The Literal Meaning of Definite Descriptions

Résumé

This thesis focuses on the semantics of definite descriptions (DDs). In chapter 1 I introduce a framework for doing compositional semantics for natural language that follows Heim and Kratzer (1998) and Fintel and Heim (2011). I also address a number of issues concerning the methodology of natural language semantics, as well as the theoretical desiderata that we aim to achieve. In chapter 2 I offer a reconstruction within this theoretical framework of three classical theories of DDs: the Russellian theory, the Fregean theory, and the Barwise and Cooper (1981) theory. Chapter 3 focuses on incomplete DDs. The incompleteness problem affects not only the Russellian theory, but also the Fregean and B&C theories. I argue that the syntactic variable approach to QDR proposed by Stanley and Szabó (2000a) offers a solution to the incompleteness problem that is equally applicable to the three theories of DDs introduced. Chapter 4 focuses on the referential/attributive distinction. Although intuitions of singularity are not part of our methodology, I point out (following Neale (2004)) that there are independent reasons why the Russellian theory predicts that a DD is semantically a rigid designator when it is used referentially. I argue that this proposal is applicable not only to the Russellian theory, as Neale does, but also to the Fregean and the B&C theories. In Chapter 5 and 6 I look at non-denoting DDs (i.e. DDs for which either uniqueness or existence fails to be satisfied). In chapter 5 I address the phenomenon called ‘presupposition’ in the literature, distinguishing between various kinds of data that are usually treated under this heading. I argue that with respect to a certain characterization of presuppositions, the Fregean theory and the B&C theory have more explanatory power than the Russellian theory. Chapter 6 deals with the truth-value intuitions triggered by utterances of sentences containing improper DDs. With respect to failures of uniqueness, I discuss Ramachandran’s (1993) argument and offer an improved version of it, which provides a compelling objection against the Russellian theory. With respect to failures of existence, I argue that the Fregean and the B&C theorist are in a better position to account for the patter of data than the Russellian. In chapter 7 I address data concerning the embedding of DDs in propositional attitude verbs. I discuss in detail the objection proposed by Heim (1991) against the Russellian theory. Again, this poses an important problem to the Russellian theory, while the Fregean and the B&C theories are not affected by the objection. The overall conclusion that this discussion leads to is that the Russellian theory is in general less prepared to account for the kinds of truth-conditional data we have considered than the alternative proposals discussed. The conclusions of chapters 6 and 7, and partially those of chapter 5, all indicate that the Russellian theory is the worst of the three options considered. The main positive contribution of this thesis is to point out that the B&C theory, although generally ignored in the literature, has the same theoretical virtues as the Fregean theory with respect to accounting for the range of data considered.