> The actual developments can be given as follows:
>
> PIE PS
> *ka/*ko *ka > ko
> *ke *ke > c^e 1st.pal.
> *ki *ki > c^I 1st.pal.
> *ku *ku > kU
>
> *ka:/*ko: *ka: > ka
> *ke: *ke: > c^a 1st.pal., ê > a
> *ki: *ki: > c^i 1st.pal.
> *ku: *ku: > ky
> *kai/*koi *kai > cê 2nd.pal.
> *kau/*kou *kau > ku
> *kei *kei > c^i 1st.pal.
> *keu *kjau > c^u j-pal. (au=au)
>
> *kja/o *kja > c^e j-pal., a > e
> *kju *kju > c^I j-pal., u > i > I
> *kja:/o: *kja: > c^a j-pal., a: > e: > ê > a
> *kju: *kju: > c^i j-pal., u: > i: > i
> *kjai/oi *kjai > c^i j-pal., ai > ei > i: > i
> *kjau/ou *kjau > c^u j-pal., au=au (and aN=aN).
I assume that by PIE you meant here Proto-Satem or the like (since
obviously PIE *kW > 'Proto-Satem' *k has taken place)
It could accounted for with these rules (I've added an intermediate
stage pre-ProtoSlavic out of convenvenience of rule ordering)
PSatem > p-PS
*i > *ji
*e > *je
*i: > *ji:
*e: > *je:
p-Ps > PS
*kj > *k^
PS >
*k^V > *c^V
*ke > *ce
PSatem p-PS PS
*ka > *ka > *ka > ko
*ko > *ka > *ka > ko
*ke > *kje > *k^e > c^e 1st.pal.
*ki > *kji > *k^i > c^I 1st.pal.
*ku > *ku > *ku > kU
*ka: > *ka: > *ka: > ka
*ko: > *ka: > *ka: > ka
*ke: > *kje: > *k^e: > c^a 1st.pal., ê > a
*ki: > *kji: > *k^i: > c^i 1st.pal.
*ku: > *ku: > *ku: > ky
*kai > *kai > *ke > ce 2nd.pal.
*koi > *kai > *ke > ce 2nd.pal.
*kau > *kau > *kau > ku
*kou > *kau > *kau > ku
*kei > *kjei > *k^ei > c^i 1st.pal.
*keu > *kjeu > *k^au > c^u j-pal. (au=au)
*kja > *kja > *k^a > c^e j-pal., a > e
*kjo > *kja > *k^a > c^e j-pal., a > e
*kju > *kju > *k^u > c^I j-pal., u > i > I
*kja: > *kja: > *k^a: > c^a j-pal., a: > e: > ê > a
*kjo: > *kja: > *k^a: > c^a j-pal., a: > e: > ê > a
*kju: > *kju: > *k^u: > c^i j-pal., u: > i: > i
*kjai > *kjai > *k^ai > c^i j-pal., ai > ei > i: > i
*kjoi > *kjai > *k^ai > c^i j-pal., ai > ei > i: > i
*kjau > *kjau > *k^au > c^u j-pal., au=au (and aN=aN)
*kjou > *kjau > *k^au > c^u j-pal., au=au (and aN=aN)
alternatively, one could skip the kj-stage and let the front vowels
and j palatalise the k's directly:
PSatem > PS
*ki > *k^i
*ke > *k^e
*ki: > *k^i:
*ke: > *k^e:
*kj > *k^
PS >
*k^V > *c^V
*ke > *ce
PSatem PS
*ka > *ka > ko
*ko > *ka > ko
*ke > *k^e > c^e
*ki > *k^i > c^I
*ku > *ku > kU
*ka: > *ka: > ka
*ko: > *ka: > ka
*ke: > *k^e: > c^a
*ki: > *k^i: > c^i
*ku: > *ku: > ky
*kai > *ke > ce
*koi > *ke > ce
*kau > *kau > ku
*kou > *kau > ku
*kei > *k^ei > c^i
*keu > *k^eu > c^u
*kja > *k^a > c^e
*kjo > *k^a > c^e
*kju > *k^u > c^I
*kja: > *k^a: > c^a
*kjo: > *k^a: > c^a
*kju: > *k^u: > c^i
*kjoi > *k^ai > c^i
*kjoi > *k^ai > c^i
*kjau > *k^au > c^u
*kjou > *k^au > c^u
Skipping instead the k^-stage and keeping the kj-stage permits one
to reduce the number of rules by making '*kj- > *k^' superfluous.
PSatem > PS
*i > *ji
*e > *je
*i: > *ji:
*e: > *je:
PS >
*kjV > *c^V
*ke > *ce
PSatem PS
*ka *ka > ko
*ko *ka > ko
*ke *kje > c^e
*ki *kji > c^I
*ku *ku > kU
*ka: *ka: > ka
*ko: *ka: > ka
*ke: *kje: > c^a
*ki: *kji: > c^i
*ku: *ku: > ky
*kai *ke > ce
*koi *ke > ce
*kau *kau > ku
*kou *kau > ku
*kei *kjei > c^i
*keu *kjeu > c^u
*kja *kja > c^e
*kjo *kja > c^e
*kju *kju > c^I
*kja: *kja: > c^a
*kjo: *kja: > c^a
*kju: *kju: > c^i
*kjai *kjai > c^i
*kjoi *kjai > c^i
*kjau *kjau > c^u
*kjou *kjau > c^u
And now I'm back at my old favorite: the claim that in PIE the kW-
series and the k^-series were both realised as a pair of allophones,
namely *k/kW and *c^/*k, respectively, in which the former allophone
occurred before front vowels (and j?), the latter elsewhere. It
would have looked loked this (and this time the first stage really
is PIE):
PIE > p-PS
*V(:) > *jV(:), /V front
*ai, *oi > *e
p-PS > PS
*kW > *k
*kjV > *c^V
PS >
*ke > *ce
PIE p-PS PS
*kWa > *kWa > *ka > ko
*kWo > *kWa > *ka > ko
*ke > *kje > *c^e > c^e
*ki > *kji > *c^i > c^I
*kWu > *kWu > *ku > kU
*kWa: > *kWa > *ka > ka
*kWo: > *kWa > *ka > ka
*ke: > *kje: > *c^e: > c^a
*ki: > *kje: > *c^e: > c^i
*kWu: > *kWu: > *ku: > ky
*kWai > *kWe > *ke > ce
*kWoi > *kWe > *ke > ce
*kWau > *kWau > *kau > ku
*kWou > *kWau > *kau > ku
*kei > *kjei > *c^ei > c^i
*keu > *kjeu > *c^eu > c^u
*kja > *kja > *c^a > c^e
*kjo > *kja > *c^a > c^e
*kju > *kji > *c^i > c^I
*kja: > *kja: > *c^a: > c^a
*kjo: > *kja: > *c^a: > c^a
*kju: > *kji: > *c^i: > c^i
*kjai > *kji: > *c^i: > c^i
*kjoi > *kji: > *c^i: > c^i
*kjau > *kjau > *c^au > c^u
*kjou > *kjau > *c^au > c^u
Does Baltic complicate the picture? I saw a remark by Schalstieg
that Lithuanian and Old Prussian palatalise (PIE *kW >) *k before
front vowels, but Latvian doesn't (and Tokharian does too).
Torsten