Dear Gérard,
(copy to the JAC email list)
Thank you for your response (below).
Detailed principles for inclusion in the alpha-2 and the alpha-3 code tables of ISO 639 are developed by the ISO 639 JAC. It hasn't been possible to make absolutely "mechanical" rules, and we do sometimes have extensive discussions. However, in the case of these three items there was consensus.
As you will know the entire situation of ISO 639 is undergoing changes. The finalization of ISO 639-3 gives us a new situation. The JAC is currently revising its principles and working methods. The new principles will be publicized on the JAC web site.
It is a strong principle that it isn't in any way "better" for a language to be encoded in the alpha-2 table than just in the alpha-3 table. Alpha-2 and alpha-3 identifiers for the same item shall be considered synonyms for all applications. Future publication of the code tables will make this even more explicit.
The principle that I quote has been decided by the JAC, and it is a very "strong" principle. The reason is that adding additional (alpha-2) identifiers to items that already have ISO 639 alpha-3 identifiers will disrupt encoding and create problems for some users. The IETF has been a very strong advocate for this principle. In the aftermath of the disrupted encoding caused by the change of value for "CS" in ISO 3166, which has in fact given "ISO" bad reputation in the IETF and the "Internet community", the ISO 639 JAC has been extremely clear about our avoidance of changes that can cause disruptions.
As to the two following issues:
Northern Sotho / Pedi is encoded nso. A proposal to add an alpha-2 identifier has been discussed by the JAC earlier, and it was not raised again at this point. I am sorry that this hasn't been conveyed to you before.
Seselwa French Creole is a new item for ISO 639. It will be discussed and decided by the JAC in due time.
The final issue, Serbo-Croatian, has been discussed extensively by the JAC. There is no clear consensus as to how to resolve the issue in the best possible way. The JAC is fully aware that this is in conflict with the principle that the alpha-2 code table is a subset of the alpha-3 table (as far as included items is concerned). The JAC will find a way to resolve this conflict, but this may have a wait until 639-3 has been finalized.
Best regards,
Håvard Hjulstad
mailto:[log in to unmask]

From: Lang Gérard [mailto:[log in to unmask]]
Sent: Tuesday, April 18, 2006 5:22 PM
To: Håvard Hjulstad
Cc: Lang Gérard
Subject: RE: ISO 639 proposals rejected: Tok Pisin, Tetum, and Palauan

Dear  Havard,
I have been very disappointed by the decisions of the ISO 639 Joint Advisory Committee concerning my proposal for identifiers
tp (tok pisin), tu (Tetum) and pu (Palauan). In fact, I found no trace of the principle you give in your message in the text of ISO 639-1 or of ISO 639-2 (and I do not even understand what could be the reason of such a principle).
I am also somewhat surprised not to know what happened with ns (Pedi or Northern sotho) and sy (Seselwa french creole).
And finally, could you also tell me what happened to my proposal about the alpha-3 code element shc for serbo-croatian, that is only the realisation of a commandment contained in the introduction of ISO 639-1 (2002) "The languages listed in ISO 639-1 are a subset of the languages coded in ISO 639-2" ana also in the introduction of ISO 639-2 (1998) "This part of ISO 639 represents all languages contained in ISO 639-1 and in addition..."; and so it should be imperative to ISO 639 Joint Advisory Committee to comply with this commandment. 
Could you please tell me what was decided ?
Trés cordialement.
Gérard LANG
-----Message d'origine-----
De : Håvard Hjulstad [mailto:[log in to unmask]]
Envoyé : dimanche 26 mars 2006 12:36
À : [log in to unmask]
Cc : [log in to unmask]
Objet : ISO 639 proposals rejected: Tok Pisin, Tetum, and Palauan

The ISO 639 Joint Advisory Committee has rejected a proposal to assign alpha-2 language identifiers to the following three items:
Tok Pisin (alpha-3 identifier in ISO 639-2: "tpi" remains the only ISO 639 language identifier).
Tetum (alpha-3 identifier in ISO 639-2: "tet" remains the only ISO 639 language identifier).
Palauan (alpha-3 identifier in ISO 639-2: "pau" remains the only ISO 639 language identifier).
The ISO 639 JAC considered that the fact that these languages have official status in their countries not to be sufficient reason to deviate from the principle that alpha-2 identifiers should not be assigned to items that already have alpha-3 identifiers.
Best regards,
Håvard Hjulstad
(secretary ISO 639 JAC)
Håvard Hjulstad
  Standard Norge / Standards Norway
  direkte tel / direct tel: (+47) 67838645
  mailto:[log in to unmask]