altamix wrote:
> I am afraid I need some suplimentary explanations here.
>
> once we have: *sw > *zw > *Dv > d, then you say "we expect the
> parallel development : *sw > *Tv >T
>
> paralell to what? Why should *sw > Tv here? And which should be
> the conditions which made this *sw > zw and the second time Tv ?
s : z = T : D
(the same place of articulation, contrast of voice)
1. PIE *s > PAlb. *z > gj before a stressed vowel
2. PIE *s > PAlb. *s > sh before an unstressed vowel
3. PIE *sw > PAlb. *zw > *Dv > d initially before a stressed vowel
4. PIE *sw > PAlb. *sw > *Tv > th (/T/) before an unstressed vowel
Changes 1., 2. and 3. are generally recognised by historical linguists;
change 4. is parallel to them in terms of phonological features.
The only point where the symmetry is not perfect is *Tv > T (not /t/),
but that's because Albanian has the now sporadic but originally
sandhi-conditioned variation of d ~ D in initial positions
(word-medially, however, old *D is reflected as -dh- /D/).
Piotr