I hesitate to mix into this conversation, because I have only followed it
intermittantly, and the discussion seems overly complicated. But I have a
couple of comments.
*Audience Languages.* The distinction between the "audience" languages and
the "document" languages is seems tenuous and artificial. I'm guessing that
the best characterization of "audience languages" is that someone who
doesn't speak one of "audience" languages would not find the document as a
whole to be understandable. For example, I could have a document that is
mostly English with a some Hebrew phrases mentioned. While both English and
Hebrew occur in the document, it would not be useful for a non-English
speaker, while it could be useful for an English speaker who didn't know
Hebrew.
*Language vs Languages. *It is also odd to talk about "the" language of a
document as if there can be only one. Even speaking of "the predominant
language" is a misnomer: look at
http://unicode.org/iso15924/standard/index.html, for example. While we can't
make a syntactic change for compatibility reasons, there should at least be
an explanation of that it is just a syntactic pidgeonhole that people have
to deal with.
Mark