Re: U-umlaut & schwa shifts

> > *The u-umlaut seen in the development of Germanic languages is
>essentially
> > labialization (/i/ -> /y/ etc); is velarization-u-umlaut - where /i/
>goes to
> > /i\/ or /M/ instead, etc - attested anywhere? If yes, is raising (eg. /O
>o/
> > -> /o u/) typically involved too?
>
>I'm of the general impression that frontness and labialisation are
>contagious, but backness isn't.
Really? What are the typical ways of deriving back unrounded vowels, then?
Unrounding somehow seems even less plausible to me. There's always cluster
reduction, but to me it seems like it would have small yield unless I reduce
stuff to /j w/ first.


>On the other hand, you could have a
>back vowel causing a /k/ to become [q], which then causes front vowels
>to be backed, all of which are attested iirc.
...and then re-collapse the [q]... Nah, sounds needlessly convoluted.


> >*Does anyone have any creativ ideas for sound shifts starting from >/@/?
>Mine are epenthetic, so deletion would be pointless. >Assimilation with the
>following or preceding vowel comes first to >my mind, but seems a little
>too obvious. I can also think of >raising it to /i\/, but is there any
>precedent for that? Any other >ideas are welcome too.
(Shifting it to [i\])
>seems to me to be less creative than assimilating to the
>surrounding vowels. If you're thinking of doing a
>velarising-u-harmony-type-thing, I vote you assimilate /@/ to
>surrounding vowels and get a nice vowel harmony system happening.
>Throw in /k/ -> [q] in back words and /k/ -> [c] in front words too
>and make it a little fun :)
>
>--
>Tristan.
That would work, except these are two different projects.


>How about
>Lenition the preceeding consonant, and then assimilation of the schwa to
>some feature of the following consonant (eg [@] > [U] before labials, [a]
>before glottals, [E] before palatals)?
>
>Pete
Not bad at all. A source of [U] would come handy.


>Proto-Austronesian is reconstructed with *i,a,u, and */@/ which does lots
>of
>odd things; it must have been a very unstable sound.
>3- In various subgroups, it can > a, e, i, o, or u; even in rather closely
>related langs. it can differ, e.g. Tagalog has /i/, Bisayan /u/;
>Proto-Oceanic has *o.
>
>4- In a few languages, it's retained: in Javanese, in both syl.; in Ml.
>only
>in penult (unstressable), but > /a/ in ult.  Some languages have penult
>/o,e/ vs. ult. /a/. Where retained, it can be [i\] or [@] or even [V].
That's good to know.


>5- Another hint at its oddness: in widely separated groups, where retained,
>it affects a following medial *C-- usually with gemination or related
>phenomena like preglottalization. (Usually in these cases it _can_ carry
>the
>stress.)
Compensatory consonant lengthening? That sounds worth considering too, but
it probably would work better with schwas deriving from unstressed vowels.

---

Thanks for the replies, anyway. I think I'll go with Pete's suggestion
mainly, but I could throw also in some assimilation ... maybe antiharmonic
;)

John Vertical