Section: New Results

Modelling and extracting discourse structures

Participants : Laurence Danlos, Charlotte Roze.

Lexical semantics of discourse connectives

Discourse connectives are words or phrases that indicate senses holding between two spans of text. The theoretical approaches accounting for these senses, such as text coherence, cohesion, or rhetorical structure theory, share at least one common feature: they acknowledge that many connectives can indicate different senses depending on their context. LEXCONN is a lexical database for French connectives [16] .

The French connectives “en réalité” and “en effet” have been the topic of numerous studies but none of them was formalized. [53] gives a formalization of the conditions the two arguments of these connectives should meet. This formalization is based on factivity information as modeled in the FactBank corpus developed by Roser Sauri.

Sometimes, the sense of connectives is unique but its arguments are hard to determine. In particular, the second argument of an adverbial connective is not always equivalent to its syntactic arguments. This raises problems at the syntax-semantics interface which are described in [52] . The method to handle theses problems in a discursive parser will be studied in the ANR project POLYMNIE, which is headed by Sylvain Podogolla (Inria Lorraine) and which started in October 2012.

Discursive annotation

We plan to annotate the French corpus FTB (French Tree Bank) at the discursive level, in order to obtain the FDTB (French Discourse Tree Bank). The methodology that will be used is close to the one used in the PDTB (Penn Discourse Tree Bank). The first steps of this long term project are presented in [48] , [49] , [51] .

This work is based on a new hierarchy of discourse relations and this new hierarchy was presented at an European workshop organized by the project MULDICO.