Richard:
> However, I don't see any analogy with English if the labialisation
> is original. /w/ also rounds in Standard English cf. 'swan',
> 'wasp', and I'm not aware of /hw/ or /W/ voicing anything but
> itself.
Perhaps, but one thing English doesn't have is a distinction between
a phoneme like *kW and two phonemes like *kw. IE did. So there must
be, in effect, two levels of rounding going on here in IE, one "light"
and the other "heavy". I don't recall English having developed this.
(but no doubt Piotr may have an interesting tidbit of info on Old
English that negates my recollection).
> However, I can't think of any examples of that causing anything
> else to be voiced.
There's nothing that prevents any voiced sound to not assimilate
a neighbouring phoneme. It happens all the time. Even if it doesn't
happen in one language, there's always another that does do it. So
(gasp!) even though I'm not catching on to the idea yet, it's possible
that *h3 could be voiced, at least in some dialects of IE, and that
it could assimilate a neighbouring sound if *hW were simply a lightly
rounded [hW]. Thus the 3ps *pi-ph3-e-ti in some dialects could have
been pronounced [pibHWeti] instead of [piphWeti], thereby being
reinterpreted as *pibHeti or *pibeti.
= gLeN