It’s true – Feynman is addressing problems in language, or problems in the way we use language to convey scientific models and understanding. By focusing on the problems, though, he isn’t discounting those fields where it blooms. In fact, he’s quite adept with the tool.
What I remember most is how he emphasizes the difference between what most people mean by “understanding” and what science means by “understanding”.
There was an article I read recently about a theory that language itself is based on metaphor at a fundamental level; memory is foggy tonight and I can’t find it with searches, though it’s out there in the zeitgeist.
… all this assuming “simile”, “metaphor”, and “analogy” as synonymous.