The software annotates text with 41 broad semantic categories (Wordnet supersenses) for both nouns and verbs; i.e., it performs both sense disambiguation and named-entity recognition. The tagger implements a discriminatively-trained Hidden Markov Model.