TALEP home page

TALEP is the Written and Spoken Language Processing group, created on Jan. 1st, 2008. The team results from the merging of the former CALN group (Natural Language Understanding, lead by Paul Sabatier) from LIF, and part of the DELIC group (Corpus-based Computational Linguistics, lead by Jean Véronis), former "équipe d’accueil" (EA 3779) of Université de Provence.

The TALEP group members work on fondamental linguistic problems, focalizing on rigorous formalism to enable non-trivial processing of language. They build and spread linguistic resources (corpora, lexica, grammars...) and provide regular updates and extensions to these resources.

The TALEP group members design algorithms and computer programs for automatically processing language. They develop generic software, like CONTEXTES, ILLICO, MACAON or MICA in order to deploy specific applications.

The TALP team covers various aspects of language modeling and resource production for:

• type of language: written and spoken;
• style of language: normalized vs unstructured, spontaneous vs prepared, formal vs informal;
• linguistic coverage: general language and sub-languages;
• languages: French, English, Japanese...;
• knowledge models involved: symbolic, statistical, mixed models;
• knowledge acquisition methods: introspection, (un-/semi-) supervised learning.

Regarding automatic processing, the TALEP group focuses on the following aspects:

• understanding levels: full understanding and partial understanding;
• processes, algorithms and architectures: analysis and generation;
• evaluation of automatic processes;
• applications: automatized linguistic resource gathering, man-machine dialog systems, language learning, assistance to disabled, machine-aided translation, indexing, information extraction and retrieval.