Lexicography is a greatly underrated branch of linguistics, yet from the information-theoretic viewpoint it is evident that the lexicon of a language contains at least a thousand times more information than its grammar (megabytes instead of kilobytes). In fact, there are significant theories of syntax, categorial grammar and type-logical grammar in particular, where the generation and analysis of sentences is entirely driven by the information stored in the lexicon. In this talk, we discuss how the information content of various forms of linguistic knowledge can be computed and outline the data structures best suited for storing lexical information, with special emphasis on lexical semantics.

Slides and video at https://nessie.ilab.sztaki.hu/~kornai/2024/Seminar/