Re: Official ISO 3166 country codes online

From: peter_constable@sil.org
Date: Fri Dec 03 1999 - 04:01:30 EST


       Michael E wrote:
>There are too many Sign Languages to expect 639-2 codes, and
       it is convenient to describe them as "sgn". The Deaf community
       approved this and the 639-2 RA is likely to. It is necessary to
       differentiate between them, however, and, since in almost all
       cases the difference is identically coterminous with ISO 3166
       entities, it is extremely sensible to describe
       them as presented in
       http://www.indigo.ie/egt/standards/iso639/sign-language.html.
       Where a country has more than one Sign Language in it
       (Australia, Belgium, Canada, US, etc.), ISO 3166-2 entities
       were used. No problems, little if any ambiguity, and everyone
       is happy.

       Michael, I'm not trying to add to your aggravation. I didn't
       see the postings that apparently got to you; I'd blather on
       knowing I'm at risk of repeating the same offensive line. I'm
       continuing on in the same vein as in another response I just
       wrote.

       I'm not involved with development of ISO 639 and so can't
       comment on what is or isn't appropriate for that standard. Your
       comments here bring two things to my mind, however:

       First, you say that the Deaf community approved the particular
       proposal. What was the process? The document referenced above
       lists over 100 putative signed languages. Did representatives
       of all of these sub-communities give their approval? Did the
       Namibian and Penang communities really give their input? If
       not, how can we be sure that the lesser-used languages are
       getting the kind of support their user community really feels
       they need (a matter you're usually quite concerned about)?
       (Again, I don't know if perhaps, for the purposes intended for
       ISO 639, what is proposed really is adequate. I represent a
       group of researchers who certainly would be better served by a
       single language tagging system that covers *all* languages
       rather than a composite language/country/subdivision tag,
       however.)

       Secondly, I wonder how I'm to make this scheme work if I apply
       it to other situations:

       There is an ISO639-2 code for Cree - cre - but this represents
       (according to the Ethnologue) 6 languages, 4 of which are
       spoken in the province of Quebec.

       There is a code for Chamic - cmc - representng 10 languages,
       several of which are spoken in Thuan Hai province and other
       provinces of Vietnam.

       There is a code for Karen - kar - representing 21 languages,
       with several spoken in Kayah state, several spoken in Shan
       state, several spoken in Mae Hong Song.

       There is a code for Zapotec - zap; there are 64 Zapotecan
       languages, most of them spoken in the state of Oaxaca. There
       are 173 Oto-Manguean languages, and except for Otomian (oto -
       11 languages), the closest code available to represent them is
       zap (except for mis - Miscellaneous). Most of these languages
       are spoken in a small number of states of Southern Mexico.

       There are 540 Trans-New Guinea languages. I really have no idea
       if *any* are represented in ISO 639-2, but I know it's at most
       a few, and most likely none. The vast majority (perhaps all - I
       haven't checked) are spoken in Papua New Guinea or Irian Jaya.
       That's a lot of languages for a small number of
       political/administrative divisions.

       There is a code for Bantu (Other) - bnt; well, I'm not sure
       exactly which languages this includes and excludes, but there
       are 646 Bantoid languages, and (going up a level of
       classification, which is reasonable since I doubt there are ISO
       codes for many of these other languages) 895 Benue-Congo
       language, and (again, going up a level, which is probably
       appropriate) 1272 Volta-Congo languages, and 1436 Niger-Congo
       languages. (There is a different code - nic - for
       Niger-Kordofanian, which overlaps here. So maybe bnt should
       just be the 646 Bantoid languages that don't have their own
       code, and nic should cover the other 700+.) I don't know
       exactly how many countries and divisions are involved, but I
       know it's a lot less than 1436.

       I think you get my drift. You've come up with a system that may
       work for speakers of various signed languages. What are the
       speakers of Atikamekw, Chru, Padaung, Isthmus Zapotec, Angal
       Heneng and Nyemba to do?

       (I had originally included Australian - aus - in my examples;
       257 languages, among them Kala Lagaw Ya, with reported 3,000 to
       4,000 fluent speakers circa 1990. Most Australian languages are
       already or are nearly extinct. Only a handful appear to have
       populations in the thousands, and not many more have
       populations in the hundres. It's depressing to read entry after
       entry reporting populations of "5 (?)", "3 (?), "1 (?)".
       Language tagging may not be an issue for Australian languages
       if there are only a few of them that are still viable.)

       If your going to propose something for tagging text to
       distinguish signed languages, it seems to me you really should
       adopt a system that can be extended to other lesser used
       languages of the world as well. The proposed system combining
       an alphabetic-3 code from ISO 639-2 with alpha-2 3166-1 code
       and an alpha-3 3166-2 code fails for simple cases such as Cree
       and is grossly inadequate for more challenging cases, such as
       Bantu, Niger-Kordofanian and Trans-New Guinea.

       I don't mean to make things difficult, but these things are
       difficult!

       Regards,
       Peter



This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:56 EDT