Deixis and points of view in media spaces: An empirical gesture

Claims are being made that videophone facilities on microcomputers allow transparent and effective use of shared work spaces by geographically separated colleagues. This can only be true if the video image helps the users understand each other's everyday speech. This study examines factors affecting the comprehension of deictic, gestural reference in videophone communication. Three camera positions were compared: the standard, 'face-to-face' view of the colleague, a 'reversed' view, and a rear three-quarters 'hind' view. Task conditions involved low referential ambiguity (where reference was verbally explicit as well as deictically indicated by gesture) and high referential ambiguity (deictic reference alone). The reference was either to an item in the workspace or a spatial relationship, and two-dimensional and three-dimensional workspaces were compared. Results indicate that the standard face-to-face view found on many systems does not allow gestures made towards shared areas of the screen to be understood when the verbal information is ambiguous. In designing systems that encourage the use of normal patterns of speech, and hence the use of deictic reference, it is necessary to understand which cues are likely to resolve ambiguities, in which dimensions, and the extent to which the cues provided are likely to achieve that end.