ABSTRACT
Spatial expressions are usually considered to reflect one of three possible perspectives: deictic, intrinsic, or extrinsic. I argue that because speakers’ and addressees’ viewpoints don’t always coincide, we need more fine-grained distinctions. I partially describe two psychological experiments on how perspectives are chosen when viewpoints do not coincide. The results show that addressees’ perspectives are indeed chosen often, and also that when people can, they often use spatial expressions that avoid the conflict in viewpoints. People also modify their spatial expressions as a result of their partner’s conversational feedback of understanding. This suggests that the generation of spatial expressions includes a component that considers the addressee’s viewpoint and what the addressee gives evidence of having understood.
