> As for the raising of unstressed vowels, the process is quite
common, not to say trivial. Many dialects of English have
unstressed /i/ derived diachronically from a variety of sources, as
in the second syllable of naked, palace, usage. Latin has a well-
known rule raising historically unstressed a's to i's, as in
cano~cecini or amicus~inimicus. The vowel in question may well be
raised along the central track, becoming first high, and only then
fronted; it need not be front at the outset.
>
> Best regards,
> Piotr
Cf. the same development in Old Prussian (at least as far as we can
see from XVI c. Sembian dialects of Catechisms): in unstressed
endings *-as > *-@... > *-Is, often rendered as -is in the Catechisms
text (in nomina clearly demostrating a-stem in oblique cases). Taking
into account the fact that OPruss and (Proto-)Slavic share some
common phonetic developements, may be this could be an explanation of
Slavic -U < *-os in masculina, where the ending was often unstressed:
*-os > *-@... > *-@ > *-U, where the drop of *-s prevented the last
stage (-Is).
Sergei