As said before, a lexical item crystallizes meaning acquired along interactions with several contexts and the lexicon is the trace of these interactions which are held between pre-representations (kinds of questioning schemes) in various contexts. It seems to us that these pre-representations could be easily rendered by c-designs in Terui’s sense. In them, loci are replaced by named actions (which could correspond to the semantic primitives (cf. types of actions represented by verbs…)). The normalization of c-designs is a generalization of the reduction of lambda-terms. That says that it makes operations of generalized application and abstraction to interact. The by-product of an interaction would be a residual, or trace, and the lexicon would be the set of them. This task concerns present attempts which follow Pustejovsky’s Generative Lexicon, as they are made by N. Asher, C. Retoré and B. Méry. It will be conducted by (or in collaboration with) them. We shall study more specifically the efficiency of cdesigns to represent this knowledge. This opens the field to a more negotiated conception of meaning of words, which allow to keep away from the inconveniences of literal meaning, in order to go more in the direction of a meaning in context. This allows us to get a better understanding of the evolution of the lexicon.
Coordinator : Alain Lecomte (SFL - Paris 8)
Members : Myriam Quatrini (IML), Laurent Roussarie (SFL – Paris 8), Claire Beyssade (IJN)
External Collaborators : Aarne Ranta (Göteborg, Chalmers) , Nicholas Asher (IRIT, Toulouse)
Content : theoretical work starting from the known approaches to formal semantics aiming at revisiting thm and offering more practical tools. Take for example the notion of epistemic modality (like in "Paul could be the culprit," Paul may be the culprit here. "). Instead of trying to define it from possible worlds and orders of preference (A. Kratzer) we will attempt to define it by means of concepts from the proof-theory. In summary : the statement is "true" if there is a procedure P to extract a plausible proof from the "facts" (locations in some loci). Necessity is defined as valid across contexts, the assertion as an anchor which is valid whatever the context be etc.. Applications to temporality, space. The semantics of the statement will be seen in discourse and dialogue. From this perspective, a comparison with the SDRT Asher et al. will be conducted. The concepts of rhetoric relations, of topics, of discourse plane will be revisited. The ideas put forward in this project are expected to give a theory of discourse and dialogue which makes the economy of the concept of intention. A book will be drawn from this sub-task.