Computational metrics like sentence surprisal and relevance predict human sentence comprehension effectively across languages.
