I thought IETF used existing standards to the extent possible. There
is an ISO standard for language codes, ISO 6391. Yes, it is the two
letter codes that some of you are complaining about. However these
codes are what is used for specifying locales on many Unix systems so
it is easy for mail to pick them up and put them in the header.
Anything else requires another table somewhere and someone to update
and distribute it.
There are a few other candidates: IBM has a three letter code table,
and there are the Library of Congress USMARC codes (also three
letters), e.g. ace=Achinese. It has about 380 languages. There is
a listing of languages called the Ethnilogue which has about 6000
languages. If all of these were given codes they would exceed a two
letter code. Some people have used telephone company country codes,
but country does not equal language, e.g. Switzerland. I believe DOS
uses these. But it also has a keyboard code which is two letters and
seems to correspond to the ISO standard above.