LINGUIST List 14.1917

Sun Jul 13 2003

Disc: Are new language classifications necessary?

Editor for this issue: Karen Milligan <karenlinguistlist.org>

Directory

1. Chris Beckwith, Are new language classifications necessary?

Re Linguist 14.1914

I agree with you that some prominent language classification theories
need to be reexamined. I would prefer to see these theories subjected
to rigorous linguistic testing first, and then decisions made about
whether or not they seem to be worth pursuing any further. In many
cases (e.g., Altaic, Sino-Tibetan, Japanese-Korean) the linguistic
relationship proposals do not stand up to serious testing. In cases
such as these the languages involved in a proposed classification
theory may be relatively poorly known, allowing weak linguistic
relationship proposals to survive much longer than they otherwise
would. If your method can support, or predict the results of, careful
historical linguistic analysis, it would be a useful tool to add to
those presently available to linguists working on languages and
language relationship theories such as those mentioned. Unfortunately,
I have not yet seen your book or Marcantonio's, neither of which is
available in our library.

If I understand your posting correctly, your analysis is that neither
Uralic nor Altaic seem to be divergently (genetically) related
language families, unlike the others in your list. I would certainly
agree with Altaic, which theory has been (in my opinion) disproven,
and I have always had doubts about the Uralic family (though not about
Finno-Ugric). What worries me a bit is that according to your figures,
Uralic, Finno-Ugric, and Altaic have very similar 'coefficients of
variance' (you say Uralic is not as compact as Finno-Ugric, but the
difference is small). Yet it has always been accepted by Uralicists
that Uralic is divided into two relatively distant (sub)families,
Finno-Ugric and Samoyedic. It is not problematic that Samoyedic is
more compact, but it is problematic that the coefficients of
Finno-Ugric, Uralic, and Altaic (the last being a totally unrelated
'group' of languages) should be so similar. Is it possible that your
method essentially measures typological variance?

If your method is based on phonological variation, it will not be of
much use for examining older language families. Tokharian, for
example, was immediately recognized as an Indo-European language when
it was first discovered, despite the fact that it ended up belonging
to a distinct, previously unknown branch of IE. Yet it has eliminated
the three-way opposition in the PIE stops, among other changes it has
made. And Armenian has other radical phonological
differences. Including these and other languages would surely give IE
a similarly high, or even higher, coefficient of variance. So, does
your method take into account convergence, and simple chronological
change?

Christopher I. Beckwith
Indiana University