Crabby Jens:
> It can hardly be styled a problem or a false conclusion that the
> analysis produces the attested result. The preforms you specify
> yourself do not differ in any interesting way from my account. Both
> explanations say that a vowel which would have become /e/ if it had
> been accented changed into some other vowel which is later found
> with o-timbre. For reasons not interesting to me anymore you choose
> to speak of schwa and vowel shift, while I call them e and o, but
> the message remains the same.
Well, phonetically we must speak of [&] and its trivial lengthening
because that is the most optimal thing that could have produced both
mid-front *e and mid-back *o without violating universals on vowel
shifts or creating unheard-of processes that are assuredly exotic.
However I realize and now accept after pondering on the phonology at
that stage of Schwa Diffusion that what I call *& must be phonemically
(not phonetically however) understood as unaccented *a. So in a sense,
we can say that unaccented *a (still phonetically [&]) became *e
(simple fronting) before voiceless consonants, and *a (simple backing)
before voiced stops.
So everything is the same as I've stated already except that I now
think it's safe to say that the schwa is phonemically unaccented *a.
As such, regardless of accentuation on or off the vowel, there is always
a two-way contrast of *e and *a in all positions in the early Late IE
period. We have plural *-es (perhaps pronounced [*-Is]), but thematic
nominative *-a-s [-&-s]. We must make the distinction otherwise
grammatical chaos ensues.
(To give an example to make sure I'm understood, the thematic nouns
should then be understood as being [&] at this stage while written
*-a-. So a thematic noun like *génha-s would be for what I've
phonetically written out earlier as *génh&-s, later becoming *génhos
on schedule after Vowel Shift.)
There is no one-vowel system anywhere here. As I'm trying to get across,
monovocalism is an illusion that has more to do with how the two vowels
evolved, merged, split and remerged. It produces the effect which we see
but this doesn't mean that there was ever a one-vowel stage that you
could compare with the situation in Sanskrit. In Sanskrit, we see the
simple merger of *e, *a and *o to *a. We don't see any such simple
evolution in IE. Rather, what I see is something that give the illusion
of monovocalism. If what was once MIE *e becomes eLIE *a (after Syncope)
and then splits up again to *e and *a (during Schwa Diffusion), later
becoming *e and *o from Vowel Shift, that in itself will paint a
compelling mirage of monovocalism without there being such a thing
during any of these stages. In my account, the perceived monovocalism
is perfectly explainable without requiring actual monovocalism in the
theory. Very optimal this is, if I do say.
To this date, the only inefficiency you have exposed in my theory is
the Final Voicing problem while I've exposed numerous problems in your
account. In the end, optimality reigns supreme.
= gLeN