>>balaur and mazãre
>>bollë and modhullë.
>
>vatrã (which reflects *wO:tra with a prothetic glide,
>presupposing earlier *O:tra, pace Alexander; the Tosk
>unrounding of /vo-/ is a late process).
Here, I'd ask myself how come that Romanian keeps /va-/ even in
those (quite vast) subdialectal areas where /vO-/, i.e. *['vOtr&]
would be possible/natural/common. But instead, as "pure" an [a]
as they come, everywhere (at least in the "Dacoromanian" dialect).
(Perhaps there was a rule for earlier stages of Romanian saying that
[O] can become [a]. I know of at least one example where this
worked, namely in a Hungarian loanword.)
>There remains the case of <cioarã> : <sorrë>, which either
>shows that the substitution was variable (BRom. *a or *o for
>PAlb. *O:), or perhaps owes its rounded vowel to the
>labialising influence of PAlb. *c^[W] < *c'W. Cf. Romanian
>-oara = Hung. -vár.
>
>Piotr
In the case of <cioara>, the diphtong [oa] (or [wa]) is there only
in the south-eastern subdialects of the DR. Especially in the north-
western "half" of the territory, people pronounce even today [O],
ie, [c^Or&]; the same people will diphtongate only when switching
to official/standard Romanian (which in this respect reflects
phonetic peculiarities of south-eastern provinces plus Bucharest).
In the non-diphtongating subdialectal areas, all words ending in
-oara are pronounced the same way [Or&], e.g. <oarã, doarã,
>moarã, comoarã, ogoarã [older+regional plural variant of
<ogoare>]> etc.
That is, there is no need for [oa] (or [wa]). It's a mere [O]
> [oa] evolution within (Northern) Romanian. Whereas in -oara
out of Hung. vár, [oa] isn't necessary even if the Hungarian
original 8-10 centuries ago was [wa:r]: just look at Rum. oraS
< Hung. vároS; besides, native-speakers of the relevant north-
western subdialects do not diphtongate [oa/wa] in TimiSoara or
SighiSoara either, they use instead [O], as in the other case I
mentioned in the previous paragraph: [timi'SOra, sigi'SOra].
(So, in case of <cioara>, methinks that [c^] + [o] are enough,
and in the case of vár, vároS, the rendering via [o] is enough,
i.e., you don't need any [w] for the transition.)
George