Bantu languages

[terrell at fmnh.org]

Colleagues,

Is there anyone on this list who has read Clare Holden's recent paper in
the Proc. Roy. Soc (25 March 2002) with whom a colleague and I may
compare notes?

We are intrigued that Holden reports using cognate sets as character
states.  On the face of it, this would seem to be a strategy that falls
between stools: what she reports doing, it would seem, is neither
numerical taxonomy (lexicostatistics), nor cladistics.

Her data set is a 92 item subset of Swadesh's famous 100-word list.  Yet
she reports using the cladistics program PAUP 4.0 to analyze this data
set-a program written to be used instead on character tables, i.e.,
tables of derived character states (what, in linguistics, we call shared
innovations).

Her paper shows that one can get results of a sort using this kind of
hybrid strategy (Holden points out, in fact, that her results are
strikingly similar to those obtained by others using lexicostatistics!),
but isn't this a case of looking through a glass darkly?

Specifically, aren't results obtained using this mixed strategy going to
be highly constrained/strongly determined by the underlying distribution
of shared innovations across the particular sample of words used?

John

John Edward Terrell, PhD
Field Museum, 1400 S. Lake Shore Dr.
Chicago, IL 60606
312-665-7822


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/an-lang/attachments/20020927/e582939c/attachment.htm>