ID for language-invariant strings

Doug Ewell doug at
Sat Mar 15 20:18:23 CET 2008

Peter Constable <petercon at microsoft dot com> wrote:

> It does seem a tad ironic for this list that a suggestion for a new 
> subtag in the registry is considered creeping featurism.

This isn't just a request for a new subtag in the Registry; as 
originally stated by Peter on Thursday, it's a request for a new code 
element in ISO 639.  And that decision has to be up to the ISO 639 JAC.

If the goal is simply to have a new subtag in BCP 47, and if ISO 639 
decides not to add the code element, then we could solve this problem 
cleanly by registering a new language subtag 'invariant'.  I believe 
that would address everyone's needs and concerns.  I can't think of a 
much better use case for an individually registered language subtag.

> Several years ago, I think when work on ISO 639-3 was just getting 
> going, Asmus Freytag advised me that it was important to get the right 
> set, and an adequate set, of special-case concepts encoded.

If Karen Broome is correct, that 'zxx' was originally presented as "not 
applicable" but was changed to "no linguistic content" before encoding, 
then that is an unfortunate narrowing of the intended semantic that has 
contributed to this problem.

Although I can see many sides to this, I'm still not convinced that "no 
linguistic content" is wholly inappropriate for the stated need. 
Language-invariant identifier strings are *based on* linguistic content, 
but they are not themselves intended to be linguistic content, as 
evidenced by their invariance.  But If Peter and others believe the 
'zxx' solution is inferior to creating a new ISO 639 code element, I 
won't argue the point further, except to say that anything other than:

* a new ISO 639 code element
* 'invariant'
* 'zxx'
* 'und'

would be flat wrong, and should not be discussed further.

Doug Ewell  *  Fullerton, California, USA  *  RFC 4645  *  UTN #14  ˆ 

More information about the Ietf-languages mailing list