The Usage of Referential Communication and Its Motivation on a Shared Drawing Space Seungwon Yang Computer Science Virginia Tech Blacksburg, VA 24060 seungwon@vt.edu Yoon Suk Lee Hyun Seung Yoo Industrial and Systems Engineering Industrial and Systems Engineering Virginia Tech Virginia Tech Blacksburg, VA 24060 Blacksburg, VA 24060 yoonlee@vt.edu hyun@vt.edu ABSTRACT METHODOLOGY We examined the use of referential communication in a simultaneous shared drawing space between a group of drawers and a guesser. The drawers’ task was to collaboratively sketch a picture that illustrates a given word, so that a guesser can verbalize the word based on the picture. The verbal communication from the drawers to the guesser was restricted to promote the visual communication through the shared drawing space. The sequence of verbal interactions and drawings were recorded and mapped out on a timeline for analyses. The referential information was identified and categorized. The motivations of referential communication were discussed. We implemented a shared drawing space using Java and IBM TSpaces on tablet PCs [3, 4]. Sixteen participants (5 males and 11 females) were recruited (4 groups of 4 people), who were undergraduate students at Virginia Tech. Each group completed ten sessions, including two practice sessions. As illustrated in Figure 1, three drawers visually expressed a given word on the shared drawing space. The guesser verbalized the word by watching the drawings from the shared drawing space on her own tablet PC and from the drawers’ gestural and facial expression through one-way mirror. Drawers could also freely discuss and strategize, since the guesser could not hear their conversation. The drawers could hear the guesser’s response. The interactions of the drawers and their sketches were recorded, using a Flip video, three webcams and Camtasia Studio 6.0. The final drawings for all sessions were also captured. Author Keywords CSCW, referential communication, shared auditory/visual information channels, tablet PC drawing, ACM Classification Keywords J.4. Computer Applications: Social and Behavioral Sciences - sociology. General Terms Design, Experimentation, Human Factors INTRODUCTION We report our preliminary results specifically focused on the usage of referential information on a shared drawing space. The significance of referential communication for effective communication has been empirically demonstrated and tested [1, 2]. However, interaction style and information type may vary depending on tasks and available communication technology. These variations require more empirical findings about how people communicate in different settings. The experiment was conducted in a unique setting, which emphasized the shared drawing space as a primary communication medium. To promote the use of shared drawing space, facial, gestural, and auditory communications were partially allowed. The findings can be used to develop suggestions specifically focused on referential communication on a shared drawing space. Figure 1. Experimental setting ANALYSIS The investigators identified the referential information from the final drawings as depicted in Figure 2 (e.g., an arrow, ‘OX’ marks). The behavior of drawers, their drawings, verbal communications, and the verbal reactions from the guesser were visualized in a timeline (Figure 3). This was to understand the contexts associated with the use of referential communications. Copyright is held by the author/owner(s). CSCW 2010, February 6–10, 2010, Savannah, Georgia, USA. ACM 978-1-60558-795-0/10/02. 493 • [(De)select entity] As soon as the word ‘3 Dimension’ was posted, a team member drew a rectangle and a cube. Then she marked the rectangle with ‘X’ and enclosed the cube with a circle. After trying a couple of words, the guesser said ‘3 Dimension’ (Figure 4-d). • [Indicate progress] The given phrase was ‘slow motion’. The team began by writing ‘2 words’ on the screen, meaning that the answer had two words. After seeing a turtle, the guesser mentioned a series of incorrect answers such as, ‘turtle’, ‘tortoise’, and ‘sea turtle’. When the guesser finally answered ‘slow’ the drawers added a check mark in front of the number ‘1’ to indicate that the guesser answered the first word correctly (Figure 4-f). Figure 2. Final drawing for ‘slow motion’. Figure 4. Various types of referential information. Referential communication played an important role in directing the guesser’s attention, indicating progress, and (de)selecting entity to support effective communication via shared drawing space. However, one interesting observation was that the drawers could not effectively use referential communication to refer to what the guesser had previously mentioned. Figure 3. Excerpt of a timeline. A ‘turtle’ was used to describe ‘slow motion’. The numbers denote the sequence of interactions from 6 to 14. PRELIMINARY RESULTS The identified referential communications were collected from the screen captures and categorized into two groups: • By visual property: circle, arrow, underline, O-X, check mark, use of color and combinations of those. • ACKNOWLEDGEMENTS Special thanks go to Dr. Deborah G. Tatar for her valuable advice, James Dustin for device support, and Dr. Scott McCrikard for allowing us to use the facility. By usage: direct attention, (de)select entity, and indicate progress. REFERENCES 1. Cherubini, M., & Dillenbourg, P. 2007. The effects of explicit referencing in distance problem solving over shared maps. In proceedings of the GROUP ’07. Circles (Figure 4-a), arrows (Figure 4-b), underlines (Figure 4-e) and combinations (Figure 4-c) were mostly used to direct the guesser’s attention to a specific entity. O-X was used to indicate a specific entity among others (Figure 4-d). When the answer was a two-word phrase, one team used a check mark extensively to inform that the guesser answered the first word correctly (Figure 4-f). The following descriptions elaborate the context of the use of referential information. • [Direct attention] The team was trying to visually express ‘blind date’. The drawers drew a face with eyes marked with ‘X’ trying to express ‘blindness’. When the guesser mentioned ‘dead’, a drawer immediately added several arrows around the eyes to indicate that the answer was associated with eyes (Figure 4-b). 2. Kraut, R., Fussell, S., & Siegel, J. Situational awareness and conversational grounding in collaborative bicycle repair. Human Computer Interaction Institute. Carnegie Mellon University. 3. Wyckoff, P., McLaughry, S., Lehman, T., Ford, D. 1998. TSpaces, IBM Syst. J. 37, 3 454-474. 4. IBM. TSpaces: Intelligent Connectionware. Retrieved from http://www.almaden.ibm.com/cs/TSpaces 494