Re: Richard's Romanian sound changer

--- In cybalist@yahoogroups.com, Miguel Carrasquer wrote:

>> I assume that the format is:
>> {Latin word} > {Real Romanian word} not {Rom. word according to
>> sound changer}.
>
> More accurately {Latin or Vulgar Latin word} > {Real Romanian word

as given

> in Bourciez} not {Rom. word according to sound changer}.

Yes, that's what I meant.

>>> lendina > lindinã not lindenã
>>
>> PBR /e/ gradually closes to CR /i/ before /n/ (like in "bine",
>> "dinte", etc.).
>
> Also when unstressed?

Oops, I was hasty with the examples. Closure still holds in these
words
before /n/, even for unstressed vocalism, but is not strict:
fraxinus > frá(p)sen > frásin
grandinem > grindenã > grindinã
plantaginem > *pãtlágenã > pãtláginã
but
galbinus > gálben (> gálbin -- regional)
mastichinus > mesteácãn (also regional mesteácen)

>>> 2) v ~ b ~ 0
>>
>> Systematic loss of intervocalic Latin -b- or -v- is still to be
>> implemented; usual confusion between /b/ and /v/ cannot be summed
>> into a simple rule.
>
> As a first step, we should have -b- > -v- as everywhere in Romance.
> Special developments seem to be: -br- > -vr- and -rv- > -rb-.

The first step is of course for intervocalic -b-: it finally gets
lost before CR, just as intervocalic -v-, so there should be little
concern about an intermediate stage without final effect (attention
still high for -Vbu- > Vg(u): nebula > negurã, rubus > rug(u), favus >
fag [> faguri pl. > fagur(e) sg.]).
Out of this case, there is indeed -br- > -vr- and -rv- > -rb-, but
confusion does not stop here: for Romanian one has berbec(e) <
berbecem
(CL vervex) [one might argue that once v..rv > v..rb, assimilation
could be responsible for v..rb > b..rb, but...], bã$icã < vessica,
bãtrân < vet(e)ranus, zbura < ex-volare; also other Romance know
this phenomenon I was referring to.

>>> deus > zeu not ze
>>> reus > rãu not re
>>
>> For these ones maybe: final /u/ in hiatus > /w/ rather than 0.
>
> These actually belong with the cases of non-loss of -u after V.

That's what I meant by hiatus (in Latin) with preceeding vowel.

>>> auricla > urechie not aureache
>>
>> Reduction of Latin au > o (> u) occurs only if already VL,
>> otherwise it remains au; there is no simple rule to give that.
>
> That's solved by assuming *oricla as the original form. The problem

is

> with the stressed vocalism: I suspect -kja > -kje first, and then

-ea- >

> -e-. Rule ordering.

Attested anciend DR are "urekie", "ureike", "urekle", "urjacle/i" (
"ja"
denotes corresponding Cyrillic grapheme): no final "a" in DR
(conserved
in AR) but some remainder of diphthong confirming your assumption.

>>> conventum > cuvînt not cunvînt
>>
>> Latin "conC" > Rom. "cuC" with regularity.
>
> There's already a rule ns > s.

That could account for C = s, but the above statement is valid
regardless
of what is C: *contremulare > cutremura, contorpire > cutropi (>
cotropi),
conpre(he)ndere > cuprinde, etc.

>>> mergere > merge not mierge [?]
>>
>> "mierge" is still regional.
>>
>>> mer(u)la > mierlã not miarlã
>>
>> "miarlã" is regional.
>
> Any reason for the StdR forms?

Well, "mierge" seems to be rather a spontaneous diphthongation not
correlated with vocalic quantity in Latin, so I would propose PBR /e/
instead of /E/ (as suggested also by spontaneous regional
dipthongation
"mearge"), with regular exitus. For the other one, the form "mierlã"
is
clearly dominant, I could think at some articulatory reasons favoring
"ie" instead of "ia" but I won't bet my income on that. :-)

Regards,
Marius Iacomi