On Mon, 30 Sep 2002 01:27:34 +0200 (MET DST), Jens Elmegaard Rasmussen
<jer@...> wrote:

>The Greek -oí and the Lith. -ì, -íe-ji are relevant in showing that plain
>*-o- + the consonant /y/ yielded acute tone (or, had short prosody) in
>word-final in the source of both languages (which can hardly be anything
>other than PIE). They also show that the concatenation of *-o- and the
>deictic particle /i/ of the locative yielded Gk. Isthmoi~, Lith. namie~
>with a different (circumflex, longer, or disyllabic) prosody from the one
>of the nom.pl.

Agreed. If we look at the Slavic reflexes of *oi/*ai in the Auslaut, we have:

circumflex (vowel + vowel):
pf.1.sg. *woid-h2a-i -> ve^de^
loc.sg. *-o-i -> -e^

acute (vowel + consonant [cluster]):
nom.pl. *-o-y -> -i
imper. *-o-yh1-s/t -> -i

It is not necessary to assume an acute / circumflex tonal contrast already in
PIE. It suffices to have a contrast between *oi and *oy.

>I used to regard the Lith. subst. nom.pl -ai~ as due to
>secondary reintroduction of the stem vowel -a-, giving -a- + -ì. Since
>then I have been persuaded by Kortlandt's explanation as the proper ending
>of "soft" adjectives, cf. medìnis 'wooden', nom.pl medìniai, where it
>fails to attract the ictus and so must be (or, have become) underlyingly
>circumflex. The whole business shows the delicacy of Auslautgesetze.

A thought. Is the non-acuteness of nom.pl. -(i)ai analogical after the
non-acuteness of the feminine in -e: (nom.sg. medìne:, nom.pl. medìne:s)? I
mean, if the hard adjectives show a contrastive pattern:

nom. sg. nom. pl.
masc. -ACUTE +ACUTE ge~ras gerì
fem. +ACUTE -ACUTE gerà ge~ros

Then the soft adjectives, by contrast, take on a non-contrastive pattern:

nom. sg. nom. pl.

with the old situation preserved in the -is/e: class (e.g. dìdelis/dìdele:, pl.
didelì, dìdele:s).

Miguel Carrasquer Vidal