Chapter 2: Background
2.1.3 Collaboration
Section 2.1.1 introduced the general Shannon-Weaver model of communica- tion, Section 2.1.2 showed how this model can be applied to describe the process of interpersonal communication. This section focuses on collabora- tive interpersonal communication, the type of human-to-human interaction that is at the centre of CSCW research.
Collaboration generally refers to the process of people being engaged in joint, interdependent activities in order to achieve a common goal3. A classi-
3For an in-depth review of different definitions of the term “collaboration”, please refer
cal example for collocated, synchronous collaboration would be a round-table meeting where participants discuss various documents and have to achieve a defined goal (find the best solution/decision for a problem) within a limited time frame. Collaboration, like interpersonal communication, is the process of creating unique shared meaning, where the body of accumulated shared meaning serves as the basis for joint action that leads to solving the task at hand. Figure 2.4 shows a simple model for collaboration, which is a further extension of Shannon-Weaver model of interpersonal communication discussed earlier. Message Channel decode Context Context noise noise noise encode decode decode encode
Shared
Situation
Model
Figure 2.4: A simple model for collaboration.
Collaboration is driven by the exchange of messages. Participants engage in both sender and receiver roles and therefore both encode/send messages and receive/decode messages from each other, including feedback. During the process of collaboration, every participant develops his or her own men- tal Situation Model which contains the available information, possible alter- native solutions, evaluation of alternatives, and so on. The largest part of the collaborative effort for participants is to expand each other’s situation models and maximise overlap. (The overlap represents participants’ shared understanding which is the common ground from where a solution for the task at hand can evolve.)
Grounding: Accumulating common ground is an interactive process and is fundamental to all collective actions (Clark and Brennan 1991). To succeed in grounding a message, that is to make it part of their common ground, it is crucial for participants to ensure that a message has been understood as it was intended. The way in which participants ground their messages depends on the individual situation.
• Verbal grounding: grounding is evident in spoken conversations. As previously mentioned, listeners may signal their understanding in form of positive feedback, also referred to as acknowledgements or backchan- nel responses. For other, more subtle forms of verbal grounding refer to Clark and Brennan (1991).
• Non-verbal grounding: Fussell et al. (2000) collected different types of visual (non-verbal) information and highlighted their relevance for three grounding subtasks. Table 2.1 is an adapted version of their table. It shows the different forms of body language people can use to coordinate each other’s attention and give feedback. Fussell et al. (2000) also considered the often neglected role of shared objects and a shared work context on facilitating grounding.
Type of Visual Information Grounding
Subtasks
Participants’ heads and faces
Participants’ bod- ies and actions
Shared task objects Shared work con- text
Establish joint focus of atten- tion
Eye gaze and head position can be used to signal area of attention
Body positions and activities can be used to signal area of attention Constrain possible foci of attention Constrain possible foci of attention; disambiguate off-task attention Monitor com- prehension Facial expressions and gestures can be used to give feed- back
Approprietness of actions can be used to infer comprehen- sion, or clarify mis- understandings
Change in state of objects can be used to infer comprehen- sion, clarify misun- derstandings Conversational efficiency Gestures can be used to point to task objects Visually shared task objects can be referred to with deixis and pronouns
Environment can help constrain domain of conver- sation
Table 2.1: Benefits of four types of visual information for three grounding subtasks (adapted from Fussell et al. (2000)).
Least collaborative effort: When trying to ground their utterances, par- ticipants strive to encode their messages in the most efficient and appropri- ate way possible. An important general rule that underlies the grounding processes is the principle of the least collaborative effort, according to which participants in conversations always try to minimise their collaborative effort – the work that both do from the initiation of each contribution to its mutual acceptance (Clark and Brennan 1991).
For example, being able to point at a visually shared object during a col- laboration allows for a very easy, brief, and concise way of referencing through a deictic utterance (e.g. “this one”) and therefore helps to reduce collabo- rative effort. It can therefore be expected that participants use indicative gestures (pointing, looking, touching) whenever possible to refer to nearby objects they attend to, as gestures come at a lower collaborative cost than having to describe the objects verbally. However, Clark and Wilkes-Gibbs (1986) point out that in real conversations minimizing effort does not nec- essarily always lead to the formulation of the shortest and most appropriate utterances. Rather, starting with an improper utterance and then repairing it together with the other participant, or splitting up one complicated utter- ance into multiple smaller parts can involve less work than having to plan and encode one flawless message in the first place.