Re: [tied] *kW- "?"

From: Grzegorz Jagodzinski
Message: 40550
Date: 2005-09-24

glen gordon wrote:
> Grzegorz challenges:
>> Why H2 here? And not H3? Cf. Latin os, costa,
>> Greek osteon, all with -o-.
> It's true that Hittite only shows /h/ in /hastai/
> which can be technically from either *h2 or *h3.
> However, the observed alternation between *q (trad.
> *k) and *h2 makes the most sense when one accepts
> that these two phonemes belong to the same "marked"
> class. Whether we want to call it "uvular",
> "pharyngeal", "velarized", etc. is another matter.
>> I'd rather see proto-IE *qW, *qu or *qo here which
>> yielded *ko ~ *H3o in PIE.
> I don't care what you'd "rather see" if it's not
> supported by facts. IE **qW first needs to be proven.

Nothing to be proven. There is an assumption that the *K ~ *H alternation is
the trace after former uvulars (you can present other explanation if you do
not like it). There is also an assumption that initial H1V- > e-, H2V- > a-,
H3V- > o-. We may also discuss whether H1V-, H2V- > o were secondary or not
but only if there exist related forms with e-, a- respectively. So, there is
virtually no base for Latin os, costa, Greek osteon < *H2o-. Because only
o-forms are known, we _must_ assume H3 here. And H3 means xW. And if it
developed from a uvular stop, we should expect *qW, *qu or *qo in proto-IE

> It's not... so get to work, man! :)

Hehe, and you should prove that *k was *q in fact, and especially that *g
was *G. And to explain why we do not observe the spirantization of it (G is
really very unstable phoneme in each language). So, get to work, man! :-)

Btw. uvulars do exist in Turkic but it is caused by back, not by low vowels.
So, it is a correlation between uvulars and /o/ rather than /a/. And you
ignore it completely.

>> *H2 can be in 'goat' instead: see Greek aig- (but
>> here -i- is unetymological).
> Fine, but if *h2 is in both, we then have a clear
> pattern of dialectal alternation, namely *q/*h2,

Not dialectal. Forms with *k and *H exist side by side, like in Latin os /
costa or Slavic koza / azIno.

> in
> a few examples like *h2ost-/*qost- "bone" and
> *h2aig-/*qag- "goat". That has greater explanatory
> power than your position.
> = gLeN

That would have had any explanatory power if you had managed to prove that
*k, *g, *gh were really uvular or what you like. I state for the tenth time:
there is absolutely no base for such a statement for some reasons.

1) plain velars fit to /a/ better than uvulars because of the same place of

2) there is not a reason to say that any *ka's (*ga's and *gha's) existed in
PIE in native words at all - all such a's can be /@/ or reduced grades of

3) even if we agree that /a/ was really a PIE phoneme, we can find it not
only in the *ka sequence but also in many other phonetic environments (see
for ex. Latin madere, cadere (k^, not k), nancisci (cf. skr. a├žnoti with
*k^, not *k) etc.)

4) if you still insist on uvular / glottal character of *k/*g/*gh, please
explain the strange behaviour of *b/*d/*g^/*g/*gW like Winter's rule etc.

5) and finally, what is evidence for *q/*G/*Gh in long-range linguistics

Comparison of your examples with EHL ( would be
really interesting, and if you find strong external evidence for uvular
character of plain velars in PIE, I will give up.

Grzegorz J.

Yahoo! Messenger - NEW crystal clear PC to PC calling worldwide with voicemail