> > So you accept that
> > at some stage of IE or pre-IE there was a velar series *[k q kW].
> > If so, it is logically unnecessarily to derive *[k' k kW] from it for
ALL
> > of IE, when there is no evidence of palatal velar stops in anything
> > but the satem languages.
I lost something here. Why must [k'] be posited for the centum languages?
Stage A: Satem / k kW q / = Centum / k kW q /
Stage B: Satem / k' kW q / Centum as above
Stage C: Satem / k' k q / Centum as above
Stage D: Satem / k' k (k) / Centum / k kW
k /
This requires both IE dialect groups to have brought [q] forward to [k], the
Centum languages doing it completely, while in one or two instances in one
or two Satem dialects it did not merge completely with the [k] from [kW].
So there never was a Centum [k'].
Peter