Identifying (In)definiteness: A Questionnaire

Veneeta Dayal
Background

This questionnaire seeks to determine whether bare XPs, those with no overt determiners, are
interpreted as definites, indefinites, or both. It identifies some of the core properties associated
with definites and indefinites and tests for them separately. The possible outcomes can be
represented in the following way:

Definite Indefinite
Context 1 Yes Yes
Context 2 Yes No
Context 3 No Yes
Context 4 No No

Examples from English, which is the language used for demonstration in this questionnaire,
illustrating three of the four logically possible contexts are given below:

1. In the classroom, a teacher/the teacher was preparing her lesson plan.

2. A boy and a girl entered the room. The boy/*a boy sat down.
3. Once upon a time, a king/*the king ruled this country.

If a bare XP is acceptable in Context 1, it does not determine its status as a definite or an

indefinite. The data is compatible with it being only definite, only indefinite, or ambiguous.

If a bare XP is acceptable in Context 2, it tells us that it is a definite. It does not tell us that it is
not an indefinite. If it is unacceptable in Context 2, it tells us that it is not a definite. It does not
tell us that it is an indefinite.

If a bare XP is acceptable in Context 3, it tells us that it is an indefinite. It does not tell us that it
is not a definite. If it is unacceptable in Context 3, it tells us that it is not an indefinite. It does not
tell us that it is a definite.

By combining results from Contexts 2 and 3, we can get a definitive answer on the status of the
bare XP.

Context 2 Context 3
Definite only Yes No
Indefinite only No Yes
Ambiguous Yes Yes

This is a simplified picture, for three reasons. One, the presence or absence of a lexical
definite/indefinite determiner may or may not affect possible readings (see blocking principle in
Chapter 1). Two, definiteness and indefiniteness may not be identifiable by a single context. That
is, there may be more than one type of definite or indefinite (see indefinites in Chapter 1). Three,

1
contexts have to be carefully controlled to safeguard against the covert introduction of
information. If the sentences in contexts in (1) and (3) are taken as discourse initial, there should
be no difference in the felicity of the teacher and the king. However, the mention of a classroom
in context 1 is sufficient to allow the use of a definite for the teacher, in a way that the mention
of a country does not seem be for the felicity of the king (see accommodation in Chapter 1).

1. Testing for Definiteness

There are two basic issues to sort out with definiteness. Does the language have a definite
determiner? Does the bare XP behave like a definite description?

1.1. Definite or Demonstrative?

Demonstratives and definites are close kin. Definite determiners are thought to have developed
diachronically from demonstratives. While most, if not all, languages have demonstratives, many
languages do not have definite determiners. The existence of definite determiners can have an
impact on the interpretation of bare XPs, so it is worth determining if a language has definite
determiners. The primary diagnostic for this tests uniqueness/maximality of the set associated
with the determiner.

In the following, X stands for the lexical item (here, the determiner) we are testing:

Context A: there is a bunch of roses and the speaker points to one of the roses in the bunch and
says:

2a. I’ll take _X_ rose.

2b. I’ll take that rose.
2c. #I’ll take the rose.

If (2a) is acceptable, we are dealing with a demonstrative, if not we are most likely dealing with
a definite determiner. The English paradigm in (2b)-(2c) provides the relevant comparison. This
shows that the definite determiner has a uniqueness requirement – the set it is associated with
must be a singleton.

This can be generalized to plurals, with the caveat that it now applies to the (unique) maximal
entity in the set – ie the totality of all the elements in the set:

Context B: there are several bunches of roses and the speaker points to one of the bunches and
says (3a). It is important that the speaker not be referring to all the roses (in all the bunches) in
the context:

3a. I’ll take _X_ roses.

2
3b. I’ll take those roses.
3c. #I’ll take the roses.

As a corollary, demonstratives have an anti-uniqueness implicature. The anti-uniqueness

implicature is clearly evident in contexts where world knowledge establishes the N-set as a
singleton (see deictic definites in Chapter 1):

4a. _X_ sun is shining today.

b. #That sun is shining today.
c. The sun is shining today.

Note that it is possible to point to the single bunch of roses in Context A and say “I’ll take that
bunch of roses” or to point to all of the bunches in context B and say “I’ll take those bunches of
roses”. The test in (4) controls for this. The following are some other possibilities:

5a. #That President of India held a press conference today.

b. The president of India held a press conference today.

Yet another context that makes the same point has to do with cases that involve a functional
relationship between an antecedent and the XP (see bridging in Chapter 1):

6a. Helen bought a car. #That steering wheel/The steering wheel was defective.
b. Mary was watching a match. #That umpire/The umpire was unfair.

These three tests should establish whether X is a demonstrative. We now turn to some other tests
for definiteness, specifically useful for bare XPs.

1.2. Anaphoricity

Here we test whether a bare XP can be classified as a definite. In administering this test, it is
important that there be two distinct entities introduced in the first sentence. This is to make the
use of a full XP, rather than a pronoun, obligatory:

7a. A boy and a girl walked into the room. [XP Girl] was wearing a red shirt, [XP boy] was
wearing a white shirt.
7b. A boy and a girl walked into the room. [XP Girl] told [XP boy] to sit down.

If the XPs in the second sentence are acceptable, the XP passes at least one of the tests for
definiteness. But there are languages which could be argued to use a (weak) demonstrative in
some anaphoric contexts so the additional test given below is necessary.

1.3. Homogeneity

This test builds on the notion of uniqueness/maximality to test whether a bare XP can be
classified as a definite. What is needed here are two predicates that are incompatible, that is, two
predicates that cannot hold of a single individual at the same time. One can vary the tense. One

3
can also change the conjunction to ‘and’ if one can safeguard against a sequential interpretation
where the two predicates are read as applying to the same individual at different times. In the
comparison provided by the English sentences in (8b) and (8c), we see that the definite
description is infelicitous but the demonstrative is fine.

Some people find it possible to make (8c) and (9c) acceptable by using “the other dog”, “the
other student” in the second clause. To the extent that this is possible, it is only possible if the
first use of the definite refers to an entity that is more salient than the others. This can be offset
by creating a context in which there are several entities of the relevant type (dogs/students) and
no single one, or no pair of such entities, stands out from the others.

Context: The speaker is describing what she/he saw when he looked into the room earlier.

8a. XP was sleeping, while XP was running around.

b. That dog was sleeping while that dog was running around.
c. #The dog was sleeping while the dog was running around.

9a. XP was working on a math problem, while XP was drawing a picture.

b. That student was working on a math problem, while that student was drawing a picture.
c. #The student was working on a math problem, while the student was drawing a picture.

2. Testing for Indefiniteness

As in the case of definiteness, there are two issues to sort out with indefiniteness. Does the
language have an indefinite determiner? Does the bare XP behave like an indefinite description?
The second question is complicated by the fact that indefinite descriptions fall into more than
one class – the specific/non-specific divide interacts with the definite/non-definite divide in ways
that calls for some nuance in creating the tests.

2.1. Indefinite or Numeral?

The indefinite determiner in English has been argued to have developed from the numeral one,
and in many other languages the indefinite determiner is homophonous with the numeral.
Arguably, every language has expressions corresponding to the numeral one, but it is still
possible to establish whether a language has an indefinite determiner. There are three tests that
can be used.

The first is the ability to scope under negation:

10a. I didn’t buy X book.

b. I didn’t buy a book.
c. I didn’t buy one book.

4
(10b) is compatible with no books being bought: ¬∃ reading. (10c) is typically thought to give
wide scope to the numeral: ∃¬. There is a book that was not bought, while others may well have
been. To the extent that the first reading is possible, it is only possible under an emphatic
reading, akin to I didn’t buy even one book. This reading can be brought out by adding prosodic
stress to one in (10c).

The second test is the ability to yield generic readings, readings that apply to all/most members
of a species:

11a. X cow is a mammal/eats grass.

b. A cow is a mammal.
c. A cow eats grass.
d. #One cow is a mammal.
e. #One cow eats grass.

These are simple generic sentences. It is not clear to me whether more complicated sentences
such as only a cow eats grass also provide a clear diagnostic for separating indefinite
determiners from numeral one.

Finally, it is possible for an indefinite to get quantificational reading:

12a. X dog is rarely vicious/usually likes meat.

b. A dog is rarely vicious.
c. A dog usually likes meat.
d. #One dog is rarely vicious.
e. #One dog usually likes meat.

The most salient reading of (12b) is that few dogs are vicious, and that of (12c) that most dogs
like meat. These readings are not possible in (12d) and (12e). Note that in this respect, some also
resists generic readings, in the singular as well as the plural (but see genericity in Chapter 1
where contrasts between someone who wants to succeed works hard and some student who wants
to succeed works hard are discussed).

These three tests should establish whether X is an indefinite determiner. We now turn to some
other tests for indefiniteness, especially useful for bare NPs.

2.2. Introducing Referents

One of the key properties of indefinites is the ability to introduce new discourse referents. To test
a bare XP’s potential to do so, story-telling contexts are useful because they rule out the
possibility of shared knowledge. It is important to test the following with singular as well as
plural terms, as the results may differ:

13a. Many years ago, an old woman used to live in this house.
b. Once upon a time, a king used to rule this country.
c. Speaker A: Why is the Principal upset?

5
 Speaker B: A teacher gave a student a very bad grade. The student’s parents are
complaining.
d. Speaker A: What do you see outside?
Speaker B: A woman is giving a toy to a child. She is telling him to be careful with it.
e. Speaker A: After the talk, the teacher introduced a student to a politician.

In all these cases, it can be assumed that this is the first mention of an old woman, a king, a
teacher, a student, a very bad grade, a woman, a toy, a child, a patient. These are indefinites that
occur in a variety of syntactic contexts. Also, these are random descriptions rather than scripted
sets (see scripted sets in Chapter 1).

2.2. Indefinites and Negation

Ordinary indefinites are assumed to have the ability to take wide or narrow scope with respect to
negation, where by ordinary indefinites I mean those with the indefinite determiner a.

14a. I didn’t finish a multiplication problem on the test. I think I’ll do well – after all, I did all the
division problems and all the other multiplication problems.
14b. Mary didn’t see a book that I had left on the table.

In both (14a) and (14b) the indefinite has scope over negation: ∃¬. The existence of a
multiplication problem and a book left on the table is not in doubt. In (14a) there are other
multiplication problems that the speaker did finish; in (14b) the book in question may well be the
only one in the context, so it is compatible with Mary seeing no books though the existence of
the book is not in question.

2.3. Partitive Specificity

Ordinary indefinites can refer back to a subset of individuals introduced in the previous sentence:

15a.There were many children in the room. A boy and a girl were playing cards.
b. Mary bought ten books and three magazines. She gave a book to her friend and a magazine
to her sister. The rest she kept for herself.

It is worth testing with plural XPs as well. In the singular case, an XP should be acceptable if it is
an indefinite and unacceptable if it is not. In the plural case, an XP should be acceptable and
refer to some of the N mentioned in the first sentence if it is an indefinite. If it is a definite, it
should be acceptable but should refer to all of the Ns mentioned in the first sentence.

2.4. Narrow and non-narrow Scope

16a. Every student read every article on XP.

b. Every student read every article on a current topic.

Three possible readings for XP:

6
Narrow scope: ∀student ∀article ∃current-topic
Every student read every article, where the article could be on any current topic.

Wide scope: ∃current-topic ∀student ∀article

There is a current topic such that every student has read every article on it. Let’s say that topic is
the repeal of ObamaCare.

Intermediate scope: ∀student ∃current-topic ∀article

For every student there is a possibly different topic, such that that student has read every article
on that topic. So student A has read every article on the repeal of ObamaCare and student B has
read every article on tax reform and so on.

2.5. Definiteness vs. Specificity

If an XP in a language without determiners has a non-narrow reading, it needs to be settled

whether it is a definite reading of the XP or a specific indefinite reading. The distinction is subtle
since in both cases, there is a unique entity being referred to. In the case of a definite, both
speaker and hearer are, in principle, able to identify that referent. In the case of an indefinite, the
speaker may have a unique individual in mind and the hearer knows that, but the referent is not
identifiable by the hearer, at least in principle.

Another way to think about this difference is that the N-set is known to be a singleton in the case
of a definite, whereas in the case of a specific indefinite there should be more than one entity that
potentially belongs in the N-set (see Specificity vs. Definiteness in Chapter 1).

This is best illustrated by an example that forces the XP to takes scope below a universal.

17a. Every teacher bought two books for a student of his/hers.

b. Every teacher bought two books for his/her student.
c. Every doctor asked for two reports from a nurse assisting him/her.
d. Every doctor asked for two reports from the nurse assisting him/her.

Here the pronoun his/hers is to be read as bound by every teacher/every doctor. This means that
the indefinite or definite has to take scope below every N. If the indefinite a student of his/hers in
(17a) takes scope above two books, we get a reading where for each teacher there is a particular
student out of several students that he/she has, such that he/she gave two books to that student.
This reading can be made explicit with the use of a certain student of his/hers. The same holds
for (17c).

With the definite in (17b) and (17d), we require each teacher/doctor to have only one
(contextually salient) student/nurse, to whom he/she gives two books or of whom he/she requests
two reports. That is, there is a uniqueness presupposition, relativized to individual
teachers/doctors, that the whole sentence has.

7
2.6. Differentiated Scope

This is a test used to separate out bare plurals in English from ordinary indefinites. It is worth
testing if the bare XP in the language under investigation has this property. It is important to test
both singular and plural XPs (if the language makes that distinction) since they may well yield
different results.

18a. Miles killed rabbits repeatedly.

b.# Miles killed a rabbit/some rabbits repeatedly.

19a. Miles baked cakes for two days.

b.# Miles baked a cake for two days.

20a. Rabbits kept dying all winter.

b.# Some rabbits/A rabbit kept dying all winter.

These tests show that ordinary indefinites cannot take scope below certain aspectual/adverbial
operators and this leads to infelicity with verbs of creation and destruction. The same does not
hold of bare plurals in English. The test will determine whether the absence of a determiner is
what is critical, or whether number specification also plays a role.

2.7. Controlling for Syntactic Position

In contexts that are equally compatible with new and old information and equally compatible
with uniqueness/non-uniqueness (ie with singleton or non-singleton N-sets), a variety of
syntactic positions needs to be checked. This has been built into some of the tests given above.
Here we use adverbials to test whether a singular XP is necessarily has a singleton N-set.

21a. I read [XP book] twice/again.

b. I gave [XP student] a book twice/again.
c. [XP child] is eating a cookie a second time/again.

If the N-set can have more than one member in its denotation, different Ns can participate in
different sub-events. If the N-set is a singleton, the sub-events will involve the same entity. It is
important to check this in different syntactic positions to see whether the adverbials allow for
different books, students, children to be involved in the two events.

8
3. Testing for Reference to Kinds

In this section we try to determine the status of bare XPs as generic and kind denoting terms.

3.1. Generic Statements

In testing for genericity, the aspectual specification is important. It should allow for the
predication to apply across times/situations/worlds in an open-ended way. In English, the simple
present is the canonical aspect for this purpose, though the simple-past can also be used in this
way. Typically, the progressive does not lend itself to this use. In testing for this, both singular
and plural forms of the XP should be tested as they may behave differently.

22a. XP barks/bark when it/they is hungry.

b. Dogs bark when they are hungry.
c. The dog barks when it is hungry.
d. A dog barks when it is hungry.

23a. XP is/are a ferocious animal/ferocious animals.

b. Lions are ferocious animals.
c. The lion is a ferocious animal.
d. A lion is a ferocious animal.

24a. XP usually drinks/drink wine with dinner.

b. Italians usually drink wine with dinner.
c. ? The Italian usually drinks wine with dinner.
d. An Italian usually drinks wine with dinner.

Many generic statements can be read as habitual statements if applied to an ordinary individual:
Dogs bark when they are hungry (generic) vs. Fido barks when he is hungry (habitual). The
definite plural also gets a habitual reading. In fact, the definite singular is also able to have a
habitual reading. The status of the indefinite singular is a bit less clear.

3.2. Reference to Kinds

In this final section we test for genuine reference to kinds by taking predicates that can only hold
of the species as a whole, not of its individual members. Such predicates include, be extinct, be
endangered, evolve from, invent/invented by. In testing for reference to kinds in a language, we
want to make sure it is a kind-level predicate by ensuring that it is ungrammatical to predicate it
of an ordinary individual: #Barney is extinct, #Fido is an endangered species/is endangered. It is
also important to test both the singular and the plural version of the bare XP, as reference to
kinds can behave differently in the singular and the plural.

9
25a. XP is/are extinct.
b. Dinosaurs are extinct.
c. The dinosaur is extinct.
d. #A dinosaur is extinct.
e. #The dinosaurs are extinct.

26a. XP is/are an endangered species.

b. Lions are an endangered species.
c. The lion is an endangered species.
d. #A lion is an endangered species.
e. #The lions are an endangered species.

27a. XP has/have evolved from YP

b. Dogs have evolved from wolves.
c. The dog has evolved from the wolf.
d. #A dog has evolved from a wolf.
e. #The dogs have evolved from the wolves.

28a. XP was/were invented by Babbage.

b. Computers were invented by Babbage.
c. The computer was invented by Babbage.
d. #A computer was invented by Babbage.
e. #The computers were invented by Babbage.

29a. Babbage invented XP.

b. #Babbage invented computers.
c. Babbage invented the computer.
d. #Babbage invented a computer.
e. #Babbage invented the computers.

The example sentences in English bring out four points:

(i) the indefinite determiner may function quite well in generic statements, but it is not a basic
kind referring term
(ii) while a plural bare XP is a kind denoting term, it is not good in the object position of the verb
invent, though it is perfectly suitable as the theme of that verb when it is in subject position
(iii) the indefinite is infelicitous only if it is meant to refer to the whole species – several of the
(d) sentences above can become acceptable if it is possible to read them as referring to one
subspecies (one of the lion subkinds, one of the dog/wolf subkind, one type of computer etc), the
taxonomic reading.
(iv) alongside the inability of English indefinite singulars to refer to kinds, is the inability of
English definite plurals to refer to kinds. Again, they may become acceptable if the sentence can
be construed to refer to (all) the sub-kinds. Note that Romance languages differ in this respect,
allowing and even requiring the definite determiner for reference to kinds.

10