'car' in different ISO 639 parts

[Doug Ewell]

I'm in the process of building a new 4645bis Registry that takes into 
account the ISO 639-5 data, which I assume I have the rights to use.

Generally, when two different parts of ISO 639 apply different names to 
the same code element, we have a policy of keeping all the names in the 
Registry, because it was considered important for traceability.  So we 
will have things like:

Type: language
Subtag: ms
Description: Malay (macrolanguage)
Description: Malay
Added: 2005-10-16
Suppress-Script: Latn

because ISO 639-2 calls it simply "Malay" while ISO 639-3, which has to 
differentiate macrolanguage codes from individual language codes, calls 
it "Malay (macrolanguage)".  This doesn't pose a problem since the 
different names seem to refer to the same linguistic concept.

Now, ISO 639-5 draws a formal distinction between "language groups" and 
"remainder groups," concepts used in (but not formally defined by) ISO 
639-2.  These are identified primarily by the words "languages" or 
"(Other)" in the ISO 639-2 name.  A code element like 'apa' ("Apache 
languages") is considered to cover all Apache languages, whereas a code 
element like 'gem' ("Germanic (Other)") covers only those Germanic 
languages that do not have an ISO 639-2 code element of their own.  Both 
the LTRU WG and ietf-languages are familiar with this concept and have 
discussed it often.

ISO 639-5 apparently redefines the 639-2 remainder groups to be full 
language groups, meaning that it considers 'gem' to cover all Germanic 
languages under 639-5, including those that have their own code, such as 
German.  I've been under the impression that this was the Way of the 
Future for ISO 639, and so it was safe to replace all "(Other)"-type 
Description fields in the Registry with the 639-5 names.  After all, it 
would be internally inconsistent to have something like:

Type: language
Subtag: gem
Description: Germanic languages
Description: Germanic (Other)
Added: 2005-10-16

For 'car' we have a unique problem.  ISO 639-5 defines this code element 
as "Carib languages" and calls it a group (not remainder) code, but ISO 
639-2 and 639-3 both define it as "Galibi Carib".  This was officially 
changed in 639-2 from plain "Carib" on 2006-11-22.  Indeed, 639-3 
identifies this as an Individual language.  This affects not only the 
Description fields in the Registry, but also the Scope field if it is 
ultimately decided to add one.

We need to decide which Description, and by extension which scope, 
should be reflected in the Registry for 'car'.  I ask for the guidance 
of both the LTRU WG and ietf-languages list, since this is not only a 
draft issue but also a registration issue, and for that reason I have 
cross-posted both lists.

--
Doug Ewell  *  Arvada, Colorado, USA  *  RFC 4645  *  UTN #14
http://www.ewellic.org
http://www1.ietf.org/html.charters/ltru-charter.html
http://www.alvestrand.no/mailman/listinfo/ietf-languages  ˆ