictionary.oed.
com.
Palmer, H. E. (1933). Second interim report on English collocations. Tokyo, Japan:
Kaitakusha.
Richards, J., Platt, J., & Webber, H. (1985). Longman dictionary of applied linguistics.
Harlow, England: Longman.
Sinclair, J. M. (1991). Corpus, concordance, collocation. Oxford, England: Oxford
University Press.
Smadja, F., & McKeown, K. (1991). Using collocations for language generation.
Computational Intelligence, 7, 229239.
Snashall, P. (n.d.). eslflow.com [Web site]. Retrieved from http://www.eslflow.com.
Sun, Y., & Wang, L. (2003). Concordancers in the EFL classroom: Cognitive
approaches and collocation difficulty. Computer Assisted Language Learning, 16,
8394.
Tsui, A. (2004). What teachers have always wanted to knowand how corpora can
help. In J. Sinclair (Ed.), How to use corpora in language teaching (pp. 3961).
Amsterdam, The Netherlands: John Benjamins.
Tyler, A. (2008). Cognitive linguistics and second language instruction. In P.
Robinson & N. C. Ellis (Eds.), Handbook of cognitive linguistics and second language
acquisition (pp. 456488). New York, NY: Routledge.
Webb, S., & Kagimoto, E. (2009). The effects of vocabulary learning on collocation
and meaning. TESOL Quarterly, 43, 5577.
Willis, D. (2003). Rules, patterns and words: Grammar and lexis in English language
teaching. Cambridge, England: Cambridge University Press.
Woolard, G. (2000). Collocation: Encouraging learner independence. In M. Lewis
(Ed.), Teaching collocation: Further development in the lexical approach (pp. 2846).
Hove, England: Heinle & Heinle.
Wouden, T. V. D. (1997). Negative contexts: Collocation, polarity and multiple negation.
London, England: Routledge.
Wray, A. (2000). Formulaic sequences in second language teaching: Principle and
practice. Applied Linguistics, 21, 463489.

30 TESOL QUARTERLY

You might also like