Yes, we should develop a new theory of language based on an information theory approach rather than a discrete symbol approach, or maybe something involving both.

But you can’t say information theory as it stands gives us a completely satisfactory model of human language. Something will need to be built on top of it.