John Cowan jcowan@reutershealth.com
Wed, 4 Sep 2002 01:06:04 -0400 (EDT)

Peter_Constable@sil.org scripsit:

> I still suspect there may be some value in this tag

I agree with all your points.

> > For software development where a distinction between es-ES and
> > *es-americas is desired, I think the simplest thing to do is to
> > choose one of the American countries, whichever one has the most
> > features of the variety of Spanish you are identifying as
> > *es-americas, and use that.
> I'd like to hear from advocates why they might consider this suggestion
> inadequate (if, indeed, there is any reason it might be considered
> inadequate).

Because it's utterly arbitrary.  As part of their informal contract
with Reuters Health, the people who translate our news articles guarantee
its acceptability in all New World countries, but not necessarily in
Spain; and indeed, our group in Spain says the language used is (subtly)
unacceptable there.

To say that the text of the first group is es-us, or es-mx, or ... or es-cl
is to import a distinction which does not exist, but to say that the
text is es, unqualified, is to blur a distinction which does exist.

There is already some arbitrariness in language tags: thus Plains
Indian SL is tagged sgn-us-sd (IIRC), as if it were restricted to
South Dakota, but at least this just prevents tags from being totally
analytic, which they can't well be anyway.  Requiring the arbitrary
choice of an overspecific tag is quite another matter.

Michael, please reconsider.

John Cowan
                I am a member of a civilization. --David Brin