My particular research interest in virtual worlds is how pedagogical communication is organised for second / foreign language learning within these environments. That is to say how the different modes available to the participants in the synthetic world work together to structure and give meaning to discourse in the target language. I hope to explore this question through studying collaborative learning scenarios in Second Life, so Wadley and Ducheneaut’s paper ‘The ‘out of avatar experience': object focused collaboration in Second Life’ was an extremely interesting read as it directly addresses the subject of multimodality and collaboration.
In this paper, the authors investigate how, in collaborative building tasks, deictic verbal references and Second Life’s detachable camera are used in the coordination of the activities. In Second Life, users can move the camera independently of the position and orientation of their avatar. Whilst the user’s ‘in-avatar’ position is visible, this ‘in-camera’ position is private and cannot be seen by the other users. The ‘in-camera’ viewpoint allows the users to gain multiple perspectives of objects more quickly than by flying or walking around them. However, Hindmarsh and al argued that because users cannot determine each others’ vistas, their ability to collaborate is lessened.
In-camera view with the avatar in the same place
Wadley and Ducheneaut’s study looked at two different tasks with 22 participants. For the first task, the researchers provided components of a house of which some were distinguishable whereas others required spatial reference to identify them. Each participant was given a screenshot of how the house should look when finished. Participants worked in groups of two or three over a 90 minute period to reconstruct the house. In the second task, each group was designated one helper who had a screenshot of a garden. This participant had no building rights but rather had to direct the other group members to build this scene. The researchers recorded participants’ screens during the tasks as well as their audio. ‘Salient events’ in the data were counted with verbal references to objects being categorized by their linguistic form. These references were compared to experience levels of participants to discover patterns of use. Four participants had experience in building in SL; ten users had experience of virtual worlds and four users were novices. Two participants’ data was excluded from the study due to the poor quality of recordings.
The study found that overall task-time was divided equally between in-avatar and in-camera modes. However, experienced builders spent more time in-camera whereas novices spent more time in-avatar. The participants used their avatars to mark positions and whereas pointing and gestures were rarely used to reference objects as would be common in the physical world. Rather, the participants would often move their avatar to stand beside the object or jiggle the objects to show which object they were referring to. The experienced builders and VW users made more references to objects, and even when they were in-camera mode they used the avatars frame of reference. Novices, on the other hand, sometimes forgot these visions were not the same when trying to communicate a spatial reference. The researchers also discovered that much of the building work was divided into sub-tasks (e.g. roof and base of the house) and rarely did they observe closely-coupled collaboration with participants working simultaneously on the same sim. Perhaps because of the difficulty of communicating spatial references or a result of the way in which visual feedback is given in SL : the user can see the object being moved whereas collaborators only see the start and end positions. In terms of the verbal references made to objects, the participants most frequently referred to the objects by name or by the object’s property, secondly they used relative frames -references from the speaker or listener’s point of view and lastly they used intrinsic frames of reference when the object. No reference was made using cardinal or Cartesian systems.
Wadley and Ducheneaut conclude that in collaborative building activities in Second Life, the users do not use their avatars as we would our bodies in the physical world to refer to objects (e.g. pointing). Rather, the users make the object more accountable including making it visible when selected, relaying its movement as it happens or changing the object to refer to it.