We tend to think that we can only know things if we are able to express them in language. E.g., if I can articulate that the word “square” refers to the shape with four equal sides and each corner of those sides meets at a 90º angle, then I have “knowledge” of that fact. But aren’t there ways of knowing that are inexpressible by or through language? By that I mean, aren’t there things that I know to be true that (even if I cannot empirically verify this fact) I am incapable of saying with words: the feeling I know I’ll feel when I see my hometown again, or the particular melody of a song, or the difference between the way the air feels in summer and winter.
So, then, if language does not comprise our ability to “know” something, what does? Perhaps in some cases it functions that way, but in other ways, language is not only inadequate, but incapable of capturing the various ways in which we know things. And this is probably because language, despite what we have been taught, does not function solely as a representative or referential system — it can do that (poorly, in many cases), but that is not how we use language. Language, instead, is functional, contingent on the context in which we are using it.
Thus, our attempts at objective philosophizing about reality, metaphysics, ethics, and so on, are doomed to failure. We can only do so with language, and language doesn’t do that really well. It cannot name concepts in with pure, one-to-one accuracy. And if that’s the case, then knowledge that we have articulated via language is necessarily contingent — in other words, our knowledge is relative to something (a community, a tradition, a social context).