Low Vowel Raising in Sinitic Languages: Assimilation, Reduction, or Both?

Many Sinitic languages respect low vowel raising (LVR) whereby the low vowel /a/ is raised to /e/ only when flanked by a palatal glide and a coronal coda. In this article, I show that LVR in Sinitic languages is a genuine process of grammar because low-level phonetic coarticulatory effects alone cannot account for crosslinguistic/dialectal variation. Specifically, my survey reveals that vowel duration, onset consonants and/or the presence of retroflexes may have a direct bearing on LVR, thus calling for a phonological analysis. I argue that the phenomenon in question is better treated as additive effects of (i) target undershoot in the wake of impoverished vowel duration (i.e. vowel reduction) and (ii) contextual influences (i.e. backness assimilation). Phonological vowel quality change (LVR here) is motivated only when the benefit of contextual neutralization exceeds that of the sum of the “cost” of implementing the above coarticulatory patterns. Analytically, Flemming’s (2003) feature co-occurrence constraints, together with phonetically driven constraints, provide a straightforward account of the relationship between coronal place and vowel backness attested in LVR. Finally, implications for Mandarin segmental phonology are also discussed.