A system for processing text captured from rendered documents is described. The system receives a sequence of one or more words optically or acoustically captured from a rendered document by a user. The system identifies among words of the sequence a word with which an action has been associated. The system then performs the associated action with respect to the user.