Trask on Ringe

[manaster at umich.edu]

----------------------------Original message----------------------------
Larry asks about a point made by Ringe (1992), viz.,
"...resemblances between the basic vocabularies of
languages commonly believed to be related occur
with clearly greater-than-chance frequency, while
resemblanes between teh basic vocabularies of languages
not commonly believed to be demonstrably related do
not occur with greater-than-chance freuquency"
(p. 80).  Since he appears to hold (p. 80-81)
that such resemblances are the only way to test
for language relatedness, this seems to be saying
that any set of lgs not ALREADY commonly recognzied
as related can NEVER pass the test, since the
probabilities involved cannot change.
 
Ofcourse, not only this and but the whole
argument is really bad mathematics, as
shown by Baxter and MR in Diachronica.
 
AMR