Excluding data

[Larry Trask]

On Wed, 22 Sep 1999, Jon Patrick wrote:

[on the current debate on selecting Basque words]

> we know that early glossaries/dictionaries do not contain words that
> did exist at their time of compilation.

Yes; this is true of all dictionaries.

> So I believe that a word in
> the list compiled by Azkue at the turn of this century should be
> used in any anlsysis unless it can be shown to be derivative from
> some foreign source. The default is "basque" not "foreign".

In "any analysis"?  No.  An analysis of the Basque existing around 1900,
fine: Azkue is a pretty good source, with a few qualifications.

But an analysis of Pre-Basque?  No.

The overwhelming majority of the words entered in Azkue's dictionary did
not exist in Pre-Basque.  So how can this dictionary, all by itself, be
a source of information about Pre-Basque?

Take any decent dictionary of English published around 1900.  How many
of the words entered in it existed in Old English, only about 1000 years
ago?  Not many.  Why should Basque be different?  (And the Pre-Basque
I'm interested in dates back to about 2000 years ago.)

> we know some words of limited distribution are in fact the earlier
> forms of words. For example, certain words with aspirations are only
> found in the north (the dialects with least speakers) but are
> believed to be earlier.

There may be some confusion here between regional variation and regional
attestation.  Many Basque words, of course, occur in variant forms in
different parts of the country.  This, in itself, is not an issue for
selection, though it may be an issue for determining the most
conservative form.

Quite different is the case of the numerous words which are only
attested *at all* in a limited part of the country.  These, in my view,
should be automatically excluded from any initial list seeking to
identify the best candidates for ancient status.  Some of them,
undoubtedly, are ancient words which have survived only in a small area,
but very many, probably the great majority, are words which are not
ancient and which have been created and used only in a small area.
Since we can't tell the difference in advance, all such words must be
excluded from any initial list, for the purposes I have in mind.

As I've pointed out before, the first goal is to identify those words
with the strongest claims to being native and ancient -- not to make
sure we include every word which might conceivably be native and
ancient, since any attempt at this must inevitably sweep up an awful lot
of words which are not ancient.  First things first.

Larry Trask
COGS
University of Sussex
Brighton BN1 9QH
UK

larryt at cogs.susx.ac.uk