On 2007-06-26 21:53, stlatos wrote:
> These and similar words show both *gYhel-xW-t.o+ and *gYhl,xW-t.o+
> so I don't think starting with 0-grade explains this change.
>
> I'd say that when x() > -syllabic between C the xW,>xW>W before t.
> in Slavic and n.(y) in Baltic. The remaining +round moved to l, then
> el-W > elW- > olW-.
Still it's curious that we have East Baltic *varda- vs. OPr. wirda-
(without a BSl. laryngeal), whereas Germanic has a nil grade, as in the
'gold' word. I suspect some kind of vr.ddhi transformation of an
inherited nil grade within Balto-Slavic. Cf. Lith. z^énklas 'sign' for
*g^n(o)h3-tlo- (Skt. jn~á:tra- 'intellectual faculty'), with the vowel
inserted in the wrong slot.
In some cases, as Jens has suggested, the full/nil variation may reflect
the old difference between a plain noun and its collective, as in
*bHérh1g^-o-s (Lith. bérz^as 'birch') and *bHr.h1g^-áh2 (Lith. bi`rz^e.
'birch grove'), with the full grade generalised in Slavic. To be sure,
*bHérh1g^-o- itself looks like a vr.ddhi formation (derived from
*bHr.h1g^-(r)ó- or the like), but it's an old one, cf. the full grade in
Gmc. *Birkjo:n-.
Piotr