[con-romance] Rumiyaan vocalism

  I "felt" that Rumiyaan should have Romanian-like umlaut -- in Romanian a
following *a causes *E and *O to break to ea oa -- similar to what happens
with *E > ie; *O > uo,ue in western Rmc, except the conditioning factors
are different.  Before I was quite clear about the vowel developments
between Latin and Rumiyaan (well I still am not *quite* clear, but I'm
clearer) I imagined a universal breaking of *E and *O to yæ and wæ, with a
lot of palatalizations and w-clusters that didn't fit the taste of the lang
as a result.  Eventually I abandoned the idea of breaking completely, but
after grasping the main developments:

   *a  *E  *e  *I  *i  *u  *U  *o  *O
    |   |   |   |   |   |   |   |   |
    æ   e   i   e   i   u   o   a   o

i.e. that the tense/lax distinction is all-important over any height or old
length distinctions, it was also clear to me that one of the exceptional
rules (Sonderregeln) is that when an *a followed *E or *O these became /æ/
but *I and *U retained the reflexes /e/ and /o/:
    *mola 'grind-stone' > _mæl_
    *korda 'cord' > _kærd_
    *oktauu '8th' > _æghdú_
    -- *au monophthongizes after the *o > /a:/ > /a/ shift and then to /u(:)/!
        Probly *au > [o:] after *o had lowered-unrounded, and then merged
with /u/ at the same time as /e:/ < *e merged with /i(:)/.  In the 18th
century Lemontaigner still distinguished "ê" and "ô" from "î" and "oû" in
one of his two informants.
    *uolare 'to fly' > _bælá_
    *bukka 'mouth' > _bok_
    *fundare 'to found' > _fondá_
    -- note that *-are in infinitives > /a/ rather than /æ/!
    *petra 'stone' > _pæhr_
    *estàte 'summer' > _æstæd
    *pira 'pear' > _per_
    *lingua 'tongue' > _leng_ -- not used for 'language' in Rumiyaan!
Where internal development looks very smooth, but external development was
tortuous!

/BP