A Classification of the Slavic Languages based on the material of the Slavic Linguistic Atlas: A Trial Run (original: Классификация славянских языков на материале Общеславянского лингвистического атласа: пробный шар)

Abstract

This study proposes a classification of Slavic languages based on an objective dialectometric analysis of the materials of the Slavic Linguistic Atlas. Unlike traditional classifications, which rely primarily on literary languages and are influenced by extralinguistic factors, this method focuses on the dialect continuum. The work is based on data from 6 volumes of the phonetic-grammatical series of the Slavic Linguistic Atlas, containing information on the reflexes of 7 Proto-Slavic vowels (*ě, *ę, *ǫ, *ъ, *ь, *o, *e) in 780 localities. For each reflex, linguistic characteristics (height, backness, length, rounding,
nasalization) were determined and presented as vectors. The key stage was a pairwise comparison of reflexes between all locations with the calculation of the Euclidean distance, taking into account structural differences (for example, between monophthongs and diphthongs). Based on the constructed matrix of linguistic distances, hierarchical clustering by Ward’s method was applied. The analysis revealed that, for the material considered, the optimal division of the Slavic area is into 15 clusters. The algorithm confirmed established views in Slavic dialectology, as it correctly separated the East, West, and South Slavic areas and accurately attributed locations outside state borders. However, shortcomings associated with relying solely on vocalism were also identified, the most striking of which was the erroneous unification of the Torlak and Sorbian dialects.
The study demonstrates the potential of quantitative methods to reveal the structure of a dialect continuum, free from subjective assessments. The resulting classification serves as an important benchmark but does not replace traditional approaches.