Corpus oraux et chunking

This paper describes a process of partial parsing of a spontaneous spoken corpus in French. It is based on a preprocessing stage that consists in reformatting and tagging utterances that breaks the syntactic structure of the text. The chunking stage uses large-coverage and fine-grained lexical resources for general language that have been augmented with resources specific to spoken. We show that it reaches a score of 84.1% f-measure