So the rule for recognizing identifiers needs to be changed, not just the class of certain characters.

I don't know what the motivation was. Contrast with my own proposal concerning identifier names, which is to allow the ® character but only for use by the implementation, to make names that don't clash with the user's program. It's not used for anything else, so doesn't not hurt the parser. Whether implementations do that or not, it's great for documenting the semantics.

—John