Abstract EN:

Very few gold standard annotated corpora are currently available for French. We present a project to build a reference tree bank for French. We annotate a newspaper corpus of 1 Million words (Abeillé et al 1998, 1999, 2000), following EAGLES recommendations (von Rekowski 1996, Ide et al. 1996, Sanfilippo et al. 1996, Kahrel et al. 1997) and developing specific annotation guidelines for French. Similarly to the Penn Tree Bank (Marcus et al. 1993), we distinguish a tagging and a parsing phase, and reach a process of automatic annotation followed by a systematic manual validation and correction. Similarly to the Suzanne Corpus (Sampson 1994, this volume), on the Prague tree bank (Hajicova et al 1998, this volume), we rely on several types of morphosyntactic and syntactic annotations for which we define extensive guidelines. Our goal is to provide a theory neutral, surface oriented, error-free tree bank for French. Similarly to the Negra project (Brants et al. 1999, this volume), we annotate both constituents and functional relations. Due to the lack of robust reusable annotation tools at the beginning of the project, we chose to develop our own.