Re: [tied] *kW- "?"

>> And by the way, I never said the theory's not
>> valid, I only said that it's probably pre-IE rather
>> than proper IE.
>
> Fine, but there is still no logical reason to theorize
> that an unstable system lasted longer than necessary.
> You follow no consistent methodology.

Of course, only your methodology is the right one...

>> Maybe your not aware that languages are seldom
>> optimally efficient.
>
> Wow. I'm floored that you missed the basic point:
> We can't depend on rarities, oddities, random
> goofball changes, etc, etc, etc to formulate an
> unbiased and logical theory. Your statement is
> frustratingly irrelevant but you repeat this rot
> over and over as if it changes good methodological
> procedure.

IE is hardly what you would call a "normal", "logical" language. There are
quite a few rarities and oddities in IE, so another one is hardly
surprising. As for "random goofball changes", I don't see any...

>> Palatalized velars are attested (almost) directly
>> by it's fricative/africate reflexes,
>
> No. **Satem** languages only attest to _*c^_!

If we are to hairsplit - they attest *c´ (ts´) rather then *c^. And I
don't beieve that *k´ > *c´ (if it were always exactly this change and not
a similar one) was a common development in all "satem" lgs. I find it hard
to believe that for instance Greek and IIr, which are otherwise quite
close, are separated by this isogloss and that this isogloss cut the
Anatolian branch into two parts.
I think that that is highly improbable - I concur with Melchert who
reconstructs *k, *k' and *kW for Proto-Anatolian. What would your solution
be? That already in Proto-Anatolian some dialects had *c^ and some had *k?

>Aside
> from violating markedness, palatal **k^ is not in any
> single IE language at all.

Neither is *q. But *k' can be reconstructed on the basis of its
reflections while *q cannot.

>While *q may equally be
> unattested directly, it does not violate markedness.

It's not "equally" unattested. Reflexes like fricatives, africates
compared with plain velars clearly point to palatalize velars while for
your supposed *q all the languages show just regular plain velar reflexes.

> In fact, what few languages truly have a palatalized
> velar stop in the first place?

None, but no lgs have *gWH either and it's not difficult to reconstruct it.

> Logic still favours *k-*q-*kW over *k^-*k-*kW.

No, it does not.

Mate