I have to say that I think this is probably overly provincial or
US-centered. In Europe and in Israel, at least, my experience has been
that people tend to be multilingual and switch rapidly back and forth
between multiple languages in many social and work contexts. I see no
...
Nathaniel,
I may have not stated my point clearly enough, but my experience is
identical to yours and I still believe that most email messages are
short, address a single issue, and do so in one language. I'm not
suggesting that of a single terminal or a single day's work: three
separate messages might easily use as many languages. And longer and
more structured materials may, and often do, contain insertions or
sections in languages different from whatever language they started in.
I'd suggest that more paper intra-office notes have these same
properties, and we have a lot of experience with them. There is
a range of exceptions, analogous to commenting on a note in language B
with a cover sheet in language A, but, for MIME purposes, we would
handle those as separate body parts so it would still be mostly one
language per body part.
Note that neither the question or the answer are especially
important except to deciding whether per-body-part language tagging is
of any use. If lots of messages use different languages interspersed
within a body part, and high-precision language identification is
needed, then we can stop worrying about "character set" as a driving
concept and start worrying about a language markup model that will, or
will not, force every message into an enhanced richtext.
Since I think "every message, no matter how trivial in enhanced
richtext" is not going to fly in the marketplace, I'm looking for a way
out that covers some rational number of cases.
john