On 11/8/05, Andre Engels andreengels@gmail.com wrote:
2005/11/8, SJ 2.718281828@gmail.com:
- tpi (tok pisin, recent conlang, 160 articles),
tpi *is* an official ISO 639-2 code. It is also not a conlang, but a creole - I think you're confusing it with Tokipona, for which the Wikipedia has been closed.
Whoops, it certainly is. Boy do I have papaya on my face... Tokipona, while it lastedas a WM project, did represent another reason some people wish to go outside the ISO.
Apart from that, I am of the opinion that it is good to go by ISO 639-2, but as a default rather than an absolute. That is, we don't always follow ISO 639-2, but for languages on the list we will include unless there is a strong argument being made against, whereas languages not on the list will not be included only if there is a strong argument being made in support. Personally I would like to be slightly more strict - ISO 639-2 for living languages, but dead languages will have to show that there is recent material written in the language as well, rather than just an ISO 639-2 code because of old work.
It should always be possible to make strong arguments for exceptions... though perhaps they should be limited to a well-defined space (perhaps even off the mailing lists :-).
As for ISO 639-3, this is not at all a patch or substitute for 639-2. Each layer of language-codes has a lower bar for what it means to be a [meaningful] language. -1 and -2 are specifically focused on l anguage designations that are useful to written work...
We should make use of the good work ISO is doing to distinguish between these language categories, where we can.
SJ