*gwh in Gmc.

[Hans-Werner Hatting]

Thanks to Miguel Carrasquer Vidal for the enlightening answers in his post
of Sun, 11 Feb 2001 17:34:00 +0100.

The question at hand seems to be what to do in cases when languages deviate
from those results we expect by the normal sound rules. I think all
possibilities (positing a new phoneme for PIE, thinking about dialect
mixing, or trying to find a rule to account for seemingly irregular changes)
are methodologically admissible.

>Well, most of the words in my list offer sporadic cases of */p/ in
>lgs. outside Germanic (as well as variants with */hw/ in Germanic).
>Arm. <leard> "liver" can be either *lepr.t or *lekwr.t.  Both *leip-
>and *leikw- occur in most IE branches.  For "oven", we have Grk.
><ipnos>, Bret. <offen> (as well as Goth. <auhns>).  *seip- is alo in
>Tocharian (and *seikw- also in Germanic).  "Wolf" has forms with *p in
>Latin and a similar root (*wlp-) exists in I-I, Grk, Arm. etc. with
>the meaning "fox" or "jackal".

A parallel to the current discussion is the case of the k/g reflexes for PIE
*k4/g4/g4h in Satem languages. In this case, as far as I know, nobody has
posited an extra series of (e.g.) half-palatalised k/g/gh; the usual
positions in this case seem to be to assume borrowing or a wave-light spread
of the palatalisation phenomenon, which left out some words in the languages
(Baltic and Slavic) farther away from the center of the wave.
So maybe we have a similar case here, and the variations between labiovelars
and labials quoted are just witnesses of an uneven spread of the
labialisation of labiovelars, while the words with /p/ for */kw/ in Satem
languages are just later borrowings (at a time when the change /kw/ > /k/
had already occurred, and /p/ was substituted for /kw/ in the borrowed
word).

>My proposal makes the most sense within a wider context where _all_
>(pre-)PIE consonants had labialized (and palatalized) variants.  This
>would be comparable to the developments that can be seen later in Old
>Irish (3-way split between "slender", "broad" and "u-coloured"
>consonants) or in Tocharian.  The Tocharian case (where *i, *u and *e
>merged as *@ (*a"), or rather: *e > *@, *i > *(y)@, *u > *(w)@) is
>especially interesting, given the lack of *i and *u in PIE
>[full-grade] root structure (so maybe **CiC > *C(y)eC, **CuC >
>*C(w)eC).  As was the case in Old Irish and pre-Tocharian, such a
>system with a 3-way opposition was inherently unstable, and was
>eventually resolved leaving a number of irregularities.  The
>alternations between *p and *kw (with Germanic mostly, but not always,
>on the *p-side, the other lgs. mostly, but not always, on the
>*kw-side) can be interpreted that way, as can other PIE irregularities
>(e.g. *t ~ *s < *tw (cf. the Greek soundlaw *tw > s) in the words for
>"month", "dawn", the pf. act. ptc. in *-wot-/*-us-, etc.; *n ~ *i <
>*n^ in roots like *nem-/*yem-/*em- and the Vedic *-i/*-n-stems; *l ~
>*i < *l^ in the "liver" word, maybe also in "yoke"; *m ~ *w < *mw in
>the 1 sg., du. and pl. of the verb, etc.).

Anyway, this is an interesting concept. Did you elaborate on this anywhere?

Best regards,
Hans-Werner Hatting