Nik Taylor wrote: >Word-final devoicing can only occur when the preceding consonant is >voiceless. Kaku can be pronounced [kakM_0], but oyogu can only be >[ojogM] or [ojoNM] (/g/ is optionally [N] when non-word-initial). And >also, the devoicing is optional. As I understand it, _su_ in such cases >is often pronounced actually as a sort of moraic /s/, such that _desu_ >is something like [des=] > How about final nasal plus vowel? I'd assume no devoicing there. What is the rule for word-medial devoicing? My impression is "between voiceless stops", but how about ...s__[vl.stop]... ? I also have the impression it involves /i, u/ more than the low vowels?? A friend once maintained that the "correct" pronunciation of _sukiyaki_ was more like "skiyaki", (I suppose [s=kiyaki]) which seems possible though I've never noticed it e.g. in Japanese restaurants.