To appreciate this dense and detailed paper, the reader will definitely need a computational linguistics background. The authors extend Levin’s work [1] on verb recognition. Some verbs are not ambiguous, but some are. They apply a relatively simple verb recognition strategy, using a statistical model instead of maintaining a large list in a database. This significantly speeds things up. It also reveals how prior attributes can contribute to meaning. This paper is best absorbed by specialists in the area. It is far more complex than an introduction.