Guiding widgets provide an intuitive way of assigning tasks and roles to co-located collaborative learners. This functionality is currently not provided by the XSS Framework, which was pre- sented in chapter 3. Currently, a typical application created with the XSS Framework provides guidance through a task title and description. However, there is no way of representing the learn- ing partners and their roles (other than mentioning them in the task description). In contrast to the Task Tokens or Ludo learners need to verbally communicate their tasks and roles which are also automatically assigned by the application. If the group only consists of two learners (as in our example application M.U.R.D.E.R. in section 3.2) the coordinative effort, which is imposed on the learners, is still acceptable. However, the larger the learning group is, the more useful are Guiding Widgets as they ease coordination. In addition, applications developed with the XSS Framework do not provide a visualization of the task sequence - they merely guide the learners through the sequence ‘blindly’ by presenting a sequence of screens each of which provides the means to fulfill the task. There is also no way of knowing when the learning partner is finished - the application just asks finished learner to wait.
6.4 Related Work 137
Consequently, the functionality of the XSS Framework (as well as of the applications devel- oped with it) could benefit from integrating Guiding Widgets. Distributed Ludo provides a good example of a solution for distributed applications (which are currently supported by the XSS Framework). The role indicator of the Ludo laptop (on the right of Figure 6.10) could be added to the XSS Dummy Clients (cf. Figure 3.10). In addition, there could be a shared representation, such as the Ludo wall (Figure 6.10). All of this could be generated automatically.
As the XSS Framework at its current state only supports distributed applications, a concept for an extension towards shareable user interfaces was proposed in section 3.5. On top of this extension, the XSS Framework for shareable user interface could be further extended with regard to Guiding Widgets. Strict Ludo provides a good example how this could look like. As in Strict Ludo, the shared workspaces (cf. Figure 3.18, right) could show a representation of the task sequence, whereas the private workspaces contain (placeholders for) role indicators and task descriptions. Again, this functionality could be automatically generated by an extended XSS framework.
6.4
Related Work
This work is related to several research areas. First, there is the large body of research on col- laboration scripts, which has been introduced in section 2.1. Of particular interest is the learning strategy ‘Reciprocal Teaching’ (Palincsar and Brown, 1984), which defines the learning tasks used in this application. Also related are other tabletop applications that support collaborative learning (cf. section 2.2.1), even though these applications are not based on collaboration scripts. Second, in the field of computer-supported collaborative work (CSCW), there are a number of studies on social protocols that many tabletop applications rely on, for example asking team members before taking action (Izadi et al., 2003). In order to avoid conflicts that might occur despite such protocols, Morris et al. (2004) established a set of coordination policies, which are graphical metaphors shown on a shared display to clarify the rights of each participant. Another approach is using personal workspaces that are either visually highlighted or implicitly defined by the proximity to the workspace owner. For instance, on the DiamondSpin (Shen et al., 2004) ownership is implicitly defined by windows that automatically turn towards the closest outside edge of the tabletop. The Task Tokens and Ludo use this research as a foundation for intuitive task and role assignment. However, to our knowledge there have not yet been any tabletop or shareable user interface applications that specifically target task distribution. A probable reason is that scripted applications are often realized on traditional personal computers (e.g. Clark et al., 2010 or Stegmann et al., 2007b). On personal computers roles are usually implicitly distributed by showing different task descriptions on different computers, which is an entirely different ap- proach than ours.
138 6 Guiding Widgets and GUI Concepts for Shareable User Interfaces
6.5
Lessons Learned
Two important aspects of scripted collaborative learning are guiding learners through a sequence of learning activities and assigning roles to different group members to induce socio-cognitive conflict. This chapter presented two GUI concepts that use Guiding Widgets to assign tasks and roles to co-located learners. At first, the Task Tokens were developed and examined in a user study. The study showed that the Task Tokens are very easy to understand, easy to use and ef- fectively guide learners through a sequence of tasks. However, from an educational perspective, further guidance regarding the assignment of different roles for each task was desirable. We therefore extended the concept and created Ludo, which assigns roles for each task. A second study showed that Ludo effectively assigns tasks as well as roles. These examples demonstrate that Guiding Widgets can successfully guide learners with regard to certain components of col- laboration scripts. Our experience also shows that more than one kind of Guiding Widgets (e.g. for tasksandroles) can be integrated in carefully designed GUI concepts without overwhelming the learners.
Furthermore, different versions were compared, which varied in the applied strictness of guid- ance. In both studies, learners preferred an automatic task (study 1) and role (study 2) assignment. They considered it fair and appreciated that the coordinative effort is taken care of by the appli- cation. Of course, this finding cannot be generalized to ‘the stricter the learning application, the better’. However, it is reasonable to assume that strict role and task assignment is generally preferred in learning settings with complementary roles and/or with roles of different popularity. Learners neither appreciated their freedom of choice in the Task Token study, where they are forced to volunteer for tasks, nor in Ludo, where there was no social pressure (as learners could simply keep their role for as long as they wanted).
Finally, while the Task Tokens and the original version of Ludo were both designed for tabletop displays, we showed that Guiding Widgets (and the according GUI concepts) can be transferred to distributed applications (e.g. to a combination of laptops and wall display) - leading to compara- ble success. Contrary to expectations the level of awareness regarding the learning partners’ roles was almost equally high in Distributed Ludo, despite the fact that the learners’ attention shifted between the private workspaces and the shared wall display. Only the awareness of the partners’ progress was higher in the tabletop versions. Consequently, the advantages of the tabletop setup were lower than expected. In addition, there were some major benefits that the participants saw in the distributed setting. For instance the laptops were found to be less distractive than the table- top application. Also, laptops are the more familiar working environment, provide easy means for text input and are more ergonomic. Furthermore, the availability of private workspaces was appreciated by the majority of the participants. In addition, there is an organizational benefit as laptops and wall displays are ubiquitously available. These findings leave serious doubt as to whether tabletops are the preferable display environment for co-located scripted collaborative learning.
Chapter
7
Using Ambient Feedback for
Mirroring and Guidance
The previous work examined guidance mechanisms, which are integrated in learning applica- tions. A different approach is using separate mirroring or guidance systems, which are dedicated to the purpose of providing feedback. Such systems collect information about the learners’ col- laboration and visualize it or give advice based on the data (Jermann et al., 2001). For instance, participation levels of all group members can be captured via microphones and displayed on the wall next to the collaborating group (DiMicco et al., 2004). We consider this form of guidance asexplicit because the feedback is unambiguously perceived as such. Furthermore, the guidance is less strict than previously discussed mechanisms because the feedback can easily be ignored - in particular because it is not integrated in the learning system. On the contrary, Guiding Wid- gets, for example, can be overruled but not completely disregarded as they are an integral part of the learning application. At the same time the independence of any learning application makes mirroring and guidance systems flexible tools that can used in combination with learning appli- cations (and therefore in combination with other guidance mechanisms such as Guiding Widgets) as well as in pen-and-paper-based learning settings.
Chapter Question:
How can learners be guided through collaboration scripts explicitly using non-interactive mirroring and guidance systems?
This chapter investigates how dedicated mirroring and guidance systems can be applied to scripted collaborative learning scenarios. After presenting related work on mirroring and guid- ance systems in general (section 7.1), we describe two studies that examined effects of script- basedmirroring and guidance systems. In the first user study (cf. section 7.2) two visualizations of the same data (metaphor and diagram) were compared to a condition without group mirror. While both visualizations had a significant positive effect on the collaboration, the metaphor led
140 7 Using Ambient Feedback for Mirroring and Guidance
to significantly faster reactions compared to the diagram. As a follow-up project we moved out of the laboratory settings and implemented a similar system in a practical seminar at a German high school. The experience in the classroom is presented in the section 7.3.