Complex edges, transparent frontiers: grammatical complexity and language spreads
The Caucasus and the eastern Eurasian steppe are geographically very different but both offer typological and sociolinguistic situations where isolation and complexity can be measured and their correlations with other structural and sociolinguistic factors tested. This paper uses cross-linguistic surveys in both areas tp measure two kinds of grammatical complexity: (1) size of inventory or structural complexity (number of phonemes, of genders, of verb inflectional categories, of verb derivational classes, of alignment types, of basic word orders, etc.) and (2) opacity (non-transparency, non-biuniqueness: e.g. allomorphy, suppletion; sandhi, fusion, multiple exponence; synthesis, incorporation; classification; etc.). In each of the two areas complexity proves to be greater at peripheries of language spread than at centers; opacity is greater in sociolinguistically isolated languages (true whether the sociolinguistic isolation is ascertained directly or through a geographical or demographic proxy). Especially favorable to low opacity and complexity is a variable standing monopoly on inter-ethnic function and spread, where several typologically and/or genealogically close languages have a back-and-forth history of language shift and accommodation now in one direction and now in another. Thus both areas support the correlation of complexity with isolation. The paper uses results of my own fieldwork in the Caucasus as well as published grammars and historical and ethnographic information.