Phylogeny of the Turkic Languages Inferred from Basic Vocabulary: Limitations of the Lexicostatistical Methods in an Intensive Contact Situation

Abstract

This article provides an attempt to revise the phylogenetic structure of the Turkic family using a computational lexicostatistical approach. The methodological framework of the present research is characterized by the following features: (1) wordlists with strictly controlled semantics; (2) step-by-step reconstruction using Swadesh wordlists for proto-languages; (3) three stages of post-processing of the input data (analysis of root cognacy, elimination of derivational drift, and optimization of homoplasy); (4) application of several computational algorithms (Starling neighbor-joining, Bayesian MCMC, and maximum parsimony). The analysis provided confirms the status of Chuvash as the first outlier and suggests a subsequent multifurcation of Proto-Nuclear-Turkic into eight branches. The Siberian Turkic group is a purely areal unity, that is, Yakut-Dolgan, Tofa-Tuvinian, Khakas-Mrassu, Sarygh Yugur and Altai do not form a clade. Altai is grouped together with the Kipchak languages as a separate taxon; it does not show a particularly close relationship with Kirghiz, which belongs to another Kipchak subgroup. Karluk is a low-level taxon inside the Kipchak clade.