I recently typeset a book with a fair number of medical terms. I can say from that experience that it would be impractical to switch \languages in any way to get different hyphenation on a word-by-word basis. It was not uncommon to have several words in the same sentence that needed hyphenation exceptions; the source would have quickly become unmaintainable, even with a one-character command to switch the language. And my book was for a layperson audience; it would have been even worse for a professional medical-chemical-whatever text.
Thus, what would have benefitted me is not a separate \language, but a (lengthy) exception list that I could \input. (That's what I ended up creating for myself, by hand, with Barbara's help.) For something of general use, there are so many chemical-medical-technical terms that I think it would be necessary to create that exception list automatically, somehow. I guess by generating "chemical" patterns and then applying them to a long word list. I wouldn't worry about the 64K exception limit until we get that far. Since the hyphenation of these terms is language-dependent (not surprisingly), it would have to be generated for each. At least as far as English goes, though, I suspect one list could suffice to cover all the variants of English (US, UK, CA, AU, ...). My $.02. --best, karl.
