*gwh in Gmc.

[Douglas G Kilday]

Miguel Carrasquer Vidal (5 Dec 2000) wrote:

>As I may have mentioned here earlier, I have been investigating the
>possible ramifications of hypothesizing that not only *k/*g/*gh had
>labialized (*kw/*gw/*ghw) and palatalized (*k^/*g^/*gh^) variants, but
>that this was originally the case for *all* (pre-)PIE consonants.

>One interesting possibility is **pw, which would have mostly merged
>with *kw (for obvious reasons, a labialized labial would have been a
>highly marked phoneme), but with *p in (pre-)Germanic.

I'm not comfortable with double stars, but *pw in Early PIE which merged
with *p in Pre-Germanic and with *kw in most other dialects  makes sense.

>This could be the case in the words "liver", "four", "-leven, -lve",
> >"oven", "wolf" and some others ("leave", "sieve", etc.).

I would add the tail-end of "five"; Goth. <fimf> suggests Early PIE *pempwe.
On the other hand I would remove "wolf"; Lat. <lupus> argues against a
labialized stop. Gk. <lukos> probably results from taboo-substitution by an
epithet 'bright-eyed', 'shining-eyed' or the like. I would refer "wolf" to
PIE *wlpos. Making all attested forms into taboo-variants, as Mr. Watkins
does, is an unnecessary shell-game.

>The existence of a parallel development *ghw (I prefer to spell it >this
>way) > Gmc. *b would be most interesting.  To be fully parallel >with **pw
>-> *f, there should be quite a number of exceptions to the >"law" (that
>it's controversial is therefore a good thing in principle).

Sadly we don't have a whole lot of PIE roots with labialized aspirates, and
some of the alleged cognate-sets are dubious, particularly in regard to the
Greek members.

>I'm not too pleased with "bane", however, being from the same root as
> >*gunT- "Kampf, Schlacht", which means a putative **bhwen- (for PIE
> >*ghwen) "to kill" is out of the question.  Not that it matters for
> >judging the etymology by its own merits...

I don't follow this. OE <bana> 'slayer' and <gu:th> 'battle' hardly look
like cognates. I would refer <bana>, OHG <bano> 'death', and probably Gk.
<phonos>, <pephamai>, etc. to PIE *bhen- 'to slay'. <Gu:th> etc., Gk.
<theino:>, and Lat. <defendo> belong with PIE *ghwen- 'to strike, fight,
defend, etc.' Ger. <Guenther> 'warrior' probably represents the zero-grade
agent *ghwntr-. I'm aware that some persons link <phonos> and <theino:>, and
I can't rule this out on strictly phonologic grounds. However, there's a big
difference between merely 'slaying' and 'being a soldier, defender'. If
correct, this analysis suggests *ghw -> Gmc. *g(u) initially. We also have:

  Lat. <frendo>, Gk. <khondros>, OE <grindan> <- *ghwren-
  OL <dingua>, OE <tunge>, OHG <zunga> <- *dnghw-
  Lat. <ungula>, Gk. <onukhos>, OE <naegl>, OHG <nagal> <- *Hnoghw-
  Lat. <levis>, Gk. <elaphros>, OE <le:oht>, OHG <li:hti> <- *Hleghw-

The last three suggest *ghw -> Gmc. *g medially unless the latter is
assimilated into *h (i.e. [x]) before *t.

In my opinion, PIE roots containing a labialized aspirate (traditional *gwh)
which becomes Gmc. *w are most easily explained by assuming that the Early
PIE root had *bhw. The phonetic realization in Early Gmc. was probably close
to [vw], and this could plausibly have been reduced to [w]. In "mainstream"
PIE, *bhw merged with *ghw. I hypothesize:

  Lat. <formus>, Gk. <thermos>, Skt. <gharmas>, OE <wearm> <- *bhwermos
  Lat. <nivis>, <ninguit>, Gk. <niphei>, OE <sni:wan> <- *sneibhw-
  Lat. <nefrendes>, Gk. <nephroi>, ME <nere>, Ger. <Niere> <- *nebhwr-

Early PIE *dhw may be represented in "deer", OE <de:or>, assuming this is
connected with Lat. <ferus>, Gk. <the:r>. The latter has the Aeolic form
<phe:r> which suggests a PIE labialized aspirate, just as Aeol. <pisures>
'four', <pempe> 'five' have <p> for Attic <t> where PIE had a labialized
stop. I would refer "deer" to *dhwer- which became *ghwer- in "mainstream"
PIE.

Doug Kilday