In the DenK project a multimodal interface is developed where natural language is combined with graphical interaction. For the design of this interface, knowledge is collected about how humans refer to objects in a task-oriented environment, by means of natural language and gestures. In this paper we report results of an experiment concerning referring behaviour in tasd-oriented keyboard dialogues. The results are compared with those of an earlier experiment we performed with spoken dialogues. The differences were all found to be related to the so-called principle of minimal cooperative total effort, which says that, within the limitations of the available modalities, the participants aim at spending as little total effort as possible on referring to a certain object on the other hand, and on identifying the object on the other hand. Based on the results, we formulate recommendations for the design of multimodal interfaces which include typed natural language.
|Title of host publication||Multimodal human-computer communication: systems, techniques and experiments.|
|Editors||H. Bunt, R.J. Beun, T. Borghuis|
|Place of Publication||Berlin|
|Publication status||Published - 1998|
|Name||Lecture notes in computer-science|