Non-moraic schwa

This dissertation proposes that schwa can be non-moraic (Kager 1989, 1990, Féry 1995, 1996), analogous to high vowels. In addition, such moraless schwas can head syllables (called ‘minor syllables’ after Matteson 1965, Lin 1993, 1997, 1998, Shaw 1994, Gafos 1998). Non-moraic, monomoraic, and bimoraic schwa can co-exist in the same phonological system. One of the major empirical consequences of this theory is that it accounts for stress systems in which stress avoids schwa. I go further in claiming that non-moraic schwa is the only means by which stress systems are sensitive to vowel quality, contra Kenstowicz (1997), de Lacy (2002, 2004, 2006), and others. I argue that non-moraic, monomoraic, and bimoraic schwa co-exist in Piuma Paiwan, an Austronesian language that has been reported to have sonority-sensitive stress (Chen 2009a, b, Yeh 2011). My fieldwork and experimental results provide acoustic evidence that stress avoids landing on a schwa. I argue that such avoidance is a side-effect of schwa’s prosodic status: schwa is usually non-moraic in Piuma Paiwan. However, schwa is required to be monomoraic when it appears in the non-head position of a foot, and bimoraic when it is forced to be in the head syllable of a foot. The different kinds of schwa have significantly distinct phonetic effects, particularly in duration and vowel quality variability.The theory proposed here predicts that stress should never avoid non-central vowels. One of the major challenges to this prediction is found in sonority-driven stress systems that seem to make peripheral vowel distinctions. However, I will present experimental evidence that the most discussed example of such a system – Gujarati – has been described incorrectly (cf. de Lacy 2004). Of the five types of phonetic evidence examined, only F1 provides clear evidence for stress, revealing stress to be consistently penultimate, and not sonority-driven. I will also show that many descriptions of putative sonority-driven stress lack robust phonetic and phonological evidence. Finally, I present an Optimality Theory factorial typology of constraints relating to schwa moraicity, and identify important rankings for grammars with various effects.