Re: *a/*a: ablaut

I think the best we can do is this:
If a PIE root in -a- has 'PIE cognates' with the ablaut vowel, it
should tentatively be considered a loan from an 'old IE' substrate
into a newer one.
If it has cognates in languagers outside of PIE, it should tentatively
be considered a loan either from those languages to PIE, or a loan
from some unknown language into both language groups.

Torsten

========

What is your proposal for :

Basque hartz "bear" Ursus
Hittite hartaka "bear"

Hittite is a loanword from Basque ?

Arnaud

=================