On Tue, 20 Apr 2004 00:54:39 +0200, Mate Kapovic
<
mkapovic@...> wrote:
>> The narrowing of -e:(r) to -i: is similar to the narrowing
>> of -o:(n) to -u:. The forms without -n, -r, if I understand
>> Jens correctly, have circumflex accentuation, i.e. they're
>> superlong (-o:n vs. -o::, -e:r vs. -e::), so perhaps the
>> narrowed Slavic reflexes -u: > -y and -i: > -i do in fact
>> both point to sandhi-variants without final sonorant. I
>> don't see how you could get -e:r > -i otherwise.
>
>Why not? We have no other examples, why is *-e:- > *-i:- in front of *-r# in
>last syllable unacceptable?
According to general phonetic principles, /r/ normally
doesn't have a closing effect on a preceding vowel, on the
contrary, it usually has an opening effect. This is quite
unlike /n/ (or /m/), which _do_ have a closing effect. If
the underlying ending had been -e:r, and assuming -r is
lost, I would expect the outcome -e^, which is normal for
word-final -e: (e.g. 1du. p.p. ve^ < *we:).
So there has to be another way to explain the -i: of
*ma:ti:, *dUkti:, and I see no other option than superlong
-e::, the sandhi-form with (PIE) loss of -r. This can then
also be used to explain -u: in kamy, although there we can
also have the alternative explanation of narrowing of /o:/
to /u:/ caused by final -n.
=======================
Miguel Carrasquer Vidal
mcv@...