Valency patterns across Slavic: insights from a token-based cross-linguistic perspective

A token-based typological perspective sheds new light on the comparative study of valency class systems in ten modern Slavic languages. Using Universal Dependencies (UD) treebanks, the analysis addresses two issues: cross-linguistic differences in valency class system complexity and the relationship between type and token frequencies of argument-coding devices such as cases and adpositions. Quantitative measures based on transitivity prominence and entropy reveal systematic intra-Slavic variation. Languages of the northern Slavic area (Russian, Ukrainian, Polish, Czech) show higher valency class complexity than South Slavic languages, despite their close genealogical relationship. Complexity metrics based on UD treebanks largely converge with questionnaire-based results from the BivalTyp database, supporting the methodological validity of the approach. Type and token frequencies of individual argument-coding devices are shown to diverge in systematic ways. Cases and case-preposition combinations associated with semantically more homogeneous verb sets and more tangible semantic roles tend to display higher token frequencies relative to their type frequencies. These findings highlight the added value of token-based evidence for understanding the structure and variation of valency class systems.