F—‘fi ____, . _ ._

PLACENBE‘I’UM Boxwmmmbmuhmywncord.
TO AVOID FINES rutunonorbdonddodm.

DATE DUE 3' DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU I. An Affirmative Action/Equal Opponunlty Instltmlon
WWI

 

A STUDY OF THE ARTICLE SYSTEM IN ENGLISH

BY

Pi-fen Liu Chen

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Linguistics
and Germanic, Slavic, Asian, and African Languages

1988

Abstract
A STUDY OF THE ARTICLE SYSTEM IN ENGLISH

By
Pi-fen Liu Chen

This dissertation aims to answer the following
questions frequently asked by adult second language learners
of English: (a) when do we use Mm? (b) when do we use
000? and (c) when do we use neither dw'nor 000? This
study discusses mass and count nouns in English, and what
makes an NP definite and what makes it indefinite. It also
discusses the generic use of English articles. It is argued
that not every mass noun can be converted into a count noun
and vice versa. Four principles are given for mass/count
conversion. For (in)definiteness, three requirements--
existence, uniqueness, and familiarity-- are posited for the
use of Mn; Subtle differences among the generic use of
Mm, 000 and O, the zero article, are discussed and their
respective distributions and restrictions are presented. An
over-all system of article usage is presented for second

language learning.

Copyright by
Pi-fen Liu Chen

1988

Dedicated to

My teacher, Dr. Barbara Abbott

iv

ACKNOWLEDGEMENTS

I would like to express my deepest gratitude to Dr.
Barbara Abbott, my dissertation director, for her teaching
and various tremendous help in the past five years. I am
particularly indebted to her for being my most frequently
consulted informant. Her speedy and helpful comments on my
dissertation drafts have led to considerable improvements in
the final version. I am grateful to her for answering my
never-ending questions. Her generosity in lending me books
and articles related to my topic is also greatly
appreciated.

I also wish to extend my appreciation to the other
members of my committee. My thanks go to Dr. Ruth Brend for
her numerous comments on my draft and for providing contexts
for sentences about which I did not have good intuitions. I
owe my thanks to Dr. Grover Hudson for his comments and
corrections. I am grateful to Professor James Wang for his
generous support and encouragement throughout my graduate
career at Michigan State University. I would like to
express my special thanks to Dr. Marcellette Williams for
serving as college representative at my oral defense and for
her comments and corrections.

Those who kindly answered my questionnaire deserve my
special gratitude, too. I am particularly grateful to

V

Rebecca Smith and Kathleen Smith for their generous help.
This study was supported in part by a grant from P.E.o.
International Peace Scholarship Fund. I am grateful for
this grant and for P.E.o. members' friendship.

Finally, I wish to thank my husband, Tung-jung, for his
support during my study and my two daughters, Shih-wen and

Hsiao-wen, for their intuitions.

vi

Chapter I

Chapter II
2.0
2.1
2.2
2.3
2.4

Chapter III
3.0
3.1

3.2

3.3
3.4
3.5
Chapter IV
4.0
4.1
4.2

4.3

4.4

Chapter V

TABLE OF CONTENTS

IntrOductioneeeeeeeeeeeeeeeeeeeeeeeeesee 1

Hess and Count Nouns

Introduction.................... ........ 15
The Mass/Count Distinction...... ....... . 16
Countness of Nouns............. ....... .. 23

Mass/Count Conversion................... 32
Implications for L2 Learners............ 43
Definiteness and Indefiniteness

Introduction....... ........ ........ ..... 52
Existence Claim......................... 53

The (Non)-Uniqueness Property of
(In)definite NPs...................... 68

Familiarity............................. 90
Indefinites............................. 98
Concluding Remarks... ...... . ..... . ...... 105
The Generic Use of Articles
Introduction............................ 108
The Generic Use of the Zero Article..... 110
The Generic Use of the Definite Article..119

The Generic Use of the Indefinite
m1CIQOOOOOOOOOOOOOOOIOCOOOOOOOOOOOOO 13S

canCJ-usioneeeeeeeeeeeeeeeeeeeeeeeeeeeeee 142

canalusionOOOOOOOOOOOO0.0.00000000000000 145

BibliwrathOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO0...... 156

vii

Chapter I

Introduction

The purpose of this study is to try to answer the
following questions frequently asked by adult second
language learners of English: (a) when do we use flu? (b)
when do we use a an? and (c) when do we use neither the
nor a/an? The choice of the definite article, the, the
indefinite article a/an or the zero article, 0, has long
been a great problem for non-native speakers of English.
This persistent problem can be seen in numerous studies on
errors committed by adult second language learners of
English with different language backgrounds.

I will mention only a few of these studies. Agnihotri,
Khanna, and Hukherjee (1984) reported that Hindi/Punjabi-
speaking undergraduates of the University of Delhi, after
seven years of studying, in a test of insertion of missing
articles, only 33% of the time correctly inserted the
definite article in simple cases such as "the door of a
bank", "the son of a rich man", and "the counter marked
'Accountant'". Another study by Kharma (1981) shows that
Arabic-speaking students, after 12 years of studying, with
the final 3-4 years spent in intensive study of English
language and literature at Kuwait University, scored only as
high as 71.2% on a test of articles which did not in fact

1

2
incorporate some of the most difficult uses. A similar
result was reported by Yamada and Hatsuura.(1982). Their
study shows that Japanese students majoring in English at
Hiroshima University, who were in their seventh to ninth
year of learning English, scored about 70% in an article
insertion test.

There are probably various reasons for the failure of
second language learners of English to make the right choice
of articles after numerous years spent learning the
language. It might be that some of them simply do not pay
attention to the usage of articles in the course of their
study. It might be the teachers' fault: they do not
emphasize the importance of articles. Or more
fundamentally, it might be due to a lack of good available
descriptions of the English article system.

To see how well the system is described, I now turn to
a review of some teaching grammars on English articles. In
an article entitled "Rules in the Teaching of the English
Articles”, Lacey (1977) introduces an "over-neat" (his own
word) system of English articles. This system involves
three notions: generality, definiteness and countability.
In selecting a correct article for a given noun, first of
all, Lacey says, one needs to ask the question: "Are we
speaking of all occurrences of what this noun refers to or
not?“ (p. 34). If the answer is yes, the noun carries
the feature [+all]: if no, it is [-all]. For a noun which
is [-all], we have a second question to ask: ”Do we know

this noun already?" (p. 34). If we do, the noun carries the

3

feature [+definite]: if we don’t, it is {-definite}. A
third question is: "Is this a countable noun?" (p. 34).
Here we get the feature [+count] or {-count}. For [+count]
nouns, we have a last question to ask: "Is the noun plural
or not?“ (p. 34). Here we get the feature [+plural] or
[-plural].

Lacey gives the following chart as a summary of his

system: (p. 34)

 

 

 

+definite +definite +definite

-count +plural -plural

the + N the + N+s the + N

-definite -definite -definite

-count +plural -p1ural

0 + N 0 + N+s a + N

+all +all +all

-count +plural -plural

0 + N CA 0 +N+s a + N
the + N

It seems that we have a simple and clear-cut system here.
There are, however, at least three problems. All of them
are related to the notions behind the features. First, we
don't know exactly what [+definite] or {-definite} means,
and hence, we don't know when to apply one or the other to a
given noun. Lacey's definition of this feature is as
follows. "Do we know this noun already? (By 'know' I mean
it has occurred in the text of speech before or it is known

by implication or deduction when it has not so occurred)"

4
(p. 34). Such a definition is not very useful because
it is really hard, in many cases, to determine whether a
given noun should be [+definite] or {-definite}. For
example, such a determination is difficult in cases like

(l)a. Despite the protest of his family and his

neighbors, John was not intimidated.

b. Despite protest from his family and his

neighbors, John was not intimidated.
Let us take (1) as a situation where the noun protest has
not occurred in the text of speech before. If in (1a) the
noun‘pNMau is known by implication or deduction and
hence it is [+definite], why not in (1b)? We can infer what
protest John met with in (1b) as in (la). Why does
prom! remain {-definite}?

Secondly, there are problems with the feature [+oount]
or {-count}, too. One problem is that "many English nouns
can happily occur as either " (p. 35). Take the phrase the
tar afspeech for example. Both text and speech are
labelled as [U] (uncountable) and [C] (countable) in
WAdvaItcedLearna’sDictionmyofCunentEngltsh (1948). To
make things worse, a most common count noun like apple can

be readily converted into a non-count noun as in
(2) He often puts apple in his salad.

Similarly, an abstract non-count noun like excess is

‘readily used with the indefinite article as in

(3) An excess of imports over exports will surely lead
to a trade deficit.

5
Besides the definiteness and countability problems,
lastly, there is a problem of generalizing with Lacey’s
system. In his article system, only when we speak of "all
occurrences” of what a noun refers to, does the feature
[+all] apply. I am not sure about what counts as all
occurrences. In cases like
(4)a. The dog is a friendly animal.
b. The tiger lives in the jungle.
apparently not each and every dog is friendly: neither is it
true that each and every tiger lives in the jungle. In
other words, strictly speaking, in (4), we are not speaking
of all occurrences of what the given nouns "dog" and "tiger"
refer to. But on the generic reading, the use of the
definite article in both cases seems to be a result of
applying the feature [+all] in Lacey’s system. Further, we
make generalizations such as:
(5)a. Candy is bad for the teeth.

b. Iron is good for the blood.
(Bolinger, 1975, p. 181)

In (5a) , we have a plural noun teeth co-occurring with

Me. But in Lacey’s chart we do not find the form

”Mw’+ N + s” for [+all] NP's. Similarly, in (5b), we have

a non-count noun blood co-occurring with the. But in

the chart, we find only "0 + N” for [+all] {-count} nouns.
A similar system is presented in another article

entitled "A Teaching Grammar of the English Article System"

by Patricia HCEldowney (1977). She says, "The presence or

6
absence of a, Mw’or "-s" in various combinations in
noun phrases communicates four main types of meaning: i)
general or particular, ii) any or special, iii) countable or
uncountable, and iv) singular or plural” (pp. 96-98). She

presents English article usage as follows: (p. 99)

 

 

 

 

 

 

 

Particular General
S ial y
Co t Uncount Co t Unc unt Count Uncount
s . P1. 89. Pl.
the the the a 0 (D a 0 t e t e D
+ + + + + + + + + + +
N Ns N N Ns N N Ns N Ns N

If we compare this chart with Lacey’s system, we find
that "Particular" is equivalent to [-all], "General" to
[+all], ”Special” to [+definite], and "Any" to {-definite}.
From this chart we see that in a general statement, the
Special/Any distinction disappears and we can use either the
singular or plural form of a count noun to make

generalizations. Hence for a count noun, there are four

7
ways to express a general statement, singular or plural, and
definite or indefinite. For a non—count noun, of course
the plural is impossible: however, the definite is not
impossible, as shown in (5b) above. But it is missing in
NbEldowney's system.

How do ESL (English as a Second Language) textbooks
present English article usage? In an article entitled "The
Article in American English: What the Textbooks Don't Tell
Us", Pica (1983) summarizes five rules that most ESL
textbooks have in common and interestingly enough, she gives
what she calls a "counterexample" to each of these five
rules (pp. 223-24). In the following, one by one I will
present each of Pica’s rules and its counterexample and then

make comments.

Rule 1: a for introductory usage of an item,
followed by Mm for second mention of the
item, e.g.:

(6) His car struck a tree: you can see the mark on the
tree.

Counterexample to Rule 1 (second mention):

(7) His car struck a tree. He was surprised to see
how much damage a car could do to a tree.

Strictly speaking, (7) is not a counterexample to Rule 1.
In (6), we have a second mention of the same tree: whereas
in (7), what we have is a second use of the same word

tum, which does not refer to the same tree. In (7), in
the complement.hmvnuwhahmugeacuraxdddomadame,

the two NPs near and atree become generic because of

8
the word amdd. Hence a real problem with Rule 1 is its

wording, but not its content.

Rule 2: the implies uniqueness, e.g.:
(8) The sun and the planets remain a mystery.
Counterexample to Rule 2 (uniqueness):

(9) A sun and some planets were sighted by a group
of astronauts during a recent space probe.

Rule 2 itself seems to be right. What might go wrong is
the application of the rule. In (8) we are talking about
the sun, the only sun that we know of, the sun that has been
sighted for billions of years: whereas in (9), we are
speaking of a new group of heavenly bodies, not the unique
sun and the planets in our shared knowledge and hence Mm

is inappropriate.

Rule 3: a for typicality or representativeness,
e.g.:

(10) A man and a boy are on a bus.
Counterexample to Rule 3 (typicality):

(11) A man named Higgenbottom and a boy with seven
fingers are on a bus.

Rule 3 as it stands is too narrow and is easily subject to
objections. In fact it may well be covered by the first
part of Rule 1, which says: a for introductory usage of
an item. Then both (10) and (11) can be covered by this

usage.

9

Rule 4: Mw’is used with nouns preceded by ordinals
and superlatives, e.g.:

(12) Which country of the world has Mw’most famous
art collection?

Counterexample to Rule 4 (superlative):

(13) Our university library has a most famous art
collection.

Here again, strictly speaking, (13) is not a counterexample
to the superlative rule because in (13), the word "most" is

not a superlative, but an ordinary adverb meaning "very".

Rule 5: the can be used with a first mention item if
the item is familiar or identifiable to both
speaker and listener, e.g.:

(14) Where did you park the car?

Counterexample to Rule 5 (familiarity/identifiability):

(15) A book on Mr. Allen's desk is yellow.

Even if (15) is a case where the book in question is
identifiable to both speaker and listener, it is not a
counterexample to Rule 5, which says: ”thefl cwnbe used...
Notice that the rule does not say: "the" should be or must

be used. The real problem with the
fwmiliarity/identifiability rule is that in many cases a
first mention item is not familiar or identifiable to either

the speaker or listener, but the is used. For example:

(16) John read a book about Chomsky and wrote to the
author.

10
(16) can be uttered in a context where the author of the
book is familiar or identifiable to neither speaker nor
listener.

Probably it is not fair to criticize ESL textbooks
which do not aim to give detailed discussion of English
article usage. Nor is it fair to criticize articles which
devote only a few pages to a rather complicated issue such
as the English article system, as I did Lacey's and
HcEldowney’s. But how about a reference grammar which
devotes an entire book to discussing the issue? I now turn
to a review of TheLbeqflAfikhsbuCbnkmnxmmylbgflfl:
written by Henryk Kaluza (1981). It is claimed in the
introduction that "there must be [a] few very simple rules
governing the whole usage [of the English articles]." The
aim of his book is to formulate these rules and show how
they work.

The overall system is presented as follows in the

conclusion (p. 83).

The articles a, fix; 0, and the nouns C, U
form combinations with the following specifying
and generalizing meanings:

specifying | generalizing
a C (specific, introductory) a C (concept)
the C (specific, referential) the C (object)
the Cs (specific, referential) the Cs (homogeneous class).
0 Cs ("more than one"
the U (specific, referential) D U (mass, abstractness)

 

[C stands for "countable noun" and U for "uncountable
noun'.]

11
This chart takes some explanation. For specifying aCL
Kaluza says, "When we have in mind a specific entity of a
class paraphrasable by 'a certain' or 'a particular' not yet
expressed or implied, we commonly introduce it by means of
a“ (p. 23). This leaves out a whole range of non-
specific uses of a, as exemplified in

(17)a. Mary wants to marry a Norwegian.
b. John wants to buy a house.

(17a) has a reading according to which Mary wants to marry
a person from Norway, but she does not have a particular
Norwegian in mind. Similarly, (17b) has a non-specific
reading according to which John wants to buy a building to
live in, but he does not have a certain house in mind.

For the generalizing uses of Hwy a and 0, Kaluza
bases all his discussions on Burton-Roberts’ (1976) work,
which I will discuss in due course. I now turn to the
itemized rules Kaluza presents. One of the rules that he
gives is this: “Practically every U can be converted into a
C with one of the following meanings: a) a unit of, b) a
kind of, and c) an instance of" (p. 10). For example, we
say “two coffees" meaning "two cups of coffee", ”two
different wines” meaning ”two different kinds of wine" and
”many small kindnesses” meaning ”many small instances of
kindness“. He says that ”a C sense is often indicated by
means of a restrictive modifier" (p. 10). Among others he

gives the following sentence as an example:

(18) Jerseys give a very rich milk. (p. 18)

12
This sentence might be acceptable, but how about a sentence
such as

(19) ?We've not had a stormy weather like this for a
long time.

I think in (19), the use of a will probably be considered
bad by most people. However, according to Kaluza's rule,
it should be acceptable because conceptually, we can argue
that "stormy", "sunny", ”cloudy", "rainy", etc. constitute
different kinds of weather. Why, then, in (19), does the
modifier stormy fail to bring out a count sense ("a kind
of“) of the uncountable weather?

Besides saying that ”restrictive modifiers change
0 U into a C (with the article a) when they bring out
the sense of 'a unit of', 'a kind of’, or 'an instance of'
of uncountable mass or abstractness", Kaluza goes on to say
that “the restrictives also cause the use of flw'[with a
first-mention item] if they answer the question 'which
one(s)?' Otherwise, restrictive modifiers narrow down the
meanings of their heads without affecting article usage" (p.
84). For example, we say "a man who writes novels", but we
say "dw'man who wrote this novel". In the former case, the
question "which one?" is not answered by the restrictive
modifier because presumably there is more than one person in
the world who writes novels. In the latter case, however,
the question is answered by the restrictive relative because
there is normally one and only one person who wrote a

particular novel.

13
This rule works with simple cases like the examples
given above: however, it does not work well with more

complicated cases. Consider:

(20) Mary has a dog whose name is Coco.

Is the question ”which one?" answered by the restrictive
relative in (20)? Apparently, yes, because we can answer,
"the one whose name is Coco". Nevertheless, the use of
the in this context is not appropriate. Besides, for
abstract nouns, in cases like

(21)e. The early arrival of Santa Claus that cheered

up everybody was itself a nice present for
b. 3::nimportance of style that stylists

emphasize a great deal is overlooked by a lot
of people.

Mn is obligatory in both (21a) and (21b), but the
restrictive modifiers do not answer the question "which
one?'. In fact, the question itself does not arise for
abstract nouns because we do not identify the denotation of
an abstract noun by picking out an individuated entity.

To give a brief summary of what we have discussed so
far, we note that we have three kinds of problems to solve.
First, what is involved in converting non-count nouns into
count nouns and vice versa? Second, what makes an NP
definite and what makes it indefinite? Third, what are the
generic uses of English articles? To make a correct choice

of English articles, a second language learner needs to

14
consider the following properties of the NP in question: i)
countability, ii) definiteness, and iii) genericness. This
dissertation will discuss these three topics and in that
order.

The choice of the, a an or 0 is a tough one for
non-native speakers of English. In sharp contrast, it is
not a problem at all for native speakers. If the system
underlying native speakers' usage can be described
adequately, non-native speakers will have good guidelines in
making the choice. The problems to be addressed are among
the factors affecting the choice. This dissertation will
conclude with a set of principles for the choice of
articles and a short discussion of the pitfalls of rule

application by second language learners.

Chapter II

Mass and Count Nouns

2.0 INTRODUCTION
This chapter is concerned with the countness of
nouns. For L2 (second language) learners, in making the
choice of a, Mw’or the zero article, a question to ask is,
”Is this a count noun?” For a count noun, the choice of
articles is a/Mw’(+ N) or oydw'(+ Ns); for a non-count
noun, the choice is D/Mm'(+ N). This correlation of
count and non-count nouns with articles is illustrated in
the following examples:
(l)a. I have a/*¢ book on fish.
b. I have *a/D books on fish.
c. The book/books I bought today was/were on fish.
(2)a. I'm looking for *a gold.
b. I'm looking for gold/*golds.
c. The gold/*golds I found is/*are worth millions
of dollars.
(1) shows that a count noun like book, if singular,
must co-occur with a as in (1a), or with Mw'as in (lo):
if plural, it cannot co-occur with a, as in (1b). A plural
count noun must co-occur with either Mm, as in (lo), or the
zero article, as in (lb). (2) shows that a non-count noun
like gold cannot co-occur with the plural morpheme "-5" nor
with a, but it can co-occur either with the zero article or
Me.
15

16

Because of these systematic co-occurrence restrictions,
L2 learners can narrow down their choices of articles if they
can decide whether the noun in question is count or non-
count. To better understand what is involved in determining
whether a noun is count or non-count, we need to know the
semantic distinctions between these two kinds of nouns, and
understand the switching back and forth between count and
non-count meanings of the same noun.

It is concluded in this chapter that if L2 learners
know the distinctions between count and non-count nouns,
they do not have to learn separately for each noun whether
it is count or non-count. Instead, what they have to learn
is only two types of nouns. The first type is always non-
count. The second type is always count. The rest of the
nouns can be used either as count or non-count, depending on
context. Here the semantic count/non-count distinction can
be of help to the L2 learner in deciding when a noun is to
be used in a count sense and when to be used in a non-
count sense. Principles for the conversion of count nouns
into non-count nouns and also the opposite are offered as.
guideposts for L2 learners.

Section 2.1 addresses the count/non-count distinction.
Section 2.2 deals with the degrees of countability. Section
2.3 is about count/non-count conversion. Finally, section
2.4 discusses some implications for L2 learners.

2.1 The MASS/COUNT DISTINCTION
Jespersen (1924, p. 198) called non-count nouns "mass

words" and he gave them this definition:

17

There are many words which do not call up the idea
of some definite thing with a certain shape or
precise limits. I call these "mass words": they
may be either material, in which case they denote
some substance in itself independent of form, such
as.dben qukknwen unwr,4hmflv,)gu, db,

etc., or else immaterial, such as leisure, music,

aafliu smxemu..

Jespersen's definition of mass words seems to be vague
because we can never know for sure what idea is called up in
the mind of a speaker of a language. Further, there are
things which come in similar shapes, but the nouns which
denote them are of different status. One may be a count
noun and another a mass noun. As Barbara Abbott has
pointed out to me, asparagus and carrots are an example of
this kind. They come in similar shapes, but the noun
asparagus is a mass word and the noun carrot is a count
noun. Given the fact that they are similar in shape, we are
not sure whether or not the two words will call up different
ideas about their shapes in the mind of language speakers
just because one is count and the other is mass. That is,
will asparagus, being a mass noun, fail to call up the idea
of something with a certain shape or precise limits? In a
broader sense, we are not sure whether or not language
affects the way we see the world. This is another
difficulty that Jespersen's definition of mass words
encounters.

Jespersen was not the only one who assumed that the
mass/count distinction is simply in the nature of the

things referred to. Whorf was another one who held

18
basically the same position. Whorf (1956) divides nouns
denoting physical things into two categories: individual
nouns and mass nouns. His remarks on the distinction
between the two are: "Individual nouns denote bodies with
definite outlines...Mass nouns denote homogeneous continua
without implied boundaries" (p. 140). To explicate
his point, he gave the following example. When we want to
talk about only a certain portion of the homogeneous
continuum of a mass noun, say milk, water, or sugar, we
have to individuate the mass noun by an individual noun like
bottle, cup and lamp, as in abottleofmilk, acupof
wau7,4ahunqusqmv, etc.

Whorf's remarks on the mass/count distinction, like
Jespersen's, are subject to attack. If the distinction is
simply in the things referred to, a noun which refers to
an identical entity through different times should be always
mass or always count. The development of the English
language, however, shows that there are nouns which were
mass earlier in history but later became count. As Barbara
Abbott pointed out to me, pea and cherry are two historical
examples. The former comes from pease and the letter from
chaise. Both pease and chen'se were 'mass, but because they
sounded like a plural, people took pea as the singular of
pease and chewy the singular of cherise. Thus pea and
tinny became count nouns. For nouns like these two, being
count is nothing but a historical accident.

Are there better theories for the mass/count

distinction? Quine (1960, p. 91) claimed that.fluwy

19
pairofshoes [count] and footwear [mass] refer to basically
the same stuff, and are different from one another solely in
how they divide their reference. Both shoe and pairof
shoe: divide their reference, differently. But footwear
does not at all. Quine claimed thatthe mass/count
distinction lies in the words themselves and not in the
stuff they name. Count nouns individuate their reference,
but mass nouns do not.

MCCawley (1975) illuminated this distinction further
by arguing that "the meaning of a count noun specifies an
individuation, whereas the meaning of a mass noun is neutral
as to individuation” (p. 314) . One of his examples is cold
and.flh. The following facts are given to support his
contention: (p. 317)

(3)a. I have a cold.

b. I have a case of the flu.

(4)a. Do you have the same cold/*flu that you had last

week?

b. Do you have the same case of the flu that you had
last week?

He says that ”a cold is a ’case' of a particular infection"
(p. 317) , and argues that the countness of cold can be
attributed to an individuation specified in its meaning. In
contrast, the word.fla does not individuate its referent,

and hence when we are talking about a particular case, we
cannot say ”I have a flu": instead we have to say (3b),
using an additional unit word like ave.

In sum, according to Quine and McCawley, count nouns

20

differ from mass nouns in that the former, but not the
latter, include an individuation in their meaning. This
theory does not have the shortcomings of Jespersen's or
Whorf’s. If the mass/count distinction lies in word
meaning, and not simply in the nature of the things referred
to, the historical development of pea and cherry can be
explained as a change in word meaning. We can say that an
individuation for their reference was added to their word
meaning at a certain point in history. And for the question
why carrots is count and asparagus is mass, we can say that
it is because the former includes an individuation in its
meaning, but not the latter. It has to be admitted, however,
that the correspondence between form and meaning is
arbitrary. That is, we still cannot explain why the word
carrot, and not the word asparagus, includes an
individuation in its meaning.

This mass/count distinction explains why two words like
surgety and operation, although similar in meaning, can
be different in countness. To most native speakers of
English, the word.aagay denotes the treatment of
injuries and diseases by operations and hence in this usage,
it is a mass noun: whereas operation denotes the act
performed by a surgeon and hence it is a count noun. Note
that acts are relatively easier to individuate than
treatment. Many Chinese learners of English do not know
that.sugay is a mass noun because when they are learning
this word, they fail to learn its exact meaning, which does

not include an individuation for its reference.

21

For abstract entities such as surgery and operations,
L2 learners cannot decide whether the nouns which denote
them are count or mass simply by logical reasoning.
However, for nouns which refer to physical objects, there is
a relatively reliable tendency. If a noun refers to a
discrete object, i.e. an object with definite outlines or a
certain shape, it tends to be a count noun. Things like
cars, houses, chairs and tables are discrete and countable
and hence the nouns refer to them tend to be count. On the
other hand, if a noun refers to an object without a natural
boundary such as water, air, milk, cotton, and sand, it
tends to be mass. Here, we see that Jespersen and Whorf are
not totally wrong.

Yet in English, we do have nouns which do not follow
this tendency. Asparagus is such a word. McCawley (1975)

gave us some more examples of this kind:

Count Mass
noodles spaghetti
3323;; """""" Q33?"
132;; """"""" £12; """
213;};""""""E;I-;IE;;;'

McCawley pointed out that rice comes in grains, which are
countable: nevertheless the word rice is a non-count

noun. Similarly, there is no physical reason why noodle is
count and not.qmghafi.

Nevertheless, an explanation was given by Markman

22

(1985) as to why words such as fiu'niture which apparently
refer to discrete objects are non-count. Markman explained
why many superordinate category terms, i.e. category terms
of relatively high levels (e.g. furniture, jewelry, money),
are mass nouns although conceptually they refer to diverse,
discrete, countable objects. Let me take money as an
example. Although we say, as many fairy tales go, ”The king
is counting his money in the palace", the word money is
itself a non-count noun. If we can count money, money must
refer to countable objects. Why is it a mass noun?

Markman's explanation was that mass nouns have the
property of being a compromise between "collections" and
”classes" and this property helps children to learn
superordinate category terms. ”Classes" have an inclusion
structure: for example, all roses are flowers, but not all
flowers are roses. The inclusion structure expresses the
"is a” relation. A rose is a flower. A doll is a toy. In
contrast, I'collections" have a part-whole structure: for
example, a tree is a part of a forest, but itself is not a
forest.

Markman said that studies showed that children find
it simpler to learn the part-whole structure than the
inclusion relation. But collective nouns (e.g. family,
anny), said‘Markman, cannot be superordinate category terms
because they do not express the inclusion relation. A
soldier is a part of an army, but himself is not an army.
However, a chair itself is furniture. A coin itself is

money. Superordinate category terms like furniture and

23
tummy’need to express the ”is a” relation. Markman argued
that mass nouns can be viewed as a compromise between part-
whole and inclusion relations. A piece of clay is part of
the whole mass of clay and each piece of clay is itself
clay.

Markman’s studies showed that children, at the age of 4,
were better able to learn a new category such as "vehicle”
if they heard "A car is a piece of vehicle" rather than "A
car is a vehicle" (p. 31). Markman’s conclusion was
that "languages tend to use mess nouns to refer to
superordinate categories because it helps children to learn
them” (p. 51) .

If Markman is right, L2 learners can expect English to
evolve in the direction of regularizing all English category
terms of relatively high levels to be mass nouns. Thus the
learner would have a good rule to follow.

To sum up this section, the mass/count distinction lies
in the meanings of mass and count words themselves: the
latter specifies an individuation, but not the former. As a
consequence, count nouns normally denote entities with a
certain shape or precise limits: whereas the reference of a
mass noun is normally a homogeneous continuum and not
individuated. Being mass or count is part of the meaning of

a word.

2.2 COUNTNESS OF NOUNS
Although traditionally nouns are classified into two

types, count and mass, there are complications in that the

24

distinction between the two is not a simple binary one. In
the very beginning of this chapter, it was noted that count
nouns admit the singular article a0», but mass nouns do
not. Yet it is found that there are mass words which
nevertheless allow 000 and there are count words which do
not allow 000-

Allan (1980) challenges the traditional binary-feature
notion of countness, which assigns either [+count] or
{-count} to a given noun. He claims that instead of two,

there are eight levels of countability: (p. 563)

REPRESENTATIVE PERCENTAGE LEVEL
car 100 7

S;i""""""""""§Z""'""""'2"
2§ZZI;""""""'""§3"""""""§"
EIQQI2§§;""'"""'"'ZZ'"""'"""Z"
321;;SLQ """""""" 23"""""""§"
;;SEI;§""""""""§E""""'""'Z"
QEQIEQZIS;"""""'“’IZ""""-""'I"
3351;;232"""""""'3"""""""3"

This chart indicates that a word like ar'is 100%

countable, a word like oak is 82% countable, and a word
like equipment is 0% countable. If a word is 0%

countable, it is on the lowest countability level, 0. In
contrast, if a word is 100% countable, it is on the highest
countability level, 7. .

Allan computes the countability of words by trying them

25
against four tests. In the following, I will take as
examples five words from the above chart, car, oak,
cattle, mankind and equipment to illustrate what
Allan’s countability tests are and how these five nouns

behave in different environments.

(i) A + N Test: to see if the form "a/an + N" is
grammatical or not, e.g.:

(5)a. Acar is a convenient vehicle for

transportation.

b. Anoak is a tree.

c. *I saw acattle in the field.

d. I'd like to see amankind full of charity and
sweetness.

e. *An equipment in our lab was destroyed by the
fire.

(ii) F(uzzy) + Ns Test: to see if a noun can be
~ preceded by a fuzzy denumerator
such as several, many,

about fifty, e.g.:

(6)a. Sawndcan'were crushed in the accident.
b. Many oaks were chopped down by the boy.
c. I saw aboutfiftycattle in the field.
d. *I have met with several mankind: and they are
all different.
e. *Severul equipments in our lab were destroyed by
the fire.

From (5) and (6) we see that car and oak pass both (i)
and (ii) tests. On the other hand, equipment fails in
both tests. In (So) we see that anfle fails in the A + N
Test, but in (Ge) we see that it passes the F + Ns Test.
Conversely, mankind passes the-A + N Test, as shown in
(5d), but fails the F + Na Test, as shown in (6d).

(iii) EX(ternal)-PL(ural) Test: to see if an NP governs

plural NP-external number
registration, e.g.:

26
(7)a. Thosecars are wonderful and I like them all.
b. Oaks are deciduous, aren't they?
c. Those cattle are dying for lack of water, aren't
7

d. Manla'nd are expected to give an account of
themselves before God, aren't they?

e. *Equipment(s) are essential, aren't they?
(7) shows that all the five words except equipment pass
the Ex-PL Test.

(iv) All + N Test: to see if the form
”a” + N + V ” is grammatical
or not, e. 3139““
(8)e. *All car is convenient vehicle for
transportation.
b. Alloak is flammable.
c. tAflcka b dying for lack of water.
d. All mankind is rational.
e. AH equipment in our lab was destroyed by the
fire.
( 8) shows that oak, mankind and equipment pass the
All + N Test: whereas car and cattle fail in this test.
Among the four tests, the All + N environment
(Test (iv) above) is an uncountable one whereas the other
three are countable. To compute the countness of nouns,
Allan gives a plus to a noun if it passes a count test
(Tests (i-iii) and he also gives a plus when a noun fails
the non-count test (Test (iv) ) . Equipment fails in all
the three count tests, and passes the non-count one.
Hence it receives no plus and is 0% countable. For nouns of
this category, there is no problem for us to assign the
feature {-count] to them. However, words on Level 6, like

oak, pass all of the count tests and also the

non-count one. In terms of feature assignment, which

27
feature, [+count] or {-count], shall we give to them?

The four countability tests show that words on level 2,
like manla'nd, pass the All + N Test and thus are non-count,
but they admit the indefinite article, which non-count nouns
normally do not. And they pass the Ex-PL Test (i.e. take a
plural verb or plural pronoun), which again non-count nouns
normally will fail. Are they count or mass? The reader
might suggest that we can treat words like mankind as either
mass or count. Yet this treatment cannot rule out bad forms
like severalmankinds.

On the other hand, although words on level 5, like
aafle, fail the uncountable All + N Test and thus they are
not non-count nouns, they do not admit the indefinite
article, which count nouns normally do. We cannot simply
assign [+count] to them. We have to say, in addition, that
they never take the indefinite article.

Being the opposite of words on level 5 (e.g. aafle),
words on level 1, like admiration, heat, sincerity,
darkness (derived nominals) , and physics (names of subjects
for study) pass the uncountable All + N Test and hence are
non-count. Like typical mass nouns, nouns of this category
do not admit fuzzy denumerators: nor do they govern plural
NP-external number registration. However, unlike typical
mass nouns, they admit the indefinite article. Some
examples are as follows:

(9)a. A phnmm in which energy is lost rather than
transferred is quite inconceivable: where would

the energy go to?
b. *There are several physics: geophysics,

28
astrophysics, nuclear physics- and I don't know
what else.
(Allan, 1980, p. 550)
c. *Thosephysics are all difficult to study,
aren’t day?

For (9b), if we say ”several types/kinds of physics"
instead of ”several physics", the sentence will become well-
formed. Similarly, in (9a), aphysics means 'a kind of
physics', but the indefinite article itself without hhd
cj'does the job. Here we see a difference between the
indefinite article and fuzzy denumerators.

Derived nominals like heat and darlmess behave
similarly.

(10) Wheat is so much more bearable than adamp

(11)e. *We got up in adaflmess.
b. We got up in a pitchy darkness.
c. Anoppressivedarlmess hung all around us.
(Allan, 1980, p. 559)

In (11a) we see that darkness does not behave like an
ordinary count noun since it cannot co-occur with the
indefinite article. However, in (11b) and (11c) we see that
with a modifier, pitchy and oppressive respectively, the
indefinite article becomes acceptable. Similarly, in (10),
the two occurrences of the non-count noun had, with the
modifiers dry and damp respectively, become countable.
Allan describes this kind of usage as "referring to
instances or occasions of particular note" (p. 559).

The preceding usage of a was treated as one of the

iblportant functions of English articles in Frank's (1972)

a:

re

29

exercises for non-native speakers. She comments on this

as follows: (p. 160)

In some sentences, noncountable abstract nouns

with adjective modifiers may be used with a.
In many such sentences a is the equivalent of

akbwhfll
If we use akindof instead of a in (10) and (11b-c) , we
get:

A dry kind of heat is so much more bearable than a

(12)
damp kind of heat.

(l3)a. We got up in a pitchy kind of darkness.
b. An oppressive kind of darkness hung all around

118.

Does this mean that akindof is a reliable test for using

a with abstract mass nouns modified by adjectives or

relative clauses? Consider:

He provided us with a kind of information that

(14)a.
only insiders can.
b. This is a kind of evidence that could be used to

persuade people to believe in God.

If we delete kindof, both (14a) and (14b) become

ungremmatical, as shown in (15):

(15)a. *He provided us with an information that only

insiders can.
b. *This is an evidence that could be used to

persuade people to believe in God.

Frank's rule does not tell us when a can replace akind
0f: Allan's eight levels of countability tell us that this

‘OCcurs when the noun in question is on level 1 (e.g.

timbre“, heat). Words on level 0, like information and

30
endear, can never be used with a.
Notice that we can drop the indefinite article in both

(10) and (11) and we get:

(10') Dryheat is so much more bearable than damp heat.
(ll')a. We got up in darkness.
b. We got up in pitchydarkness.
c. Oppressivedarkness hung all around us.
In (10') and (11') we see that after the dropping of the
indefinite article, all the grammatical sentences remain
well-formed and the ungrammatical one, (11a), becomes
acceptable. Nevertheless, if we have restrictive relative
clauses modify the underlined NP's, the indefinite article
has to be put back again. Take (11’) for example:
(16)a. We got up in a/*¢ darkness that was really
scary.
b. We got up in a/*¢ pitchy darkness that was
really scary.
c. An/*¢ oppressive darkness that was really
scary hung all around us.
Restrictive relative clauses seem to have a stronger effect
on individuating the whole mess of darkness into different
types than prenominal adjectives. Perhaps this is why the
indefinite article is obligatory in (16), but optional in
(11b) and (11c).
Besides derived nominals like darkness, and heat,
and names of subjects for study like physics, Allan gives

EMflhh, as a name for a language, as an example of words

on level 1, i.e., abstract mass nouns that admit a:

(17) He speaks an EMflBh that I can barely

31
understand at all: and I was born in London.
(p. 558)
Some native speakers find (17), with the proper name
Ehflhh turned into a count noun, not acceptable. There
is definitely no problem if, instead of (17), we say

(18) He speaks akindofEnglish that I can barely

understand at all: and I was born in London.
This indicates that a language change, moving.EMflBh from
level 0, where mass nouns can never be used countably, to
level 1, where mass nouns with modifiers can co-occur with
a, is not yet complete.

To sum up this section, Allan's discussion shows that
the grammatical correlates of the mass/count distinction are
complicated by the fact that this distinction is not a
simple binary one. There are words like cattle which are
[+count] except that they do not admit the singular
indefinite article 000- There are words like ham which
are {-count} except that they admit 000 under certain
circumstances. Further, there are words like oak which are
either [+count] or {-count]. Allan's four tests of
countability, however, are of little help to L2 learners
because the grammaticality judgements are exactly what is in
question. Further it would be a great burden for L2
learners to learn which word falls on which level, so the
existence of the levels is of little help. In the next
section, I will discuss an alternative to the solution

to the complications of the mass/count distinction offered

32

by Allan.
2.3 MASS/COUNT CONVERSION

The mass/count distinction, as has long been observed
in the literature, is better taken as a distinction among
word-senses, or ways of using words, rather than a
distinction among words themselves. A good illustration
of this point is found in nouns which denote either the
animal or its flesh as food. For example:

(l9)a. I don't eat chicken because I like chickens.

b. Lamb is delicious and lambs are lovely, too.

The singularity and plurality of chicken and lamb in
(19) are determined by the different senses of the same
words: and in turn, the grammatical correlates reveal the
different senses of the words. When referring to the
animal, chicken or lamb denotes a discrete countable
object, and hence it is used as a count noun and thus has
to take the form "a/the + N” or "O/the + Ns". In
contrast, when referring to the flesh of the animal as food,
lamb or chicken becomes non-count and thus has to take the
form nayth+ N”.

Besides the above animal/meat example, we have a
lot more instances of shifts in sense leading to shifts in

countness. Jespersen (1924, p. 199) gives us the following

examples:
a parcel in brown paper state papers
little talent few talents

it is hard as ban a hot ban (flat iron)

33
Were (1979) in "Some bits and pieces" gives more examples of
this kind:

(20)a. His politics are atrocious.
b. Politics is not his bag.

(Po 15)

(21)a. His faculties are intact.
b. how much faculty he has for the project

(Po 20)
(22) Many glasses do not have any glass in them.

(p. 17)

From these examples we see that many English nouns have
both mass and count meanings. According to Allan's
computation, these words, like oak (wood/tree), will fall on
'Level 6, with 82% countability. They pass all the three
countable tests and also pass the uncountable test. In
other words, they are either [+count] or [-count]. It is
suggested here that words like these be treated as two words
instead of one. Take the mass/count--meat/animal words as
an example. Chuien is a name for both the animal and the
meat. Yet as in the case of pork/pig or mutton/sheep, it
could have been the case that for chicken too the animal and
the meat were named by two different words. The word that
named the animal, a discrete object, would be a count noun
and the word that named the meat would be a mass noun.

Another example is found in the language/people words
such as Chinese, Italian, and Greek. When referring to the
jpeople, the noun is count and when referring to the
language, it is mass. Words such as authority

(quality/person) constitute still another example of this

34
kind. The word ammofiy'has two distinct senses. When we
are talking about special knowledge as in write with I
authority it is non-count: whereas when we are talking about
a person with such knowledge as in an authority on nuclear
physics, it is count.

In sum, ambiguities between a mass and a count
meaning are likely to be found in wood/tree, meat/animal,
language/people, and quality/person words. The same noun
can be treated as two words. One that denotes the discrete
object is count and the other that denotes the constituent
substance or a peculiar quality of the discrete object is
mass.

words such as those we discussed above are clearly
ambiguous between two distinct meanings. Other words may
not involve an ambiguity, but they have both the count and
the mass use. Candy, hair, stone, and wine are some
examples. When we are talking about the substance or
material, they are mass: whereas when we are referring to
shaped pieces (instances) or kinds of the substance, they
are count. For example:

(23)a. Do you want a candy?

b. Candy is bad for your teeth.

(24)a. I found a hair in my soup.
b. The cat has a fine coat of hair.

(25)a. The box is filled with heavy stones.
b. The wall was made of stone.

(26)a. This is a French wine.
b. I do not drink wine.

We have identified two groups of words which can be

8—

to

in

be

C01;

35
readily used either in a count or a mass sense. Let us

state this fact in the form of a conversion principle:

Principle 1:

HISB'NI Count: If a noun is used to denote a
discrete entity, it is count: if it is
used to denote the material content or
a particular quality of the discrete
entity, it is non-count.

On the other hand, there are many words like an and
book which we believe behave only as count nouns. Notice
that these words are 100% countable according to Allan's
computation. But some linguists point out that given the
right context, they can be used as non-count. For example,
Gleason (1965) asks us to imagine an animal story "featuring
a mother termite concerned over her child: 'Johnny is very
choosey about his food. He will eat book, but he won’t
touch shelf'" (pp. 136-37). In this context, the mother
termite is talking about the material constituents of books
and shelves, but not the discrete objects themselves. Thus
book and shelf become mass nouns. Conversely, a word like
icecream which we think behaves typically as a mass noun
can be used countably. Gleason says, "...a customer, unable
to choose between two brands, might say: 'I don't care: one
ice cream is as good as another'" (p. 136). Here urcmaan
is used as a count noun. He concludes that "every noun,
given the right context, can occur in either type of usage,
count or mass" (p. 137).

Is it true that every mass noun can be used as a

count noun and every count noun can be used as a mass

36
noun? Different linguists (or philosophers) hold different
positions. Pelletier (1979, p. 5) says:
I think that reflection on the example of
above, [How many oetmeals are in your kitchen?]
provides convincing evidence that every word

which would normally be called a mass noun can be
given a perfectly clear count sense.

Pelletier, like Gleason, holds that every mass noun can be
converted into a count noun and also the opposite. He
describes a thought experiment to persuade people that all
count nouns can be given a mass sense. He asks us to
imagine a machine, the Universal Grinder. The machine can
chop and grind anything, say dogs, cats, cars, or men. Put
whatever object you wish to, say a porcupine, into one end
of the grinder. After the grinder chops the porcupine and
grinds it up into a homogeneous mass and spews it onto the
floor from the other end of the grinder, ask what is on the
floor. The answer: "There is porcupine all over the
floor”. In real life, we do see porcupines, raccoons, and
squirrels smashed by cars.

Pelletier is aware that his machine can only grind
physical objects, and that there remains a problem for
those count nouns which denote non-physical things.
Nevertheless, he argues that the thing to be put into the
grinder does not have to be grindable. If a normal sentence
can use a count noun in a mass sense, his theory holds. His
example is the word number. He uses this count noun in a

mass sense in the following sentence: (p. 6)

37
(27) If numbers were physical objects, and if we
were to put one into the grinder, there would be
number all over the floor.
I will leave this issue to philosophers. I think that it is
reasonable to say that practically every count noun which
denotes an object with material content can be used in a
mass sense, given the right context, that is, when we use
the word to refer to the material constituent or the mass of
the object, and not to the discrete object itself. I agree
with Ware's (1979, p. 19) position:
I do not think that all homophones with count
occurrences have mass occurrences and vice versa.
Words for orifices seem to have count but not
mess occurrences, e.g. opening, hole, mouth.
Obviously there is no mass of openings or holes to be talked
about. Furthermore, some nouns that denote abstract
entities like idea, trick, and characteristic, do not
seem to have mass occurrences.
To L2 learners, the conversion of count nouns into mass
nouns causes fewer troubles, because after all cases like
(28)e. The scrapyard is full of smashed car awaiting
recycling.
b. Emmy finds squashed spider more nauseous than
the thing alive.
~ (Allan, 1980, p. 547)
are unusual and if the count nouns remain count, as shown in
(29):
(29)a. The scrapyard is full of smashed cars awaiting
recycling.

b. Emmy finds a squashed spider more nauseous than
the thing alive.

III

at

cer

38
the sentences still are well-formed. In other cases like
Therewascataflovertheroad describing, e.g. a poor
cat smashed by a car, failure to convert the count noun into
a mass noun, as in Therewasacatallovertheroad,
might lead to a semantic anomaly.. However, if we state the

conversion principle as follows:

Principle 2:

Count --> Mass: When we are talking about the
undifferentiated mass of a physical
object, but not the discrete object
itself, the count noun denoting that
physical object should be converted
into a mass noun.

since the principle is well-defined, it will not cause too
much confusion for L2 learners.

The conversion of mass nouns into count nouns, on the
other hand, is much more troublesome. First of all, is it
true that every mass noun can be used as a count noun?
Recall the Universal Grinder that turns count nouns into
mass terms. Bach (1986, p. 10) suggests an opposite switch,
a machine called the Universal Packager that is capable of
packing all substances into precise units and hence
converting mass nouns into count nouns. In the mass/count
conversion, the Universal Packager is supposed to have the
same function as the Universal Grinder, though working in
the opposite direction. However, it turns out to be not so
plausible.

Since every concrete count noun denotes something with

certain material content, when we want to talk about the

39
mass instead of the form denoted by a count noun, the right
context arises for the mass use of the count noun. On the
other hand, in a hypothetical world, we can have each and
every substance denoted, e.g. by water, milk and gold,
whatever you wish, packed by the Universal Packager. When .
the new product, say milk, comes out of the packager, what
will be the answer to the question, "What is on the floor?"
Will we answer "There is a milk on the floor" (or "There are
milks on the floor")? Probably not: instead, the answer
most likely will be "There is an X of milk on the floor" (or
”There are Xs of milk on the floor"). A unit word x, e.g.
bottle, glass, is still needed.

Why would the Universal Packager fail to convert mass
nouns into count nouns? If we do not take milk as
denoting a bounded and discrete object, we will not use it
as a count noun. ‘The Universal Packager can pack everything
in natural units, but our intuition about word meaning does
not change accordingly.

Although the Universal Packager fails to convert mass
nouns into count nouns, the real world packaging does create
some conversions. Jespersen (1924) claims that ”in English,
hmad is only a mass-word" (p. 200). In the twenties,
probably it is true that.bmad was used only as a mass-
word, but it is no longer true now. We find a lot of
occurrences of bread as a count noun. To cite only a few
of them:

(30)a. coarse whole grains blended in a light brown
bnmd (American Meal Bread ad)

40

b. For many years, so-called "diet” bmmds
attempted to create the illusion... (Story of
Less, Schafer's Less ad)

c. Try the other delicious Country Hearth.Bmmds.
Look for the Country Hearth family of quality
variety breads. . . (Country Hearth ad)

d. The Mackinaw Milling Co- family of breads are
all made with...to create a line of breads which
can please...Whichever variety you choose, feel
confident you're serving a bread which. . .
(Mackinaw Milling Co. ad)

From the occurrences of bread in singular and plural
forms in (30), it might be inferred that bread manufactures
perceive the different kinds of bread they produce as well-
defined, individuated objects, and the word.bnma is used
countably to denote one "kind” of the mass.

Bread makers use abread to denote a kind of bread
and breads for different kinds of bread. L2 learners,
however, cannot jump to the conclusion that all instances of

'x kind(s) of bread" can be reduced to "X bread(s)", as in
(31) I went to Shop-Rite today. ?I bought two breads.

To some people, (31) is just ungrammatical. To some, at
best twobreads can be taken as "two loaves of bread",
but not "two kinds of bread".

Bread denotes something edible and it can be packed
into discrete units. _Words denoting abstract entities such
as.mugay also can gain an individuation in their meaning
and eventually gain a count sense. As I mentioned in
section 2.1, surgery as opposed to operation is a non-count
noun. However, in a survey that I did in which 50 subjects

were asked to choose the one they preferred in the following

41
pair of sentences:

(32)a. I had a CHOLELITHOTOMY, which is surgery, and

it was covered under surgical expense benefits.
b. I had a CHOLELITHOTOMY, which is a surgery, and
it was covered under surgical expense benefits.
25 subjects chose (a) and 25 chose (b). In random
interviews following the survey, one subject said that the
reason he chose (a) was that muggy is basically a mass
noun. Another subject said that he chose (b) because
CHOLELITHOTOMY is not a term for surgery in general.
Still another one said that although he chose (a), (b)
was possible. A last one said that he chose (a), but he
preferred a form ofsurgery than surgery alone. From this
equal split of 25 to 25 and the comments made by the
subjects, I conclude that nugay is undergoing a semantic
change, moving from being.mass to count.

In fact, a health insurance company worker, while she
was explaining that different types of surgery are covered
under different policies, did say this to me: ”If you have
another surgery. . .". Non-medical people probably would say,

“If you have surgery again..." in this case. I think to
health insurance workers, the word augay has gained a
discrete reference through constant application of the word
to well-defined categories of surgery, exactly like the word
band to bread manufacturers.

In addition .to bread and surgery, the real world
"packaging" has brought about other count uses of mass nouns.

In restaurant orders, we have occurrences of "a large coke",

42

"a small coffee", or simply "two coffees", "three cokes",
etc. This is because in such places as fast food stands or
stores, these drinks typically come in cups. Instead of
saying "two cups of coffee" or "three cups of coke", the
elliptical forms, "two coffees" or ”three cokes" are used..
However, when we are at an American friend's house, the
host/hostess will not ask, "Would you like a afifim?"
(meaning "would you like a cup of coffee?"): nor will we
answer, "Yes, a small coffee, please". The elliptical forms
presumably are not used in these contexts because these
drinks are not packed in some standard containers at home.

Summarizing the above discussion of'bmmd, muggy,

aim and mafia, let us state another conversion principle:

Principle 3:

Mass -- > Count: In commercial contexts, a mass noun,
through constant application of the
noun to well-divided instances of the
referent, can gain an individuation for
its reference and thus can be used as
a count noun.

In this section, the complications of the binary
mass/count distinction are resolved by treating the uses of
a noun and not the noun itself as being mass or count.

In contrast to Allen's approach, the countness of a noun is
not treated as a question of percentage of countability.
Instead, a noun is taken as being basically count (e.g.
book, car) or mass (e.g. bread, surgery) or both (e.g.

chicken, authority), and then a conversion principle is

offered to account for its converse use. Three such

43
principles have been offered.

In the preceding section, it was noted that Allan
identified a group of abstract nouns (e.g. ahaafly)*which
are basically mass, but admit the singular article a0» when
they co-occur with a restrictive modifier. Let us state

this in the form of a conversion principle:

Principle 4:

Mass -- > ?Count: Abstract non-count nouns such as a
derived nominal or a name of a
subject for study, when modified by
'a restrictive modifier, admit ah»,
which is the equivalent of akind of
in this context.

2.4 IMPLICATIONS FOR L2 LEARNERS

What implications does the preceding discussion

have for L2 learners in their choice of English articles?
In English, there is a systematic distinction in the

choice of articles between two classes of nouns, count and

COUNT MASS
the + N/Ns the + N
a + N O + N
0 + Ns

For mass nouns the choice is limited between 0 and Mw'
whereas for count nouns it is more complicated. It has to
be decided first whether the noun in question is singular or
plural. For singulars, the choice is between a and Mw'

and for plurals O and Mn» 0 or a on the one hand

indicates indefiniteness and Mm on the other hand

44
indicates definiteness. Definiteness and indefiniteness
(including singulars and plurals) will be discussed in the
next chapter. In the above chart, we see only one
overlapping, i.e. in the form Mm N. We cannot tell whether
a given noun is used countably or uncountably just by
looking at the form the N. In other words, only in cases
where a noun is definite and at the same time it is
singular, we do not have to make a distinction between count
and mass nouns for the choice of articles. Except for
cases like this, the question ”Is this noun count or mass?"
has to be answered first.

A L2 learner might want to answer this question through
some kind of logical inference. A first hypothesis might
be that nouns referring to concrete objects (e.g. ”milk”,
'book") are countable and nouns referring to abstract
entities (e.g. "idea", "music') are uncountable. But this
is not true of English. A second hypothesis might be that
only those nouns referring to discrete, differentiated
concrete objects (e.g. ”lettuce", "pillow") are countable.
But this is not true, either.

In English, even for two concrete (material) nouns
referring to objects with similar shapes, one can be count
and the other mass. The same thing happens with abstract
nouns. For two abstract nouns with similar meanings, one
can be mass and the other count. This arbitrariness is

exemplified as follows:

45

CON TE ABSTRACT
CO 8 CO 8
carrots aseragus qugestions adLice
noodles spaghetti proofs evidence
beans rice jobs work
chairs furniture ideas knowledge

The arbitrariness of noodle being a count noun and

spaghetti a mass noun, on the one hand, and idea being

count and lorowledge mass, on the other hand, suggests that
the mass/count distinction is not simply in the nature of
the things referred to. Being mass or count is part of the
meaning of a word. A count word includes an individuation
in its meaning, but not a mass word.

Furthermore, Allan's eight levels of countability
suggest that the mass/count distinction is not a simple
binary one. The traditional view that a noun is either
[+count] or {-count] is inadequate. Not all English nouns
have a fixed feature [+count] or {-count]. There are words
like oak which are both [+count] and {-count]. There are
words like her which are {-count} except that they admit
a0» under certain circumstances.

The discussion of mass/count conversion shows that
every concrete count noun can be converted into a mass noun
(recall the Universal Grinder), but not vice versa. Gleason
(1965), and Pelletier (1979) went wrong in saying that every
mass noun can be converted into a count noun. Bach's (1986)

Universal Packager cannot convert every mass noun to count,

47

overall picture of the mass/count distinction as I see it
for L2 learning. I start with a common sense. The things
in the world can be divided into two categories, abstract
and non-abstract. Abstract entities do not have physical
forms and hence do not occupy any space. Since they do not
have physical forms, we cannot count them perceptually.
Therefore, let us assume initially that nouns which denote
abstract entities are all non-count. On the other hand, non-
abstract entities can be divided into two subcategories.
One contains discrete, differentiated objects, i.e. objects
with definite outlines or precise limits. The other
contains substances or masses that do not have natural
boundaries. The former is presumably countable and not the
latter. Therefore let us assume that nouns which denote the
former are all count and nouns which denote the latter are
all non-count. In brief, the three assumptions that I made
are:

Assumption 1: Nouns which denote discrete objects are

all count.

Assumption 2: Nouns which denote undifferentiated
substances are all non-count.

Assumption 3: Nouns which denote abstract entities are

all non-count.

Obviously, all of the three assumptions need to be
mmdified. Assumption 1 that nouns which denote discrete
objects are all count has a problem. There are nouns the
referents of which come as discrete objects, but they are

always non-count . Asparagus , spaghetti , rice , lettuce ,

48
fluniture, and equipment are some examples. The last two
are superordinate category terms which tend to become mass,
as Markman’s (1985) claimed. Yet in English we still have a
lot of superordinate terms which are count such as vehicles,
and “as. ‘Under our assumption, nouns which denote discrete
objects should be count. Hence we still have to treat words
like furniture as exceptions, although they have good reason
to be mass, if Markman is right. It is suggested that for
these exceptions, the learner learn each noun together with
a unit word that it goes with, e.g. agrainofrice, ahead
oflettuce, marticleoffirmiture. The countness of this
type of noun has to be learned by rote.

If the above-mentioned exceptions can be taken care of,
Assumption 1 and 2 can stand as they are, if we do not
regard a noun as having a fixed feature of [+count] or
{-count]. It is the sense that a noun has, not the noun
itself, that determines whether it is count or mass. We
have nouns with two distinct senses, such as chicken and
oak. ‘We have nouns with both the count and the mass use,
such as candy and hair. We-have nouns which are normally
count but given the right context can be used uncountably,
such as car and book. For all these nouns, when used to
denote a discrete object itself, they are count: whereas
when used to denote the material constituent or the mass of
the discrete object, they are mass, as made clear in
Conversion Principles 1 and 2.

Although we have expressions such as alarge coke and

alightbrown bread, assumption 2 that nouns which

49
denote undifferentiated substances are all non-count is
still valid, because Conversion Principle 3 states that only
in commercial contexts do these expressions occur.

Parallel to our assumption about discrete objects,
Assumption 3 that nouns which denote abstract entities are
all non-count has a problem, too. There are nouns which
denote abstract entities, but they are always count.

Tricks, ideas, and characteristics are some examples. It is
suggested that the learner learn the plural form Ns or the
singular form a0» N, and not simply N, for these words.
This is another type of noun whose countness has to be
learned by rote.

There is a second problem for Assumption 3. Allan
(1980) identified a group of abstract nouns which are
normally non-count, but when they are modified by a
restrictive adjective or relative clause, they admit the
singular article aou, which is equivalent to a
kindof in this kind of context. This group of nouns
includes names of subjects for study like physics and
chemistry, and derived nominals like sincerity, admiration
and.han. To maintain Assumption 3, we have to treat this
group of words as exceptions. Conversion Principle 4 takes
care of it.

Nevertheless, not all derived nominals that are
modified by restrictive modifiers allow a(n). Information,
evidence, and lorowledge are some examples. These are
abstract nouns which are always non-count, exactly the

opposite of those abstract nouns like tricks and ideas which

50
are always count. They have to be learned by rote. Again,
it is suggested that learning the word together with a unit
word that it goes with (e.g. a piece of evidence) might be
of help.

For the rest of the abstract nouns, they are all non-
count except for two situations. First, there is a group of
nouns which has two distinct senses. One of the two senses
denotes a certain abstract quality, and the other denotes a
person that possesses this quality. Authority and gossip
are two examples. When a noun is used to denote the person,
it is count: when it is used to denote the quality, it is
non-count. In fact, if we treat this kind of noun as being
ambiguous in the sense that what we have is two different
words instead of one, then they are not a problem for our
assumptions.

The second situation where abstract nouns are not non-
count is when separable instances of a certain quality or
action, and not the quality or action itself, are referred
to. Suwion, discussion, difi‘iculty, and experience
are some examples of words that have both the count and
the non-count use. Unlike concrete nouns such as army and
hab, which also have both uses, for abstract nouns, it is
harder to decide when to use which. The general principle
is that when things are done at different times or are of a
different nature, they are individuated and hence the count
use is the right choice. For example:

(34)a. After several long discussions, we finally
reached the conclusion that...

51
b. Chomsky’s discussions of transformational
grammar and the theory of government and
binding were boring.
On the other hand, consider:

(35)a. What implications does the preceding discussion
have for L2 learners in their choice of English
articles?

b. What implications do all the preceding
discussions have for L2 learners in their
choice of English articles?
The context for (35) is that I am writing a thesis on the
topic of English articles. Since the discussion is on one
topic and done by one person and within a single unit,
sentence (a), the non-count use, is a better choice.

In brief, the whole picture of the mass/count

distinction presented above is as follows:

”\‘H

Abs ract o rete

[ -count] /\5
discrete ma s

 

[+count] {-count]
VI
exceptions: exceptions:
a) bums, etc. ram, etc.

b) a sincerity, etc.
o) dhuwnums, etc.

Chapter III

Definiteness and Indefiniteness

3.0 INTRODUCTION

This chapter addresses the question: What makes an NP
definite and what makes it indefinite? We have noted that
the grammatical correlates of articles with the distinction

between count and mass nouns are as follows:

Count Mass
Definite the + N/Ns the + N
Indefinite a + N O + N
0 + Ns

If an NP is definite, regardless of whether the head noun is
used countably or uncountably, dw'is used. On the other
hand, if an NP is indefinite, 000 is used for singular
count nouns and the zero article is used for plurals and
mass nouns. If we can make explicit what makes an NP
definite and what makes it indefinite, it will help L2
learners in their choice of articles.

In this chapter, it is argued that "existence",
”uniqueness" and "familiarity" are three characteristics of
definite NPs that distinguish them from indefinite NPs.
Section 3.1 addresses the problem of existence claims
carried by statements containing definite and indefinite
NPs. Section 3.2 discusses the uniqueness and non-

52

53
uniqueness properties of definite and indefinite NPs.
Section 3.3 is about familiarity. Section 3.4 deals with.
indefinites. Finally, section 3.5 is some concluding

remarks.

3.1 EXISTENCE CLAIM

Russell's (1905) well-known analysis for flw'contains
two claims about the definite article. He analyzes (1a) as
having (1b) as its logical structure:

(l)a. The king of France is bald

b. 3x (Kx8Vy (Ky->y=x&Bx)
In (lb), Kx stands for "x is king of France" and Bx stands
for ”x is bald". (1b) says that there is an individual such
that he is king of France and that no one but him is king of
France and that he is bald. To assert (1a) is to assert
three things: (a) there is a king of France, (b) there is
not more than one king of France, and (c) there is nothing
which is king of France and is not bald. Among these three,
what concerns us now is the first. To Russell, the definite
description ”the king of France" in (1a) entails the
existence of the individual, king of France. If there is no
king of France, (1a) is false.

Strawson (1950) disagrees with Russell’s analysis. He
argues that if there is no king of France, the question
whether the statement of (1a) is true or false simply does
not arise. Strawson says that when someone says to you,

”The king of France is wise", you will not say, "That's

54
untrue". Instead you will probably say, "There is no king
of France.” From this, Strawson argues, we can see
that for the statement of (1a) to be judgeable to be true or
false, there must be a background assumption that there is a
king of France. For a statement to be true or false, its
presupposition(s) must be true. When presuppositions fail,
statements have no truth value.
In brief, Strawson differs from Russell in that
under Strawson's analysis, the existence of individuals
referred to by definite descriptions is presupposed, and not
part of the assertions of the statements containing those
descriptions.
Quine (1960, p. 113) takes a similar position to
Strawson's. He says:
Sentences like "Mama sings" and "I saw the lion",
which contain definite singular terms, may indeed
be said to depend for their truth on the
existence of objects named by those terms,
but...they do not clearly become false failing
such objects. Failing objects of reference for
their definite singular terms, such sentences are
likely to be looked upon as neither true nor
false but simply as uncalled for.
From the above remarks, we see that Quine, like Strawson,
holds that definite NPs presuppose the existence of
objects denoted by those NPs.
At this point, I think that two semantic relations,
entailment and presupposition, need to be clarified .
A sentence A semantically entails a sentence B if and only

if every situation that makes A true, makes B true. For

example:

55

(2)a. I saw the lion.
b. There was a lion.

(3)a. The person wearing a green hat killed Smith.
b. There was a person wearing a green hat.

In both (2) and (3), if sentence (a) is true, sentence (b)
must be true, i.e. we cannot find a situation where (a) is
true and (b) false. Therefore (2a) entails (2b) and (3a)
entails (3b).

On the other hand, a sentence A semantically
presupposes a sentence B if and only if A entails B and the

negation of A also entails B. For example:

(4) I did not see the lion.

(5) The person wearing a green hat did not kill Smith.

we have seen that (2a) entails (2b). Does (4), the
negation of (2a), also entail (2b)? That is, can we
imagine (4) being true and (2b) false? No. Whenever (4) is
true, (2b) must also be true. Hence (4) entails (2b).
Since (2a) entails (2b), and its negation also entails (2b),
we say that (2a) presupposes (2b). Similarly, (3a)
presupposes (3b). Although there are problems with the
semantic notion of presupposition illustrated above (e.g.
the projection problem for presuppositions: see Levinson,
1983, pp. 199-204), those problems do not affect the claim
that definite NPs presuppose the existence of objects
denoted by those NPs.

We see that a definite description like the lion or

theperson wearingagreenhat not only entails the

56
existence of the referent of the description, but also
presupposes it. How about indefinites? Consider:

(6)a. I saw a lion.

b. I did not see a lion.

c. There was a lion.
(6a) entails (6c), but (6b), the negation of (6a), does not.
Hence (6a) does not presuppose (6c). Affirmative statements
containing indefinites such as (6a) entail the existence of
the entities denoted by the indefinite NPs, but they do not
presuppose it.

The hypothesis that indefinite NPs do not carry
existential presuppositions can be further supported by
examples such as:

(7)a. You must give her a call.

b. I should have written him a letter.
It is impossible for the indefinite NP acafl in (7a) or a
letter in (7b) to carry an existential presupposition
because in (7a) the call has yet to come into existence by
virtue of your calling him: and in (7b) the counter-factual
verb form should have written implies a negation and hence
the non-existence of the letter in question.

Karttunen (1969, 1976) points out that if we go on to
refer to such non-existing entities as those in (7) by using
the definite article, which carries an existential
presupposition, a semantic anomaly arises, as shown in

(8)e. You must give her a call. ?She is expecting the

call.
b. I should have written him a letter. ?The letter

57

was crucial for my promotion.

Note, however, if the continuation is in the same "world" as
its preceding sentence, the semantic anomaly will not arise,
as shown in

(9)a. You must give her a call. She will be very

happy to receive the call.
b. I should have written him a letter. The letter
would have been crucial for my promotion.
In (9a), the modaltmun in the first sentence creates a world
and the second sentence keeps the same world through the
model will. Similarly, in (9b) , both sentences occur in a
counter-factual world.

This indicates that the existence which a definite NP
presupposes can be of various kinds. We can have existence
in the actual world, in a future world, in a counter-factual
world, in a dream world, etc. The semantic well-formedness
of the anaphoric use of Mw’depends on the consistency of
the world in which it occurs with that of the antecedent.

When we say that definite descriptions carry
existential presuppositions, the existence we talk about is
always "discourse existence", and not necessarily real
existence in the real world. A speaker can presuppose the
existence of the entity he is talking about, whether the
entity at issue really exists or not. The discourse creates
a world of its own. For example, the sentence Thebhgof
Franceiswise is grammatical, although there exists no
king of France. The king of France exists in the discourse

world. Lack of real existence in the real world does not

58
lead to ungrammaticality or any semantic anomaly.

From the above discussion, we get one semantic
distinction between definite and indefinite NPs. That is,
definite NPs presuppose the existence of entities denoted by
,those NPs whereas indefinite NPs do not: they merely entail
it. For L2 learners, this means that in deciding which
article to use, definite or indefinite, they must ask
whether or not the existence of the entity in question can
be brought into the discourse in some way. If the answer is
no, the use of the definite article will be infelicitous.
For example, a person can say, "Is the king coming?",
presupposing there is a king, but if the king in question
has not been introduced into the discourse in some way,
that is, without a proper context, his question will be
infelicitous, although it is grammatical.

How does a certain entity come into existence in
discourse? The most obvious way is through an explicit
previous mention. For example:

(10)a. She checked out a book on fish yesterday. And

she has finished reading Mw’book.

b. I saw a unicorn. The unicorn was hurt.
In (10a), the first sentence entails that there was a book
on fish that she checked out yesterday. The second sentence
goes on to talk about the same book. In (10b), the first
sentence, like that in (10a), entails that there was a
unicorn that I saw. Although in reality there exists no
unicorn, the existence of a unicorn is brought into the

discourse as a consequence of the statement Innvawuaawn.

59
This has been labelled ”the anaphoric use" of flwu

A second way is through an association with an entity
that has been mentioned previously. For example:

(11) She checked out a book on fish yesterday. And she

has read ten books by the same author.
In (11), if there exists a book, the person who wrote it
must have existed at some point. By mentioning a book, the
existence of its author is introduced into the discourse by
inference.

A third way is through the immediate situation where
the discourse occurs. For example:

(12)a. Pass me the salt, please.

b. Watch out for the dog.

In (12), the addressee may or may not be aware of the
existence of the salt or the dog, but its existence can be
brought into the discourse through the command because if it
is a felicitous command, the salt or the dog in question
must be in the immediate situation.

The existence of the entity in question can also be
brought into the discourse through an immediate linguistic
follow-up, as in

(l3)a. This chapter addresses Mm question: What is

the responsibility of intellectuals?

b. Here is Mw’answer: The responsibility of
intellectuals is to expose the lies of
governments.

In (13), the existence of the entities in question is

brought into the discourse by what follows immediately after

60
the entities are mentioned.

A fourth way is through shared knowledge between
speaker and addressee. For example:

(14)a. I'm going to the library.

b. The mayor visited our school yesterday.
In (14a), the existence of the library is brought into the
discourse through the mutual understanding that there is a
library that the speaker goes to. In (14b), it is the mutual
knowledge that there is a mayor in the city where the
speaker (and the addressee) lives that brings the existence
of the mayor into the discourse.

For sentences like (10)-(14), the above four ways
account for the fulfillment of the first requirement of the
use of due since Mw’carries existential presuppositions,
the existence of the entity in question must be brought into
the discourse in some way. A common characteristic of these
four ways is that the entity denoted by the definite NP must
find a so-called ”anchor" for its existence. The anchor may
be an explicit previous mention, or an association with such
a mention, or the immediate discourse context or shared
knowledge between speaker and addressee.

Besides these simple cases, we have numerous
complicated ones. By simple and complicated, I am referring
to the internal structure of the noun phrase. How do we
account for the existence of the entities denoted by the
definite NPs in the following cases (quoted from McCawley's

(1976) introductory remarks on Karttunen’s paper "Discourse

61
Referents", pp. 363-64):

(15)a. Karttunen's paper can be described as a summary
of Mwwwwkonrakemmeflhdgaummhe
gnanmmmhnswenndkmuuodofiWVhenaufiWe
yams.

b. .. .existential quantifiers have thedual
,fiawdanofasethgednaueamdof

LMMmhwagwuanuumufluucanfigmebr
ndmamumfldbanmun

In (15a), the underlined definite NP is composed of a head
NP followed by a restrictive relative clause, and in (15b),
it is composed of a head NP followed by a conjunction of two
cj'phrases. What can be the anchor for the existence of the
two entities denoted by the definite NPs? None of the above
four ways can be applied to these cases. For (15a), it has
been argued by Smith (1969), Vendler (1967) and others that
restrictive relative clauses can serve as a grammatical
previous mention.

For vendler, Mw’can be used with a noun only when the
noun is “identified" by a sentence. According to Vendler, a
sentence identifies a noun if it connects the noun with a
definite NP. In some cases, Vendler notes, "nouns are
identified by the mere presence of a verb in the past tense"
(p. 64). Consider:

(16)a. The man that I met wore a hat.

b. I met a man.

According to Vendler, the noun man in (16a) is identified by
(16b) because (16b) connects the noun with a definite NP, I,
and the verb nu! is in the past tense. We see that a

relative clause can serve as a grammatical previous mention

62

in the sense that a sentence like (16a) can be interpreted

as
(17) I met a man. The man wore a hat.

Following the line of the argument, sentences like
(15a) can be interpreted as:

(15)a'. Generative grammarians were about to do some

work on reference for the next five years.
Karttunen’s paper can be described as a
summary of that work.
Along the same line, we can say that (15b) can be
interpreted as:
(15)b'. Asserting existence and introducing a constant
that can figure in subsequent discourse is
a dual function. Existential quantifiers have
this dual function.
Thus, of phrases, like restrictive relative clauses, can
also serve as a grammatical previous mention.

All relative clauses miss one constituent, the head NP,
and hence structurally, to fill the missing slot, the head
NP can appear in the relative clause as a first mention of
the entity. However, semantically, a relative clause may
fail in bringing the existence of the entity in question
into discourse, and hence Mm is impossible. For example:

(18) He welcomed me with a/¢/*the warmth that he had
never shown before.

(18) cannot be interpreted as

63
(18') He had never shown (any) warmth before. He
welcomed me with that (kind of) warmth.
We see that if a relative clause contains a negative
predicate, it fails to serve as a grammatical previous
mention, because the existence of the referent of the head
NP is explicitly denied by the relative clause.
In fact, even in a case like
(19) He welcomed me with a/¢/*the warmth that was
not at all surprising.
where the existence of the entity denoted by the head NP is
not denied by the relative clause it still fails to serve as
a grammatical previous mention. (19) cannot be interpreted
as
(19') Warmth was not at all surprising. He welcomed
me with that (kind of) warmth.
In (19) the relative clause does not delimit a particular
kind of warmth and hence it fails to serve as a grammatical
previous mention. In contrast to (19), consider:

(20) He welcomed me with the warmth that was
characteristic of him.

(20) can be interpreted as

(20') Warmth was characteristic of him. He welcomed
me with that (kind of) warmth.

Note that (19) and (20) are similar in meaning. But one
allows Mw'and the other does not.

In brief, restrictive relative clauses can serve as

64
grammatical previous mentions except when they contain a
negative word which either contributes to an explicit
denial of the existence of the entity denoted by the head NP
or deprives the relative clause of its delimiting power.

As for of phrases, the relation between NPl and NP2 in
the form "NPl of NPZ' is different. It can be of various
kinds. Only when they stand in a particular relation can
the of phrase serve as a grammatical previous mention. For
example:

(21)a. In this article, we discuss thesignificance of

intuition.s... (Gass, 1983, p. 274)
b. Theinterpretation ofsuch phrases is a matter
of considerable difficulty... (Russell, 1905,
c. iIzhefai'liire ofpresuppositions leads to truth-
valueless sentences.
In (21a), the of phrase can function as a grammatical
previous mention in that the sentence can be interpreted as:

(21)a'. Intuitions are significant. In this article,'

we discuss that (kind of) significance.
In (21a') we see that the existence of the entity denoted by
the definite description the significance of intuitions is
brought into the discourse through an association with an
explicit mention of the concept of the entity expressed in
the form of a statement, Intuitions aresignificant.

Similarly, (21b) and (21c) can be interpreted as:

(21)b'. We (people) interpret such phrases. That

(kind of) interpretation is a matter of
considerable difficulty.

c'. Presuppositions fail. That failure leads to
truth-valueless sentences.

65

Summarizing the three cases in (21), we find three
different relations held between the head NP and the of

phrase:

(21a) the significance of intuitions < intuitions are
significant

NP1 of NP2 < NP2 be adjective

(NP is a derived nominal of the adjective.)

1

(21b) the interpretation of such phrases < we
(people) interpret such phrases

NP1 Of NP2 < Verb NP2

(NP is a derived nominal of the transitive verb.)

1

(21c) the failure of presuppositions < presuppositions
fail

NP of NP < NP Verb

1 2 2

is a derived nominal of the intransitive
verb.)

(NP1

We see that when a head NP stands in any of these three
relations with the object of the of phrase, the of phrase
can serve as a grammatical previous mention. Let us call
these relations a thematic relation.

Besides the thematic relation, we see another kind of

relation in (15b), repeated here:

(15)b. . ..existential quantifiers have thedual

.fiuuaoncfamwnhgwmmumweandcf

° ' acanmmutMucanfigweui

admapumtdhanmun

b'. Asserting existence and introducing a constant

that can figure in subsequent discourse is a
dual function. Existential quantifiers have
this dual function.

66

In (15b), the relation is:

(15b) the dual function of asserting... <
Asserting...is a dual function

NP1 of NP2 < NP2 is NP1

(NP2 is in apposition to NPl)

Two more examples of this relation:

(22)a. Thesubjectofdenoting is of very great
importance... (Russell, 1905, p.93)
b. I shall now attempt partially to clarify Mm
nodanofnfbnhg. (Searle, 1969, p. 26)

(22) can be interpreted as:

(22)a’. Denoting is a subject. This subject is of
very great importance.
b’. Referring is a notion. I shall now attempt
partially to clarify this notion.'

Let us call this kind an appositive relation. Finally we

have a relation of possession as in:

(23)a. It is widely accepted that the language of
second languagelearners...is a system in its
own right. (Gass, 1983, p. 273)
b. Presuppositions can be created or destroyed in
thecourseofaconversation. (Lewis, 1979,
p. 172)

D

(23) can be interpreted as:

(23)a'. Second language learners have a language of
their own. It is widely accepted that this
language is a system in its own right.

b'. A conversation has a course of its own.
Presuppositions can be created or destroyed in
this course.

67

To sum up, in the form of "NPl of NP ", if NP1 and
NP2 have a thematic, appositive, or possessive relation,
the of phrase, like the restrictive relative clause, can
serve as a grammatical previous mention. Now consider:

(24)a. In order to get *the custody ofhis daughter,

he made an appointment to see his lawyer.

b. The Joneses are getting a divorce, but it has
not been decided who will have *the custody of
flui'oMMhen.

In (24), we do not find a thematic or appositive or
possessive relation between custody and his daughter/their

children. Thus the of phrase cannot serve as a grammatical
previous mention in this case. In addition, there is no
other proper way available to bring the existence of the
entity, custody of his daughter/their children, into the
discourse. As a result, the first requirement for the use
of the cannot be fulfilled.

Many other forms of restrictive noun modifier, such as
infinitives, present participles, past participles, and
propositional phrases headed by between, from, at, etc. , can
be regarded as a reduced form of restrictive relative
clauses and hence can serve as grammatical previous
mentions.

Summarizing this section, the definite article carries
existential presuppositions and hence the first requirement
for the use of Mw’is that the existence of the entity
denoted by the definite description must be brought into the

discourse in some way. It can be introduced by an explicit

previous mention, association with such a mention, or a

68
grammatical previous mention. The immediate discourse
context or shared knowledge between speaker and addressee
can also bring the existence of an entity in question into
the discourse . The fulfillment of this first requirement,
however, does not necessitate the use of flux It only makes
‘it possible, as a first step, so to speak, to use Mm. A
proper use of Mn needs to meet two other requirements,
uniqueness and familiarity, which will be discussed in the

next two sections.

3.2 THE (NON)-UNIQUENESS PROPERTY OF (IN)DEFINITE NPs

In the preceding section, we started our discussion with
Russell's analysis of Mn» We mentioned that he makes two
claims about the definite article. First, it entails the
existence of the entity named by a definite description. We
have also noted that this claim is challenged by Strawson
and others. we came to the conclusion that it is
existential presuppositions and not simply entailments that
definite NPs carry. We now turn to discuss Russell’s second
claim. Recall the content of the second part in his logical
structure for IhekingofFranceisbald is that there is
only one king of France. This is the well-known uniqueness
analysis of Mw'by Russell.

If uniqueness means one and only one individual,
obviously it can apply only to the use of Mm in singular
count nouns, but not to plurals or mass nouns. Seeing
this, Hawkins (1978) argues that uniqueness is not part of
the meaning of Mn» Instead, in his theory, dw’implies

69

inclusiveness and a0» exclusiveness. By inclusiveness, it
is meant that a referring expression with the linguistic
form ”Mw’(modifier) noun (modifier)" refers inclusively to
the totality of the set of objects designated by the
definite description. And by exclusiveness, it is meant
that an indefinite description refers exclusively to ”not-
all', i.e., there exists at least one object meeting the
description to be excluded from the reference of the
indefinite description.

Some examples to illustrate this point are as follows:

(25)a. Fred lost a leg/?nose in the war.

b. Fred lost some fingers/?arms in the war.
(26)a. Fred is the/*a taller of the two.
b. we went to see a house, but we didn't buy the
house because the/*a roof was broken.

In (25a), Fredlostanose is odd because Fred has only
one nose, but the use of 0 requires that there should exist
at least one other nose of Fred's to be excluded from the
reference. In (25b), Fredlostsomearms is odd because it
means Fred lost at least two arms, but the use of the
indefinite swneirequires that besides the two arms that Fred
lost, there should exist at least another arm of Fred's to
be excluded from the reference. Similarly, in (26a), within
a set of only two people, the number of the taller people
must be exactly one and in (26b), a house usually has only
one roof: hence in both cases, it is impossible to exclude
at least one object from the reference of the indefinite

description.

70
Does the inclusiveness and exclusiveness theory of Mw'
and a make correct predictions of the use of the two
articles? I pointed out in my master's thesis (1985) that
the theory runs into trouble with articles as head NP
determiners of restrictive relative clauses. Hawkins tries
to distinguish two subtypes of restrictive relative clauses
to prevent the exclusiveness meaning of a from applying
where it should not apply. Consider:
(27)a. A girl who lives in Chicago sent me a nice
Valentine card.
b. A book that I read last night gave me a big
headache.
In (27a), there is no problem for the exclusiveness meaning
of a to apply: there is at least one other girl living in
Chicago. In (27b), however, Hawkins is aware that his
exclusiveness meaning of a should not apply because (27b)
does not necessarily indicate that I read more than one book
last night.
fiawkins' two subtypes of restrictive relatives are
called "establishing relative clauses" and "non-establishing
relative clauses". Two major diagnostics are presented for
these two types of relative clause. The first diagnostic is
as follows:
...sentences with establishing relative clauses
permit semantically identical paraphrases in
which the content of the establishing relative

clause becomes a main clause.
(p. 267)

For example:

71

(28)a. A story that I read ten years ago contained the
same plot.
b. A story which was very long contained the same
plot.

(28a) contains an establishing relative clause, and (28b) a
non-establishing relative clause, because (28a) can be
paraphrased as (28a’), but a similar paraphrase of (28b) is
unacceptable as shown in (28b'):
(28)a’. I read a story ten years ago and it contained
the same plot.

b’. ?A story was very long and it contained the
same plot.

The second diagnostic for establishing relatives is as

follows:

[An establishing relative clause] functions
pragmatically to relate the new referent to
previously mentioned or known objects, to
participants in the talk exchange, or to objects

in the immediate situation.
~(p. 263)

For example:

(29)a. What’s wrong with Tom? on, he flunked in a
test he took yesterday.
b. What's wrong with you? Oh, a tape-recorder
that I bought last week was broken.
c. Would you please go and get me a book that is
over there on the shelf in the right-hand corner?
The relative clauses in these examples are all establishing
relatives. In (29a), the relative clause he tookyesterday
relates the new referent test to a previous mentioned
object, Tom: in (29b), the relative clause I boughtlast

week relates the new referent tape-recorder to a participant

72
in the talk exchange, I: and in (29c), the relative clause
thatisoverthereontheshelfintheright-handcomer'
relates the new referent book to the immediate situation.

By distinguishing these two subtypes of restrictive
relatives, Hawkins has the following to say with regard to
the exclusiveness meaning applying to indefinite head NPs
followed by restrictive relative clauses:

...indefinites with establishing relatives do not
carry the exclusiveness presupposition...these
sentences [containing establishing relatives] do
not presuppose that there are other excluded
referents, but only that there may be such
referents, whereas a truck which isblue [a non-
establishing relative] does presuppose the
existence of more than one.
(pp. 266-67)
As can be seen from the above quotes, it is clear that
Hawkins holds that indefinites with non-establishing
relatives carry the exclusiveness presupposition, whereas
indefinites with establishing relatives do not.

Unfortunately his diagnostics fail to make correct
jpredictions. According to the two diagnostics, sentences
like the following:

(30)a. A zoologist who lives in Detroit is coming to see

me this Sunday.
b. A linguist who works on Black English
Vernacular is giving a lecture tonight.
c. A man who became quite well-known for receiving
an artificial heart died yesterday.
all contain a non-establishing relative clause. They fail
in the paraphrasability test (Hawkins' first diagnostic for

establishing relatives), as shown in the following:

73

(30)a’. ?A zoologist lives in Detroit and he is coming to
see me this Sunday.
b'. ?A linguist works on Black English Vernacular
and he is giving a lecture tonight.
c'. ?A man became quite well-known for receiving an
artificial heart and he died yesterday.
They also fail in Hawkins' second diagnostic because in
(30), none of the relative clauses relates the new referent
to previously mentioned or known objects, or to participants
in the talk exchange, or to objects in the immediate
situation. However, the exclusiveness meaning of a should
not apply here because none of the sentences in (30)
indicates that there necessarily exists at least one excluded
referent meeting the description of the indefinite NP.
In my 1985 thesis I concluded that NPs of the form
“000 x which Y’s" do not actually carry the exclusiveness
presupposition. They only suggest that there is a
possibility of more than one X satisfying the description of
Y’ s . The linguistic form a girl who lives in Boston itself
does not carry the exclusiveness presupposition. Rather, it
is the world knowledge that makes us infer that there is
:more.than one girl living in Boston. The indefinite article
can.be used with a restrictive relative even if there is
only one object meeting the description of the restrictive
relative.
Hawkins’ distinction between establishing and non-
establishing relative clauses does not‘work'with‘Mw as head

HP determiner of restrictive relatives, either. Hawkins

claims that the definite article with an establishing

74
relative can function successfully as a first-mention,
whereas the definite article with a non-establishing
relative cannot. For example:
(31)a. The game that we played in gym today was

terrific.
b. The man who was from the South was disgusting.

According to Hawkins, in (31a), the relative clause is an
establishing relative because it relates the new referent
game to participants in the talk exchange, we, and because
it can be paraphrased as:

(31)a'. We played a game in gym today and the game was

terrific.

Hawkins says, "The purpose of the establishing relative
is thus to do what a previous mention would do" (p. 138).
The in (31a) hence can function successfully as a first-
mention. In (31b), however, the relative clause is a non-
establishing relative because the following paraphrase is
unacceptable:

(31)b'. ?A man was from the South and the man was

disgusting.

The in (31b) hence cannot function as a first-mention. In
other words, the in (31b) must be anaphorical.

The best attack on the above claim of Hawkins' is to
find two syntactically-similar sentences containing
tau: restrictive relatives which relate the head nouns to
objects of the same nature, but one turns out to be able to

‘use Mw'as a first mention and the other fails to do so.

75
Allan (1986, p. 128) provides such an example:

(32)a. The woman I saw collecting tickets at the
station this morning looked just like my
mother.

b. The woman I saw at the station this morning
looked just like my mother.
Both sentences in (32) pass Hawkins’ paraphrasability test,
as shown in

(32)a'. I saw a woman collecting tickets at the

station this morning and she looked just like

my mother.

b'. I saw a woman at the station this morning and

she looked just like my mother.
And both relatives relate the head noun to a participant in
the talk exchange, I. Hence both relative clauses are
establishing relatives. However, in (32a), the underlined
Mw’can function as a first mention: whereas in (32b), it
would not usually be acceptable as a first mention.

Taking a closer look at (32), we find that whether Mm
can be a successful first mention or not depends on whether
or not the head NP together with the restrictive relative
clause ends up referring to only one individual. In (32a),
uniqueness is guaranteed because there is usually only one
station functionary: whereas in (32b), it is highly possible
that a person sees more than one woman at a station at a
certain time and hence uniqueness is not guaranteed. If
uniqueness is not guaranteed, without an explicit previous
mention, the addressee will not know which one is being

talked about.

Summarizing my discussion on Hawkins' inclusiveness and

76
exclusiveness theory of the definite and indefinite article,
I have pointed out that his distinguishing two subtypes of
restrictive relative clauses to prevent his exclusiveness
meaning of 000 from applying where it should not apply
fails to do the job for him. I have also pointed out that
his distinction between the two subtypes also fails to
predict when Mw'can be used as a successful first-mention.
Consequently, his exclusiveness theory for indefinites and
his distinction between establishing and non-establishing
relative clauses should be abandoned.

I argue for Russell's uniqueness theory, but first of
all, the notion of uniqueness must be expanded to cover
plurals and mass nouns. Uniqueness means one and only one
for singular count nouns, and it means all for plurals and
mass nouns. For example, the book indicates that there is
one and only one book: the books refers to all the books
there are in the relevant domain of interpretation; and the
nu“? refers to all the water there is in a discourse
context. This treatment does not mean that Mm has two
meanings. Basically, the uniqueness defined here is like
Hawkins' inclusiveness. A definite description of the form
'Mw*+ HP(s)" refers to the totality of the set of objects
designated by that description. For singular count nouns,
there is one and only one member in the set, and hence the
total is only one object. For plurals and mass nouns, the
total is all that there is in the set.

Note that although Mm presupposes uniqueness, a does

not presuppose non-uniqueness. This is exactly where

77
Hawkins’ exclusiveness theory went wrong. In (29) and (30),
repeated here:

(29)a. What’s wrong with Tom? Oh, he flunked in a

test he took yesterday.

b. What’s wrong with you? Oh, a tape-recorder
that I bought last week was broken.

c. WOuld you please go and get me a book that is
over there on the shelf in the right-hand corner?

(30)a. .A zoologist who lives in Detroit is coming to

see me this Sunday.
b. ,A linguist who works on Black English
Vernacular is giving a lecture tonight.
c. ,A man who became quite well-known for receiving
an artificial heart died yesterday.
in all cases a is used, but none of the cases indicates that
there necessarily exists at least one excluded referent
satisfying the description of the indefinite NP. This is
a piece of evidence that a does not carry the exclusiveness
presupposition or that it is not necessarily non-unique, to put
it in another way. In all the cases in (29) and (30), a is
used because in none of these cases, is uniqueness
guaranteed. »In (29a) it is possible that a person takes
more than one test in a day: in (29b) it is possible that a
person bought more than one tape-recorder last week, etc.

On the other hand, when uniqueness is not guaranteed,
it does not follow that the referent of an indefinite
expression is necessarily non-unique. (29a) is consistent
with the situation where he took only one test yesterday and
(29b) is consistent with the situation where I bought
only one tape-recorder last week. But because uniqueness is

not guaranteed, a is used.

Quite contrary to Hawkins’ exclusiveness theory, Kadmon

78

(1987) reported that to some people, the following discourse
(33) A cat walked in. It sat down.

is not felicitous in a situation where there is more than
one cat that walked in. This means that acat here must be
unique in some way. It has been argued in section 3.1 that
the first requirement for the use of Mm is that the
existence of the entity in question must be brought into the
discourse in a proper way. In (33), although to the
speaker, the uniqueness requirement is met, since the cat
has not been properly introduced, the cannot be used.

In sum, my analysis differs from Hawkins’ in two
ways. First, he treats definites and indefinites as exactly
opposite to each other. To him, the carries the
inclusiveness presupposition and hence 000 carries the
exclusiveness presupposition. I have shown that a0» + N
does not necessarily indicate that there exists more than
one object designated by the indefinite description in the
relevant domain of interpretation. Under my analysis,
the form ”000 N” is used, not because there exists more
than one entity designated by H, but because either the
existence of the entity denoted by N has not been properly
introduced into the discourse or because uniqueness is not
guaranteed in the sense that the addressee has no way of
knowing that the entity in question is the only one. In
Kadmon’s example, (33) above, although the speaker knows
there was only one cat that came in, the addressee has no

way of knowing it without shared knowledge.

79

Secondly, to Hawkins, the referent of "flw’x which Y’s"
can be a first-mention only when the relative clause is an
establishing relative. I have shown that his
(non)establishing relatives fail to make correct
predictions. Under my analysis, the question again lies in
whether or not uniqueness is guaranteed, as shown in (32).

My proposal for the choice between the definite
article Mw*and the indefinite a00 or the zero article
depends on whether the existence and the uniqueness
requirement, plus another one, the familiarity requirement
which I will discuss in the next section, are all met or
not. If yes, Mw'is the right choice; if not, 000 or the
zero article is to be used. The question now is how to
decide whether a given entity is unique or not.

In the following, I will focus on NPs modified by of
phrases and NPs with restrictive relative clauses. In the
literature, it is often noted that when a noun is modified
by an of-phrase, the definite article is often used.
Christophersen (1939, p. 43) says:

A great many of-phrases demand a the. The
curious thing is that the same is not true to
such a wide extent of other prepositional phrases.
Christophersen (pp. 148-9), however, mentions a counter-
tendency to the above-cited remarks:
When the first substantive is an action-noun
having as its object the regimen of "of".
Examples: ...The Somme butchery led to

discussion of other slaughters / . . .They were
drawn together by examination of the scanty

80
flora...

This counter-tendency, unfortunately, is in effect cancelled
by Christophersen’s (p. 149) own remarks immediately
following his examples for that tendency:

In many of these examples the dwaorm might as

well have been used. In fact, in slow and

careful style the article is now more often used

in these cases...
With all these remarks, Christophersen’s position on
articles with of phrases amounts to saying that a great many
of phrases demands the, and for action nouns, the is
optional, depending on speech style. His observations might
be correct, but we want to know exactly when Mw’is required
and when it is not allowed.

Following the theory we have been developing here, to
use Mme two requirements have to be met. The first one is
the existence requirement. An of phrase will not demand the
if this requirement is not met, as in the example I gave in
the preceding section: In order to get custody of his
daughter, hemadeanappointmenttoseehislauyer. Since
custody and hisdauytter do not stand in a thematic, or
appositive, or possessive relation, the of phrase here
cannot serve as a grammatical previous mention and hence the
existence of the entity in question cannot be brought into
the discourse.

The second requirement is uniqueness. Consider:

(34)a. Describe procedures for protecting against or

minimizing potential risks and anwessment of
their likely effectiveness. (HSU regulations on

81
research on human subjects, 3A )
b. If methods of research create potential risks,

describe other methods... (38)
In (34a), assessments of the procedures in question can be
more than one, and hence uniqueness is not guaranteed: the
researcher is asked to describe only one of the possible
assessments. Similarly, in (34b), methods of research can
be more than one and some, not necessarily all, methods may
create risks.

For nouns which are basically count, like car and book,
it is relatively easier to know whether the entity in
question is unique or not, because there is an element of
individuation in count nouns, and in a sense we can count
their reference. For nouns which are basically either
[+count] or {-count}, it is a little bit harder to decide.
For example:

(35) A student who has been on an F-l visa for eight
consecutive years must apply for an extension
of stay. (News&Notes 16.1, 1987, Office of
International Students and Scholars, HSU, p. 3)

To make the right choice of articles for the phrase
extension ofstay, first of all, it has to be decided
whether the noun extension is count or mass. Basically it
can be either count or mass, as in

(36)a. The extension of Soviet influence in Central
America is a threat to the United States.

b. I’m glad I’ve got an extension of my summer
holidays.

The context of (35) indicates that students must apply for

82
their individual extension of stay and it can be done at
different times: hence individuation is possible for
extension here. After it is decided that the word can be
used as a count noun in this context, we ask, ”Is the
referent of extension ofstay here unique?” The answer is
no because a student applies for his own extension and he
applies for one extension at one time.

For nouns which are basically mass, because there is no
natural boundary implied for the reference of a mass noun
and especially, for abstract mass nouns, because there is no
real tangible thing out there, it is much more difficult to
decide whether a given entity is unique or not.
Christophersen’s observation that a great many of phrases
demand Mw'is true of abstract mass NPs modified by of
phrases. Consider:

(37)a. This phase is marked by the advent of terms

like ’roundness’... (Quine, 1960, p. 118)

b. We saw that the emergence of abstract singular

terms is not to be... (p. 120)
In (37) , the abstract mass nouns advent and emergence by
themselves denote homogeneous continua without implied
boundaries. The of phrase, however, delimits their
reference to a limited, specific territory. Within this
specific territory, the abstract entity, advent of terms
like ’roundness’, or emergence of abstract singular terms,
is all there is. This is why the uniqueness requirement is
always met.

Note that other prepositions such as from and in do not

83
have such a delimiting power. It has long been observed
that we say music fiom the south, but we say the music of
thesouth and we say lifeinthenineteenth century, but we
say thelife of the nineteenth century. Prepositions such
as in and fi'om are locative. In the phrase music from the
south, the prepositional phrase fiom the south only
specifies where the music in question is from. It does not
delimit the music to a particular kind: music from the south
can involve various kinds of music. Further, notice that in
the preceding section it was noted that a phrase like dw'
music of the south can be taken roughly as the south
hasitsomrmusic, i.e. the two nouns stand in a possessive
relation. If we possess something that cannot be
individuated, the possession is always inclusive. If the
south has its own music, it has all of it. Therefore, in
the form "N of NP", if N and NP stand in a possessive
relation, and N is used as a mass noun, the uniqueness
requirement is always met.
However, in cases like
(38)a. This is O evidence of the existence of dinosaurs.
b. I find O evidence of this second class to be even

more convincing... (Bickerton, 1981, p. 146)
In (38a) the phrase evidence of the existence of dinosaurs
can be taken roughly as ’the existence of dinosaurs is
evident’: that is, in the form "N of NP", N and NP stand in
a thematic relation. The uniqueness requirement is not met
in (38a) because the sentence means ’This is a piece of

evidence of the existence of dinosaurs’. On the other hand,

84
in (38b) evidenceof this second class can be taken roughly
as ’this second class is evidence’: that is, N and NP stand
in an appositive relation. The uniqueness requirement is
not met in (38b) because not all instances of this second
class as evidence have been given. A major difference
between abstract mass words such as evidence and those such
as advent and emergence in (37) is that the former can go
with a measuring word like piece, but not the latter. The
uniqueness requirement is always met when an abstract mass
noun which cannot go with any kind of measuring word is
modified by an of-phrase, as illustrated in (37) . In cases
like (38), where the abstract non-count head noun can co-
occur with a measuring word, the uniqueness requirement is
not always met.

How about Christophersen’s remarks on action nouns,
which are abstract mass nouns, that dw’is optional,
depending on speech style? A similar hypothesis is made by
Tang (1986). He says (pp. 114-5) that if the head noun
comes from an intransitive verb dw'is used, whereas if it
comes from a transitive verb either Mm or O is used. For
example:

(39)a. Our efforts succeeded.

b. the success of our efforts

(40)a. We accept such a policy.

b. the/8 acceptance of such a policy (is unlikely)
(p. 115)
To test Tang’s hypothesis, I asked 50 subjects to choose the

sentence they like better from each of the following pairs

85

of sentences:

(41)a.
b.
(42)a.

(46)a.

The use of a dictionary is not allowed in this
test.
Use of a dictionary is not allowed in this test.

Interpretation of a text may differ from person to
person.

The interpretation of a text may differ from person
to person.

The recognition of ambiguities is very hard for
second language learners.

Recognition of ambiguities is very hard for second
language learners.

Prevalence of bribery among officials often leads
to the fall of a government.

The prevalence of bribery among officials often
leads to the fall of a government.

Some linguists hold that the failure of
presuppositions leads to truth-valueless sentences.
Some linguists hold that failure of presuppositions
leads to truth-valueless sentences.

Failure of crops often results in famine.
The failure of crops often results in famine.

The only difference in (a) and (b) in each pair is the

choice of articles: one contains Mm and the other O.

In (41)-(43), the head noun of the NP containing the of-

phrase each comes from a transitive verb, i.e. uun

interpret and recognize, respectively. In contrast,

in (44)-(46), the head nouns come from intransitive verbs.

In (44), it is prevail and in both (45) and (46) it is

fail.

The result of this survey is shown in the following

chart:

86

the O both
(41) 25 24 1
50% 43% 2%
(42) '22 27 1
44% 54% 2%
(43) 1a 31 1
36% 62% 2%
(44) 33 16 1
66% 32% 2%
(45) 34 1s 1
63% 30% 2%
(46) 20 28 1
40% 56% 2%

The result shows that for the pair of sentences in (41), the
choice between Mn and O is almost equally split: 50% of
the subjects, i.e. 25 people, chose Mme 48%, i.e. 24
people, chose O and 2%, 1 person, chose both. For (42), the
choice was close, too. (43), however, shows a significant
difference in the choice. Most people prefer O to the.

For nouns which come from transitive verbs, out
of the three cases tested, two have an approximately equal
chance of co-occurring with either the or O, and one has a
greater chance of co-occurring with O. As for nouns
which come from intransitive verbs, the survey shows the
majority prefer Mw’in (44) and (45), but in (46) slightly
more people like O better.

Tang’s hypothesis that nouns which come from
intransitive verbs co-occur with.dw’was not borne out

Perfectly because in (46) we have 56% of the people

87

preferring the use of O. His hypothesis that nouns which
come from transitive verbs co-occur with either dw'or O was
supported, but we find that in (43), 62% of the people
prefer the use of O.

How can we explain all these results? Recognition in
(43) differs from use in (41) and interpretation in (42) in
that the former is much more an action noun than the latter
two. Recognition is mostly used as a non-count noun because
of the sense of action it carries from its verb counterpart:
whereas use and interpretation can be used as count nouns
in many cases. The generalization seems to be that if a
noun is used in its action sense, the demand for the use of
Mn is weakened. On the other hand, if it is used in a non-
action sense and in a very specific situation, the is
preferred. For example, in contrast to (42a), (repeated
here), which is a general statement of a certain state of
affairs, in (47) dw’is much more preferable:

(42)a. Interpretation of a text may differ from person

to person.
(47) The interpretation of this sentence comes out
like this: ...

In a word, action nouns followed by of phrases indeed
' weaken the demand of dwu However, it is not simply a matter of
speech style, as Christophersen suggested. Tang’s
hypothesis that it is transitivity that determines the
choice between dw’and the zero article was not supported,
either. Instead, the data above shows that it is activity

that affects the choice. As for failure in (46) , the

88
reason why 56% of the subjects prefer the use of O might be
that without Mm, the sentence is more law-like. Generic
NPs will be discussed in the next chapter.

Summarizing, in the form "N of HP", if the of
phrase can serve as a grammatical previous mention, and if
the uniqueness requirement is met, Mw'can be used. For
abstract mass nouns which cannot go with any kind of
measuring word, with the modification of the of phrase, the
uniqueness requirement is always met. Action nouns,
however, weaken the demand for Mm»

How I turn to the discussion of NPs containing
restrictive relative clauses. As with NPs containing of
phrases, it seems easier to determine whether a given count
noun is referring uniquely or not. For example:

(48)a. ?God spoke to a man who begot Isaac. (Vendler,

1967, p. 51)

b. I know the men who fought in Korea. (p. 50)

c. I know men who fought in Korea. (p. 51)
In (48), manwhobegotlsaac must refer to only one person:
it refers to the individual, Abraham. Uniqueness is
guaranteed in (48a), and hence Mw’is required. In (48b-c),
menwhofoughtinKorea refers to more than one person: if
the speaker knows all of them, “m is used: if he knows some
of them, O is used.

How about mass nouns? Consider:

(49)a. The food that he ate this morning made him

, sick.
b. He eats food that makes him sick. (Abbott,

p.c.)

89
(50)a. The coffee that I drank this morning is keeping
me wide awake.

b. Coffee from which extract is made is grown in
the lowlands. (Quine, 1960, p. 111)

In (49a), foodthatheatethismorning specifies a
particular quantity of food and within this specification,
that is all there is: thus the uniqueness requirement is
met. In (49b) , foodthatmakeshimsick does not specify a
particular kind (or quantity): it can be of various kinds of
food. Similarly, in (50a), cofl‘eethatldrankthismoming
has a specific reference, but in (50b) , coffee from which
extractismade does not: it can be of various kinds of
coffee.

Now consider abstract mass nouns:

(51)a. I don’ t understand the Chinese that he speaks.
b. He speaks *the Chinese that I don’t understand.

(51a) can be interpreted as:

(51)a’. He speaks Chinese. I don’t understand that
kind of Chinese.

(51b) , however, cannot be interpreted as:

(51)b’. I don’t understand Chinese. He speaks that
kind of Chinese.
Thus, in (51a), the relative clause can serve as a
grammatical previous mention whereas in (51b), it cannot.
Furthermore, in (51a), Chinese thathespeaks specifies a
particular kind of Chinese whereas in (51b), Chbumefluul
don’t understand cannot refer to a specific, natural

90
category of Chinese. It has been noted in the preceding
-section that a negative word in a relative clause to some
extent deprives the relative clause of its delimiting power.

To sum up this section, I have argued that to use Mm;
a second requirement, the uniqueness requirement, has to be
met. I discussed the possibilities for the fulfillment of
the uniqueness requirement for NPs containing restrictive
relatives and of phrases. For count nouns, if there is
only one individual referred to by the whole NP, the
requirement is met. If the whole NP refers to all
individuals in the discourse context, the requirement is
also met. For mass nouns, the requirement is met only when
the whole NP specifies a specific, natural portion of the
mass. CU'phrases always bring about this specification when
modifying an abstract mass noun which cannot go with any
kind of measuring word.

One dounterexample to the uniqueness requirement is
mentioned in Kadmon (1987). It is cases like Hekissed her
onthecheek. Despite the fact that everybody has two
cheeks and we do not know which cheek received the kiss, dw’
is used. Note that we have similar expressions like He took

herbythear'm. We can treat these exceptions as

collocations.

3.3 FAHILIARITY
Familiarity has long been held to be crucial in
the use of Mn; Christophersen (1939), Jespersen (1949) and

Heim (1982) have argued for it. The problems still open to

91
debate are: what counts as being familiar with the referent
of the definite NP? To whom is it familiar, only the
speaker or both the speaker and addressee? How is
familiarity achieved?
Pica (1983, p. 226) reported an investigation she made:
Upon arriving in an unfamiliar community, the
researcher asked a young woman for directions:

A: Can you tell me where I can find a drug store?
B: Oh. That’s right inside the terminal.

A: Terminal?
B: Yeah. You don’t know the terminal?

A: NO.

B: Oh. You’re not from around here?
In this short dialogue, in the first exchange, the young
woman took the location of the terminal to be shared
knowledge between her and the researcher. But it was not.
The researcher needed to be able to identify its location in
order to get to the store. Thus clarification is needed.
In the second exchange, the terminal was mentioned again.
This time it becomes "familiar”, although the researcher
still could not identify it: "familiar" in the sense that it
had just been mentioned. .

I mentioned in section 3.1 that restrictive relative
.clauses can serve as a grammatical previous mention. The
familiarity that they bring about is different from an
explicit previous mention. Consider:

(52)a. I met a poet last night. The poet is giving a

lecture today.

b. The poet that I met last night is giving a
lecture today.

92
In (52a), we have an explicit mention of the poet and hence
the second mention of it is "completely familiar": whereas
in (52b), without an explicit previous mention, Mwy to most
people, cannot be a successful first-mention. Consider,
however, the following case:

(53) I’m fed up with Mw'plumber that came to fix my

kitchen sink this morning.
The plumber himself, like the poet in (52), may be
unfamiliar to the addressee to the same extent, e.g., he
does not know anything about the plumber except that he is a
plumber. Yet, to most people, flw’in (53) is acceptable as
a first-mention.

The familiarity difference involved in (52b) and (53)
is not with the individual that the whole NP refers to but
with the event it refers to. A plumber’s fixing kitchen
sinks is not something unusual, and hence it is easy for the
hearer to accept dw'(plumber) as a first-mention: whereas
it is rather unusual for a person to meet a poet downtown
and hence it is hard for the hearer to accept Mw'(poet) as
a first-mention.

This phenomenon is called "bridging" (e.g. Clark,
‘1977) or "accommodation" (e.g. Lewis, 1979) in the
literature. Although a relative clause can serve as a
grammatical previous mention, if the information provided is
unusual, it is harder for the addressee to adjust his

assumptions. Consider:

(54)a. The zoologist that I talked to this morning

93
told me that the monkey is man’s best friend.
b. .A zoologist that I talked to this morning

told me that the monkey is man’s best friend.
The relative clauses in (54) can serve as a grammatical
previous mention, and hence the existence of the entity
in question has been brought into the discourse. Our first
requirement for the use of Mn is met. In the preceding
section, it was argued that in sentences like (54), because
uniqueness is not guaranteed in the sense that there is a
possibility that the speaker talked to more than one
zoologist on a certain morning, our second requirement for
the use of dw’is not met. Therefore, the subject NP in
(54a) cannot be a first mention. Here we have another
explanation. The relative clause in (54) provides rather
unusual information. If the addressee does not know
anything about the speaker’s talking to a zoologist or he
does not expect this kind of event at all, "accommodation"
or "bridging” is less likely to occur.

In (54b), the indefinite a is used because of the
unfamiliarity of the event. It does not indicate that there
necessarily exists more than one zoologist spoken to that
morning by the speaker. However, in a case like

(55)a. The test that I took yesterday was really hard.

b. .A test that I took yesterday was really hard.
if the addressee knows that the speaker is a student and it
is, say, finals week, the NP in (55a) can be a successful
first mention. Although the addressee does not know either

that the speaker took a test or that it is exactly one test

94
that the speaker took, nevertheless, when he hears (55a), he
can readily accept.Mw'(test) as a first mention because the
gap he has to fill in is not too big. That is, ”bridging”
or “accommodation” is possible. Notice that in this
situation, if the speaker uses a as in (55b), it implies
that he took more than one test.

we see that if accommodation is possible, there is no
reason to use a when the entity in question is unique.

This is in sharp contrast to the case in (54b) where a is
used not because of non-uniqueness but because of the
unfamiliarity of the event. Given cases like (55a), our
uniqueness requirement has to be modified a little bit.
According to the preceding section, we would say that in
(55) Mw’(test) cannot be a successful first-mention because
uniqueness is not guaranteed in that it is likely that the
speaker took more than one test on a certain day. Now we
have to allow a situation in which the uniqueness
requirement is met so long as the speaker, although not the
addressee, knows that the entity in question is unique. To
use Mm; nevertheless, the familiarity requirement also has
to be met. (55a) meets the familiarity requirement, but not
(54a). Thus we can allow Mw’in (55a), but not in (54a), to
be a successful first-mention.

Besides the familiarity difference involved in the
events referred to by the head NPs together with their
restrictive relative clauses, linguistic expressions
themselves might bring about some differences in

familiarity. Tang (1986, p. 108-111) says that in the form

1i

3‘1

95
'whh NP + restrictive relative clause”, if the verb in the
relative clause belongs to the "expected" category (e.g.
expect, characteristic), the is used with the head NP:
whereas if the verb belongs to the "surprising" category
(e.g. surprising, incredible), a is used. The rationale
behind this rule may be that the first category of verbs
suggests familiarity, and the second unfamiliarity. But he
says that without the preposition wuh, there is no such a
restriction. (He does not explain why there should be such
a difference.) His examples are as follows:

(56)a. The amateurs defeated the pros with the/*a/*O
regularity that was characteristic of this group
of amateurs.

b. The amateurs defeated the pros with *the/a/*O
regularity that was surprising.
(p. 110)

(57) The/An/*O impression that is
characteristic/surprising is that he is not out
of mind. (p. 111)

To see how much familiarity is brought about by verbs
belonging to the ”expected” category, and whether the
preposition wuh makes a difference, the following pairs
of sentences were given to 50 subjects:

(58)a. He welcomed me with a warmth that was

characteristic of him.
b. He welcomed me with Mm warmth that was
characteristic of him.

(59)a. At the welcome party, he showed Mw'warmth that

was characteristic of him.
b. At the welcome party, he showed a warmth that
was characteristic of him.

If Tang is correct, in (58), all (or almost all) the

subjects should choose Mw’because the predicate

96

chanumnfific in the relative clause belongs to the
”expected” category. The result, however, is that 54%,
i.e. 27 subjects (more than half) chose a: 42% chose Mw'
and 4% chose both. Further, there is not much difference in
the choice between the two pairs. In (59), 54% chose Mw’
and 42% chose a, 4% chose both. This indicates that the
Huh-phrase does not affect the choice.

Warmth that was characteristic of him indicates that
the speaker is somewhat familiar with the referent, and
hence‘Mw is possible: however, to approximately half of the
subjects, a was preferred. This suggests that although an
entity is familiar to the speaker, if he thinks that it is
unfamiliar to the addressee, a is preferable. We see that
in cases like this, the familiarity requirement is met for
some speakers, but not for others. Those who prefer
the probably assume that "bridging" or "accommodation" is
likely to occur on the part of the addressee. And those who
prefer a probably do not.

In numerous cases, familiarity with the referent of a
definite NP such as Quine’ s the advent of termslike
hmwubumr and memumwmeeqfidmaaaefiguhvnanu
(cited in (37)) comes to us through the denoting phrase
itself. In a sense the description is specific enough to
satisfy our curiosity. In (32), repeated here:

(32)a. The woman I saw collecting tickets at the

station this morning looked just like my
mother.

b. The woman I saw at the station this morning
looked just like my mother.

97

in (a), the description about the woman in question is
sufficient to bring about familiarity in that normally there
is exactly one person fitting the description. Although the
addressee may not be able to identify the woman, he knows
which one is being talked about. In contrast, in (b), the
description is not sufficient because it is highly possible
that there is more than one person fitting the description.
In cases like (32a) and (37) familiarity is brought

about by uniqueness.

In sum, familiarity can be achieved through uniqueness.
Specific descriptions of the whole_NP which refers uniquely
bring about familiarity in the sense that the addressee’s
curiosity is satisfied with the descriptions. A unique
explicit previous mention (e.g. abook and later the book)
or association with such a mention (e.g. abook and later
theaufluv) also brings familiarity by virtue of mentioning.

The familiarity requirement is met when the speaker
assumes that "bridging" or "accommodation" is likely to
occur on the part of the addressee. This assumption might
be based upon somewhat objective judgement of the
familiarity ofi»the event the speaker is referring to. For
example, a plumber’s fixing the speaker’s kitchen sink is
supposed to be more familiar to most people than the
speaker’s meeting a poet downtown. The assumption that
"bridging" is possible might also be based upon personal
relationship between speaker and addressee. For example, to

a friend who knows that he is a student and that it is

98
finals week, the speaker may well say, "I’m upset with the
test I took yesterday", but to somebody who knows neither,
he would probably say, "I’m upset with a test I took
yesterday".

At this point, let me clarify the relations among the
three requirements posited for the use of Mn. To use Mm,
first of all, the existence of the entity in question has to
be introduced into the discourse in some way. If what is
introduced is not unique, the second requirement is not met.
For example, in ”I met three persons. ?The person wore a
hat“, although the existence of the entity in question has
been introduced by an explicit previous mention, Mm is not
possible here because the entity in question is not unique
with respect to the discourse, the addressee does not know
which one out of the three is referred to.

If uniqueness is guaranteed, i.e. it is known to both
speaker and addressee, familiarity is achieved. If
uniqueness is known only to the speaker, but not to the
addressee, only when "bridging" or "accommodation" is

assumed to be possible can familiarity be achieved.

3.4 INDEPINITES

I have argued that to use Mm, three requirements, the
existence requirement, the uniqueness requirement, and the .
familiarity requirement, all have to be met. Further, if
all of them are met, Mw’has to be used. Hence the three
requirements posited here are necessary and sufficient

conditions for the use of the definite article. To meet the

99
existence requirement, the existence of the referent has to
be brought into the discourse by an explicit previous
mention, or association through such a mention, or shared
knowledge, or the immediate discourse context, or a
grammatical previous mention. To meet the uniqueness
requirement, the referent has to be unique in the sense that
in the relevant domain of interpretation, for count nouns
there is only one individual fitting the description or all
the individuals fitting the description are included: for
mass nouns the description must be specific enough to
individuate a natural portion of the mass (e.g. the waterin
theglass, theloveshefeltforhim). To meet the
familiarity requirement, the referent must be familiar to
the hearer in the sense described.

I now turn to the discussion of the use of a0» and the
zero article. It has been noted that if any of the three
requirements for the use of the is not met, the cannot be
used. This entails three things. First, the use of a or
the zero article occurs when the existence requirement is
not met, i.e. when the existence of the entity in question
has not been properly introduced into the discourse. For
example:

(60)a. I read a book today.

b. I had dinner with fi'iends last night.

c. I used to have tea after dinner.
Assuming that the addressee does not know anything about the
event that is going to be reported, the speaker uses an

indefinite to start a conversation.

100

Secondly, the use of a or the zero article occurs when
the uniqueness requirement is not met, i.e. when there is
at least one entity fitting the description excluded.

For example:
(61)a. I know a person who works in the Administration
Building.
b. I know people who work in the Administration
Building.
According to the theory developed in this chapter, in (61),
the‘existence of the entity in question, i.e. people who
work in the Administration Building, has been introduced by
the relative clause which serves as a grammatical previous
mention. Presumably there is more than one person working
in the Administration Building. If the speaker knows only
one person there, as in (61a), a is used: if he knows more
than one person, but not all the people there, as in (61b),
the zero article is used.

Thirdly, the use of a or the zero article occurs when
the familiarity requirement is not met, i.e. when the
speaker thinks that the entity in question might not be
familiar to the addressee. For example:

(62) Azoologist that I talked to this morning told me

that the monkey is man’s best friend.
(62) can be a case in which the speaker talked to only one
zoologist, but since it is quite unusual for a person to
talk to a zoologist, the speaker thinks that the addressee
might not be familiar with the event and uses a, instead of

due

101
Note that the difference between a and the zero

article is not that the latter is simply the plural
counterpart of the former. This is because a and the zero
article have their own meanings. An NP determined by a can
be specific or non-specific: whereas a plural NP determined
by the zero article can only be non-specific. Carlson
(1977) points out that a NP and ONPs differ in several ways.
In the following I will discuss two of the differences that
Carlson mentioned to make clear what is meant by "specific"
and "non-specific".

One difference that Carlson points out is the opacity

phenomena. A sentence like:
(63) Mary wishes to meet a violinist.

has two readings. On one reading, there is some particular
violinist that Mary has in mind, and she wishes to meet him.
This is the specific or transparent reading. On this
reading, the NP a violinist is specific, denoting a
particular violinist who is known to Mary, but not
to the addressee. On the other reading, the non-
specific or opaque reading, Hary’s desires are fulfilled by
meeting anyone, so long as that person is a violinist. On
this reading, the NP a violinist is non-specific, denoting
an unspecified violinist.

Carlson points out that this ambiguity between a
specific and non-specific reading disappears when the
singular NP a violinist turns into a plural determined by

the zero article, as shown in

102
(64) Mary wishes to meet violinists.

(64) has only the non-specific reading according to which
Hary’s desires are fulfilled by meeting more than one
person, so long as those persons are violinists. What is
absent is the specific reading according to which there is a
particular group of violinists that Mary has in mind, and
she wishes to meet them. We see that a plural NP determined
by the zero article like violinists can only be non-
specific.

Carlson points out that if we use other plural
determiners like many, all, three, and sm (reduced some),
we can have both the specific and the non-specific reading.

For example:
(65) Mary wishes to meet three violinists.

In (65) , three violinists can denote either three particular
violinists or any three unspecified violinists.

Notice that in the above examples we have an "opacity-
creating" verb, wish. This kind of verb, including want,
believe, and think, expresses our propositional attitudes
such as desire and belief. The non-specific readihg arises
in sentences with this kind of verb because the object of
our wish or belief does not have to exist in the actual
world. By contrast, in the following sentences

(66)a. Hary met a violinist.

b. Mary met sm violinists.
c. Hary met violinists.

103

we do not have an opacity-creating verb. All of the three
sentences entail that there was something that Mary met.
Nevertheless, the indefinites a violinist and sm violinists

have both the specific and the non-specific use whereas
violinists has only the non-specific use. In the non-

specific use, all of the three indefinites are attributive
in the sense that they are used to attribute the property of
being a violinist to the kind of person Mary met. And the
difference among the three sentences is only in the number
of violinists that Mary met. In (a), it is exactly one: in
(b) it is three or more: and in (c), it is more than one.
In the specific use, since the singular article a can mean
’a particular one’, a violinist can be used specifically to
denote a particular violinist. Since an can mean a certain
number of particular ones, sm violinists can be used
specifically to denote a particular group of violinists.
Yet the zero article does not have a meaning which
specifies a particular one or a particular group: it does
not have a specific use.

The non-specific use of a and an in non-opaque contexts
is most clearly seen in imperatives and questions. For
example:

(67)a. Do you have a pen?

b. Take sm apples.
In both (a) and (b) the natural reading for the indefinite
NP is the non-specific, not the specific.

Another difference between a NP and ONPs that Carlson

104
mentioned is found in cases like
(68)a. An accident happened today at 3, 4:30, and 6.
b. Accidents happened today at 3, 4:30, and 6.
(p. 421)
In (68a), the prominent reading for anaccident is the
specific one, i.e. there was one accident, the same one,
which happened three times: whereas in (68b) accidents is
non-specific, i.e. it was a different one (or a different
group of accidents) that happened at the three different
times. Hence (68a) is semantically odd while (68b) is not.
A L2 learner needs to know that an NP with a favors a wider
scope reading than a plural determined by the zero article,
so that he can avoid using sentences like (68a).
Concerning the choice between indefinite singulars and
plurals, there is another question often asked by L2
learners, i.e. "Do we need agreement in singularity and
plurality for related nouns?" This question is raised
because of instances like the following:
(69)a. *The student used the zero article for the
abstract noun "invasion” and "reality".
b. The student used the zero article for the
abstract nouns "invasion" and "reality".
(70)a. He has a good 'e.
b. Thqrhave good 'es.
c. They have a good life.
(69) indicates that agreement in plurality is obligatory,
but;(70) shows that it is optional. When we consider more

cases, we find that if there is a lack of agreement, there

is a special purpose. For example, a case like the

105
following indicates that to avoid ambiguity, agreement must
be avoided:

(71)a. I’ll show you rooms with aTVset.
b. I’ll show you rooms with TVsets.

(71a) clearly indicates that there are a number of rooms,
each with one TV set whereas (71b) is ambiguous between this
interpretation and another according to which each room has
more than one TV set. Similarly, in (72)-
(72)a. Cats have a tail.
b. Cats have tails.

the singular, but not the plural, tells that one cat has one
tail.

3.5 CONCLUDING REMARKS

It is hoped that when an L2 learner wants to check
whether he has made the right choice of articles, the
discussion above is of some help. When he does not know
whether to use Mm or not, he can ask, "Are all of the three
requirements, existence, uniqueness, and familiarity,
. fulfilled?" If any of them is not met, the is not to be
used. If all of them is met, dm’is the right choice. This
means that the three requirements together are necessary and
sufficient conditions for the use of Mm. This, however,
does not suggest that for any utterance in a given context,
only one choice of articles is possible.

It has been noted in section 3.3 that the judgement

whether the familiarity requirement is met can be different,

106 '
depending on whether or not the speaker thinks "bridging" or
"accommodation" is likely to occur on the part of the
addressee. Besides, the judgement whether the uniqueness
requirement is met can also be different in some cases. For
example:

(73)a. He provided us with the/a kind of information

that only insiders can.

b. Intellectuals should try to seek the truth lying

hidden behind the/a veil of distortion.
In (a), to some people, the information that only insiders
can provide constitutes only one kind: to others, it may be
viewed as constituting various kinds. Similarly, in (b),
to some people, there is only one kind of veil of
distortion, but to others, there can be more than one.

The use of articles with proper names has not been
discussed above because traditional grammars (e.g.
Christophersen, 1939, Jespersen, 1949, and Quirk, et al.
19732) have offered some useful rules. In fact some of
the uses of Mn can be treated as part of the name, as in
theHague and theMITPress. When a proper name loses its
property of referring uniquely, it is used as a common noun
and hence its use is governed by the same rules discussed in
this chapter.

One thing that might be confusing to L2 learners is that
when a proper name is used together with a common noun, the
whole NP still behaves like a common noun, and not a proper
noun. For example:

(74) the Watergate scandal: the Reagan administration:
the American dollar

107

We see that in the form "proper name + singular count noun",
du’is always used because the proper name makes the whole
NP definite. Because of the proper name, the addressee

knows what is referred to.

Chapter IV

The Generic Use of Articles

4 . 0 INTRODUCTION

In chapter two we discussed the mass/count distinction
and in chapter three we discussed another distinction, that
between definiteness and indefiniteness. It was concluded

that if we know these two distinctions, we can determine the

choice of articles as follows:

Count Mass
Definite the + N(s) the + N
Indefinite a(n) + N O + N
O + Ns

This chapter is concerned with what is traditionally
called "generic” uses of articles. A typical example of
'this kind of use of articles can be illustrated by the

following example:

(l)a. The dog is a mammal.
b. ,A dog is a mammal.
c. O dogs are mammals.
In (la-c) we have a generic use of the, a generic use
of a, and a generic use of O respectively. In (1) , all

the three sentences seem to express more or less the same

idea and the three subject NPs all seem to refer to more or

108

109
less the same thing, i.e. the species of dogs. This is
different from the non-generic uses of articles we
discussed before. In sentences like the following:
(2)a. I saw Mw*dog.
b. I saw a dog.
c. I saw O dogs.
the three object NPs refer to quite different things. In
(2a), the definite NP thedog refers to a particular dog
”familiar" to both speaker and addressee. In (2b), the
indefinite NP adog refers to a specific dog "unfamiliar" to
the addressee. In (2c), the indefinite plural dqg'refers
to some non-specific dogs.
Hass nouns, like count nouns, are also used to
generalize over a whole class of entities. For example:
(3)a. Water is essential to life.
b. Gold is precious.
c. Snow falls.
(3) shows that we use the zero article with a mass noun to
make a generalization over the class designated by the mass
noun. (1) and (3) are typical examples of generic uses of
articles. To generalize over a whole class of entities, for
count nouns, we use either a or Mm followed by a singular
form, or we use the zero article followed by a plural form,
as shown in (l): for mass nouns, we use the zero article.
Comparing these forms with the non-generic uses of .
articles as listed in the very beginning of this chapter, we
find two forms missing, i.e. "Mw'+ mass nouns" and

'Mw’+Ns'. Do we use Mw'with a mass noun in a general

110
statement? Yes. we have sentences like:
(4)a. Mary is a girl who loves the water.
b. Iron is good for'Mw’blood.
(Bolinger, 1975, p. 181)
Do we use dw'followed by a plural noun to refer to a
generality? Yes. We have sentences like:
(5)a. Most people enjoy the movies. (idem.)
b. The males are usually stronger. (p. 183)

We see that all the forms of the non-generic uses of
articles have their counterparts in generic uses. Is the
generic use of articles related to the non-generic use in
any way? How does the generic use of articles arise? Are
generic Mm; generic a and generic O always interchangeable
as in (1)? What are the differences among these three uses
of articles? These are the main questions to be addressed
in this chapter. If these questions can be properly
answered, it will be of help to L2 learners in their choice
of articles.

Section 4.1 discusses the generic use of O. Section
4.2 deals with the generic use of am» Section 4.3 is about

the generic use of a. Finally, section 4.4 is a conclusion.

4.1 THE GENERIC USE OF THE ZERO ARTICLE

Carlson’s work (1977, 1982) has made significant
contributions to our understanding of the semantics of what
he calls "bare plurals", i.e. plural NPs without any

determiner, such as books and cats. This section draws

111
heavily on his discussion of bare plurals and generic
sentences. It is hoped that this will lay a foundation for
my discussion of the generic use of the other two articles.
Carlson (1977) distinguished two major uses of bare
plurals. One is what is traditionally regarded as the
plural counterpart of a singular NP determined by the
singular indefinite article a(n), such as abook and acat.
This use of bare plurals is found in sentences like the
following:
(6)a. I had dinner with friends last night.
b. Students at Stanford University debated over a
course called Western Culture.
c. Chm were put to sleep and they died two or
three days later.
In (6), the underlined bare plurals all seem to have an
”existential” reading, i.e., they all have essentially the
force of awne. Traditionally, this kind of bare plural is
called the indefinite plural.
This use of bare plurals is different from the use
we find in sentences like
(7)a. Chm are mammals.
2. figs; l151:5]:wings.
In (7), the underlined bare plurals all seem to have a
"universal” reading, i.e., they all have essentially the force
{of a”. These bare plurals are traditionally grouped as
generic NPs.
Carlson argued that the indefinite plural use and the

generic use of bare plurals should receive a unified

112

analysis. He argued that indefinite plurals are not the
plural counterpart of an indefinite singular. Instead, they
act like a proper name of a "kind" of thing, just as generic
NPs do. Carlson claimed that a bare plural itself is not
ambiguous: the existential or universal reading of a bare
plural can always be attributed to some aspect of the
environment where it occurs. 'Consider:

(8)a. Dogs are barking in the backyard.

b. Dogs bark.

In (8a), we have an existential reading of the bare
plural, and in (8b), the bare plural is generic. In (8b)
the generic reading arises because the verb is in the
present simple tense, which suggests that barking is a
permanent or characteristic trait of the species of dogs.
On the other hand, in (8a), the verb is in the present
progressive form. The temporal aspect of the verb implies
that the sentence is not a statement about some permanent
property of the whole species and hence the existential
reading arises.

we see that verb tense or aspect determines whether a
bare plural is being used generically or not in cases like
(8). In other cases like:

(9)a. Owls are awake.

b. Owls are intelligent.

we have a clear intuition that (9b) is a general statement
about the species of owls whereas for (9a) we do not. The

generic or non-generic sense of the bare plural is brought

113
out by the respective adjectives. Awake denotes a fairly
temporary property and hence the existential reading arises:

whereas intelligent denotes a more permanent property and
hence the generic reading arises. The former kind of

adjective, including hungry, drunk, available, etc. , are
The latter, including tall,

called "states" by Carlson.

he calls ”properties". We see that

fat, clever, etc. ,
in cases like (9) , the nature of an adjective determines

whether or not a bare plural is being used generically.

Carlson mentioned two more classes of predicates which

can determine the generic or non-generic use of a bare

plural. Consider:
(10)a. Hens are female chickens.
b. Hens are in the backyard.

Predicate nominals such as female chickens in (10a) refer to

more permanent properties and hence select the generic

reading: whereas locative prepositional phrases such as in

the backyard in (10b) refer to temporary states and hence

select the non-generic reading.
The main point of the above discussion of Carlson’ s

1977 work is that the generic use of the zero article is
The two uses of the same

related to its non-generic use.
article are determined by the context where the bare plural
occurs. Genericness is not inherent in the article itself:
it is the predicate of a sentence that determines whether

the subject NP has a generic use or not.
Carlson (1982) elaborated his hypothesis that a bare

114
plural acts like a proper name of a "kind" of thing by
drawing parallels between habitual sentences about

individuals such as John and generic sentences about kinds

of things such as dogs. He pointed out that a sentence like

(11) is ambiguous.

(11) John walked to work.

Carlson noted that (11) has two readings. One is what he

calls ”episodic" (or "event”) reading, according to which

(11) is a report of what John did on a certain day at a

certain time. The other reading is a habitual reading,

according to which (11) is a report of a habit or a

customary activity of John's. Note that adding a phrase

like ”in those days" brings out this reading. Carlson said

that in the episodic reading, the predicate is attributed

not to the individual John, but rather to some "space-time

instance" of him. In contrast, in the habitual reading, the

predicate is attributed to the individual John, not to some

temporal part of his "space-time realization".

In Carlson's view, an individual is an abstraction of

all his spatio-temporal instances (or stages). People live

through numerous stages, undergoing different processes and

events. And yet on the other hand, they have more stable

properties obtained from some constant activities they do
such as their habits (John smokes: Mary writes with her left
hand), their characteristic appearance (Bill wears a beard),

and their place in the world (Tom is a store manager).

Episodic sentences are space-time sentences: they are about

llS
Habitual

some temporal realization of an individual.
sentences, however, are about the more stable individual

itself.
Carlson argued that there is a strong parallel between
habitual sentences and generic sentences. He noted that a

sentence like ( 12) exhibits the same episodic/non-episodic
ambiguity as in (11):

( 12) Dogs barked .
(12) is understood as a report

On the episodic reading,
of what some unspecified dogs did on a given day at a given

In this reading, the subject dogs is non-generic.

time.
On the non-episodic reading, we have to assume, for example,

a perspective in the future where the species of dogs no

longer exists and then (12) can be understood as a report of
This

a trait of a kind of thing in the past named "dog".

reading is a generic reading and in this reading, the

subject dogs is a generic NP.
It is noted above that in the habitual reading of (11) ,

the predicate is attributed to the individual John.

in the generic reading of (12), Carlson

similarly,
suggested, the predicate is attributed to a "kind", dogs.

In this sense, a kind of thing such as dogs and cats
By contrast, in

is also an abstract individual like John.
the episodic reading of (12) , the non-generic dogs refers to

some non-specific members of the species, not to the

abstract individual Dogs.
It has been noted above that Carlson argued that the

116

generic or non-generic use of NPs like dogs in (12) is

determined by context. In (12) , we have an ambiguous
context. The utterer of (12) can be either talking about an

event of barking, i.e. we can have an episodic reading of
the predicate, or he can be talking about a trait of dogs,

i.e. we can have a non-episodic reading of the predicate.

The generic or non-generic use of the zero article is

determined by the episodic or non-episodic reading of the

predicate.
To sum up Carlson’ s insights on generics, a generic

sentence with a bare plural as its subject attributes the

predicate to the "kind" of thing designated by the bare

plural. This generic use of a bare plural arises only
when the predicate of a sentence denotes a more permanent or

stable property, for example, a predicate in the simple

present tense, not the progressive: adjectives denoting

"properties” such as intelligent, not "states" such as

awake: predicate nominals, not locative prepositional

Without contexts such as these, bars plurals are to

phrases .
be interpreted as denoting some non-specific members of a

kind of thing, not the "kind".
In Carlson's theory, the notion "kind" plays an
important role.‘ A kind of thing is different from just a

set of things. Consider:

(13) a. Dogs are widespread.
1:. Dogs that wag their tails at people are

widespread.
c. *Dogs that are running in the backyard right now

are.widespread.

117
In (13a), we have the species of dogs: in (13b), it is not
the whole species of dogs, but a subset of the species, yet
it is a kind of its own because wagging tails at people is
a common characteristic of these dogs. In (13c) , however,

running in the backyard right now is a temporary state and
Hence

not a permanent property of the dogs in question.
these dogs only form a set of things, not a kind.

Widespread, a "class predicate" which attributes a

property to a kind, is thus disallowed.

To sum up this section, the form ONPs (the bare
A generic ONPs

plural) can be generic or non-generic.
denotes a certain kind of thing, e.g. generic dogs denotes

the kind, dog: whereas a non-generic ONPs denotes more than

one unspecified member of a certain kind of thing, e.g. non-

generic dogs denotes more than one dog which is ”unfamiliar"
to the addressee. Whether ONPs is used generically or non-
generically depends on context. Let us assume without

argument that ONPs is basically used non-generically. When

it occurs in a special context, the generic use arises. The

special contexts mentioned in this section include two
When

In both, the form occurs in subject position.

kinds.
the predicate of the sentence is a "class predicate" such as

widapread , numerous , and come in (difi'erent shapes or
.n'za), the subject ¢NPs is forced by the predicate to be
interpreted generically. For example, when we say that
horses are numerous, because of the semantics of numerous,

in uttering that sentence, we cannot be talking about some
Rather, we must be talking

specific or non-specific horses.

‘6

118
about horses as a kind and hence the generic use arises.
Secondly, the generic use of ONPs arises when the

predicate denotes a property which can be characteristic of

a whole class. Such predicates include "property"

adjectives like intelligent, as opposed to "state"
adjectives like awake, and predicate nominals as in "Dogs

are mammals", as opposed to locative prepositional phrases

as in "Dogs are inthe backyard". Predicates in the simple

present tense, as opposed to the progressive, can also be a
trigger of the generic use of ONPs. In brief, when what the
predicate denotes is not limited to a certain (period of)

time or space, the generic use arises. For example, when we

say that owls are awake, it is impossible that all owls,
including all the owls in the actual world and owls in any

possible world, are awake at the same time. In contrast,

when we say that owls are intelligent, the statement may not

be true, but it is possible that all owls, i.e. members of
the kind, owl, are intelligent. In other words, the

generic interpretation of owl: is available due to the

semantics of the predicate intelligent.
Up to this point we have been assuming that the non-

generic use of ¢NPs can occur in any context except for

those special contexts mentioned above where the non-generic
When

sense of ¢NPs disappears and the generic use arises.

¢NPs is forced to be interpreted generically by a predicate,
it lust denote a kind. If it does not, a semantic anomaly
If we

arises. This is why (13c) , repeated here, is bad.
substitute a "state" adjective like hungry for the "class"

119

predicate widespread, as shown in (13d) , the sentence is

well-formed .

(13)c. *Dogs that are running in the backyard right now

are widespread.
Dogs that are running in the backyard right now

are hungry .

d.

Here we see a distributional restriction for the generic use
of ONPs, i.e. when the predicate is a class predicate, the

subject of the form "ONPs which Y’s" must denote a kind,

not only a set of things.

4.2 THE GENERIC USE OF THE DEFINITE ARTICLE
It was noted above that in subject position, a generic

sense of the zero article arises when the predicate has a

“characteristic" (non-episodic) reading. In an unambiguous

context such as

(14) Horses work hard.

the verb is in simple present tense: we have only a

"characteristic" reading of the predicate, not an "event"

reading. In this context, horses is unambiguously generic.
How about the definite article the in this context?

-Consider:

(15) The horse works hard.

the context is unambiguous like that of (14) , but

In (15),
It can be either

the definite article here is ambiguous.

specific or generic. In its specific sense, the horse

120

refers to a particular horse mutually known to speaker and

 

addressee. In its generic sense, the home refers to the

species of horses.
In subject position, the generic sense of the arises in

the same environments as those where the zero article is

used generically. For example:

(16)a. The lion is awake.
b. The lion is courageous.

 

(17)a. The tiger is in the cage.
The tiger is a dangerous animal.

(18)a. The cow is giving milk.
The cow gives milk.

All the subject NPs in the (a) sentences are non-

generic: whereas in the (b) sentences, on the prominent

reading, all the subject NPs are generic. In (16b) we have a

”property" adjective, courageous, as opposed to a "state"

adjective, awake. In (17b) we have a predicate nominal, a

dangerousanimal, as opposed to a locative prepositional

phrase, inthecage. In (18b) we have the present simple

tense as opposed to the progressive. These are generic

environments .
Is there any difference in meaning between generic the

and generic at We noted in section 4.0 that both the
singular and the plural definite can be used generically.

Bolinger (1975) noted that there are two ways to refer to a

generality by using a plural noun. One is to use a bare

plural and the other is to use a definite plural. His

examples are (p. 181):

121

(19)a. Airlines charge too much.
b. The airlines charge too much.

Generals usually get their way.

(20)a.
The generals usually get their way.

Bolinger remarked that in (19a), airlines, being

indefinite, refers to all and any airlines: if it is an

airline, it will charge too much. This is similar to

Carlson's position. Airlines, a bare plural, refers to a

Any and all airlines will charge too much

kind of thing.
In (19b), theairlines, being

by virtue of being airlines.
definite, refers to "those actually in existence, out there

in the world, forming a subclass of common carriers" (p.
181). Similarly, in (20a), gaumdk refers to all generals,
extant and those yet to come: whereas in (20b) thegenerals
refers to those actually existing in the world, forming a
subclass of officers.

Bolinger's main point is that the definite article
Jhere is to ”single out (make definite) the thing mentioned

against the background of a more inclusive whole" (p. 181) .
Hence when we say "The males are usually stronger", we view
males as a subclass of a class that also includes females.

And when we say "Candy is not good for the teeth”, we refer

to the teeth of the human body. Mass nouns act like

plurals. When we say "Iron is good for the blood", we refer

also to the blood of the human body.
Bolinger said that this view of a larger whole as a

backdrop is also true of definite count singulars when used

to generalize. Consider:

122

(21)a. The hens lay eggs.
b. The hen lays eggs.
The larger whole here is all the domestic animals.
According. to Bolinger, the difference between (21a) and
(21b) is that the plural form'the hens refers to the
totality of a subclass of domestic animals and the singular
form thehen refers, not to the totality of the subclass,
but to a single item which is taken to represent the
subclass. In the generic sense, the hen refers to an
abstract typical hen, not a particular one.
I have found that the difference between generic "the
N” and "the Rs” is not as simple as Bolinger suggested.
Burton-Roberts (1976, p. 442) claimed that (22) does not have a

generic interpretation.
(22) The beavers build dams.

I think that a generic interpretation for ( 22) is hard to get
for most people because it takes a special context to be
interpreted generically. ( 22) needs to be viewed in a picture
where beavers' building dams is contrasted with other
rodents' peculiar activities, which are incidentally not easy

to think of. By contrast, in a sentence like
( 23) The cows give milk.

if it is viewed with a picture of a farm where the cows give
milk, the hens lay eggs, and the horses haul carts, etc. ,

:hen a generic interpretation for ( 23) arises. Note that for

 

123

singular definites, however, such a demand for contrasting

is not so strong. For example:

(24)a. The cow gives milk.
b. The beaver builds dams.
c. The lion is the king of beasts.

In (24) so long as the subject NPs denote a well-defined
species, they can be readily interpreted generically without

a strong demand for a contrast with other species within the

same larger class.
Vendler (1967) has a different way of saying almost the

same thing as Bolinger with respect to generic the. He

claimed that "the definite article-always presupposes a

restrictive clause” (p. 56). For example:

The man [that I saw] wore a hat.

(25)a. I saw a man.
I hate the man [who

A man keeps bothering me.
keeps bothering me].

The [animal that is a] tiger lives in the

jungle.
The Incas did not use Mw’[instrument that is
(ibid., pp. 56-7)

(26)a.

b.
a] wheel.

In (25) we have non-generic the to pick out a particular

individual: whereas in (26) we have generic the to single

out a representative of a class. Vendler argued that a

restrictive clause is a necessary condition for both uses

of the. In the generic use, this means that a larger class
is presupposed and the referent of the definite generic NP
is a subset of this larger class. The genus of tigers is a

subset of the genus of animals and so are wheels to

instruments .

 

 

 

124
According to Vendler, the generic use of the requires a
superior genus. This requirement is similar to Bolinger's
suggestion that generic the needs an appropriate backdrop, a
larger whole. Lacking such a background, a definite NP has
only a non-generic interpretation. For example:
(27)a. Objects are in space.
b. The object is in space.
(Vendler, 1967, p. 57)
(27a) is a generic sentence but (27b) is not. The object
in (21b) can only refer to a particular object, not a
representative of the genus of objects because it does not
fall under a superior genus.

A superior genus or a backdrop is a necessary condition
for the use of generic the. It explains the use of the in
cases like the following:

(28)a. This book is written for the [person who is a]

mathematician. (Vendler, 1967, p. 57)

b. There are two kinds of large cat living in
Paraguay, the [kind of large cat that is a]
jaguar and the [kind of large cat that is a]
puma. (p. 58)

However, this condition is not a sufficient one. It

does not rule out cases like:

(29)a. *Bill likes to drive the [vehicle which is a]

sports car when he gets a chance. (Lawler,
1973, p. 114)

b. *On Mother's Day, people wear the [flower which

is a] carnation in their lapel.

Why is generic the blocked in sentences like (29)? Both

sentences in (29) are habitual sentences. The referents of

 

125

both nouns, sportscar and carnation, fall under a superior

genus, vehicles and flowers, respectively. And intuitively,

we can say things like:

Bill likes to drive a kind of car called

'sports car' when he gets a chance.
On Mother's day, people wear a kind of flower

called 'carnation’ in their lapel.

(30)a.
b.

One thing clear is that in object position, generic use

of the does not require that it occur in a generic or

habitual sentence. For example, we have sentences like:

Han invented the wheel in protohistoric times.

(31)a.
b. Euclid described Mm parabola. (Vendler, 1967,
p. 58)
c. In a TV interview, Saul Bellow talked about
the novel .

In (31) , all the verbs are in the past tense and further we have

specific time adverbials, in protohistoric times in (a) and

inaTVinterview in (c). All these indicate that the

sentences here are not generic sentences. However, the use

of the in all three cases is generic because none of the

definite NPs refer to a particular entity: instead they

denote a ”kind”.
Bolinger (1975, p. 184) has the following to say about

the generality of a noun in object position:

It is a fact that when a noun is object of a verb
or is some other kind of complement, it is far
more often than not partitive rather than
general. In a sentence like He eats sweets or
Theyhaul coal it is pretty clear that he eats

only the sweets he eats, not sweets in general,
and they haul just what coal they can load on
their trucks, not coal in general.

126

Bolinger classified verbs like eat and haul as

"manipulative" verbs. They involve physical actions.

Bolinger said, "we can act only on so much of it as we can

reach or manipulate" (p. 184) . He noted, however, there are

some contexts in which a manipulative verb may be used in a

non-manipulative sense. For example:

(32)a. Why do you (always) spend money like that?
b. I saw him spending money.
(ibid.)

In (32a), the verb spend is used in a non-manipulative

sense: the sentence means 'Why are you so wasteful of

money?’ In (32b) , spend is used in a manipulative sense:

the use of the progressive indicates that the action is

manipulative.
In (29), repeated here:

(29)a. *Bill likes to drive the sports car when he gets

a chance. (generic the)
b. *On Mother's Day, people wear the carnation in

their lapel.

we have' two manipulative verbs, drive and wear. Drive and

wear involve physical actions: Bill drives only the sports

cars he can reach, not sports cars in general and people

wear only the carnations they wear, not carnations in

general.
On the other hand, in (31) , repeated here:
(31)a. Han invented the wheel in protohistoric times.

Euclid described the parabola.
In a TV interview, Saul Bellow talked about

the novel .

C.

127

we have verbs of a different nature. When we drive or wear
something, we actually do something to it, but when we
describe or talk about something, we do not. I think they
can be used in a non-manipulative sense. Saul Bellow could
talk about the kind, Novels, and Euclid could describe the
kind, Parabolas, but Bill cannot drive the kind, Sports
Cars, and people cannot wear the kind, Carnations. If we
turn the definite generics in (31) into bare plurals, we can
have a different meaning of each sentence.
(31')a. Han invented wheels in protohistoric times.
b. Euclid described parabolas.
c. In a TV interview, Saul Bellow talked about
novels.
(31a) talks about the invention of the wheel against other
human inventions: (31’a) does not have such an implication.
Further, in (31a) the wheel implies the kind, wheel,
whereas in (31'a) wheels implies various forms of wheels
such as cart wheels and wagon wheels. It is clear that
(31b) implies that Euclid described a kind of curve that is
a parabola: what he described was the kind of thing itself.
In (31’b), however, it may well be some non-specific
parabolas that he described. Similarly, in (31c) , the novel
implies novels as a literary genre whereas in (3l'c), nawfls
can refer to some non-specific novels that Saul Bellow
talked about.
The point here is that the manipulative sense of a verb

blocks the generic use of Mw'with its object. If a verb is

 

128

used in a non-manipulative sense, generic flw'is possible:
if it is used in a manipulative sense, generic Mw’is
impossible. For example:

(33)a. To hunt the elephant, you need special guns.

b. ?John is hunting the elephant.
(34)a. Their task is to eradicate the wolf.
b. ?John eradicates the wolf.
(Bolinger, 1975, p.182)

We can infer that (33a) is about elephants as a class
because the sentence means that to hunt a typical elephant
out there in the world, you need special guns. In contrast,
this generality cannot be inferred from (33b). It does not
have a reading according to which John is hunting a
typical elephant. Similarly, (34a) expresses an idea that
'individualizes the noun as representative of its class"
(Bolinger, 1975, p.182): whereas (34b) does not. Why so?

Note that the generality inference of (33a) and (34a)
is made possible by the infinitive. In this untensed form,
both verbs are used in a non-manipulative sense and hence
generic Mw’is allowed. By contrast, in (33b) we have the
progressive, and hence John must be hunting some specific or
non-specific elephants. In (34b) we have the simple present
tense, and as before, John eradicates only the wolves he can
manipulate, not wolves in general. In both cases, the verb
is used in a manipulative sense and hence generic Mw'is
disallowed.

Note also that the infinitives in (33a) and (34a) are

different from that in (29a), repeated here:

‘h

A."

 

129

(29)a. *Bill likes to drive Mw'sports car when he gets
a chance. (generic mm)
In (29a) the infinitive serves as the complement of the verb
flhs: in (33a) the infinitive can be interpreted as 'in
order to' and in (34a) the infinitive is a predicate
complement. Generality cannot be inferred in the former
case, but it can in the latter two cases. Although (29a) is
bad, (35) is not:
(35)a. To drive Mm sports car, you need special
skills.
b. Their task is to improve Mm sports car.
Besides the above two cases, generality can also be
inferred when an infinitive serves as the subject of a
sentence. For example:
(36) It is always helpful to ask.flw’librarian if
you have any question on reference books.
Generic dw'is possible in (36), but without a special
context, it is not possible in
(37) When you absolutely positively have to know, ask
Mw'librarian.
As a message from a public library printed on a bookmark, a
is a better choice than the in (37) . In (37) , the verb ask
occurs in a command. I think the use of a command is a
clear indication that the verb is used in a manipulative
sense.

In sum, the generic use of fin requires a superior

i...

 

130
genus or a larger whole for the NP at issue, but if the NP
is the object of a verb which is used in a manipulative
sense, generic Mn is disallowed. Manipulative verbs
involve physical actions. The contexts in which a
verb is used in a manipulative sense include the use of a
progressive, a command, and a simple present tense. On the
other hand, the indication of a non-manipulative sense of a
verb includes the use of infinitives which serve as a
a subject as in (36) or as a predicate complement as in
(35b) or are interpretable as 'in order to'.

The zero article, however, is allowed with either the
manipulative or the non-manipulative use of a verb. In the
cases of (33) and (34), bare plurals are allowed in all four
cases, as shown in the following:

(38)a. To hunt elephants, you need special guns.

b. John is hunting elephants.
(39)a. Their task is to eradicate wolves.
b. John eradicates wolves.

I now turn to another point about generic Mwu It is
frequently mentioned in the literature (e.g. Quirk, et al.,
1972, and Lawler, 1973) that definite generics, like
ordinary definite NPs, carry existential presuppositions.
It is true that a difference between the sentence Ekmns
workhard and the sentence Thehorse workhard is that the
indefinite generic horses refers to any and all horses or
the species of horses: by virtue of being horses they
(normally) work hard: whereas the definite generic the horse

refers to an item which is taken to be representative of the

 

 

ho:

of

pr!

V0.

in.

km

sp.

Dd'

ge:

ge;

in‘

an

13].
horses existing out there in the world, forming a subclass
of domestic animals. However, the existential
presupposition requirement does not rule out sentences like:

(40)a. The unicorn is a popular theme in children's

literature.
b. The unicorn cleans water with its horn.
c. The dragon is a symbol of nobleness in ancient
China.
Although unicorns and dragons do not exist in the real
world, they exist in a world of imagination. This world of
imagination is presupposed to be part of our world
knowledge. Since unicorns and dragons are well-defined
species in stories, they have a definite generic form like
natural species such as horses and snakes.

Generic Mm, nevertheless, does have a restriction that
generic 0 does not have. Carlson (1983, 1985) points out
that a non-natural kind term does not have a definite
generic form. For example:

(41)a. Dandelions are widespread.

b. The dandelion is widespread.
(42)a. Weeds are widespread.
b. *The weed is widespread.
Dandelions are a well-defined natural species, but weeds
involve a variety of species memberships: they themselves
are not a natural kind. Another example:
(43)a. Dogs that wag their tails at people are
widespread.
b. ?The dog that wags its tail at people is

widespread.

(44)a. Birds that eat fish are widespread.

ms

1“

132
b. ?The bird that eats fish is widespread.

Dogs and birds are natural kinds, but dogs that wag their
tails at people and birds that eat fish are not. It was
noted above that generic the presupposes existence: and
further the predicate nmhsnmmd applies only to ”kinds".
Given this, the (b) sentences imply that dogs that wag
their tails and birds that eat fish are two well-

defined natural species already existing out there in the
world. But they are not. They are kinds set up by the
description of the sentence. And in fact we can have
sentences like:

(45)a. The dog that wags its tail at people is

generally friendly.
b. The bird that eats fish generally has a long
beak.
We can create kinds of things by description and attribute
some characteristic properties to the kinds that we create.

In brief, generic Mw’is blocked in a sentence with a
"kind" predicate such as rare, common,and widespread that
applies only to kinds, if the subject term does not
designate a natural kind. Generic 0, however, is not
subject to this restriction.

To sum up this section, like ONPs (the bare plural),
both the form.'Mw N" and "M2 Ns" can be either generic or
non-generic. Non-generic "fiwrN” denotes a particular
entity known both to speaker and addressee. Non-generic
'Mw'Ns” denotes inclusively all the entities in the

relevant domain of interpretation. On the other hand,

133
generic "dw'N (a count noun)" denotes an item which is
taken as representative of a subclass of a larger class.
Generic "Mw'Ns' or generic "Mn N (a mass noun)" denotes
the totality of a subclass of a larger class, and usually
demands a contrast with other subclasses of the same larger
class.

Let us assume again that the non-generic use is the
basic and the generic use arises only in special contexts.
In this section, three environments were mentioned where the
generic use of "Mw'N" arises. First, like the bare
plural, when it co-occurs with a class predicate, the generic
use arises. In this environment, the noun must be a natural
kind term. For example, we say "The ladybug is common”, but
we do not say ”The bug is common" because ladybug is a
natural kind term, but.bug is not.

Secondly, the generic use of "flw’N” arises when it
co-occurs with a predicate which denotes a property
distinctive of the whole class designated by the noun. If
the property is not distinctive of the whole class, ”the N"
can only be interpreted non-generically. For example, when
we say ”The cow gives milk", it is possible that we are
talking about a particular cow, and it is also possible that
we are talking about the species of cows. In contrast, when
we say “The cow eats hay", the sentence can only be
interpreted non-generically, i.e. a particular cow known
both to speaker and addressee eats hay. The generic
interpretation is not available because eating hay is not a

distinctive property of all cows.

134

In the two environments above, "Mw’N" occurs in
subject position. The generic use of ”Mw’N" also occurs in
object position. When a verb is used in a non-manipulative
sense, its direct object, 'Mw’N', can be interpreted
generically. For example, when we say "In a TV interview,
Saul Bellow talked about the novel", it is possible that we
are referring to a particular novel known both to speaker
and addressee. And it is also possible that we are
referring to novels in general. On the other hand, if a
verb is manipulative, its object does not have a generic
interpretation because normally we cannot manipulate a whole
class of thing. Thus we do not say sentences like ”People
wear the carnation on Mother’s Day". And when we say "John
likes to drive the sports car", the sentence can only mean
that John likes to drive a particular sports car known to
both speaker and addressee. The generic interpretation of
thesportscar is not available in that sentence.

The plural definite, ”Mw'Ns", is used non-generically
in most cases. Its generic use arises in the same
environments as that of the singular definite, Mw'N. Both
the plural and the singular definite, when used generically,
require a higher genus as a backdrop. However, when we say,
for example, "The cows give milk", it needs to be viewed as
a contrast to "The hens lay eggs: the horses haul carts,
etc." Without this contrast, the generic sense of the
plural definite can hardly arise. Therefore, L2 learners
have to be especially careful when they use a plural
definite generically.

135

4.3 THE GENERIC USE OF THE INDEFINITE ARTICLE

It was noted above that a generic sense of the zero
article or the definite article arises if it occurs with a
subject term and the predicate attributes a characteristic
property to the subject. In object position, a generic
sense of the definite article arises if the verb is used in
a non-manipulative sense. Some tested cases are:

(46)a. Horses work hard.

b. The horse works hard.

(47) To hunt the elephant, you need special guns.

How about the indefinite article 000? When does a
generic sense of this article arise? Consider:

(48)a. A horse works hard.

b. To hunt an elephant, you need special guns.
What interpretations does (48a) have? Is an elephant in
(48b) generic?

Burton-Roberts (1976) argued that in subject position a
generic NP determined by a is derived from a subjectless
predicate. Thus he claimed that (49a) is derived from
(49b):

(49)a. A whale is a mammal.

b. To be a whale is to be a mammal. (p. 430)
He claimed that generic NPs determined by a are like
predicate nominals (e.g. John is ateacher) in that

they both are non-referring: they both represent abstract

concepts, not objects. Hence generic a is about what

136
constitutes membership in a class, not about the class
itself.

This point can be further illustrated by an example

provided by Dahl (1975, p. 108):

(50) Amember of this club does not drink whisky:
hence, since you will now be accepted as a
member, you will have to stop drinking.

Using a, the first clause of the above sentence states that
"there is an obligation for members of the club not to drink
whisky, or at least that it is expected of them that they
will not drink whisky” (p. 108). Given this interpretation,
(50) is a valid argument. Note that (50) can be

paraphrased as:

(50') To be a member of this club is to not drink

whisky: hence...

Lawler (1973) noted that generic a is most natural

in definitional sentences. He gave us examples like:

(51)a. A madrigal is polyphonic.
b. *A madrigal is popular. (generic a)
(p. 112)

To be a madrigal is necessarily to be polyphonic, but to be
a madrigal is not necessarily to be popular. (51b) implies
that popularity naturally comes to a madrigal just because
of its class membership. Nunberg and Pan (1975, p. 415)
noted that although (51b) is not acceptable, (52) is not

problematic:

(52) A football hero is popular.

137

If a person is a football hero, he is popular. Popularity
is a natural result of being a hero.

From the discussion above we see that it seems to be
generally agreed that generic a is about class membership.
This sense of the indefinite singular article arises when it
occurs with a subject term, and the predicate attributes to
this subject a property which is a natural result of class
membership. Sentences with this use of a state that the
property denoted by the predicate is an automatic and hence
necessary property of the subject by virtue of its being a
member of the class designated by the subject term.

With this in mind, let us return to (48a), repeated

here:
(48a) A horse works hard.

Like the definite article in this context, a is ambiguous.
It can be either specific or generic. On the specific
reading, (48a) states that a particular horse known to the
speaker but not to the addressee works hard. On the
generic reading, it states that working hard is a necessary
property of being a member of the species of horses.

Now I turn to (48b), repeated here:
(48b) To hunt an elephant, you need special guns.

What is the status of a in object position? It was noted in
the preceding section that generic flm'is blocked with

manipulative verbs. We discussed cases like (29), repeated

138
here:
(29)a. *Bill likes to drive Mw'sports car when he gets
a chance. (generic Mm)
b. *On Mother's Day, people wear Mm carnation in
their lapel.
Although generichw is not allowed, we can have a in these
cases, as shown in (29'):
(29’)a. Bill likes to drive a sports car when he gets
a chance.
b. On Mother's Day, people wear a carnation in
their lapel.
In cases like these, a is clearly not generic: rather, in
(29'a) it is either specific or non-specific and in (29’b) it
is non-specific. In the previous section, we also noted that
generic Mw’is possible if a verb is used in a non-
manipulative sense. We discussed cases like (31), repeated
here:
(31)a. Han invented Mw'wheel in protohistoric times.
b. Euclid described.dw'parabola.

c. In a TV interview, Saul Bellow talked about
Mw’novel.

If we turn Mw'into a, we have:

(53)a. ?Han invented a wheel in protohistoric times.
b. Euclid described a parabola.
c. In a TV interview, Saul Bellow talked about
a novel.
In these cases, a is specific. (53a) states that man
invented a specific wheel in protohistoric times: (53b)
states that Euclid described a particular parabola: and

(53c) states that Saul Bellow talked about a particular

139
novel in a TV interview. Hence we see that as the direct
object of a verb, "000 N“ cannot be generic. Instead, it
is either specific or non-specific.
So far we have discussed the genericness of "fiwrNT,
"000 N”, and ONPs as subject NPs and as direct objects of
verbs. Now I turn to a discussion of these NPs as the
object of a preposition. In a general statement such as:
(54) Yet until now much of the work in this field has
not been easily accessible to the student, and
often written at an intimidating level of

technicality. -
(Cambridge Textbooks in Linguistics,

Pragmatics, 1983, back cover)
we can substitute students for thestudent, but astudent
would not as good as the other two forms.
This does not mean that only "Mw*N" and ONPs, but not
"000 N", can be used when they occur as the object of a
preposition. Consider:
(55)a. Defoe was an important figure in the development
of the novel.
b. *Defoe was an important figure in the development
of novels.
(Allan, 1986, Vol.2, pp. 139-140)
(56)a. This book is suitable for the first grader.
b. This book is suitable for first graders.
c. This book is suitable for a first grader.
(Abbott, p.c.)
(55) shows that ONPs sometimes cannot occur in this position,
and (56) shows that in this position, "000 N" sometimes can
readily occur, just like the other two forms.

Recall that the bare plural has two uses, the generic

and the indefinite plural use. In the generic use the bare

140
plural denotes a kind of thing and in the indefinite plural
use, it denotes an unspecified number of objects designated
by the noun. It was also noted that it is the environment
where the bare plural occurs that determines when the
generic or the indefinite plural use arises. In (55b) the _
generic sense of novels does not arise and hence it is bad.
(55) is a context in which the "kind", and not members of
this kind of thing, is at issue. It requires an NP which
denotes a kind. A.generic sense of "flw’N" in this position
is always available whenever the context requires it: but
this is not true with the bare plural, as (55b) clearly
indicates.

Given the above claim, how can we explain (56b)? If I
say that the bare plural in (56b) is not generic, the reader
might object to it because it obviously denotes first
graders in general. I'd argue that the indefinite plural
use of firstgraders here denotes an unspecified number of
first graders and this number is so big as to amount to the
whole class of first graders.

How about the status of "000 N" in this position? If
we define generic "aha N” as denoting class membership,

a firstgmder in (56c) is clearly not generic. Instead, it
is non-specific, denoting an unspecified first grader and in
this particular context, it amounts to any first grader.

In sum, as the object of a preposition, "Mw’N" can be
generic when the context requires it whereas the bare plural
or ”hvn N" cannot. However, the bare plural can be

interpreted as denoting an unspecified number of objects and

141
this number can be so big as to amount to the whole class
and thus it achieves a general use. Similarly, "000 N” can
be non-specific and by denoting an unspecified object it can
be taken as meaning 'any N' and thus achieves a general use.

Although generic a has a quite restricted distribution,

non-specific a is widely used in general statements where
generic Mw’would not be as appropriate. For example:

(57) .A student who has been on an F-l visa for eight
consecutive years must apply for an extension of
stay. (News and Notes 16:1, 1987, HSU)

The use of astudent here is better than the student because
applying for an extension of stay requires individual
actions: the use of generic Mm (meaning 'the kind of
student') implies that this kind of student as a group must
do something. Again, a is paraphrasable as any in (57) .

To sum up this section, the form «an» N" can be used
either generically or non-generically. Non-generic an»
N can be either specific or non-specific. Specific 000 N
denotes a particular entity known to the speaker but not to
the addressee: non-specific 000 N denotes an unspecified
entity. Generic 000 N denotes membership in a class. The
generic use of 000 N arises only in subject position. We
have a generic use of 000 N when it occurs with a predicate
which denotes a necessary property of the class designated
by the noun. Thus it is most appropriate in a definitional

sentence.

142
4.4 CONCLUSION
When an NP determined by the zero article or Mw’or a
is used generically, it does not refer to a specific or non-
specific item: rather it denotes something more or less on
an abstract level, a ”kind" of thing, or an item which is
taken to be representative of a class, or class membership.
The three articles are interchangeable in cases like (1),
repeated here:
(l)a. The dog is a mammal.
b. .A dog is a mammal.
c. 0 dogs are mammals.
However, they have their own distributional restrictions.
Among the three articles, generic use of the zero article
has the least restrictions.
When the predicate of a generic sentence is a "class"
predicate such as rare, common, widespread, and extinct
the zero article is allowed, but the article a is completely
blocked because a generic NP determined by a does not refer
a class itself. With a "class" predicate, the definite
article is possible for an NP that designates a "natural”
kind, but impossible for a kind that involves a variety of
species memberships (e.g. weeds, bugs) or a kind that is
created by the description of the sentence itself. Thus we
have:
(58)a. Dogs are common.
b. *A dog is common.
c. The dog is common.
d. *The dog that bites people is common.

(59)a. Dandelions are common.

143
b. The dandelion is common.
c. Weeds are common.
d. *The weed is common.

In generalizing over a kind of thing, bare plurals
allow the predication of properties which hold only on a
statistical basis, but for definite generics the properties
attributed to them must be distinctive of the whole class.

Thus we have:

(60)a. Cows give milk.
b. The cow gives milk.

(61)a. Cows eat hay.

b. ?The cow eats hay.

In object position, bare plurals or indefinite
singulars are allowed if a verb is used in a manipulative
sense, but definite generics are blocked. On the contrary,
if a verb is used in a non-manipulative sense, dw'is the

most appropriate. Thus we have:

(62)a. Bill likes to drive {a sports car, sports cars,
*the (generic) sports car} whenever he gets a
chance.

b. Euclid described {the parabola, ?a parabola,
?parabolas}.

Generic a is limited to subject position, but non-
specific a, joining generic Mw’and 0, is widely used in

general statements. Thus we have:

(63) There are sensible exercises to most chapters,
and adequate references for {the reader,
readers, a reader} who want(s) to go deeper.

(Cambridge Textbooks in Linguistics,
LogicinLinguistics, reprinted 1986,

back cover)

144

If L2 learners know the general properties of generic
sentences and the subtle differences among generic NPs
determined by'qu a and the zero article, they will have
better guidance in making choices of articles. If they know
the distributional restrictions for each article, they can

avoid a wrong choice.

Chapter V

Conclusion

Krashen (1982) claims that L2 learners can make use of
their knowledge of rules for language use only when they
have time to think about the rules and only when they focus
on form. This study has aimed to present the English
article system in such a way that when L2 learners want to
check whether they have made the right choice of English
articles, they can follow step by step the principles
presented in this dissertation. The use of articles may not
be categorical because native speakers of English may not
agree with one another in all cases. The purpose of
this study has been to predict the most appropriate use of
articles, that is, what educated American people will use in
a non-casual style in a given context.

The procedures of checking suggested in this
dissertation are as follows. First of all, ask if the noun
in question is to be used as a count or a mass noun.
Secondly, ask if the given noun is to be used generically or
not. If not, ask whether the given noun is definite or
indefinite. If yes, see what the grammatical function
(subject or object) of the noun is and ask what meaning is
to be expressed.

Thus the overall picture of the system is:

145

146

Count Mass
Definite the + N(s) the + N
Indefinite a + N O + N
¢ + Na

Each of the above forms has a generic use also. The
mass/count distinction is still valid in the generic use of
nouns, but the definite/indefinite distinction disappears.
In this system, overall, there are three questions to be
answered: a) count or non-count? b) definite or indefinite?
and c) generic or non-generic?

In chapter two we discussed the mass/count distinction.
According to the findings of Master's (1987) study, this
distinction "causes the most persistent difficulty in article
acquisition” (p. 181). In Master’s study, he analyzed the
spoken English of speakers of five different native
languages, three with no article system, [-Art], (Chinese,
Japanese, and Russian) and two with article systems, [+Art],
(Spanish and German). He found that "acquisition of a and
the [icount] feature takes place more slowly for the [-Art]
than the [+Art] group" (p. 91). Why this distinction is
difficult to acquire can be explained by the hypothesis that
being count or mass is part of the meaning of a word, as
discussed in chapter two. The countness of nouns in some
cases is arbitrary and thus has to be learned word by word.
Besides arbitrariness, the complications caused by the fact
that the mass/count distinction is not a simple binary one

also contribute to the difficulty. L2 learners cannot learn

147
a fixed feature, [+count] or {-count} for each noun.
Despite these disadvantages, I offered three
assumptions and four conversion principles in the hope that
they will be of some help to L2 learners in dealing with the

mass/count distinction. The three assumptions are:

Assumption 1: Nouns which denote discrete objects are
all count.

Assumption 2: Nouns which denote undifferentiated
substances are all non-count.

Assumption 3: Nouns which denote abstract entities are
all non-count.

And the four conversion principles are:

Principle 1:

Mass nJ Count: If a noun is used to denote a
discrete entity, it is count: if it is
used to denote the material content or
a particular quality of the discrete
entity, it is non-count.

Principle 2:

Count --> Mass: When we are talking about the
undifferentiated mass of a physical
object, but not the discrete object
itself, the count noun denoting that
physical object should be converted
into a mass noun.

Principle 3:

Mass -- > Count: In commercial contexts, a mass noun,
through constant application of the
noun to well-divided instances of the
referent, can gain an individuation for
its reference and thus can be used as
a count noun.

148

Principle 4:

Hass -- > ?Count: Abstract non-count nouns such as a
derived nominal or a name of a
subject for study, when modified by
a restrictive modifier, admit any,
which is the equivalent of akind of

' in this context.
The four conversion principles are to deal with the
complications that the mass/count distinction is not a
simple binary one. The three assumptions serve as a
starting point in conquering the arbitrariness of this
distinction. These assumptions stand as valid with some

exceptions. The overall picture of this distinction is

presented as:

 

discrete m s
[+count] [-count]
exce tions: exce tions:
a) bums, etc. nke, etc.

b) caiwunuh‘whkh was
not surprising, etc.

o) discussions, etc.

Chapter three provides answers to (in)definiteness. To
use as» three requirements all have to be met. Details for

the fulfillment of each requirement are given as follows.

Requirement 1: Existence

a) The existence of the entity denoted by the noun is

149

introduced into the discourse by i) an explicit
previous mention or association with such a mention,
or ii) by the immediate situation or an immediate
linguistic follow-up, or iii) by shared knowledge
between speaker and addressee.

b) The noun has a grammatical previous mention.

1)

ii)

(1)

(2)

(3)

(4)

The noun is followed by a restrictive relative
clause or another form of restrictive modifier,
including infinitives, present participles, past
participles, and prepositional phrases which are
not headed by of.

Exception: If we have the form "x which...NEG...Y",
i.e. if the relative clause contains a
negative word, it cannot serve as a
grammatical previous mention.

The noun is followed by an of-phrase, i.e. we
have the form ”N of NP". The N and the NP must
stand in one of the three relations: thematic,
appositive and possessive.

flhmmnkvdhdmn:

the significance of intuitions < intuitions are
significant

N of NP < NP be adjective
(N is a derived nominal of the adjective.)

the interpretation of such phrases < we
(people) interpret such phrases

N of NP < Verb NP
(N is a derived nominal of the transitive verb.)

the failure of presuppositions < presuppositions
fail

N of NP < NP Verb
(N is a derived nominal of the intransitive verb.)

zhnmmhbendadmn:
the subject of denoting < denoting is a subject

N of NP < NP be N

150

ltnmsmhendawmn:

(5) the course of a conversation < a conversation
has its own course

N of NP < NP has its own N

It is suggested that to check.whether the use of dw’is
possible, as a first step, the learner see if the existence
requirement is met. If either one of the (a-b) situations
occurs, this requirement is fulfilled. Then the learner can
go on to consider the second requirement. If the existence
requirement is not met, the use of dw’is inappropriate.

The second requirement is uniqueness. The details for

its fulfillment is given as follows.

Requirement 2: Uniqueness

The noun refers inclusively to all the entit(ies)
that there is/are in the discourse context.

a) Uniqueness is guaranteed by shared knowledge.

b) In the form "N of NP", if N is an abstract mass
noun which cannot go with any kind of measuring
word, the uniqueness requirement is always met.

Exception:

If N is an action noun, the demand for using the
is weakened.

c) In the form "N which...", if N is non-count, and the
restrictive relative clause specifies a specific
amount or a specific kind of the entity denoted by
the N, the uniqueness requirement is met.

Note: If uniqueness is known only to the speaker, but
not to the addressee, we say the uniqueness
requirement is met, but it is subject to the
screening of the familiarity requirement.

If any of the situations in (a-c) occurs, the second

151
requirement for the use of Mw’is met. After this, the
learner can go on to consider the last requirement,
familiarity. The details of its fulfillment are given as

follows:

Requirement 3: Familiarity

a) If any of the (a-c) situations listed under
Requirement 2 occurs, the entity denoted by the noun
is familiar.

b) If uniqueness is known only to the speaker,
but "accommodation" is possible, the entity denoted
by the noun is familiar.

In chapter four we discussed genericness. The
distributions and restrictions of the generic use of each
non-generic form is as follows:

Generic use of "Mw’Ns": refers to the totality of a,
subclass of a larger class:
must be viewed as a contrast
with other subclasses of the

same larger class.

Generic use of ”Mw'N": is taken as representative of
a subclass of a larger class.

a) Mw’N(s) + class predicate
(N must be a natural kind term.)

b) the N (s) + predicate
(The property denoted by the predicate must be
distinctive of the whole class designated by N.)

c) verb + the N(s)
(The verb must be non-manipulative.)

d) preposition + the N(s)
(The sentence must be a general statement.)

Generic use of ONPs (the bare plural):

Generic use of ON (mass nouns):

152

a) ONPs/ON + class predicate
(0NPs must denote a kind, not only a set of things.)

b) ONPs/ON + predicate
(The property denoted by the predicate can be
statistically true (say 80%) of the reference of N.)
Generic use of ”000 N":
a) a(n) N + predicate
(The predicate must denote a defining property of the
class designated by N.)
Besides the generic use, ONPs and "000 N" have general use,
which is something between the generic and the non-generic use.
In this use both ONPs and "000 N" is non-specific. They
occur in general statements. ”000 N" is paraphrasable as
”any N" in this general use. ONPs denotes an unspecified
number of members in the class designated by N.

The English article system is a complicated one. Few
non-native speakers have a complete acquisition of this
system. The mass/count distinction is a tough one because
for every single use of a noun the learner has to know
whether it is count or non-count. If this first step goes
wrong, the choice of articles will not be correct.

Although chapter two provides an overall picture of this
distinction, the exceptions listed there still remain a big
burden for the learner. If rules can be found as to what
kind of abstract entity counts as discrete and hence
countable, the exceptions can be greatly reduced.

There might be pitfalls in the application of the rules
for the use of dw*provided in chapter three. Let us take

the following text as an example:

153

(6) The following 20 questions contain two sentences
each. The only difference in the two sentences is
in the use of articles. For each pair of
sentences, assume that there is no previous
mention of the item in question.

The question here is why the underlined noun disallows dw’
to go with it. The underlined word.ummmas seems to have
been explicitly mentioned twice in the preceding sentences.
Why shouldn't it be definite? The tricky thing here is that
although the entity has been introduced, it is not unique.
We have a pair of sentences for each of the 20 questions.
When we talk about each pair of sentences, the sentences in
each pair remain indefinite because we still do not know
which two out of the forty are referred to. This is similar
to a case like

(7) I drank 20 cups of coffee yesterday. Each cup of
coffee cost 50 cents.

The underlined word adfiw remains indefinite although it
seems to have been mentioned.

The choice of articles in a longer text other than
isolated or discourse-initial sentences might involve more
complicated factors than those considered in this study. Du
Bois (1980) mentioned that in story telling, there are two
modes, descriptive and narrative. In the descriptive mode
the speaker focuses on introducing salient objects of
the story and thus initial mentions are indefinite. On the
other hand, in the narrative mode, the speaker focuses on

advancing the story line, and hence he usually fails to make

154
indefinite introductions for objects which have not been
introduced. In this case, we will have the definite article
for initial mentions.

The rules for article usage summarized in this
conclusion may look long and complicated. Nevertheless,
they present English article uses as a system, and not a set
of unrelated rules. The findings of this study clarify
several vague points made in popular ESL textbooks on the
grammar of English articles. For example, it was noted in
chapter three that in Frank (1972, p. 160), an exercise was
given for the use of a with non-count nouns. The rule was
stated in this way:

In some sentences, noncountable abstract nouns

with adjective modifiers may be used with a. In
many such sentences a is the equivalent of athd

qfl
With such a rule, the learner would not know exactly when a
can be used with a non-count abstract noun. Take Frank’s
first sentence in this exercise as an example.

(8) He has simplicity which is seldom met with

these days. (p. 160)
According to Frank's rule, the learner can only try to put a
kind of in the blank and see if the sentence is good. It
has been pointed out that this is not a reliable test.
In the system presented in this study, the learner has two
clues for the choice of articles for (8). First, snuwkhy
is a derived nominal (simple, simplify, simplicity). Our

fourth conversion principle says that it admits a in this

155
context. Secondly, we have a negative word, nmfiwn, in the
relative clause here. Our existence requirement tells us
that it is not possible to use Mw’in this context. Thus
the use of a is confirmed.

Brodkey (1969) and Master (1987) both reported that the
results of their experiments suggested that "systematic
instruction can foster accelerated learning of the article
system” (Master, 1987, p. 187). Master emphasized
repeatedly that ”the articles must be taught as a system
over a considerable period of time" (p. 188). It is hoped
that this study can bring about a better understanding of
the English article system from L2 learners’ point of view
and its findings can be used to construct carefully
sequenced lessons so that the learner can have a better

grasp of the whole system.

BIBLIOGRAPHY

BIBLIOGRAPHY

Agnihotri, Rama K.: A. L. Khanna: and Aditi Mukherjee.
1984. The use of articles in Indian English: Errors
and pedagogical implications. International Review of
Applied Linguistics in Language Teaching 22.115-29.

Allan, Keith. 1980. Nouns and countability. Language
56.541-67.

----- . 1986. Linguistic meaning, Vol. 1. London:
Routledge 8 Kegan Paul.

Bach, Emmon. 1986. The algebra of events. Linguistics and
Philosophy 9.5-16.

Bickerton, Derek. 1981. Roots of language. Ann Arbor,
Michigan: Karoma.

Bolinger, Dwight. 1975. A postscript to Poston on the
article. Modern Language Journal 59.181-85.

Brodkey, Dean G. 1969. A self-instructional program in
English article usage for Chinese, Japanese, Korean,
and Thai college students. PhD dissertation.
University of California, Los Angeles.

Burton-Roberts, Noel. 1976. On the generic indefinite
articles. Language 52.427-48.

Carlson, Greg N. 1977. A unified analysis of the English
bare plural. Linguistics and Philosophy 1.413-57.

----- . 1982. Generic terms and generic sentences. Journal
of Philosophical Logic 11.145-81.

----- . 1983. Natural kinds and the denotations of common
nouns. Unpublished draft.

----- . 1985. werlds and natural kinds. Unpublished draft.

Chen, Pi-fen, L. 1985. Determiners and relative clauses in
English. Unpublished master thesis. Michigan State
University, East Lansing.

Christophersen, Paul. 1939. The articles: A study of

156

157

their theory and use in English. Copenhagen: Einar
Munksgaard.

Clark, Herbert H. 1977. Bridging. Thinking: Readings in
cognitive science, ed. by P. N. Johnson-Laird and P. C.
wason, 411-20. Cambridge: Cambridge University Press.

Dahl, Osten. 1975. On generics. Formal semantics of
natural language, ed. by Edward L. Keenan, 99-111.
Cambridge: Cambridge University Press.

Du Bois, John W. 1980. Beyond definiteness: The trace of
identity in discourse. The pear stories: Cognitive,
cultural, and linguistic aspects of narrative
production, ed. by Wallace L. Chafe, 203-74. Norwood,
New Jersey: ABLEX.

Frank, Marcella. 1972. Articles. Modern English:
Exercises for non-native speakers, Part I: parts of
speech, 135-61. Englewood Cliffs, New Jersey:
Prentice-Hall.

Gass, Susan. 1983. The development of L2 intuitions.
TESOL Quarterly 17.273-91.

Gleason, H. A. 1965. Linguistics and English grammar. New
York: Holt, Rinehart and Winston, Inc.

Hawkins, J. 1978. Definiteness and indefiniteness: A
study in reference and grammaticality prediction.
London: Croom Helm.

Heim, Irene R. 1982. The semantics of definite and
indefinite noun phrases. Ph.D. dissertation,
University of Massachusetts at Amherst.

Jespersen, Otto. 1924. The philosophy of grammar. New
York: Henry Holt and Company.

----- . 1949. A modern English grammar on historical
principles, Part VII. Copenhagen: Ejnar Munksgaard.

Kadmon, Nirit. 1987. On unique and non-unique reference
and asymmetric quantification. PhD dissertation,
University of Massachusetts at Amherst.

Kaluza, Henryk. 1981. The use of articles in contemporary
English. Heidelberg: Julius Groos Verlag.

Karttunen, Lauri. 1969. What makes definite noun phrases
definite? The RAND Corporation.

----- . 1976. Discourse referents. Syntax and
semantics 7: Notes from the linguistic underground,
ed. by James D. McCawley, 363-85. New York: Academic

158
Press.

Kharma, Nayef. 1981. Analysis of the errors committed by
Arab university students in the use of the English
definite/indefinite articles. IRAL 14.333-45.

Krashen, Stephen D. 1982. Principles and practice in
second language acquisition. New York: Pergamon
Press. -

Lacey, Andrew. 1977. Rules in the teaching of the English
articles. English Language Teaching Journal 32.33-37.

Lawler, John M. 1973. Studies in English Generics.
University of Michigan Papers in Linguistics, Vol. 1:
No. 1.

Levinson, Stephen C. 1983. Pragmatics. Cambridge:
Cambridge University Press.

Lewis, David. 1979. Scorekeeping in a language game.
Formal semantics of natural language, ed. by Rainer
Bauerle, Urs Egli 8 Arnim bon Stechow, 172-87. Berlin:
Springer.

Markman, Ellen M. 1985. Why superordinate category terms
can be mass nouns. Cognition 19.31-52.

Master, Peter A. 1987. A cross-linguistic interlanguage
analysis of the acquisition of the English article
system. PhD dissertation. University of California,
Los Angeles.

McCawley, James D. 1975. Lexicography and the count-mass
distinction. Proceedings of the 1st annual meeting of
the Berkeley linguistics Society, 314-21.

McCawley, James D. (ed). 1976. Syntax and semantics:
Notes from the linguistic underground. N.Y.: Academic
Press.

McEldowney, Patricia L. 1977. A teaching grammar of the
English article system. IRAL 15.95-112.

Nunberg Geoffrey: and Chiahua Pan. 1975. Inferring
quantification in generic sentences. CLS 11.412-22.

Pelletier, Francis J. 1979. Non-singular reference: Some
preliminaries. Mass terms: Some philosophical
problems, ed. by Francis Pelletier, 1-14. Dordrecht,
Holland: Reidel.

Pica, Teresa. 1983. The article in American English: What
the textbooks don’t tell us. Sociolinguistics and
language acquisition, ed. by Nessa Wolfson and Elliot

159
Judd, 222-33. Rowley, Mass: Newbury House.

Quine, Willard. 1960. Werd and object. Cambridge,
Massachusetts: MIT Press.

Quirk, Randolph: Sidney Greenbaum: Geoffrey Leech: and Jan
Svartvik. 1972. A grammar of contemporary English.
London: Longman.

Russell, Bertrand. 1905. On denoting. Rpt. in
Contemporary readings in logical theory (1967), ed. by
I. M. Copi 8 J. A. Gould, 93-105. Macmillan.

Searle, John R. 1969. Speech acts. Cambridge: Cambridge
University Press.

Smith, C. 1964. Determiners and relative clauses in a
generative grammar of English. Rpt. in Modern studies
in English: Readings in transformational grammar
(1969), ed. by D. Reibel 8 S. Schane, 247-63.
Englewood Cliffs, N.J.: Prentice-Hall.

Strawson, Peter F. 1950. On referring. Rpt. in
Contemporary readings in logical theory (1967), ed. by
I. M. Copi 8 J. A. Gould, 105-26. Macmillan.

Tang, Ting-chi. 1986. The articles in English: Meaning
and usage. Teaching and Research 8.85-140.

Vendler, Zeno. 1967. Singular terms. Linguistics in
philosophy, 33-69. Ithaca, New York: Cornell
University Press.

Ware, Robert x. 1979. Some bits and pieces. Mass terms:
Some philosophical problems, ed. by Francis Pelletier,
15-29. Dordrecht, Holland: Reidel.

Whorf, Benjamin L. 1956. Language, thought, and reality.
Cambridge, Massachusetts: MIT Press. '

Yamada Jun: and Nobukazu Matsuura. 1982. The use of the
English article among Japanese students. Regional
English Language Center Journal 13.50-63.