On Thu, 11 Oct 2001 11:30:01 -0000,
tgpedersen@... wrote:
>I thought the results of Cavalli-Sforza etc showed that for the most
>part, genes and languages do (and please don't use that African-
>American example; its scope is unprecedented)?
If you're referring to the infamous diagram in Cavalli-Sforza et al.
1988 (reprinted in several of Ruhlen's publications), it shows no such
thing. The diagram is cleverly drawn up to give the impression that
genes and languages correlate, but upon closer inspection it turns out
that the layout is only there to trick us, and the whole thing is a
deception. There are 38 genetic groupings listed, and assuming the
left hand side of the diagram gives us an accurate account of the
genetical relationships (Cavalli-Sforza is the specialist, not me),
the expected vs. the observed linguistic groupings would be:
EXPECTED OBSERVED
Congo-Saharan Niger-Congo
Niger Nilo-Saharan
Pygmy
Khoisan-Cushitic Khoisan (?)
Irano-Semitic
Berber Afro-Asiatic
European Basque
Sardinian Indo-European
Indo-Dravidian Dravidian
Saami Uralic
Uralo-Mongol Altaic[-Korean-Japanese (?)]
Tibeto-Koreo-Japanese Sino-Tibetan
Ainu !Ainu
Turkic
Eskimo-Chukchi !Eskimo-Chukchi (?)
S-C-Amerind
N-Amerind Amerind (?)
Na-Dene !Na-Dene
Sinitic
Austric-Thai Austric (?)
Malayan Austronesian-Thai (?)
Indonesian
Philippine
Polynesian
Micronesian-Melanesian Papuan (?)
Papuan-Australian Australian (?)
-- Kartvelian
-- NE Caucasian
-- NW Caucasian
-- Burushaski, etc., etc.
Which means that in the whole diagram there are only *two* (2) hits
(Ainu [duh!] and Na-Dene[*]) and one possible hit (Eskimo-Chukchi).
The rest of the linguistic groupings fail to correlate with the
genetics.
[*] without comitting myself on the status of Haida.