Language codes can be either very coarse, like codes for "English",
"German" etc. or they may be very detailed, like codes for
"American English" or "West Yorkshire English" or "Swiss German"
etc.
In order to decide which kind of language codes are suitable in
mime headers, one should consider to what use such codes may be
made. I can see two obvious uses of such codes:
(1) To filter out messages in languages I cannot read
(2) To direct mail to a translator (human or automated)
based on the input language for the traanslation
(not the output language).
In both these cases, a coarse language code is enough.
So I support the idea of using the ISO standardised
language codes. An additional advantage is that this
makes co-working with X.400 simple, since X.400 uses
the ISO codes.