--- In
cybalist@yahoogroups.com, enlil@... wrote:
> [...] Now, since part of the
> issue here involves taking abstraction way too far, I will
indirectly
> quote the sensible Alexis Ramer's own words via Marc Hamann's
account
> of them whilst pretending it was my own amidst the confusion :P
>
> "'(a) the so-called monovocalic hypothesis treats */e/ and */o/ as
> variants of the same MORPHOphoneme (i.e., the same UNDERLYING
> segment), not the same PHONEME, in PIE, and to count these
two
> as one phoneme would be like counting all the vowels of
Hebrew
> or Arabic as one phoneme, [...]'"
>
> Precisely. Or English for that matter. So maybe /u/ in 'run'
and /a/
> in 'ran' are the same vowel! Of course, it's completely absurd.
There
> has to be a limit but some people feel less bound by logical
constraints
> than they should be by the constraints of a tight-fitting
straightjacket.
You still appear to be missing the point that we were in fact
talking morphophonemics. And in morphophonemic analysis there is a
scale of options from the most timid to the most intrepid. Somewhere
along the line lies the observation that, in IE *roots*, there is a
tremendous preponderance of the vocalism /e/ which must be taken
stock of in any account of the grammar.
This certainly points in the direction of a monovocalic system,
without *completely* getting there because there are fringes on the
carpet we cannot explain. The easy explanation is that the fringes
are post-monovocalism innovations, and the only thing dissuading
that is the parti pris that such an option should not be used,
however open it may lie.
I have shown that such an analysis is quite viable for Sanskrit,
even in traditional phonemic terms. You don't need extra vowels any
more in Sanskrit than you need thorn in German or dorsal spirants in
English. Duden's German grammar posits a dental spirant for final of
the German word commonwealth, and a thousand pronunciation guides
say that [x] is the sound of English loch. In like fashion one must
acknowledge Sanskrit /o/ in opposition to /av/ on the basis of rare
pairs like toyam : avyas, and /e/ : /ay/ for s'eya- adj. 'to be lain
on' : s'ayya:- f 'couch' (probably originally the same word-stem).
For way over 99 percent of the material the one-vowel analysis is
impeccable. I posted an example of 113 words of running text without
a single exception. It was only my laziness that prevented me from
going on.
If I were to do the same with Old Persian I wouldn't even know where
to look for deviations. There are no e- and o-vowels in OP which has
retained Indo-Iranian ai, au, a:i, a:u. There are no environments
opposing /y/ and /i/, or /v/ and /u/, or /r/ and /r./ to each other.
Thus as far as Old Persian is known it fully qualifies as a one-
vowel language.
My analysis was anticipated, without my knowledge, by Ivanov who, in
his grammatcal sketch "Sanskrit" (in Russian 960, English transl.
1968) wrote:
"There are sufficient reasons to postulate only one vowel pphoneme
(in the proper sense of the word), _a_, for the earlier Indo-Aryan
period (before the change of the diphthongs _ai_ and _au_)" (p. 46
of the Eng. version).
Ivanov does not state the reasons, and the importance ascribed to
the diphthongs is of course wrong. As Peter has now convinced you,
there is no opposition between [e] and [ay] the only thing that held
Ivanov from assigning a one-vowel system to Sanskrit itself. If we
do not have German /T/ (thorn) or English /x/, one can take that
step in good faith.
Like you said, if I understood you, that has no immediate
repercussions for the prehistory of either Sanskrit or PIE. It does,
however, affect the situation indirectly, since it extinguishes the
red light. Now it is, again, a matter of objective inspection and
analysis whether any of these stages is to be reconstructed without
phonemic vowel oppositions.
Jens