Re: [tied] Greek labiovelars

To return to the original digression, the outcome of labiovelars in
Celtic, I guess the following would be fairly close to the communis
opinio (with notable exceptions):

1) From PIE to Proto-Celtic: *gW > *b (according to McCone, due to
the almost complete lack of PIE *b, thus filling the gap).

2) Loss of aspiration on stops, producing a merger between *g and
*gH etc. etc. This produces a new *gW from earlier *gWH.

3) Loss of *p (or maybe just *p > *h)

---

In Irish we then find delabialization of *kW and *gW (> c and g).
However, one can still tell the existence of earlier labialization
on the rounding effect of a following /a/ (> /o/) and /i/ (> /u/),
e.g. *gWaneti > gonaid, *gWediti > *gWidhith (raising) > guidid,
etc. And of course <Q> (= /kW/) and <NG> (= /gW/) in ogam.

The situation in British Celtic a more complicated, since, as has
already been mentioned, initial *w- anyway gives *gw-. So we may
assume both a retention of PCelt. *gW- or a development of *gW- to
*w- and back to *gw- again. The alleged exmaples of initial PCelt.
*g- > Brit. *gw- are simply wrong.

Word-internally the situation is more complicated, but it is
possible that the outcome is /v/, the same result as from lenited *b.

The case of *kW in British Celtic is well-known, namely *p.

There is some evidence for PCelt. *gW > *w- in Gaulish, eg.
uediiumi 'I pray'(?) < *gWediyu:+mi (OIr. guidid) < PIE *gWHedH-ye/o-

As can be seen, quite a lot has happened since Jackson wrote his
book.
The better understanding of the development of the labiovelars is
due to the research of Cowgill, Sims-Williams and McManus among
others.

Anders