Perceptual systems often include sensors from several modalities. However, existing robots do not yet sufficiently discover patterns that are spread over the flow of multimodal data they receive. In this paper we present a framework that learns a dictionary of words from full spoken utterances, together with a set of gestures from human demonstrations and the semantic connection between words and gestures. We explain how to use a nonnegative matrix factorization algorithm to learn a dictionary of components that represent meaningful elements present in the multimodal perception, without providing the system with a symbolic representation of the semantics. We illustrate this framework by showing how a learner discovers word-like components from observation of gestures made by a human together with spoken descriptions of the gestures, and how it captures the semantic association between the two.