Criteria for languages?

Peter Constable petercon at
Thu Dec 3 10:15:09 CET 2009

I am talking about macrolanguages, given the full context of their usage in applications of ISO 639, of which I see BCP 47 as being of particular importance. 

It was not the original intent to introduce macrolanguage as a new concept in ISO 639; we started with a very different principle in mind that would have applied to every macrolanguage in 639 with quite different results: e.g. "zh"/"zho" would have been documented as denoting precisely Mandarin, period. It was realities of pre-existing usage that forced us to abandon that approach -- a reluctant but necessary decision.

So, I mean the whole ball of wax, not just extlangs. Even apart from use of extlangs, the very existence of "lvs" and "lv" as alternate codings for Latvian (or "ar" and "arb" for Arabic, ...) creates more inconvenience than it does convenience, IMO.


-----Original Message-----
From: ietf-languages-bounces at [mailto:ietf-languages-bounces at] On Behalf Of Doug Ewell
Sent: Wednesday, December 02, 2009 8:45 PM
To: ietf-languages at
Subject: Re: Criteria for languages?

Peter Constable <petercon at microsoft dot com> wrote:

> IMO, macrolanguages are generally not desirable. Case in point: the 
> messiness they can create led us to debate for months/years over 
> Chinese, and in the end we barely arrived at a compromise that left 
> nobody thrilled (and it remains to be see to what extent interop is 
> less than all we might like to see).

I assume you're talking about the BCP 47 extlang mechanism, not the ISO
639-3 macrolanguage mechanism, which is not our doing.

The years-long debate was over whether or not to create the extlang mechanism, and if so, with what rules.  We ended up creating it, with the rule that any extlang must be mirrored as a primary language subtag, which is preferred over the extlang.

I don't recall that much of the debate was over the desirability or undesirability of adding to the original list of seven primary languages (including 'sgn') that can have extlangs.  If we had wanted to prohibit that, we would have swiftly added a sentence saying so.

> With that in mind, I don?t as much concern about a lack of 
> consistently-applied principles; the principle I?d hope for is to 
> avoid them if possible, and consider on a case-by-case basis if there 
> is legacy that in some way prevents us from doing so.

Is this another instance of creating a mechanism and then saying "don't use this"?  We were never "prevented" from avoiding the extlang mechanism for Chinese and Arabic languages either.  We applied it because we thought it would provide the best tagging options.

Doug Ewell  |  Thornton, Colorado, USA  |
RFC 5645, 4645, UTN #14  |  ietf-languages @ ­

Ietf-languages mailing list
Ietf-languages at

More information about the Ietf-languages mailing list