Principles of Operation (was: LANGUAGE SUBTAG REQUEST FORM, Erzgebirgisch)

Doug Ewell dewell at
Sun Jan 27 07:24:15 CET 2008

Frank Ellermann <nobody at xyzzy dot claranet dot de> wrote:

>> Upper Saxon, the language, exists.
> You say.  It's not like "gsw" or "nds", where
> my intuition is "different enough".  Without
> linguistic background I could make a case for
> "nds" as a separate language, but I couldn't
> for "sxu" (as of today, not centuries ago).

I don't say, ISO 639-3 say.

ISO 639 has historically been the standard upon which language tagging 
has been based.  RFC 1766 was limited to ISO 639-1 (plus registered and 
private-use tags).  RFC 3066 expanded the list to include ISO 639-2/T, 
and this was carried over into RFC 4646.  With RFC 4646bis we have 
planned to add seven thousand more language subtags from ISO 639-3.

In its attempt to catalog more than 7,600 languages, ISO 639-3 cannot be 
expected to be perfect.  As comprehensive, publicly available listings 
of human languages go, ISO 639-3 is the worst except for all those 
others that have been tried from time to time.

>> "sxu", the subtag,  doesn't yet exist, but
>> will in a matter of months.
> Maybe.  The registration request is now, and
> now our referene list is ISO 639-2, where I'm
> confident that it's no nonsense modulo "tlh".

This is the first hint I have heard that the very addition of ISO 639-3 
code elements to BCP 47 might be seriously challenged.

> IOW if "we" (or in fact Michael) would never make
> our own judgement we could hand over the registry
> maintenance to Ethnologue, and maybe get "ork".

I think, and this only my opinion, that our judgment and that of the 
Reviewer should be limited to what is not covered in the core ISO 
standards.  If they do not address an issue, we are free to apply our 
own interpretation.  If they do not provide a code element for a 
language, we are free to provide our own subtag.  If they, or their 
oversight agencies, assert that Erzgebrigisch is a dialect of two 
languages, I think we are free to choose one of the two for BCP 47 
tagging purposes.  But if they do the choosing, I think we must adhere 
to their choice.

> The zh-* registrations weeks
> before 4646 made them redundant were not rejected.

That's because 4646 did not change anything about the nature of data 
tagged as "zh-Hant".  Yesterday it was a registered tag, today it is a 
generative tag consisting of language and script subtags.  Content 
tagged "zh-Hant" never changed meaning; the tag never became "incorrect" 
or a suboptimal choice.

What is being proposed here is that 'de' be selected as the prefix for 
'erzgeb', not necessarily because it is the best choice, but because the 
discussion is being held a few months before the alternative choice 
'sxu' becomes available.  A possibly incorrect or imappropriate choice 
is being advocated in preference to waiting a few months.  That is what 
I object to.  I don't think the analogy with the "zh-Hant" tags is at 
all accurate.  A better analogy might be the "zh-cmn" tags which were 
expected to become redundant, but will now be grandfathered due to the 
decline and fall of extlangs.

> If the proponent feels that what he really wants
> is "sxu-erzgeb", then waiting until "sxu" exists
> is the best he can do.  Registering "saxon" under
> RFC 4646 rules as ersatz-"sxu" would be messy.

Agreed, and I don't remember anyone proposing 'saxon'.

Doug Ewell  *  Fullerton, California, USA  *  RFC 4645  *  UTN #14  ˆ

More information about the Ietf-languages mailing list