How many subtags is ideal?

Doug Ewell dewell at
Sat Feb 5 20:50:57 CET 2005

JFC (Jefsey) Morfin <jefsey at jefsey dot com> wrote:

> your point is well taken. We had an extensive debate on this at the
> occasion of the "RFC 3066bis" proposed Draft. The lang3tag does not
> obviously support all the flexibility needed to register all the
> languages variations,

John was talking about regular spelling variations used by groups of
speakers thousands of kilometers apart.

> the first one being necessary to the very Internet operations
> is the support of the DNS IDN Tables,

The DNS tables require a language tag to identify the language with
which a given DNS is associated.  They have no requirement for tags with
exhaustively detailed variations.  Even John's example of "East Canadian
Inuktitut" vs. "West Canadian Inuktitut" probably would not need
separate consideration in the IDN tables.

> that should be the default for the 3 elements tags as they are the
> only generic (IANA existing) registration on the matter by:
> - the national community trustee (defined by the RF 1591 rules using
> the ISO 3166) codes

You keep mentioning RFC 1591 in this context.  I have read RFC 1591 over
and over, and it makes NO mention of establishing a special profile of
usage guidelines for ISO 3166 codes.  All it says is that two-letter
ccTLDs are based on ISO 3166, because ISO has a procedure for
determining what is and is not a "country."  If you see something else
in RFC 1591 about these "rules," please share it with us.

Country codes used in language tags are not related to country codes
used as ccTLDs.  They serve entirely different purposes.

> The proposition is to use two additional elements:
> - the style in which the language is to be used. Default would be IDN

IDN is by no means the only user of language tags, nor even the most
important user.

> - the author(ity) of reference having registered the language
> description. By default the ccTLD manager of the country code. As
> already registered in the IANA base.

ccLTD managers are not linguistic experts, or experts on coding systems
in general, and certainly are not suitable "authorities of reference"
for languages.

> These lang5tags give probably the whole flexibility necessary to go
> down to full vernacular support (including individual/cultural
> variations). In this perspective the current requests should be
> understood as including their requester name as authoritative element
> (the one who compiles the sources and documents the dictionary, etc.
> for web services interinteligibility for example).

Individual variations.  Let's wait until the "several meetings" you
mentioned, and find out at that point how much support there is for
tagging "Russian as spoken by Boris" differently from "Russian as spoken
by Natasha."

-Doug Ewell
 Fullerton, California

More information about the Ietf-languages mailing list