Andrew Jarette wrote, in a web-hostile format:
(It took me three attempts to get your post into a format to which I
could make a reply!)
> So are you implying that the lack of voice or the presence of
aspiration makes the articulation of /kW/ and /gWH/ clearer than that
of /gW/, and therefore more likely to retain separate velar and labial
elements, since they are more audible due to greater clarity (from
lack of voice, clearer explosion, or presence of aspiration)? Also,
do you believe the idea that Celtic speakers began to pronounce /gW/
as /b/ so as to balance the IE deficiency of /b/?
On one hand, the distinction between [kW] and [p] is greater than that
between [gW] and [b], so it is easier not to learn the distinction.
It is even easier if one never hears a contrast between [gW] and [b] -
quite possible if [b] is rare.
> To me that sounds like they deliberately selected /gW/ against /kW/
and /gWH/ for aesthetic reasons, which I would find rather
unbelievable (and plus they developed a deficiency of /p/ which was
only compensated for in the Brythonic branch, while remaining in
Goidelic which had enough /b/ -- so why didn't they (Goidelic
speakers) start pronouncing /kW/ or some other sound as /p/ to make up
for this deficiency?
Well, [gW] does not seem to be very stable. As far as I can tell,
word-initially PIE *gWH merged with *g^H and *gH in Germanic -
possibly after affecting the following vowel. [kW] is much stabler,
or at least the cluster [kw] is.
> Of course, I suppose chronology comes into play here, that perhaps
at one stage there was a balance between /p/ and /b/ in Goidelic).
> Unless it were an entirely natural process of equilibration -- but I
would expect that /kW/ and /gW/ would still be subject to
labialization just as /gW/ was, in this equilibration process. I
guess that's where your explanation of the degree of sonorance of /gW/
would be a factor.
There's also the fact that some gaps in phonetic systems are stabler
than others. For example, if a language has only 5 of /p/, /t/, /k/,
/b/, /d/ and /g/, the most likely missing consonant is /p/ or /g/.
Classical Arabic lacks both!
Richard.