Gemination and segmental patterns with reference to Sanskrit and Assamese: An OT account

Abstract

This paper addresses the issue of segmental distribution and patterns with reference to the gemination processes in Sanskrit and Assamese within an Optimality theoretic model (Prince and Smolensky, 1993). Segmental properties inherent in a segment and cross-linguistic well formedness conditions play a significant role in triggering gemination. The underlying liquids and approximants /j/, /r/, /l/, and /w/ trigger gemination of the preceding obstruents in Assamese. However, what has been conspicuously observed in synchronic Assamese grammar is that, in the surface forms, /j/ gets dropped in gemination and a process of metathesis occurs in the form of an insertion of a vowel before the geminates. The process is analyzed in an Optimality theoretic model and I have proposed the ranking *CCG>>MAX C>> Linearity to represent the phenomenon. However, we need one more markedness constraint *V[back, round]glides to address the issue of obstruent geminates followed by approximant ‘w’. In such cases ‘w’ gets dropped in the surface form, and we do not get any process of metathesis operating here. Actually, what happens is a consequence of the fact that metathesis is not possible because Assamese does not allow / ɔw / or /Vu/ diphthongs. Hence, I propose that the constraints *V[back, round]glides, *Cw are higher ranked over [Max Round>>Max W] and Linearity.