Perhaps you are right.
My proposal is based on the assumption that VRML needs to support text
natively , and as you rightly note, this may, or may not be, the case,
though I think *some* native support for multilingual text will be
necessary.
I imagine that what will eventually evolve is a hybrid, where native
text will be used for things like neon signs and whatnot, and that
actual textual information will be delivered via SGML (possibly within
a VRML-defined object). As such, I do not think we need to define
overly much for the native text support, which is why my notes do not
present overly much detail (and thinking about what is necessary took
more time than writing the notes themselves).