On Sat, 24 May 2003, Miguel Carrasquer wrote:
> [...]
> As far as I know, in the whole of Balto-Slavic the original
> o-stem/pronominal PIE genitive (*-osyo; *esyo, *tosyo) only survived
> in the single Slavic form *kWesyo > OCS c^eso "of what?". (Modern
> Slavic lgs. have generally normalized this to c^ogo or c^ego, but
> Slovenian still has c^ésa).
It is also the form of the genitive of pronouns in Old Prussian: stas
'that, the' has gen.sg. stessei or stesse (also steise and once steises),
ains 'one' has gen.sg. ainassei or ainessa. Trautmann explains the form
from *-e/osyo with or without added deictic -i. This supports the
impression that the BSl. gen.sg. of pronouns is from the IE genitive, not
from the ablative. This speaks against an attempt to unite Slavic -ogo
with -a and Lith. -o (from circumflex long a) and rather favors a
derivation from the IE genitive which is well known. The stem of 'what' in
Slavic reflects *ki- and *ke-, while that of 'who' reflects *ku- and *ko-
(all earlier labiovelar of course). Therefore OCS kogo, which is directly
opposed to a neuter c^eso, would be expected so somehow reflect *kWoso as
opposed to *kWeso (taking these to be reduced variants of *kWosyo/
*kWesyo). Now, c^eso is phonetically regular, but kogo looks like nothing
we know. I have ventured the guess that it is analogical on c^eso which
was perceived of having two related consonants, the second of which was a
reduced form of the initial; therefore, if the second consonant of *koso
is replaced by a weak form of k-, the result cannot be very different from
kogo. So I suggest that that was its story, and that the other pronouns
like togo, inogo, jego followed suit by a second analogy. It should
perhsps be pointed out that in the adjective forms ending in -ogo, -ego,
-ago are from the *definite* adjective and so reflect forms like OCS
nova-jego with old ablative (-a) and genitive (-go) superimposed. This
further restricts the Slavic g-forms to the grammatical sector of pronouns
where OPruss. indicates that the old consonant was -s-. I know too little
about the details of the reduction the velar to Russ. [-v-] to express an
opinion about it. I find it a little strange that, while unaccented -ajego
ended up [-&v&], the accented variant is [-ov&] although there was never
an /o/ in the form. Surely we cannot reanalyze PIE on such a basis,
especially if we are not sure we are talking about the correct item.
Jens