An Optimality Theoretic Approach to Accent Insertion

Abstract

This paper explores the accentuation in the words consisting of inherently unaccented morphemes in the languages which distinguish between inherently accented and unaccented morphemes. In this paper, a variety of accent insertion in non-Indo-European languages (Turkish, Tokyo Japanese, Cupeno, and Thompson River Salish) is presented and compared with the word-initial accent insertion in Vedic Sanskrit, which is an Indo-European language. Then, from the viewpoint of Optimality Theory (OT) (Prince and Smolensky 1993 and McCarthy and Prince 1993a, b), it is argued that the parametric difference in the ranking relation of ALIGN-R(ACC, PrWd) and ALIGN-L(ACC, PrWd) between Vedic Sanskrit and Turkish causes the difference in accent insertion between these languages. Further, the examination of accent ual systems of Cupeno and Thompson River Salish shows that Root Faith » Affix Faith as a met a-constraint is too fixed to explain a cross-linguistic variety of places for an inserted accent. This examination also shows that the difference in accent insertion among languages should be explained by the different rankings among domain-specific DEP-IO-faithfulness constraints.