R-U-Typing-2-Me? Evolving a chat tool to increase understanding in learning activities

Abstract Very often, when using a chat tool where more than one participant is talking simultaneously, it is difficult to follow the conversation, read all the different messages and work out who is talking to whom about what. This problem has been dubbed BChat Confusion.’’ This article investigates this problem in debate sessions in an online university course. Chat Confusion has been singled out as the main limitation to using chat in educational activities. Confusion needs to be reduced for understanding to increase, making it easier to track what is being discussed during a learning activity. This study investigated the phenomena responsible for causing this confusion. A version of the Mediated Chat tool was developed for each problem identified and was subsequently tested in online courses. This article describes the Mediated Chat development process, the problems identified, and the results obtained from the case studies. p. 1

in a chat session with various participants talking at the same time, situations occur that prevent the conversation from being followed with ease. This problem has been dubbed BChat Confusion’’ (Pimentel, Fuks, & Lucena, 2003; Thirunarayanan, 2000) and is referred to in the literature as Bchaotic flow of conversation,’’ Binteractional incoherence’’ or Black of coherence and mutual understanding’’ (Cornelius & Boos, 2003; Herring, 1999; McGrath, 1990; O_Neill & Martin, 2003). p. 2

The term groupware, coined by Johnson-Lentz and Johnson-Lentz (1982), refers to computer applications designed to support collaborative work. Developing group- ware requires an understanding of collaboration. Collaboration, from the Latin co [together] + labore [labour] + action, means the action of working together, the accomplishment of common tasks undertaken by two or more people. Collaboration has been investigated in this research based on the 3C Model, which highlights that a group has to establish adequate communication, coordination and cooperation in order to collaborate. p. 2

Communication, communicare [to make common] + action, means the action of making common, to exchange messages for the purpose of mutual understanding, to converse, to dialogue. During collaboration, members of a group normally communicate towards action: they negotiate, make decisions and reach agreements (Winograd, 1989). A group contains people with different viewpoints who can supplement individual understanding (Gerosa, Pimentel, Fuks, & Lucena, 2005). p. 2

Coordination, co [together] + ordinare [order] + action, means the action of disposing of something according to a particular order and method, to organize, to arrange. The coordination of collaborative work aims at organizing the members of p. 2

the group so that the agreements reached through negotiations are realized in the right order and timescale, reaching their objectives within their anticipated limitations. It also aims at ensuring that the effort put into communication and cooperation is not wasted (Raposo, Pimentel, Gerosa, Fuks, & Lucena, 2004). Cooperation, co + operare [operate] + action, means the action of operating together. Members of the group act in conjunction on shared objects within a shared space to perform tasks defined and organized during coordination. In cooperating, individuals need to communicate to renegotiate and make decisions on unforeseen situations, reinitiating the cycle of collaboration. The 3C Collaboration Model, originally proposed by Ellis, Gibbs, and Rein (1991), has been used to analyze, classify, and develop groupware (Baker, Greenberg, & Gutwin, 2001; Ellis, 2000; Laurillau & Nigay, 2002). Groupware applications can be classified according to the degree of support given to communication, coordination and cooperation, as can be seen in the triangle illustrated in Fig. 1 (Borghoff & Schlichter, 2000; Teufel, Sauter, Mu ̈hlherr, & Bauknecht, 1995). As shown in Fig. 1, the applications closest to the communication vertex are classified as Communication Tools—applications that aim at establishing the exchange of messages among members of a group with a view to argumentation, negotiation, decision making, etc. p. 3

Communication tools can be organized into two large groups according to the time in which communication is established: synchronous, when the sent message is received instantly; and asynchronous, when the sent message is received at a later moment (DeSanctis & Gallupe, 1987). Lately, the expression Bquasi-synchronous,’’ coined by Garcia and Jacobs (1999), has been used to distinguish Btrue’’ synchronous dialogue, such as face-to-face conversation, and full-duplex phone conversation and videoconference from typed-text conversations, such as chat p. 3

and instant messaging (O’Neill & Martin, 2003; Zemel, 2005). For the remainder of this text, synchronous stands for both: synchronous and quasi-synchronous. p. 4

Analyzing a typical chat tool (Fig. 2), three main components can be identified: an area used to type the message, enabling the user to communicate with other participants (a Communication support); a list of participants, indicating who is connected and available for conversation (a Coordination support); and an area presenting a record of sent messages (a Cooperation support). p. 5

Communication Coordination Cooperation Language Transmission Size and Quality Dialogue structure Categorization Topic Session Access Availability Roles Turn to speak Message frequency Message visibility Addressing Indication of turn-inprogress Evaluation Session record Pre-formulated Languages generally used to establish communication: textual, spoken (audio), pictographic (images and animations) and gestures (video and avatars). Message transmission is intermittent (after the sender formulates the entire message) or continuous (continuous transmission of video and audio, or character-by-character as the message is being formulated). Restrictions on the size of the message limiting the quantity of characters (text) or the message_s duration in seconds (video and audio). Video and audio quality is reduced for transmission in the Internet. A linear dialogue structure is usually adopted in synchronous communication tools: one message presented after the other in chronological order. Other forms of structuring the discussion: hierarchical (tree, threads) or in network (graph, maps). Labels for characterizing the messages, such as: type of speech (whisper, speech, cry, question, reply, agree, disagree etc.); type of discourse (direct or indirect), type of emotion (happy, normal, angry) etc. Topic to be discussed Length of time for duration of chat Who and how many people can take part in the chat Availability of participant: connected, absent, busy, etc. Definition and attribution of roles: Operator, Mediator, Moderator etc. Who can speak at a given moment Limit to the quantity of messages in an interval of time Public (visible to all participants) or private (restricted to two participants) Indication of message recipient Information that the participant is formulating the message (before its transmission as one block) Qualification of messages, participants or session Storage, recovery and display of published messages Messages that are pre-formed and shared by participants Springer messages to be exchanged during the conversation p. 6

AulaNet (Filippo, Fuks, & Lucena, 2005; Lucena et al., 1998) is an environment based on a groupware approach for teaching/learning on the Web that has been being developed since June 1997 by the Software Engineering Laboratory of the Catholic University of Rio de Janeiro (PUC-Rio). The AulaNet environment is freely available in Portuguese, English, and Spanish versions at http://groupware. les.inf.puc-rio.br. p. 7

Over the six years (12 semesters from the first semester of 2000 to the second semester of 2005) in which the Mediated Chat tool has been used to hold debates in the ITAE course, participants frequently demonstrate their enthusiasm for this Bdifferent and interesting’’ activity, although they also often find the conversation to be confusing. It is interesting to note the terms used by participants to describe the problem: Bconfusion,’’ Ba mess,’’ Bturmoil,’’ Bbabble,’’ Bchaos,’’ Bpandemonium,’’ Ba frenzy,’’ Ban uproar,’’ and so on (the texts transcribed in this article were originally produced in Portuguese and translated into English—the originals can be obtained from the authors). p. 8

Chat Confusion. The main factors pointed out by interviewees for the sources of this confusion were: the large number of posted messages and continuous screen scrolling, which made it difficult to read all of the messages; the excessive number of learners and mediators (an average of 19 people were involved in each debate); and parallel conversations mixing up messages on different topics and leading to participants Bmissing plot lines.’’ p. 8

Learners also reported developing strategies to follow the debate, such as: focusing on messages sent by the moderator and mediators, by oneself, and by preferred interlocutors; trying to focus on one subject at a time; and trying not to repeat what others had already said, etc. The use of these strategies shows that, over time, participants acquire experience and improve their participation, making it possible to navigate, and somehow tolerate, the confusion. It seems that, in the literature, Chat Confusion is overrated, given that in many papers it is not taken into account that participants are aware of the potential for misunderstandings and therefore develop strategies to manage turns and threads, thereby producing a coherent conversation. p. 8

avoid being the passive victims of technology (Cornelius & Boos, 2003; Herring, 1999; O_Neill & Martin, 2003). p. 9

HyperDialog: Conversation threading to avoid Co-text loss Co-text Loss occurs when a participant is unable to establish the thread of the conversation; when a participant is unable to identify the earlier message to which a particular message is responding. BCo-text’’ designates surrounding text written before or after a statement, and provides elements towards understanding it. It differs from Bcontext,’’ which designates textual and extra-textual factors, such as the situation in which the text is produced or the reader’s presuppositions (Crystal, 1985). p. 9

This research into co-text loss is based on hypertext literature. The identification of the problem and the design of the solution were inspired by the analogy between the non-linearity of a chat session and the non-linearity of a hypertext and the related BLost in hyperspace’’ problem (Conklin, 1988). p. 10

Analyzing the organization of the conversation in the debates of the ITAE 2000.1 edition, where Mediated Chat version 1.0 was used, it was found that the text resulting from these sessions is predominantly non-linear: only 20% of messages refer to the immediately preceding message and, on average, the messages continue a conversation with a message located 5 or 6 positions earlier. The topics are also not discussed linearly, since the subjects are discussed in parallel and tackled alternately in the sequence of messages with topical confluence taking place (Pimentel, Fulks, & Lucena, 2003). The low level of linearity in the chat session was identified as one of the main causes of Co- text Loss. p. 10

In order to reduce the problem of Co-text Loss, the HyperDialog tool was developed (Pimentel, 2002), p. 10

In the case study conducted in 2001.1, the HyperDialog tool failed to lessen Co- text Loss, as Fig. 4b shows. In part, losses continued to occur because participants p. 11

committed too many mistakes in establishing the thread between messages (7.5% of messages were either not threaded or wrongly threaded). p. 12

Although message threading has the potential to solve the problem of Co-text Loss, the mechanism implemented in the HyperDialog tool introduces new problems in the group’s communication, coordination, and cooperation. In terms of communication, the conversation becomes unsuitably more formal, since the participant has to make explicit the message to which he or she is replying. In terms of coordination, the message tree disperses the focus of the participants along different conversational branches, making coordination of the debate even more difficult. And in terms of cooperation, the display of messages in two main views (chronological and associative) and the recovery of the message’s thread in a separate window make the HyperDialog interface much more complex, introducing problems in the shared space. The results obtained from the use of HyperDialog corroborate the results obtained with the use of the Threaded Chat tool (Smith, Cadiz, & Burkhalter, 2000), whose users declared it to be worse than a typical chat tool. On the other hand, the preliminary findings with the use of Academic Talk (McAlister, Ravenscroft, & Scanlon, 2004) showed that the argumentation process was more coherent than when using a non-threaded chat tool. p. 12

In order to carry on using threads without introducing coordination and cooperation problems, the revised proposal is similar to the one implemented in ConcertChat (Mu ̈hlpfordt & Wessner, 2005). The mechanism implemented in Mediated Chat 6.0 only shows the chronological view displaying arrows between related messages. p. 12

Mediated Chat 2.0: Conversation techniques to avoid interruptions In the ITAE course debates, a pre-selected learner performs the role of moderator and is responsible for coordinating the debate. Until the ITAE 2002.1 edition, the moderator’s main function was to present topics related to the seminar to be p. 12

discussed by the learners (Fig. 5). Based on an analysis of the records of these debates and interviews with participants, it became apparent that the moderator frequently has difficulties in coordinating the conversation. When the moderator is unable to conduct the debate adequately, the discussion may become highly confused, appearing unproductive and pointless. p. 13

With the aim of facilitating and systemizing coordination, a social protocol was defined in which more structured stages are established for the ITAE course debates (Fig. 5). In this dynamic, the debate is organized into three parts, each one discussing a question previously addressed in the course’s seminar. The moderator presents the question and each learner, in alphabetical order, sends a comment on the question. All the learners then choose a comment to be discussed freely. After discussion of the selected comment, the learners close the discussion, presenting their conclusions about what was discussed. This dynamic is repeated for each seminar question. Overall, this format allows a clearer definition of the debate’s objectives and how the participants should be coordinated in order for these objectives to be attained (Pimentel, Fuks, & Lucena, 2004). This dynamic was implemented from the ITAE 2002.2 edition onwards. Comparing this edition with the preceding ones, it was observed that manifestations of Co-text Loss were cut by half. This result indicates that, by itself, the use of a more structured dynamic makes the conversation much less confused. p. 13

Scripting of chat session p. 13

The Mediated Chat 2.0 tool was developed (Rezende, 2003), as seen in Fig. 6, with a set of conversation techniques for specifying who can speak at any given moment (turn to speak): Free Contribution, in which all the learners can speak at any time; Circular Contribution, in which learners are organized into an ordered p. 13

queue and allowed to send one message apiece; Single Contribution, where each learner can send just 1 message and there is no specific order; and Blocked, where only the mediators can send messages while learners cannot. p. 14

Like in other structured chat tools, rules for the interaction process are implemented for improving coordination and coherence (Lonchamp, 2005). p. 14

The case study conducted in the ITAE 2002.2 edition showed that the number of interruptions remained practically unchanged when Mediated Chat version 2.0 was used, as Fig. 7 shows. Based on an analysis of these sessions, modifications to the conversation techniques were identified that could reduce the occurrence of the interruptions that were still taking place (Pimentel, Fulks, & Lucena, 2004). The need was identified to overcome exceptional situations that occur during the implementation of techniques—for example, in Circular Contribution, the need was identified to skip p. 14

the turn of those participants who had no messages to send. The need to implement new techniques was also identified, such as in Mediated Contribution, in which the mediator authorizes or cancels the publication of sent messages. The conclusion of this experiment was that the use of a well-structured dynamic organizes the debate and thereby considerably reduces Chat Confusion. However, the social protocol alone is incapable of implementing the dynamics adequately, since many interruptions still occur. Conversation techniques need to be used to force the implementation of the dynamic, but the implementation of these techniques should be sufficiently flexible to overcome exceptional situations. p. 15

Mediated Chat 3.0: Publication queue to avoid Message Overload One of the problems frequently cited by participants of the ITAE course is the difficulty in reading all the messages during the debate. The problem occurs when several messages are sent in a short period of time, which makes reading all the messages impossible, causing anxiety and generating the possibility of Chat Confusion. Identified in the research, this phenomenon was called Message Overload and can be seen in the declarations made by participants during interviews, such as BI find it difficult to keep up with the speed of the debate. I don’t think I’ll ever adapt;’’ BI only know that I can either read or write. By the time I formulate a reply, the subject has already changed;’’ BWe can see that ideas are lost during the flood of messages. A question, statement or reply can go unnoticed and the learner loses the rhythm and his or her line of reasoning, affecting the person’s performance.’’ p. 15

Compared with spoken conversation, Message Overload is similar to Overlapping Voices, a phenomenon that occurs when two or more interlocutors are speaking at the same time. In spoken conversation, the social protocol Bonly one speaker at a time’’ is used to avoid or get around this problem. However, the overlapping voices phe- nomenon does not occur in chat tools since the messages are presented all at once, masking the process of production and making the use of the social protocol Bonly one speaker at a time’’ unviable as a means of organizing the turn to speak (Herring, 1999). The lack of visibility of the turn-in-progress (Smith et al., 2000; Garcia & Jacobs, 1998; Viegas & Donath, 1999) is identified as one of the causes for Message Overload, but not as the underlying problem. Even where the turn-in-development is perceived, various messages can still be sent in a short period of time, thereby generating Message Overload. p. 15

The Mediated Chat 3.0 tool (Fig. 8) was developed to avoid Message Overload. After publishing a message, the chat server waits a period of time before publishing the next message (an interval of time, estimated to be sufficient for reading the previous message, based on its number of characters). During this interval of time, the new messages sent by participants are queued on the server for later publication. This mechanism distributes the publication of messages over time so that the participants manage to read all the messages without being taken by surprise by message bursts (several messages published over a short period of time). p. 16

In the Participants List, a grey bubble can be seen pulsating next to the participant’s name while he or she is typing. After the participant sends the message, a black bubble is displayed next to the name indicating that the participant has already sent a message that is now in the queue waiting to be published. While the message is in the queue, its publication position queue is displayed to the sender, the typing area is blocked, and the sender has the option to cancel publication of the message. When a message is published, the black bubble blinks for a period of time, indicating that the participant is Fspeaking_ at that moment. Messages sent by mediators receive a higher priority for publication. p. 16

The Mediated Chat 3.0 tool was tested in the ITAE 2004.1 edition. Hypothetically, the message publication queue should lessen Chat Confusion by allowing all the messages in the debate to be read. The results, however, were inconclusive. The interviews conducted with participants revealed that many did not have an adequate understanding of the message queue and some even thought that the tool had become p. 17

slower because the messages took time to be published (the messages were actually waiting in the queue, but participants did not perceive or understand the mechanism). To avoid this problem, the message publication queue will be represented directly in the Participant List through the ordering of participants, as implemented in the PalTalk tool. By making the queue more visible, the mechanism can hopefully be better understood and its impact on reducing Chat Confusion re-investigated. On the other hand, the participants rapidly understood the indicator about who was typing (pulsating grey bubble next to the typing person’s name on the Participant List). In the interviews, they declared that this mechanism helped in the coordination of the debates, particularly in their decision about when to write or send a message during the debate (self-coordination of their participation). An analysis of the session record shows that interruptions caused by the lack-of- visibility-of-the-turn-in-progress problem (Garcia & Jacobs, 1998), as illustrated by message 20 of the debate in Text 4, were avoided. p. 18

The interface modifications introduced in this version aimed at facilitating the processes of reading and writing messages during the debate. The typing area comprises 3 lines of visible text instead of a single line, which helps the process of revising and editing before sending the message. To help the reading process, text formatting was used to clarify the moment when the message was published (dimmed time-stamp); visually differentiate the sender from the content of the message (sender in bold and content in normal font style); differentiate par- ticipants’ messages from system messages (alerts to participants entering and leaving are displayed in gray); and better differentiate the limits of each message (indented margin for the first line of each message and a small extra space after the last line of each message, increasing the separation between messages). Only the first name of the participants is displayed instead of the whole name, re- ducing the amount of text needed to identify the sender. The scroll bar only scrolls automatically if located at the foot of the page, a situation that occurs when the participant is following the most recent messages, and stops scrolling when the participant changes the position of the bar to read the messages that are no longer visible on the screen (automatic scrolling makes reading earlier messages difficult). p. 19

The participants found the conversation less confusing due to the changes made in the chat register, as the declaration made by Carlos exemplifies: One thing that I found is that the tool helps considerably towards the success of the debate. The interface of the first debates made things more complicated. Every text seemed packed and difficult to read. It was difficult to keep track when that flood of messages began. This interface improved things a lot. p. 20

Whenever a participant enters in the middle of a debate session, whether because of late arrival or loss of internet connection, the other participants are already engaged in discussion and the participant may encounter difficulties in entering the conversation—in the present research, this problem was called Decontextualization. p. 20

In relation to the time taken to engage in the chat, two premises had been formulated: provided with the register of the complete session, the participant would rapidly become contextualized and would engage into the conversation; or the opposite would occur, and the participant would lose more time reading the previous messages in order to become contextualized and take longer to en- gage in the conversation. Analyzing the interval of time between the participant’s re-entry and the posting of their first message engaged in the conversation, it was found that message registration had no influence on the participant’s engagement in the conversation: it neither helped (reducing the time interval) nor hindered (increasing the delay). At least in this edition, no impact on the learner’s par- ticipation was identified in terms of his or her disconnection and reconnection in the debate. In relation to interruptions, no learner manifested decontextualization after reconnecting to the debate, even in the absence of the session history. The ab- sence of interruptions could be explained in many ways: learners are instructed to avoid interrupting when they arrive late (social protocol); the short time interval between disconnection and re-entry (few seconds) does not generate any decontexualization (context remains in the learner’s memory); and learners were only disconnected during the free conversation stage to avoid upsetting the dy- namics of the debate given that it is easier to engage in conversation during this stage. p. 22

Baker, K., Greenberg, S., & Gutwin, C. (2001). Heuristic evaluation of groupware based on the mechanics of collaboration. Proceedings of the 8th IFIP Working Conference on Engineering for Human-Computer Interaction (ECHIF01). Toronto. Baron, N. S. (1984). Computer mediated communication as a force in language change. Visible Language, 18(2), 118–141. Borghoff, U. M., & Schlichter, J. H. (2000). Computer-supported cooperative work: Introduction to distributed applications. New York: Springer. Collins, A., Joseph, D., & Bielaczyc, K. (2004). Design research: Theoretical and methodological issues. Journal of the Learning Sciences, 13(1), 15–42. Conklin, J. (1988). Hypertext: An introduction and survey. In I. Greif (Ed.), Computer supported cooperative work: A book of readings (pp. 423–476). San Mateo, CA: Morgan Kaufmann. Cornelius, C., & Boos, M. (2003). Enhancing mutual understanding in synchronous computermediated bommunication by training: Trade-offs in dudgmental tasks. Communication Research, 30, 147–177. Crystal, D. (1985). A dictionary of linguistics and phonetics. Oxford: Blackwell. DeSanctis, G., & Gallupe, B. (1987). A foundation for the study of group decision support systems. Management Science, 33(5), 589–609. Ellis, C. A. (2000). An evaluation framework for collaborative systems (Technical Report CU-CS901-00). Boulder, CO: Colorado University. Ellis, C. A., Gibbs, S. J., & Rein, G. L. (1991). Groupware—Some issues and experiences. Communications of the ACM, 34(1), 38–58. Ferraz, F. G. (2000). Framework Canais de Comunicac ̧a ̃o. Undergraduate Computer Science Department, Rio de Janeiro: Catholic University of Rio de Janeiro. Filippo, D., Fuks, H., & Lucena, C. J. P. (2005). AulaNetM: Extension of the AulaNet Environment to PDAs. CEUR Workshop Proceedings, 10—Context and Groupware, Proceedings of the Fifth International and Interdisciplinary Conference on Modeling and Using Contex, Paris. Fuks, H., Gerosa, M. A., & Lucena, C. J. P. (2002). The development and application of distance learning on the internet. Open Learning—The Journal of Open and Distance Learning, 17(1), 23–38. Fuks, H., Raposo, A. B., Gerosa, M. A., & Lucena, C. J. P. (2005). Applying the 3C model to groupware development. International Journal of Cooperative Information Systems (IJCIS), 14(2–3), 299–328. Garcia, A., & Jacobs, J. (1998). The interactional organization of computer mediated communication in the college classroom. Qualitative Sociology, 21(3), 299–317. p. 24

Garcia, A., & Jacobs, J. (1999). The eyes of the beholder: Understanding the turn talking system in quasi-synchronous computer mediated communication. Research on Language and Social Interaction, 32(4), 337–367. Gerosa, M. A., Pimentel, M., Fuks, H. & Lucena, C. J. P. (2005). No need to read messages right now: Helping mediators to steer educational forums using statistical and visual information. In G. Stahl & D. Suthers, (Eds.), Proceedings of Computer Supported Collaborative Learning 2005 (pp. 160–169). Mahwah, NJ: Lawrence Erlbaum Associates. Gerosa, M. A., Pimentel, M., Raposo, A. B., Fuks, H., & Lucena, C. J. P. (2005). Towards an engineering approach for groupware development: Learning from the AulaNet LMS develop- ment. Proceedings of the 9th International Conference on CSCW in Design (CSCWiD) (pp. 329– 333). Coventry, U.K. Herring, S. (1999). Interactional coherence in CMC. Journal of Computer-Mediated Communica- tion, 4(4). Retrieved from http://jcmc.indiana.edu/vol4/issue4/herring.html Johnson-Lentz, P., & Johnson-Lentz, T. (1982). Groupware: The process and impacts of design choices. In E. B. Kerr & S. R. Hiltz, (Eds.), Computer-mediated communication systems: Status and evaluation. New York: Academic Press. Laurillau, Y., & Nigay, L. (2002). Clover architecture for groupware. Proceedings of the Computer- Supported Cooperative Work (CSCW) (pp. 236–245). New York: ACM Press. Lonchamp, J. (2005). A structured chat framework for distributed educational settings. In G. Stahl & D. Suthers, (Eds.), Proceedings of Computer-Supported Cooperative Learning (CSCL) 2005 (pp. 403–307). Mahwah, NJ: Lawrence Erlbaum Associates. Lucena, C. J. P., Fuks, H., et al. (1998). AulaNet—An environment for the development and maintenance of courses on the web. Proceedings of ICEEF98—International Conference on Engineering Education. Rio de Janeiro. McAlister, S., Ravenscroft, A., & Scanlon, E. (2004). Combining interaction and context design to support collaborative argumentation using a tool for synchronous CMC. Journal of Computer Assisted Learning, Special Issue: Developing dialogue for learning, 20(3), 194–204. McGrath, J. E. (1990). Time matters in groups. Intellectual teamwork: Social and technical foundations of cooperative work (pp. 23–61). Hillsdale: Lawrence Erlbaum. Mu ̈ hlpfordt, M., & Wessner, M. (2005). Explicit referencing in chat supports collaborative learning. In G. Stahl & D. Suthers, (Eds.), Proceedings of Computer-Supported Cooperative Learning (CSCL) 2005 (pp. 460–469). Mahwah, NJ: Lawrence Erlbaum Associates. Oikarinen, J., & Reed, D. (1993). Internet relay chat protocol (RFC 1459). Network Working Group. Retrieved from http://www.faqs.org/rfcs/rfc1459.html. O_Neill, J., & Martin, D. (2003). Text chat in action. Proceedings of the 2003 International ACM SIGGROUP Conference on Supporting Group Work (pp. 40–49). New York: ACM Press. Pimentel, M. (2002). HiperDia ́logo: Ferramenta de bate-papo para diminuir a perda de co-texto. Master_s dissertation, Computer Science Department of Federal University of Rio de Janeiro. Pimentel, M., Fuks, H., & Lucena, C. J. P. (2003). Co-text loss in textual chat tools. Proceedings of the 4th International and Interdisciplinary Conference on Modeling and Using Context (pp. 483– 490). Stanford. Pimentel, M., Fuks, H., & Lucena, C. J. P. (2004). Mediated chat 2.0: Embbeding coordination into chat tools. Proceedings of the 6th International Conference on the Design of Cooperative Systems (pp. 99–103). Hye`res: IOS Press. Pimentel, M., Fuks, H., & Lucena, C. J. P. (2005). Mediated chat development process: Avoiding chat confusion on educational debates. In G. Stahl & D. Suthers, (Eds.), Proceedings of Computer Supported Collaborative Learning (CSCL) 2005 (pp. 499–503). Mahwah, NJ: Lawrence Erlbaum Associates. Raposo, A. B., Pimentel, M. G., Gerosa, M. A., Fuks, H., & Lucena, C. J. P. (2004). Prescribing e-learning activities using workflow technologies. Proceedings of the 1st International Workshop on Computer Supported Activity Coordination (CSAC) (pp. 71–80). http://www.iceis.org/ iceis2004/workshops/csac/csac2004-cfp.html Rezende, J. L. (2003). Aplicando te ́cnicas de comunicac ̧a ̃o para a facilitac ̧a ̃o de debates no ambiente AulaNet. Master’s dissertation, Computer Science Department of the Catholic University of Rio de Janeiro. Smith, M., Cadiz, J. J., & Burkhalter, B. (2000). Conversation trees and threaded chats. Proceedings of the 2000 ACM Conference on Computer Supported Cooperative Work (pp. 97–105). New York: ACM Press. Teufel, S., Sauter, C., Mu ̈hlherr, T., & Bauknecht, K. (1995). Computerunterstu ̈tzte gruppenarbeit. Bonn: Addison-Wesley. p. 25

Thirunarayanan, M. O. (2000). Cutting down on chat confusion: A proposal for managing instructor- controlled chat systems. Ubiquity, 1(38). Toffler, A. (1970). Future Shock. New York: Random House. Viegas, F. B., & Donath, J. S. (1999). Chat circles. Proceedings of the Conference on Human Factors in Computing Systems. New York: ACM. Vilhja ́lmsson, H. (2003). Avatar augmented online conversation. Ph.D. dissertation, Program in Media Arts and Sciences, Cambridge: Massachusetts Institute of Technology. Vilhja ́ lmsson, H., & Cassell, J. (1998). BodyChat: Autonomous communicative behaviors in avatars. Proceedings of ACM Autonomous Agents ’98 (pp. 269–276), New York: ACM. Werry, C. C. (1996) Linguistic and interactional features of internet relay chat. In S. C. Herring (Ed.). Computer-mediated communication: Linguistic, social and cross-cultural perspectives (pp. 47–63). Amsterdam/Philadelphia: John Benjamins Publishing Company. Winograd, T. (1989). Groupware: The next wave or another advertising slogan? Proceedings of CompCon 89 (pp. 198–200). Washington, DC: IEEE Computer Society Press. Zemel, A. (2005). Texts-in-interaction: Collaborative problem-solving in quasi-synchronous computer-mediated communication. In G. Stahl & D. Suthers, (Eds.), Proceedings of Computer< Supported Collaborative Learning 2005 (pp. 753–757). Mahwah, NJ: Lawrence Erlbaum Associates. p. 26