I once fancied a girl who was doing postgrad work in linguistics, and one time she made a remark like "I love lambda calculus!" And I was like, really? That's a programming/CS thing, how does it apply in your field? Turns out LC is used as a representation to normalize semantics in linguistics.

It would be interesting to see a Scheme-like underpinning to the semantics of any language -- maybe not to be used for communication in its own right, but to achieve things like more intelligent translation, or NLP machine-learning applications that extract meaning from a text. I don't see much interest in something like this emerging, however, with the current trend in AI being "throw more statistics at the problem".