1It is commonly assumed that the primary function of the definite article is to show that a reference is made to an entity that both the Speaker and the Addressee can find (i.e. identify) in the context. By this property, the definite article is distinguished from the indefinite article, which is used when a reference is made to a non specific entity belonging to a class of similar entities. There is thus an opposition between definite and indefinite DPs, which can be illustrated by the minimal pair given in (1): in (1a), the Speaker asks the Addressee to take a book, without giving any indication about which book, while (1b) is an utterance appropriate only in a context where the Speaker and the Addressee are referring to a particular book. Either this book has been identified in the preceding context, or it is the only one present in the material environment, in the current situation.
(1)
|
a. Take a book !
|
|
b. Take the book !
|
2Since the 1980s, many studies have focused on indefinites and have led to distinguish several types of indefinites, on the basis of the meaning they convey and their syntactic, semantic and discourse properties. An important distinction between two classes of indefinites, weak and strong indefinites, was introduced. On the one hand, strong indefinites refer to a specific entity, which is presented as new in the discourse. Indeed, strong indefinites introduce a new discourse referent, which can then be referred back to by an anaphora in the succeeding discourse. On the other hand, weak indefinites don’t introduce a discourse referent, they merely constrain the range of a variable that has been previously introduced in the discourse by another linguistic expression. Most often, this variable has been introduced by the verb itself, which asserts the existence of its arguments. The weak/strong distinction is due to Milsark (1977) and was revisited among others by Diesing (1992): it has significantly renewed the studies about indefinites and indefiniteness and it has led to a better understanding of similarities and differences between indefinite DPs and bare nouns, especially in languages which have these two types of DPs.
3These works on indefinites also clarified, indirectly, by contrast, the meaning of the definite article and of definite DPs: taking as a starting point the formal contrasts between definites and indefinites in the languages which have both types of articles, researchers have tried to find out the propertie(s) that would be shared by definites and that indefinites would be lacking. As a matter of fact, it seems that there is an asymmetric relationship between definites and indefinites: one can assume that the definite article entails the indefinite article, since it seems that if a declarative sentence built with a definite article is true, then the same sentence in which the definite article is replaced by an indefinite article is also true. But the reverse is not true: the indefinite article doesn’t entail the define article, since to be appropriate, the definite requires some conditions to be satisfied, conditions which don’t play any role if the Speaker uses an indefinite article. It is generally assumed that the definite differs from the indefinite because the former conveys a double presupposition of existence and uniqueness, which is not the case of the latter. This double presupposition is what guarantees the identifiability of the referent. It is a presupposition, not a mere assertion, as shown by the usual tests: these presuppositions resist negation and interrogation (from (2a), (2b) and (2c) it can be inferred that (2d) is true) and these two presuppositions cannot be canceled unless they appear within the scope of metalinguistic negation (see (3a) and (3b)).
(2)
|
a. John read the book.
|
|
b. John didn’t read the book.
|
|
c. Did John read the book?
|
|
d. There is one and only one book relevant in the context.
|
(3)
|
a. John didn’t read the book, because there was no book to read where he was.
|
|
b. John didn’t read the book, he read a book.
|
4However, it seems that not all definite DPs presuppose the existence and uniqueness of their referent. Many examples have been found, in which a definite article is perfectly appropriate although it doesn’t presuppose the uniqueness of its referent. In (4a), it is clear that if John made a change, he didn’t take one train but two different trains; similarly, (4b) refers to “the side of the mountain”, without presupposing that the mountain has only one side. Every mountain has at least two sides, and sometimes more than two. This belongs to common knowledge.
(4)
|
a. John took the train and he made a transfer in Berlin.
|
|
b. The village is located on the side of a mountain.
|
There would therefore be (at least) two classes of definite DPs: those which satisfy the uniqueness presupposition (called strong definites) and those which don’t (called weak definites).
5This volume addresses the issues of the existence and characterization of weak definites. The term “definite” is taken in its narrowest sense: as a matter of fact, we are only interested in DPs built with a definite article, in English the, or in French le, la, les. We won’t take into account other DPs such as DPs headed by demonstrative or possessive determiner, even if these DPs are usually considered as instances of definite descriptions. The various papers in the volume base their analysis on data taken from various languages (English, French, Italian, Hebrew) and try to answer the following questions:
-
In which contexts do weak definites occur? It seems that definite descriptions aren’t systematically ambiguous between a weak and a strong reading and that weak readings arise only when a particular noun co-occurs with a particular verb (like in take the train), or when the definite description is built with a relational noun and embeds a genitive complement (like in the student of a linguist).
-
What is the syntactic and semantic structure of weak definites? Are there two definite determiners, unrelated to each other? Or is there only one definite determiner? In the latter case, is the strong reading of a definite DP obtained by a semantic enrichment from a weak reading of the definite DP, the uniqueness presupposition coming not from the definite article itself but from another linguistic element present in the sentence or in the context? Or on the contrary, is the weak reading obtained by weakening the strong reading, the uniqueness presupposition being suspended or canceled in context? Is it the case that the difference between strong and weak readings is a true semantic difference or is it anchored into a syntactic distinction, the syntactic structures of weak and strong definites not being the same?
6These questions give rise to others. Is the weak/strong distinction the same when it is applied to definites and to indefinites? What properties are shared by weak indefinites and weak definites, if any? Do weak definites and weak indefinites behave similarly with respect to the issue of discourse referent management and to anaphora resolution? In other words, what characterizes the weak reference in general? And is there a difference between weak reference and generic reference? What does the contrast between indefinite and definite become, and how does it cross the weak/strong distinction? In the examples (4), it seems that one could easily replace the definite article by an indefinite article without fundamentally changing the meaning of the sentence. If definites and indefinites are commutable in these contexts, how can we maintain the idea of a binary opposition and what does the demarcation line drawn between definiteness and indefiniteness become? Works on weak definites suggested rather that there is a continuum between strong reference and weak reference and that if the two extreme positions are occupied on the one hand by weak indefinites and the other hand by strong definites, there is an intermediate area, more difficult to characterize, in which weak definites and strong indefinites can be found. It is only though a precise characterization of these semantic notions that we may hope to account for the large variety of determiner systems found across languages and for their correspondence, which has to be cautiously established, for instance to solve translation tasks. Let us remind that there are languages with and without articles, but also languages with definite articles but without indefinite articles, languages with just the indefinite article, languages with several definite articles (see Ebert 1971, Löbner 1985) and also languages with several indefinite articles (see von Heusinger, Klaus & Klein, Udo 2012). Last, but not least, there are the bare nominal phrases and we should try to understand their relation to the notions of (in)definiteness.
7The volume consists of six contributions. The first three papers (Carlson, Klein, Gegg-Harrison & Tanenhaus; Aguilar & Zwarts; Donazzan) focus on weak definite descriptions that appear in a VP and denote activities describing “routines”: they are simple DPs consisting of a definite determiner followed by a noun, as the train in the VP take the train. The three contributions are concerned with the same set of examples and they agree about the characterization the whole class shoud receive: weak definite DPs are definite DPs which legitimize a sloppy interpretation of the DP in the case of VP ellipsis. However they differ in the analysis they propose.
8Carlson et al. suggest that weak readings be analyzed as cases of incorporation. In support of their thesis, they present a series of experiments whose findings is that compared to strong definites, weak definites need not refer uniquely and readily trigger semantically enriched readings that compete effectively with normal inferences one might draw from a sentence. According to them, there are two different definite determiners, one which gives rise to strong readings and another which gives rise to weak readings.
9Aguilar and Zwarts, on the contrary, consider that the definite determiner is the same in weak and strong readings, and that what varies is the denotation of the noun. They assume that in weak readings, the definite determiner combines with a kind referring noun. Then kinds are instantiated by ordinary individuals because weak definites combine with object-level predicates. This combination is made possible by a lexical rule that lifts object-level predicates to kind-level predicates, and incorporates into their denotation a predicate that represents the stereotypical usages of the kinds. The major advantage of this analysis is to account for the semantic enrichment usually associated to weak reading.
10Finally, Donazzan discusses both Carlson et al. and Aguilar et al. proposals on the basis of observations from Italian. She shows that in Italian weak readings are not necessarily associated to a semantic enrichment. According to her, the common property to all weak readings is that the VP has to be interpreted as referring to a prototypical activity. Donazzan uses weak familiarity from Roberts to account for prototypicality restrictions, which are culturally dependent. She claims that the definite determiner is the same in weak and strong definite NPs and that it is associated with a double presupposition of familiarity and uniqueness also in weak readings. She explains that the prototypical interpretation associated with the VP is what explains the absence of presupposition failure.
11The other three contributions (Corblin; Beyssade; Doron & Meir) extend their investigations to cases of complex weak definite descriptions, built with the definite article followed by a relational noun and a genitive complement, such as the daughter of a farmer. They seek to provide a unified analysis of all the weak definite descriptions, both single DPs involved in a VP and complex DPs built with a genitive.
12Corblin shows how to include weak definites within a general theory of definiteness in which the concept of accommodation plays a central role. According to him, the use of a definite NP is legitimized if the definite NP refers to an entity which is identifiable by both the Hearer and the Speaker. And there are as many different sub-classes of definites as there are different identification processes. Corblin analyses weak definites as relational definites and compare them with associative definites, whose interpretation needs the accommodation of a bridging between the definite DP and its antecedent as proposed by Hawkins (1978). The thesis defended by Corblin is that weak definites are based on a relational meaning containing free variables which are bound within their own clause. The proposal defines weak definiteness as a constructional meaning involving the whole host clause and triggering a meaning enrichment based on the telic quale (Pustejovsky 1995) of the head noun.
13In her paper, Beyssade proposes to unify the analysis of weak definites by using the opposition between types and tokens. According to her, weak definites always refer to types (and never to tokens): types of entity in the case of weak definites including a relational noun and a genitive complement, and types of activity in the case of weak definites which are arguments of a VP. She shows that if one recasts the content of the presupposition associated with the definite determiner and proposes, as Coppock & Beaver (2012), to replace the double presupposition of existence and uniqueness by a weak presupposition of uniqueness, in which existence and uniqueness are dependent one from another, then one can claim that both weak and strong definites trigger the same presupposition of weak uniqueness. And as a matter of fact, one can explain why definite and indefinite convey the same content in weak readings and why, in many cases, definites are obligatory and indefinites sounds as inappropriate.
14Finally, Doron & Meir propose an analysis of weak definites in Hebrew. They show the role of Semitic state inflection in determining strong and weak readings. In Hebrew, weak definites often take the form of noun phrases headed by a noun marked with construct-state inflection. Doron et Meir discuss the interpretation of all construct-state nouns as relational. In colloquial Hebrew, the type-shifted definite determiner used in the formation of weak definites may take the form of a numeral (or other amount nouns) marked with emphatic-state inflection. The paper focuses on a subclass of DPs, weak definites headed by emphatic-state amount nouns, called amount definites. The paper compares the properties of amount definites to those of definite DPs where the amount noun is marked with construct-state inflection. So the contribution of this paper is twofold: it presents a unified analysis of weak definites and it extends the study of weak definiteness to a new class of examples built with amount nouns.
15To conclude, the volume presents a new panorama of weak definiteness. It reviews the main theoretical frameworks elaborated to account for weak definites, it gives a large set of data from various languages which are typologically different, it opens new perspectives of the role of definite and indefinite articles in establishing the reference of DPs, and on the issue of how to achieve the effects of definite and indefinite articles in languages which lack those articles. In one word, this volume shows that the issue of definiteness is not solved but it has been recently largely renewed by the discovery of important data, found in corpora and across languages, which challenge the thesis according to which definites are always associated to a uniqueness presupposition.