Talking about places, spaces and faces.
Recently, the issue of how language is grounded has recently been the focus of much attention (e.g., Barsalou, 1998; Glenberg, 1997; Harnad, 1990). In this talk I explore the relationship between language and perception, beginning with a focus on spatial language. A series of experimental studies is outlined, which shows that the comprehension and production of spatial prepositions is underdetermined by geometric relations, and is influenced by kinematic-dynamic relations between the objects being described. These results suggest that flexible interactions between language and perception are required in order to establish the situation specific meaning of spatial expressions. A functional geometric framework for the meaning of spatial prepositions is presented (Coventry and Garrod, 2004), which exhibits such flexibility. Furthermore, in comparison with computational models which focus on processing of geometric relations for spatial prepositions (Logan and Sadler, 1996; Regier, 1996; Regier and Carlson, 2001), a new computational model for spatial language comprehension will be outlined which computes how objects are functioning in context as well as where they are in relation to each other. Following on from this model, differences between languages in how they carve up space are considered, and the use of space as structuring domain for describing emotional states (such as facial expressions) is explored.