This PhD proposal is in the field of Sign Language processing, specifically machine translation from text (in English or French) to Sign Language (SL).
Sign Language is an oral gestural language involving many simultaneous articulators, which we prove is not describable in sequences of lexical units. We have designed a Sign-specific model at LIMSI to build formal grammars and cover SL properties better. These grammars are based on recursive rules, each of which carries a semantically-loaded function that can be triggered automatically by processing an input text to be translated. Triggered portions must then be combined to form the output translation. The work here will be to explore this combination issue, and eventually build a full text-to-sign translation system. Applications clearly exist in the domain of web accessibility to the Deaf public.
For further details, please download and read the PDF file here (in French).
Candidates are expected to apply for a grant from EDIPS, before summer 2014