Identifying collaboration patterns in collaborative telelearning scenarios
Anders I. Mørch
Project DoCTA (Design and use of Collaborative Telelearning Artefacts) aims to bring a theoretical perspective to the design of ICT technologies that support the sociocultural aspects of human interaction and to evaluate its use. In the first phase of project DoCTA (June 1998 - December 1999) we focused on the use of technological artefacts to support collaborative telelearning aimed at teacher training (Wasson, Guribye & Mørch, 2000). The research was not limited to only studying these artefacts per se, but includes social, cultural, pedagogical and psychological aspects of the entire process in which these artefacts are an integral part. This means that we both provided and studied virtual learning environments that were deployed to students organised in geographically distributed teams.
Various scenarios utilising the Internet were used to engage the students in collaborative learning activities (Wasson & Mørch, 1999). Through participation, the teachers gained experience with not only collaborative learning, but with collaborative telelearning through the collaborative design of a textual or visual artefact. Scenarios IDEELS (see Guribye, 1999) and Demeter (see Junge, 1999)), involved European inter-cultural simulations where the goal was to design a textual artefact (such as a treaty or policy statement). Scenario VisArt (Wasson, 1999) had the goal of designing a visual artefact to be used in teaching a subject of choice.
There are a number of research areas that have influenced project DoCTA (see Wasson, Guribye & Mørch (2000) for details). The two most significant are the conceptual framework offered by sociocultural perspectives (Wertsch, del Río & Alvarez, 1995) and the emerging area of computer-supported collaborative learning, in particular Salomon’s (1992) work on genuine interdependence. We also take inspiration and guidance from CSCW theory, in particular ideas on awareness (Dourish & Bellotti, 1992; Gutwin et al., 1995) and coordination science (Malone & Crowston, 1994). These theoretical perspectives have influenced our choice of groupware tool, the design of the VisArt scenarios and its collaborative task, and the design of our research studies (Guribye & Wasson 1999; Mørch & Wasson, 1999).
The theoretical, or conceptual approach, is rooted in a sociocultural perspective and the methodology was influenced by ethnographic studies, favouring naturalistic and qualitative research methods. The main research question has been formulated to ask how these students, teachers and facilitators organise their work.
In order to collect data about the activities in which the student are engaged, different methods and techniques are being used. The most important sources of information are derived from observing the students as they collaborate using artefacts, from interviewing them, and from the artefacts they create/design. These ethnographic flavoured studies were augmented with more traditional computer science oriented usability studies, evaluations of computer logs, and questionnaire studies. The following specific research questions have been fleshed out in a number of Master’s dissertations (Andreassen (2000); Baggetun (in preparation); Guribye (1999); Higraff (in preparation); Meistad (2000); Pedersen (in preparation); Rysjedal (in preparation); Underhaug (in preparation) ; Wake (in preparation)):
These exploratory studies carried out within DoCTA had an unexpected side effect in that they provided us with enough insight into the processes of collaboration to enable us to identify collaboration patterns. That is, the identification of collaboration patterns was not an initial objective of the project, rather the patterns emerged as we inspected and discussed the data from the individual thesis studies.
This paper continues with a brief introduction to the theoretical underpinnings for identifying collaboration patterns. Then, the VisArt scenario, where the collaboration patterns were found is described. After collaboration patterns are introduced, the paper concludes with a discussion of the implications for the design of collaborative telelearning.
Theoretical Underpinnings for identifying Collaboration Patterns
Collaboration patterns define sequences of interaction among members of a team (such as students) that satisfy established criteria for collaborative behaviour. For example, Salomon (1992) is concerned with effective collaboration and he argues that collaboration will only be effective if there is genuine interdependence between the collaborating students. Genuine interdependence is described as 1) the necessity to share information, meanings conceptions and conclusion, 2) the necessity for division of labour into complementary roles, and 3) the need for joint thinking in explicit terms. Thus Salomon’s description for genuine interdependence is one source for criteria for collaborative behaviour.
Another example is the need for coordination. Bourdeau & Wasson (1997) report that viewing collaborative telelearning from a coordination theory (Malone & Crowston, 1994) perspective offers a means of understanding the inter-relationships between actors and entities and how these relationships can and should be supported. Adopting Salomon’s ideas about genuine interdependence and a coordination science approach, they modelled (inter)dependencies between actors in collaborative telelearning scenarios and they have extended the definition of coordination to be managing dependencies between activities (Malone & Crowston, 1994) and supporting (inter)dependencies between actors. Wasson (1998ab) proposes a set of actor (inter)dependencies and related coordination processes for collaborative telelearning. This set of (inter)dependencies is another source for criteria for collaborative behaviour.
Mørch & Wasson (submitted) discuss collaboration patterns in light of previous work in CSCW and in other areas. They have identified the following candidates for collaboration patterns in this literature:
Mutual learning was found to be important in the Florence project (Bjerknes et al., 1985) as a strategy for joint system development between nurses and system developers. It was adopted to create system descriptions of a future hospital information system by methods of participatory design. Without each group first teaching the other about its professional language, the shared descriptions would have been difficult to develop (Bjerknes et al., 1985).
In a case study with the ShrEdit system (Olson, et al., 1992), Dourish and Bellotti (1992) identified shared feedback as a feature that is central to successful collaboration in multi-user (shared) text editors. Shared feedback makes information about individual activities apparent to all participants in a group by presenting feedback on operations within the shared, rather than the private workspace (Dourish & Bellotti,1992).
The VisArt scenario involved students taking courses at three geographically dispersed educational institutions in Norway, the University of Bergen (UiB), Nord-Trøndelag University College (HiNT), and Stord/Haugesund University College (HSH). Teams comprised of 3 students (1 student from each institute for 5 of the teams; 2 from HSH and 1 from UiB for 7 of the teams) collaborated to design a learning activity over the Internet. There were no opportunities for the teams to meet face-to-face. As the HSH students were distance education students, they never meet face-to-face within their course either. TeamWaveWorkplace(TW) was used as the main information and communication technology. The VisArt activity took place during February and March 1999 and provided an opportunity for the DoCTA project to study an authentic collaborative telelearning activity. One week of training (Underhaug, in preparation) in using the TW tools and in collaboration proceeded three weeks of design activity. This section gives a brief description of the scenario. For more details see Wasson, Guriby & Mørch (2000) and visit http://www.ifi.uib.no/docta/VisArtDemo/VisArtOverview.html for a guided visual tour of VisArt.
The VisArt Assignment
Salomon’s ideas influenced the design of the collaborative task given to the students – effort was placed in designing a task that created genuine interdependence between the students. In the VisArt design activity the teams were to design a TeamWave Workplace room for learning about some topic. The students were given literature on CSCL (including Salomon (1992) & Gutwin et al. (1995)) and were asked to organise a team effort thinking about Salomon’s definition of genuine interdependence (sharing information, division of labour, and joint thinking). More details about the design task were given as follows:
Evaluation of Groupware Systems
Higraff (in preparation) evaluated four existing groupware systems and frameworks: BSCW (Bentley, Horstmann & Trevor, 1997), Habanero (Chabert et al., 1998) Lotus LearningSpace (Lotus, 1998) and TeamWave Workplace (TeamWave, 2000). The volunteer participants were given a collaborative modelling task that was connected to an undergraduate course they were taking. The task required that they design a use-case diagram and a class-diagram for an information system to be used in a Dental Centre. Each pair was given a short introduction to the groupware system they were to use. Then they were given the task assignment and could begin to work, each at a geographically dispersed computer. The study addressed the usability, the interface and the relative usefulness of each system. One of the systems, TeamWave Workplace, has been used in two longer scenarios, one of which was VisArt.
TeamWaveWorkplace(TW, http://www.teamwave.com) is based on the metaphor of shared networked places. Using real life physical team rooms (Johansen et al., 1991) as inspiration, the notion of virtual team rooms has been adopted in TW. These virtual team rooms provide a permanent shared space where teams distributed over the Internet can have meetings, store documents, share URL links to web sites, coordinate and communicate with one another and can carry out collaborative activities such as brainstorming or participation in a discussion forum. Each team can build a set of different rooms according to need. For example, they might create a common project room and resource room, and in addition, individual rooms for each team member. Each room can be customised by the team to suit their specific needs and tasks by using any of the 19 tools provided in TW. The use of the tools is relatively intuitive just by seeing the name of the tool. The tools include: Address Book, Brainstormer, Calendar, Chat, Concept Map, Database, Doorway, File Holder, File Viewer, Image Whiteboard, Meeting Roster, Message Board, Personalised Message, PostIt, ToDoList, URLRef, Vote, Web Browse, and the on-line help. For a thorough review of TW see (Wasson, Guribye & Mørch, 2000).
Included in the design of VisArt was the preparation of the learning environment in TeamWave Workplace. Several rooms, including the Classroom, the Training Room and the Help Room, were created for the students. When a student logs onto the TW server, they find themselves in the Classroom. The Classroom, see figure 1, was created by the head instructor.
In the Classroom there is a To Do List which lists a number to tasks for the students to carry out (e.g., change their password, add their name to the Address Book, go to their Team room and leave a message on a PostIt), etc.). There is an Address Book where students, instructors and researchers can enter contact information, a class Calendar where important deadlines are indicated (e.g., on March 3 Oppgave 4 legges ut means Assignment 4 will be available) and a Message Board where messages can be left for the instructor or researchers. There are also a number of doorways to other rooms including: the Help Room, the Training Room, and private Team rooms (Team01 – Team11). There is a File Holder that contains the VisArt assignment (VisArt Oppgave) and four URL links to the profile, pre-, self-evaluation, and post-questionnaires. Similar rooms for training and help were designed.
Another aspect to which a lot of attention was paid was how to provide help and assistance to the students. In addition to the Help Room where students could post questions, find pointers to tips, etc., we used an assistance model where all help requests could be sent to a head assistant via email. The head assistant then either answers the help request, or sends it further to either a content assistant or a technical assistant who become responsible for answering the question. If the assistant answering the question feels that the question and answer are relevant to all the students, the answer can be emailed to all students and not just the student asking the question. Help pages (http://www.ifi.uib.no/docta/VisArt/help.html), in Norwegian, were also developed.
The 32 students (12 females & 20 males) participating in the VisArt activity had different backgrounds, ranged in age from 23 to 68 years, and many had family responsibilities and full time work.The 11 University of Bergen students were taking a graduate course in pedagogical information science where one of the topics of study was collaborative telelearning. They were a blend of pedagogical information science graduate students (with a teacher’s background) and information science graduate students (with a social science background). After participation in VisArt, they had to produce an individual report which included a reflection on the experience. The 6 students at Stord Teacher’s College, were senior undergraduate students training to be teachers. They were taking a distance learning course on pedagogical information science and were also learning about collaborative learning. They participated in VisArt and wrote about their participation in their “electronic workbook”. Finally, the 5 Nord-Trøndelag students were taking an undergraduate introductory course on ICT in learning and had a choice to participate in the scenario to learn about a telelearning application. These students used the VisArt experience as part of their final course project.
Deployment of VisArt
The VisArt activity was deployed for one month beginning on February 25th and ending on March 26th. Following five days in which the students were to download TW, test their accounts and team email addresses, one week of training began. This was followed by three weeks of design activity.
There were 11 teams and the topics they chose to design learning rooms for included: endangered species, gothic art, publishing on the internet, geometry (triangles), the big bang, travelling in Denmark, renewable energy sources, between the world wars, polar bears, and astronomy.Figure 2 shows the TW learning room designed by Team07.
Figure 2. Room for Learning about Polar Bears in TeamWave Workplace
In the learning room for polar bears (isbjørn) shown in figure 2, the targeted students are told that they will learn about polar bears. Before beginning the four assignments (oppgave 1 – oppgave 4), they can look at some pictures on the web (URL - Bilder) of polar bears in their natural habitat. Also they can take a polar bear test (Isbjørnn-Quiz) to check how much they already know about polar bears. There are 4 assignments each with 2 or 3 questions that they will find the answers to. For example, in assignment 2 (oppgave 2) they are to find out “Why don’t polar bears freeze?” and they are asked to “find their secret to keeping warm”. Assignment 3 (oppgave 3) is about enemies of the polar bear and the students are asked to find out how many polar bears there are on the Norwegian/Russian Island of Spitsbergen.
Evaluation of VisArt
The evaluation of the VisArt scenario has been carried out on several levels and from several perspectives. In addition to the student’s own theoretical reflections which were given in an individual report after the completion of scenario, VisArt is being evaluated as part of eight Master’s theses. Numerous data and data collection techniques have been used including questionnaires, semi-structured interviews, log files produced automatically by TW, email send among team members, chat files saved by the teams, documents and artefacts produced by the teams, participant observations, video taping of activities, self-evaluations and the participants own interpretation of the experience. As we discussed the findings of the individual Masters’ theses, we became aware of a number of collaboration patterns that were emerging. Thus, in addition to the individual studies, we are beginning an integrative analysis aimed at gleaning collaboration patterns from the data. The remainder of this paper focuses on four of the collaboration patterns that we have identified.
Collaboration patterns define sequences of interaction among members of a team that satisfy established criteria for collaborative behaviour. We have identified four instances that we believe can be characterised as collaboration patterns: adaptation, coordinated desynchronisation, constructive commenting and informal language. The reader should keep in mind that our results thus far are preliminary as the majority of the data collected is part of a set of Master’s theses that are in various stages of completion. We see great promise, however, in identifying collaboration patterns and expect them to be useful in further work on virtual learning environments. We use examples from the data to illustrate the patterns by showing concrete, but representative, activities.
Pattern 1: Adaptation
Synopsis: This pattern describes how students gradually adapted to each other’s practices when working together to solve a common problem.
There were no established protocols for how to use the different tools in TW. The students used a wide variety of tools when communicating with each other (e.g., a message board, shared rooms, chat, post-it notes, etc.). Indeed, several of the tools could actually be used to accomplish the same end and since the students preferred to use tools they felt comfortable with they would choose differently. After an initial phase of interaction, however, the more or less arbitrary tool use pattern changed. Here is an excerpt from an interview that illustrates this change:
Sam: “… but also Pat had a tendency to use post-it notes inside his room”
Interviewer: “…Yes, they are really popular.”
Sam: “We had to run into his room and see what new post-it notes had been posted there instead of looking at the message board or the group room.”
In this excerpt one of the students (Sam) has adapted to another student’s (Pat) way of communicating by placing post-it notes in his private room and waiting for people to come by and read them instead of using the shared room (group room). The shared room had been explicitly designed to be a common meeting area. In spite of this, the student (and later the others) adapted to one student’s ways of using a communication tool and ignored the initial set-up of the public room. The facilitator of the above scenario (i.e., the observer who noticed it) summarises it in the following way:
“A person enters into his own room looking for messages. He then takes a roundtrip through the other team member’s rooms to see if any messages have been posted there. He then returns to his own room”.
This adaptation to a new team behaviour that followed as a result and persisted throughout the scenario we have identified as the adaptation pattern. We call this a collaboration pattern because we have found similar interaction sequences in the other teams as well.
Pattern 2: Coordinated desynchronisation
Synopsis: This pattern describes how coordination of activities between team members changes after they have identified a common goal.
Many of the teams would start working synchronously, for example by having a simulated real-time meeting. In the meeting the team members would agree upon a goal to pursue, divide the work and then work more or less asynchronously to accomplish it. The latter kind of working meant fewer same-time meetings but required regular “coordination points” to ensure that the divided work would progress towards the common goal. TW tools supported this kind of coordination — they served as coordination mechanisms.
Here is an excerpt from a session that started with a conventional (synchronous) discussion of meeting times:
S1: I vote that we work individually and that we meet Monday or Thursday morning.
S1: Yes, we must meet before Monday at 5pm.
S3: I think tomorrow is too soon. I’ll go for Monday at 5pm.
S2: (…) Monday is better for me too.
S1: OK, Monday it is.
When a meeting time later had to be rearranged because one or more of the students could not make it they would leave messages for each other in their TW rooms, or they would send emails. The data from later activities shows a gradual shift from synchronous to asynchronous communication. This is further evidenced by a comment from S2, expressed as a post-it note:
After having deciding upon the goal, the students would divide the work among themselves, each being responsible for one part and work asynchronously on this. A reason for this shift from synchronous to asynchronous communication can, in part, be explained by the following comment by the facilitator:
“They (the students) would work on their own — creating the rooms — and the others would drop in and comment on the work that had been done and proposing changes they thought should be made.”
Not only were the students working on their own in an asynchronous fashion, they were coordinating the activity by visiting each other’s rooms and giving feedback to each other. The post-it notes in TW were frequently used for this purpose. In this situation the asynchronous communication is also carried out in a way satisfying the conditions of the adaptation pattern described above, or in other words the coordinated desynchronization pattern uses adaptation as one of its component parts. This part/whole structure can be found in many examples and is similar to how patterns are organised in other “pattern languages” (e.g., Alexander et al’s (1977) architectural patterns and Gamma et al’s (1995) software design patterns).
Pattern 3: Constructive commenting
Synopsis: This pattern describes commenting behaviour. Comments that are neutral (e.g., just to the point) are perceived to be less useful than comments that are also constructive (e.g., suggesting what to do next) or supportive (e.g., encouraging).
This pattern is related to Pattern 2 and often used as one of its components. We found that lack of feedback was perceived to be problematic in several of the teams. Students wanted to get feedback from each other on the artefacts they have made in their rooms. Feedback is also required to keep each other up to speed and moving in the right direction towards the common goal. This could be accomplished by using the tools such as a message board, or post-it notes, etc. A chat dialog that took place between students illustrates this pattern.
While collaborating in TW to create a class diagram (see section 4), two students were using the chat tool to make proposals to each other and waiting for responses before continuing.
Initially – student B asks for permission to execute her proposals:
B: What use cases do we have?
A: What about “Making Appointment”?
B: I agree. But we can have two kinds of appointments: through dentist and through assistant. Will this require two sub-cases?
A: Let’s assume that making appointments can only be done through a secretary
Later this interaction sequence (of getting a confirmation before continuing) changed. One would post a diagram and keep on working until they heard something from the other person. The following excerpt illustrates this as well as the linguistic mechanisms used to coordinate the activity.
Later – both of them are commenting and encouraging each other:
A: I have changed the diagram. Do you agree?
B: Yes, it is great! You have created a connection between “report” and “dentist”
A: You impress me!
The initial commenting is somewhat of a hindrance (time consuming) since feedback is requested and given at every atomic step in the interaction. However, in the latter case the student A is given the opportunity to change something that has built up over some time (a diagram) and the changes are likely to have more impact compared to the previous version of the diagram. The author of the changed diagram will therefore be subject to more severe commenting than in the one-step-one-comment case. When commenting of the latter kind is constructive and encouraging it is perceived to be more supportive of collaborative behaviour. Thus, we call this pattern Constructive commenting.
Pattern 4: Informal Language
Synopsis: This pattern describes how interaction often starts in a formalistic style and gradually becomes more informal as team members get to know each other. Frequent use of slang words or dialects local to the community working together is common in instances of this pattern.
This dialog is extracted from a dialogue between 2 students (bruker1 and bruker2) who are, as in the previous example, creating a class diagram together. In this excerpt and the next example, the original Norwegian dialect is given in Italics with an English translation in parentheses. In this first part, the two are “getting to know” each other:
[12:52:40] bruker1 says: Skal jeg lage flere aktører? Tannlege osv? (Shall I make more actors? Dentist, etc.?)
[12:53:15] bruker2 says: Greit (OK)
[12:54:45] bruker1 says: Enig i at vi har alle ansatte på høyresiden? (Agree that we have all the employees on the right?)
The next excerpt is taken from the dialog about half an hour later. The two seem to feel more comfortable with each other. They both use slang or “dialect”. In Norwegian there are 2 formal languages and an unknown number of dialects. Every little place in Norway has their own dialect. The two students in this example come from somewhere outside Bergen (on the western coast) and somewhere close to Bodø (in the north) – their dialects are very different! Also, they both use abbreviations and uncompleted sentences:
[13:31:39] bruker1 says: Se! (Look!) (Subject points at new connection-arrow)
[13:31:48] bruker2 says: Kossen? (How?)
[13:31:51] bruker1 says: Vi har laga Inheritance-pil! (We made an inheritance-relation!)
[13:32:14] bruker1 says: Den ligg nederst i "tabellen" av link-typer. (It’s in the bottom of the “table” of link-types)
[13:32:21] bruker2 says: Aha! (Aha!)
[13:33:23] bruker1 says: Ka no då? (What now?)
[13:33:49] bruker2 says: Forbindelsar... (connections...)
[13:34:08] bruker2 says: Ml.klient - tannlege? (between Client - Dentist?)
[13:34:35] bruker1 says: Ja, selvfølgelig... E litt sein i hodet i dag... (Yes, of course…My head works a bit slow today…)
When the students had “warmed up”, they were more comfortable with each other. They would often shift to an informal language and use slang words, local dialects and uncompleted sentences during the interaction. In the excerpt above one of the students used the word “Kossen” (How) while communicating using the chat tool. This term is not common in written Norwegian, but it is common in regional speech. By using the word Kossen, the student (bruker2) identifies himself as someone coming from a certain region in Norway (in this case Bodø). This makes the interaction much more personal and informal.
Our findings and subsequent reporting of the patterns are, at this point, only descriptions, hence there are no direct links to whether or not they were productive or disruptive to collaborative learning. As the collaboration patterns emerged from a number of individual studies, our integrative analysis is only beginning. The patterns have been observed at least one time, and several of them have been observed more than once. As these observations have been made informally we wish to wait until further investigations have been made before we report on their frequency. Furthermore, as our initial studies were not designed to search for collaboration patterns, we did not attempt to measure their effect on collaboration, thus we can only speculate on whether they are productive or disruptive. For example, the adaptation pattern may be a result of poor guidelines for how to use individual tools (Baggetun & Mørch, 2000) to support coordination, or, it may be that people like to get to know each other before they begin working together. Coordinated desynchronization could be both disruptive and productive. Attempts to try to learn the skills of distributed learning from a prior perspective of face-to-face learning may lead the students to believe they need to collaborate synchronously. This could be disruptive in that finding times for synchronous meetings can become difficult as the different lives of the students set so many constraints that it becomes impossible to find meeting times. This could lead to dividing their tasks such that there is less genuine interdependence between the students and lead to a more cooperative than collaborative way of working. On the other hand, this pattern could lead to more efficient use of their time as it is possible to collaborate asynchronously. One team member pointed this out in the interviews and stated that they, as a team, should have been more clever in working asynchronously, but early on their behavious was influenced by the misconception that collaboration meant they needed to work synchronously.
Implications for design of collaborative telelearning
Andreassen’s study (Andreassen, 2000) is an example of how identification of collaboration patterns can implicate the design of collaborative telelearning environments. In his preliminary analysis Andreassen dealt with aspects such as coordination, communication mode, division of labour, and feedback and identified a few aspects of work organisation that occurred within one team in VisArt. The emergence of an explicitly formulated common goal seemed to reduce the need for same-time meetings, leading to a shift in communication mode, from largely synchronous to asynchronous. This shift was accomplished through a process identified above as coordinated desynchronisation. In addition, a more asynchronous form of work organisation may be ascribed to the division of labour. This, and coordinated desynchronisation, may, but need not, lead to a cooperative rather than a collaborative form of work (cf. Dillenbourg, Baker, Blaye & O’Malley, 1996). Each team member having his/her own area of responsibility, without the final outcome being dependent on the results of the work of the other team members, may be unfortunate in that it reduces the incentive to provide feedback. Feedback, leading to reflection on the work and learning of both oneself and others, may be seen as an integral part of a concept of collaborative learning, and settings promoting collaborative work organisation should therefore be provided (Mørch & Wasson, submitted). Particular emphasis might be placed on encouraging a team to formulate a common goal and to divide their task so that genuine interdependence is present.
In the initial phases of a collaboration effort, a sort of double communication might occur. For example, more than one tool is used to inform other team members about a changed meeting time. This type of adaptation pattern was observed in VisArt and lead to inefficiencies. This sort of communication may be reduced or disappear with improved technical understanding or changed work coordination over time, but might be avoided with sufficient training and examples on how different tools can be used for coordination purposes. For a further discussion of the relationship between coordination, collaboration, and communication in the VisArt scenario see Baggetun & Mørch (2000).
What these examples illustrate is that training and practice in the art of collaboration is equally, if not more important, than just being trained in using the tools. This may be even stronger in collaborative telelearning settings where the only means for collaboration is through collaborative technologies. The tools need to be seen as an integrated part of an activity that entails coordination, collaboration and communication. This view is promoted by an activity theory (Leontev 1978; Engeström 1987) perspective which had a strong impact on the design of several of the DoCTA research studies (see Guribye (1999), Guribye & Wasson (1999), Mørch & Wasson (1999), and Andreassen (2000)).
Conclusions and Further Work
Our interest in collaboration patterns is less on instruction for design and more on identifying desired relationships between people, tools, and tasks that can stimulate collaborative behaviour, such as genuine interdependence. A team of students needs to be allowed to find their own way of collaborating instead of being forced to behave in a particular manner. A collaboration pattern should therefore be seen as a recommendation for how (or how not) to interact in collaborative telelearning activities and not as instructions for enforcing such behaviour. We feel that both the provision of a number of enabling tools and the accesses to a number of examples of fruitful collaboration patterns is necessary. This requires patterns to be both usable (understandable by people other than developers) and useful (explicit and shared among the collaborators). Lack of usability has often been cited as a limitation of the pattern languages approach since design patterns are primarily a tool for developers (Gamma et al., 1995). Collaboration patterns have the potential to be both usable and useful. These issues, however, need to be investigated further.
This work on identifying collaboration patterns is in its infancy. We feel, however, that it is promising and that our continued scrutiny of the collected data will most likely reveal additional patterns. Thus, the work on collaboration patterns will continue in the second phase of the DoCTA project (2000 – 2003). In this phase we will continue to search the data for instances of the collaboration patterns presented above. We will also continue our literature search to identify other criteria that can be used to identify new patterns. In addition, the implications of these collaboration patterns on pedagogical, technological and organisational design will be studied. Finally, work has already begun on how electronic (intelligent) agents can identify incomplete collaboration patterns and then advise the participants of this discovery.
DoCTA is a multidisciplinary research project administered and co-ordinated by the Department of Information Science at the University of Bergen, Norway. The project is funded by KUF’s (the Norwegian Department of Churches, Education and Research) ITU programme (IT in education). Phase I of DoCTA is a collaboration with HiNT (Høgskolen I Nord-Trøndelag), HSH (Høgskolen Stord / Haugesund) and Telenor FOU, Kjeller. The 19 researchers and graduate students involved have various backgrounds including computer science, psychology, sociology and education. Finally, the authors sincerely thank the anonymous referees for their constructive criticism that has lead to a much better paper.