Milos Rancic millosh at
Wed Jun 16 21:06:18 CEST 2010

On Wed, Jun 16, 2010 at 20:04, John Cowan <cowan at> wrote:
> Milos Rancic scripsit:
>> If I could dream for a little bit, a proper tag of, let's say, Serbian
>> language could be for example:
>> indo_european-slavic-balkan-synthetic-shtokavian-serbian and for
>> Pidgin (a creole language, not a type of languages)
>> mixed-(sino_tibetan-chinese-cantonese-...)-(indo_eurpean-germanic-anglo_frisian-english)
>> or similar. Probably, with some numbers in one repository which would
>> describe closeness inside of the hierarchical group or, better, with
>> more descriptive explanations inside of the repository.  But such
>> notification is not supported by any system.
> ISO 639-6 (which is still in draft, and is not part of the BCP
> 47 evolving standard) is based on exactly such a system.  However,
> explicitly putting such relationships into tags is not only politically
> sensitive, but unstable from a strictly scientific viewpoint.
> Indo-European relationships (except at the topmost level) are pretty
> well accepted, but the same is not true in general around the world.
> Even such a question as "How many top-level language families exist?"
> does not have a definite answer.

Thanks for the information. I was thinking that ISO 639-6 is a system
for tagging all language varieties, not with making links between

I don't care if we would have a descriptive categorization or just
codes connected to the categorization. So, if a ISO 639-6 code for
some language is "wxyz" and the tag correlates with some place inside
of the categorization, it is fine. That would allow creation of new
standards and their proper categorization. It is not happening often,
but it happens from time to time.

> A macrolanguage is a group of language varieties which are treated as
> a single language for some purposes and as multiple languages for
> other purposes.  That definition is intentionally broad so that it
> can cover many particular cases.

Yes, a useful definition from a political point of view. But still not
useful enough to tell to the reader which type of the relation it is.

More information about the Ietf-languages mailing list