Re: [tied] PK

From: Miguel Carrasquer
Message: 14885
Date: 2002-09-01

On Sat, 31 Aug 2002 14:33:08 -0700 (PDT), guto rhys <gutorhys@...> wrote:

>A few months ago Piotr touched on the PIE phoneme which gave both 'p' and 'k' in
>later reflexes. He mentioned something about a 'p' with some velar element (I think).

Must have been me. There is no doubt that PIE possessed a series of labiovelar
consonants (*kW, *gW, *ghW), which become plain velars (satem-group; Old Irish;
Greek), remain as labiovelars (Latin, Gothic, Ogham Irish, Linear B Greek) or
give labial consonants (Brythonic, Osco-Umbrian, Romanian, Greek).

In Germanic, however, there is a small number of roots which show *f, *b where
one would have expected *hw, *g(w). Some examples are: wolf, oven, leave,
liver, bid, from PIE *wl.kWos, *h2aukW-, *leikW-, *(l)ye(:)kWr-, *ghWedh-. My
hypothesis is that these words originally had labialized *pW, *bhW (note the
lack of *bW, as expected), which later mostly merged with *kW, *ghW, except in
Germanic, where the reflexes are as for *p, *bh.

It was objected that labialized labials are extremely rare in the world's
languages, which they are, but in languages where *i and *u have been lost
relatively recently (as e.g. in the NW Caucasian lgs., or as in PIE [where *i
and *u are indeed almost lacking as vowels]), one may expect the palatal and
labial feature of the lost vowels to have been transferred to the neighbouring
consonants, leading, amongst other things, to labialized labials. It is also to
be expected that the labialized labials will be the first to go (followed by the
labialized dentals), while labialized velars are a lot tougher to get rid of
(again, as seen in PIE, where only the *kW series maintained full phonological
status up to the breakup of PIE). Another objection was that there are no
parallels for a development *pW > *kW, which may or may not be true (in any
case, I don't know of any documented cases). However, the fact that the
development *kW > *p is extremely common, and that it's due to the large
acoustic similarity between *kW and *p, makes me feel quite confident that in a
language having *pW and *kW [of which there are not many, see above], the more
marked consonant (*pW) will tend to merge with the less marked and acoustically
quite similar *kW (just as in a language having *p and *kW, the more marked
consonant (*kW) will tend to merge with the less marked and acoustically quite
similar *p).

Note further that "Germanic *p (*b) ~ non-Germanic *kW (*ghW)" is actually an

- We have Germanic *kW > *ghW (Verner) in ON. ylgr < *wl.kWi: "she-wolf"
- We have *wl.p- in Latin lupus "wolf"; volpe:s "fox", Av. urupis "dog", raopis
"fox, jackal", Skt. lopa:s'á- "jackal, fox", Arm. aluês, Grk. alope:ks, Lith.
lãpe:, Latv. lapsa "fox".

- We have Germanic *kW > *h(w) in Gothic aùhns "oven"
- We have *aupn- in Greek (h)ipnós, Bret. offen "Steintrog".

- We have Germanic *kW > *hw in Gothic leihWan, E. loan, etc.
- We have *leip- in Toch. lip- "übrigbleiben"

- We probably have *lepr.t (although *lekWr.t is also possible) in Arm. leard

- We have *bheidh- Toch. B peti "Verehrung".

In other words, the tendency for **pW to become *kW also worked in Germanic,
while the retention of *pW (as *p, with loss of labialization) is not unknown
outside of Germanic either.

Miguel Carrasquer Vidal