Glen,
One of the things that may happen in
convergence areas is "cumulative creolisation", where diffusion taking place
over the millennia simply drowns the detectable genetic signal in the long run,
so that secondary (areal) affinities replace family features. This is what Dixon
argues for Pama-Nyungan and for the Australian languages in general, if you need
a present-day example. This is not "a wild possibility" but something that
happens every now and then. I'm pretty sure, for example, that the alleged
close relation between Germanic and Balto-Slavic which makes some linguists
propose a Germano-Balto-Slavic genetic unit or claim that Germanic is a
secondarily "desatemised" language, is in fact a purely areal phenomenon. As a
matter of fact, we don't see neat family trees very often -- the model works
best in cases like Polynesia, where there is a natural barrier between any pair
of languages. Just consider Germanic, Slavic or Romance, and the way divergence
and convergence have operated in those microfamilies. What does the abstract
entity we call Modern German derive from? Which "continental" element
(Anglian, Saxon, Jutish?) predominates in English? Handbook
classifications favour binary or tripartite divisions, so that we get
taxonomic figments like West Germanic, East Germanic, or South Slavic. I
completely agree with Ed that excessive concentration on genetic taxonomies
gives us a false picture of linguistic evolution.
Piotr
----- Original Message -----
Sent: Monday, March 05, 2001 4:19 PM
Subject: [tied] Greenberg and Nostratic
>[Ed:]
In addition, over-concentration on the genetic side of things also gives a false
picture because it is never all that is going on. Creoles never happened in
ancient history then?
[Glen:] Hmm, seems unlikely if you're talking about
"creole" in a non-layman sense. I would have thought that you need some precise
conditions for this to occur, like, say, mass-slavery of indigenous populations
or colonial expansion. Hmm... I don't know of any hunter-gatherer societies that
fit this bill :) Usually, a language is predominantly comprised of features of
one language group with the usual [ad/sub/ca/pro]-strate influences and so I
don't see this creole arguement as a serious one that needs to be
addressed.
>It may also well be that Dixon's model applies to IE's
relationship with its nearest 'relatives' because of the relatively greater
social equilibrium of the historical period in question.
No need
stressing out over every wild possibility. It's more logically economical to
presume that everything worked the same in the past as it does
now.