Piotr:
> Here's a follow-up to my first response.
>[...]
> Let me also point out that the Hittite reflex of the labiovelar >series
>is k(u)w. Given the position of Anatolian relative to its IE >sisters, this
>is another little piece of data that adds to the >cumulative evidence in
>favour of reconstructing the PIE >pronunciation as [kW].
There is more, Piotr, when one takes the Nostratic theory into account. IE
sequences like *kWeC (C means "any consonant") may derive from a Pre-IE form
*kuC (a syllable like IE *keC without labiovelar would derive from *kiC). In
other words, the occasional labial qualities given to these IE velar sounds
*k, *g and *gh may in fact be pre-historical remnants of a lost neighbouring
vowel *u, shedding light into older stages of IE pronunciation.
A real example of this might be the IE particle *kWe (Latin que) which has
connections to Uralic *ku- (Finnish kuka), showing that Uralic contains the
old *ku sequence but IE has moved the labial quality of the original vowel
over to the neighbouring consonant *k, producing *kWe with an initial
labiovelar.
If we were to imagine IE *kPe instead, we would put everything out of
alignment both in the direction of IE's past (*ku) and its future (Latin
que). And so, on a Nostratic standpoint, *kPe is fun to entertain but just
doesn't work as well as *kWe does, for many reasons.
Of course, this doesn't mean that there weren't some goof-ups from time to
time, like Piotr suggests with /apa/ "water", since /kW/ can still easily
become /kP/ or maybe /hW/ > /f/ > /p/... or is /apa/ evidence of early
borrowing from a different IE dialect where /kW/ became /p/? Language mostly
evolves in a regular fashion but not always and sometimes there are
exceptions to rules because of the unpredictability of the people and
societies who speak it.
In the grand scheme of things though, *kW is a safer bet.
As for pronunciation of *kW, I suggest the 'qu' in 'quack'. On the other
hand, for words like *kwon "dog", we have to understand that it is really
*k- plus *-won _without_ labiovelar *kW. It is in fact two sounds *k and *w
put together to form *kw (not *kW!). This is why *kwon- becomes Sanskrit
/s'van-/ and not /*kan-/. The sounds *kW and *kw may have been kept apart by
pronouncing *kw differently, with a short vowel in between like the 'a' in
'sofa'. Thus, *kwon would have been pronounced /kaWON/ with heavy stress on
*-won.
PS: The verbs *ghWen- "to slay" (Sanskrit han-) and *ghWermo- "warm"
(English warm // Greek thermos) are examples of roots with *ghW- to answer
the other question.
Hope that helps everyone. Just call me Mother Theresa.
- gLeN
______________________________________________________