Postnasal voicing, where an underlying voiceless nasal-stop sequence is mapped to a voiced nasal-stop sequence (NT > ND), is a common process that has a well-known phonetic motivation (e.g. Pater 1999, Hayes & Stivers 2000). The converse of postnasal voicing, postnasal devoicing (ND > NT), is uncommon and often claimed to be phonetically unnatural (e.g. Beguš 2019). In this talk I argue against this characterization by showing that postnasal devoicing is perceptually advantageous. The idea (following Stanton 2017) is that, all else being equal, the contrast between a nasal (N) and ND is less distinct than a contrast between N and NT. Speakers employ post-nasal devoicing as a form of enhancement, in order to render the N-ND contrast more distinct.
Evidence for this claim comes from a perception study, which has two main results. First, the N-NT contrast is more distinguishable than the N-ND contrast, both prevocalically and word-finally. Second, both the N-ND and N-NT contrasts are more distinguishable prevocalically than they are word-finally. I demonstrate that this second finding has parallels in the typology of PND: prevocalic PND entails word-final PND. I present an analysis of PND that references auditory factors and show how it can be extended to account for cases of postnasal aspiration. Time permitting, I discuss differences between the present proposal and that of Beguš (2019).
