This is a lively debate in linguistics, so while it’s not objectively true (i.e. it depends on how you want to use the word), it’s much more justified than you’re implying here. As others have said, it’s about distinguishing the kind of communication humans do from simple animal communication — a common hook for this topic is the fun fact that “chimpanzees can learn sign language, but they’ve never asked a question”. It’s somewhat analogous to how animal communication can be seen as categorically distinct from simple signaling done by flowers, scent markers, etc.

I’m in a bit of a rush but suffice to say that Chomsky is probably the best champion of the “language is compositional” view, or as he puts it, “language is the generation of an infinite range of meaningful outputs from a finite range of inputs”. There’s dozens of great, layperson-friendly talks of his on this topic on YouTube, for anyone who’s curious!