VVNC in Kamituumbi and Kikongo

Abstract

Vowels before NC sequences in Bantu languages are often long. In certain languages, morphophonemic length alternations provide support for a synchronic rule turning VNC into VVNC; in other languages, we find root-internal VVNC, but no length alternations which directly support a lengthening rule. The question arises whether, in such languages, vowel length before tautomorphemic NC is governed by phonological rule or by redundancy statement on underlying forms. In this study Kikongo and Kimatuumbi are compared on this point. It is shown that the languages are alike in having root-internal VVNC, but no direct vowel length alternations. However, these languages give different solutions to the redundancy problem. In Kimatuumbi vowels are simply redundantly long within roots before NC clusters. In Kikongo, analysis of tonal alternations demonstrates that long vowels before NC derive from underlying short vowels. Owing to borrowing, there are many stems in Kikongo where short vowels survive before NC, a fact which on the surface weighs against any rule lengthening vowels before NC. However, a study of tone rules shows that at some level of analysis, C[Vacute]C[Vacute] and C[Vacute][Vacute]NC[Vacute] nouns are prosodically equivalent, in contrast to C[Vacute][Vacute]C[Vacute] The evidence for treating Kikongo VVNC as deriving from VNC therefore comes from a relatively abstract property of tones in the language, namely, how adjacent identical tones are represented. Thus the system of rules points to a different analysis than would be constructed solely on the basis of surface contrasts. What must be remembered is that a child learning Kikongo has access not just to surface patterns, but must construct a grammar accounting for all of the data, including tone alternations which may reveal something quite at variance with surface observations.