Much of what we know and say refers to the dynamics of our world. We have a large class of linguistic objects -- verbs -- devoted entirely to expressing dynamics. In this paper we present a dynamical representation sufficiently general and expressive to capture both the meaning of individual verbs and the more abstract commonality of sets of semantically related verbs. Our purpose is to develop a complete theory of language acquisition that can be implemented on a physical platform, such as a mobile robot. The representation of the semantics of a word or set of words is constructed incrementally based on what is happening in the agent’s environment when the word or a member of the set of words is heard.