alexandru_mg3 wrote:
> As you know Piotr, the Whole Reconstruction of PAlb c^, c, dz,
> g^ are mainly based on the Attested Romanian dz, c and c^ still
> present 'in the PAlb/Dacian? loans of the Balkan Latin' and not Only
> on the Albanian dh, th, s, z ...so please do not open this subject
> (when even you, didn't trust in it)....because this will affect the
> WHOLE MODEL only because you cannot justify the preservation of
> intervocalic dh based on your own local rule...
The fact is that the preform of <dh> was never lost if it represented an
older afficate (*3) rather than PAlb. *d, so that there may have been a
contrast between the product of the lenition of *d and the reflex of the
affricate. Romanian evidence is nice indeed. It confirms the
independently established reconstruction of PAlb. as having two types of
affricates, one from the old *K^ series, the other from palatalised *KW
(later merging with a few other sources of Mod.Alb. /s, z/; but it
cannot be used as the cornerstone of PAlb. reconstruction, which must be
done primarily on the basis of inner Albanian evidence.
How did *3(') change into /ð/, avoiding merger with /z/ from *3^(W)? If
the latter passed through a stage resembling *[3], old *3 must have
changed into something different by that time, and that "something" must
have been both sufficiently contrastive and relatively similar to /ð/.
Hamp reconstructs a non-strident affricate like *[dð], and its voiceless
counterpart *[tþ], even for the earliest stages of Albanian; these look
like plausible intermediate steps between *c('), 3(') and modern <th,
dh>. Proto-Romanian *3 (modern /z/ ~ /3/) may very well reflect this
stage. If the lenition of *d yielded [ð], its similarity to [dð] would
have prompted the same kind of phonemic substituion in Romanian.
Piotr