Fw: Erzgebirgisch Classification Question

Frank Ellermann nobody at xyzzy.claranet.de
Tue Mar 25 18:08:13 CET 2008


Doug Ewell wrote:

> I thought we had this discussion already about adding a
> second prefix, and frankly I'm surprised to see it come
> up again.

I wrote "might be possible", because that's as it is, a
variant can have more than one prefix, and that's the case
for various existing variants.

> The purpose of multiple prefixes is to allow a single
> variant subtag to be associated with multiple languages,
> NOT to provide "aliases" for the same language variation.

And that point isn't obvious for cases like erzgeb, which
can be seen as "de" dialect, or as "sxu" dialect, assuming
that "sxu" is a language and not another "de" dialect.

An assumption I don't share, but as I'm not planning to use
this tag for an Accept-Language "de" audience this is not
relevant for the erzgeb registration - Thomas decided to 
wait for "sxu", and show that it is wrong later, problem
solved.

Somebody intending to tag de-erzgeb for various plausible
reasons can still differ.  Nobody is forced to believe in
ISO 639-3 classifications when they make no sense for the
intended usage.

> it would imply the existence of both an 'erzgeb' variant
> of Upper Saxon AND an 'erzgeb' variant of German, which
> is not the situation here.

It would imply that "sxu" is an arguably dubious concept 
not confirmed by ISO 639-2.

> We have enough trouble with duplication as it is, with
> registered tags that are superseded by ISO-based subtags,
> and ISO-imposed overlaps like no/nb/nn.

Yes, ISO 639-2 has its own warts, and some grandfathered
i-anything etc. are also odd.  If folks have no compelling
reason to introduce new overlaps their request won't make
it in the review.  
 
> We need to understand this and not promote the idea that
> we can always add a "better" alias for the same variation
> later on.

I disagree with treating ISO 639-3 as sacrosanct, as source
I'd treat it with the same caution like say Wikipedia, and
a correct info following ISO 639-3 rules is not necessarily
useful for the practical purpose of tagging Web pages.

 Frank



More information about the Ietf-languages mailing list