Perception makes sense only as, and is a subjective view of the world from given point of view. Implicitly it posits the possibility of subjects which have different perspectives of the same world.
As such we have situation that in subjective perception there are posited things which transcend the subjective points of view, and transcend the points of view of all possible subjects, (including the possibility of such subjects). Of course that should maybe not be surprising, as anything transcendental, including math, geometry, logic, and so on, as presented by Kant, will be such only as going over the limits of the subjective.
This situation opens the possibility for communication of multiple subjects about the world and about their ideas, in the way we experience it every day… as talking about the same things. That is, it opens the possibility for direct reference through language.
But this doesn’t mean that we can abstract from everything that contributes to this possibility, and think that there is nothing but the language and the reference. Ignoring the base which makes this kind of referencing possible, will create problems when one will try to consider deeper problems of metaphysical or epistemological nature.
Note 1:Of course possibility of illusion is still there, so what looks like another subject to us might not be having perception, might not be “proper” subject, but just behave like one. For example, it might be unconscious AI, as in the Turing Test. But as noted in previous post, possibility of illusion is included in the semantics of perception.