What does "Code for the representation of names of languages" mean ?

Lang Gérard gerard.lang at insee.fr
Fri Jun 19 17:38:31 CEST 2009

Dear Peter,

I thank you very much for this interesting message.
I certainly do no share the view that "It's time to put the "names of languages" canard to rest once and for all", but maybe this an acceptable view.

Nevertheless, it remains true that:
1-(As you recognize it) this has been the model followed by ISO639 for ISO 639:198, ISO 639-2:1998 and ISO 639-1:2002 until ISO 639/RA-Jac adopted another model (when and why ?)
2-This model is the official one by the official title of the ISO 639 series as well as by the letter of the unchanged official normative text of these international standards

So, at least, it is certainly not a personal "canard" of mine, but an official international "canard" with some  merits that, in my opinion, are not less interesting and useful (inclusively for international information systems) that the alternative model you seem to have in mind with others, that I am considering much less efficient and definite. And this is another far more interesting rationale. In fact and going to the very heart of the matter, I never read any presentation of explanations of the "obvious reasons why, at least for ISO 639, this is not a useful model", but only assertions like yours offering no new (or old) argumentation. And I really would be happy to understand these obvious reasons, and eventually to be convinced so as to change my mind on this question.

That autonyms can in some cases present alternatives or historic variations is nothing that decisively alter the process for choosing a "Code for the representation of names of languages".

Bien cordialement.
Gérard LANG

-----Message d'origine-----
De : Peter Constable [mailto:petercon at microsoft.com] 
Envoyé : vendredi 19 juin 2009 17:04
À : Lang Gérard; John Cowan
Cc : ietf-languages at alvestrand.no
Objet : RE: Valencian Language Tag registration request

From: ietf-languages-bounces at alvestrand.no [mailto:ietf-languages-bounces at alvestrand.no] On Behalf Of Lang Gérard

> Not exactly so, because the title is "Code for the representation of 
> names of languages", so that:
> (i)-the only veritable name of the considered language is the autonym

Or autonym_s_, in many cases. And when there are alternatives in use who is to decide which is to be preferred?

I heard a fascinating presentation by an Indian linguist about ten years ago in which he described rapid changes in linguistic self-identity in India as inferred by looking at census data over successive decades. The labels reported for language in use at home have been seen to change dramatically over two or three decades. The only possible interpretations are that linguistic populations have grown and shrunk at astounding rates or that the autonyms have changed, and only the latter is plausible.

So, do we code all of those different autonyms? That would only be useful if the purpose of coding was to document the autonyms that are or have been in use -- but coding serves no purpose in that recording exercise. On the other hand, if coding is done to provide metadata elements for declaring linguistic attributes of information objects, then coding those different autonyms is decidedly unhelpful.

> (original name inside ISO 639:1988; vernacular form inside ISO 
> 639-2:1998; indigenous form inside ISO 639-1:2002; and should be 
> reference name inside ISO 639-3: 2007)

While autonyms may be preferable, reference name in ISO 639-3 should not be constrained to only autonyms. First, it is simply impracticable logistically to rapidly determine a complete set of autonyms for all of the languages spoken in the world. Secondly, there will be many cases in which the autonym cannot be discovered (e.g., do we know what the Persians of the 5th century BC called their language?), and cases of historic varieties for which the autonyms would be the same yet ICT applications require distinctions (e.g., Modern vs. Middle English).

> , and this is the object that should whose representation should be coded inside ISO 639.

You continue to repeat the same assertions, but the only rationale you ever give is that this was the historical model for ISO 3166 and ISO 639. Even if that is true, others see obvious reasons why, at least for ISO 639, that is not a useful model. Yet, in several years of making your case you have not offered any new argumentation. As John said, 

>> It's time to put the "names of languages" canard to rest once and for all.


More information about the Ietf-languages mailing list