> >Actually, you are postulating *-oi have yielded BSl. *-ai while *-oy
> >have yielded BSl. *-ei, aren't you?
>
> No, I can't do that: plural -i as well as locative -e^ have
> 2nd palatalization in Slavic, so it couldn't have been *-ei
> already in Balto-Slavic.
Of course. Sorry for a silly question. Just out of curiosity: if a
distinction between the reflexes of PIE *-oy and *-oi had been kept at
least up to the time of the second palatalization (which -- having been
failed in Krivichian -- seems to be a rather late phenomenon and can be
dated to the end of the 5th c. AD), what phonetic (segmental or
prosodic) contrast would you ascribe to these reflexes in Late Common
Slavic? The traditional explanation employs a prosodic (circumflex vs.
non-circumflex) contrast.
> I can't understand why Lith. has -ì [< *íe -- ST] in the adjectives
(except
> those in -iai), but -ai~ in the nouns (and -ie~ in the pronouns).
>
Just a little addendum to what Jens has already said: *ai > ie is normal
in East Baltic (though the exact conditions are still disputed); that
could facilitate the analogical interplay ai <--> ie. It seems to me
that what we have here is the contrast between nominal -ai~ and
pronominal-adjectival -íe (-ie~ of the forms like <s^itie~> being
introduced from monosyllables like <tie~> where circumflex is
phonetically regular).
Sergei