Both substance and structure of each of the members are relevant for the classification of a language family; therefore both genetic (chiefly of statistical nature) and typological criteria should be employed. Typological criteria are subordinated because: (1) a language family is a genealogical category and (2) typological elements merely specify the realization of substance. Of course, in comparing the members of a language league typological criteria alone are relevant because a league is a typological category. Naturally, a genetic classification often displays elements and features characteristic of a league. Thus in the East Slavic group a considerable percentage of common traits derives from typological convergence. This is true for the Romance family even to a greater extent. On the other hand, a partial league having no genetic counterpart may appear within a family (such as Polish-White Russian-Ukrainian league within the Slavic family). It would be ideal if diasystems only were considered in comparative operations, both genetic - in families, and typological - in leagues. Since such diasy- systems are not available in linguistics as yet, standard languages and dialect materials serve the purpose. This should not prove a serious drawback if one is aware of standard languages acting in such cases as substitutes and due corrections are made in the process of comparison (e. g. on historical or dialectal levels), but it may be so if applying the equation »standard language = language in general«. Language in g en eral can function as a diasystem of language »rank« only, i. e. of rank higher than that of dialectal diasystems. Complications arise in the rare cases of diasystems having language rank but without its Standard member (Sardinian, Kasiubian), or having more than one (Rumanian and Moldavian standard languages), or if the area of a diasystem covers several standard languages (Low German diasystem comprising Dutch and High German languages). Varieties of a standard language (Serbian and Croatian, Flemish and Dutch, British and American) as well as the independence of nations (ethnological-social formations) from both standard languages and diasystems should be borne in mind. As a matter of fact, in the area of its use, each standard language forms a league sui generis with all its organic idiolects, related or not. A classification of standard forms of languages of a family would facilitate the understanding of the problem. Such classifications are not to be done on the basis of substance or structure but on the basis of standardness of the respective languages. As a matter of principle, these criteria should be of the type which is easily applicable to any family. The work on a classification of Slavic standard languages following these principles is in progress.
