ABSTRACT
When trying to understand a verbal description of a spatial configuration, a listener or reader normally goes beyond the information given, and forms an elaborate mental model in which various attributes of the described scene are inferred (Struiksma et al., 2009; see also Taylor and Zwaan, this volume). When parsing a spatial description and forming a mental model, one always needs to choose a reference frame (RF), which denotes the point of view that is taken when perceiving spatial relations (see also Miller and Carlson, this volume). The observer can choose his or her own viewpoint, take that of an object that is part of the scene, or take one that relies on omnipresent cardinal directions or local/distal landmarks. Depending on the frame of reference, the representations and descriptions of spatial relations within and between objects differ. In the spatial literature two particular RFs are generally dissociated: the egocentric RF in which locations of objects are represented with respect to the observer; and an allocentric RF in which locations of objects are represented within an external framework (i.e. configuration of landmarks). In the language literature other taxonomies have been used to denote linguistic RFs, i.e. viewer-centred, object-centred or environment-centred RFs (see Miller and Johnson-Laird, 1976). In this chapter we use the terms intrinsic, relative and absolute RFs as proposed by Levinson (1996, 2003). A relative RF defines the relations among objects with respect to the observer’s position. Note, however, that the relative spatial location of (an) object(s) change(s) when the observer changes his or her position. Consequently, the use of a relative RF may create ambiguity. The intrinsic RF takes the perspective of one of the objects in the spatial scene, labelled the ‘ground object’ and relates the locations of other objects (‘figures’) to it. For this RF the location of the observer is irrelevant and the definition of a spatial relation does not change when the observer moves. Finally, the absolute RF uses a fixed point of view that is located on an invariable axis, such as the cardinal axes. Since the application of this frame to a scene remains unaltered upon the rotation of either the observer or the object(s), it is essential to be aware of its orientation when perceiving or describing a spatial configuration.
