alexandru_mg3 wrote:
> Note: I agree that there weren't any: *dek^s(i)no- and *dek^s(i)tero-
> , I was the first to say (I mean in this discussion) that we have
> here a 'simple' *dek^s-)
... which is unattested in IE despite the wide distribution of this
word-family. Then, if you claim that the reflex of *k^s was still a
bisegmental sequence like *-cs- in Proto-Albanian, I should challenge
you to show at least one certain example of *e > *ja before *-k^s-
followed by a vowel. It seems that proto-Albanian had generally
eliminated obstruent clusters other than *st and *sp (whose special
status is cross-linguistic), so I'd be most surprised to see -cs- there.
Piotr