alexandru_mg3 wrote:
> In conclusion Piotr: So many 'analogical restaurations or
> retentions' that you can see above could sound finally for you 'like
> a rule'...?
You are just recycling your original argument ad nauseam without
offering any new evidence. No matter how many such verbs you quote, it's
simply more of the same thing: the restoration took place _generally_ at
the end of verb stems, where intervocalic stops alternated with final
ones. If there was no such alternation, as in nouns like <thëri> (Geg
<thëni:>) 'nit' < *k^oníd-ah2 or <be:> 'vow' < *bHoidHah2, the stop,
never exposed, could not be restored. As for the final *-d(H) being
reflected as <dh>, it shows an Old Albanian final sandhi rule: *d was
lenited in this position. The same happened word-initially, though here
the lenition is variable, and it seems that initial *d- originally
changed to /ð/ only when it followed a vowel or *r at the end of the
preceding word or prefix.
The loss of intervocalic *-d- vas probably via *-ð-, though we have no
direct evidence of the intermadiate stage, which must have been rather
short-lived. It seems to be the most natural path of development and is
consistent with the observation that *-d(H)- often ends up as Mod.Alb.
<dh> in non-intervocalic positions.
> And the 'stupid idea' that Alb va is inherited could appear now as a
> reality....?
I didn't say the idea was stupid. I just don't believe the final *d
would have been lost from pre-Alb. *wad-(os), which is why I much prefer
the loanword solution (also admitted by Demiraj -- note his uncertainty
about the PIE reconstruction).
Piotr