1 Introduction

1.1 The realization of definiteness

The general structure and purpose of this study is Popperian. The

starting-point is neither a corpus of data nor a particular theoretical
position, but a problem, or rather a set of problems. My basic strategy is
first to describe the problems, then consider the solutions that have been
proposed to date. I then try to show in what respects these solutions are
inadequate, how they neglect or misinterpret certain kinds of data and
finally suggest my own analysis. In Popper's terms I thus proceed from
Problem to Hypothesis to Test to New Hypothesis (see Popper 1972). It
then remains to be seen what new problems and further, more refined,
hypotheses will follow from where this study leaves off.
The general issue is: what is definiteness? This is first of all a semantic,
conceptual question: what does the term 'definiteness' mean? Conse-
quently, and more specifically, we may ask: what does it mean to say that
something is definite or indefinite?
A second level of discussion then has to do with recognition: given a
definition of definiteness, definite and indefinite, how do we actually
recognize that something is definite or not? This is a matter of realization,
in two senses of the word. In linguistic terms, how is definiteness realized,
i.e. expressed? And in psycholinguistic terms, how does the hearer/reader
realize, or infer, that something is definite or not? In other words, what
needs to be examined is both how definiteness is encoded and how it is
decoded.
In outline, the main arguments are as follows. I shall claim that
definiteness is not a semantic primitive but a cover-term comprising a
number of oppositions which are more basic. This componential analysis
of definiteness shows that 'definite' and 'indefinite' are not merely polar
opposites, but qualitatively different concepts. Definiteness is not binary,
but composite; it is also scalar.

Cambridge Published
https://doi.org/10.1017/CBO9780511519710.001 Books Online ©Cambridge
online by Cambridge University
University Press Press, 2009
 2 Introduction

I start with an analysis of the English articles, as prototypical

realizations of definiteness. I argue that the English article system consists
of five terms: the, a and unstressed some, as 'surface articles', plus two 'no
article' categories: zero (indefinite, with mass and plural) and null
(definite, with singular proper nouns and some singular count nouns).
Each of these five terms imposes a distinct meaning on the NP. As
expressed in the English articles, definiteness can be analysed as a matrix
of three binary features, which suffice to distinguish the five articles:
locatability in a shared set (having to do with familiarity), inclusiveness
(quantity), and extensivity (abstractness and generality). The first two
features are adapted from Hawkins' location theory, and the third is
based on Guillaume's analysis of the difference between any surface article
and no surface article. The description proposed is then placed in an
informal set-theoretical framework. It is general enough to account for
both referentials and non-referentials, and also incorporates a theory of
genericness as a non-uniform derived reading.
In order to examine the implications of this analysis for languages
which lack articles I then turn to Finnish, a language which has a very
different structure. Here, definiteness is expressed or inferred in a variety
of ways, via indications either of familiarity or of quantity. The various
methods constitute a hierarchy: inferences based on word order can be
overruled by the use of article-like function words or by certain case-
endings, but the predominant factor is context.
These detailed treatments of English and Finnish are then followed by
briefer references to studies of a number of other languages, which appear
to corroborate my general conclusions.
The arguments about the nature of definiteness are also of relevance to
some aspects of linguistic theory in general: the applicability (or otherwise)
of standard logic in linguistic analysis; the relation between grammar and
pragmatics; fuzzy grammar and prototype theory. I conclude with the
suggestion that linguistics needs a non-Aristotelian paradigm to cope with
fuzzy concepts such as definiteness.

1.2 The theoretical problems

We start, then, with the question: what is definiteness? Something

of the nature of this problem is illustrated by the following definition by
Kramsky (1972: 30) of what he calls 'determinedness': 'By the term
" determinedness" we understand the fact that nouns are classified

Cambridge Published
https://doi.org/10.1017/CBO9780511519710.001 Books Online ©Cambridge
online by Cambridge University
University Press Press, 2009
 The theoretical problems 3

according to whether the content expressed by the noun is clear and

identifiable in a concrete way or not. In topical utterances [sic] this
category is realized in the positive case by "determinedness", in the
negative case by "indeterminedness'V Definitions of this kind are not
helpful. Kramsky evidently means that nouns are either determined or
undetermined in this way, but this does not amount to a definition of
definiteness or determinedness as such. Furthermore, his definition is
couched in terms that are themselves undefined. To say, for instance, that
a determined noun is one that is 'identifiable' is circular: it begs the
question of precisely what is meant by 'identifiable'. 'In a concrete way'
does not take us any closer - how would this be distinguished from an
'abstract' way? And what is meant by 'clear'? Is it really the content that
must be 'clear' in order for a noun to be determined? Precisely what is the
category that is 'realized by determinedness'? Is it not determinedness
itself which is the category in need of a definition? Indeed, in what sense
does it constitute a 'category' (whether universal or not) in the first place?
Related to this central question of definiteness are a number of other
issues which will also receive attention. One of these will obviously be
reference, which has traditionally posed a labyrinth of problems; however,
one of my points will be that definiteness cannot be restricted to matters
of reference alone. It will also become clear that reference, like definiteness,
is difficult to conceive of in any simple, unitary sense; rather, the term has
a wide and varied scope.
Another related problem concept is genericness, to which a good deal
of space is given. Again, I argue that this is a cover-term for a variety of
readings with different quantitative extensions. In particular, I argue that
the traditional view that the concept must be restricted to 'whole-species'
readings is mistaken. I also find it significant that in research on generics
appeal is often made to data of extraordinary dubiousness. Acceptability
judgements seem to vary enormously, not only regarding genericness but
also as regards basic grammaticality. Such discrepancies of opinion
among native speakers suggest that genericness is a particularly undefined
area of semantics.
Article-languages express definiteness through articles, but languages
lacking articles use a variety of other resources. A number of further
theoretical issues arise here. How do these other resources work? Are they
shared between unrelated languages? Is the definiteness they express
identical with that expressed by articles? What is the status of definiteness
as a universal category?

Cambridge Published
https://doi.org/10.1017/CBO9780511519710.001 Books Online ©Cambridge
online by Cambridge University
University Press Press, 2009
 4 Introduction

Ultimately, these theoretical problems do not relate to one particular

model of grammar or another but to certain general characteristics of
linguistic theory as a whole. In particular, they have to do with the rather
limited degree to which language can be said to be a well-defined system,
with clear-cut categories and black-and-white distinctions. They are also
of relevance to the problem of the empirical status of linguistics as a
(possibly) cognitive science, although this last is an issue that lies beyond
the scope of the present study.

1.3 The English problem

1.3.1 A great deal has been written about definiteness in English. It

must be stressed at the outset, however, that I shall focus exclusively on
the articles, and omit consideration of definiteness elsewhere in the
grammar. It is via the articles that definiteness is quintessential^ realized,
and it is in analyses of the articles that the descriptive problems are most
clearly manifested. Moreover, it is largely on the basis of the evidence of
articles in article-languages that definiteness has been proposed at all as a
category in other languages. Since the articles constitute the prototypical
core of definiteness expression in English, an adequate description of this
core must be a necessary precondition for any more comprehensive
account of definiteness in English as a whole. However, any description of
this core should also be capable, in principle, of extension at least to other
determiners and quantifiers, as I shall seek to show.
The word-class ' articles' has been a puzzle to grammarians of English
right from the start. An extreme illustration is the view of Gardiner that
the articles are no more than 'useless ballast', 'old rubbish':

It is sometimes said that such relatively insignificant words [i.e. as the articles] are
grammatical tools. But the function of tools is to achieve some specific end. That
is precisely what, in many cases, the article does not do, or at all events does only
in a very slight and uncertain degree. Often it is mere useless ballast, a habit or
mannerism accepted by an entire speaking community. The accumulation of old
rubbish is so easy. (Gardiner 1932: 47. Cited in Hewson 1972: 78-9)

The word 'article' itself derives from the Greek arthron, which in Greek
covered relative pronouns and originally also personal pronouns. Latin,
however, had no articles, and because the early seventeenth-century
grammarians of English based their descriptions on Latin the English
articles presented a problem (see e.g. Michael 1970; Vorlat 1975). Most of

Cambridge Published
https://doi.org/10.1017/CBO9780511519710.001 Books Online ©Cambridge
online by Cambridge University
University Press Press, 2009
 The English problem 5

them did not dare to recognize the articles as a separate part of speech,
and gave them a variety of descriptive labels such as 'nominal note',
'particle', 'sign of the substantive', 'sign corresponding to the Latin
cases', and later 'adjective'. For instance, Michael (1970: 351) quotes
from J. Clarke's Rational Spelling Book of 1772: 'As there is but one real
Case in our Tongue, viz. the Genitive... we are obliged to have Recourse
to Articles to decline our nouns.' It was understood that a was a weakened
form of the numeral one, and that the derived from the demonstrative that,
but there is very little analysis of the different functions of these forms.
The definite article the was said to 'individuate', and often implied
previous reference, while a was used with first mentions.
It is really not until Lowth (1762) that articles are taken as a separate
word-class in their own right. As such, Lowth was also interested in the
characteristics of this word-class as a whole, as well as the differences
between the articles. He wrote that the articles - i.e. the and a - are used
before nouns 'to shew how far their signification extends'. And: 'A
substantive without any article to limit it is taken in its widest sense'
(Lowth 1762: 15ff.: quoted in Michael 1970: 361). We shall have reason
to return to this insight later, and also to the connection between the
articles in English and case in other languages.
This kind of historical perspective illustrates how the early grammarians
looked at the English articles through Greek and Latin spectacles. At the
beginning they were not seen as a self-evident well-defined category native
to English itself.

1.3.2 As an illustration of the standard contemporary view of the

articles, and of the problems which arise from this view, I take the analysis
given in Quirk et al. (1985) as representative. This is summarized as
follows (1985: 265), for common nouns with 'specific reference' (as
opposed to 'generic reference').

DEFINITE INDEFINITE
Count Non-count Count Non-count
Singular the tiger the furniture a tiger (some) furniture
Plural the tigers (some) tigers

In addition, say the authors, the, a and 'the zero article' can be used for
generic reference, in contexts which vary to some extent for each of these
articles.

Cambridge Published
https://doi.org/10.1017/CBO9780511519710.001 Books Online ©Cambridge
online by Cambridge University
University Press Press, 2009
 6 Introduction

Underlying this standard analysis are a number of assumptions which

are too rarely challenged (see also Chesterman 1990). Such assumptions
include the following:

(a) English has two articles proper, the and a; plus (sometimes)
unstressed some.

(b) The set of articles also includes the member 'no article', or 'zero
determination'. This is at least sometimes in free variation with
some in non-generic contexts.

(c) The distribution of the articles (and of' no article') is determined

by a combination of three binary oppositions: definite vs
indefinite, count vs non-count and singular vs plural.

(d) There is something called 'generic reference' which can be

expressed by the, a and zero.

(e) Proper names are a separate category, and do not take articles at
all in the singular.

There are problems with all these assumptions, which will be discussed
at length in subsequent sections. The main issues can be outlined as
follows.

Concerning assumption (a): What is the precise status of

unstressed some! Is it or is it not an article proper? On what criteria are
the articles proper to be defined? (See 3.2 below.)

Concerning assumption (b): What is the status of'no article'? If

it is indefinite, why do Quirk et al. also feel it necessary to refer (1985: 276)
to uses of 'the zero article with definite meaning'? (See 2.1.3, 3.2.) If, as
the chart above suggests, zero and some are (at least sometimes) in free
variation, why is it that they are not always? (See 2.4.)

Concerning assumption (c): A binary division of definite vs

indefinite suggests that, according to the chart, a, some and zero are all
indefinite 'in the same way', or at least subcategories of a single category
of indefiniteness; but is 'indefiniteness' a unitary concept? (See 4.1.)
Presumably not, since the overlap between some and zero is by no means
perfect. Compare (from Quirk et al. 1985) Tve just bought some melons /
^melons and They have become vegetarians / *some vegetarians. And if
zero is indefinite, again, how can it also have definite uses?

Cambridge Published
https://doi.org/10.1017/CBO9780511519710.001 Books Online ©Cambridge
online by Cambridge University
University Press Press, 2009
 The English problem 1

Concerning assumption (d): Is 'generic reference' a unitary

concept? Again, presumably not, since not every 'generic' article can be
used in every generic context: *A tiger is becoming almost extinct. (See 2.5.)

Concerning assumption (e): It is surely a weakness of this

standard description that proper nouns are not incorporated into the
description in any systematic way. (See 4.4.)

Another severe drawback of this view is the enormous number of

exceptions it gives rise to, which plague not only learners of the language
but also grammarians and language teachers trying to make sense of the
article system. This standard analysis states that the distribution of the
articles is restricted in the first place by the class of the noun: only certain
nouns (count or non-count/mass) accept given articles. But consider the
wealth of 'exceptional usage' here. First, there is the generalization that
a only occurs with count nouns. Any other usage - such as a vicious anger,
a surprising determination - must therefore be somehow exceptional.
Second, since neither a nor the is supposed to occur with proper names,
uses like There was a Tom Jones on the phone for you are also exceptions,
together with the Freddie I knew, a second Milton, and even the River
Thames. Third, singular count nouns are not supposed to take 'no article',
and so the many uses where this happens are also relegated to the status
of exceptions or 'rather special uses'. Some examples are: captain of the
team, in bed, a girl of good family, on piano tonight we have..., a funny kind
of person, sailor he may be, but..., doctor will see you now, part is given
below. Exceptions such as these are significant because they all represent
productive types, not one-off uses. As exceptions, they are all in fact
counter-evidence to the standard rules.
One conclusion to be drawn from this is that it is not helpful to link
article usage too directly to noun class, and hence to the distinction
between count and non-count. The custom of so doing goes back at least
to Jespersen (1924), but there is another research tradition which starts
from the very opposite assumption: that, given an appropriate context,
almost any noun can occur with any article. One major source of this
tradition is the work of Guillaume (1919), which will be discussed further
in section 2.3 below. In this view, the likelihood of a given noun class
occurring with a given article is more a matter of statistics than syntax:
some article + noun combinations will occur more frequently than others
because certain types of context are, for pragmatic reasons, more
frequent. Thus in our culture dog is normally thought of as a discrete

Cambridge Published
https://doi.org/10.1017/CBO9780511519710.001 Books Online ©Cambridge
online by Cambridge University
University Press Press, 2009
 8 Introduction

object and hence count, but in another culture some dog might denote
desirable food and thus become a mass noun. And imagine the giant in
' Jack and the Beanstalk' calling out' I smell boy!' Apart from the obvious
impossibility (for historical reasons, a deriving from one) of a + plural, the
exceptional cases would be no longer those with an unusual article, but
those nouns which consistently reject a given article. This rejection would
thus be the result of a clash between something in the meaning of the noun
in question and something in the meaning of the article. (This topic is
discussed further in 3.1.)
It seems, then, that the problem remains of producing a coherent
description of article usage in English, one that does not need to
incorporate such a host of exceptions. My own analysis (in chapter 4) will
seek to combine insights from several research traditions into a more
comprehensive theory of the articles than any of the traditions can provide
on its own.

1.4 General outline

The outline of the book is as follows. The main problems have

been sketched in preliminary terms in the present chapter. Chapters 2-4
deal with English. Chapter 2 is a critical review of the major research
traditions, and concludes with a statement of issues that remain
unresolved. Chapter 3 looks at English article usage in some detail,
focusing particularly on less common or so-called exceptional uses. It also
discusses the question of how many articles there actually are. Chapter 4
presents my own unified description of the English articles, based on a
componential analysis of deflniteness that incorporates three distinct
semantic or pragmatic oppositions. Deflniteness and indefiniteness are
thus taken as composite cover-terms for a complex of inter-related
distinctions, rather than labels for a single binary opposition. Deflniteness
itself is argued not to be a primary notion at all, but compositional. The
chapter includes brief indications of how the suggested description can be
usefully applied to some related issues, and ends with some speculation on
the possible explanatory value of two key concepts in the analysis.
Chapters 5-7 then look at Finnish. Chapter 5 introduces the language
and the kinds of data that are relevant to definiteness. Chapter 6
summarizes the Finnish research tradition on deflniteness from the
beginning of the century. Most of the significant work has been done in
response to a proposal by Siro in the 1960s, and the chapter accordingly

Cambridge Published
https://doi.org/10.1017/CBO9780511519710.001 Books Online ©Cambridge
online by Cambridge University
University Press Press, 2009
 General outline 9

discusses this response in some detail, according to each particular line of

argument used. In chapter 7 I then develop my own analysis of the ways
in which definiteness can be expressed or inferred in Finnish. I also claim
that some of the ways traditionally proposed must be rejected, such as
stress and, to some extent, word order.
Chapter 8 has an explicitly contrastive focus. It lines up the analyses of
the two languages side by side and demonstrates the extent to which they
can both be stated in the same terms. This common descriptive
denominator is then used to establish a number of correspondences
between given structures in the two languages. The chapter also points out
several diachronic similarities in the historical development of some
English articles and some Finnish function words.
Chapter 9 discusses a number of wider theoretical issues arising from
the study. These partly have to do with the theory of definiteness and
reference in general. Similarities are noted with descriptions of definiteness
and/or reference in several other, unrelated, languages. Attention is
drawn to the fact that evidence from many languages suggests that both
definiteness and reference are scalar phenomena. This in turn has
implications for the relation between language and logic. The study also
illustrates how the borderline between grammar and pragmatics is drawn
differently in different languages. The evidence presented is thus pertinent
to the whole issue of the delimitation of grammar as what Levinson (1987)
calls 'frozen pragmatics'. Finally, a comparison is drawn between the
kind of fuzzy grammar that appears to be necessary for an analysis of
definiteness, and more general theoretical principles that seem to indicate
the existence of a new paradigm in linguistics.

Cambridge Published
https://doi.org/10.1017/CBO9780511519710.001 Books Online ©Cambridge
online by Cambridge University
University Press Press, 2009