I share that perspective about the correspondence theory of truth, and the pragmatic basis of our knowledge. At least, that's how I'd frame the skepticism.

In the article I make the same point you do about language, in the section on the hierarchy of abstractions, where I say that our concepts, too, are models that simplify, and that our artificial languages consist of even more abstract symbols. Math is thus paradoxically both rigorous in its applications and anti-realistic.