A reexamination of the Philippine-type voice system and its implications for Austronesian primary-level subgrouping
This dissertation investigates the nature of the Philippine-type voice system and two associated diachronic questions: (i) what is the nature of noun-verb (nominalizer-voice affix) homophony, a common trait of Philippine-type languages?, and (ii) does the synchronic variation of this voice system among higher-order Austronesian languages constitute valid evidence for Austronesian primary-level subgrouping? Using novel comparative data from Puyuma, Amis, Seediq, and Tagalog as the empirical starting point, I explore the shared syntax of Philippine-type languages and its implications for these two questions.
In the first half of the study, I argue that Philippine-type languages are best analyzed as exhibiting a nominative-accusative case system with prominent topic-marking that overrides morphological case. I then provide a novel account for the nature of the four-way division of the Philippine-type voice morphology: the four affixes are best analyzed as the spell-out of four different bundles of Agree relations that agree with the topic of a clause. Under this analysis, Philippine-type “voice” is fundamentally different from Indo-European “voice”. The latter is valency-rearranging morphology, while the former is topic-indicating morphology. Building on this analysis, I argue that Philippine-type languages are best characterized as discourse configurational languages (Li & Thompson 1976; Kiss 1995; Miyagawa 2010, 2017), whose topic-prominent nature is manifested both in prominent topic-marking and in articulated verbal morphology that indicates the Agree relations of the topic in a clause. I conclude that Philippine-type languages are best analyzed as hosting a topic-feature on C and the φ-feature on T, with topic-agreement spelled-out as verbal morphology.
In the second half of the study, I demonstrate how this synchronic syntactic analysis enables a simpler solution to two central issues in Austronesian diachronic linguistics (i)-(ii). Building on the conclusions outlined above, I argue that the presence of either (a) nominalizer-voice affix homophony, or (b) Philippine-type voice distinctions in root-clause environments does not constitute valid evidence for Austronesian primary-level subgrouping, as the absence of both features reflects independent morphological erosions in innovative languages. I conclude that phonological innovations are better evidence for Austronesian primary-level subgrouping than the morphosyntactic variation among higher-order Philippine-type languages, as the latter is best viewed as a product of independent drifts, rather than shared innovations.
[a finalized version will be uploaded in Dec. 2017]