Boundary tones and the phonetic implementation of tone in Chichewa

Abstract

In Chichewa (Bantu, Malawi), there are three differences in fundamental frequency (£0) between questions and statements: (a) questions have a fmal rise, while statements have a final fall, (b) questions do not display the strong downdrift trend found in statements, and (c) questions are produced in a higher pitch range than statements. A quantitative model is proposed that captures these generalizations through a dependency of the bottom of the current pitch range on the boundary tone. It is argued that such a model accounts for the distinction between question and statement better than a model in which pitch range distinctions are encoded in the phonological representation.