Contenu du sommaire : Les sources du savoir et leurs marques linguistiques, sous la direction de Patrick Dendale et Liliane Tasmowski

« Evidential values of the French tense system » A great number of languages encode by means of grammatical markers the speaker's distance with respect to the situation expressed by the proposition : the speaker explicitly signals that his information is derived from general reputation, hearsay, inferences , reasoning. . . Although French is not traditionally thought of as a language endowed with an evidential system, several temporal forms (especially the conditionnel, the passé composé and the présent) can take evidential meanings. This paper presents some comments on the phenomenon.

« Epistemic devoir, modal or evidential marker ? » It is generally recognized that the French verb devoir has deontic and epistemic senses. We focus on the epistemic sense and claim that the basic value of epistemic devoir is an evidential one whereas the modal values (probability, quasi-certainty) actually derive from it. We consider the nature of evidential devoir to be inferential and show how inference is to be understood as a complex operation including the search for premisses and the evaluation of competing conclusions.

« Epistemic pouvoir, an evidential marker » Taking into account the close relationship between devoir and pouvoir, we argue that epistemic pouvoir should be considered an evidential marker in the same way epistemic devoir is, which means that epistemic pouvoir signals information gained by inference. After a characterization of epistemic pouvoir along these lines, we address the problem of the influence of the conditional mood on both verbs.

« Savoir and croire » It is generally assumed that savoir « to know » and croire « to believe » belong to the same class of attitudinal predicates. In this article it is shown however that, from a syntactic point of view, their behaviour is quite different (for example, savoir can be in the scope of croire, but not vice versa). It is also shown that, when used « directly » (first person, present tense), croire serves to modify the truth value of the proposition, whereas savoir can be used to correct what the hearer thinks the speaker knows.

« The perceptual access to information : on the expressions un homme arrive / on voit arriver un homme » This paper aims to provide an analysis of a particular type of existential sentences based on the perceptual access to information. The purpose of the analysis is twofold : first, to describe the semantic and cognitive (perceptual) content of the sentences where the perceptual relation of access is implicit ; secondly, to establish, on the basis of the preceding analysis, the conditions of use for the explicit perceptual marker on voit in the same sentences.

« The linguistic dilution of responsibilities. A polyphonic description of the evidential markers il semble que and il parait que » Assertion guarantees truth based on evidence available to the speaker. The speaker may indicate the source and type of his evidence. This paper deals with the two French evidentials il semble que and il paraît que. Both indicate hearsay evidence and are equivalent to the English evidential it seems that. Although very similar at first glance, closer analysis reveals important, though often rather subtle differences as to the type of evidence. A polyphonic study permits a detailed explication of their functioning, and may explain observable distributional differences.

« Proverbs and proverb-like sentences » Proverbs and proverb-like sentences show characteristics that distinguish them from non-proverbial forms. The first part of the article is devoted to the study of such properties. In the second part, it is claimed that the function of proverbs is one of evidential marking. Proverbs appear to be one linguistic expression among others of common knowledge. Moreover, the genuine function of proverbs is argumentative : they serve as warrants for the continuation of the discourse.

« The "creation of the world" by the definite article. Le : an evidential marker ? » To use/interpret a definite description, the speaker/hearer has to justify the uniqueness conveyed by the definite article. To do this, he has to construct information by making a number of inferences, by default or in accordance with Levinson's principle of informativity. Since these inferences affect the degree of reliability of the utterance, the definite article, whose linguistic meaning triggers them, is to be considered a marker of evidentiality as defined by Dendale, but, as can probably also be maintained for other evidential markers in French, this evidential value is not to be equated with its basic meaning.