Re: Not phonetic but ___???

John C:

> And Rosta scripsit:
>
> > as for /ju/, it would be counterproductive to treat it as a diphthong,
> > since its behaviour is so much to the contrary.
>
> Indeed, it would be plausible, if SPE-ish, to treat it as underlying /y/,
> realized by breaking.
I was thinking rather that the standard analysis, /j/ in onset
+ nucleus /u:/, looks to be by far the best, since the /j/
participates in yod-dropping and yod-coalescence -- characteristic
onset-affecting phenomena -- while the u element has exactly the
allophony of the GOOSE vowel, when due allowances are made for the
fronting effects of the preceding [j]. Countervailingly, though,
/j/ is a phonotactic anomaly both in the kinds of onset clusters
it can belong to and in the arbitrary-looking cooccurrence
restrictions on what the following nucleus can be. The anomalies
would largely disappear if /ju/ or "/y/" were a unitary phoneme,
but I find that unpersuasive, both because yod-dropping makes
sense as a change to shed anomalous onset clusters and because
arbitrary-looking cooccurrence restrictions are often just the
dead relics of long-past diachronic processes.

--And.