Richard:
> Why? The initial cluster tkn- exists in either Russian or Polish.
Yes, that's great. And we know how twisted Polish is ;) but the thing
is that even though we might say this "cluster" exists in some languages,
to call it a cluster to me seems a bit misleading phonetically. I
honestly don't know how one's tongue can contort in such a way as to
pronounce tkn- without a vowel somewhere in it. Again, I'm speaking on a
more subphonemic level. It might be analysed as a cluster but I don't
think it truely is.
Actually, thinking more on Jen's version of EA, I think I have an idea.
We can think of a schwa in EA as the "non-vowel", his subphonemic vowel.
It is placed in syllables which otherwise would be zero and awkward. So,
in the case of a root *nat- and a suffix *-R, it can only become *nat&R
because **natR would violate the syllabics. (Yes, I'm aware of French
"quatre" but I'm not speaking of French so go with me for a sec on this).
So, now, let's say that you want to add another "consonant-only" suffix
like *-m. We would get *natR&m because the *& is only a go-between vowel
to even out the syllables in the otherwise blech-looking **natRm (that
would never have existed in any prestage of EA). Here *& goes between *R
and *m because this evens out the syllables optimally to produce C&C
in the second syllable.
However, with *alu- + *-R, we get *aluR, and with *alu- + *-R + *-m,
we'd get *aluRm. Why? Because a syllable with a real vowel *u _can_
tolerate a CVCC syllable. I think I get it.
Would that now make sense, Jens? In this way, we don't need a ridiculously
vowelless prestage and we get to keep my ideas of Proto-Steppe intact.
Between Proto-Steppe and EA, there must have been a reduction of full
vowels to schwa or null in these suffixes.
= gLeN