kentum/satem built-in already in PIE?

I've propagated this idea before,
I've added a few details now.

Leaving the plain velars aside, which I think occur
only in loans, PIE has

palatal velars, /k^/ etc, supposedly
-> /s/ etc in satem languages,
-> /k/ etc in kentum languages

labiovelars, /kW/ etc, supposedly
-> /k/ etc in satem languages,
-> /kW/ etc in kentum languages


My idea was that the /k^/'s and /kW/'s had
allophones in PIE, one before front vowel,
another in other positions, and that the
first of these allophones was generalised
in satem languages, the other in kentum
languages.

Thus,
the supposed traditional /k^/'s had
the allophones, say, /c^/ and /k/, and
the supposed traditional /kW/'s had
the allophones, say, /k/ and /kW/.

Especially the thematic inflection, with its
alternating thematic vowel /e/ vs. /o/
would cause stems to alternate on the
final consonant.

I had the further idea some of that PIE
alternation might have survived in Slavic;
the traditional explanation is that it
was caused much later by subsequent
Slavic palatalizations.

Ex. of (P)PIE alternation *k/*kW:

1st sg. *pakWám
3rd sg. *pakWát

->

*päkWã´
*päkWä´t

->

*pekWõ´
*peke´t

-> (Russian)

*pekú
*pec^ët


As for the PIE *c^/*k alternation, that
worried me, but then I noticed something:

1st sg *pi:kam
3rd sg *pi:kat

->

*peikã
*peikät

->

*peikõ
*peic^et

->

*pis^u
*piset

which is not quite Russian, since the
3rd sg there is pis^et. But the "second
stem", that of inf. and pret. is pisa-
with an /s/, so I assume the /s^/ of
3rd sg etc is analogous?

Please object.


Torsten