Section: New Results

Development of the Corpus de Référence du Français

Participants : Stéphane Riou, Benoît Sagot.

The `Initiative Corpus de Référence du Français' (ICRF) is a project of Institut de Linguistique Française (ILF-FR2393 CNRS), coordinated by its director Franck Neveu and by Benoît Sagot.

The purpose of the ICRF is the development of a first prototype of the future French Reference Corpus, so as to assess the feasibility of this project and evaluate its potential impact. ICRF reuses existing freely-available French corpora, supplemented by additional data in an opportunistic fashion (e.g. a French media critic corpus and the corpus of talks given at an workshop on ethics and neurodegenerative diseases). ICRF preserves copyright and authorship of all corpora used. These corpora have been or will be part-of-speech tagged with MElt, converted to XML (TEI-P5-compliant) and made accessible via a web interface. The aim of ICRF is not to replace individual corpora and the interface will therefore allow, whenever possible, to easily recover access to each individual corpus. ICRF adds 5 metadata tags to categorize each individual corpus: spoken/written, text type and genre, linguistic competence level, date and linguistic area.

In 2015, the normalisation, tagging and conversion to XML of individual corpora has started, following the design of format specifications. The development of the web interface has already started, and a prototype is now available. Users can perform queries (search by tokens and/or POS) and use basic linguistic tools on the corpora (e.g. a concordancer). It is therefore more than a simple search interface or a download site: it improves research and selection of corpus.