Reconciling the debate about final obstruent voicing: The phonology of Lakota obstruent lenition

Abstract

Word-final obstruent voicing has been claimed to be unavailable as a synchronicphonological process due to universal markedness constraints. However, Evolutionary Phonology (Blevins 2004) predicts the possibility of such a phonologicalrule, and purports to show it in various languages, touching off a debate withKiparsky (2006, 2008). Searching for a more robust example, Blevins et al. (2020)provide a phonetic study of voicing in Lakota. They show that Lakota has phoneticalternations between voiceless and voiced stops, with the voiced stops appearingin final position. They therefore claim the language has a synchronic phonologicalrule of final obstruent voicing. We call into question the assumed phonological status of final voicing in Lakota. Phonetically, Lakota does have only voiced stops infinal position, but at the same time final fricatives are neutralised to the voicelessseries. We show that, rather than neutralizing stops toward a marked or complexstructure, Lakota has only a process of phonological lenition in weak positions:sonorisation of stops and devoicing of fricatives. Under the assumption that manner of articulation is a structural property, a single lenition process can producefinal voiceless fricatives but final voiced stops.