Basque statistics - methodological contradiction

[Larry Trask]

On Fri, 3 Sep 1999, Jon Patrick wrote:

[snip quote from me]

> This is presented after a long and detailed response to Lloyd
> Anderson's revision of his 6 rules for deciding what should be used
> as acceptable words in the study of early basque words. Larry's
> rules are highly restrictive and my counter argument is that they
> are too restrictive and they do not let the data speak for
> themselves.

The problem with "letting the data speak for themselves" is that ancient
Basque words are vastly outnumbered today by words of more recent
origin.  So, if we can't systematically exclude the newer words, we have
no reasonable hope of picking out the ancient ones.

> It seems to me, unless I am misreading something, that Larry's final
> comment objects to someone else entertaining an a priori model of
> the data as making assumptions, but doesn't perceive that he is
> making assumptions from his own expectations of what a basque word
> should look like.

But, Jon, I am *not* making any assumptions in advance as to what a
Basque word should look like.  Observe that not one of my proposed
principal criteria has anything to do with the *phonological form* of a
word.  My principal criteria are distribution, date of first
attestation, and absence of the word in neighboring languages: nothing
to do with form at all.  My secondary criteria exclude apparent nursery
words and imitative words, which admittedly have something to do with
form, but both of which can be identified by independent criteria having
nothing to do with what I hope or expect to find in Pre-Basque.

Larry Trask
COGS
University of Sussex
Brighton BN1 9QH
UK

larryt at cogs.susx.ac.uk