Limits of variability in complexity of valency class systems
From January 2026, Sergey Say and Ilja Seržant have been working on the project D05 “Limits of variability in the complexity of valency class systems” as part of the DFG-funded Collaborative Research Centre (CRC) 1287 “Limits of Variability in Language: Cognitive, Computational and Grammatical Aspects”.
This project investigates cross-linguistic variation in the complexity of valency class systems, combining insights from argument coding, linguistic complexity, and token-based typology. Using corpus data from 70 languages, it develops information-theoretic metrics based on entropy and surprisal. The study explores local factors shaping argument coding and examines cross-linguistic limits on valency system variability. It tests the hypothesis that natural languages avoid both overly simple and overly complex systems and shows that valency class complexity is constrained by its interaction with other typological features.
The current Phase III of the project (planned for 2026 – mid-2029) builds on a preparatory stage (2023–2025), in which we developed statistical techniques for measuring valency class complexity using both questionnaire-based and token-based cross-linguistic data. Some of the results are presented in the following publications.
Say, Sergey & Ilja Seržant. 2025. A Frequency-Based Algorithm for Argument Extraction from Russian Treebanks. Neophilologica 37, 2025. DOI: https://doi.org/10.31261/NEO.2025.37.07
Say, Sergey. 2025. Degrees of complexity in valency class systems: implications for efficiency. STUF - Language Typology and Universals 78 (4). https://doi.org/10.1515/stuf-2025-2022
Say, Sergey. Valency patterns across Slavic: insights from a token-based cross-linguistic perspective (accepted for publication in Slovo).