The Role of Users' Concepts of the Robot in Human-Robot Spatial Instruction

Spatial instructions are always delivered for a particular communication partner. In this paper I investigate the the role of users' concepts of their communication partner in human­robot interaction by analysing the spatial language choices speakers make in three comparable corpora with three different robots. I show that the users' concepts of their artificial com­ munication partner is only mildly shaped by the appearance of the robot, and thus that users do not mindlessly use all clues they can get about their communication partner in order to formulate their spatial instructions. Instead, spatial instruction in human­robot interaction also depends on the users' models of the communication situation, as well as on external variables, such as gender. The problem I address in this paper is the role of speakers' concepts of their communication partner in verbal spatial instruction. Speakers have a range of different ways of referring to spatial configurations at their disposal, and in a particular situation they have to decide for one variant over another. Among many other possibilities, speakers may refer to the spatial scenario in the following figure in different ways, for instance: the square on the right (from your point of view)