Rob:
> Assuming that the original PIE gender distinction was an animate-
> inanimate distinction, a house is hardly capable of animacy. Now,
> that doesn't mean it can't be classified as animate, but not on a
> logical basis.
No, but why then do you expect rigid logic from illogical creatures
called 'human beings' who speak these languages?
> I accept that. However, grammar has to be logical,
No, _rules_ have to be logical in the sense that they must logically
explain the origin of the phenomena that we observe or at the very
least describe the pattern that we see in optimal fashion. The
phenomena themselves however don't need to be logical.
If grammar were logical, we'd all speak one unchanging language since
the slow change of language that naturally occurs across the globe is
a very product of our human imperfection. Any changes that occur in a
language could be deemed illogical by nature since they are in effect
random. If they weren't random, we could predict what English will
be like in 2050. We can't. Randomness defies logic.
> I think this is very possible -- I didn't reject the possibility that
> *wlkWos was earlier an adjective. Is there indeed an attested root
> *welkW- "howl"?
Not that I know of, however the word is thought to have been taboo
given other variants like *wlpos, so perhaps the form of the verb
I cite is incorrect and it is instead *welp- or even something I
haven't considered yet.
= gLeN