It's deliberate that we do not use "value representations". There are environments in which the same number can be exactly represented as both a normalized and a non-normalized floating point value, with distinct value representations, and we did not intend to require that they be treated as different.

I agree that the exact rules are unclear, but I think that's not something that can be fixed editorially. (It's also hard to fix this non-editorially since the C++ core language currently has no notion of NaN values, normalization, and other similar things you would need to take into account to specify rules here.)

It seems that C (at least for C23) has sufficient notions. Perhaps we can adopt C wording as DR(s).