If you read the introduction, it does. In particular, it explains how normalizing for parts of speech (all nouns, verbs, etc) zipf's law does not emerge. This method predicts that phenomenon as parts of speech being a latent variable, which they do not believe other models can do.