Francois Recanati

It is often claimed that, because of semantic underdetermination, one can determine the content of an utterance only by appealing to pragmatic considerations concerning what the speaker means, what his intentions are. This supports ‘inferentialism' : the view that, in contrast to perceptual content, communicational content is accessed indirectly, via an inference. As against this view, I argue that primary pragmatic processes (the pragmatic processes that are involved in the determination of truth-conditional content) need not involve an inference from premisses concerning what the speaker can possibly intend by his utterance. Indeed, they need not involve any inference at all : communication, I argue, is as direct as perception.

Philosophie du langage et philosophie de l'esprit constituent désormais un tout indissociable. Les expressions linguistiques «signifient». Qu'est-ce que cela veut dire? François Recanati distingue trois réponses possibles. Selon la première, signifier c'est (pour une expression linguistique) être associée à des représentations mentales. Selon la deuxième, signifier c'est «faire référence» et renvoyer à quelque chose dans le monde – une réalité extralinguistique. Selon la troisième, enfin, signifier, c'est jouer un rôle distinctif dans cette activité sociale qu'est la parole. La première réponse renvoie aux représentations mentales. Mais qu'est-ce, pour une représentation mentale, que d'avoir un contenu? La vraie question, est-on tenté de penser, est plus générale : qu'est-ce que signifier ou avoir un contenu? Qu'est-ce qu'une représentation (linguistique ou mentale)? Les philosophes contemporains recherchent une théorie du contenu qui soit suffisamment générale pour s'appliquer à la pensée aussi bien qu'au langage. François Recanati nous introduit à leurs efforts, et conclut en faveur de l'approche «pragmatique» inspirée de Wittgenstein. Ce qui fait que la pensée et le langage représentent le monde, c'est avant tout le fait que la pensée et le langage sont dans le monde, qu'ils y ont leur place et y jouent leur rôle.

In a series of papers, Sperber provides the following analysis of the phenomenon of ill-understood belief (or 'quasi-belief', as I call it): (i) the quasi-believer has a validating meta-belief, to the effect that a certain representation is true; yet (ii) that representation does not give rise to a plain belief, because it is 'semi-propositional'. In this paper I discuss several aspects of this treatment. In particular, I deny that the representation accepted by the quasi-believer is semantically indeterminate, and I reject Sperber's claim that quasi-belief is a credal attitude distinct from plain belief.

The received view about meteorological predicates like ‘rain’ is that they carry an argument slot for a location which can be filled explicitly or implicitly. The view assumes that ‘rain’, in the absence of an explicit location, demands that the context provide a specific location. In an earlier article in this journal, I provided a counter-example, viz. a context in which ‘it is raining’ receives a location-indefinite interpretation. On the basis of that example, I argued that when there is tacit references to a location, it takes place for pragmatic reasons and casts no light on the semantics of meteorological predicates. Since then, several authors have reanalysed the counter-example, so as to make it compatible with the standard view. I discuss those attempts and argue that my account is superior.

First person thoughts are the sort of thought one may express by using the first person ; they are also thoughts that are about the thinker of the thought. Neither characterization is ultimately satisfactory. A thought can be about the thinker of the thought by accident, without being a first person thought. The alternative characterization of first person thought in terms of first person sentences also fails, because it is circular : we need the notion of a first person thought to account for the reference rule governing the first person in language. The paper offers a new characterization of first person thought. A first person thought is a thought which deploys the first person concept, where the first person concept is construed as a special kind of ‘mental file’. Mental files are based on, and their reference determined by, epistemically rewarding (ER) relations in which the subject stands to entities in the environment. In the case of the SELF file, the relevant ER relation is identity. This guarantees that the first-person concept refers to the thinker of the thought in which it is deployed.

In a recent paper (Linguistics and Philosophy 23, 4, June 2000), Jason Stanley argues that there are no `unarticulated constituents', contrary to what advocates of Truth-conditional pragmatics (TCP) have claimed. All truth-conditional effects of context can be traced to logical form, he says. In this paper I maintain that there are unarticulated constituents, and I defend TCP. Stanley's argument exploits the fact that the alleged unarticulated constituents can be `bound', that is, they can be made to vary with the values introduced by operators in the sentence. I show that Stanley's argument rests on a fallacy, and I provide alternative analyses of the data.

Putnam is known for having demonstated the existence of a new form of context-dependence, namely that which characterizes natural kind terms. Terms like ‘tiger' and ‘water' are indexical, Putnam says, since their conditions of application varies with the context of use — in a suitably broad sense of ‘context'. In this talk I focus on the relation between Putnam's semantics and a body of views I call ‘contextualism'. Contextualism generalizes context-sensitivity : it claims that sentences carry contents only in the context of a speech act. This view was put forward by ordinary language philosophers in the mid-twentieth century, and it has re-surfaced in recent times in the works of philosophers like John Searle, Charles Travis, and myself. In the talk I argue that Putnam's semantics has strong affinities with contextualism.

For Perry and many authors, de se thoughts are a species of de re thought. In this paper, I argue that de se thoughts come in two varieties: explicit and implicit. While explicit de se thoughts can be construed as a variety of de re thought, implicit de se thoughts cannot: their content is thetic, while the content of de re thoughts is categoric. The notion of an implicit de se thought is claimed to play a central role in accounting for the phenomenon of immunity to error through misidentification. Lewis has attempted to unify de re and de se in the opposite direction: by reducing de re to de se . This, however, works only if we internalize the acquaintance relations. I criticize Lewis's internalization strategy on the grounds that it rests on Egocentrism (the view that every occurrent thought is ultimately about the thinker at the time of thinking). In the conclusion, I suggest another way of unifying de re and de se , by extending the implicit/explicit distinction to de re thoughts themselves.

What characterizes indexical thinking is the fact that the modes of presentation through which one thinks of objects are context-bound and perspectival. Such modes of presentation, I claim, are mental files presupposing that we stand in certain relations to the reference : the role of the file is to store information one can gain in virtue of standing in that relation to the object. This raises the communication problem, first raised by Frege : if indexical thoughts are context-bound and relation-based, how is it possible to communicate them to those who are not in the same context and do not stand in the right relations to the object? Following Frege, I argue that the solution comes from an important distinction between linguistic and psychological modes of presentation. Psychological modes of presentation are mental files. They are perspectival and context-bound. But linguistic modes of presentation are fixed by the conventions of the language and they are shared by the language users. They are public and serve to coordinate mental files in communication by constraining them to contain the piece of information they encode. In this way communication takes place even though the indexical thoughts entertained by the speaker are, in some sense, private and cannot be shared by the audience.

It has often been observed that the meaning of a word may be affected by the other words which occur in the same sentence. How are we to account for this phenomenon of 'semantic flexibility'? It is argued that semantic flexibility reduces to context-sensitivity and does not raise unsurmountable problems for standard compositional accounts. On the other hand, it would be a mistake to assume too simple a view of context-sensitivity. Two basic forms of context-sensitivity are distinguished in the paper. The second form — sense modulation — shows that, in a sense, there is more in the meaning of the whole than can be derived from the meanings of the parts.