Prominence from complexity: Capturing tianjin ditonal patterns

This paper explores the six ditonal sandhi that spawn from an inventory of four tones (Low, High, Rising, and Falling) in an attempt to provide an account for why only certain ditonal combinations trigger alternation but not others. Earlier accounts have relied on Obligatory Contour Principle (OCP) applying to adjacent syllables both at the level of the full tone contour and also of the tonal features that comprise the contour. Another account has explained sandhi in terms of leveling across syllables so that excessive contours are avoided. A careful exploration reveals that neither of these approaches is adequate, nor are some of their stipulations necessary. Instead, a more viable solution might be found if we accept that Tianjin prosody is right-headed such that prosodic prominence is reflected through tone complexity. To this end, I propose the Head Tone Complexity (HTC) constraint that partners with OCP to generate the attested ditonal sandhi patterns.