No, not really. The comparative method (which is what allows us to make
statements about language relationships) doesn't have the time-depth to reach that
far back. The farthest back it can reach with confidence is around 10,000 years
(and that's really stretching it -- many would say half that is the limit). But, for
instance, there have been humans in New Guinea and Australia for around 40,000
years, and the language patterns there couldn't be more different. New Guinea is
the most linguistically diverse place on earth, as you note, while in Australia the vast
majority of languages all over the continent (except the part close to New Guinea)
belong to only one family. Part of the diversity in New Guinea is simply due to the
fact that there hasn't been enough work on most of the languages, which are hard to
get to; but there's still plenty of diversity to go around. And in North America,
California is the most diverse place linguistically, with representatives of just about
every Amerind family except Eskimo; apparently it's always been a place where
people like to live, but nobody would suggest it as an Urheimat.
If there's
historical (or late pre-historical) records of migrations, like the Bantus in Africa or the
Austronesians in Oceania, then the norm of the most diverse geographic site is
useful for determining homelands (e.g, NW Africa for the Bantus, and Formosa for
the Austronesians). But we don't even know whether language originated only once
and spread, or whether it had many origins. And we probably never will. Too bad.
Good question, though.
-John Lawler http://www.umich.edu/~jlawler
Thinking
is more interesting than knowing,
but less interesting than looking.
-Johann
Wolfgang von Goethe
