High Vowel Devoicing in Tohoku Japanese is Conditioned by Foot Structure

I argue that high vowel devoicing in Tohoku Japanese constitutes evidence for sonority-driven foot structure in natural language. While the application of high vowel devoicing in Tohoku Japanese is determined by seemingly unrelated factors, I show that these factors can be unified by making reference to sonority-sensitive foot structure. Based on a modified vowel sonority hierarchy where voiceless vowels are the least sonorous, I introduce two foot well-formedness constraints — one that bans voiceless vowels (low sonority) in a foot head, and another that bans non-high vowels (high sonority) in a non-head position. Devoicing is blocked in order to avoid violations of the foot well-formedness constraints, when a target vowel for devoicing is in a foot head. Conversely, devoicing is licensed when the target high vowel is not in a foot head and no longer subject to these constraints. In addition, I show that loanword accentuation patterns provide further evidence for sonority-sensitive footing. Specifically, I propose that accent shift from the default position is attributed to the effect of the same foot well-formedness constraints used for the analysis of high vowel devoicing. In conclusion, this study provides novel evidence for sonority-sensitivity in foot structure, independent of stress assignment.