Cette thèse s'intéresse à la structure, aux fonctions, et aux bases cognitives des termes d'offense (tels que le terme "boche"). Les termes d'offense, ainsi que leurs équivalents psychologiques, posent des problèmes intéressants et possiblement fondationnels à propos de la nature de la signification, de l'expressivité dans les langues naturelles, du rôle des émotions dans la catégorisation. Ce travail discute de ces questions - ainsi que de nombreuses autres - en s'intéressant à différentes théories existantes ou originalesdu phénomène. De nouvelles données linguistiques sont mises en avant qui remettent en cause des théories linguistiques telles que les visions vériconditionnelles ou présuppositionnelles du phénomène, et de nouvelles théories non-linguistiques du phénomène sont développées, invoquant les concepts de qualité seconde ou la notion d'essence. Les propriétés linguistiques particulières des termes d'offense, telles que la projection ou l'expressivité, apparaissent dans ce travail être des conséquences linguistiques d'un phénomène essentiellement psychologique : la possibilité d'une composante émotionnelle ou évaluative dans la structure même des concepts. / The present work investigates the structure, function and cognitive underpinnings of slurring terms (such as "boche"). Slurring terms, and the mental correlates that I posit they have, raise interesting and possibly foundational issues about the nature of meaning, about expressivity in natural language, about the role of emotions in categorization. I discuss these questions - among many others - by studying different existing or original accounts of the phenomenon. I present novel linguistic evidence against linguistic views such as truth-conditional or presuppositional accounts, and develop new psychological (i.e. non-linguistic) theories of the phenomenon based on a connection with responsedependent concepts, or with essentialist concepts. The interesting linguistic properties of slurs, such as projection and expressivity, appear to be the linguistic consequences of the essentially mental fact that concepts may be loaded with emotional or evaluative content.

This thesis investigates the distinction, or distinctions, between response-dependent and response-independent concepts or subject matters. I present and discuss the three most influential versions of the distinction: Crispin Wright’s, Mark Johnston’s, and Philip Pettit’s. I argue that the versions do not compete for a single job, but that they can supplement each other, and that a system of different distinctions is more useful than a single distinction. I distinguish two main paradigms of response-dependence: response-dependence of subject matter (Johnston and Wright), and response-dependence of concepts only (Pettit). I develop Pettit’s ‘ethocentric’ story of concept acquisition into an account of concept evolution that suggests answers to a range of hard questions about language, reality, and the relation between them. I argue that while response-dependence theses of subject matter can be motivated in very different ways, the resulting theses are less different than they might seem. I suggest that the traditional ways of distinguishing response-dependent subject matters from response-independent ones – in terms of a priori biconditionals connecting facts of the disputed class with responses in subjects in favourable conditions, and fulfilling some further conditions such as non-triviality and sometimes necessity – may not be the best approach. I also discuss two general problems for response-dependence theses: the problem of ‘finkish’ counterexamples, and the problem of specifying the ‘favourable conditions’ a priori, yet in a non-trivial way. The discussion of response-dependence is informed by a framework based on the idea that some realism disputes can be viewed as location disputes: disputes over the correct location of the disputed properties among several levels of candidate properties. The approach taken in this work is a charitable one: I try to make sense of response-dependence. The conclusion is the correspondingly optimistic one that the idea(s) of response-dependence makes sense.

La thèse examine, à partir d’une relecture de Peirce et de certaines de ses interprétations des philosophes et des problèmes antiques, les arguments qui peuvent permettre à une conception pragmatiste des croyances et de la signification de parer aux accusations de conduire au relativisme. Ces arguments résident dans la façon dont ces conceptions s’articulent entre elles, et acceptent une forme de réalisme / This thesis examines arguments taken from Peirce’s reading in Ancient philosophy, which could be used to block accusations of relativism being latent in a pragmatist conception of belief and concepts. The argument lies in the articulation of the two conceptions and their compatibility with a realist view