A subsegmental correspondence approach to contour tone (dis)harmony patterns

A long-standing question in phonological theory is whether there are parallels in the behavior and representation of contour tones and contour segments—that is, segments with distinct gestures sequenced in time (e.g., prenasalized consonants, diphthongs; Sagey 1986; et seq.)—and, if they exist, what can be gained from the understanding of these parallels (e.g., Yip 1989). Zhang (2001, 2004), for example, noted distributional and phonetic similarities between contour tones and contour segments (i.e., diphthongs). In this paper, we argue that there are also parallels in the syntagmatic behaviors of contour tones and segments: namely, that contour tones sometimes participate in harmony (spreading) and disharmony (OCPtype restrictions) as whole units and, at other times, their internal components act independently. Formally, such schizoid behavior from both contour tones and segments in (dis)harmony patterns has challenged phonological theory (for tone: Duanmu 1994; a.o.; for segmental phonology: Lombardi 1990; a.o.). As a solution, we propose to extend a novel phonological representation for subsegmental units couched in surface correspondence theory (Agreement by Correspondence (ABC); Hansson 2001; Rose and Walker 2004; a.o.) to contour tone patterns in harmony and disharmony systems. The resulting approach, which we term ABC+Q, makes it possible for the first time to offer a united treatment for the behavior of both contour tones and contour segments across observed patterns of (dis)harmony. Building on insights from Autosegmental Theory (Goldsmith 1976), Aperture Theory (Steriade 1993), and Articulatory Phonology (Browman and Goldstein 1989; Gafos 2002), the representational component of ABC+Q subdivides all segments Q into a maximum of three quantized subsegmental units q, which are temporally-ordered and host unitary sets of distinctive features: