Facilitating knowledge convergence in videoconferencing environments: The role of
external representation tools
Frank Fischer & Heinz Mandl University of Munich, Germany [email protected] ABSTRACT
The study investigates the process of knowledge convergence in computer-mediated and in direct communication.
More specifically we analysed how two types of graphical representation tools influence the way in which learners use shared and unshared knowledge resources in these two different collaboration scenarios, and how learners represent and transfer shared knowledge under these different conditions. Moreover, the relation between the use of knowledge resources, representation, and the transfer of shared knowledge was analyzed. We varied the type of graphical representation (content-specific vs. content-unspecific) and the collaboration scenario (videoconferencing vs. face-to-face). 64 university students participated. Results show that the learning partners converged in their profiles of resource use. With the content-specific graphical representation, learners used more appropriate knowledge resources. Learners in the computer-mediated scenarios showed a greater bandwidth in their profiles of resource use.
A relation between discourse and outcomes could be shown for the transfer but not for the knowledge representation aspect.
Keywords
Collaborative knowledge construction, external representation, shared knowledge, videoconferencing
BACKGROUND OF THE STUDY
Knowledge Convergence
A question central to research and practice of computer-supported collaborative learning is, how locally distributed learners manage to converge with respect to their knowledge. In this paper we therefore focus on a theoretical aspect which seems both, highly relevant for the field and so far neglected by empirical research: The aspect of knowledge convergence. Following Roschelle (Roschelle, 1996), convergence - not conflict - is the crucial point in collaborative learning: two or more learners, whose activities have impact on those of the partners - which - in turn - have impact on the own activities. Salomon and Perkins (Salomon & Perkins, 1998) spoke of "Spirals of Reciprocity" to characterize these collaborative interdependencies. The psychology of knowledge acquisition has up to this point dealt first and foremost with the single individual. Even when analyzing cooperative learning processes, the focus of attention was, how individuals represent their knowledge, how they solve problems, etc.
What the learning partners do exactly, how they represent their knowledge and solve problems, has up to this point played a subordinate role. Little research has been done on this point. Even less is known on how convergence is affected by specific characteristics of different collaboration scenarios. We believe that a more elaborated concept knowledge convergence could advance research on computer-supported collaborative learning.
In our analysis of cooperative learning we consider three main aspects of knowledge convergence: (1) Process convergence. If two or more learning partners cooperate, they use shared and unshared knowledge resources. It can be investigated, how two or more group members use the knowledge available to them (e.g. from their prior knowledge, from learning material and so on) to collaboratively construct new knowledge in discourse.
From studies in collaborative decision making we know that groups often show a tendency to neglect unshared resources, i. e. knowledge and information which only one or a small proportion of the group members has access to (e.g. Wittenbaum & Stasser, 1996). Moreover, it we analyse how learning partners converge in their use of resources or the content focus, for example situational or strategic focus. As a final aspect of process convergence we assume, that learning partners may influence each other concerning the functional modes of their discourse, like for example externalization, elicitation, conflict-oriented negotiation. It is plausible that cooperation partners develop a kind of collaborative stile - even in short-term problem solving activities (e. g. Roschelle & Teasley, 1995). (2) Outcome convergence. If group members learn together they can construct shared cognitive representations. Here, for example, it can be of interest, to what extent the learning partners construct similar declarative knowledge. The pioneering study of Jeong and Chi (Jeong & Chi, 1999) showed that only a relatively small portion of the
knowledge, which a dyad constructed during collaboration, is actually represented by both of the learners. Moreover, it the relation of knowledge representation and knowledge application is increasingly seen as very complex (Renkl, Mandl, & Gruber, 1996). Therefore, we also consider the transfer of shared knowledge to be an important aspect. A main question is to what extent former learning partners are similarly able to apply the shared knowledge in new contexts.
Facilitating knowledge convergence with shared external representations
Shared external representation tools might help to improve their discourse in computer-supported collaborative learning scenarios (e. g. Fischer, Bruhn, Gräsel, & Mandl, in press). In this study we investigated the effects of different types of shared representation tools on knowledge convergence. We distinguish three theoretical perspectives to external representation: (1) The memory aid/ processing aid perspective (e. g. Larkin, 1989) assumes that external representations might function as external working memories, might enhance metacognitive processing, and foster organizing processes. (2) The constraints-and-affordances perspective. Being inspired by Gibsons ecological psychology, constraints and affordance approaches emphasize that external representations helps information processing by reducing possible degrees of freedom in the learners' activities or by initiating activities by providing salient structure. This salience assumption has recently been extended as representational bias by Dan Suthers.(2000). (3) Group-specific approaches. This third group comprise approaches, which are genuinely collaborative in that they make the interaction - not the individual processing - to their starting point. Examples for collaboration-specific approaches are (1) the collaborative abstraction and (2) the grounding (further advanced recently by Baker (Baker, Hansen, Joiner, & Traum, in press).
Types of shared graphical representation tools. We distinguish between two forms of external graphical representation: (a) Content-unspecific representation: The widespread shared whiteboards (mostly simple graphic editors) should support interaction between remote collaborators by providing them with the possibility to collaboratively visualize graphical elements as well as written notes. The subject area (e. g. medical diagnosis, botanical classification) as well as the task type (e. g. discussion, decision making, learning) do not play a role in the design of this tools. In our study, learners in the content-unspecific visualization condition worked on a computer tool which comprises the functionality of a simple graphic editor. The learners could type in and edit text, draw lines, circles and rectangles and change the colors of these items and drag the items across the screen. In (b) content-specific representation, the degrees of freedom of the external representation are constrained by task-relevant structures. Dyads in the content-specific representation condition were provided with a computer-based graphical mapping tool, that includes concept cards for case information and cards for theoretical concepts, in which text could be typed in directly. Positive and negative relations can be used to connect cards. Moreover, the screen of the CoStructure-Tool is divided into an empirical and a theoretical level. Both learners were provided with a keyboard and a mouse and could access the different objects on the screen virtually simultaneously.
We expected that the provision of these task-related categories in the content-specific representation would promote the construction of shared knowledge because of the representational bias and because less coordination effort will be necessary. We assume that content-specific external representations provide an initial coordination of activities in the case learners have to collaboratively represent knowledge externally. Moreover, when we use Suthers (Suthers, 2000) representational bias assumption of external representation it should be possible to guide the discourse focus of the learners with a content-specific external structure.
Videoconferencing
It is unclear, to what extent the conditions of videoconferencing have an impact on process and outcome convergence. Up to this point, no systematic studies on this topic have been conducted. A smaller amount of convergence is possible, for the development of similar positions might be mediated through nonverbal and para- verbal aspects. Although nonverbal and para-verbal signals can be partly transported through audio and video connections, important differences do exist between face-to-face communication and videoconferencing (Fussell &
Benimoff, 1995; O´Connaill & Whittaker, 1997). For example, the lack of eye contact and gaze awareness, differences in the visual fields of the partners, as well as the reduced possibility to make deictic gestures in a video conference could serve as hindering factors. Overlapping turns and unwanted interruptions can often occur under these conditions.
However, empirical studies on problem solving and decision making rarely showed differences between videoconferencing and face-to-face conditions concerning the outcome (O´Malley, Langton, Anderson, Doherty-
Sneddon, & Bruce, 1996): In spite of partly different process characteristics cooperation partners frequently come to qualitatively similar solutions in the setting of a video conference as compared to face-to-face settings. There are even Web pages on "the no significant differences phenomenon" collecting further evidence of outcome similarities between the two collaboration scenarios.
Research questions
(1) Do learning partners converge with respect to the content focus, discourse modes, knowledge representation, and knowledge transfer?
(2) Which effects do the kind of external representation, the collaboration scenario, and their combination have on these variables?
METHOD
Sample and design. Sixty-four students of educational psychology volunteered in this study. The participants were separated into dyads and each dyad was randomly assigned to one of the four experimental conditions in a 2x2 factorial design. We varied (1) the cooperation scenario (face-to-face vs. computer-mediated) and the type of external representation tool (content-unspecific vs. content-specific). Time-on-task was held constant in all four conditions.
Learning environment. Students in both conditions had to work on complex cases in the domain of education. In these cases, teachers describe a plan for an instructional unit and ask the participants for an evaluation of the plan from a theoretical perspective. The students’ task was to prepare a common evaluation of the case. While working on a case, students were provided with a collaborative visualization tool to represent their developing solution graphically. After each case, students were asked to give a short oral evaluation of the case from a theoretical perspective.
Type of representation tool. Dyads in the content-specific representation tool condition were provided with the CoStructure-Tool, a computer-based graphical mapping tool, that includes concept cards for case information and cards for theoretical concepts, in which text could be typed in directly. Positive and negative relations can be used to connect cards. Moreover, the screen of the CoStructure-Tool is divided into an empirical and a theoretical level. Both learners were provided with a keyboard and a mouse and could access the different objects on the screen virtually simultaneously.
Learners in the content-unspecific graphical representation condition worked on a computer tool, which comprises the functionality of a simple graphic editor. The learners could type in and edit text, draw lines, circles and rectangles and change the colors of these items and drag the items across the screen.
Procedure. After a pre-test consisting of a content-specific declarative knowledge test and a case task, students were made familiar with the learning environment, especially with the use of the representation tools. Next, learners worked together on three cases. During their work on the cases learners were allowed to use a text with a number of relevant theoretical concepts and their descriptions. The collaboration was followed by an individual post test which paralleled the individual pre-test.
The effects reported here have been tested with ANOVA or Kruskall-Wallis, and Mann-Whitney procedures.
RESULTS
First, we analysed, to what extent learning partners converged concerning the process of collaborative knowledge construction. Fig. 1 shows the standardized amount of specific content focus categories like case information, concept or relation between them.
z-values Shared resources
theoretical concept
rel. theor. concept - case information
prior knowledge concept case information
rel. prior kn. concept - case information Unshared resources
Nominal dyad Eva Alexa
z-values
2,0 1,5 1,0 ,5 0,0 -,5 -1,0 -1,5 -2,0
Shared resources
theoretical concept
rel. theor. concept - case information
prior knowledge concept case information
rel. prior kn. concept - case information Unshared resources
z-values
4 3 2 1 0 -1 -2
Nominal dyad Lukas Peter
Fig. 2c
Fig. 2d
z-values
2,0 1,0 0,0 -1,0 -2,0
Alexander Leonhard Shared resources
theoretical concept
rel. theor. concept - case information
prior knowledge concept case information
rel. prior kn. concept - case information Unshared resources
Fig. 2a
z-values
2,0 1,0 0,0 -1,0 -2,0
Katrin Sabine Shared resources
theoretical concept
rel. theor. concept - case information
prior knowledge concept case information
rel. prior kn. concept - case information Unshared resources
Fig. 2b
Fig. 2.
Profiles of resource use in four dyads. Fig. 2a and 2b show real dyads with high (a) and low (b) degrees of shared knowledge representation. Fig. 2c and 2d are examples of nominal dyads. Learners are typically less similar in regard to their resource use profiles.
We look at two typical patterns: dyads differ highly with regard to that patterns but in most of the cases, there is a high similarity among learning partners of one dyad, indicating the high interdependency of the cognitive processes of the learning partners. In Fig. 2, the learners focus heavily on contextual information or case information and spend less efforts on elaborating and applying concepts. But again, very high similarity is characteristic for the pattern or profile of resource use.
We compared this to nominal dyads - that are two learners out of the same experimental condition, who have not learned together, whom we put together post hoc. Figure 3 shows a prototypical pattern of a nominal dyad.
There is much higher divergence at nearly every content focus category as well as for a global similarity measure based on these categories (p < .05).
Results concerning outcome convergence showed that real dyads do not differ from nominal dyads with respect to the representation of shared and unshared knowledge (p > .2). However, more shared knowledge is transferred in real dyads as compared to nominal ones (p < .05). These findings could be interpreted such that co-construction within the dyads could only be effective concerning higher order processes in connection with transfer (Salomon & Perkins, 1998).
Research question 2
We now turn to the effects of the two factors - collaboration scenario and type of external representation on the process variables. We decided to illustrate the effects with some quantitative case studies.
Fig. 2: Effects of the external representation tools on conceptual focus and on the focus on relations between concepts and context.
Effects of the type of shared external representation. Compared to the content-unspecific representation, the content-specific representation fosters the use of conceptual elaboration and the application of theoretical concepts onto case information (Fig. 2; both p < .05). This indicates a representational bias effect of the content- specific structure given with the representation tool. Even more interesting is something we call the initial coordination effect. Figure 3 shows the graphical representation of the content focus of one dyad during one case.
We distinguish 4 foci: contextual, conceptual, relation between contextual and conceptual aspects and (4) strategic.
17,62
42,25
3,75
26,25
0 5 10 15 20 25 30 35 40 45
Conceptual Relations concepts-context
Content-specific repr.
Content-unspecific repr.
Fig. 3.
Quantitative single case study of the process of collaborative knowledge construction. Moving averages (10,10) for 4 of the content focus categories are shown.
Many of the successful and converging pairs start with a phase of intense coordination. Afterwards there are - often with a kind of rhythm - shorter and less intense phases of coordination throughout the case. In between there are phases of conceptual activity and application of concepts to case information. In the content-unspecific
representation condition, dyads without such a starting sequence of strategic planning were often neither successful nor convergent with respect to the knowledge. In contrast, there are pairs in the content-specific representation condition who had high achievement and high knowledge convergence without intensive initial strategic activity.
The content-specific condition seems to provide learning partners with an scaffold for initial coordination. In these pairs, just one or two speech acts at the beginning are concerned with strategic activities, there seems to be no kind of rhythm with strategic focus later - those discourse patterns are more frequently found with the content-specific representation.
Summarizing we found two interesting effects of the content-specific external representation: a representational bias leading to higher conceptual activity and an initial coordination effect - compensating for inadequate planning at the beginning and maybe scaffolding the content-focus.
Effects of the collaboration scenario. With respect to the collaboration scenarios two interesting effects occured. Firstly, in videoconferencing we found a higher variety of collaborative knowledge construction styles.
Remember the dyadic profiles of the content focus: In videoconferencing, there is the same tendency to converge concerning the two learning partners. However, there are a more different patterns between the dyads (p < .05).
Secondly, single case studies showed that there seems to be a continuous base rate of contextual and strategic speech acts. The learning partners seem to continuously sustain their common ground concerning their proceeding and concerning the givens of the common problem. In most cases, we could hardly find a sequence, in which there is neither strategic nor contextual activity. This is not the case for the physical co-presence condition.
Non-verbal monitoring the common ground may be more easy in this condition.
Collaborative knowledge construction Pair 46/47
Speech act number
246 211
176 141
106 7 1
3 6 1
1,0
,8
,6
,4
,2
0,0
Content focus
contextual conceptual
rel. context.-concep strategic
Finally, we turn to the question of outcome convergence: We analysed the quantitative divergence of the learning partners concerning knowledge application in the individual transfer case. Interestingly, for content-specific representation, the divergence is similarly high in both collaboration scenarios. However, for the content-unspecific representation, divergence is very high in physical co-presence and very low in videoconferencing (p < .05 for main effect type of external representation and for the interaction effect). This effect pattern highlights the more general aspect, that the same tool might have quite different effects in different collaboration scenarios. We found some preliminary process-related explanations for this effect. Pairs in the videoconferencing condition with content un- specific representation tended to produce textual representation and used graphical elements rather sparsely, whereas pairs in physical co-presence rely heavily on graphical elements in their external representation. We speculate, that to collaboratively produce a text might be less excluding as a drawing task, which might be more dependent of diverging individual skills.
However, knowledge convergence in this case was a quantitative measure and not sensible to qualitative differences in the knowledge of the learning partners. Therefore, we analyzed in the next step the qualitative dimension of knowledge convergence, i. e. shared knowledge representation and shared knowledge transfer.
Remember that in the comparison of real and nominal pairs we showed an effect of the collaborative knowledge construction only for shared knowledge transfer and not for shared knowledge representation. However, no effects of the collaboration scenario (p > .05) nor of the type of shared representation (p > .05) on the qualitative knowledge convergence - or shared knowledge - could be shown.
CONCLUSIONS
Our conclusions of this study are the following: Firstly, we found clear evidence for collaborative styles even in short term collaboration: Learning partners strongly converge to a common profile of content focus in collaborative knowledge construction. Secondly, our findings concerning the shared representation tools could be seen as support for Suthers (Suthers, 2000) representational bias assumption: Learning partners talk much more about conceptual aspects, if the external representation provide this conceptual structure. Thirdly, content-specific representation tools might provide an initial coordination for learners in that they have some task-relevant categories already in their joint problem space as a preliminary common ground (Clark & Brennan, 1991). Fourthly, collaborative knowledge construction and knowledge convergence is neither hampered nor facilitated by the characteristics of our videoconference. One compensating mechanism might be the sustained situational grounding which virtually all of our pairs displayed. Fifthly, the same external representation tool might fulfill quite different functions for the process of knowledge convergence in different collaboration scenarios.
ACKNOWLEDGMENTS
Part of this work was funded by the Deutsche Forschungsgemeinschaft
REFERENCES
Baker, M., Hansen, T., Joiner, R., & Traum, D. (in press). The role of grounding in collaborative learning tasks. In P. Dillenbourg (Ed.), Collaborative Learning: Cognitive and computational approches : Elsevier Science/Pergamon.
Clark, H. H., & Brennan, S. E. (1991). Grounding in Communication. In S. D. Teasly (Ed.), Perspectives on socially shared cognition (pp. 127-149). Washington: American Psychologie Association.
Fischer, F., Bruhn, J., Gräsel, C., & Mandl, H. (in press). Fostering collaborative knowledge construction with visualization tools. Learning and Instruction.
Fussell, S. R., & Benimoff, N. I. (1995). Social and cognitive processes in Interpersonal communication:
Implications for advanced telecommunications technologies. Human Faktors, 37(2), 228-250.
Jeong, H., & Chi, M. T. H. (1999). Constructing Shared Knowledge During Collaboration and Learning. Paper presented at the AERA Annual Meeting, Montreal, Canada.
Larkin, J. H. (1989). Display-based problem solving. In D. Klahr & K. Kotovsky (Eds.), Complex information processing. The impact of Herbert A. Simon (pp. 319-341). Hillsdale, New Jersey: Lawrence Erlbaum Associates.
O´Connaill, B., & Whittaker, S. (1997). Characterizing, predicting, and measuring video-mediated communication:
A conversational approach. In S. B. Wilbur (Ed.), Video-mediated communication (pp. 107-132). Mahwah:
Lawrence Erlbaum Associates.
O´Malley, C., Langton, S., Anderson, A., Doherty-Sneddon, G., & Bruce, V. (1996). Comparison of face-to-face and video-mediated interaction. Interacting with Computers, 8(2), 177-192.
Renkl, A., Mandl, H., & Gruber, H. (1996). Situated learning in instructional settings: From euphoria to feasibility. Educational Psychologist, 31, 115-122.
Roschelle, J. (1996). Learning by collaborating: Convergent conceptual change. In T. Koschmann (Ed.), CSCL:
Theory and practice of an emerging paradigm (pp. 209-248). Mahwah, New Jersey: Lawrence Erlbaum Associates.
Roschelle, J., & Teasley, S. D. (1995). The construction of shared knowledge in collaborative problem solving. In C. O'Malley (Ed.), Computer supported collaborative learning (Vol. 128, pp. 69-97). Berlin: Springer-Verlag.
Salomon, G., & Perkins, D. N. (1998). Individual and social aspects of learning. Review of Research in Education, 23, 1-24.
Suthers, D. (2000). Initial Evidence for Representational Guidance of Learning Discourse. Paper presented at the Proceedings of International Conference on Computers in Education, Taipei, Taiwan.
Wittenbaum, G. M., & Stasser, G. (1996). Management of information in small group. In J. L. Nye & A. M.
Brower (Eds.), What´s social about social cognition? Research on socially shared cognition in small groups (pp.
3-23). Thousand Oaks: Sage Publications.
