Bambara-to-French Statistical Machine Translation System

The Bambara, an unwritten but orally preserved SOV language, is used as the Lingua Franca in Mali, it is spoken in neighboring countries like Ivory Coast, Guinea, Burkina Faso and Gambia. With 80% of Mali's population speaking Bambara language, efforts to prevent its extinction led to a proposed Bambara to French machine translation system. To facilitate this, a Parallel corpus was gathered from the New Testament bible, and the translation model was constructed using statistical approaches such as Bayes' Theorem, IBM Model 1, IBM Model 2, and a search algorithm. The system, developed using the Unified Modeling Language (UML) and Python, underwent performance evaluation using the BLEU metric. Testing involved 100 sentences from various biblical contexts. The results indicated a 66% BLEU score for translating the SOV to SVO sentences, with 100%, 40%, and 20% accuracy in translating sentences from the New Testament, the Old Testament, and non-Biblical words, respectively. Moreover, the study emphasized the importance of a diverse and extensive corpus in enhancing the system's training and improving its accuracy. Ultimately, the Bambara to French machine translator was recommended as a tool for preserving the Bambara language.