Contenu du sommaire : Temps, référence et inférence, sous la direction de Jacques Moeschler

This paper is a radical criticism of formal semantic approaches to the sequencing problem in temporal reference assignment. The authors show that the main problem for the calculation of temporal reference is not the sequencing one, but the interval and causality problems, and that a pragmatic criterion (the criterion of consistency with the Principle of Relevance) explains the assignment of temporal reference.

The aim of this paper is to show how the use of Relevance Theory can simplify traditional analyses of a range of temporal phenomena. First, it is argued that logical descriptions of the past which treats it as ambiguous can be replaced by a pragmatic analysis which treats the past as univocal. Second, it is shown that the temporal progression characteristic of the so- called narrative past falls out from relevance-theoretic considerations, without the need for logical stipulation. Third, it is argued that a partial explanation for the interpretation of present tense sentences containing dynamic verbs can be derived from the exploitation of the idea of « interpretive use ».

This paper argues for a radical pragmatic and inferential approach of temporal reference assignment. It shows first that temporal expressions allowing the calculation of temporal reference are undetermined and must be non-literally interpreted and second that it is possible to formulate, via the notion of causality, a coreferential solution to the sequencing problem.

This paper gives first a detailed criticism of classical approaches to the imparfait (more precisely the aspectual and coreferential theses) and second a radically new description of this verbal tense. The solution stipulates an anaphoric relation of the kind part-whole between the antecedent-situation of the past and the situation described by the imparfait. This proposal allows to put together the questions of temporal reference and « associative » anaphora.

This paper is about the cooccurrence of deictic markers and past tenses in narratives within the framework of a theory of fiction which distinguishes two main configurations : the restricted configuration, in which the temporal reference point is movable and coincides with the related events, and the extended configuration, in which the temporal reference point is fixed, exterior to the fiction, and corresponds by convention to the date of publication of the narrative.

This paper is about the treatment of temporal referential expressions within the domain of man-machine communication, in which the problem is not only to allow the identification of the good referent to referential expressions, but also to account for the processes of changes of states, which supposes to have access to an ontology necessary for the possible reasonings on time.

This paper approaches the question of referents that are subject, along the discourse, to transformations, although they can be referred to anaphorically. The question raised is thus the possibility of a coreferential relation, whereas the material identity is not conserved. This contribution discusses the main solutions given to this problem (in particular the mental models thesis) and leads to the conclusion of the necessary distinction between coreference and material identity : pronouns, although they imply coreference, are sensitive to transformations of their referents.