The paper sums up some of the results of the previous research on this topic with regard to lexical, morphological and phonetic parallels, points out some of the problems of formal and semantic interpretation of these parallels and proposes a new approach to the comparative study not only on the basis of individual lexemes but on the basis of established models. The relatively high number of regular phonetic parallels coupled with a number of what appears to be irregular correspondences, does not allow the interpretation of these parallels in terms of a classical language family, but rather in terms of a prehistoric linguistic relationship filtered and altered through an ancient linguistic area, or areas.