*hwergh-, *hwerg- or *hwerk-?

[X99Lynx at aol.com]

JoatSimeon at aol.com wrote:
<<The agreement between Hittite and Tocharian -- very widely separated IE
languages -- would suggest PIE status for this word as well.>>

In a message dated 2/12/00 10:12:09 PM, mcv at wxs.nl replied:

<<Only for the root, strictly speaking.  The words are formed quite
differently (Hitt. *HwrK-is, Toch *HwerK-ontos).  It does seem to
indicate that this (*hwergh-, *hwerg- or *hwerk-?) was the
preferred word for "to turn, to roll" at quite an early stage.>>
("mcv at wxs.nl earlier wrote:  <<...the Hittite word for "wheel" is <hurkis>.
No relation to the *kwel-words.>>)

Let me ask, does evidence of (*hwergh-, *hwerg- or *hwerk-?) appear in any
other IE languages?

I guess it would have to.  Otherwise you'd have at best a Tocharian/Hittite
connection, but not necessarily PIE.  Am I right?

After all, you wouldn't want to be finding PIE roots every time just two
ancient IE languages showed cognates - that would mean that PIE would need to
have a a larger lexicon than any of its daughters, since it would always have
an original word for as little as two cognates among the daughters.  This
seems to be too much to ask of a real or even a hypothetical proto-language.

(This idea is interesting though.  It makes PIE look like a language of
nuggets of abstractions (e.g., to turn, to roll) unrelated to any concrete
object, waiting for some practical application.  Of course this might be
conceived of both very foresightful and very thoughtful of those early
PIEists, forseeing the needs of all those future daughter languages.)

I wouldn't of course think - if these two languages were the only evidence of
root mentioned above - that a PIE root or root-stem would need to be
conjectured simply because of some assumption that Hittite and Tocharian had
no contact or common ancestor after PIE split.  I can't take seriously the
idea that it is ENOUGH to say that Hittite and Tocharian are "very widely
separated IE languages."  For one thing, all IE languages are geographically
widely separated from Tocharian, but that couldn't always have been true.
And for another thing, some trees at least (e.g., the UPenn tree) have
Hittite and Tocharian right next to each other in terms of relatedness.

Now it may be that evidence of (*hwergh-, *hwerg- or *hwerk-?) might be found
dispersed throughout the IE languages in which case all this may be moot.

But if it isn't - my question is:

Just because (*hwergh-, *hwerg- or *hwerk-?) is reconstructable, why
reconstruct it and  then call it "the preferred word for "to turn, to roll"
at quite an early stage?"   (Why could it not have been a later innovation,
born out of one of Stefan Georg's "expressives" and then formalized as a
non-expressive?)

I mean, IF the word is only found in Hittite and Tocharian, why not swing the
conclusion around to its simpliest form and simply take it as evidence of
contact between Tocharian and Hittite?  Do you lose valid historical evidence
of contact when the presumption is that such words must have had PIE origins?

Regards,
Steve Long