Jens:
> It *may* suffice to assume /s/ -> [z] __# , provided something else
> is assumed for s/t -> [s] __#, as in the 2sg in *-e-s. However,
> nothing proves that this is the (or, a) correct solution. In case it
> is not, and *-o-s vs. *-e-s does reflect two earlier discrete
> phonemes, the "superior stance" leads to "[n]othing but" loss.
Wrong. We don't know whether the *o in the nominative singular thematic
*-o-s and the *e in plural *-es derive from the same vowel. Common
sense would have us begin by presuming what we see: two _different_
vowels. We also begin with only *s (no **z)... again, that which we
see.
It seems that the oscillation of *o with *e in the thematic paradigm
would appear to guarantee that we're dealing with an originally
single vowel that was transformed before voiced *s. This conclusion
merely necessitates an allophone [z] of *s which already exists in
*nisdos.
To me, this is a seperate vowel *&, the process was lengthening as we
find in a number of languages, and only this vowel caused the
e/o-effect. Simplicity is key.
That's how I can explain *-es without **z since it merely derives
from earlier *-es [-ez] while *-o-s is from earlier *-&-s [-&-z].
(Further comparisons outside of IE confirm the *e since it reflects
Proto-Steppe *i. Thus Steppe plural *-it > Tyr *-er; IE -es; Uralic
*-(i)t; Altaic *ir^. Now, the ENTIRE array of facts are accounted
for.)
So your statement, "However, nothing proves that this is the (or, a)
correct solution. In case it is not, [...]," is misleading. My
above account cannot be disproven, nor is it unlikely, and yet it
contradicts the need for **z while strengthening IE's connection
to external language groups. Your theory falls flat and doesn't
appear to have any purpose, the least of which is an adequate
explanation of all the facts.
While I don't see anything disproving either of our viewpoints on this
so far, it still remains that **z is an added indulgeance that is not
shown to be necessary in PreIE.
You continue to insist to us that it is necessary. Yet, the unanswered
question persists: "Why?" Why must **z be anything other than an
allophone of *s.
= gLeN