On Sat, 3 Nov 2001 23:38:01 +0100, "Piotr Gasiorowski"
<
gpiotr@...> wrote:
>>> _Final_ *-oi became *-i (via *-e.:) in the nom.pl. of thematic masculines.
>
>> Curiously enough, this is often ascribed to the effect of the _circumflex_ accent (which is rather unnatural for me if we accept at least the assumed original Baltic phonetical rendering of a diphthong's circumflex: it's the _first_ component which is acoustically more prominent, not the second one).
>
>... which simply means that there are still problems waiting to be solved.
Well, this particular one has been bothering me for a long time. Let's
go back to basics. In which cases is Slav *ai/*oi > i (instead of
regular e^) attested?
Bräuer's Slav. Sprachwissenschaft gives only three cases:
1. Dat. sg. u-stems synovi < su:nowai < su:newai. This is simply
wrong: the dat. sg. ending was of course *-ei.
2. Nom. masc. pl. o-stems -i < *-oi. A nuisance.
3. 2/3 sg. imper (< opt.) -i < *-oih1s, *-oih1t. Surely, the
laryngeal and the Szemerényi effect of -s would have conspired to make
these endings *-o:is and *-o:it, and then eventually Slav. *-y as in
the ins. pl. What we have here might thus be analogical after the
athematic optative in *-jeh1-/*-ih1- > -(j)i, or after the verbal
(V)je-stems (i-stems), where *-joih1- > *-je(:)i- > -(j)i.
This (unless I, or Bräuer, are missing something) leaves the o-stem
nom.pl. as the only clear instance of final *-oi > -i. It would be
very convenient to get rid of it. If we next turn to Baltic .... we
end up much more confused than we were:
Lith. Latv. OPr.
loc.sg. *-oi -è / -ie~ -a: -ai
nom.pl. *-oi -ai~ -i -ai
dat.sg. *-oi -(i)ai/-iui (-im) ?
dat.sg. *-o:i -ui (-am) -u (?)
ins.pl. *-o:is -ai~s (-ie~m) ?
That's not much help. Szemerényi in a footnote says something about
Eichner arguing for Lith. nom.pl. from -a:-i < n. pl (what's with the
-i?). That doesn't solve our Slavic problem either (Slavic kept the
neuter gender, and -a:i > -ai > -e^, not -i).
One solution is to assume a Slav. nom.pl. -oi-s, with raising caused
by -s [an -s that would be analogical after the -s's elsewhere in the
plural]. But this is not backed by Baltic, where -s would have been
preserved.
OK, then: influence from the jo-stems, where -joi > -jei > -(j)i is
regular.