Can Grammar Define Similarity of Human Natural Languages?

The aim of the present study is to show that similarity of human natural languages can be conveyed not only by phonetic data, but also by grammar. The paper regards the largest typological database WALS and its possibilities in the sphere of genealogic relationship of languages. Using the method of two-objective optimization and data mining, which is new for linguistic studies, we show that grammatical (structural) data, as well as phonetic data, can deliver information on the similarity of languages. Language isolates and micro-families do not have genealogic relatives based on phonetic information, but they do have genealogic relatives based on grammar information.