The Hypermedia Conversation: Reflecting Upon, Building and Communicating Ill-defined Arguments
Milton Campos, Université de Montréal

Abstract
This paper shows how an argumentation process carried out on a hypermedia conversation in a post-secondary course led to collaborative knowledge-building and conceptual change. Conditional reasoning was assessed through a technique specifically developed to analyze transcripts of online asynchronous conversation. Results suggest that when proper teaching strategies are present, knowledge building and collaboration are achieved. The study also suggests directions for new forms of representation and of meaning-linking in asynchronous conferencing systems.

1. Introduction
This paper shows how reasoning upon questions and hypotheses built collaboratively can lead to knowledge building and conceptual change in hypermedia conferencing systems. We studied one sub-conference of a biology mixed-mode undergraduate course on mammals whose topic was evolutionary biology. The professor applied Socratic maieutic by posing online questions to the students, reasoning upon the answers, and raising new questions in face-to-face encounters.

Results demonstrate the knowledge building process that led some of the participating students to achieve high order reasoning and conceptual change. We understand high order reasoning as the ability to logically operate on the modules of content, and conceptual change as the ability to correct an argument in logic that was flawed because its premises were inconsistent with the conclusion.

The circumscribed collaborative knowledge building found in the students' learning processes, reified by the transcripts as objects of knowledge (Popper, 1994; Bereiter, 1994; Scardamalia, & Bereiter, 1996), was assessed by transcript analysis. The meaning implication transcript analysis technique was applied to identify the chains of interconnected meaning implications, and to show how collaborative conditional reasoning and hypotheses formulation evolved in the hypermedia conversation (Campos, 1998).

2. The Study
Our goal was to obtain circumscribed qualitative evidence of collaborative learning in a hypermedia asynchronous conversation that was triggered by a teaching method using interrogation and reasoning about scientific dilemmas.

2.1 The Course

Seventy-three persons, including students, the professor and the teaching assistant , took part in the course, delivered in the fall semester of 1998 in a French speaking Canadian university. It was a traditional face-to-face course in which a networked-enhanced component was included: parallel non-mandatory online conferences.

Note : All persons involved provided their written full informed consent. Identity and gender of the subjects are concealed to preserve anonymity and confidentiality.

The online teaching strategy was designed to trigger discussion through questions related to unsolved scientific dilemmas in mammalogy. Different hypotheses explaining those dilemmas, formulated by biology theorists, were presented to the students by the professor through traditional face-to-face lecturing. The students were then encouraged to discuss the topic in the hypermedia conferences and propose alternative solutions for the dilemmas by reasoning upon the evidence and using their knowledge about mammals and biology.

The instructor, a full professor who taught this course without significant changes for more than 20 years, was not knowledgeable about technology, did not want to get involved in learning how to use computers, and did not participate directly in the hypermedia part of the course. The hypermedia component was introduced experimentally. A graduate student, expert in educational technology, worked as an exclusive online teaching assistant, responsible for posting the questions and providing instructions to the students. No facilitator or moderator was assigned to mediate the conferences. The students were left free to take charge of the discussions.

The professor marked the messages according to the level of plausibility of the hypotheses formulated and the quality of argumentation. Those marks were neither made available to the students nor were they taken into consideration for evaluation purposes. The professor marked the messages to help him to prepare the face-to-face discussions, and later they were made available for research purposes. Together with the marks, there were, sometimes, written comments. The following marks were given:

• No mark - no meaningful contribution,
• ¼ - minor contribution,
• ½ - reasonable contribution, and
• 1 - outstanding contribution (plausible hypothesis and/or argumentation).

Later, the students' contributions were discussed in class. The professor commented on the most plausible solutions presented, raised further questions, and discussed the scientific state of the dilemmas.

2.2 Procedure

2.2.1 Hypotheses
We hypothesized that if implications among meanings could be found in arguments built across different messages, then the quality of the premises and conclusions marked "1" could be explained by previous conditional contributions given by others. In addition, it would indicate that the high order reasoning found in the knowledge building process (attributed by the professor through the expert marking system) was achieved through collaboration. Furthermore, such results would provide qualitative evidence that the process of interrogation and reasoning about scientific dilemmas would be an effective method for teaching evolutionary biology.

2.2.2 Research Instruments
The following instruments were used in the research:

• Transcript analysis,
• In-depth interview with the professor and the teaching assistant, and
• Analysis of the comments written in and the marks given to the messages provided by the professor.

2.2.3 The Research Process
After reading the texts of all conferences, we identified all meaning implications that sustained the arguments found and chose, for a closer study, an excerpt of one sub-conference discussing dilemmas related to characteristics of mammals. The excerpt

• was part of the most interactive of all sub-conferences both in terms of implications among meanings and in terms of a reasonable (although poor) use of the threading feature of the conferencing system used (Virtual-U),
• had a number of messages with high marks, and
• discussed bone features of mammals, specifically, the evolution of the number of phalanges.

Figure 1. The conferencing system used, Virtual-U, allows users to sort messages by threads.

External links to the maker of Virtual-U software: Virtual Learning Environments Inc.

2.3 Meaning Implication Analysis

2.3.1 Introduction
We focused on backtracking the meaning implications between the arguments and hypotheses marked "1" of the chosen excerpt in order to understand how they were built (identify knowledge building), and to examine whether they had been gestated in previous messages (identify collaboration). We studied six messages. Four students participated in this part of the discussion.

2.3.2 Procedure
The first step of the meaning implication analysis was an attentive reading in which all conditional words (like if, then, would, could, might, perhaps, etc.) of the text messages were highlighted. It is important to note that, in the case of courses with multiple conferences (as the one being presented here), all messages (of all conferences) were chronologically organized. We chose to ignore the boundaries of specific confererences because people's meanings could travel from one conference to another.

Afterwards, we examined all phrases in which conditional words occurred and evaluated whether their occurrence really indicated that conditional reasoning was present. In addition, phrases in which no conditional words were found, but which seemed to have a conditional meaning, were examined. As we applied meaning implication analysis in our examination of the texts, we found that, sometimes, the use of a conditional word did not mean, necessarily, that the person was making use of conditional reasoning. Conversely, a phrase in which conditional words were absent did not mean that conditional reasoning was not present. However, in most cases, conditional words indicated conditional reasoning.

The third step was to make links between the meaning implications of a given phrase and those found in previous phrases. Through this process, we were able to build a chronological map of inter-connected meanings. The more ill-defined an argument was (in other words, the less coincidence between logical rules and the facts found in a group of conditional phrases with premises and a conclusion), the less inter-coder reliability was found. The explanation for this is simple: many words have multiple meanings, and context is not always enough to guarantee that one of the meanings of a given word (interpreted by the coder) is necessarily the one intended by the writer. This phenomenon is quite well-known in cognitive science literature (for a review, see Gibbs 1994).

Finally, and when applicable, we applied truth and falsity values to the conditional structure of the phrase to evaluate whether conclusions were derived logically from the premises. However, we found, in most cases, logical rules do not apply at all in ill-defined domains, especially in hypermedia written conversation, (although it is possible they may, as in the case of syllogistical reasoning).

2.3.3 The Chain of Thought
The messages, sequentially presented here, had the following characteristics:

A new window will open up : refer to the Data by clicking here

1st message - Message number 4 was published with the following question: Apply the three elements method that determines the function of a character to try to explain why we observe phalange reduction. The professor's question requested memory retrieval of the method (presented in a face-to-face lecture) to guide reasoning upon a specific unsolved evolutionary dilemma.

2nd message - Message number 7 was written by student A without making use of the threaded feature. He/she opened a new tree (nobody continued it later). The student retrieved correctly the method, and built an argument with three meaning implications, one related to the other. The first two were the premises. They led to a third premise and a general conclusion. They were all chained to each other, and linked to the question posed in Message number 4. Student A defended the idea that reduction of phalanges occurred either due to locomotion of mammals during long periods or due to increase of speed. The professor considered the quality of argumentation reasonable (mark: ½).

3rd message - Message number 8 was written by student B without making use of the threaded feature. It was explicitly built upon message number 7. Student B seems to agree with the three elements method presented by student A, but, in fact, she/he changes two elements and contests the third. By doing this, student B contests the whole argumentation of Student A. Student B reasons that, on the contrary, the reduction of phalanges allowed mobility, and that bigger mammals had the tendency to have fewer and not more phalanges. The refutation was marked "1" by the professor.

4th message - Message number 20 was written by student C without making use of the threaded feature. It was explicitly built upon message 7 by Student A and message 8 by student B. Student C agrees with the method presented by student A, and tries to refute student's B argument. The argument gets a negative written comment from the professor. The professor writes that Student C is just repeating what he/she said in her/his lecture: reduction of the number of phalanges would be due to an evolutionary agreement between solidity and flexibility. Student's C refutation is inconsistent because (1) some conclusions just confirm Student's B hypothesis (that C tries to dismantle) concerning the reduced number of phalanges in mammals like the cow, and (2) C asserted that it would be logical that human beings should have more phalanges than cows because of the use of hands. The professor writes another negative comment concerning this statement because the number of phalanges is stable across species, and marks ¼ for the contribution.

5th message - Message number 37 was written by student B without making use of the threaded feature. It contests assumptions of message number 20. The answer starts with an attempt to build upon the plausibility of Student's C hypothesis. Nonetheless, the hypothesis is contested. Student B adds that to discuss the number of phalanges the number of fingers and toes should also be discussed. In what we consider evidence of conceptual change, Student B says that C was induced by his/her own error because she/he had followed his/her line of thought: B's previous contribution (message number 8), marked "1" by the professor, had an aside pointing out exactly this error, considered by her/him misinterpretation of the evidence. Student B suggests the need of an inter-species observation, inter-fingers / inter-toes observation, and comparison between hands and feet. The professor marked ½.

6th Message - Message number 47, written by student D, is a threaded answer to message 37. Student D clarifies student B's doubt about the similarity of the number of phalanges of hands and feet by pointing out their different use. D argued that the reduction could be found in all mammals, although there were exceptions in the organization (cows and horses among others). This organization would correspond to an evolutionary agreement that would bring efficacy to the use of hands by mammals. Because mammals are young in evolutionary terms, timing was not enough for those characters to develop, reasons student D. The professor marked "1".

2.4 Summarizing results

The meaning implication analysis shows that hypotheses were built upon not-plausible hypotheses formulated in previous messages in both messages marked "1". The plausibility indicates high order reasoning because the students who wrote those messages were able to come up with hypotheses to solve the ill-defined problems that were pertinent to the topic. They considered (1) the premises (evolutionary facts guiding reasoning) and (2) the possible conclusions (inventory of reasons to explain the facts: the features of the mammals). In addition, message number four indicates a process of conceptual change in which student B realized that her/his own previous message was based upon a misinterpretation of the evidence (highlighted by the professor in a comment) and he/she decided to re-build his/her own argument. Student B accommodated her/his own learning by changing the previous conceptualization of the problem. The fact that his/her last message was marked ½ does not change the fact that conceptual change occurred. The analysis clearly indicates that knowledge building was achieved through collaboration.

2.5 Implications

2.5.1 Course Design
The study suggests that the teaching strategy, the Socratic maieutic, might be strongly related to the results. The importance of interrogation in educational processes is common sense. For Socrates, interrogation was a teaching method (Mondolfo, 1972) through which problems about the empirical world could be responded to inductively (Jaeger, 1987). The professor posed questions to the students; they reflected upon the answers provided, and explanatory arguments were built and communicated both in the hypermedia conferences and in face-to-face encounters. Questions were challenging, and are still to be answered by evolutionary biology. This aspect seemed specifically attractive for the students because, as novices, they were given opportunities to exercise their own sense of how to apply background knowledge to scientific dilemmas, to solve actual, ill-defined scientific problems, and to provide a contribution to the discipline. The success of the strategy suggests that maieutic questioning is an appropriate teaching method for exploring scientific dilemmas.

2.5.2 Conferencing systems design
An attentive analysis of the chain of thoughts shows that the agreement of built-in threads and meanings does not emerge by itself in hypermedia conversation. Our analysis of the way the students linked their chain of thoughts shows clearly that the threading feature is neither necessary nor sufficient for knowledge building through collaboration. This finding is consistent with previous studies (Campos, 1998) and is a strong indicator that hypermedia conferencing systems still need further development and increased flexibility. Their underlying formal structure should be adapted to the way we use natural language to mold our symbol systems and assimilate the ideas of others, and thereby construct knowledge. The architecture of conferencing systems can be enhanced if it were created to mirror more closely the functioning of our neural systems. A reply-to-many feature could be a first step to allow more interaction (Campos, 1998). An interactive helping tool might be another important feature to guide participants in the use of threads. An example of such a tool is a set of pop-up menus that respond to each action of the user with questions about what he or she wants to do in the conferencing system. The possible answers would be represented as a number of links among which the user might choose one, leading to a place in the system in which the user could satisfy his or her needs.

3 Conclusion

The meaning implication analysis shows how conditional reasoning governed the inferences students made upon the contribution of their peers. In addition, the way the students built their chain of thoughts shows clearly that the threading feature, as it is today, is neither necessary nor sufficient for meaning linking. Furthermore, the study suggests that the Socratic maieutic, as a teaching strategy, might be strongly related to the results.

Thus, we suggest that when students are free to take charge of their own learning processes without any of the award systems found in pedagogical behaviorism, collaborative knowledge building is achieved. The fact that the professor did not participate at all in the discussion is additional indication of the inner learning possibilities of hypermedia conferencing. In addition, it seems that the mixed-mode opportunities enriched students' social interaction and learning (Laferrière, Breuleux, & Campos, 1999). The teaching and the learning processes of this mixed-mode course seem to show that when proper teaching strategies are present, knowledge building and collaboration are achieved (Campos, & Harasim, 1999).

4. References
Bereiter, C. (1994). Constructivism, socioculturalism, and Popper's world 3. Educational Researcher, 23 (7), 21-23.

Bereiter, C., & Scardamalia, M. (1993). Surpassing ourselves. An inquiry into the nature and implications of expertise. Chicago: Open Court.

Campos, M. N. (1998). Conditional reasoning: A key to assessing computer-based knowledge building communication processes. Journal of Universal Computer Science [online serial], 4 (4), 404-428. Available: http://www.iicm.edu/jucs_4_4/conditional_reasoning_a_key/paper.html (October 21, 1999)

External links to the reference: Campos 1998

Campos, M. N., & Harasim, L. (1999). Virtual-U: results and challenges of unique field trials. The Technology Source [online serial], 6. Available: http://horizon.unc.edu/TS/vu/1999-07.asp (October 21, 1999)
External links to the reference: Campos 1999

Gibbs Jr., R. W. (1994) The poetics of mind: Figurative thought, language, and understanding. Cambridge (NY): Cambridge University Press.

Grize, J.-B. (1996). Logique et langage [Logic and language]. Paris: Ophrys.

Jaeger, W. (1987). Paideia: Los ideales de la cultura griega [Paidea: The ideals of Greek culture]. Mexico: Fondo de Cultura Economica. (Original work published 1953)

Laferrière, T., Breuleux, A., & Campos, M. (1999). L'évolution des métiers et des formations dans les nouvelles méthodes de production des connaissances. L'apprentissage en réseau, une réalité pédagogique à définir [The evolution of professions and professional formation in the new methods of knowledge production. Networked learning: a pedagogical reality to be defined] .Paper presented at the colloquium Initi@tives99, Universités Virtuelles: Vers un Enseignement Égalitaire [Initiatives, Virtual Universities: Towards an Egalitarian Teaching], Edmunston, NB. [Online]. Available: http://www.aupelf-uref.org/initiatives/colloque/ (October 21, 1999)
External links to the reference: Laferrière 1999

Mondolfo, R. (1972). Socrates. São Paulo: Mestre Jou. (Original work published 1959)

Popper, K. (1994). Knowledge and the body-mind problem. In defense of interaction. London: Routledge.

Scardamalia, M., & Bereiter, C. (1994). Computer support for knowledge building communities. The Journal of the Learning Sciences, 3 (3), 265-283.

5. Acknowledgements
The Canadian TeleLearning - Network of Centres of Excellence and the Université de Montréal funded the present study. I would like to thank Dr. Linda Harasim for making it possible, the students and the professor who provided the materials and their informed consent, Sandrine Turcotte (Université Laval, Canada), Dr. Thérèse Laferrière (Université Laval, Canada), Jean Gunderson (Douglas College, Canada) Sylvia Currie (Simon Fraser University, Canada), and Dr. Brian Fisher (University of British Columbia). I would also like to heartily thank the theoretical guidance, support and encouragement of Dr. Jean-Blaise Grize (Université de Neuchâtel, Switzerland).