GENEALOGY-DNA-L Archives

Archiver > GENEALOGY-DNA > 2004-08 > 1092276939


From: (John Chandler)
Subject: Re: [DNA] Scythians, Indo-Europeans & Neolithic Farmers
Date: Wed, 11 Aug 2004 22:15:49 -0400 (EDT)
References: <20040812001910.2117.qmail@web52102.mail.yahoo.com>
In-Reply-To: <20040812001910.2117.qmail@web52102.mail.yahoo.com> (message fromellen Levy on Wed, 11 Aug 2004 17:19:10 -0700 (PDT))


Ellen wrote:
> I'm not sure I see enough evidence to support the idea
> of a single, unified language, not even back in Africa
> 100,000 years ago.

No, there doesn't have to be a single, unified language at any time in
the past. Nonetheless, the same processes seen in Y-DNA (extinction,
expansion, and differentiation) clearly operate in language, and there
is no reason to believe that modern-day languages are different enough
from each other to require multiple origins. Note that, just as
"mtDNA Eve" or "Y Adam" was not the only living human female or male
at any time, a precursor of all *surviving* languages need not have
been the only language spoken in its time.

> Your posting also addressed the problems surrounding
> rate change among languages. I'm afraid the same can
> be said for DNA mutational changes.

That's quite true.

> We like to think
> there is a basic, constant rate to mutational changes,
> and IN GENERAL this is true.

No, not in general, but only ON THE AVERAGE in the very long run and
with large numbers of individuals. That's the key to population
genetics -- taking averages over many generations for many lineages.
As far as genealogy is concerned, we always hasten to say that an
exact match does not prove any relation within a specified time
period, and a severe mismatch only makes a relation so improbable as
to be not worth looking for (if we don't already have proof of it) or
makes us challenge the lab that the test results must be in error
because we have proof of relation. For genealogical purposes, we are
definitely stuck in the regime of small-number statistics.

> We just don't know why one lineage mutates, sometimes
> quite quickly, while another doesn't.

There needn't be a "reason" at all. It hasn't been established that
environmental factors can have *any* effect on STR mutations. In the
small-number regime, you have to expect large statistical
fluctuations. That's just Nature at work.

> The same can be said for linguistic change. It
> changes can be predicted at a fairly constant rate in
> a GENERAL sense, but then random events sometimes
> cause faster or slower changes, as you noted with the
> English language itself.

It's very different in language because there is no large-number
regime at all. No such thing as a linguistic analog of population
genetics. No "general" sense anywhere -- just a modest collection of
PARTICULARS. Indo-European is a single language family that underwent
dramatic expansion and differentiation, bumping into particular other
languages along the way. Need I point out that the Indo-Europeans on
the fringes would naturally have bumped into more than their share of
other languages and thus appear to have branched off earlier than they
really did?

> If you are saying that Atkinson's approach is
> seriously flawed because there is no way that he can
> accurately reflect the random changes that sometimes
> happen in languages

That's exactly what I'm saying. I see no justification for their
assumption that an average rate of change even exists, much less that
they can guess what it is. Unexpected changes are always happening in
languages, but there is no obvious way of describing them as "random".
It is known that languages *are* influenced by contact with outside
cultures, for example.

> then I think you are condemning DNA research as
> well by this very argument.

Certainly. I have gone on record as condemning unrealistic and
uncalibrated DNA "research". If someone were to tell me he had used Y
STR tests to ascertain that the MRCA of two individuals was born 620
years ago, plus or minus 60, I would be equally scornful. I have
expounded at length on the misleading marketing of DNAprint. Earlier
today, Ken made the point that even population genetics may be flawed
by substructure found in populations. I'll have to think about that
one, but he could be right.

John Chandler


This thread: