--- In
cybalist@yahoogroups.com, "Richard Wordingham" wrote:
>>> The vowel of the first syllable. If we counterfactually ignore
>>> the fall of /v/, Latin _gingí:va_ should give Romanian _*gengívä_.
>>
>> If first /i/ was short, maybe;
>
> Italian _gengi:va_
Well, Italian word should be written rather "gengiva" since there is
no phonological distinction of quantity in modern language (but this
is just a minor detail).
> and Western Romance *ginci:va (> French _gencive_, Spanich _encía_)
> argue for the first /i/ being short.
Still don't forget that we deal with is unstressed vocalism and the
confusion between /i/ and /e/ (long or short) appears (as irregular)
in inscriptions from the 3rd century a.D. on. So one can expect a few
words not fitting canonical rules.
>>> Latin _*gíngi:va_ (not a possible stress until stress became
>>> lexical) would have given Romanian _*gíngivä_.
>>
>> Latin stress doesn't migrate so easily on the way to Romanian.
>> Only some particular classes of words make it change. OTOH, with
>> your same arguments, why didn't you reconstructed from the second
>> a "Romanian" *gíngevä?! :-)
>
> I had grave doubts about *gíngevä > _gíngie_. (Smiley ignored.)
Besides the well-known fall of intervocalic /v/ in Romanian which was
consciently neglected in this analysis, there is another issue needing
some further elaboration. You proposed a Latin reconstructed *gíngi:
va.
This word doesn't fit Classical Latin (stress rules forbid it). In
Late
Latin the stress could technically be possible because of quantity
loss;
ipso facto, the word shouldn't be written with /i:/. But in that case,
the first /i/ should have evolved already in (closed) /e/. Otherwise:
the quantity loss (and consequently /i/ > /e/ in the first syllable)
must precede any hypothetical stress shift. So this possibility cannot
realistically give "*gíngivä" but "*géngivã".
Another one would be to infer that stress shift has preceded Latin
quantity loss; in that case, for some reason the second /i:/ _must_
have turned short /i/; therefore its natural exitus would be /e/ for
the very same arguments as for the first /i/ above. Hence, you should
have reconstructed some something like "*gíngevã" (that's why I wrote
this form in the text with smiley above)
I fail to see how can one generate a reconstructed "*gíngivä" from
the
given starting point "gingi:va".
Regards,
Marius Iacomi