About
SlaViCoN investigates the structure of Slavic verb systems using methods from computational network analysis. Rather than treating verbs as isolated words, the project models relationships between verbal elements, such as prefixes, roots, and theme vowels, as interconnected networks.
Slavic languages are particularly well suited for this type of analysis because their verbal systems are highly combinatorial. Verbs are formed through the interaction of multiple morphemes that can combine in different ways and contribute distinct semantic and grammatical properties. For example, Bosnian, Croatian, Montenegrin, and Serbian verbs such as do-ček-a-ti ‘welcome’ and sa-ček-a-ti ‘wait’ share the same root ček and theme vowel -a-, but differ in the prefix, resulting in different meanings.
Similarly, Slovenian verbs such as rumen-e-ti ‘become yellow’ and rumen-i-ti ‘make yellow’ share the same root while differing in their theme vowels, again producing systematic semantic contrasts. SlaViCoN explores how such relationships can be represented computationally and what they reveal about the organization of Slavic morphology.
The project draws on large annotated corpora of Bosnian, Croatian, Montenegrin, Serbian, and Slovenian verbs, including the database WeSoSlaV, which contains more than 8,000 verbs annotated for morphological, semantic, syntactic, and phonological properties.
Computational network measures such as connectivity, centrality, clustering, and similarity are then used to identify structural regularities and productivity patterns within the verbal system.
A central aim of the project is to address broader theoretical questions concerning the nature of morphemes and the relationship between morphology, syntax, and semantics.
More broadly, bridging symbolic and stochastic approaches to language is a key contribution of SlaViCoN.