Phonology of vowel devoicing: A typological perspective

Abstract

This dissertation presents a systematic phonological analysis and typology of vowel devoicing based on a large-scale cross-linguistic survey of 70 processes from 53 languages. While previous literature on the phenomenon has primarily focused on its phonetic properties and mechanisms, I argue that vowel devoicing’s interaction with other phonological processes in a number of the languages in which it occurs necessitates an account of its phonological mechanisms as well.I propose a typological model of vowel devoicing that focuses on the distinct roles of prosodic structure and segmental phonology in conditioning its application, which I treat as two separate, but potentially interacting parameters. This yields three categories of vowel devoicing. When the prosodic parameter operates alone, it yields processes sensitive only to prosodic structure. Specifically, processes in this category are restricted to certain prosodically weak positions. When the segmental parameter operates alone, it yields processes sensitive only to the local segmental environment. The processes in this category are restricted to vowels adjacent to particular voiceless and/or aspirated consonants. Finally, the interaction of the two parameters yields processes sensitive to both prosodic structure and segmental phonology. The processes in this final category are restricted to vowels in specific weak positions that are also adjacent to specific types of consonants. Crucially, since all three scenarios are well-attested cross-linguistically, I argue that a complete phonological analysis of vowel devoicing cannot reduce the phenomenon to a single mechanism, but rather, must include separate prosodic and segmental mechanisms. I then propose a formal, constraint-based analysis of vowel devoicing that attributes the two distinct mechanisms to two different types of markedness constraints. While this analysis is general enough to account for all processes in the relevant typological categories, I also illustrate its language-specific applications with case studies from diverse families and geographic regions, including Cheyenne (Algonquian, North America), Comanche (Uto-Aztecan, North America), Niuafo’ou (Malayo-Polynesian, Tonga), and Wallaga Oromo (Cushitic, Ethiopia). Finally, I demonstrate that when vowel devoicing is considered from a phonological perspective, it yields crucial implications for larger issues in phonological theory. These include the nature of laryngeal features in Feature Theory, the distinct roles of prosodically strong and weak positions in segmental phonology, and the nature and role of domain generalization in the distribution of phonological processes within a given language.