Re: uh?

Zhen Lin wrote: 
Bart Mathias wrote:

[...]
/u/ lacks accuracy, but that's fine - it's a broad transcription. [...]

How can you have a "broad transcription" in phonemes? /u/ doesn't mean anything without reference to how it is pronounced, so it is very vague in a cross-dialect discussion.

[...]
?  To me, [ü] represents a high front rounded voiced vowel;

I was going for centralised high back vowel - the diacritic ¨ is used for centralisation in IPA. (_" in X-SAMPA)

Yes, I see that goes way back to the 1947 standard, the one I supposedly learned! Maybe [ü] is used for IPA [y] in American phonetics? Anyway, it's a high central *rounded* vowel in IPA, so the one after /s/ or /t/ has to be [ï], with a little circle underneath for /suki/.

Also /o/ and /a/, and probably /e/, though I can't provide an example. I'm a bit suspicious of [ju jü] here. You wouldn't use a [j] in transcribing things like /syuku/ | /Suku/, would you?

/sjuku/ > [SM_0_"kM_0] (or [Sü_0ku_0] for simplicity) - that's about the only place where I can imagine an underlying /ju/ being realised as palatalisation + [ü].

I've never seen Japanese [S] analyzed as /sj/ before, although it is of course only a variant of the very common /sy/. Of course, you're free to do what you want between "/"s, as long as you state the rules correctly. I am solidly in the /S/ camp, if only to avoid the problem of new-Japanese [Se] (including [tSe]).

I'm puzzled over [E:]. I took the [E] above as the sound in what I might spell as [mEt:ani], but when would the higher "e" occur long, distinct from [e:]?


/E/ is mid-open, /e/ is mid-close. I would say /ee/ 'yes' is [E:] whereas /eieN/ 'forever' is [e:e:~] (or maybe [e:::~] !). Not sure if there are any words which contrast [E:] and [e:] though, as with [E] and [e]...

A frightening idea. Maybe two frightening ideas. Are you saying [E] and [e] (short) belong to different phonemes, that they distinguish words? Examples? I'll *probably* believe that the first two moras in your /ee/ and /eieN/ (/e:/ for me in both cases) are different when I see several spectrograms of each that show it, but I hope that never happens.

[ju: jü:] is also a distinction I can't figure out yet.

/kjuu/ 'nine' [k_ju:] vs. /sjuu/ 'week' [Sü:]

OK, maybe. It has been my belief that long /u/s are about equally back, but I haven't seen any conclusive discussion of the matter yet. It's not like the vowel of "/kjuu/" is that far back either. (I really need a good spectrograph program for my computer...)

Bart
.