Frisian problems (was: New language subtags)

[Doug Ewell]

(adding LTRU back to the list because of RFC 3066bis registry philosophy 
questions)

John Cowan <jcowan at reutershealth dot com> wrote:

>> Note also that "frr" does not currently exist
>> in ISO/DIS 639-3, although it does not conflict with it.
>
> Where do you get this from?  Both the Ethnologue and the ISO 639-3
> site show 'frr' as the code for Northern Frisian.

You're right, I screwed up.

Actually a bit of explaining is in order.  As some may know, I have 
already gone ahead and mocked up a possible registry for the anticipated 
future "RFC 3066ter" era, in which each ISO 639-3 code element would be 
utilized as either (a) a primary language subtag or (b) an extended 
language ("extlang") subtag, depending on whether it (b) is or (a) is 
not covered under a macrolanguage code element.  I hope this work will 
end up being of some use to whoever is involved in the RFC 3066ter 
effort.  (I am not making it available at present, so please don't ask 
for it.)

In checking whether the new ISO 639-2 code element "frr" existed in ISO 
639-3, I actually cheated and looked in the RFC 3066ter registry mockup 
instead.  I did not find a primary language subtag "frr", but had I 
looked a little deeper I would have found "frr" as an extlang subtag. 
That is because, as John points out, "frr" (Northern Frisian) is 
actually a member of the ISO 639-3 macrolanguage "fry" (Frisian).

This raises an interesting question.  If "frr" is, in fact, going to be 
an extlang subtag under RFC 3066ter, then it probably doesn't make sense 
to add it as a primary language subtag under RFC 3066bis.  After all, 
there is no provision for reclassifying a subtag from primary language 
to extlang.  I would have to recheck this, but I don't believe there is 
another ISO 639-2 code element that is covered under a macrolanguage in 
639-3.

So if "anp" and "gsw" are to be added to the RFC 3066bis registry 
through the registration process, maybe it would be best NOT to add 
"frr", but rather to wait until the role of that code element in the RFC 
3066ter framework is agreed upon.

> In looking this up, I spotted two serious problems with Frisian...

I have nothing to add to John's very lucid explanation of these issues.

--
Doug Ewell
Fullerton, California, USA
http://users.adelphia.net/~dewell/