Anders J. Schoubye

Ghosts, Murderers, and the Semantics of Descriptions

It is widely agreed that sentences containing a non-denoting description embedded in the scope of a propositional attitude verb have true de dicto interpretations, and Russell’s (1905) analysis of definite descriptions is often praised for its simple analysis of such cases, cf. e.g. Neale (1990). However, several people, incl. Elbourne (2005; 2010), Heim (1991), and Kripke (2005), have contested this by arguing that Russell’s analysis yields incorrect predictions in non-doxastic attitude contexts. Heim and Elbourne have subsequently argued that once certain facts about presupposition projection are fully appreciated, the Frege/Strawson analysis of definite descriptions has an explanatory advantage. In this paper, I argue that both Russell’s analysis and the Frege/Strawson analysis face a serious problem when it comes to the interaction of attitude verbs and definite descriptions. I argue that the problem observed by Elbourne, Heim, and Kripke is much more general than standardly assumed and that a solution requires a revision of the semantics of definite and indefinite descriptions. I outline the conditions that are required to solve the problem and present an analysis couched in dynamic semantics which can provide a solution. I conclude by discussing some further issues related to propositional attitude verbs that complicate a fully general solution to the problem.

Noûs (Online First), doi:10.1111/j.1468-0068.2011.00836.x

Descriptions, Truth Value Intuitions, and Questions

Since the famous debate between Russell (1905, 1957) and Strawson (1950, 1964) linguistic intuitions about truth values have been considered notoriously unreliable as a guide to the semantics of definite descriptions. As a result, most existing semantic analyses of definites leave a large number of intuitions unexplained. In this paper, I explore the nature of the relationship between truth value intuitions and non-denoting definites. Inspired by comments in Strawson (1964), I argue that given certain systematic considerations, one can provide a structured explanation of conflicting intuitions. I show that the intuitions of falsity, which proponents of a Russellian analysis often appeal to, result from evaluating sentences in relation to specific questions in context. This is shown by developing a method for predicting when sentences containing non-denoting definites elicit intuitions of falsity. My proposed analysis draws importantly on Roberts (1996, 2004) and recent research in the semantics and pragmatics of focus.

Linguistics and Philosophy (2009) 32, 6: 583-617, doi:10.1007/s10988-010-9069-y

'The Argument from Convention' is not ambiguous

In recent years, a novel argument in favor Donnellan's (1966) semantic distinction between attributive and referential descriptions has become increasingly popular in the literature on definite descriptions. This new argument is based on putative observations about statistical frequency of use and processing ease. In this paper, I attempt to show that there is a significant lack of empirical justification for the main premises of this argument and that the argument overgenerates. In particular, I show that if the argument were sound, it would entail that a very wide range standardly assumed pragmatic phenomena in natural language are in fact semantic ambiguities. I also show that when this novel way of characterizing the distinction between referential and attributive descriptions is properly understood, it becomes unclear what data the distinction is supposed to explain. I conclude by demonstrating a more general point, namely that determining literal meaning on the basis of statistical frequency of use and intuitions about processing ease is not a reliable methodology.

On Describing

The overarching topic of this dissertation is the semantics and pragmatics of definite descriptions and its central question is whether a sentence such as 'the king of France is bald’ literally asserts the existence of a unique king (and therefore is false) or simply presupposes the existence of such a king (and thus fails to express a proposition).

One immediate obstacle to resolving this question is that truth value judgments seem particularly unstable with regards to these cases (viz. sentences containing non-denoting descriptions). Some of these elicit a strong intuition of falsity whereas others seem simply awkward or strange. Because of this, truth value judgments are generally considered unreliable. This is the central topic of Chapter 2 where I attempt to develop an explanation of this phenomenon. I observe that when these types of sentences are considered in the context of a discourse, a systematic pattern in judgments emerges. This pattern, I argue, should be explained in terms of certain pragmatic factors, e.g. whether a speaker's utterance is interpreted as cooperative. I then lay out a detailed and general explanation of the phenomenon which draws importantly on recent research in the semantics and pragmatics of questions and focus. I show that the behavior of these judgments can be systematically explained, that truth value judgments are not as unreliable as standardly assumed, and that my proposed explanation best supports the conclusion that definite descriptions presuppose rather than assert existence.

If definite descriptions are assumed to literally assert existence, a sentence such as 'Hans wants the ghost in his attic to be quiet' is incorrectly predicted to be true only if Hans wants there to be a ghost in his attic. This prediction is often considered evidence against Russell’s quantificational analysis and evidence in favor of the referential analysis of Frege and Strawson. In Chapter 3, I argue that this problem is a general problem about the existence commitments of natural language determiners, i.e. not an argument in favor of a referential analysis. I show that in order to avoid these undesirable predictions, quite radical changes to the semantic framework are required. For example, it must be assumed that a sentence of the form ‘The F is G' has the open sentence 'x is G’ as its asserted content. I outline a uniform quantificational and presuppositional analysis of definites and indefinites which by exploiting certain features of so-called dynamic semantics unproblematically assumes that the asserted contents indeed are open sentences.

In view of the uniform semantic analysis defended in Chapter 3, I use Chapter 4 to resist a recent argument due to Reimer (1998) and Devitt (2004) that definite descriptions are ambiguous between attributive and referential uses. In contrast to Donnellan's (1966) pioneering arguments in favor of this distinction, Reimer's and Devitt's argument rests on the assumption that definite descriptions are regularly used to communicate singular thoughts and that the standard meaning of definite descriptions therefore must be fundamentally indexical and directly referential. I argue that this argument relies crucially on tacit assumptions about semantic processing for which there is no empirical evidence. I then show that the argument is too general; if sound, it would be an argument for an indexical treatment of most, if not all, other determiners. I conclude by demonstrating that the view does not explain any new data and thus has no clear motivation. I conclude by providing a rough outline of how the uniform analysis presented earlier is capable of explaining referential uses.

These three chapters are preceded by a short general introduction to definite descriptions. There I explicate Russell’s highly influential theory of descriptions and I outline its motivations. I also briefly introduce four of the standard objections to Russell’s theory since these are relevant to the subsequent chapters.

The dissertation is supplemented with an appendix where I discuss a wide range of data relevant to Chapter 3. In principle, this could have been included there, but since Chapter 3 deals with a large number of theoretical issues already, I thought it best to postpone a comprehensive survey of the data to a later stage in order to maintain the focus of the chapter.

Summarizing, I have aimed at providing a detailed pragmatic explanation of a long-standing puzzle about truth value judgments and then at outlining. a novel dynamic semantic analysis of definite (and indefinites) descriptions. This analysis solves a significant problem about existence commitments -- a problem that neither the standard Russellian analysis nor the Frege/Strawson analysis are equipped to handle. This proposed analysis is then defended against the claim that definite descriptions are lexically ambiguous between attributive and referential uses.

Doctoral Dissertation (2011), University of St Andrews, St Andrews