> Notably, this design doesn't _prevent_ anyone changing hyphenation > language when they switch environments -- it just doesn't do it > automatically. That is, it covers the common case, but gives the user > flexibility to handle other cases. With the scope of both hyphenation > mode and hyphenation language being documented (the latter less > explicitly, but consistent with how other global parameters are > documented), the user knows what to expect.
Indeed. There's nothing really wrong with the way it is now, and arguments can even be made for keeping it this way. (E.g., you're setting text in two languages and you have an environment for footnotes, then you don't want that environment to be associated with either language, but instead you want to be able to switch languages at will in both the footnotes as well as in the body text.) Given that so many attributes (for example, the hyphenation mode) are in fact associated with the environment, I just found it surprising that this isn't, (and the info file doesn't mention it).
