The labelling of phonemes is fairly arbitrary: we say English has an /d/
phoneme, for example, but it may not ever be realized as a true perfect IPA [d]
. Reconstructed languages often have phonemes not named after any IPA symbol, such as Proto-Indo-European's /*ǵ/
(what would it even mean to put a high-tone mark on a velar plosive?) or Proto-Slavic's /*ъ/
(not even a Latin character). Some reconstructions (such as Alice Kober's early analysis of Linear B) have even just given each phoneme a number, to avoid committing to any ideas of what it might have sounded like.
So the phoneme in Japanese is usually written /u/
or /ɯ/
, whichever the individual doing the analysis prefers. Neither is wrong: you could just as well call it "vowel number three" and write it /3/
, and that wouldn't be wrong either. What's important is just that readers understand what phoneme you mean, and neither /u/
nor /ɯ/
(nor even /3/
) is likely to be mistaken for any other phoneme.
Phonetically, the most consistent features of Japanese /u/
is that it's high and non-front. It sometimes moves as far forward as central [ɨ]
after palatals, and might have the lips rounded, unrounded, or compressed ([ɯ̟ᵝ]
). The version used in the prestige dialect of Tokyo in careful, formal speech tends to be more like [ɯ̟ᵝ]
, so we might well call the phoneme /ɯ̟ᵝ/
, but that's just too annoying to write over and over.