> That is, there is no need for [oa] (or [wa]). It's a mere [O]
> [oa] evolution within (Northern) Romanian.
You confound (I mean related to their inner timeframes) 2 distinct
phenomena in Romanian that took place in different moments in time.
1. One is wa > wã > wo > o
Lat. una (Aromanian unã) > uã > uo > o (fem. 'a')
with attested:
nuor (attested sec XVI) > nor 'cloud'
uo (attested sec XVI) > o fem. 'a'
So we need a 'wa' to explain Romanian 'orash' and Magh. varos.
We also need a 'wa' to explain Romanian 'oare' and Alb.
vallë 'perhaps?maybe?" < PAlb. *wala
(same situation with Rom. 'hora' Alb. 'valle' 'circular dance')
2. The second one is the diphtongation of o>oa that is a later
process in Romanian (affecting first wave of Slavic loans too so
around sec X BC)
See: Rom. coasa - Sl. kosa
When the dipftongation o>oa was started in Romanian the output of
*wa was already *wo (uo in Romanian)
If we take the example for 'oare' 'maybe?perhaps?' we have:
PIE wel-2 'to wish' > PAlb *wala (before Latin contacts)> PRom
*wãre (after Latin Contacts) > ORom *wore (before Slavic Loans) >
Rom. woare (diphtongation) (even today the real pronunciation
is /woare/ not /oare/(that sound very artificial for a Romanian
speaker) with a retraction of diphtongation (or a missing of it) in
Transylvania /wore/ )
Especially Lat. 'una' clearly shows us that the output of *wa is *o
in Romanian (at least in an initial position).
Only the Bests,
Marius