Once again we have had the "tempest in a teacup" discussion about ISO 639
and its development. Sometimes the discussion is fun, sometimes it is just

The ISO 639 Registration Authorities' Joint Advisory Committee (JAC) does
NOT have plans to alter the rule that states that no alpha-2 language
identifiers will be added if an alpha-3 identifier already exists. For some
of our users that rule is of great importance; although it is a problem for
other users. In the lifetime of ISO 639 (first edition published 1988) uses
and users have changed. There is no reason to expect that the next 14 years
will involve fewer changes. Decisions were made in 1988 that are overruled
by decisions today. Decisions will be made 14 years from now that will
overrule today's decisions. It doesn't take a prophet to predict that; and
it makes no difference how much Michael objects.

No, Michael, I am not going to propose to change JAC's policy now. I have
trust in the committee to make the right decisions. We recently had a
proposal to add an alpha-2 identifier for one language that already has an
alpha-3 identifier. The proposer submitted some information/opinions in
support of the proposal. As acting chairman of the JAC I found it consistent
with a proper procedure to circulate the proposal to the committee. The
feedback from the committee was clear: No, this is not sufficient to change
our "freeze" rule. So the process worked; there is nothing to worry about.

But how can I say that 13 years from now, when technology has changed, the
JAC has changed, the IETF language tag reviewer has changed, that there
still is no reason to change alpha-2 registration rules?

Right now there seems to be a problem for some of ISO 639 users that they
need rules to avoid synonyms. If one language has already been tagged "de",
it causes a problem if the same language in another document is tagged "deu"
or "ger". Other users don't have this problem. Their systems allow synonyms.
Even "near synonyms" may be built into the system, allowing e.g. "no",
"nor", "nb", "nob", "nn", "nno" (Norwegian in various forms) to be used

Other users have the problem that their system allows only alpha-2
identifiers for reason of space.

I hope that the "freeze" discussions give some new users the understanding
that it might be a good idea to build into their systems features like
"synonyms", "near synonyms", and "hierarchies". And at the same time I hope
that no new implementations will allow alpha-2 language identifiers only.
Language identification will need variable-length strings.

ISO/TC37 plans to continue its work with ISO 639, language identifiers, and
language identification mechanisms. We will listen to the needs of all
users, and we will seek cooperation with all relevant expertise.

Håvard Hjulstad    mailto:[log in to unmask]