Overall Objectives
Scientific Foundations
Application Domains
New Results
Other Grants and Activities

Section: Scientific Foundations

Lexical semantics

Participants : Christian Bassac, Mauro Gaio, Renaud Marlet, Bruno Mery, Christian Retoré [ correspondent ] .

One of the most exciting challenges in computational linguistics is the question of lexical semantics, that is a proper treatment of word meanings and the way they relate one to another and finally how to handle the minimal interaction with knowledge representation. This part of semantics is relevant, not to say mandatory, for computing the semantic counterpart of composition be it lexical or syntactic.

The Generative Lexicon [64] is one of the most common frameworks for representing the internal structure of the meaning of words and morphemes, an alternative being the lexical functions of Mel'c̆uk and Polguère. The former is better suited for the logical apparatus developed by Signes, since Pustejovsky's set up can be viewed as an extension of Montague semantics, with which it shares the compositionality and the type theoretical formulation.

The information which depicts the sense of a word or morpheme is organised in three layers: the argument structure (related to logical semantics and syntax), the event structure, and the qualia structure. The argument structure provides types (in the type-theoretical sense) to the arguments encoded in the qualia structure regardless of whether they are syntactically mandatory or optional. The event structure follows [46] . It unfolds an event into several ordered sub-events with a mark on the most salient sub-event. Events are typed according to the typology of Vendler: state, process and transition, this latter type including achievement and accomplishment. The qualia structure relates the argument structure and the event structure in roles: formal, constitutive, telic, agentive.

This information and its organization into the generative lexicons allows an explanation of, for instance, polysemy and of compositionality (in particular in compound words or in simple phrase structure). This kind of model relates knowledge representation to linguistic organization and thus is especially useful for word sense disambiguation during (automated) syntactic and for computing the semantics of a compound, a phrase, a sentences and a discourse.

Signes is for instance interested in the so-called logical polysemy, that is how some occurrences may refer to one or another aspect (corresponding to a semantic type) of a given word. In order to get a better interface with syntax, our research rather try to extend logical and compositional sentence-semantics like Montague semantics and lambda-DRT, than to encode the structure that one finds in dictionaries and lexical studies.


Logo Inria