The Bielefeld-Abidjan documentation project:
Information types and dissemination media

Dafydd Gibbon, Universität Bielefeld

The Bielefeld-Abidjan documentation project `Encyclopédie des Langues de Côte d'Ivoire' is funded by the Deutscher Akademischer Austauschdienst (DAAD, German Academic Exchange Service) with the goal of developing a specification and design for a multimedia encyclopedia of the four language groups of the Ivory Coast (Kru, Kwa, Mande, Gur). Initially, the Kwa languages are being focussed on, in particular the closely related langauges Baule and Anyi. Four exploration methods are being developed:

1. Detailed questionnaire on structural and functional aspects of language, with CGI-based form input and hypertext output.
2. Systematic acoustic phonetic analyses to extend the phonological analyses under (1).
3. Specialised phonetic databases, e.g. diphone database for testing models via speech synthesis.
4. Text collection with - audio - transcription - annotation (selected) - illustrated lexicon - HyprLex dynamic concordance (for models, see coral.lili.uni-bielefeld.de/HyprLex/)

The prototype tools for these tasks are being made available for use in the field. Close attention has been paid to appropriate levels of analysis for tools in this area. For example, the variety of mutually incompatible fonts for African languages, IPA, etc., involve much unnecessary engineering overhead in automatical processing, e.g. in concordance generation, and are not conducive to free information interchange. General computational principles suggest that this is an overly concrete level of representation and that a more generic and standardisable platform-neutral transcription format, such as the X-SAMPA machine-readable (ASCII) IPA coding, is more suitable, with conversion to reader-friendly fonts at a post-processor stage. We have developed X-SAMPA encodings for Bete (Kru) and Abbey (Kwa), and by the end of the project will have codings for the other languages currently being treated.