Xitsonga tone: The syntax–phonology interface

Abstract

AbstractPatterns of prosodic-structure-sensitive high tone spread in the Bantu language Xitsonga reveal mismatches between syntactic and phonological/prosodic constituency. A modular account of these mismatches is proposed: Match constraints (Selkirk 2011) are re-construed as spell-out constraints relating the output representation of the morphosyntax to the input representation for the phonology. It’s argued that in Xitsonga the spell-out constraint MatchPhraseLEX relates only phrases with lexical category heads to phonological phrases in the phonological input representation. In the phonology per se, a novel class of prosodic-structure faithfulness constraints interacts with prosodic-structure markedness constraints to produce further constituency mismatches in the output phonological representation.