Re: genetics & linguistics

On Thu, 1 Jul 1999, alypius wrote:

> See _The Great Human Diasporas_, by Luigi & Francesco Cavalli-Sforza, pp.
> 196-202.  ~alypius
I've received this book yesterday from the interlibrary loan system.
(Curiously enough, it appears to be in the possession of a hospital
library).

It's not a very scholarly work, more popular in nature, but extremely
interesting nonetheless. I can see his argument, but I can also see
the flaws. The author tries to prop up his general theory by using a
linguistic theory (by first defining his genetic populations as defined by
the subgrouping of languages - using a suspect subgrouping as a starting
point), and then uses the results of the genetic theory to prove the
linguistic theory.

If I understand Cavalli-Sforza correctly, then genetic drift indeed
proceeds at a regular rate, and he's naturally taken to the theory that
linguistic drift also proceeds at a regular rate, much like genetic drift,
a theory  which he takes as proven. Johanna Nichols (1992:214) correctly
points out that this basic assumption is wrong:

  There are enormous differences between the material foundations of
  population genetics and those of linguistics: population genetics
  has a secure mathematical basis because there are two alleles at each
  gene locus, the chances of one or the other being inherite by an offspring
  are precisely calculable, and the effects of natural selection can at
  least be modeled. Linguistics has no analogs to alleles, genes, chromosomes,
  etc., or to inheritance or dominance; there is very little natural
  selection operating on language, and nothing resembling the two-allele
  inheritance.

She also points out some broad analogies, however.

Cavalli-Sforza's argument means he has to use some rather suspect linguistic
sources, like Swadesh and the later Greenberg - and he condemns
critics of those two not with arguments, but by saying the criticism is
incomprehensible to him, that history will prove him right, the critics
are not really interested and besides, don't know what they are talking
about (182-183). I am rather more impressed by Matisoffs (Matisoff 1990)
arguments against Greenberg, myself.

Of course, Cavalli-Sforza may be right, if only in the broadest, general
terms, but even he admits his theories break down in the chaos prevalent
in, for instance, the Himalayas.

I finally came to the conclusion that while a correlation between genetic
grouping and the group of speakers of a certain language is established
as being often present, there is no necessary corrolation, and that in
every specific case both groupings have to be established independently
and then the areal extent can be compared: genetic evidence cannot
prove a linguistic hypothesis, nor can linguistic evidence prove a
genetic hypothesis, but if both types of evidence independently give
the same results, the whole theory is strengthened. Since the linguistic
affinity of Korean and Japanese seems still not to be satisfactorily
determined, where not at that point already.

Perhaps the central issue is that people like Cavalli-Sforza (and
Greenberg) tend to want to look at the broad issue, and are rather
disdainful of the details. It appears that they often think that
the actual quality of the data they use is relatively unimportant,
because the broad sway of their research will cancel the errors
out, and as a consequence they seem to be intolerant of evidence
that contradicts their theory, especially if that is on a level of
detail they do not work with, themselves.

References

Matisoff, James, 1990. On Megalocomparison. Language, 66. 109-20

Nichols, Johanna. 1992. Linguistic Diversity in Space and Time. Chicago:
University ofChicago Press

Boudewijn Rempt  | http://www.xs4all.nl/~bsarempt