--- In cybalist@..., "Piotr Gasiorowski" <gpiotr@...> wrote:
> That does not make any sense. I must have misunderstood you
somewhere?
>
>
> [Piotr:] Yes, you did. I'm speaking exclusively of forms that have
*K^ in a non-alternating environment. For example, the IE root *k^leu-
has the allomorphs *k^lou- and *k^lu-, but not **k^elu- or the like.
There is no form which could serve as the basis for "satem
regularisation" here. You could say (as you did, in fact) that such
words were "infected" with a change initiated elsewhere, but this
loose idea of "infection" is arbitrary, since other similar roots
were not infected (e.g. *kreuh2- 'blood') and you don't provide an
explanation for that.
Yes that occurred to me, but too late yesterday. But you're
right, "infection" is what I had in mind.
Torsten