Caucasian languages and Asia Minor

[Glen Gordon]

DR WOLFGANG SCHULZE:
>The problem is that people do not accept language isolates. What is an
>isolate? It's a language that obviously has no documented relatives.

Yes there is a problem but it's not the definition of "isolates". I
fully understand that this means that a given language MIGHT have
relationships but that the language is difficult to relate to others,
pure and simple. This could be for many reasons, one of them being that
there may be only very distant siblings that survive to be analysed, as
you say.

However, I have trouble with the idea that seems to arise on this
subject that because a language group can't be 100% proven to be related
to another language group that we should take up the NULL hypothesis in
place of it, ignoring the question completely of genetic relationship
and the probabilities and uncertainties inheirant in its answer (whether
this be NEC or IE). There still remains the most likely relationships
and I would be concerned of the reasoning skills of anyone who considers
the possibility of IE being closely related to Uralic as equal as the
possibility, say, of a strong relationship with Algic, half-way 'cross
the world. There remain high probabilities and low probabilities despite
lack of what you would accept as certain "proof".

This is the probabilistic nature of comparative linguistics. The idea
that IE is the ancestor language of the languages we find in Europe and
India now (as well as the ubiquitous English), is all based on
probability too. We don't find actual, _physical proof_ of people
speaking this supposed IE language but it is highly probable that one
existed in some form at least 5,000 years ago based on a damn good
theory that has been refined better as we learn more, a theory that,
because of alot of speculation beyond what we had known in the past, has
moved forward.

>Relatives are established by means of regular sound correspondencies
>based on lexical *words* (and not *stems*) as well as on morphological
>correspondencies that match established sound correspondencies.

And wouldn't it be wonderful in an ideal world? However, in the case of
IE, (and I suspect similarily in NEC) much of the recoverable language
is verb roots, whether that be nouns or adjectives based on verb roots
or extended verbs built on the more basic ones. Atomic nouns are not all
that common in the IE language and inevitably due to declension and the
nominative *-s, are ALL "stems" save those 1% who have no marker in the
nominative. Thus, morphological correspondances by your criteria must
hold most of the weight of the proof. So how much proof is proof enough
to make it credible? This is a very subjective thing. What we SHOULD be
concerned about is finding the relative probability of a hypothesis,
based not only on the data that one can supply for it but on the
probabilities of other possible theories of relationship in comparison
to it.

At that, we can conclude very apparent things such as Algic's
relationship with IE is extremely remote in comparison with a Uralic
relationship (as we should be concluding intuitively!) as well as more
opaque concepts such as the most likely pre-IE interactions with
neighbouring languages. Unfortunately, we can't even begin to fathom
answers to these questions, not because we don't know enough, but
because we've gotten stuck in a rut over the impossible-to-attain "100%
probability" and not on "the BEST probability that we can achieve at
present".

>[An isolate] simply is an orphan with unknown parents. [...]
>I think that's just what the situation is like with respect to East
>Caucasian.

Unless you're saying that NEC invented its own language, it's related to
something (even if it is remote as I suspect similarly). So there must
be some candidates that stick out amongst the random chaos of world
language groups like Benue-Congo, Austronesian, Ainu, etc. I'm willing
to wager that NWC is one of those better candidates even lacking your
absolute "proof". And because I know that IE is likely to be related to
something for the same reason as NEC above, I'm willing to wager that IE
(and Etruscan) are more closely affiliated with languages like Uralic
and Altaic as opposed to things like Basque (Larry will agree :), Ainu,
NEC (you'll even agree), etc. This is just common sense and taking the
NULL position on this subject isn't.

>These nagging questions are quite trendy, but that does not mean that
>they are on safe grounds with respect to method and language theory...
>But paradigms [hopefully] change, as Kuhn told us....

Humans answering nagging questions has been trendy for millions of
years. It's all logic and probability with a hint of imagination.
Hopefully this is a paradigm that will never change.

--------------------------------------------
Glen Gordon
glengordon01 at hotmail.com

Kisses and Hugs
--------------------------------------------