Vowel harmony in Stau

Abstract In this paper, we propose that in Stau (Rgyalrongic, Sino-Tibetan) there is a system of four vowel pairs (/i/-/ə/, /e/-/ɛ/, /æ/-/ɑ/, /u/-/o/) that undergo regressive vowel harmony. This system of vowel harmony produces root morpheme forms such as [æCæ] and [ɑCɑ], whereas forms like [æCɑ] or [ɑCæ] are remarkably absent. Vowel harmony is also observed when combining morphemes to create complex words. Unambiguous cases show that the vowel of the first morpheme assimilates to the vowel of the second (e.g. /æCɑ/ → [ɑCɑ]), if the vowels of the two morphemes belong to the same vowel pair set (e.g. /æ/-/ɑ/). At the same time, there are several situations where vowel harmony is consistently not observed. Notable among these blocking mechanisms are lexemes that contain semantically-heavy first syllables. In this situation, we observe the interplay of semantics and phonology. Analysis of the vowel harmony system in Stau yields a way to measure the distance between morphemes.