Odp: Labiovelars

From: Piotr Gasiorowski
Message: 1873
Date: 2000-03-16

 
----- Original Message -----
From: Gregory L. Eyink
To: cybalist@eGroups.com
Sent: Wednesday, March 15, 2000 7:29 AM
Subject: [cybalist] Labiovelars

Hi, again! Here is a second question. It is always stated that the
articulation of labiovelars in PIE is unknown with precision. However,
they are commonly represented orthographically as something like "kw"
for unvoiced and "gw" for voiced (as in Piotr's system). Yet, it has
always seemed to me strange that the sound literally represented by
"kw" could become the labial "p", or "gw" become "b" , as it did in
so many branches of IE (p-Celtic, Osco-Umbrian, Greek, or Germanic
in certain cases, such as Gothic "fidwor" and "wulfs"). Admittedly,
phonology is not my strong suit. However, it has has always seemed
to me that something more like "kp" or "gb" would represent better
the original pronunciation and lead more naturally to the splits
"kp"-->"kw","p" and "gb"-->"gw","b". By a little trial and error,
it does seem possible to articulate something like "kp" or "gb" in
a unitary fashion (as the early IE scripts seem to indicate speakers
felt labiovelars to be single sounds).

All of this leads me to my question: does anyone know of examples
in which the transformations "kw"-->"p" or "gw"-->"b" are actually
attested, in historical records of any language? Can it actually occur?
I know the example of Rumanian, in which words like Latin "aqua"
appear as Rumanian "apa". However, it seems that this may be a
substrate influence of the Dacian languages of the area. As far as
I know, there are no written documents tracing "aqua" --> "apa"
in Rumanian. Are there other examples? And, for those who are more
knowledgeable in phonology, is the transformation "kw"-->"p"
and "gw"-->"b" at all plausible?
 

Hi, Gregory, welcome to the list,
 
What you're talkig about is, technically speaking, the difference between labialised velars (/kW/, etc., i.e. velar consonants accompanied by lip-rounding) and labiovelar (doubly articulated) stops (let's transcribe them /kP/, etc., to keep them apart from /k/+/p/ sequences). Both types of sound are pretty common in the languages of the world, the latter especially in Africa, e.g. in Yoruba (kPakPa 'field'); the very name of Igbo (a major Kwa language, spoken in Nigeria) contains /gB/.
 
The contrast between the two is hardly sharp in acoustic or articulatory terms; in either case you have a combination of a velar closure with some kind of secondary labial activity. The difference depends on the strength of that secondary element. A weakened allophone of /gB/ may well sound like [gW], and by giving some extra emphasis to the labial element in /gW/ you can turn it into something like [gB]. If the labial articulation comes to dominate the velar component, you can get [gB] > [b] and [gW] > [w] or [b] (either directly or via [gB]); [b] inherits the manner of articulation of the labialised velar and its labial place of articulation, promoted to primary status. All these changes are perfectly natural from the phonological point of view. A comparable change (with fricatives instead of stops) is [xW] > [f], very amply attested (even in English -- compare the spelling and the pronunciation of laugh, enough, rough, etc.; the [f] developed historically from a velar fricative [x] which acquired lip-rounding from the vowel it followed).
 
As a matter of fact, it has been proposed before (by A. Martinet, I think), that the PIE labiovelars were realised as doubly articulated stops. But there are some good arguments in favour of the traditional interpretation. There is some evidence of strong lip-rounding influencing the quality of an adjacent vowel (e.g. in Greek); /w/ also produces such an effect, but labial stops don't. The Latin change of */gW/ into /w/, or Germanic */gWH/ > /w/, square better with the standard view. Perhaps most importantly, in the Centum languages biphonemic sequences such as */k/ + */w/ nearly always develop exactly like the reconstructed labiovelars; e.g. ekwos 'horse' (NOT *ekWos! -- cf. Sanskrit aśva-) > P-Celtic *epo- (cf. the Gaulish horse goddess Epona).
 
As for my personal opinion, I would admit the possibility of [kP] etc. being an occasional realisation of the corresponding labialised velar and making it more confusible with [p] -- at least in those IE dialects that gave rise to the various Centum branches.
 
Best regards,
 
Piotr