'titrebSummary’/titrebThis study is part of a series of work on the automatic generation of moving image indexing. Textual media related to audiovisual production (production scenarios, subtitles for hearing impaired, audiovision for visually impaired) are used here for the purpose of extracting vocabulary allowing the indexing of sequence films. The context of this research, the method adopted and the results obtained from a specific example are presented and discussed. This work provides a solid theoretical basis for the exploitation of texts created during the pre-production, production and post-production processes of audiovisual documents as a material source.