This dissertation investigates the phonetics and phonology of laryngeal contrast in Nepali, an Indic (aka Indo-Aryan) language with a four-way laryngeal contrast in its stops and affricates between classes often described as voiceless, voiceless aspirated, voiced, and voiced aspirated. Specifically, it brings Nepali data to bear on three phenomena central to the study of laryngeal contrasts: representation of laryngeal contrast, the effects of laryngeal specification on preceding vowel duration, and patterns of neutralization and weakening of laryngeal contrasts. Despite the diversity in language-specific phonetic cues used to realize laryngeal contrasts (voicing, aspiration, etc), each of these phenomena is sometimes referred to as phenomena of "voicing": voicing contrasts, the voicing effect on pre-coda vowels, and final devoicing. There is now a body of literature that grapples with the connections between these phenomena and the language-specific cues used to realize contrast, and their theoretical implications. But despite an ever-increasing amount of phonetic and phonological work in languages with complex laryngeal contrasts, these theoretical conversations and proposals often still focus on languages with two-way contrasts, asking: do these phenomena behave the same in languages with voicing contrasts vs. languages with aspiration contrasts? Should the segments be specified for the same laryngeal features in both types of languages? This dissertation asks: how do these phenomena behave in a language with both a voicing contrast and and aspiration contrast, including a class with both voicing and aspiration?

Each chapter of this dissertation addresses this question in light of a different phenomenon typically described as "voicing". Chapter 2 presents a production study examining the phonetics of "voicing" contrasts in initial position and measures of speaker control proposed to diagnose feature representation in initial and intervocalic positions, finding evidence for representations using [voice] and [spread] features. Chapter 3 engages with the longstanding question of how and why laryngeal specification of a consonant affects the preceding vowel's duration. In order to assess a variety of explanations for the source of this effect, it examined the effects on vowel duration of not just voicing specification but aspiration specification as well. The results are problematic only for exclusively articulatory explanations, but consistent with perceptual explanations, and with analyses in which multiple factors and forces affect vowel duration. Through both production and perception studies, Chapter 4 shows that the label of "final devoicing" is insufficient to capture the patterns of contrast maintenance and weakening that target Nepali's aspiration and voicing contrasts in word-final position. It finds that the voicing contrast remains robust, while the aspiration contrasts is weakened, confused, and shifted.

This dissertation proposes that Nepali motivates a tighter connection between phonetics and phonological representation than whether segments should be specified for the same laryngeal features in voicing vs. aspirating languages, arguing for laryngeal specification on a subsegmental level.