Re: Voiced Velar Fricative

John Cowan:
> And Rosta wrote:
>
> > Cool fact. How trustworthy/well-documented is it?
>
> It's in *The World's Major Languages_, which is generally reckoned reliable.
>
> > It's as though final
> > /h/ (from /s/) transmogrifies into an autosegment that triggers ATR-ish
> > harmony (with stressed vowels immune?), which suggests that H-ness
> > and ATR-ness are phonologically the same thing.
>
> I think the historical sequence is that the fronting was associated with
> /h/ in the coda, and when /h/ went away, the fronting remained.
Okay; but still an explanation is invited for why final /h/ triggers some
sort of ATR-like effect on vowels. My very limited readings in phonology
have not encountered anything that links /h/ and ATR-ness explicitly,
though I'm hoping that Dirk will entertain us by disserting on this
issue in due course.

> Why "stressed vowels immune?"  The "s" in "casa" survives because it is in
> the onset, not because the preceding vowel is stressed, IMHO.
Because your original message said:

> 'la casa' /lakasa/ 'las casas' /l&kas&/.
not /l&k&s&/. But I realize now that I apprehended the data too hastily and
thought there was leftwards harmony, but in fact the /h/ has affected only
the vowel in the syllable it is coda to. So it's not *quite* as cool as I
thought, but still interesting.

--And.