Exploitation Of The Lexical Environment In The Automatic Processing Of The Possessive Ka In Bamanankan

To process bamanankan (bambara) in an artificial intelligence program, it will first be necessary to solve the problem of homograph units of distinct lexical category. The ka is a good illustration of this spelling overlap phenomenon: ka (marker of the qualifying adjective), ka (infinitive), ka (possession), ka (imperative). It is therefore necessary to have material indices to distinguish the ka occurrences from each other in a text. As such, the present study seeks to unmask the lexico-syntactic indicators capable of disambiguating the ka of the possession from other ka with different categories. The potential uses of the artificial processing of possession ka are numerous, as these techniques are involved at least in machine translation. This work formulates hypotheses on lexical indicators that disambiguate the ka. The hypotheses are verified through appropriate search terms which extract the respective occurrences from a corpus using the Bosolan software. The occurrences are studied to distinguish between confirming and disproving cases of the hypothesis. Indeed, material indicators that can guide a computer program have been found through the exploitation of the lexical environment. About 6 stated hypotheses, more than the average occurrences were confirmed.