2.2 Work practices around tables
2.2.2 Communication
Collaboration around tables is different from collaboration at other devices, mainly because of the unrestricted, face-to-face style of work that tables afford. This style of collaboration leaves open a space for direct human interaction and a greater reliance on verbal and non verbal communi- cation channels. To understand how people collaborate, it is thus necessary to observe not only the actions that are directly related to the accomplishment of a task, but also hidden aspects of communication, or even the absence of actions, that might have contributed (Hollan et al., 2000). My observations included conversation, body position, gaze and gestures.
2.2.2.1 Conversation
Inevitably, conversation was the principal means of collaboration. Notably, most of the conversa- tion took place between two participants in each group who undertook the majority of the work. Following Morris and Winograd (2004), it is possible to categorize the different types of conversa- tion in collaborative tasks according to whether they are related to the details of the task, planning of the task, or management of the resources at hand. Talk that is related to annotating or summa- rizing (details of the task) was the main component of conversation, though at the beginning of each stage participants spent some time discussing how to proceed. In stage 2, the initial discus- sion was about who was going to do the actual annotation, and for stage 3, it was about who was going to dictate and who was to do the transcription. Because of the small group size, the small number of shared resources, and the size of the table used, no conflicts or explicit coordination effort were expected or observed. Access was simply coordinated through the usual “standards of polite behaviour” (Morris et al., 2004).
2.2.2.2 Body position
Posture at the table conveyed implicit information about the coupling styles of participants. Dur- ing stage 1, while working individually, all participants at some stage adopted an up-right posture, and leaned slightly on the table (Figure 2.2). During collaboration, all participants leaned more significantly toward the centre of the table (Figure 2.3). The contrasting position of leaning back- ward indicated task completion, or disengagement from the task (Figure 2.6). Interesting postures were observed while handling the public document when placed in the public space (Figure 2.7) during attempts to increase the awareness of others of the type and location of action.
2.2.2.3 Gaze and gesture
Three of the functions of gaze that are identified by Knapp and Daly (2002) were observed during the collaborative stages: (1) information gathering; (2) looking at the face of another person to establish an obligation to interact or to signal turn-taking (regulating the flow of communication);
Figure 2.6: Leaning backward: idle-state (stage 1: finished annotating the local copy)
Figure 2.7: Special handling of the public document (stage 2: annotating the public document)
and (3) looking at others after making a suggestion in an anticipation of a non-verbal reply (mon- itoring feedback). Deictic gestures (i.e. pointing) played a major role in promoting awareness and coordination. Participants pointed to the public copy, to their local copies, and to a lesser extent, to the local copies of other participants to attract focus to a certain page or a certain position on a page (Figure 2.5). Certain actions were exaggerated to draw attention to the participant’s status or change of status. For example, aligning pages and/or capping the pen in a particularly demonstra-
tive manner to declare a completion of a task (Figure 2.8), and tapping on the table with the pen or fingers to indicate involvement in the task. Gestures and actions were also used in coordinating turn-taking (Tang, 1991). In one case a participant even threw the public document to another participant across the table informing him that it was his turn (Figure 2.9).
Figure 2.8: Gestures: The use of stylised actions to indicate a state transition.
Gutwin and Greenberg (2000) and Pinelle et al. (2003) considered such verbal and non-verbal behaviours to be part of the low level actions and interactions that must be carried out to complete a shared task, and called them the mechanics of collaboration. Gutwin et al. divided group work into task related work (which applies to individual work as well as group work), and team related work. Team related work can be further categorized according to whether it relates to social and group dynamics, or the mechanics of collaboration. Our interest in this section is in the mechanics of collaboration, which can be either communication or coordination (see the updated categorization (Pinelle et al., 2003)). Communication can be explicit or implicit through the gathering of information. Some of the actions discussed previously, such as verbal, gestural, and manifested (exaggerated) actions, fall into explicit communication; while actions intended to increase ones awareness about the process by observing other people in the a group, their activities, and changes made to objects in the space fall into implicit communication (i.e. information that is communicated through observation and not stated explicitly). Coordination is required when
Figure 2.9: Throwing the public document to another participant
managing access to resources or when transferring objects. When such coordination is not done physically, it will be through one of the communication techniques discussed, such as verbally, through body position and movements, or using gaze and gestures. The mechanics of collaboration identify observable, low-level interactions that can help in analysing collaboration and breaking it down into specific actions that can be evaluated one at a time.
2.2.2.4 Implications on digital tabletops design
The variety of functions served by verbal and non-verbal communication highlights the impor- tance of enabling barrier free communication between collaborators around a table. Gutwin and Greenberg (2000) considered the degree to which a groupware system supports the mechanics of collaboration to be a measure of its usability. They suggested the mechanics of collaboration as a discount evaluation method for shared-workspace distributed groupware systems. Traditional tables provide a barrier-free communication environment with natural support for awareness and verbal and non-verbal actions (Rogers and Lindley, 2004; Tang, 1991; Bly, 1988). Therefore, groupware systems based around tables rank high in their support for the mechanics of collabora- tion. Digital tabletop interaction designers must take care not to introduce features or obligations on users that hinder these channels of communication (e.g. visual or auditory barriers). Poten-
tial visual barriers include the use of personal devices such laptops in front of each participant. The presence of the vertical screen can affect both explicit communication by hiding some of the communicative gestures, such as pointing (Scott et al., 2003), and implicit communication by reducing awareness about others, their activities, and changes made to objects on these personal screens or on areas obscured by these screens. Auditory interference can be introduced when providing public or private audio feedback. For example, private instructions provided through headphones negatively affect implicit communication (information gathering) and may result in responses on the part of the receiver that are distracting and unintelligible to collaborators. More- over, private and public feedback that is not timed appropriately contribute to breakdowns in the interaction (Morris, 2006; Scott et al., 2003).
The analysis so far has been guided by distributed cognition theory to bring into focus different aspects of the interaction. As mentioned in section 2.2, it is useful to apply activity theory (AT) for the detailed analysis of the verbal and nonverbal behaviour to lead to the same conclusions regarding the provision of the barrier-free communication environment. Verbal and non-verbal behaviour of users can be categorized into conscious actions that serve a specific goal and sub- conscious operations that result in as reactions to different conditions (refer to section 1.3.1). Ac- cording to AT, subconscious operations transform into conscious actions because of breakdowns in the process or when a tool that is used in the interaction brings focus to itself rather than to the process. Consequently, it is possible to study the categories of actions and their frequency, and how the introduction of new circumstances (such as introducing personal devices or audio feed- back) affect the distribution and frequency of actions and operations. As an example, observing a person’s gesture and reacting to it is usually done subconsciously, but if the gesture was obscured by a visual barrier forcing the observer to take an intentional action to observe it, such as moving aside or explicitly asking the action initiator to repeat or explain what they have done, then this is an indication to a breakdown in the interaction process which requires a modification to the design. A similar analysis can be applied to the provision of audio feedback and whether it leads to breakdowns in the process that shift subconscious operation to conscious actions or not.