On definiteness and the co-occurrence of the definite article with other determiners in Modern Greek

Full text

On definiteness
and the co-occurrence of the definite article
with other determiners in Modern Greek
Dimitra Lazaridou Chatzigoga
A doctoral thesis
presented to the Departamento de Filologia Española
Universidad Autónoma de Barcelona
Doctorate program: Ciencia Cognitiva y Lenguaje
Directors:
Dr. Louise Elizabeth McNally Seifert
Dr. Josep Maria Brucart Marraco
May 2009

!"#$ µ%&'()

2
nouns to nouns
wan
wan
too nons too
and
and
nuns two nuns
w an d
ering
in sin
g
ular untheknowndulous s
pring
E.E.Cummings

3
this is a common-place for acknowledgments
in the first place i would like to thank my directors, louise mcnally and josep maria brucart.
since i met louise we started a dialogue that was not initially planned to have such concrete
results. now i am pretty convinced that if she had not been so involved in the writing and
shaping of this work, it would most probably not have become an instantiation of the kind
thesis. she showed me the way to think and argue as a linguist, challenged me and discussed
with me everything that came up as an issue and i thank her for being open enough to
incorporate other thoughts. josep maria on the other hand kindly accepted to become involved in
a thesis that was meant to be more ‘semantic’ than ‘syntactic’ and helped me a lot with his view
as a syntactician, his thought as a grammarian and his always so well-placed words.
i received inputs from discussions with many people, who brought up issues, commented on my
work-in-progress and challenged me with questions. one subset of them includes teresa espinal,
josep macia, maribel romero, josep quer, anastasia giannakidou, melita stavrou and cleo
condoravdi, so i hereby thank them all. my informants form another subset, without whom this
research would not have had solid grounds. another subset includes younger linguists and
friends, of which i want to especially thank cristina real puigdollers and ía navarro for making
the autonoma circle more enjoyable, as well as víctor acedo matellán for being such a thoughtful
and caring linguist friend.
the academic environment, wherein i found myself, the Universitat Autònoma de Barcelona, has
been a place where to arrive, but also a place from which to depart. i thank all the people in the
Centre de Lingüística Teòrica. special thanks are due to teresa espinal, who has been my
dedicated tutor, since i arrived in barcelona. this thesis was supported by a research grant FI-
IQUC from AGAUR, which i hereby acknowledge1.
other people and close friends that have been around all these years, but do not fall under the
scope of linguistics, deserve my deep gratitude. they can be found in other places and not within
the preface of a thesis..
i would like to warmly thank my parents, (*+ and ,%'$$)-, and my sister, ./"0%$1, for always
being there, supporting me in diverse ways and following all this procedure from a distance.
anyone interested in understanding the reason or the possible links between me, linguistics and
the philosophy of badiou can only get a clear representation thereof by meeting µ%&'()-. he is
and has been here and there, throughout this work over the years, but, most importantly, over
each moment. he has been the locus of constant inspiration and insistent elaboration of a
handful of my thoughts, i am simply so grateful and charmed by his generous presence in my
life.
1 Amb el suport del Comissionat per a Universitats i Recerca del Departament d’Innovació, Universitats i Empresa de
la Generalitat de Catalunya.

4
TABLE OF CONTENTS
Acknowledgments........................ .................................................................................................3
Chapter 0
Introduction-Overview of the thesis...........................................................................................8
Chapter 1
A theory of definiteness..............................................................................................................11
1.1 Introduction-Overview...........................................................................................................12
1.2 Definiteness intertwined with the concepts of uniqueness and familiarity............................12
1.3 Uniqueness-based theories.....................................................................................................17
1.3.1 A Russellian perspective.....................................................................................18
1.3.2 Uniqueness, maximality, inclusiveness..............................................................21
1.3.3 Existential claim, reference and specificity........................................................22
1.3.4 Domain restriction..............................................................................................24
1.3.5 Summary of the uniqueness-based theories........................................................24
1.4 Familiarity-based theories......................................................................................................25
1.4.1 Grammarians.......................................................................................................26
1.4.2 Contemporary semantic and pragmatic theories: Hawkins (1978), Heim (1982)
and Roberts (2003)..........................................................................................................28
1.4.3 Summary of the subsection.................................................................................35
1.5 Discussion..............................................................................................................................35
1.5.1 Problems with the uniqueness-based theories....................................................35
1.5.2 Problems with the familiarity-based theories.....................................................38
1.6 A new proposal for a theory of definiteness..........................................................................44
1.7 Summary of the chapter.........................................................................................................51
Chapter 2
The definite article in Modern Greek.......................................................................................52
2.1 Introduction-Overview...........................................................................................................53
2.2 Introduction to the syntactic characteristics of the definite article in Greek.........................53
2.2.1 Syntactic position and syntactic independence...................................................55
2.3 Excursus on Grammars of Greek...........................................................................................56
2.4 Uses of the definite article.....................................................................................................58
2.4.1 Pragmatically motivated presence of the definite article....................................59
2.4.1.1 Anaphoric use....................................................................................................59

5
2.4.1.2 Situational use: visible or immediate..................................................................61
2.4.1.3 Situational use: specific or general knowledge/associative................................61
2.4.1.4 Associative anaphoric use...................................................................................62
2.4.1.5 Endophoric use...................................................................................................63
2.4.2 Semantic motivation for the presence of the definite article..............................63
2.4.2.1 Proper nouns.......................................................................................................64
2.4.2.2 Generics..............................................................................................................68
2.4.2.3 Superlatives and ordinals....................................................................................71
2.4.3 Other uses...........................................................................................................72
2.5 More on the syntactic properties of the definite article: the DP hypothesis..........................74
2.6 Polydefiniteness.....................................................................................................................77
2.7 Co-occurrence of the definite article with other elements: Determiners, Demonstratives and
Quantifiers in Greek...............................................................................................................83
2.7.1 Obligatory co-occurrence of the Greek definite article with other
determiners/quantifiers....................................................................................................84
2.7.1.1 Demonstrative determiners.................................................................................85
2.7.1.2 Possessive determiners.......................................................................................88
2.7.1.3 Universal quantifier olos.....................................................................................89
2.7.1.4 Other pronouns...................................................................................................91
2.7.2 Optional co-occurrence of the definite article with determiners and quantifiers
in Greek...........................................................................................................................92
2.7.2.1 Numerals and polar contrastive adjectives and a note on the incompatibility of
the definite article with determiners/ quantifiers.............................................................93
2.7.2.2 Challenging cases of optional co-occurrence of the definite article with other
determiners......................................................................................................................94
2.8 Some remarks on the indefinite article..................................................................................97
2.9 Summary of the chapter.........................................................................................................99
Chapter 3
The co-occurrence of kathe with the definite article.............................................................101
3.1 Introduction..........................................................................................................................102
3.2 Morphological properties.....................................................................................................102
3.3 Syntactic properties..............................................................................................................103
3.4 Semantic properties..............................................................................................................116
3.4.1 Singular/ plural and complement phrases.........................................................117
3.4.2 Characteristics of the quantifier........................................................................120
3.4.2.1 Scope and distributivity....................................................................................120

6
3.4.2.2 Distributivity.....................................................................................................122
3.4.3 Genericity.........................................................................................................135
3.4.4 Summary of the similarities and differences between [kathe NP] and [o kathe
NP].................................................................................................................................145
3.5 Analysis of [o kathe NP] .....................................................................................................146
3.5.1 Analysis of [o kathe NP] ..............................................................................................146
3.5.2 Discussion of previous analyses....................................................................................148
3.6 Summary of the chapter.......................................................................................................151
Chapter 4
The co-occurrence of opjosdhipote with the definite article..................................................152
4.1 Introduction-Overview.........................................................................................................153
4.2 Morphological properties.....................................................................................................155
4.3 Syntactic properties..............................................................................................................156
4.3.1 Syntactic position and co-occurrence with other elements in the noun phrase
.......................................................................................................................................156
4.3.2 Syntactic analysis with the indefinite and the definite article..........................159
4.3.3 A note on partitives and opjosdhipote..............................................................160
4.3.4 Polydefiniteness................................................................................................161
4.4 Semantic properties..............................................................................................................163
4.4.1 Non-veridicality and anti-episodicity...............................................................163
4.4.2 Cancellation of the anti-episodicity constraint.................................................168
4.5 Semantic analysis: Opjosdhipote as a Heimian indefinitew.................................................171
4.5.1 Arguments in favor of the indefinite analysis of FCIs.....................................175
4.5.2 Caveat on differences.......................................................................................177
4.5.3 A note on the scope of opjosdhipote.................................................................179
4.6 Discussion of [o opjosdhipote N]........................................................................................179
4.7 Analysis of [o opjosdhipote N]...........................................................................................181
4.8 Summary of the chapter.......................................................................................................184
Chapter 5
Diachronic and typological issues...........................................................................................185
5.1 Introduction-Overview.........................................................................................................186
5.2 On categories: lexical, functional, linguistic categories/ parts of speech............................186
5.2.1 Lexical and functional words/categories................................................................186
5.2.2 Linguistic categories/ parts of speech.....................................................................189
5.3 Cross-linguistic considerations and grammaticalization.....................................................191

7
5.4 Are [o kathe/o opjosdhipote N] innovative? .......................................................................193
5.5 The definiteness cycle..........................................................................................................196
5.6 Summary of the chapter.......................................................................................................207
Chapter 6
Conclusions...............................................................................................................................209
Appendix: On Badiou...............................................................................................................211
Bibliography.............................................................................................................................215

8
Chapter 0
Introduction-Overview of the thesis

9
The topic of this thesis is the notion of definiteness as manifested in Modern Greek. On the one
hand we are interested in the conditions that describe the contribution of a definite noun phrase
introduced by the definite article o, i, to ‘the’, while on the other hand we have studied two
cases of co-occurrence of the definite article with other elements in the noun phrase, namely,
with the universal distributive quantifier kathe and with the free choice item opjosdhipote.
Given the apparent optionality of this co-occurrence, the initial motivation for our inquiry has
been to identify whether the definite article has any contribution in these constructions, and if
any, to specify its characteristics.
In the first chapter of this thesis we will be concerned with the notion of definiteness, providing
a critical review of the two notions that have been associated with it, namely, uniqueness and
familiarity. This critical review will lead us to consider representative accounts of both lines of
thought, which will be called uniqueness-based theories and familiarity-based theories. The
discussion to be presented will reveal data that challenge the viability of an account that relies
on only one of these two notions, while the reinterpretation of both will be shown to be
necessary: on the one hand, uniqueness will be re-interpreted as count-as-unique (on the basis of
the ontology of Badiou 1988), and on the other hand, familiarity will be interpreted as weak
familiarity in the sense of Roberts (2003). Finally, the definition of definite to be proposed here
will be closely connected with Greek data that appear to be challenging for a theory of
definiteness, namely, the co-occurrence of the definite article with the distributive quantifier
kathe and with the free choice item opjosdhipote.
In the second chapter I focus on the expression of definiteness and the definite article in Modern
Modern Greek, and thus a comprehensive and in depth presentation of the characteristics of the
Greek definite article and its uses is called for. I first present the uses of the definite article in
Greek on the basis of Hawkins’ (1978) characterization, dividing them into uses that are
motivated pragmatically on the one hand and on semantic grounds on the other. I think discuss
the syntactic characteristics of the article, including its position, its syntactic independence, the
phenomenon of polydefiniteness and, finally, the possibility of co-occurrence of the article with
other elements in the noun phrase.
In the third chapter I provide a thorough discussion of the relevant data with [(o) kathe NP] in
Greek. The presentation of the morphological, syntactic and semantic characteristics of the
universal distributive quantifier kathe and its interaction with other elements in the noun phrase
will be paralleled by the presentation of the contexts where the co-occurrence with the definite
article can take place, which will be shown to be restricted, and thus the issue of the difference
in interpretation that comes about with the definite article will be addressed.

10
In the fourth chapter I study in more depth the free choice item opjosdhipote and its co-
occurrence with the definite article in Greek. The discussion of the morphological, syntactic and
semantic characteristics of this item will provide sufficient grounds to evaluate the contribution
of the definite article. The semantic properties of (non)veridicality and anti-episodicity will be
shown to form the two licensing condition on the free choice item (FCI) opjosdhipote, which
will be given an analysis as a Heimian indefinite (Heim 1982). A comparison of [opjosdhipote
N] and [o opjosdhipote N] will lead us to the conclusion that their differences are belong to the
realm of pragmatics, while the contribution of the definite article will be identified with the
count-as-unique condition.
In the fifth chapter I attempt to situate the description of the linguistic data and the formal
proposal of the previous chapters of this thesis within two perspectives I did not consider in
great detail previously, namely, that of diachronic evolution and that of cross-linguistic
evidence and comparison. I will first provide some discussion of the lexical/functional category
distinction and of the evolution of the category of article in traditional grammatical and
linguistic theory. This discussion will be followed by the consideration of the presence/absence
of this category in different languages, moving from Greek to other Indo-European and non-
Indo-European languages, ending finally with the proposal Greenberg (1978) elaborated
concerning the definiteness cycle, as refined later by Lyons (1999), with special attention, for
the purposes of this work, to Greek.
In the sixth chapter I will draw some conclusions on the basis of the discussion and proposal of
the theory of definiteness defended in this thesis.

11
Chapter 1
A theory of definiteness

12
1.1 Introduction-Overview
In the first chapter of this thesis we will be concerned with the notion of definiteness providing
a critical review of the two notions that have been associated with definiteness, namely,
uniqueness and familiarity. This critical review will lead us consider representative accounts of
both lines of thought, which will be called uniqueness-based theories and familiarity-based
theories. The discussion to be presented will reveal data that challenge the integrity of an
account that relies on only one of these two notions, while the reinterpretation of both will be
shown to be necessary; on the one hand, uniqueness will be re-interpreted as count-as-unique
(on the basis of the ontology of Badiou 1988), and on the other hand, familiarity will be
interpreted as weak familiarity in the sense of Roberts (2003). Finally, the definition of definite
to be proposed here will be closely connected with Greek data that appear to be challenging for
a theory of definite, namely, the co-occurrence of the definite article with the distributive
quantifier kathe and with the free choice item opjosdhipote.
1.2 Definiteness intertwined with the concepts of uniqueness and familiarity
A definite noun phrase is either unique or familiar. The current state of the investigation on
definiteness usually associates definiteness with uniqueness or familiarity. We may ask then,
what are the properties of the notion involved? What are the interpretations speakers give to
definite noun phrases and what have linguistic theories said on this topic? The objective of this
chapter is on the one hand to review the notions that have been associated in one way or another
with definite noun phrases in language and on the other hand to provide a new perspective
within this landscape, which will be the point of departure for a new understanding of
definiteness.
The characterization of definiteness is associated with the investigation of two other
notions, to which definiteness has been quite often attached. The emblematic notions in the
history of the definition and in the study of definiteness have been uniqueness and familiarity2.
The moment we start reading a section about definiteness in a semantics textbook or a grammar
of any language possessing definite articles3 the words unique and familiar appear almost
immediately. The application of these two notions originates in the attempt to explain what it is
2Definiteness has been also attached to other notions like presuppositionality (of existence), referentiality, specificity,
strength/weakness, salience, notions that will be mentioned throught the discussion, but will be not treated in detail.
For definite descriptions governed by salience see for example Lewis (1973,1979).
3 I treat the definite article as the basic, simple manifestation of definiteness. I would like to point out here that
definite is only one, still the most common one, among the various names that have been proposed in the literature to
designate the function of the article. In Christophersen (1939: 50ff) we find a rather long list of other possible terms:
determinative, defining, individualizing, particularizing, specializing, descriptive, familiarizing, actualizing,
concretizing, substantivizing, etc. He says that the term «definite is presumably meant to signify that the noun to
which the article is attached stands for something definite».

13
for a noun phrase to be definite, to understand what the conditions are under which people use
or rather choose to use a definite over an indefinite noun phrase and what people interpret when
they hear definites. The interpretation of definites is an issue situated on the border between
semantics and pragmatics and this situation, of being between, of depending on both levels, is
visible in the theories on definiteness that linguists have proposed. On the one hand, theories
that are more focused on logical semantics are usually based on uniqueness, while pragmatic
theories tend to treat familiarity as the central notion for definiteness. Except for this difference
with respect to which notion is claimed to be the most relevant one, another feature marks a
clear distinction between these two theoretical choices: uniqueness-based theories, on the basis
of standard logical tools, often treat (in)definite noun phrases as quantificational, while
familiarity-based theories often treat them as non quantificational, as variables, as, for instance,
in the theory of Heim (1982).
The threads that unite definiteness with these notions can be appreciated if we consider some
definitions of these two notions. In the first place, an intuitive and pre-theoretic interpretation of
these notions will be helpful. Taking as a starting point the notion of uniqueness, a first
definition could be the following:
1. An entity is unique if it is the only one (exactly one and no more) of its kind.
The definition in 1 provides us with two basic insights; in the first place that the category of
quantity is particularly relevant and in the second place that an existential predication
concerning the entity in question (manifested with the predicate be) is involved. Another way to
define uniqueness would to be to claim the following:
2. Uniqueness indicates that there is one and only one (of some property).
Definition 2 is used to indicate the existence of only one entity with a certain property. A
recurrent employment of the notion of uniqueness is present in mathematics and logic, where
the category of quantity and number are crucial. The representation of this notion in
mathematical logic makes use of the unique existential quantification and is formalized as
follows:
3. !!x " F
(with F representing a property)
The exclamation mark in 3 serves to indicate uniqueness -if we don’t include it, the formula
turns into regular existential quantification. This formula can be read as follows “there is exactly

14
one entity x, such that x is F”. What we are interested in in the interpretation of the exclamation
mark we observe in this logical formula, of this exactness of the number one applied to this
concept.
After this brief introduction to the notion of uniqueness, we can turn now to the investigation of
the link between definiteness and uniqueness, which has been drawn since Russell (1905) along
the following lines: When someone uses a definite noun phrase s/he does so because the noun
phrase denotes exactly one entity in the universe of discourse. In contrast, if s/he would use an
indefinite s/he would pick one of the possibly many entities the common noun designates.
Suppose a speaker utters the following:
4. The trumpet is on the table.
The speaker uses the definite article4 the to talk about an entity, in this case the trumpet, which
is supposed to be exactly one, that is, unique in the universe. Otherwise she would use instead 5,
6 or 7:
5. Trumpets are on the table.
6. A trumpet is on the table.
7. The trumpets are on the table.
If she had used 5 she would have been committed to the fact that there is more than one trumpet
on the table given that there is a plural involved, namely, trumpets and if she had used 6 she
would have implied that there might be more trumpets somewhere else. By uttering 6 the
speaker would not want to commit herself to the claim that there is only one trumpet. Coming
back to the utterance with the definite noun phrase in 4 we can say that the speaker predicates
something of the only trumpet in the universe. The utterance of 7 instead would lead the speaker
to commit herself to the claim that there is more than one trumpet; there is a set of trumpets that
is supposed to be the unique maximal set of trumpets. Stated along these lines, uniqueness
seems too strong a claim, since it is hard to believe that someone would commit herself to the
existence of exactly one trumpet in the whole universe. Recognizing this fact has lead to the
loosening of the strength of the link between definiteness and uniqueness, so that uniqueness of
a definite noun phrase is not to be understood in a global, universal manner, but as restricted to
the specific universe of discourse that is under play, to the specific context of utterance, which
4 Throughout the thesis I will use only definites employing the definite article, but in the literature other kinds of
definites have been also identified such as demonstratives, possessives, proper names and personal pronouns, which I
will discuss with respect to Greek in the second chapter.

15
is a subpart of the universe. This restriction of the domain helps us to understand what happens
when a speaker utters a sentence like 4 above or 8 below:
8. The cat is asleep.
The speaker is not saying that there is only one cat in the whole universe, but rather that in the
specific part of the universe comprised by the speaker and hearer at the moment of the relevant
discourse there is only one cat or that there is only one salient cat or that the speaker is aware of
only one cat. I will take up this issue, which can be termed domain restriction later in this
chapter.5
At the beginning of this chapter I alluded to another concept that needs to be scrutinized,
namely the concept of familiarity. Considering first the notion independently of definites in
natural language, familiarity is said of something when close acquaintance, reasonable
knowledge or intimacy with someone or something is present. A familiar person is someone we
are closed to and a familiar situation is one of which we have some experience, some
knowledge of what it is about. If we transport this notion to the field of definiteness, the
rationale proceeds as follows: when we use a definite noun phrase we do so because we assume
that we know what we are talking about, we are familiar with the entity described by the definite
noun phrase and because, crucially, we think that the hearer knows it too, in this sense it is
something both participants of the discourse are aware of or have acquaintance with. How are
we familiar with the entity in question? The most immediate answer that comes to mind is that
familiarity is assured on the basis of the previous discourse, which includes some explicit
mention of the entity. If we weren’t acquainted (in the sense of being also able to identify it)
with the entity we were talking about we would probably use an indefinite noun phrase because
we would lack the information needed for a felicitous use of a definite6. It is interesting to notice
that definites act like an understanding between speaker and hearer and that both interlocutors
need to be aware of that understanding. If it were for example the case that the speaker knows
what she is talking about, but the hearer doesn’t, a sentence like 8 above would provoke an
answer like the following:
9. What cat are you talking about?
5 See Abbott (2004:8) for possible solutions in order for shrinking the domain of evaluation of the NPs in questions:
Barwise & Perry (1983) speak in terms of resource situations, Westertahl (1985) of context sets, Hawkins (1991) of
pragmatic sets. Stanley & Szabó (2000) propose indexing nominals with functions from discourse entities to
restricted sets.
6 I remind the reader that here I am only discussing the notions in an intuitive manner and that a more precise
discussion will follow in the subsequent pages of this chapter.

16
If the hearer were not acquainted with the cat, 9 would be a perfectly plausible continuation of
the discourse. According then to this line of thought, when we use a definite noun phrase we
presuppose that we share with the hearer a common ground, a group of entities that we both
have knowledge of and to which we can refer via language. When someone uses a definite noun
phrase she has the intention to refer to the entities shared by herself and the hearer, so that they
can both identify the referent of a noun phrase. It would seem then that if the speaker utters a
definite with no previous context provided, the hearer has no basis to interpret the definite, so
she will not be able to understand to what the speaker is referring to. Discourse in everyday life
is not so rigid though and there are a lot of situations where hearers prove themselves more
adaptive and imaginative than this previous discussion shows or linguistic theories consider
them to be; it common to disregard the fact that there is also context outside the strictly
linguistic one.7
The notions of uniqueness and familiarity have provided the point of departure for many
theories of definiteness, which could be divided according to authors that provide an
explanation of definiteness relying on either uniqueness or familiarity or take one of them as the
basic notion and derive the other one depending on which they take to be the fundamental one.
There are also other authors that claim that both notions co-exist or that neither is valid. This
typology, as we see, leaves us with a lot of possible theoretical combinations and almost all of
them have been exploited in the literature. We will refer to the theories that are build on the
notion of uniqueness, as uniqueness-based theories and to the theories relying on familiarity as
familiarity-based theories. Historically, it seems that there first emerged a version of a
familiarity-based theory, but in modern linguistics the uniqueness-based theories are considered
to be the classical theory of definite descriptions (for such a claim see for instance Abbott
1999). On the side of the uniqueness-based theories we encounter thus Russell (1905), Löbner
(1985), Kadmon (1990), Neale (1990), Hawkins (1991) and Abbott (1999 and subsequent
work), while on the side of familiarity-based theories we find Apollonius Dyscolus (2nd century
AD), Christophersen (1939), Jespersen (1949), Bolinger (1977), Hawkins (1978), Prince (1981,
1992), Heim (1982), Kamp and Reyle (1993) and Roberts (2003)- to name only some of the
most representative and influential accounts. Farkas (2002) and Farkas and deSwart (2006) use
the umbrella term determined reference to refer to both uniqueness/maximality8 and familiarity,
while Poesio and Vieira (1998) claim that work based on corpora has shown that neither notion
can account for all the data on its own and that the relevant conclusion is that both are needed in
order to explain the data or that a new concept (individuation) is needed, as argued also by
Birner & Ward (1998). This landscape needs to be explored in more detail in order for us to
7 These observations about the nature of everyday discourse and the means that interlocutors employ in order to
interpret language are connected to more dynamic theories of meaning in language, which rely on the notion of
discourse as well as on the study of extralinguistic factors. See for example the dynamic theory of Heim (1982).
8 Maximality can be seen as a version of uniqueness. See later on in this chapter for some discussion on this point.

17
decide what lies behind this notion, its association to definiteness and which theory describes in
the most comprehesive way the data we encounter in languages. The description and discussion
of the theories will be centered around the crucial concepts for each line of thought and is going
to be the main focus of this chapter from now on, leading finally to a new proposal for a theory
of definiteness.
Anticipating slightly the discussion, the proposal I am going to elaborate is a reconsideration of
both notions, uniqueness and familiarity, and is an appeal for the necessary combination of the
two in order to account for the vast amount of contexts where the definite article can be used in
languages, and specifically, in Greek. Theories that rely either on uniqueness or familiarity
alone appear unable to provide a coherent explanation for the whole range of data, since there
are empirical data involving definite noun phrases that question the full coverage of each line of
thought and show that theories of definiteness founded only on one notion seem incomplete,
since there always remain some cases that cannot be treated by uniqueness or familiarity alone.
As pointed out above, this conclusion has been reached by other authors as well (Poesio&Vieira
1998, Birner&Ward 1998). The proposal we are pursuing in this thesis relies on the claim that
both notions are needed, that they should be treated as two conditions, independent the one from
the other (and therefore we are not going to derive one from the other). Uniqueness and
familiarity will be rethought as two conditions, whose co-existence is a unavoidable fact. The
position we adopt is based on empirical data that demonstrate that definite noun phrases cannot
be fully explained on the sole basis of one notion, that the combination of the two gives us a
more precise view on definiteness and that their re-elaboration helps us understand understudied
uses of the definite article in Greek. The notions of uniqueness and familiarity will not remain
intact through the discussion, so towards the end of this chapter modified versions of both
notions will lead to the theory of definiteness advanced in this thesis.
In the next sections we will critically review each line of thought, focusing on the central
assertions the main adherents of each type of theory have made.
1.3 Uniqueness-based theories
After having presented grosso modo both the notions of uniqueness and familiarity, as well as
their imanigable relation(s) to definiteness, let us now endeavour to situate ourselves within the
vast amount of literature that establishes a strong connection between definiteness and
uniqueness, positioning the latter as the explanatory feature of definite noun phrases.
In the famous Port Royal grammar (1660), which is considered the first grammar to have a
modern approach to language especially due to its attempt to present a universal grammar,

18
Arnauld and Lancelot claim that the definite article (in French, which is the language for which
most examples are provided in the Port Royal Grammar) indicates uniqueness. One of their
examples- chosen to be impressive, as they say- is the following:
10. Charles est le fils de Louis.
With respect to 10 they claim that it shows that Charles is an only child, that Louis has no more
children, given that the definite article is associated with what they term singularity/ uniqueness,
which rules out multiplicity, in the sense of being more than one, while the indefinite (un fils de
Louis) would indicate oneness, that is, some number, which would presuppose multiplicity.9
1.3.1 A Russellian perspective
About two and a half centuries later, in the modern era of philosophy of language, we can find a
revival of the notion of uniqueness as introduced in the work of Bertrand Russell10(1905).
Russell’s work focused on the importance of the use of logic, the tools that predicate logic
provides, such as quantifiers, for the clarification of several issues in mathematics and
philosophy. His analytic method can be seen in “On denoting”, where he employed logical
formulae of sentences in order to determine the interpretation of definite descriptions. The
fundamental idea of Russell lies in substituting the view that definite as well as indefinite
descriptions are referring expressions with the view that both kinds of expressions are to be
represented by existential quantifiers, their difference being placed in that definite descriptions
also involve a uniqueness condition, which is represented via a universal quantifier. This
proposal marked the whole subsequent literature on the subject in uniqueness-based theories,
which on the basis of standard logical tools, almost always treat (in)definite noun phrases as
quantificational. According to Russell’s account when we use a definite description such as the
King of France as in the following sentence we make the three subsequent claims:
11. The King of France is bald.
a. There is a king of France.
b. There is not more than one king of France.
9 In fact, le fils de Louis can be also used even if Louis has more than one children. An example they use from the
Bible in order to strengthen their claim is le fils de Dieu, which can only designate the unigenitus son of God
stressing again the relevance of uniqueness in definite noun phrases.
10 Russell’s motivation for discussing definite descriptions can be divided into metaphysical, semantical and
epistemological. The metaphysical motivations concern the issue of how someone can use a nondenoting expression
like Pegasus in a meaningful sentence, the semantical motivations concern descriptions and their possible
interchangeability and the epistemological motivations concern the issue of knowledge. Russell distinguished
between objects that we are directly acquainted with and objects that we only know under a description.

19
c. This individual is bald.
The first claim (a) is an existential claim: the object referred to by the definite description the
king of France is claimed to exist.11 The second claim (b) is a claim about the uniqueness of this
individual: there is only one such entity that satisfies the description the king of France, while
the third claim (c) asserts that the predication of the sentence is.bald is taken to hold of that
individual. Since Russell was working within the realm of truth-conditional12 semantics, what is
to be determined by a representation of a proposition is its truth or its falsity. All the three
claims above are thus taken to determine together the truth or falsity of the proposition. If one
(or more) of the claims fails to hold, then the proposition is rendered false. This means that if
there is no king of France or if there are two or more, or finally if there is exactly one king of
France but he is not bald, the proposition is false. For Russell13, someone that asserts the
proposition the king of France is bald does actually make the three claims as exemplified above.
The representation of the king of France is bald in predicate logic runs as follows:
12. !x [(F(x) & #y(F(y) 2 y=x) & G(x)]
with F standing for be the king of France and G for be bald
This logical formula above can be analyzed into the three following statements:
a. there is an x such that x is the King of France (!x(Fx))
b. for every y, if y is the King of France, then y equals x (i.e. there is at most one
King of France) (#y(F(y) 2 y=x)
c. x is bald. G(x)
The sentence above thus would receive the following representation:
13. !x [(King-of-France(x) & #y (King-of-France(y) 2 y=x) & Bald(x)]
11 Claim (a) is also true for indefinite descriptions according to Russell.
12 Truth-conditional semantics is defined as the branch of semantics that takes the meaning of a sentence to be
describable in terms of its truth conditions. According to this line of thought to know the meaning of a sentence is to
know its truth conditions, which amounts to the need to know what the world must look like if the sentence is to have
the value ‘true’. Otherwise the sentence is rendered ‘false.’
13 Although Russell is the main defender of uniqueness-based theories of definiteness, in On Denoting we trace some
doubts with respect to the absolute explicative power of uniqueness, a point, which to our view is of great
importance. These doubts are expressed by Russell as follows:
Now the, when it is strictly used, involves uniqueness; we do, it is true, speak of `the son of So-and-so' even when So-
and-so has several sons, but it would be more correct to say `a son of So-and-so'. Thus for our purposes we take the
as involving uniqueness.

20
In Principia Mathematica (1910-13), Russell and Whitehead affirm that descriptions, which are
‘incomplete symbols’, with ‘no meaning in isolation’, do not stand for things and they introduce
an inverted Greek iota, hence the recurrent term iota operator, in order to formally represent the
interpretation they give to definite descriptions. A description is then represented by a quasi-
singular term of the form (%x).x, which can be read as ‘the unique x which is .’14.
Russell’s article on definite descriptions has been highly influential, leading many philosophers
to adopt or criticize his theory. If we observe the analysis of the formula Russell gives to
definite descriptions as presented above, we can easily discern the different nature of the claims
involved. This point has been the focus of the criticism made by Strawson (1950), who
challenged thus the truth conditions given by Russell for definite descriptions. Strawson
discussed the different nature of the first two claims, the existential (a) and the uniqueness claim
(b), and the third claim, that is, the surface predication (c). To illustrate this difference, he asked
himself what would happen if one of the relevant claims did not hold. He argued that if claim
(c), that is, the predication, does not hold, then we regard the utterance false and the negation of
the sentence (The king of France is not bald) to be true. If, on the other hand, one of the first
two claims, the existential or the uniqueness one does not hold, then according to Strawson we
do not have a false utterance, but rather a sentence that lacks a truth value; it is neither true nor
false. So, claims (a) and (b) seem to be background assumptions or presuppositions of definite
descriptions instead of assertions as Russell defended. Subsequently, for the sentence to be true
or false, for the matter of truth-value to arise, we need to make sure first that the presuppositions
hold15. According to Strawson thus the three claims differ in strength and, on the basis of
judgments of native speakers, he argues that when someone claims that the King of France is
bald, claims (a) and (b) act as presuppositions16. Strawson’s arguments and his treatment of
uniqueness reveal that for him uniqueness is sort of a pre-condition, a presupposition for the use
of definites.17
More recent work in the literature of philosophy of language and logic reveals that there have
been scholars that continued to argue for a uniqueness account of definites. Barwise and Cooper
(1981:181) in their paper on generalized quantifiers18 argue that “the blond man does not denote
14 Superficially, the iota operator (%x) is a variable-binding device for forming a singular term from a formula .x.
15 This was the view actually held by Frege (1892) before Russell presented his theory on definite descriptions.
16 The issue of whether to treat uniqueness as a presupposition, entailment or implication will be not treated here in
detail.
17 Other criticisms to Russell (1905) within the phisolophy of language literature include Donnellan (1966) and Searle
(1969), which will not be handled thoroughly here.
18 The aforementioned theories use standard predicate calculus, employing the two basic quantifiers to natural
language, namely, the existential (!) and the universal (#) one. A major departure with respect to quantification in
language was initiated by Barwise and Cooper, who developed the notion of generalized quantifiers, based on the
work by Mostowksi (1957) and Montague (1974). The motivation for the critique of standard logic was that many
natural language quantifiers, for example, many, could not be represented with the means of predicate logic, so a new

21
anything at all unless there is a unique blond man in the state of affairs represented by the
model” and that the determiner ||the|| is a function with domain the set of sets with exactly one
element and, consequently, a noun phrase containing a definite article is undefined if it does not
denote exactly one entity. According to Barwise and Cooper the definite article belongs to the
group of definite determiners (comprising the, demonstrative pronouns and both), which form a
subset of strong quantifiers19 and are argued to presuppose the existence of their referent. A
revival of the uniqueness-based theory of definiteness can be also found in Neale (1993), who
provided an extended version of the Russellian theory, while a recent defender of a uniqueness-
based treatment of definite descriptions is Barbara Abbott, who has dedicated a series of articles
to this issue (1999 and subsequent work). Her account basically is based on arguments against
familiarity, so the details of her proposal will be discussed after the presentation of familiarity-
based accounts.
What are the consequences of treating a definite description as unique, what are thus the
consequences of claim (b) of Russell’s analysis? Namely, that if we have a definite noun phrase
that can be applied to more than one individual the whole utterance is rendered false. Were we
to find two individuals and not one satisfying the descriptive content of the predicate, this would
necessarily question the uniqueness of the definite. The use of logical quantifiers makes it
explicit there can be no more individuals involved, the one that satisfies the descriptive content
of the definite noun phrase is to be unique, in the sense that if we find an x that satisfies the
descriptive content and if we find an y that also does, it can only be the case that x and y are the
same object. The rigidity of logical formulae does not though correspond to the actual use of
definite descriptions in language, an issue to on which we focus in the subsequent sections.
1.3.2 Uniqueness, maximality, inclusivenes
Despite the fact that the examples we have considered in the previous pages, as well as
Russell’s examples, all involve count nouns in the singular, that is, definite descriptions that
denote discrete and singular entities, the definite article can also co-occur with count nouns in
the plural as we saw in 7 as well as with mass nouns as below:
14. The sand was very hot.
conception of the notion of quantifier was necessary, a notion that would both represent all the natural language
quantifiers and be closer to the syntactic structure of sentences. This notion was termed generalized quantifier. The
basic claim of this theory is that quantifiers correspond directly to noun phrases (not only to determiners like all), that
is, they correspond to the whole phrase all men. A quantifier can be thought of as making an assertion about sets.
19 The weak/strong distinction as a property of quantifiers was the distinction on which Milsark (1977) placed his
generalizations about quantifiers. Depending on whether a quantifier can appear in there-insertion contexts or not,
the quantifier is termed as weak or strong respectively. The weak ones include {some, a, two, many, no} and the
strong ones include {the, every, neither, most, not all, both}.

22
Here lies thus a further criticism that can be made of the Russelian account. In order to account
for plural or mass nouns co-appearing with the definite article20 we would then be forced to
stretch the limits of uniqueness, since when applied to plural or mass nouns the resulting
interpretation fails to describe the interpretation attested. A possible account of these facts
would be to expand the notion so as to include these cases or to replace it for inclusiveness,
totality or maximality (the corresponding suggestions have been made by authors like Hawkins
1978, Sharvy 1980, Kadmon 1987). Hawkins (1978:161) gave the following definition for the
notion of inclusiveness as an extension of uniqueness:
This property of the definite article to refer to all the objects or all the mass in the
pragmatically limited domain of quantification, whereupon the sentence as a whole
makes some claim about these objects, I shall refer to as inclusiveness.
Sharvy (1980:623) pointed out that “the primary use of ‘the’ is not to indicate uniqueness.
Rather, it is to indicate totality; implication of uniqueness is a side effect”. His proposal,
maintaining the spirit of the Russellian proposal, consisted in the claim that the definite article
(as employed with singular, mass or plural nouns) denotes the sum or totality of all that to which
the predicate applies and that uniqueness implications is only a side effect. The introduction of
the notion of maximality is applied in this way: a definite description is maximal if it denotes
the set of things that satisfy the definite description. Maximality defined as above would include
thus plurals and mass nouns, as well as singulars with the caveat that in the case of singulars it
would amount to a singular set, that is, to uniqueness (see Kadmon 1987 for details).
1.3.3 Existential claim, reference and specificity
The issue of whether the entity denoted by the definite description actually exists in the world
we live in is answered affirmatively by someone retaining a Russellian account, for whom the
presupposition of existence of the referent of the definite noun phrase is part of the truth
conditions that determine the interpretation of a definite description. As can be seen in claim (a)
of the Russellian proposal we saw above, an existential statement about the referent in question
is made, a statement concerning objects in the world. The referent of a definite noun phrase
need not be a real object, with material existence. It may be deduced, but it need not be treated
as a necessary part of the meaning of the definite article. It may not be part of the assertion
made, but, as Strawson (1958) suggested, it may be a presupposition. Were we to ask, whose
uniqueness is predicated to be unique? The referent’s, Russellians claim. What is the referent?
20 I will not go into detail here since in this thesis I will be mainly dealing with singular definite descriptions for
reason of space and simplicity. I assume that my claims will remain valid for plural definite descriptions as well.

23
Some object in the actual world. Skepticism about reference as the basic function of noun
phrases has lead to the reconsideration of the existential claim and the introduction of a new
concept that has been called discourse reference on the basis of work of Kartunnen (1976),
which was further developed by Heim (1982), as we will see below. The actual existence of an
object tied to the use of definite descriptions is questioned on the basis of more complex
examples than the ones we have seen thus far. These examples involve quantification and
definite descriptions, as for instance in
15. Every child got the present he deserved.
The speaker here cannot and need not to be aware or be in the position to identify the actual
present every child got. Examples with superlatives can also make the same point. Does the...
presuppose that such a thing exists? And if the answer is affirmative, where does the thing
exist? In the actual world? Or in the discourse? Such entities may be called discourse referents
and may be represented as referential indices or numbers depending on the version of the theory
in question. How linguistic expressions depend on the real world and how they are related to
each other, that is, the possibility of co-reference between linguistic expressions is closely
linked to the issue of definiteness, since definite noun phrases are supposed to co-refer with
other linguistic expressions mentioned before in the discourse.
What happens when the entity in question does not exist? Since Russell argues that when we
use a definite noun phrase we assert the existence of its referent (we may add in the real world),
if we use a phrase that does not denote anything, the utterance should be judged to be false with
respect to its truth conditions. Russell’s judgments seem to be very strict. An alternative answer
would be to claim that the issue of the truth or falsity of the utterance in question does not arise,
if the existence claim is not satisfied, as Strawson (1958) has already proposed.
Another related issue is specificity. The speaker may felicitously use a definite noun phrase
without knowing actually to whom he is referring to, since definites seem to be able to be
specific or non-specific. A non-specific definite according to some authors would be the
occurrence of the culprit as in the following sentence:
16. The police look for the culprit.
In 16 the speaker employs an intensional verb look for with the definite the culprit, and it may
either be the case that the speaker knows who the culprit is and refers to him/her by the definite
noun phrase or it may also be that the culprit does not refer to a certain known individual, that

24
is, it is non-specific. In the latter case the definite noun phrase receives an attributive
interpretation, which amounts to the description whoever individual satisfies the property of
being the culprit, a point that has been discussed especially by Donnellan (1966), but will not be
examined in more detail in this thesis. Enç (1991) provided also a thorough discussion of this
issue of specificity. Another set of non-specific definites, to which we will return later on in
more detail is the one comprised by weak definites as the ones attested in examples (examples
from Carlson & Sussmann 2005) like the following:
17. Mary went to the store.
18. I’ll read the newspaper when I get home.
In the above examples the store or the newspaper do not refer specifically to a particular object,
but rather receive an attributive interpretation.
1.3.4 Domain restriction
At the beginning of this chapter we alluded to the fact that the original analysis of Russell
(1905) seems to be very inflexible in the sense that unique definite descriptions are thought of
as applying to one object evaluated from a perspective that encompasses the whole world. Let
us look at the following example:
19. The table is covered with books.
A sentence like 19 can be used even though the speaker does not think that there only exists one
table in the world; s/he may even be sure that there exist more. According to Russell, the
uniqueness implication is determined strictly and solely by the descriptive material in the
definite NPs, a fact that can be overridden by a more ‘realistic’ uniqueness (to use Kadmon’s
1990 term), which would limit the uniqueness of the definite to the specific universe of
discourse, a common assumption made nowadays in the literature on definiteness. In what
proceeds I will follow Roberts (2003:292), who introduces the term pragmatic enrichment of
the descriptive content of the NP to describe this phenomenon, seen as an instance of the
pervasive phenomenon of domain restriction in the interpretation of logical operators.
1.3.5 Summary of the uniqueness-based theories
The above discussion has served to show the basic motivation for the use of the notion of
uniqueness for the treatment of definite noun phrases. Some basic criticisms of Russell’s

25
analysis have been pointed out and a set of crucial issues for uniqueness accounts have been
identified. Uniqueness was found to be easily modified to maximality in order to extend to
plural or mass nouns, the notion of reference was discussed, while the existential claim of the
referent was taken to be rather a presupposition than an assertion. Before entering more into the
discussion of the problems, some of which have been already foreshadowed, that a uniqueness
account may face, let us turn to the other side, the one of familiarity.
1.4 Familiarity-based theories
Familiarity is a notion that has been associated to the use of the definite article since antiquity.
The roots of familiarity lie in the first attempts to describe and explain definiteness, which are
found in the work of Apollonius Dyscolus (2nd century AD) on Greek. Familiarity remained an
recurrent notion in the work of English grammarians like Christophersen (1939) and Jespersen
(1949), while it was given a even fuller account in the work of Hawkins (1978) and it has been
shaped within a recent formal account in the work of Heim (1982), continued by the work of
Roberts (2003).
The hypothesis of familiarity (although not defined in these terms) goes back to the work of
Apollonius Dyscolus, who studied the occurrence of the definite article in Ancient Greek and
tried to establish a basis for the distribution of definite noun phrases in contrast with bare noun
phrases21. Apollonius Dyscolus argues (Book 1.43(VI) p. 33) that:
The special function of the article [...] is anaphora (reference), that is, the presentation
of a person [i.e. the referent of an NP] previously identified [i.e. known to speaker and
hearer]22.
In his analysis we discern the concept of familiarity or identification (to the speaker and hearer)
on the basis of explicit previous mention in the discourse.
Familiarity-based theories fundamentally try to account for the fact that in a discourse like the
one below, where the speaker B uses the definite the book, s/he is referring to an entity known
to both speaker and hearer, which is the one referred to by speaker A in the preceding
discourse23:
21 It is of crucial importance to pinpoint here that Ancient Greek lacked indefinite articles, possessing only the
definite article o,i,to.
22 Comments in square brackets are by the editor of the translation in English, F. Householder.
23 In the discussion of this type of theories we will immediately see that the explicit mention of a referent is not the
only way to license a definite.

26
20. A:Yesterday I bought LeClezio’s first book, called The interrogation.
B: That’s nice! Could you pass me the book to have a look at it?
Speaker B cannot use the definite the book to refer to another book; in this discourse it is
supposed to refer to LeClezio’s book mentioned by his/her interlocutor.
1.4.1 Grammarians
The idea of familiarity, whose first employment was traced to the Ancient times, was further
developed by Christophersen (1939), who was probably the first to present a coherent theory of
definiteness based on the notion of familiarity defined in these terms. Christophersen notices
that the employment of the notion of familiarity goes back to the work of Brown (1861) or
Hansen (1927), but that a common core based on familiarity can be found under different names
in other authors as well. Christophersen used the critical concepts of speaker and hearer, the
basis of linguistic communication, and grounded his analysis on the common ground these
individuals -the speaker and the hearer- form together. He argued that the use of the definite
article in English directs the hearer to the referent of a noun phrase by indicating that this
referent is familiar to hearer as well as to speaker. He writes (Christophersen 1939: 28):
Now the speaker must always be supposed to know which individual he is thinking of;
the interesting thing is that the the-form supposes that the hearer knows it too. For the
proper use of the form it is necessary that it should call up in the hearer’s mind the
image of the exact individual that the speaker is thinking of. If it does not do that, the
form will not be understood.
A condition of the use of the is that there is a basis of understanding between speaker
and hearer. This basis comprises the subjects and things known by both parties, and the
speaker as the active party must consequently adapt his language to the hearer’s state of
mind. If he wants to be understood it is important that he should not use words and
phrases which the hearer is likely to misinterpret.
Christophersen argues that the creation of a common basis for understanding, which is essential
for any communication and, especially, for the understanding of definite noun phrases, may be
founded on different types of contextual bases.24 For Christophersen the function of the definite
article may be called familiarization, if by this we mean ‘the indication (not the creation) of
24 Contextual bases may be explicit, implicit or situational. See Christophersen (1939) for more discussion of each
case.

27
familiarity’ (1939:70). As for an attempt to define familiarity25 we may look at the following
passage (Christophersen 1939:72):
The article the brings it about that the potential meaning (the idea) of the word is
attached a certain association with previously acquired knowledge, by which it can be
inferred that only one definite individual is meant. That is what is understood by
familiarity.
Commenting on the above citation, we see that for Christophersen familiarity means that
previously acquired knowledge is essential for the proper understanding of definite noun
phrases. On the basis of this knowledge, the uniqueness of the entity in question can be
described as an inference, an implication of that first claim.
As can be appreciated by a passage from Essentials of English grammar (1933), the thread that
conducts Jespersen’s investigations on definiteness also rests on the notion of familiarity. We
read the following with respect to the definite article (1933:162):
The chief use of the definite article is to indicate the person or thing that at the moment
is uppermost in the mind of the speaker and presumably in that of the hearer too. Thus it
recalls what has just been mentioned... Or else the whole situation is sufficient to show
what is meant.
In the above passage what is important is the uppermost salience Jespersen gives to the referent
of the definite noun phrase both in the speaker’s and the hearer’s minds and the fact that he
refers to the immediately previous discourse as the locus where this recollection takes place.
The notion of familiarity has been further developed by Jespersen in his modern English
grammar based on historical principles vol. VII, where he distinguished three “stages of
familiarity” with respect to noun phrases and the presence/absence of determiners. Familiarity
is defined as “knowledge of what item of the class denoted by the word is meant in the case
concerned” and three stages are recognized on the basis of the employment of articles: complete
unfamiliarity, nearly complete familiarity and complete familiarity. While complete
unfamiliarity requires the indefinite article in English, the definite article is needed in nearly
25 Christophersen (1939:73) acknowledged the fact that familiarity may not be always direct, but that a rather indirect
association may take place in the hearer’s mind. He provides an example of associative anaphora, while talking of a
certain book uttering The author is unknown.

28
complete familiarity and no article is necessary when complete familiarity has taken place26.
Jespersen’s second stage, the stage of nearly complete familiarity, where the use of definite
article is obligatory, is the most interesting one for the purposes of this thesis. He argues that the
determination may come from the linguistic context (in cases of anaphoric use of the article) or
from the extra-linguistic context (in the case of things we are in general familiar with like the
president of a certain country or the sun of our solar system), opening up the possibility for a
weakening of the licensing of familiarity. Jespersen provides us with the following description
of the function of the definite article (1949: 479):
The definite article plus a substantive in the singular denotes one individual (supposed
to be) more or less familiar to the speaker or writer: Some image or notion of the thing
or person denoted by the substantive is (supposed to be) already found in the
consciousness of the speaker or writer before he makes the statement.
For Jespersen the familiarity ascribed to speaker or writer is essential for the understanding of
the definite article. He alludes to the consciousness of the speaker or writer as the locus of an
image or notion of the thing or person denoted by the noun.
1.4.2 Contemporary semantic and pragmatic theories: Hawkins (1978), Heim (1982)
and Roberts (2003)
In modern linguistics with its further subdivisions of labor, we find ourselves in a situation that
distinguishes different levels in grammar such as syntax, semantics, and pragmatics. Within the
semantics as well as the pragmatics literature (which is closely linked to the philosophy of
language and therefore often treats the same subjects through a slightly different angle) we
evidence that there is a lot of work associated with the definite article. In this subsection I will
26 Jespersen (1949: 417) writes:
Stage 1: complete unfamiliarity (or ignorance): indefiniteness
i. unit word (singular). indefinite article (e.g. an apple)
ii. mass word. Zero article (e.g. he drinks milk every morning)
iii. unit word (plural). (e.g. he eats apples every morning)
Stage 2: nearly complete familiarity. The word in question still requires the. The necessary determination is given
either by the (linguistic) context or by the whole (non-linguistic) situation:
i. by the context: once there lived an old tailor in the village. The tailor was generally known in the
village as the crook.
ii. by the whole situation: the government [of the time, of the country concerned], the sun [that one
among many suns of which it is most natural to speak]
Stage 3: complete familiarity: familiarity so complete that the use of the definite article is not needed, it is redundant,
as with proper nouns, vocatives and a few other cases. The most important cases:
i. direct address (e.g. come along, boy)
ii. proper names (e.g. John, Mr. Brown)
iii. God
iv. Father, uncle, baby, nurse and other members of the family circle
v. Dinner and other regular meals
Church, prison, town (chiefly in prepositional phrases denoting “professional tours” like go to college)

29
review the proposals by Hawkins (1978) and Heim (1982), as well as the development of the
latter one by Roberts (2003).
Working within the framework of speech act theory27, Hawkins (1978) proposed a theory for
(in)definiteness and reference in language grounded on the importance of familiarity for definite
noun phrases. He argued that the typology of the uses of the points to the generalization that the
referent of a definite noun phrase must be part of a shared set, that is, the entities known by the
speaker and hearer to constitute either the previous discourse, the immediate or larger situation,
or an association set. According to Hawkins the definite article instructs the hearer to infer
which shared set is intended and locate28 the referent in it (that is, understand the referent to be a
part of it). The importance Hawkins ascribes to familiarity for the interpretation of definiteness
is structurally linked with the rejection of uniqueness (1978:158) as part of the meaning of the
definite article. He argues that uniqueness is only a result of the fusion of the meaning of the
definite article with singularity or oneness of the count noun in question, making the count noun
responsible for the quantity conveyed.
Hawkins’s thesis had a major influence on subsequent work- descriptive and beyond. The
classification of different uses was used as a guide for the presentation of the occurrence of
definite articles across languages and they will provide the basis for the presentation of the
Greek data to be given in the next chapter. I do not enter here in the discussion of each use,
since this will be a substantial part of the next chapter on definite noun phrases, where this
classification will provide the basis for the presentation of the Greek data to be discussed in this
thesis. The main uses Hawkins identifies are the following: anaphoric, situational (based on the
visible or immediate situation), larger situational (based on specific or general knowledge) and
associative anaphoric uses. Hawkins furthermore points out unfamiliarity uses, that include
establishing relative clauses, associative clauses, NP-complements, nominal modifiers
(examples of which we will see in the section discussing the problems familiarity-based theories
face), and he also outlines the use of the definite article with ‘unexplanatory’ modifiers like
27 In pragmatic theory, one of the most important accounts given to discourse has been proposed by Austin in his
seminal book How to do things with words (1962). The influential insight that this work offered to pragmatic theory
was the idea that can be captured by the following motto when we say something, we do something, which
emphasizes the fact that sentences do not always state facts, but they are also used to perform different kinds of acts.
Austin emphasized this fact by looking closely at perfomative sentences of the form “I baptize this child of God
Petro” and he argued that when the circumstances are appropriate in certain ways, then the utterer of this sentence
will have done something special, namely, s/he would have performed the act of naming the child. In How to do
things with words he outlined the theory of speech acts, which is founded on the distinction between locutionary,
illocutionary and perlocutionary acts- the illocutionary acts being the only true “speech acts”- this theory was later
developed by John Searle, who offered a classification of speech acts in his book Speech Acts. An Essay in the
Philosophy of Language in 1969.
28 The importance of the act of localization in his theory has lead Heim (1982) to call Hawkins’ theory location
theory.

30
ordinals, superlatives and adjectives like same, only or unique, which all show uniqueness
effects.
As mentioned before, Hawkins’ approximation to the issue of the definite article uses speech act
theory. So, the summary of his location theory (1978:167) includes the speech acts a speaker
performs when using a definite article29. The speaker
(a) introduces a referent(s) to the hearer
(b) instructs the hearer to locate the referent in some shared set of objects
(c) refers to the totality of the objects or mass within this set which satisfy the referring
expression.
The pragmatic account of Hawkins was further developed by a formal semantic and pragmatic
account that can be found in Heim’s (1982) theory, which is considered an important revival of
the familiarity theory of definites. Heim argued against the view that indefinites and definites
are quantificational expressions and treated both of them as free variables to be bound by some
operator available in the context. Heim developed this view and gave a formal semantic account
of definiteness incorporating many pragmatic aspects of interpretation within a theory of the
discourse30 she introduced and called File Change Semantics. This theory introduces a new
conception of semantics and pragmatics that Heim endorsed and is a variety of the notion of
dynamic semantics. In the field of Montague semantics (1974) the standard core semantic
property has been truth conditions, which was replaced by Heim by the notion of context
change potential. Sentences are, according to Heim, not only to be interpreted with respect to
their truth conditions (amounting to how the world must be in order for a proposition to be true),
but they also have a strong relation to the context, in which they are uttered, which is constantly
changing, since each new utterance adds something new to the context. Sentences are thus
treated as denoting functions from contexts to contexts and someone can treat presuppositional
conditions (such as the familiarity condition to be presented here) in terms of saying that a
sentence can denote a partial function from contexts (just those that meet the familiarity
condition) to contexts.
29 The successful application of the above-mentioned speech acts is conditioned by the following appropriateness
conditions:
(1) set existence condition: the speaker and hearer must indeed share the set of objects that the definite referent
is located in
(2) set identifiability condition: the hearer must be able to infer either from previous discourse or from the
situation of utterance which shared set is actually intended by the speaker
(3) set membership condition: the referent must in fact exist in the shared set which has been inferred
(4) set composition conditions: (i) there must not be any other objects in the shared set satisfying the
descriptive predicate in addition to those referred to by the definite description, i.e. there must not be fewer
linguistic referents referred to by the definite description than there are objects in the shared set; and (ii) the
number of linguistic referents referred to by the definite description must not exceed the number of objects
of the appropriate kind in the shared set; and (iii) the hearer must either know or be able to infer that the
intended object has the property that is used to refer to it in the descriptive predicate.
30 Hans Kamp (1981) almost in parallel with Heim introduced his theory of discourse called Discourse
Representation Theory which shares important points in common with the theory of Heim.

31
Heim argued against the view that indefinites are to be represented as existential quantifiers, but
also against the view that definites are to be represented by the combination of an existential
plus a universal quantifier (the basic claim of a Russellian analysis). Indefinites and definites are
for Heim similar in the sense that they are both non-quantificational. She argued that (in)definite
noun phrases introduce a free variable31 to be bound by some operator available in the
discourse. An indefinite for instance would be represented as follows:
21. A cat came in.
cat (x) $ came.in (x)
A definite would receive basically the same treatment, thus the following would amount to the
same open formula, differing in the conditions attached to each nominal:
22. The cat came in.
cat (x) $ came.in (x)
The difference between definite and indefinite NPs does not lie thus in the nature of the
quantifier(s) attached to them, but rather, on the one hand, on the different conditions attached
to them. indefinites are associated with novelty and definites with familiarity, and on the other
hand, on the different requirements on their descriptive content, that is, in definites, the
descriptive content of the NP is presupposed, whereas in indefinites it is (merely) asserted
(1982:233). The combination of both conditions, novelty/familiarity and presuppostion (or not)
of descriptive content gave rise to the following version of her theory of definiteness
(1982:36932):
23. The Extended Novelty-Familiarity Condition:
For a logical form . to be felicitous w.r.t a context C it is required for every NPi in .
that:
(i) if NPi is [-definite], then i % Dom(C)33,
(ii) if NPi is [+definite], then
(a) i " Dom(C), and
(b) if NPj is a formula, C entails NPj.
31 Here we only give the representations with the open formulas, which are then bound by some available operator
(the existential closure introduced by Heim or some other operator). In the fourth chapter we will return to Heim’s
analysis of indefinites presenting it in more detail, given that we will rely on it for the analysis of the indefinite
opjosdhipote (see chapter 4, page 170ff).
32 I use here Robert’s (2003) slightly modified version of Heim’s condition, which basically substitutes context for
Heim’s file with the understanding that this preserves her intentions.
33 Dom(C) is treated here as a set of indexed variables (corresponding to referents in the domain of discourse).

32
This condition amounts to the following: a definite is used to refer to an entity that is already
old or familiar at the current stage of the conversation, something that is essentially part of the
domain of the discourse or is entailed by the prior context it is included in, while an indefinite is
used to introduce a new referent, so it cannot be part of the domain34.
Heim critically discusses (1982:299) the traditional view of definite descriptions that takes
definites to be referring expressions35, a condition that can be easily shown not to hold. Heim
grounds this critique in Kartunnen (1976) and she claims that there are uses of definites such as
in the following sentence, where reference to some particular entity is not needed for the
sentence to be felicitous:
24. Every girl got the present she deserved.
As we already commented before, in order to solve this problem, the introduction of a new
notion, discourse reference, came to replace the more general notion of reference. Heim’s
familiarity theory in essence dissociates familiarity from reference understood in the “classical”
sense and makes use of discourse referents, which in her model are represented analogous to
file cards. A mental file containing cards is the metaphor Heim used for modelling the
interpretation of utterances. A card contains information about a particular object, a particular
discourse referent. The difference between definite and indefinite noun phrases lies then in their
relation to the cards36, so within her File Change Semantic theory Heim argues the following
(1982: 276):
25. For every indefinite, start a new card; for every definite, update a suitable old card.
As we will elaborate in more detail in the section discussing the problems familiarity-based
theories, Heim’s innovative treatment of definites cannot cover all the cases of definites and
therefore she introduced an additional operation called accommodation, as defined by Lewis
(1979), in order to be able to account for instance for cases that involve associative uses of
definites and cases where the relevant situation amounts to something larger than the actual
discourse, as in expressions like the president.
34 According to Heim the context is a file of information held in common by the participants of the discourse. She
adopts a modified version of Stalnaker’s (1979) notion of the common ground, which includes the domain of the
discourse, that is, technically a set of indices, which are considered to be discourse referents that the interlocutors
know.
35 See above the short discussion on the presupposition of existence and referentiality of definites defended by
Russell.
36 This does not amount to arguing that cards are only influenced by the actual discourse. A new file can be created if
a new entity enters the universe of discourse. For example, if there are two speakers and a cat enters the scene, they
can refer to the cat by the definite noun phrase the cat considering that a file already was created when the cat
entered.

33
On the basis of the theory of Heim, Roberts (2003) proposed a theory of definiteness she called
informational uniqueness, which will provide the grounds for our own conception of familiarity.
The main contribution of Roberts lies according to our view in that she builds on the theory of
Heim, refining the notion of familiarity and clarifying a lot of confusion around this notion. On
the basis of examples that clearly show that there is no need for explicit mention in the
preceding discourse in order for a felicitous use of a definite noun phrases (we will provide
some relevant examples below and more examples in Greek in the next chapter), Roberts argues
for a re-elaboration of the notion of familiarity. She proposes to distinguish between strong and
weak familiarity37 and argues that strong familiarity is taken to involve explicit previous
mention of the entity in question, while within weak familiarity she embraces other uses,
wherein the existence of the entity in question need only be entailed by the (local) context of
interpretation. Roberts’ notion of weak familiarity subsumes strong familiarity but is more
inclusive in the sense that discourse referents may be introduced not only linguistically (via
explicit mention) but also non-linguistically on the basis of contextual entailments (and
perceptual accessed information) alone. Her taxonomy of familiarity is presented as follows
(2003: 304):
26. Familiarity:
(a) strong familiarity: the NP has as antecedent a discourse referent introduced via the
utterance of a (usually) preceding NP
(b) weak familiarity:
i. the entity referred to is perceptually accessible to the interlocutors
ii. the entity referred to is globally familiar in the general culture or at least among
the participants in the discourse, although not mentioned in the immediate
discourse
iii. introduction of the NP’s discourse referent is licensed solely by contextual
existence entailments38
iv. weak familiarity is guaranteed by giving a functional interpretation to the definite
description (which function may have to be accommodated) with the intended
argument(s) both familiar and highly salient (bridging)
(emphasis in italics mine)
So, weak familiarity would be crucial to predict examples like the following (I introduce in
parentheses the relevant subclasses of weak familiarity):
37 As Roberts (2003) also notices this employment of the notion weak familiarity differs from the use of the same
label used by Condoravdi (1992) to refer to other phenomena of discourse.
38 This is relevant for definite pronouns (see Roberts 2003 for more discussion).

34
27. Look at the cat! (i)
28. I don’t like the colour yellow. (ii)
29. I don’t remember the beginning of the tune. (ii)
30. I will go to the store later. (ii)
31. Every motel room has a copy of the Bible in it. In this room, it was hidden under a pile
of TV Guides. (iii)
32. We had dinner around nine. The dessert was delicious. (iv)
33. What’s the matter with Ino? Oh, the book she is expecting has not arrived yet! (iv)
On the basis of the taxonomy of familiarity presented above, we see that Roberts argues that all
definite NPs must have antecedents, but that the required antecedent are not only NPs present in
the discourse, but they can also be discourse referents in the (common ground) context of the
discourse. So, according to weak familiarity the existence of the relevant entity needs only to be
entailed by the interlocutors’ common ground, although an appeal to the operation of
accommodation is still necessary for some remaining cases (see below for more discussion).
The characterization of Roberts’ theory in terms of informational uniqueness is supported by the
assumption she posits that uniqueness effects are explained via pragmatic, Gricean principles.
The uniqueness presupposition of definite descriptions is expressed in her theory as the
requirement that sufficient information has been given to uniquely indicate the intended
discourse referent antecedent among all those in the common ground of the participants (2003:
307).39 Roberts’ proposal is stated as below (2003: 308), a statement on which we will base our
own theory:
34. Informational Existence and Uniqueness of Definite NPs
Given a context C, use of a definite NPi presupposes that it has as antecedent a
discourse referent xi which is:
a) weakly familiar in C, and
b) unique among discourse referents in C in being contextually entailed to satisfy the
descriptive content of NPi
39 Roberts argues that robust uniqueness effects arise on the basis of special epistemic circumstances, which involve
examples with definite descriptions whose informational uniqueness must be guaranteed in contexts where the
interlocutors may have knowledge of the entire extension of the relevant CN, examples with titles like The Ohio State
University or the Queen of England or instructions of the sort Find the clown in the next page. Strong familiarity fails
for instance with superlatives, where consistently a functional interpretation arises or due to a Quantity-based
conversational implicature.

35
1.4.3 Summary of the subsection
In the preceding discussion of some of the basic adherents of the familiarity-based account of
definiteness we have seen its evolution from Ancient Greek grammars through its elaboration
by Christophersen (1939) and Jespersen (1933, 1949), its establishment within a speech act
theory as the one by Hawkins (1978) until its development and formal definition by Heim
(1982), which is the basis of the refinement by Roberts (2003) that constitutes also the view on
familiarity to be followed in this thesis, which will be presented after the discussion of the basic
counterexamples each line of thought is confronted with, if taken as the only explanatory feature
of definiteness.
1.5 Discussion
In this section we will provide examples that demonstrate that each of the two lines of thought
we have been discussing cannot account on its own for the whole range of data attested. After
presenting these examples, we will have sufficient grounds, on which to hinge our option for
maintaining modified versions of both conditions, familiarity and uniqueness.
1.5.1 Problems with the uniqueness-based theories
I will discuss here some counterexamples to uniqueness-based accounts of definiteness.
Suppose we follow a uniqueness account for definites. We can explain utterances like the
following:
35. The King of Spain is tall.
35 is unique, true of one individual, because the king of Spain is one. There is though a caveat
to be addressed here: the king of Spain can only be one, the property of being king of some
country can only be ascribed to one person, so that this noun phrase can only denote a
semantically unique entity. Uniqueness is somehow presupposed for royalty. Another example
of this sort would be the following:
36. The sun is shining.
We normally take it that there is only one sun, so 36 is explained if we follow a uniqueness
account. If we limit ourselves to examples like the above, it seems that we can follow a
uniqueness-based theory of definites. Looking closer though to these examples it seems that we

36
need to possess knowledge about how the world is, for example, that kings are normally one per
country or that there is only one sun in our stellar system. Relying on uniqueness as the central
notion for definiteness begins to be problematic though in cases, where the use of definites is
with non-unique entities, for instance, when someone utters:
37. The cat is lovely.
Our knowledge of the world instructs us that there are more than one cats in this world and the
speaker of 37 is most likely not to believe that there is only one cat in the universe. We need to
restrict the domain of quantification to the relevant context each time in order for the utterance
to be felicitous, as we commented in the beginning of this chapter. If the definite the cat is
pragmatically enriched (Roberts 2003), the claim above can be taken to be about the relevant
cat, which is salient is the discourse.
The next paragraphs will serve to show that there are different kinds of examples that involve
the definite article and are not unique, though they are familiar. An argument against
uniqueness-based theories is that definites have frequently an anaphoric use, which depends on
the discourse. In a discourse like the one in 38 (example due to Heim 1982), the wine glass is
not supposed to be unique:
38. A wine glass broke last night. The glass had been very expensive.
Heim (1982) argues on the basis of this example that uniqueness implications seem to disappear
very easily and takes this fact as an indication that they cannot be part of the semantics of the
definite article. If uniqueness were a inherent notion involved in definites, we would expect it to
be a part of every definite description independently of the context.
A similar problem is posed for uniqueness-based theories by examples that present uniqueness
under quantificational phrases, since the definite article bound by the quantifier does not denote
a unique entity as in the example below (taken from Roberts 2003):
39. Every bicycle had a spoke missing from the wheel.
In 39 the wheel does not denote something unique, it seems to have a relational reading; the
implicit argument of this relation is a variable bound by the quantificational subject every
bicycle. Under a uniqueness-based account we would have to assume that bicycles have only

37
one wheel, a fact that as we know is not true for bicycles (where actually the very lexical
meaning of the word bicycle is informative, since it that refers to two wheels).
Another set of examples is comprised by the use of definites with entities that under ‘normal’
circumstances are not unique, for example, body parts like eyes, arms, legs etc. as in 40 below
(see Levinson 2006, who argues for an existence of gradual uniqueness effects in non-unique
definites) or examples that involve ‘stereotypical’ expressions like go to the hospital.40 Let us
see an example like the one below:
40. John was hit on the arm.
In 40 the more common interpretation is not that John has lost one hand, so he has one, or that
he was hit on that unique arm of his, but it seems rather that the definite the arm does not refer
to something unique. Crucially, it seems that in these types of examples, it is not relevant for the
discourse whether John has one or two arms, or on which of the two he got hit. The same
happens with expressions like the one below:
41. Yesterday I was feeling sick and I finally went to the hospital.
In 41 the most probable case is not that the city the speaker lives in has only one hospital, but it
seems that it is not necessary to differentiate among hospitals. It refers to some non-unique and
not necessarily familiar entity, yet the use of the definite is felicitous. These NPs are usually
used to refer to locations that are not relevantly differentiable from other locations denoted by
the same NP. Notice that the speaker must be involved in some stereotypical role in relation to
the hospital, for example, as a patient or as a doctor, but the example would be infelicitous if
s/he would had been there as a choreographer working with a group of patients. Other examples
of the same kind, that frequently involve the movement verb go, are the following:
42. go to the bank, go to the bakery, go to the park, put something on the bookself
Summarizing, the problems that uniqueness-based theories face have to do with examples like
the following:
i. anaphoric use of definites like the one in 38
ii. uniqueness under quantification as in 39
iii. non unique definite descriptions, as in 40-42
40 On this issue and its relevance to forms that involve contracted prepositions and the definite article in German, see
Puig-Waldmüller (2008), where a lot of examples that involve contracted forms of preposition and determiner are
attested with this kind of examples: zur Schule gehen ‘go to.the school’, ins Theater gehen ‘go to the theatre’ or am
Arm sich verletzen ‘to get someone’s arm hurt’.

38
The counterexamples to the theories presented are not language-specific, they do not concern
only English, but also other languages, such as Greek, into which we can reproduce the relevant
examples, as we will see in chapter 2.
There are furthermore some examples that show that definite noun phrases may not be unique,
though they are familiar. These examples turn uniqueness into a non necessary condition for the
use of the definite article, despite the fact that it is a sufficient condition. These are examples
like:
43. (In a room with three windows). It’s hot here. Could you please open the window?
In 43 despite the fact that there are than one windows in the room in question, the speaker asks
the hearer to open one of them, the identity of which is of no interest here, so the use of the
definite the window is felicitous.
1.5.2 Problems with the familiarity-based theories
What a handful of authors, including Fraurud (1990), Birner and Ward (1994) and Poesio and
Vieira (1996) argue is that both the notions of familiarity and uniqueness with respect to definite
noun phrases can be easily questioned on empirical grounds. We have seen reasons to question
uniqueness and now we move on to discuss cases that question familiarity. With respect to
familiarity, Fraurud (1990) and Poesio and Vieira (1996) examine definites in naturally
occurring corpora and on the one hand Fraurud’s results show that the 61% of the definite NPs
she finds are first mention (roughly, without an NP antecedent in the previous discourse),
whereas on the other hand Poesio and Vieira (1996), after classifying the occurrences of
definites, draw the conclusion that more than half of the definite NPs they encounter belong to
first mention NPs, comprising examples of associative anaphora, larger situation use and
unfamiliar ones. If we consider for instance sentences like the following:
44. Aghorasa ena vivlio xtes. O sighrafeas tu ine katalanos.
bought.1sg a book yesterday. the author it-GEN is catalan
‘I bought a book yesterday. Its author is Catalan.’
45. Simera idha to feghari apo to balkoni mu.
today saw.1sg the moon from the balcony mine
‘Today I saw the moon from my balcony.’

39
The definite NP in 44, o sighrafeas ‘the author’ does not refer to an author previously
mentioned in the discourse, but it is a DP that appears to be first mention and felicitous. These
examples could be untangled with an invocation to the operation of accommodation (cf. Heim
1982). The idea lying behind this operation is that there may exist definite noun phrases
referring to new discourse referents, a fact that seems to be opposed to the idea that definites are
familiar, but that these uses may be accommodated as being linked to familiar discourse
referents or as making reference to the context of the utterance. Heim’s operation of
accommodation (based on Lewis 1979) involves a linking or bridging (Clark 1975) operation.
She offers the following requirement on accommodation:
When a new file card is introduced under accommodation, it has to be linked by cross-
references to some already present file card (associative anaphoric) or to the utterance
situation (immediate and larger situation uses).
This necessary resort to an additional operation in order to deal with a wide range of definite
noun phrases is one of the weak points of Heim’s account. The attraction of such a move is
easily debilitated, when we encounter the big number of occurrences of such ‘novel definites’ in
corpora that register natural occurring speech. The possibility of accounting for these cases
under a revised version of familiarity has been showed in the work of Roberts (2003), who gave
a new definition to the notion of (weak) familiarity, under which a lot of novel definites can be
accounted for. Roberts needs though also to resort to the operation of accommodation, in fewer
cases than Heim though, namely, in cases involving endophoric use of the definite as in 46 or in
restrictive relative clauses like 33.
Moving away from cases of associative anaphora and larger situation uses of definite noun
phrases, there still remains another set of examples that addresses difficulties for theories of
definiteness that rely on familiarity as their explicative feature and that resists explanation even
if we adopt the notion of weak familiarity. One characteristic of this set of examples is that they
are comprised by definite noun phrases that contain modification of the noun in question. This
modification can have the form of an adjective, a prepositional phrase or, more usually, of a
restrictive relative clause as in 46 and 47:
46. Stin omilia tis, i Maria anepktikse tin theoria oti ta glosika simia dhen ine afthereta.
in.the talk hers the Maria developed.3sg the theory that the linguistic signs not are
arbitrary
‘In her talk, Maria developed the theory that linguistic signs are not arbitrary.’

40
47. An pas sto domatio, fere mu se parakalo, to meghalo vivlio pu afisa
if go.2sg to.the room, bring.2sg-IMPER me-GEN you please the big book that left.1sg
mesa stin dulapa.
in.the closet
‘If you go the room, please bring to me, the big book I left in the closet.’
The felicitous use of the definite article in cases like the above, where the definite NP refers to
entities that neither have been previously evoked in the discourse (strong familiarity) nor are
assumed to be otherwise familiar to the hearer or inferable from the context (weak familiarity),
provides sound evidence that familiarity is not a necessary condition for felicitous use of the
definite article. The DP tin theoria oti ta glosika simia dhen ine afthereta ‘the theory that
linguistic signs are not arbitrary’ is new information, as well as the DP to meghalo vivlio pu
afisa mesa sti dulapa ‘the big book I left in the closet’. Both examples signal to the same
direction: the definite noun phrases are not familiar, but crucially they are uniquely identifiable.
In fact in example 47 there can be in principle any number of big books in the room, but the one
left in the closet is believed to be the only one possessing the property of being the big book left
in the closet. Birner and Ward (1994) and Roberts (2003) stress the fact that the examples above
cannot be accounted in terms of accommodation since (transferring their arguments for the case
above) books for instance are not usually associated with closets, but with shelves or libraries,
so that there lacks a sufficient trigger for the associative anaphora to take place. As we have
pointed out before it is significant that this set of examples involves modification, that is, rich
descriptive information. The sufficiently rich description, that are often notable for their length
(see Abbott 1999 for a study of written texts) furnishes the ground for the hearer to be able to
fully and uniquely identify a relevant discourse entity, which nevertheless is unfamiliar to him.
In other words, all the examples above display semantic uniqueness effects and furthermore
there is an attributive relation between the NP and the complement of the sort: “her theory is
that linguistic signs aren’t arbitrary”.
The insufficiency of the notion of familiarity alone to furnish a full picture of definiteness may
be evaluated also from a different angle. Familiarity alone cannot license definite noun phrases;
a clear indication that it does not constitute a sufficient condition for definiteness. Suppose for
example we have the following situation, where in the first utterance two new books are
introduced in the discourse and therefore establish two new discourse referents (example
modeled on Birner&Ward 1994):

41
48. I Ino ke i Ariadhni dhulevun sto tmima filosofias sto panepistimio,
the Ino and the Ariadni work.3pl in.the department philosophy-GEN in.the university
ke i kathemia eghrapse ena kenurjo vivlio. #Ta ipolipa meli tu tmimatos
and the eachone wrote.3sg a new book the other members the-GEN department-GEN
zilepsan poli epidhi to vivlio ine poli endhjaferon.
got.jealous.3pl much because the book is very interesting
(‘Ino and Ariadni work for the philosophy department at the university, and each of
them wrote a new book. # The other members of the department got very jealous
because the book is very interesting.’)
In spite of the two books being familiar in the discourse of 48 and, therefore, possible
candidates for subsequent ‘definite’ reference, we observe that the definite singular to vivlio ‘the
book’ is not used in a felicitous way. The hearer has no way to disambiguate between which of
the two books the definite noun phrase is referring to, to the new book written by Ino or the new
book written by Ariadni? Without more descriptive content added to the definite noun phrase,
the use of the book here turns out to be infelicitous providing stable grounds to argue that
familiarity alone is not sufficient for the felicitous use of the definite article.
We draw thus the following conclusion: In some cases, uniqueness proves to be a sufficient
condition in the sense of denoting an entity that is ‘identifiable’ by the hearer, meaning that it is
distinguishable from all other discourse entities, regardless of the fact of being identified. Birner
and Ward (1994) employ examples that involve superlatives, where the tallest boy in my class
refers to a uniquely identiable entity whether or not the hearer is able to attach a name or other
attributes to the referent. The term ‘identify’ is further discussed by Birner & Ward (1998),
since it suggests that an addressee is able to pick the referent out in the world at large. They
argue instead that what is required for felicitous use of the definite article (and most uses of
other definites) is that “the speaker must believe that the hearer is able to individuate the
referent in question from all others within the discourse model” (Birner & Ward 1998:12).
According to them, the term uniquely identifiable is to be replaced by the notion of individuable
within the discourse model, which is a concept very close to the notion we will introduce and
elaborate in this thesis, which we will call count-as-unique.
A fourth set of examples that challenge the sufficiency of familiarity in the explanation of
definiteness involves focus phenomena. Abbott (1999) brought up this particular point in the
discussion by observing that while the definite article is an unstressed element in its regular

42
function, it can be stressed according to pragmatic needs41. According to theories of focus (see
e.g. Rooth 1992), elements in speech are stressed when they are associated with focus, which
alludes to alternatives. When someone focuses the definite article s/he contrasts it with its
alternative, which would supposedly be the indefinite article (or no article at all). Abbott
develops her criticism towards familiarity-based theories by arguing that when a speaker utters a
sentence with a stressed definite article, s/he conveys that the entity in question is unique and
that familiarity seems not to play a role in the interpretation of the noun phrase. To show her
point, she employs examples like the following:
49. The person who could have convinced me to join that club just quit himself.
The speaker in 49 stresses the definite article, because s/he thinks that the entity that has the
property of being able to convince him/her is only one, that is, unique. Abbott (1999) uses
further examples to show that a stressed definite article contrasted with a stressed indefinite one
clearly shows a contrast between uniqueness and plurality and not between unfamiliar and
familiar discourse referents:
50. That wasn’t a reason I left Pittsburgh, it was the reason.
51. Aftos dhen itan enas loghos ja ton opio efigha apo ti Thessaloniki, aftos itan o loghos.
this not was a reason for the which left.1sg from the Thessaloniki, this was the reason
‘This was not a reason for which I left from Thessaloniki, this was THE reason.’
Although in her examples the contrast seems to hold and the observation look to me pertinent, it
should be noted that in 49 the definite noun phrase is modified by a relative clause, so this
example can be compared to the above-mentioned examples with rich descriptive content,
which we have seen that give rise to semantic uniqueness. Except for the cases Abbott uses to
illustrate her claim there are other examples of stressed definite articles as shown below:
52. That was the movie.
In examples like the above the definite noun phrase seems to be equivalent to a superlative of
the sort “the best movie I’ve ever seen”, which is semantically unique, we see therefore that an
attributive reading emerges.42
41 See more on the unstressed nature of the definite article as a functional category in chapter 5. See also there for the
possibility to stress it in Greek.
42 In classical rhetoric this reading is called antonomasia.

43
To summarize, counterexamples to familiarity-based theories43 involve:
i. associative use of definites as in 44
ii. first mention definites (that do not involve a trigger) as in 46 and 47
iii. familiar but not unique referents as in 48
iv. focus phenomena as in 49-52
What emerges from the discussion of the problems uniqueness- and familiarity-based theories
face, is the need for the concurrence of both notions, a point that has also been made by Farkas
(2002) and Farkas and deSwart (2006), who use the umbrella term determined reference to refer
to both maximality and familiarity. I quote (2006:9):
In dynamic approaches such as ours, both properties concern the discourse referent
introduced by the NP in question. A discourse referent u is maximal iff it ranges over all
entities that satisfy its descriptive content. In the absence of a plural condition, maximality
amounts to uniqueness. Whether a discourse referent is maximal or not is encoded in the
discourse referent feature [±Max]. A discourse referent that is marked by [+Max] must
have unique reference if atomic, and maximal reference otherwise. Familiarity concerns
the question whether the discourse referent is present in the input context or not. The
feature [±Fam] on a discourse referent encodes its status with respect to this parameter. A
discourse referent characterized as [+Fam] must be familiar in the relevant situation or
accommodatable therein. For convenience, an NP is said to be maximal or familiar when
the discourse referent it introduces is maximal or familiar respectively. We assume that
both maximality and familiarity play a role in definiteness. As an umbrella term for these
two semantic properties, we adopt the dynamic notion of determined reference (from
Farkas 2002).
Bearing all the preceding discussion in mind, we will present in the next section a new proposal
for a theory of definiteness based on the one hand on the modification of uniqueness as count-
as-unique, that is grounded on the ontology of Badiou (1988) and will be the focus of the
discussion to proceed, and on the other hand on weak familiarity as defined by Roberts (2003),
which we have described in the section about familiarity-based theories above.
43 Despite the fact that in this thesis we are focusing on the definite article, other kind of definites such as personal
pronouns should also be taken intro account. Prima facie counterexamples to familiarity can be also constructed with
personal pronouns, as in the example below taken from Heim 1982:390–391:
i. Every motel room has a copy of the Bible in it. In this room, it was hidden under a pile of TV Guides.
Roberts (2003) also discusses personal pronouns within definites. Our employment of definite comprises only noun
phrases headed by the definite article, especially in the singular.

44
1.6 A new proposal for a theory of definiteness
The theory of definiteness we pursue here could be seen as an attempt to articulate an account of
definiteness in coherence with the development of recent philosophical thought. The focus of
this section will be to show how the philosophical assertions of the French philosopher Alain
Badiou (especially, 1988) can be interpreted with relation to the subject that we are discussing
here (the reader interested in a short introduction to the philosophy of Badiou is referred to the
appendix in the end of the thesis). Let me now introduce the part of Badiou’s work that will be
of interest here. Badiou’s concern with ontological issues is centered on his fundamental interest
for mathematics and set theory. What there is and what the structure is of what there is are
central concerns of any metaphysical consideration. Our interest in the theory of Badiou is
centered on the qualifications he gives to the One and the Multiple, the fundamental issues of
ontology, which we take to be closely connected with the discussion of the notion of uniqueness
in the domain of linguistic definiteness. Badiou’s advances on ontology open up new
possibilities for a theory of definiteness.
According to Badiou, we find ourselves in an impasse if we accept that Being is One (or
unique), since accepting that Being is One, one must claim that what is not One, i.e. what is
Multiple, is not, [a claim that corresponds to the proposal that being is not multiple], a
proposition that is falsified by the fact that what there is, what is presented, is in fact Multiple.
In order to provide a way out from this impasse, Badiou posits that the One is not and that it
exists (merely) as an operation, which he denominates count-as-one. In order to give this
answer Badiou relies heavily on the advances of set theory, which represents the most
formalized and complete framework of axioms of the multiple today. He argues thus that being
is fundamentally pure multiplicity, including infinite chains of multiplicities. Given that the One
is not and on the basis of this definition it follows that multiplicity is the general form of
presentation. Presentation, which roughly amounts to being, is multiple, thus it is something
that cannot be applied to the One, which exists only as an operation, the count-as-one.
Multiplicities are considered to be inconsistent at first and to receive structure by the operation
of the count-as-one, through which multiplicities become consistent. The result is a presented
multiplicity (consistent, since it has passed through the operation of count-as-one), which can be
called a situation.
Bearing all these qualifications of the One and the Multiple in mind we will try to apply
Badiou’s affirmations to a new theory of definiteness. Adopting Badiou’s ontological law that
the One is not and building on his insight of the One being a result of an operation called
“count-as-one” I will first sketch an application of it to uniqueness effects with definite noun

45
phrases. Establishing an analogy between the denial of the being of the One and its replacement
by the operation of the count-as-one, I propose re-interpret the notion of unique involved in
definiteness as “count-as-unique”. This line of thought opens a new path in the theory of
definiteness, namely, the possibility to defend the idea that definite noun phrases are not to be
interpreted as one or unique (strictly speaking), but rather as results of an operation. We may
speak of a unique thing presented (denoted by a definite noun phrase) but the unity by which we
grasp it is extrinsic to its sheer occurrence. This operation can be seen as instructing us to
consider entities, to treat them as if they were to be one or unique, this operation can be seen as
a linguistic indication of this kind of thinking, and languages that have articles provide us with
this indication.
We use thus a definite when we want to refer to something familiar (in the sense of weak
familiarity as understood after Roberts 2003) that furthermore is to be “counted-as-unique” for
the purposes of the discussion taking place. I envisage count-as-unique as a refinement of
uniqueness resembling the corresponding weakening of the notion of familiarity. So, the
account of definiteness I pursue is in need of modified versions of both notions, familiarity and
uniqueness. Uniqueness would amount to the fact that the speaker and hearer are to focus only
on the relevant referent at the time of utterance, which is to be counted unique for the purposes
of communication and, as we will see, this counting operation is of great importance.
This conception of uniqueness is analogous to the restriction to the universe of discourse
employed by semanticists in order to explain that the uniqueness of a definite noun phrase is not
to be understood as applying to the whole universe, but only to the relevant context, to the
universe of discourse comprised by the speakers present at the time of utterance. So, uniqueness
in various accounts has been modulated as context dependent. This is a more general
phenomenon though, related to the domain of individuals (universe of discourse), with respect
to which we interpret a quantificational noun phrase. The domain of discourse generally
depends on the utterance situation and often constitutes only a small subset of all actually
existing individuals. The problem of evaluating uniqueness for definite noun phrases in this
sense is parallel to the problem of evaluating quantificational phrases like every glass was
broken or not a single chair was comfortable, which in everyday speech are not understood with
respect to the whole universe, but only restricted to the universe of discourse. The count-as-
unique condition, as we posit it, incorporates the insights of the context dependence uniqueness,
but it furthermore concerns a different kind of relation to context and reference, namely, it
stresses the procedure taking place when a definite is used.

46
The operation of count-as-unique takes place whenever a speaker uses a definite, since it is a
condition for a felicitous use of a definite to be counted-as-unique. Let us reflect on the nature
of this operation. It is important that Badiou takes count-as-one as an operator that is applied to
multiples in order to structure them as ones (that is, multiple-ones). If we transfer his concept to
the field of linguistics, with the concomitant move from the One to the Unique, the count-as-
unique can be seen as an operator. Badiou’s philosophical idea that ontology is mathematics –
echoing the formalization of linguistics and theories of natural language – provides the grounds
for mathematical operations such as counting to be instructive and we will try to prove the
importance of the counting operation in the employment of the definite article in Greek. We
envisage count-as-unique as a condition that accompanies definite noun phrases in the spirit of
Heim (1982), who associated conditions of novelty/familiarity with indefinite and definite noun
phrases. Retaining from Heim the idea that (in)definite noun phrases are associated with the
introduction of a variable, which is to be bound by some operator, we claim the following.
We have already seen that the cat / a cat provide both a variable, the only difference being in
the conditions they impose (novelty/familiarity) and in the presupposition of their descriptive
content. The proposal we put forward in this thesis associates the condition of count-as-unique
and the weak familiarity condition and renders the following result (modelled on the definition
of Roberts 2003):
53. definition of definite
Given a context C, use of a definite NPi presupposes that it has as antecedent a
discourse referent xi which is:
i) weakly familiar in C (understood as defined above by Roberts 2003)
ii) counted-as-unique among discourse referents in C in being contextually
entailed to satisfy the descriptive content of NPi
Suppose we have the following sentence and apply the above conditions to a specific example:
54. Epeksa olo to apogevma me tin ghata ke meta vgika ekso.
played.1sg all the afternoon with the cat and then got.1sg out
‘I played all the afternoon with the cat and then I got out.’
The use of the definite NPi tin ghata ‘the cat’ presupposes that tin ghata has an antecedent, xi, in
the discourse that is (weakly) familiar to the participants of the discourse and is furthermore
counted-as-unique in the context in question, in the sense that the descriptive content of tin
ghata is satisfied by an entity that is counted-as-unique.

47
The study of the theories of definiteness that have been proposed reveals, I think, the fact that a
great amount of the literature on uniqueness has emphasized and considered mainly the result of
this operation, but the need to take a closer look to the realization, the process that concludes in
this result is of equal, if not greater, importance. The shift from uniqueness to count-as-unique
is not just a terminological change, it is rather a change of position that indicates how we
perceive what the definite article does. In order to understand uses that are not very common
cross-linguistically, I propose to shed some light on how we end up having the interpretation we
have. This counting operation is what enables us to understand the use of the definite article in
its recurrent uses, but also in other uses, not so frequent cross-linguistically. The employment of
the definite article with the universal distributive quantifier kathe ‘every/each44’ and with the
free choice item opjosdhipote ‘any45’ may serve as illustrations of the count process. Since the
use of the definite article with them does not appear freely in all contexts, we need to account
for the fact that the most probable hypothesis is that it has a repercussion in the
semantic/pragmatic interpretation of the nominal; otherwise we might hypothesize that speakers
would not resort to it; they employ the article because they want to express something in a
different way, and this difference is not to be equated with emphasis, as most descriptive
grammars would have it. The question amounts to: What does the speaker want to convey when
s/he adds the definite article with kathe and with opjosdhipote? Or put differently, what does the
article contribute to the nominal in question? Our working hypothesis is that the definite article
in these constructions adds the count-as-unique condition. It seems that speakers stress or rather
reveal the counting operation that underlies the use of the definite article. It is a plea for the
hearer to make sure that s/he fully distributes the counting to all the members of the set, one by
one, be it the kathe-set or the opjosdhipote-set. In the case of kathe the hearer already has to
interpret the property in question distributively, but the definite article tells him that s/he has to
be careful when she applies this property, because each and every entity must satisfy the
predicate individually- or that otherwise s/he would not understand what the speaker wants to
convey. In the case of opjosdhipote the hearer has again to consider every possible entity that
satisfies the description of the noun, without letting any one of them unconsidered. This is also
mirrored in the claim Lee & Horn (1994) make with respect to FC any (which could be
transferred to the case of opjosdhipote) being associated with a scale: even the least probable
individual is to be taken into account in order for the utterance to be felicitous.
44 From now on I will not provide the translation of kathe to English every time I refer to this item, since it can be
tranlated either as every or as each depending on the context of appearance, a fact that proves to be relevant for the
analysis adopted here. I will only refer to this quantifier as kathe both in the text as well as in the glosses, making a
decision about its translation only in the translation of the examples used to illustrate the facts.
45 Opjosdhipote would correspond to the English determiner any only in sentences where this last item has a free
choice reading. Greek has a different paradigm for negative polarity. For more details see chapter 4, and more
specifically on the comparison between these two items see Lazaridou-Chatzigoga (2006).

48
Under the view pursued here, definiteness is associated with two conditions, weak familiarity
and count-as-unique. As we have already suggested here, it is the latter one that proves helpful
when it comes to determine the semantic interpretation of noun phrases containing kathe or
opjosdhipote with the definite article. Familiarity is not relevant in these cases, since on the one
hand we have a universal distributive quantifier, an element for which reference is not tied to
the discourse, and on the other hand we have an indefinite, which is specifically tied with non
referentiality, free choiceness. We end up then with the other condition that is involved in the
interpretation of the definite article, namely, count-as-unique. This condition provides the
grounds for us to articulate a formal analysis of the concurrence of the definite article with kathe
and opjosdhipote. The contribution of the definite article in these cases is to be represented as an
independent condition, that does not constitute part of what is asserted nor is a presupposition or
background information. We seem to be discussing here an area of meaning that escapes
assertion, a dimension of meaning close to the work of Potts (2005) on expressive meaning46.
The condition of count-as-unique is to be represented formally as below:
55. count-as-unique condition
condition: [#x: P(x)] [count-U (x)]
to be read: for every x that has the property P, x is counted-as-unique
We are in the position to illustrate this condition as applied to kathe and opjosdhipote. We
suggest that kathe appearing with the definite article ends up with the following interpretation.
Following common assumptions on the semantics of quantifiers (based on Montague 1973 and
Barwise & Cooper 1981) kathe denotes a generalized quantifier, a set of sets. Here we provide
the interpretation in (-notation kathe:
56. [[kathe]]= (P (Q [#x: P(x)] [Q(x)]
and the definite article adds the count-U condition: [#x: P(x)] [count-U (x)]
If we move now to opjosdhipote, which is treated as an indefinite in the sense of Heim (1982), it
gets the following interpretation:
46 This assumption, that the meaning expressed here is not part of the assertion, can be shown by the fact that
negation does not affect the interpretation of a kathe-phrase with the definite article. If we look at an example with a
negation and an instance of kathe with the definite article, the interpretation of o kathe is not altered by the negation:
i. Dhen ine alithia oti to kathe pedhi efaghe ena biskoto.
not is truth that the kathe child ate.3sg a cookie
‘It is not true that each child ate a cookie.’

49
57. [[opjosdhipote N]]= Nw(x)
(the subscripted w indicates domain widening)
and the definite article adds the count-U condition: [#x: P(x)] [count-U (x)]
In the case of kathe then, we distribute the counting to each and every element of the set in
question and we predict thus the strict/total distribution of the events, as well as the reference to
instances of the kind, actual instantiations and not the kind itself, characteristics of these
nominals we will discuss in detail in chapter 3. In the case of opjosdhipote we have a variable
that has a widened domain (that is associated with opjosdhipote) and the contribution of the
definite article eliminates the possibility of not considering some entity satisfying the
descriptive content of the nominal, that is, there is no tolerance for exceptions. Each entity is
individuated and taken under consideration. The semantic interpretation does not change, but
the effect is seen in the realm of pragmatics. This effect can be only evaluated if we provide
some continuation of the discourse, examples of which we will discuss in chapter 4.
The proposal I have put forward in the preceding paragraphs should be tested on empirical data
and, therefore, I provide here some examples in order to show a possible application of the
theory I have in mind. More examples and a more precise characterization of these nominals is
the focus of chapters 3 and 4, but I consider it necessary for the reader to start being familiar
with the kind of expressions we will be discussing in the rest of this thesis. Here is an example47
with o kathe:
58. Pistevo oti to mistiko se tetja pechnidhia ine to pathos pu tha diskun
believe.1sg that the secret in such games is the passion that fut. show.3pl
i pektes tis kathe omadhas.
the players the.GEN kathe team
‘I think that the secret in this kind of games is the passion that the players of each team
will show.’
In example 58 we notice that the distribution is applied to each and every team. The
observations and assumptions of the proposal I