If all our acceptable theories are on similar evidential footings (neither the theory nor the theory language is fixed by the relevant evidence) then the distinction between notationally indeterminate theories and underdetermined theories comes down to just the fact that intertranslation is possible in the former but not the latter.