Based on Mande languages as an example, the article examines the possibilities of using regular polysemy and regular homonymy for solving problems of comparative studies and semantic typology. Three cases are investigated. Regular polysemy noted only in Mande languages can be used for semantic reconstruction of Proto-Mande. Regular homonymy, noted only in the Mande languages, makes it possible to distinguish phonetically similar roots of Proto-Mande. Regular homonymy, noted not only in the Mande languages, but also in other branches of the Niger-Congo, makes it possible to discover regular phonetic correspondences between the languages of various branches of the Niger-Congo without any intermediate reconstructions.