Re: [tied] Why is PIE more centum than satem?

From: Piotr Gasiorowski
Message: 12259
Date: 2002-02-03

 
----- Original Message -----
From: george knysh
To: cybalist@yahoogroups.com
Sent: Sunday, February 03, 2002 7:36 PM
Subject: Re: [tied] Why is PIE more centum than satem?


--- Piotr Gasiorowski <gpiotr@...> wrote:
> As the old dictum goes, "Sanskrit consonants + Greek
> vowels = PIE", which of course mostly reflects the
> traditional bias of IE scholars.

*****GK: Is this traditional bias justified?*****
 
PG: Partly, perhaps, like many other stereotypes. Greek vocalism is indeed quite conservative, and so is the Sanskrit system of consonants (e.g. in preserving the *t : *d : *dH contrast more or less faithfully). Having said which, a conscientious linguist must add: HOWEVER, ... [numerous reservations follow]. For example, we have the Satem shift, Indo-Iranian palatalisations, merger of *r and *l, emergence of voiceless aspirates, Grassmannian deaspiration, Indo-Aryan retroflexion, and a lengthy list of other changes that do not allow us to _identify_ the Sanskrit system with that of PIE.


*****GK: This is not a satisfactory position. After
all people should know on what basis they have
"reconstructed" PIE, and what factors are involved in
the "deep transformations" you mention. If
arbitrariness rules over objectivity in one direction,
why should we assume it did/does not also operate in
the other direction? And if one cannot adequately
judge how far a particular language has diverged from
PIE how is it possible to feel confident that one has
actually "reconstructed" PIE?******
 
PG: I am not being evasive. Any specialist can tell you how to derive the Sanskrit or Greek phonological system, inflectional endings, etc. from a common prototype. The changes involved can be enumerated and clearly defined. You can count them if you like, but it's an idle pastime. What has evolved in either case is a multi-level _system_. Quite simply, a language is an immensely complex object, and it is far from evident that an objective "similarity metric" can be defined for such objects. What matters is not just how far Sanskrit or Greek have diverged from PIE (in quantifiable terms), but by what particular steps they have evolved out of it, and how the PIE reconstruction itself is arrived at.
 
Even a simple example should suffice: The PIE word for 'family, clan' is *g^enh1os. Latin reflects it as <genus>, and Greek as <genos>. Both look much more "similar" to the reconstructed protoform than Skt. janaH. If we take another word, say, *kWetwores 'four', one's subjective judgement might be that Lat. quattuor is more "primitive" than Gk. tettares or even Skt. catva:raH. But take still another form, e.g. *bHeronti 'they carry'. Here Skt. bHaranti looks at first glance more ancient than Gk. pHerousi or Lat. ferunt. Such impressions may be "suggestive" but are not very revealing in themselves, and often contradict one another. The important question is whether we can account in a logically consistent way for regular structural correspondences between Sanskrit, Latin and Greek forms -- and that's what we _can_ do.
 


*****GK: Surely one is not condemned here to endlessly
wallow  in the "everyone is entitled to their own
opinion" morass? Surely there must be a way to move
beyond "intuitive opinion". Are there not some
positions which can be clearly demonstrated to be
right or wrong? Or is the situation in the scientific
community of linguists that hopeless? *****

PG: No. What I'm trying to say is that it is only the notion of global "similarity" that is hard to define for large and complex structures, and which has no real heuristic value in historical linguistics. In comparing related languages we look for systematic correspondencies, which may or may not produce "similarities" (very often they don't). Comparative method is precisely what allows us to move beyond intuitive opinion and replace "similarity" with something that is objectively verifiable.
 
Piotr