1) The PIE o-stem abl. ag. was still uncontracted *-ood,
which gives PBS *-aa(d) ~ *-ã:(d), and further Slavic -a,
Lith. -o~ (instead of +-uo~).
The result of thematic vowel + /e/, on the other hand, was
already (super)long /õ:/, as in the dat.sg. *-o-ei > *-õ:i >
Lith. -ui~, Slav. *-u:i > *-ô > -u.
If this is correct, then the gen.pl. cannot have been simply
PIE *-o-om, because that would have given PBS *-aam ~ *-ã:m
and further Slavic -U (correct) and, if I'm not mistaken,
Lith +-aN~ (in actual fact -uN~ < *-uoN~). Is that correct?
A reconstruction *-o-em for the gen.pl. is unattractive.
The best solution (which I've borrowed from Jens), would
seem to be *-oy-m, with the thematic plural oblique suffix
*-oy- plus the gen.pl. ending in the zero grade as *-m.
This would have been contracted at an early stage already
(before syllabic resonants) to *-õ:m. The same solution
then applies to the acc.pl. *-oy-ms > *-o:ms (not *-õ:ms,
presumably because of the -ms# cluster).
2) The Latin imperfect in -bam, -ba:s, -bat, etc. looks like
an originally periphrastic construction with as final
element the auxiliary *bhuah2- in a past tense (cf. eram,
era:s, erat from the *es-auxiliary). If this is so, then
what comes before the auxiliary must be some kind of
participle, verbal noun/adjective, infinitive or something
like that. Normal thematic stems and i:-stems, mostly,
indeed show an element -e:- between the root and the
auxiliary. This is absent in a:- and e:-stems, in Old Latin
sometimes also in i:-stems, but this can probably be
explained phonetically (a:e: > a:, e:e: > e:, i:e: > i:).
Since we would _not_ expect the mere root here, this is
good. However, there is no evidence that I know of for a
participle, verbal noun/adjective or infinitive in -e:
anywhere in Indo-European, except in the same indirect way
as in Latin, namely the Slavic imperfect in -êax- (<
*-e:-e:s-?). What can be the origin of this deverbal
derivative in -e:, and what happened to it elsewhere? Any
ideas?
=======================
Miguel Carrasquer Vidal
mcv@...