ANR "Cognitive Origins of Vagueness": Paul Egré (2008-2011)
Vagueness concerns the problem of specifying the limits of applicability of our linguistic and perceptual categories. Mathematical concepts such as "prime number" or "acute angle" have a precise extension in principle, but it is not so for everyday concepts such as "red", "tall", "heap" or "house". These concepts are often flexible in that they are applicable to objects that look quite dissimilar from each other upon closer examination. More fundamentally, vague concepts admit borderline cases, namely cases for which we are uncertain as to whether the concept should apply or not. In relation to this, vague predicates also typically give rise to logical puzzle or paradox (such as the sorites paradox, according to which if x and y are only slightly different in respects of how tall they look, then x and y should either both be tall, or both be not-tall).
The present project aims at exploring new and interdisciplinary approaches concerning the sources of vagueness, in particular through collaborative meetings involving not only philosophers, but also linguists, psychologists and decision theorists. A substantial part of the project so far has been devoted to the examination of the link between logical and psychological models of imperfect discrimination and non-transitivity (such as signal detection theory, in particular in relation to Williamson's margin for error models). More recently, collaborations have taken place to reassess the link between vagueness and ambiguity (cf. Raffman), to relate vagueness to psychological models of similarity in categorization (Gärdenfors), and to connect semantic and psychological models of comparison with the logic of vagueness (Goodman, Luce, van Rooij).
ERC Advanced Grant "Context, Content and Compositionality": François Récanati (2009-2012)
Over the past fifteen years, François Récanati has argued that the effects of context on content go well beyond what is standardly acknowledged in semantics. This view is sometimes referred to as ‘Contextualism’ or (more technically) ‘Truth-Conditional Pragmatics’ (TCP). The key idea is that the effects of context on content need not be traceable to the linguistic material in the uttered sentence. Some effects are due to the linguistic material (e.g. to context-sensitive words or morphemes which trigger the search for contextual values), but others result from ‘top-down’ or ‘free’ pragmatic processes that take place not because the linguistic material demands it, but because the literal meaning of the sentence requires adjustment or elaboration (‘modulation’) in order to determine a contextually admissible content for the speaker’s utterance. Such processes, e.g. free enrichment, sense extension, metonymic transfer, etc., arguably affect the intuitive truth-conditions of utterances even though they take place for pragmatic reasons, without being triggered by the linguistic material in an obligatory manner.
The claim which TCP makes regarding the role of free pragmatic processes in the determination of intuitive truth-conditional content is an empirical conjecture about natural language. Other philosophers of language (Stanley 2000, 2007, Szabo 2000) have made the opposite conjecture, more in line with traditional ways of thinking about meaning and truth-conditions. Since it gives up those traditional assumptions, TCP sounds revolutionary, and many theorists (e.g. Predelli 2005) are suspicious of it because they take it to threaten the very enterprise of semantics. In the literature, one often finds arguments to the effect that, if Contextualism is right, then systematic semantics becomes impossible. More precisely, the claim that is often made is that TCP is incompatible with the Principle of Compositionality, upon which (arguably) any systematic semantics must be based.
The aim of this project is to defend Contextualism/TCP by demonstrating that it is not incompatible with the project of constructing a systematic, compositional semantics for natural language. This demonstration is of paramount importance given the current predicament in the philosophy of language. We are, as it were, caught in a dilemma : formal semanticists provide compelling arguments that natural language must be compositional, but contextualists offer no less compelling arguments to the effect that « sense modulation is essential to speech, because we use a (mor or less) fixed stock of lexemes to talk about an indefinite variety of things, situations, and experiences » (Recanati 2004 : 131). What are we to do, if modulation is incompatible with compositionality ? Our aim is to show that it is not, and thereby to dissolve the alleged dilemma.
"Context-Dependence: New Perspectives", PAI Picasso en collaboration avec le groupe LOGOS de l'Université de Barcelone. (2008-2009)
"Semantic Content and Context Dependence": Isidora Stojanovic (2010-2012)
Semantic theories provide a systematic description of the meanings of the sentences of particular natural languages, or at least of simplified models of such languages, for a number of explanatory purposes. The notion of the content or proposition expressed by a sentence is central to most semantic theorising: the predictions a semantic theory yields concern which semantic content (or which proposition) each sentence expresses in a possible context of use. In the last ten years or so, philosophers and semanticists have engaged in a heated debate about the best way to model various novel forms of context dependence. These disputes have reached an impasse. It is the hypothesis of this project that they cannot be resolved without a clarification of the notion of a semantic content, which is central to the disputes. The project aims to re-assess the motivation for the notion of semantic content both historically and in the light of current theorizing, to articulate the theoretcial role and purpose of this notion (if any), and to use the resulting clarified theoretical framework to make progress in resolving the disputes.