We can use language to achieve goals by performing speech acts. Speech acts may be factored into propositional content and illocutionary force. “Put the cup on the table” and “Is the cup on the table?” share the same propositional content but differ in illocutionary force. The connection between propositional content of language and the world provides the foundation for language use. Our long term goal is to develop a computational framework that explains how propositional content of language is about the physical world, and further, to understand how speech acts may be used together with physical acts to achieve goals. Two motivations drive our effort. First, such a framework could guide the development of embodied, situated, spoken language systems. Second, the framework could help us reflect on the nature of language, cognition, and embodiment in humans. Over the past few years, my group has built a series of systems that connect language processes (learning, understanding, generation) to perceptual and motor control processes (Roy 1999; Roy and Pentland 2002; Roy 2002; Roy et al. 2002; Gorniak and Roy protectsubmitted to Journal of Artificial Intelligence Research; Hsiao, Mavridis, and Roy 2003). In this abstract, I summarize our view of what it means to “ground” language in the world which has emerged from building these systems. A sketch of our evolving cognitive architecture that supports language grounding is also included and will be expanded upon in the workshop presentation.