Reconstructing non-contrastive stress in Austronesian and the role of the mora in stress shift, gemination and vowel shift

Competing schools of thought on the reconstruction of Proto-Austronesian stress contend that primary stress was either regular (falling on the penultimate syllable with possible phonetic conditions that triggered stress shift to the final syllable) or lexical (falling unpredictably either on the penult or ultima). In this study, I argue that the comparative evidence supports the first position: that primary stress fell regularly on the penultimate syllable and was not lexical. Further, primary stress was repelled to the final syllable if the penultimate syllable was open and contained a schwa nucleus. Three Austronesian first-order subgroups, Malayo-Polynesian, Western Formosan, and Paiwan, are shown to directly continue the reconstructed stress system of Proto-Austronesian, with stress falling regularly on the penultimate syllable but shifting to the final syllable after a schwa. I also argue that the inability of schwa to hold stress is a result not of quality, but rather of quantity, as it is shown that schwa was a zero-weight vowel in Proto-Austronesian. Words with a schwa in the penultimate syllable, CəCVC, are shown to be sub-minimal, containing only a single mora. Daughter languages in Malayo-Polynesian underwent multiple cases of phonologically motivated drift, including consonant gemination, the deletion of penultimate schwa in three-syllable words, and vowel shift. These sound changes are argued to be part of a phonological conspiracy whose outcome is the addition of a mora to sub-minimal words. This study therefore offers both a reconstruction of Proto-Austronesian stress as well as a phonological explanation for these various sound changes in Malayo-Polynesian.