From: Arnaud Fournet
Message: 61099
Date: 2008-10-31
----- Original Message -----
From: Andrew Jarrette
To: pieml@yahoogroups.com
Sent: Friday, October 31, 2008 3:15 AM
Subject: [pieml] Labiovelars versus Palatals + Labiovelar Approximant
For PIE often the combinations *k^w, (*g^w), *g^hw are posited for
some words, usually based on Balto-Slavic and Indo-Iranian evidence
(e.g. words like Slavic *sveNtU "holy" < *k^went-/*k^wnt-, *sve^tU
"light" with which cf. Skt <çvetá> "white, bright", and the common
root *g^hwe:r- "wild animal", in Lith <z^ve.ris> "wild animal", Russ
<zverI> "wild animal, beast, brute", Greek <the:r> "wild beast, beast
of prey", and Latin <fe:rus> "wild, savage"). From the latter root
one can see that these combinations of palatal + *w developed
identically to the labiovelars in Greek and Latin. Thus in these
languages both the labiovelars and the palatals + *w must have been
pronounced as velars + *w.
============
You're asking a very interesting set of questions.
I suppose you'll get the official dogma from somebody else.
As I have already pointed at on cybalist,
the supposed distinction in PIE between :
kW
k + w
k + u
k^w
makes very little sense from a pure phonological point of view.
I guess there exists no real language with labio-velars and plain velars,
where the mono-phonemic kW should be different from the bi-phonemic k+w.
The graphic invention of k^w distinct from kW is a way of "saving" a couple
of words that are very dubious in the first place.
As I mentioned quite a lot of times,
dog k^won is bad, it should be kuH2on as confirmed by all the rest of the
world but the indo-europeanist dogma won't move an inch, I suppose.
horse ek^wos is worse, as this is not even a PIE-stage word and internal
correspondances within IE data are horrendous.
I think the opposition of k^ and k does not exist
but the opposition between g and g^ and gh and g^h exists.
Arnaud
============
What is original in these combinations,
palatals or velars? If, as many believe, the Proto-Indo-Iranian and
Proto-Balto-Slavic (etc.) palatals were allophones of original velars,
why do we find palatals before *w? If the palatals were merely
allophones of velars, that means that there must have been some
qualitative difference between *kW, *gW, *gWh on the one hand and *kw,
*gw, *ghw on the other. This has been discussed on Cybalist, and the
idea of timing of onset of the labiovelar approximant has been
suggested as the difference between the two series of sounds. But if
that were the only difference, why would the velars become palatalized
in this position?
Doesn't this mean there must have been original
palatals beside completely separate original velars, as well as
labiovelars?
===========
I believe the answer is yes.
Arnaud
=======
Or are original velars palatalized across the
intervening *w (perhaps a labiopalatal *w, like French <u> in
<huit>?)?
=======
This would require that PIE should _first_ have vowel ü
something that looks highly unlikely (not to say impossible)
But Attic treatment of labio-velars suggests it had ü as a glide.
It has been discussed on cybalist too.
Arnaud
=======
To me the only evident explanation is that there must have
been original palatals beside original velars, the two of which
however merged in Greek and Latin (and Celtic and Germanic). But the
fact that both are represented by velars (or their later developments)
across such a wide area over several language families, instead of by
e.g. palato-alveolars or simple alveolars, which might be more
expectable, would then be very hard to explain, and would suggest that
indeed there were only velars and labiovelars, no palatals.
========
No,
some languages have three series and you can't merge them into only two.
I mean Yeniseian for example.
the name "palatal" and "velar" may be misleading :
the opposition between g^ and g can be voiced velar g and glottalized k?.
all are originally velar.
The change of place of articulation may be a secondary development of
eastern dialects.
Arnaud
==========
What explains the mystery of palatals + *w, according to the
Balto-Slavic and Indo-Iranian evidence? (Not to mention the
suggestion of a third series of velars + *w, i.e. *kw, *gw, *ghw,
distinct from both *kW, *gW, *gWh and *k^w, *g^w, *g^hw, which seems
to be implied by some Balto-Slavic words? I don't see why this series
should exist, when no compatible distinction of e.g. "labiodentals"
vs. "dentals + *w" (**tW, **dW, **dhW vs. *tw, *dw, *dhw) exists.)
I really want to know this and therefore also the question of the
legitimacy of "palatals" for the sake of establishing the plausibility
and credibility of the IE-derived conlangs that I am inventing and
have invented.
Andrew
=========
What is this ?
Arnaud
======