Vers une lexicographie mandingue sur la base de grands corpus annotés

Availability of big electronic annotated corpora for Maninka and Bamana languages lets us think about development of corpus-driven dictionaries, i.e. dictionaries of a new generation. In a corpus-driven dictionary, elaboration of the polysemy, collocations, syntactic valencies, stylistic, regional and dialectal characteristics of each lexeme (and of each sense of a lexeme) will be based on the analysis of a concordance extracted from the corpus. Henceforth, it becomes possible to evaluate statistically each feature. Two sample examples of lexicographic analysis of Bambara lexemes, bánban and jàhadi, are given.