Structure and Discourse in a telematic learning environment
The author has found considerable difficulty in creating and sustaining meaningful educational dialogue in computer assisted collaborative learning, and computer mediated conferencing environments. The author believes in common with others discussed below, that there is considerable benefit to be gained from computer based collaborative work, however there are issues about engagement and structure of discourse that need to be addressed to improve the situation for learners and their tutors. This paper will discuss a specific methodology, the use of on-line reciprocal reading, that has attempted to put some structure into collaborative conversations in the context of a project that involved collaborative discussions and exchange.
The REM project (Reseau D’Enseignement Multimedia) was funded by the European Commission (EC) as part of its Telematics Application Programme (TAP), which aims to provide a telematic service for European learning experiences for students and teachers. These experiences are modelled on the EC programmes of education and training exchange: Erasmus (for higher education), Leonardo da Vinci (for training), and Comenius (for teachers and other educationalists). The basis of these programmes is to draw on European diversity and share, exchange, and pool experiences in the fields of education and training. The user groups for the design and experimental implementation of REM have been teachers undergoing further professional development and pre-service teachers in training. The members of the experimental network had, by and large, been institutions who had previously collaborated in conventional Erasmus intensive courses in the field of teacher education.
Professional education and development is increasingly taking place within schools rather than universities or colleges, but it is believed that teachers and trainee teachers still require good peer-peer and peer-tutor interaction. Due to time constraints and geographical dispersion, it is difficult for teachers and trainers to attend whole day (or days) training. European collaborations between institutions involving face to face meetings are time consuming and expensive for the participating parties. Firstly we had to clearly articulate what is valuable teaching and learning experience, and secondly, to develop and implement a specification for a telematic learning environment, which would enable a constructive design activity to take place over networks. REM was a collaborative activity whose motto was “when we create new tools we create new conversations”.
Aims of Project REM
REM aims to show how the training of teachers and other professionals can be supported by multimedia resources, which have been developed and made available on the Information Superhighway. Thus, the project will pioneer a “virtual university”. Specifically, the REM consortium is developing a set of multimedia learning and teaching resources; an online database of multimedia materials; an integrated telematic learning and development environment for learner and staff use; and tools for training, and the development and evaluation of teaching material. Project REM seeks to explore how professional development needs can be met by supporting innovative learning and training opportunities with emerging technologies. The two specific facets of technological support addressed by the project were collaborative conferencing activities and the use of a metadata rich pool of shared multimedia resources.
The Pedagogic Model
REM as a project has a theoretical underpinning, which is based on a number of ideas. This model has come from many years of consideration by a number of researchers and practitioners in higher and professional education. This research and reflection in itself provides a major basis for our understanding of user needs. We believe in active learning, which engages students in making their own meanings and understanding of the contexts they are in. We believe that human interaction is an important factor in learning, and that sharing ideas and collaborating are an essential components of learning and further professional development. We believe that learning is at its best when it is task based, and when students and professionals are engaged in a reflective exploration of an authentic task which builds new skills, knowledge and understanding for the student or professional. Garbinger and Dunlap (1995) define a “Rich Environment for Active Learning (REALS), which describes the basis for a comprehensive instructional system evolving from, and consistent with, constructivist philosophies and theories. Garbinger and Dunlap (1995) describe a system, which promotes study and investigation within authentic contexts. The system encourages the growth of student responsibility, initiative, decision-making, and intentional learning; it also cultivates collaboration amongst students and teachers. This system also utilises dynamic, interdisciplinary, generative learning activities that promote higher-order thinking processes to help students develop rich and complex knowledge structures; and it assesses student progress in content and learning-to-learn within authentic contexts using realistic tasks and performances.
The REM project also addresses the argument proposed by Kinchloe and Steinberg (1993) that modern analysis of our world understanding needs to establish cognitive theories that transcend deterministic Newtonian-Cartesian certainties. They postulate a need for post-formal thinking (in the Piagetian sense). According to Kinchloe and Steinberg (1993), the features of a curriculum that cultivate this thinking include epistemology – the exploration of the forces that produce what the culture validates as knowledge; pattern – the understanding of the connecting patterns and relationships that undergird the lived world.; process – the cultivation of new ways to read the world that attempt to make sense of both ourselves and contemporary society; and contextualisation – the appreciation that knowledge can never stand alone or be complete in and of itself.
REM technology and organisation addresses these issues, and enables such environment to allow choice, and a range of potential problems and solutions to be presented, with each problem examined from a variety of contrasting view points involving collaborative teaching and learning. REM provides a telematic environment for such learning experiences based on the partners experiences in teaching, and in particular on collaborative teaching in a European context on Erasmus intensive programmes. REM was not about telematic distributed courseware. It does not concentrate on conventional instructional strategies based on computer based tutorials (CBT), which while being well planned and exhaustive in their coverage on content of material, do not sufficiently contextualise the material to allow it to transfer in practice (see Collins et al. 1991), for there is evidence that skills gained in this way do not travel easily from one context to another (Clark and Vogel, 1985). Furthermore, we have still to see any evidence that the considerable investment which has been made over three decades into developing CAI for higher education in Europe has made any major inroads into teaching in Europe’s universities.
A Constructivist Model
There is a considerable body of literature, which gives form to the concept of constructivist learning. The roots can be found in the writings of Rousseau in the 18th century, in the pragmatism of John Dewey (1910), the developmental psychological traditions of Jean Piaget and Jerome Bruner (1961), and in the biological understanding gained from the work of Maturana and Valera (1992). In the context of learning technology there are a number of recent sources which describe the relationship between constructivism and learning technologies (see Jonassen, 1994). Learning or knowledge acquisition is an active and evolutionary process where knowledge becomes embedded in a rich web of existing knowledge and frameworks, and learning in a social and collaborative process of negotiation of meaning.
Based on these premises, we need to develop learning environments that have specific characteristics. It is proposed that students can take responsibility and show initiative in their learning (see Scardamalia and Bereiter, 1991). Students are given opportunities to “externalise” their knowledge in a variety of contexts that allow them to be self reflective - putting an interpretation on one’s own actions (Von Wright, 1992, p.61), that is, reciprocal teaching (Palincsar and Klenk, 1992). Students engage in activities, which generate ideas and knowledge (they are not consumers). Students engage in authentic learning activities that take place in the real world and are not abstracted from it. Students engage in assessment assignments that are authentic in real world contexts. Students work in collaboration with others, their ideas are under the review and in the social negotiation space of other learners and their teachers, and they, in turn, observe and reflect on the insights of others. These user-needs mean that REM should provide a telematic environment in which the learning experience of students maps onto these described activities.
The Educational Model
A particular formulation for the constructivist model of learning is the notion of cognitive apprenticeship. In a cognitive apprenticeship, students learn to emulate the activity of a master learner or knower (the Icelandic term for teacher “Kener” is useful here). In this case, the role of the teacher is to be that master knower with the learners as apprenticeships. Historically, the apprentice followed tasks set by the master craftsman, which eventually brought the apprenticeship to their masterpiece demonstrating their mastery. The specific roles for teachers referred to in the literature on cognitive apprenticeship include that of a modeller, a coach, and a scaffolder (see Collins et al 1991). The teacher as a modeller is an externaliser of their own expert knowledge. This is the material that is presented to stimulate and indicate to the learner what is to be understood. It is a presentation of the authentic activity. As a coach, the teacher makes decisions based on observation and reflection of students’ work, and provides feedback to the student on their performance, how they deviate from mastery, and how they can change their activity to be closer to the model. It is essentially a process of exchange of skill and technique. Coaching demands that the teacher sees the learner engaged in working. It is akin to the mathematics teacher, who needs to see more than the end result but also the working out, and the students’ progress through a problem-solving approach. Finally, scaffolding is the provision of a conceptual framework within which the learner can build their own understanding. The concept should be one of providing the support necessary to reach the next stage in development. It decomposes tasks as necessary for students to carry them out.
Tasks are set for learners, which are based on their current state of knowledge, and their potential development is activated through problem-solving tasks under the supervision of adult guidance or more capable peers. The process develops in such a way that the coaching and scaffolding gradually disappear allowing the learner to achieve the tasks unsupervised.
In terms of REM, the project needs to include materials that display “model” behaviour, and exemplify “best” practice with annotation and explanation. For coaching, there is a need to have a system that can monitor (asynchronously) the learner’s activity in the system, what they have done within the scope and context of authentic and generative learning activities. The telematic system can also provide “frequently asked questions and frequently provided answers”. REM also scaffolds a series of documents that guide students generically including planning documents, which stage their learning; documents that guide self and teachers evaluation by explicitly showing the criteria on which success might be judged on a variety of facets of any given task; templates for providing responses; and decision guides for particular activities. Further scaffolding involves the design and banking of documents appropriate for students at varying levels of development. Lastly we use scaffolding procedures to help sustain and develop reflective dialogue about these processes.
This paper will focus in detail on one such scaffolding process, the use of reciprocal reading structures to address the collaborative understanding of documents.
Implementing the REM Project
The aim of REM is to bring practitioners, and trainees together to solve problems, converse, and reflect together to create new educational resources together. There are three components, which are closely integrated. First, a space for rich structured conversation between the learners and trainers, which makes full use of the technology available. Second, a system for tutors, mentors, and coaches to observe and comment on the activities of the learners, and allow them to provide further structures and information to learners to support their learning. Finally, a well shared, and structured information and learning resources, which provides a framework and source for structured discussions and interactions. This is seen as a growing resource, created and added to by the participants as well as drawing in resources from outside. In this paper we focus on the conversation space.
REM makes use of a structured groupware environment based on the premise that learning is a conversational process of rich dialogue as a negotiation and exploration of shared meanings. However, there are significant issues raised by typical day-to-day interactions between users of email and computer-mediated conferences (CMC). The use of such technology needs to be better structured to be effective, and REM aims to achieve this. There is a large literature on the use of CMC for learning (Mason and Kay, 1990). However, there are problems with CMC as a form of discourse. Writers indicate that there is a need for quality in teaching in these environments. (see Salmon (1998), Jones, (1998)) Hilz (1986) has pointed out that “only those experienced in the world of computer conferencing should take a course using this medium” (p.103), and the need for structure in theses cases is important (Soby, 1992; Perkins and Newman, 1996; Paulsen, 1995). Grint (1992) has also pointed out the problem caused by the lack of cues and expectations of action by users.
Typical experience of many CMC interactions are a stimulus to discussion followed by a flowering of discussion, which reaches a crescendo and finally dies. Good teachers know that lessons need structure. Most lessons from the literature cited above are that CMC needs moderation or steering. Our anecdotal experiences in Bangor supports this. When we used a simple model of web delivered reading supported by responding to questions within and raised by the reading we found it was ineffective. Typically the "good" student provides "good" answers and the rest of the students are denied the ability to participate.
The approach taken by the REM project has been an evolution of systems suggested by Medina-Mores, Winograd, Flores, and Flores (1993). These systems are based on the notion that in collaborative workplaces, there are flows of work between the collaborators.
Computer supported co-operative work (CSCW) is closely connected with workflow and organisational processes. CSCW is concerned with tools, technologies, and protocols for automated support of group work practices. Many of these tools will be essential for enterprises in large networks. Most of the existing tools, however, are concerned with sharing some kinds of objects such as workspaces or decision-support systems, and not with the actual work processes in which these objects are useful. Workflow models and maps can help make the context for other CSCW tools explicit. It is intended to develop our concepts of workflows for learning, which make a significant variation from classic CSCW. These are often seen as virtual shared spaces (e.g. designers share a virtual drawing board or there is a classic integration of screen and audio-conference for synchronous co-working) or common actions on computer artefacts. REM’s focus is in the conversations for learning action between the learners themselves around a shared developing knowledge pool.
Workflow analysis and support are now being used in a wide variety of businesses and organisations (e.g. Agostini et al., 1993; Schael and Zeller, 1993; Fischer, 1995). In the REM project, which is based around conceptions of collaborative development of teaching across Europe, there is a workflow to support collaborative production of multimedia teaching materials. The workflow involves four key stages. Planning, which involves sub-processes like brainstorming, refining and defining the nature of the task, breaking down the task into manageable components and task allocation. Transforming or identifying information needs and selecting information, finding existing information and creating, structuring, and publishing new information. Evaluation and implementing the activity, evaluating impact, and identifying new needs and shortcomings, successful features, and adding activities to the resource base. Reinvestigation or following through from the evaluation process to iteratively go through the cycle again to redevelop successful practices. This is a model freely adapted from the work of Lehrer (1994) on the co-operative learning through the development of multimedia projects.
The workflow in the REM project is implemented currently in FirstClass (a Groupware product). However, the workflow is product independent of any given groupware system as long as there is flexibility in the product implementation. It is supported in the system by structuring using graphical user interfaces imposed on the software; organising specific conference structures; providing pro-forma stationery, which ensures a commonality in approaches to response to sub-tasks to aid direct collaboration; and direct access to additional support tools pertinent to the task. There are many potential workflows. One of the key contributions to the notion is that of Searle’s (1969) concept of speech acts. Searle (1969) identified over eighty different kinds of speech acts, which are part of dialogue systems. A typical business dialogue identified by Denning and Winograd (1996) is: Request-Negotiate- (fulfilment of request)-Perform-Complete. These are not complete models of workflow, and neither is Searle’s typology sufficient to describe all dialogue. However, they serve as a basis for starting to prototype the kinds of negotiation and learning that need to take place at a generalisable starting point for subsequent personalisation and localisation. Denning and Winograd (1996) note that “traditional workflow management is well suited to highly structured heads-down paper processing, but is not adequate for support of newer modes of managing work... a more useful blueprint employs a description based on language-action model”. We have hypothesised that the same is true for CMC-based learning environments.
It is these conversation structures that provide scaffolding in the first instance. It is also the direct observation of learners in the process of discussion, creation, and reflection, which allows the mentor to coach and model activity as appropriate in the process. The interactions of the group in the “cyberworld” are open to the inspection of the tutor.
We investigated a number of different strategies for managing conversational workflow across particular tasks. A key task in many situation is arriving at a shared perspective around key documents (see for instance Brown and Duguid, 1977 ) . Documents, indexes and
lists are more than delivery mechanisms to represent knowledge. They are powerful resources for constructing and negotiating social space. Brown and Duguid suggest that documents have the ability to form a community around them. From scholars, to hobbyists, to practitioner communities, to political underground movements, the placement of a marker in cyberspace is an effective method of establishing contact with like-minded people. By linking our own documents to anchors of other documents, we identify ourselves within a community. By publishing artefacts of our own making, provide scaffolding for peers who share a similar stage of development within a knowledge domain. At the same time, objects and responses to them offer anchors for others to appropriate (or not) in the formation of their own online identity.
Clearly in developing collaboration, consensus and identity, paying attention to specific documents and having an understanding of the understanding of others on the issues within the documents is important. Finding a workflow that is inclusive of all a collaborative group is therefore a priority.
Reciprocal reading is a technique described as a classroom activity by Palinskar and Scott, and has been successfully adapted by REM to be an on-line collaborative activity. It is a particularly useful activity when learner consensus and understanding of a particular document or issue is needed. It encourages deep reading, shared understanding and an exchange of perspectives on a given issue.
The process is based on the shared reading of some material. In the multimedia worlds we are not confined to using text as the material.
The stages in a reciprocal reading exercise are:
At any time the tutor can offer support to the group or individual. This may be in the form of introducing new material to the group ( harder material, easier material, extension material).
This exercise also needs good time sequencing for the participants. It is not totally asynchronous.
The implementation here in Bangor has not been under strictly controlled experimental conditions. We have been working with real teaching groups with varying experiences and background. These groups have not been screened, pre-tested, matched and neither is there a control group undertaking what could be a myriad of alternative teaching methodologies. We do not pretend to present water tight experimental results based on study of responses from a series of variables (Computer experience/gender/ mother tongue and facility in English) many of which we consider relevant to understanding the way that reciprocal reading may or may not succeed as a method. What is presented here is based on an analysis of what has happened and follow up interviews and discussions with groups who have tried the practice. The aim of these studies was not to validate the methodology with water-tight certainty, but to see if we overcome some of the issues we have encountered in earlier methods and to see what new or recurrent problems arise when implementing this method.
Quantitative data is given, however, as most of the feedback was solicited as an integrated part of the process itself the data is not quantifiable as it is dependent of the insight of the respondent to identify their own opportunities and problems.
Lessons from Practice
The exercise has been tried about 12 times with groups of students ranging from 7 to 18 (109 users in total). Some of the groups have repeated the exercise with a different document. The lessons we have learned from the practical implementation and talking to students include both comments about the surface features of using asynchronous learning like threading and synchrony, as well as their ability to engage with content and other students.
Some of the evaluative comments from participants are salient.
a) "Martin I'm lost"
"The biggest problems that I found whilst reading these messages was that i found it difficult to find martin's messages among the others, especially after a few more came in."
or more problematically
"It does feel rather disjointed to discuss these through so many threads in the conference - especially as XXXX keeps adding bits onto threads that I have already passed through. I just end up reading old bits and having to review the thread again to keep up with the conversation"
This was voiced as a concern from about 10% of respondents in their early use of FirstClass. We would infer that it is a problem for more participants however which lead us to undertake some remedial training action.
The use of the methodology requires strict adherence to threading conventions in a conferencing system If they are not strictly adhered to the whole system falls easily apart. In the subsequent iterations provided strong colour coded messages to use "reply" in response to the letter. This had some but not total success. Further iterations we have used facilities within the conferencing system we are using (SoftArc's First Class) to have a big button with "REPLY TO THIS" in the message header space.
Alternative strategies have involved starting with a training activity with a much simpler message structure requiring the use of REPLY before starting the reciprocal reading exercise. The user training method has proved very successful and no evaluation following user training has commented on being lost. However, there are comments on the problems encountered with threading are not unique to this situation ( see Lewis and Knowles 1997), and this particular work flow, that depends heavily on threading has not generated any more problems than typically encountered with more open ended exercises which have not depended on threading.
b) "I'm already behind"
How synchronous must asynchronous learning be? A typical comment from a proportion of students is:
“I'm already behind and I don't quite have the time, and the colleague who was going to do this with me is busy, and someone knocks on my door, and then …….,"
This comment would be typical of about 25% of users, and it should also be set against the 2 students who has gone through the exercise in a day without any feedback from their colleagues ie who completely missed the point of collaboration
The exercises have been constructed as part of a mainstream student activity, that timetable time has been set aside for it, and that time spent on this activity has replaced time that would have been spent on another activity. A principal difference is that the student is at liberty to manage his/her own time. There is no fixed point at which they need to be in the same room as other students. This is cited as an advantage for on-line learning. However, when there is an element of collaboration, having something to collaborate with, and being in pace with your collaborators is a key issue. Students who did fall behind were unable to easily catch up, as their contribution was inevitable of a qualitatively different nature in that they were more post-facto summations, than those who were in a near synchronous debate in a dialogue mode.
The complaints about threading and apparent disjoints, re-reviews and re writes quoted above seem to be less significant when the time scale is shorter and the exchanges are more like a dialogue. It is perhaps easier to keep track of who said want and when in a shorter time span.
The exercises took place between seven and 14 days of activity. Given our restricted sample the longer the activity, the more people became asynchronous. The 14 day activity had 9 participants and only two students kept in a genuine collaborative mode through the whole period, with daily postings. 5 students chose the participate in the first 2 days and then participate infrequently until the last two days. This would be our intuitive response. There is more opportunity for diversion from the activity the longer the time-scale. A highly structured activity like this needs to be highly synchronous to be successful
c) Poor quality of responses
Responses from students were often weak or inadequate in terms of their length. By example in response say to the note:
'Write down up to 5 points that need clarification for you in the National Curriculum IT Document."
(this refers to the statutory regulations demanded in teaching information technology in Wales)
Students responses were as terse and incomplete like :
"Who decides what level the pupils should be up to at a certain key stage?"
or a more questioning:
"How can make sure that all pupils have the same opportunities to attain their optimum level of competence.?
How can challenging work be set for mixed abilities within the Key Stages without modifying the National Curriculum
Why is the Curriculum Cymreig significant in the use of IT?
What if, as is true in many cases in schools, there are insufficient resources to enable the teacher to carry out the requirements of the National Curriculum: Information Technology?"
This clearly varied from student to student and it is noted that when the activity was undertaken both in the early part of a course or in the period before there was a training pre-exercise in the teachers opinion there were more weak responses (about 50% of responses could be classified as week in those circumstances).
There is clear need to highlight and cue the expectations from students in terms of the depth and extent of their responses throughout the exercise. One technological affordance is to include a specific space that anticipates a particular size of answer. The experience of using such approaches in text indicates that they do not always work (see Hartley 1994). Here there is a clear need for cognitive apprenticeship to be enacted. Students need to be given the expert model to follow and receive direct feedback on their personal performance in the activity. This is another case for introducing a smaller scale activity with more tutor feedback to cue expectations. The other feedback is of course from the learner's reaction to other learner's contributions
It is possible to say that on the evidence of the exercises run thus far that more students have an opportunity to respond and more importantly in a format to respond not based on the first person with the right answer. Students can feel that their contribution is meaningful, despite in many cases misgivings about the medium.
The typical evaluatory comments are like:
"Its a fun and "easy" system to get on with and I'm sure I am only at the start of the quest for knowledge so any top tips will be greatly appreciated any time, any place, any where, that's the taste of first class...."
"I think you (the tutor) should console yourself by remembering how bad and disappointing many f2f meetings are! What we are all looking for here is something better than that and I think it will take time to find ways of having productive discussions in these spaces. In the face of these high aspirations, we should seize on what does work rather than worry too much about what doesn't.
"I've enjoyed participating in a discussion forum about these issues and I've also enjoyed reading the range of views and writing styles around the various topics under discussion which continue to occupy attention for professional development. There seemed to be some convergence around some of the points, the time scale was workable and the technology wasn’t a barrier. I've also found it interesting to read about others' evaluations of the experience and hence gain a clearer sense of 'voice' from the other participants"
d) "in response to….."
An important question relates to what can be considered "collaboration" in CMC environments. What evidence can we have that demonstrates a collaboration is active. It is possible to say that reading the comments made by others is a form of collaboration at one level. The environment we are using does give us access to information of who has read what and when. This is very revealing, because it does indicate that even in conferences were there is a tail off in responses ( largely through falling out of synch), there is almost 100% reading of the postings by members of the group ( the lowest recorded was 80%).
Greater collaboration can be seen in evidence of use of:
b) requests for clarification;
c) rebuttal or agreeing with argument;
d) answering direct questions; and
e) other semantic patterns.
The structures we seek in our analysis of collaboration may include messages with structures which appear in stages 2, the initial response and 4 , the requests for clarification, or the method as described above like those below.
"I feel that the commonality in what friends are saying is that we are looking towards notions of communities who will exchange knowledge,……" (male, mature student, UK)
Response to expectations
"I liked Jonathon's "how to facilitate active participation" so I'll put that as well"(female, mature student, Spain)
Response to clarifiaction:
"I agree - there appears to be very little of substance in the introduction, certainly nothing to tell us what really makes a good teacher. "(male, UK, undergraduate)
However these make for obvious inclusion as "responding to others". Other evidence needs to be sought which does not come so well labeled. One needs to look to see if there begins to be evidence of tacit convergence of thought or understanding where the divergence is.
We have messages like
"I think the common issues are :
1) 'Active' education rather than 'passive' teaching
2) Make what's being taught relevant to the learner's life
3)Critical reflection (still not exactly sure what this is but seems to be buzz phrase at the moment so I'll join the bandwagon)" (male teacher , UK)
and in response to this we have:
"One needs perhaps to have a sense of the educational values and viewpoint which one is arguing from as a way of anchoring and rooting one’s practice and to have the capacity to be critical about one’s own practice. Being critical would be being continually looking for opportunities in a situation., opportunities for professional development and opportunities for listening to and managing diversity. Critical thinking would be seeing a situation as something to be engaged with in a conversational dialogue".
(Female, teacher, UK)
Further examples are:
“How do you guard against replication when some topics recur across key stages? The topics are pretty broad and open to several interpretations. Does the teacher have the authority to decide?”
(Mature female postgraduate trainee teacher UK)
"Perhaps the over-arching common issue is how we get teachers/trainers themselves (ourselves?) to be strong learners."
(Female, teacher, UK)
"Reading through the collected texts I found interesting. My feeling about it was that having reached a point where…."
" gain a clearer sense of 'voice' from the other participants."
These statements can be found across a range of the discussions and provide evidence that there is a positive attempt to engage with others in genuine collaboration on issues.
We have clear evidence that more people post and read others' postings (80% minimum) in this structured environment. There is some evidence that people post as a result of what they gave read other people have said. There are clearly more invitations to post than in a simple read and respond as the formulation of the questions that the learner is responding to demand a personalised response to an issue.
We will continue to develop a body of experience in the use of dialogue structures and we will be able to use a range of techniques to analyse these dialogues. However there is within the notion of group annotation and questioning of texts the possibility if creating methods for collaborative composition of texts. The method builds up the identities within collaborative learning groups that Brown and Duguid suggest surround electronic structured documents. In fact the totality of the discussion (including the original document) begins to approximate fairly strongly to their notation of an annotated living document produced by a community of practice.
REM is a project that aims to support constructive collaborative learning using telematic technologies. To be constructive we felt that there were certain characteristics required in the system such a co-production of knowledge, student responsibility for their knowledge, self evaluation, generative activity, reciprocal teaching and social negotiation. To achieve this we required good access to resources and a communications environment that supported high level interaction between learners. Previous experience with both CMC and Web based learning, in line with the literature, indicated that this is a problem area. The key problems were reluctance to participate, the intellectual argument being completed by a few individuals, and the lack of opening up of real debate.
To find a way of using the available technologies the project adapted the concepts of speech acts and work flows proposed for CMC by Winograd and Flores. A number of workflows have been used to try and encourage collaborative activity. Different activities clearly need different workflows. One specific activity that is common in many learning situations is the need to develop a shared consensus and understanding of key documents. We have adopted a process of reciprocal teaching and reading of documents described by Palinkscar(1992) as the basis of one such work flow. Reciprocal reading requires learners to read documents, develop questions about the documents and to pose those questions to their co-learners.
We encountered a great deal of success in the application of the method, particularly in relation to the reading of other learner's postings, and clear evidence of learners actually responding to other learners views and opinions. These were inferred by quotation, requests for further information, rebuttal of arguments and answering direct questions. Further confirmation of collaborative activity was inferred by analysis of specific exchanges and the personal feelings of the learners involved.
The method is not without problems. It addresses some of the issues of inclusion and structure that have been problematic in the previous experiences of the author and those cited above. There are some issues that are capable of resolution in software design (eg threading and navigation of threads), and other issues that require participant discipline and moderator action. These are largely about synchrony in asynchronous environments and cueing the expectations about responses. Where authors have pointed out the need for "good moderation", this can be an interpretation of the remark. However the framing and structuring of questions, and their sequencing is also a key. The reciprocal reading technique provides one such framework. There are others (see http://d2k.bangor.ac.uk/online/online0.html) and the work of Paulsen. The role of different structures and their effectiveness in building learning through collaboration needs further experimentation. Thinking that a high quality computer mediated conference/collaborative learning system is sufficient to generate good collaborative dialogue is clearly an inadequate model.