4.1 Grounding mechanisms are use to built shared sub-spaces

4.1.5 Grounding Representations

The information put on the whiteboard contains essentially facts (see 4.1.1) and - to a minor extent - inferences (see 4.1.2). However, one aspect which we did not address so far are the different uses of non-verbal codes (colours, positions, marks,...). This information is important to ground because it is persistent and misinformation could lead to misinterpreting facts or inferences.

The cost of grounding varies according to the medium. These codes can be grounded in the whiteboard itself: if A uses some colour codes, and if B continues to draw objects and uses A's code in a consistent way, A can infer that B has inferred the correct code. Sometimes, the code is simply not grounded, probably because it is conventional. for example crossing out suspects to indicate elimination from consideration. Sometimes, grounding was explicit, either by a legend (example 18) or by a dialogue (example 19). It seems that code negotiation was easier in spoken conversation - especially pair 1: "Why did you change colour?"; "You write the time in hours... until 24"; Do you want the same colour as him". However, we did not study these differences systematically.


Grounding in Multi-modal Task-Oriented Collaboration - 3 SEP 1996
Generated with Harlequin WebMaker