While studying the language of Shakespeare in my first year of postgraduate study at the University of Cambridge, I envisioned "The Syntagmatic Browser." Drawing from the theories of Saussure, this new kind of thesaurus visualisation would create associations among the contexts of words rather than the words themselves.
Although I developed a prototype in GZigZag, I never built the software. I really ought to someday. It would be a fantastic utility for lexicographers and literary scholars.
It wasn't until much later that I came across Statistical Natural Language Processing in my roles at Texperts, Dressipi, and TouchType.