Consonant mutation and reduplication in Seereer-Siin

Abstract

Among the many types of non-concatentative morphemes are auto- segments or floating features which, in order to be realised phonologically, must attach to a root node in the stem. Akinlabi (1996) terms this process FEATURAL AFFIXATION . Featural affixes are of particular interest for models of reduplication, which make varying predictions as to whether or not they will be copied. In segmental models such as that of Marantz (1982), where nothing other than the underlying segmental base of the melody is copied, we would not expect the featural affix to be ‘transferred’ (Clements 1985a, Myers & Carleton 1996) between base and reduplicant.In models where the entire base is copied, such as that of Steriade (1988), featural transfer depends on the particular conception of the grammar. Within serial models of the grammar featural transfer depends crucially on the ordering of word-formation rules and phonological rules, while in a non-serial model like Optimality Theory (Prince & Smolensky 1993) featural transfer depends on the rank ordering of relevant constraints. The goal of this paper is to contribute to an understanding of featural affixation and featural transfer by presenting a constraint-based account of consonant mutation in Seereer-Siin, an Atlantic (Niger-Congo) language related to Fula, and by focusing specifically on the interaction of consonant mutation and reduplication. In the analysis presented here I consider mutation to be the result of featural affixation to the root node of a stem-initial consonant.