Grammar-based phylogenetic trees: prospects and disadvantages

Makarova Elena Andreevna

The present work aims at proving that the information on the language structure can be used as a tool to classify it. Despite grammar being more prone to borrowings and changes than lexical and phonetic data, the history of linguistics provides a big number of examples when hypotheses on the kinship of languages were suggested basing on one or several grammar features. Two sets of languages from the typological database “Languages of the World” of IL RAS are used to calculate the Hamming distances and build two phylogenetic trees: one for eight different families (Uralic, Mongolic, Turkic, Northeast Caucasian, Mande, Dravidian, Semitic, and Indo-European), and the other for five branches of the Indo-European family (Slavic, Germanic, Romance, Baltic, and Anatolian). Further analysis of the results will allow us to answer the question: how accurately grammar-based trees represent the real state of affairs and, consequently, can they be used to classify languages.

Keywords:quantitative linguistics, phylogenetics, “Languages of the World” of IL RAD database, computer linguistics.