Describing and Enhancing Collaboration at the Computer

Computer-based learning materials differ from classroom practice in that they seldom explicitly offer opportunities for collaboration. Despite this, students do collaborate, helping one another through the content and affordances of computer materials. But, in doing so, students meet with challenges. Paradoxically, these challenges can either inspire or discourage learning and second-language acquisition. This paper, based on research with twenty Hong Kong university students in a controlled experiment, evaluates challenges to collaboration at the computer as evidenced by discourse. The students were videotaped and their discourse transcribed and evaluated both qualitatively and quantitatively, according to a set of discourse markers created to describe collaborative, non-collaborative and ambiguous strategies. The paper begins by exploring the differences between collaboration and similar terms such as teamwork and cooperative learning then goes on to define collaboration in the context of computer-assisted learning. It ends by presenting practical suggestions for software designers, teachers and students to enhance collaboration at the computer.

These same points are mirrored in a list presented by Jacobs (1998), which summarizes some of the above by offering ten potential advantages of group activities in language instruction: • The quantity of learner speech can increase Several of Jacobs' points are identical to the advantages Hoogeveen (1995) ascribes to multimedia learning. This is perhaps because the computer naturally invites collaboration through its various affordances, including the public nature of the display or screen (versus the relatively private nature of the textbook and sheet of writing paper). This is noted by Cazden (1988) who suggests: There may be several reasons why placing computers in classrooms seems to result in increased collaboration among peers. One reason is a permanent feature of the technology: work in progress on the screen is public in a way that paper on a desk is not. Other reasons may be more temporary. Most classrooms today have one computer at most, and that makes it a scarce resource whose use can be doubled by asking children to work at terminals in pairs. Expertise in the new technology is also a scarce resource, and student experts can supplement the limited availability of the teacher. (p. 148).
Cazden would seem to outline a paradox: a limited number of computers might in some cases be preferable to a class set, as more limited access is likely to promote collaborative learning. Collectively, the ideas and concerns of the above authors point to a broad range of benefits of collaboration. But for collaboration to be successful, it may require a structure, although that structure may be facilitated through a computer interface rather than being imposed by a teacher.

Structuring Collaboration
One debate within collaboration studies is the degree to which a teacher should structure collaborative activities and offer skills training. Hamm (1992) suggests that the teacher must structure collaboration on computers through: • assigning students to mixed-ability teams • establishing positive interdependence • teaching cooperative social skills • insuring individual accountability • helping groups process information. (p. 95) Hamm (1992) goes on to suggest that an aim of successful collaboration is to promote feelings that "no one is successful unless everyone is successful" (p. 96). He suggests that this may occur through: • Goal interdependence -stating clearly what each member of the group should know how to do upon completion of the task • Task interdependence -clearly defining the group goal, and what the team should agree on or be able to produce • Resource interdependence -specifying parameters, materials, the team's task • Role interdependence -reviewing the individual roles for the group members; keyboarder, checker, reporter, summarizer, encourager, and so on. Set up the expectation that all are responsible for explaining how they came up with the answer. Explain the grading procedures, group credit as well as credit for how well each student performs his or her group job. (p. 96) However, in the context of CALL, many of Hamm's suggestions are better left to the computer. For example, Susman (1998) argues: Software can be designed to give some interaction support features. CBIs (Computer-Based Instruction materials) can support 1) reflection, 2) exploration of multiple perspectives, and 3) the integrated use of multiple resources. These supports can be accomplished by a note-taking section or frame and providing links to a variety of media presentations and sources (e.g., links on the World Wide Web). Rysavy and Sales suggest pauses in the materials that direct group members to interact on their understanding. They also suggest that CBI could remind students to monitor their performance. Even though CBIs cannot assure a cooperative environment, CBIs can help stimulate group processing by encouraging the group to analyze their progress and by teaching them how to effectively communicate. (p. 317) CALL programs should also provide ways to make motivation implicit instead of explicit. It is certainly true that some learners (especially younger ones) may need training in the skills necessary for working in a group, but many of the advantages of collaboration are lost if learners lose their autonomy and the task becomes too teacher-centered and teacher-directed, adopting a behaviorist model of instruction.
In particular, Hamm's assignment of individual roles for each learner discourages the collaborative advantages of a learning activity. Learners are likely to be more concerned with fulfilling their individual roles and tasks rather than the overall process of socially engaging in a process of problem-solving through negotiation of meaning. Dillenbourg (1999) discusses the impact of forcing roles upon learners: Negotiation can occur only if there is space for negotiation (Dillenbourg, Baker, Blaye & O'Malley, 1996), i.e.
if something can actually be negotiated. Negotiation at the meta-communicative level can be inhibited by forcing partners to play well-defined roles (a growing trend in research in collaborative learning). (p. 15) Dillenbourg (1999) expands on this by suggesting that trivial tasks similarly inhibit collaboration or deny the benefits of collaboration: Negotiation at the task level can be inhibited in trivial tasks, in which there is nothing to disagree upon, and in which there is nothing to misunderstand. The boundary between misunderstanding and disagreement is shallow. If we do not understand each other, we cannot say that we properly agreed. (p.15) In any case, from the point of the view of learners engaged in a learning task, a conversation is often about creating roles for each person. Richards and Schmidt (1983) explain: Conversation is more than merely the exchange of information. When people take part in conversation, they bring to the conversational process shared assumptions and expectations about what conversation is, how conversation develops, and the sort of contribution they are each expected to make. (pp. 119-120) If a teacher proceeds to make these assumptions and raises the corresponding expectations of roles for learners, the learners are deprived of the opportunities to develop appropriate conversational and negotiation skills. By extension, they may also be denied the attendant opportunities for language acquisition. The above concerns and ideas serve to explain features, parameters and advantages of collaborative activities, but none provide a definition or show the difference between collaboration , cooperation and other terms, such as teamwork .

Differences Between Collaboration and Other Terms
The difference between collaborative learning and cooperative learning is not well defined and both have many shades of meaning. Biggs and Moore (1993), for example, suggest cooperative learning is an activity set by the teacher while collaborative activities are ones spontaneously set up by the learners. Kohonen (1992) uses cooperative and collaborative interchangeably. Nunan (1992) defines collaboration as an activity in which learners have greater control over the design of their learning while he considers cooperative learning as merely a mode of instruction. Other authors, such as O'Neil (1994) avoid the use of cooperation or collaboration and use the term teamwork . Dillenbourg, Baker, Blaye & O'Malley (1996) provide perhaps the clearest functional difference between cooperation and collaboration: Cooperation and collaboration do not differ in terms of whether or not the task is distributed, but by virtue of the way in which it is divided; in cooperation the task is split (hierarchically) into independent subtasks; in collaboration cognitive processes may be (heterarchically) divided into intertwined layers. In cooperation, coordination is only required when assembling partial results, while collaboration is...a coordinated, synchronous activity that is the result of a continued attempt to construct and maintain a shared conception of a problem. (p. 189) This difference between collaboration and cooperation is made plain in Hamm's earlier suggestion for teacher-imposed division of roles among learners in a group engaged in a common task. Such an imposed division places Hamm's suggestions within the definition of cooperation. This paper's definition of collaboration is based on what Dillenbourg, Baker, Blaye & O'Malley (1996) consider to be collaborative practices and situations, regardless of the terms used by other authors in describing pair or group learning.
However, various authors' quoted texts retain their original choice of terms.

The Range of Collaboration and Call
CALL programs, or other computer programs that can be used to encourage language learning 1 , often address several of the cooperative, teamwork or collaborative skills mentioned above by various researchers.
Adaptability(recognizing problems and responding appropriately) is found in various software programs that present learning as a quest. This is particularly appropriate in materials developed for younger learners in which a series of correct answers move one along a map or serve to save a creature in danger. This requires coordination and decision-making and interpersonal and communication skills. Such activities often work best with group members of different language and cultural backgrounds, such as in a mixed ESL classroom where English (however limited) is the only common language. In such situations, collaborative interpersonal skills such as consensus and queuing must be negotiated as well as the target content. However, the focus of the investigation in this paper is a different but commonly observed collaborative phenomenon: oral/aural collaborative language use among pairs or small groups of learners working at a single computer to complete a task or a series of tasks.
Some teachers worry that working at a computer can be a socially isolating experience for learners. However, as Crook (1994) states: "Yet when we examine actual classroom practice we find that, in certain significant respects, the computer has facilitated socially organized learning rather than inhibited it" (p. 121). A commonly observed collaborative phenomenon is pairs or small groups of learners working on their own outside of a class at a single computer to complete a task or a series of tasks. This type of collaboration is sometimes teacher-initiated but is more often learner-initiated. In some cases, such collaboration is a prelude to broader international communication or collaboration through email.
In collaborating, learners sometimes work together at one computer because of limited access to enough computers. But my own observations over the past eight years of learner use of computers in computer centers, computer lab areas and design studios 2 suggest learners commonly and naturally work together at the computer, despite an abundance of computers that would allow them to work individually. Argyle (1991) suggests three possible reasons, or motivations, for people to collaborate: • For external rewards • In order to form and further relationships • In order to share activities they are involved in. (cited in McConnell, 1994, p. 13) Argyle's external rewards suggest collaboration simply for extrinsically motivated reasons such as classroom marks. But Argyle's second and third reasons suggest more intrinsic motivations. These include the desire to offer assistance where one learner of the pair has already completed an assignment, to help with problems tangential to the assignments such as the operation of unfamiliar hardware/software or, less nobly, to simply plagiarize assignments.

Investigation
In an investigation of collaboration at the computer (Beatty 2001), twenty subjects were asked to answer seven questions using the resources of a CD-ROM developed around the novel Frankenstein by Mary Shelley. The subjects were second-year students at City University of Hong Kong, who had a general level of computer literacy and keyboard skills, but were unfamiliar with the Frankenstein story. The students were paid HK$100 (about CAN$20) each and arranged themselves randomly in pairs on a sign-up sheet. Each pair was given an hour to answer the seven questions.
The seven questions (see Appendix 1) were intended to encourage both discourse and exploration of the learning materials on the CD-ROM, which included the annotated novel, source materials for the novel, art, poetry and video organized into 1,042 linked pages. Part of the research was to determine whether interface was a significant influence on collaboration. To this end, two interfaces were created. The interface designs followed ideas presented by Jonassen, Wilson, Wang, and Grabinger (1993) whose ideas are summarized in the following One interface featured a behaviorist model of instruction in which the questions and hints were easily accessible and plainly stated. The second interface used a game-like constructivist model in which the same seven questions and hints were organized into a game set in a surrealistic graveyard, complete with interactive graphical elements and sound effects, for example for fire and an atomic explosion. In keeping with the game metaphor in which points are lost when one requires extra resources, a variety of different screams were heard as a gentle punishment for clicking on the help and hints. Most of the hints were arranged as apples on a tree growing from a skeleton. Some apples were empty, without hints. Other hints were obtained by clicking items that identified themselves as interactive as one passed the mouse cursor over them. To increase the subjects' perception of the need for self-reliance, the warning "Every time you take an apple, you lose a bone" was added, although no actual score keeping was done. Five of the pairs worked on the behaviorist interface while the other five worked on the constructivist interface.
While the students worked, their keystrokes were recorded through the use of Lotus ScreenCam , which allowed for a replay of each one-hour session indicating exactly where the students had gone in exploring the interfaces and also successive revisions to their answers to the seven questions.
To collect the discourse, the subjects were also videotaped. The ten hours of videotape were transcribed indicating both what was said as well as paralinguistic actions, such as pointing at the screen. Analyzing the discourse involved transcription notations -writing out the utterances and describing the paralinguistic actsand interpretive notations, deciding the motivations, or strategies, behind the utterances. The strategies, listed below, were developed after consideration of a set of strategies by Kumpulainen and Wray (1996). These are strategies learners use to engage in collaboration, strategies learners use to engage in or avoid collaboration and ambiguous strategies which may work either way, depending on the context or the respective attitudes of the speakers.
Ambiguous strategies are the most problematical in terms of strategy identification as both their intentions and effects may differ in any situation. For example, a learner might employ humor to dispel tension, but his or her interlocutor might perceive it either as an attempt to avoid the task or to engage in lateral thinking. The reverse might just as easily be true.
Strategies used to engage in collaboration: • Determine participants' expertise: Determining expertise is classified as a collaborative strategy because it helps to clarify what each partner knows or does not know about a task. Learners who start off by determining expertise are better able to collaborate because they are better positioned to evaluate what they and the other person knows.
• Explain the text/task/ideas: Explaining the text, the task and the ideas in a collaborative situation helps provide opportunities for negotiation of meaning and create a common understanding. If learners in a collaborative situation do not arrive at a common understanding of the text, the task and the ideas behind what they are studying, they are likely to work at cross-purposes.
• Offer suggestions: Offering suggestions is often marked by the phrase . think. A suggestion differs from a judgment in that the former offers a qualification and invites comment, while a judgment is presented as a final word on a subject.
• Direct attention: Directing attention to text or images on the computer screen or to something else such as the keyboard or mouse is classified as a collaborative strategy because it is a strategy that tries to involve one's partner in some aspect of the program.
• Solicit suggestions/support: Soliciting suggestions and support are collaborative actions because they directly ask one's partner's involvement.
• Solicit clarification: Soliciting clarification occurs when one learner asks the other for more information on a statement. It is a collaborative strategy essential to joint negotiation of meaning.
• Signal interest in/show support of another's ideas: Signaling interest or showing support of another's ideas are collaborative strategies that allow the learners to indicate a common direction in what they are doing or discussing.
• Solicit support for or suggest actions : Soliciting support for or suggesting actions most often occur when one learner is navigating or about to navigate in the program. Soliciting support is typically marked • by phrases such as, should we…, shall I… and so on and is collaborative because it gives the partner a chance to discuss the working process.
Social strategies learners use to avoid collaboration: • Ignore the text/task: Ignoring the test or a task is a strategy used to avoid collaboration because it often marks a learner's preference for pursuing individualistic or competitive goals. In some cases, the task may be too difficult for the learner, but not to even try to solicit suggestions signals a failure to take advantage of the collaborative partner's skills and makes ignoring the task a sign of avoiding collaboration.
• Interrupt: Interrupting is a strategy that avoids collaboration because it signals that one learner does not value what the other learner wants to say. In some cases, interrupting may be seen as a supportive strategy if, for example, one interrupts to supply information one is struggling to recall or formulate, but such interruptions are usually signaled by a pause in the speech by the other speaker. In this investigation, such interruptions were classified as offering suggestions.
• Ignore ideas: Ignoring ideas signals that one learner does not value what the other learner wants to say and does not care to discuss it. An idea might be a suggested answer or a suggested action, such as directing attention to something on-screen.
• Offer judgments: Judgments are statements without qualifying phrases such as . think…. Sometimes a judgment is presented as a learner's simple statement of a fact, but it is often a strategy that avoids collaboration because judgments to not invite the collaborative partner to present opinions or negotiate meaning.
Ambiguous strategies in collaboration: • Offer humor: In addition to the above methods of engaging in and avoiding collaboration are the sometimes negative and sometimes positive aspects of humor. Humor can be used negatively as an avoidance strategy or positively as a way of soliciting lateral thinking, smoothing social relations and dispelling tension.
• Read aloud: Reading aloud may be a neutral strategy for holding space in the conversation while one learner reads what is on the screen or it may be a collaborative strategy for dictating or keeping the partner informed.
As it is difficult to assess the first learner's intention and the second learner's perception, reading aloud is classified as an ambiguous strategy.

A Sample Episode from the Investigation
Learners are sometimes preoccupied with factors that have little or nothing to do with the task. These can include concerns about completing the task as a priority over completing it properly or successfully. In some cases, an unwillingness to enter into a task may be related to feelings of insecurity over one's expertise. The following episode is representative of the session involving Dominic, a generally conscientious and hard-working learner, and Eunice, an academically successful and intelligent learner but one who, nonetheless, balks at completing the tasks in the session. They are using the constructivist interface. This behavior seems somewhat feigned and was perhaps for the purpose of impressing Dominic with her rebellious attitude, but is also surprising, considering that she was aware (as were all subjects) that she was being paid HK$100 for participating in the one hour session. Eunice's conduct suggests that while the quasi-experiment might not have been a totally authentic task, it was at least authentic enough to make Eunice forget herself and relax into a natural behavior.
Eunice's attempts to denigrate the task (lines 286, 290, 292, 294, 296, 298, 300, 302) in the above episode are largely (and diplomatically) ignored by Dominic who takes the lead in addressing the questions and patiently working through the materials. Throughout, Dominic attempts to share what he has seen and understood (lines 283a, 285, 293, 301, 303). Seaton (1993) writes "...knowledge emerges from a collaborative process of active dialogue among those who seek to know" (p. 51). However, the above episode is clearly a situation where despite the assignment of a task in which learners are encouraged to collaborate, one subject (i.e., Eunice) simply does not seek to know.
Although Eunice exhibits moments of collaborative behavior, she generally maintains her bored and contemptuous attitude throughout the session.
In discussing guidelines for determining the locus of instructional control in CAI, Hannafin (1984) suggests older students perform more effectively under guided learner control while younger students perform best under lesson control. He goes on to say that more able students perform best under learner control while less able students perform best under lesson control. But the above episode raises the question of what happens when students of varying abilities are put together to work in a collaborative environment. Does the more motivated learner encourage or discourage or the less motivated one? Does the better learner encourage or discourage the poorer learner? In fact, both subjects in this session are good learners; it is simply that one of the subjects chooses to not be interested in the program and tasks.
In the case of the above episode with Dominic and Eunice, it may be that Eunice's negative attitude over self-perception of not being up to the task (especially seen in line 300) and poor motivation bring out better-than-usual behaviors in Dominic as he adopts, in the face of confrontation, a sympathetic, teacher-like nurturing role and explaining the question (ET) as seen in Lines 283a, 293 and offering a suggestion (OS) in line 301.
This supports the idea put forward that challenges to collaboration may promote different behaviors in different subjects. In another pairing, Eunice's negative attitude might have discouraged her collaborative partner to the extent that he or she might have become similarly demotivated. But in this case, Dominic rises to the challenge.
Line 300 also suggests a missed opportunity for scaffolded learning and negotiation of meaning. Eunice identifies her inability to understand "three new vocabularies". Had Dominic challenged her on them, they might well have been able to work out the meanings from context. Wegerif and Dawes (1998) suggest that a mixture of cooperation and conflict is the best way to promote interaction and note that such discourse is likely to include criticism, explanation, justification, clarification and elaboration. Implicitly, these are all ways in which one learner acknowledges the ideas or presence of another learner. However, as seen in the above episode, criticisms may be hostile and without any attempt to offer another idea or support. Such criticism is negative in that it deters participation. Wegerif and Dawes' other categories (explanations, justifications, clarifications and elaborations), are basically ways of engaging in collaboration by following McConnell's (1994) hypothesis that collaborative learning serves to make public those parts of our learning which are blind, hidden and unconscious. Therefore, other challenges are built on an unwillingness to offer explanations, justifications, clarifications and elaborations.
Based on the ideas elaborated above, four social challenges to discourse can be summarized as: • an unwillingness to engage in the activity; • an unwillingness to accept the collaborative nature of the activity (i.e. pursuing individual or competitive goals); • an unwillingness to offer suggestions or explanations; • an unwillingness to offer or accept justifications, clarifications, elaborations, criticism (i.e. groupthink) with supporting evidence or alternatives.
In addition to these social challenges, additional challenges can be summarized as: • the complexity of the program's content; • the navigability of the program's interface; • the difficulty of program's model of instruction.
A examination of collaborative, non-collaborative and ambiguous strategies in terms of their frequency of use by students using the two interfaces (Beatty 2001) led to the creation of a model (see Appendix 2), focusing on challenges to collaboration with respect to behaviorist and constructivist interfaces.But although the strategies are useful for examining discourse at the computer, the findings of the investigation were inconclusive because, as discussed above, learners may treat various challenges in different ways; rising to them, or being discouraged by them. A consideration of the strategies and challenges leads to pedagogical implications for software development and classroom practice reflecting on what might be done to make collaboration at the computer more effective.
into Chandler's (1984) Locus of Control(describing user vs. computer control of the interactions) and suggest how and why a learner might use particular software. In a school setting, these reviews might include the reviewer's name so that users might gauge the reviewer's opinion. In a classroom or self-access learning center, these reviews should be posted or otherwise made easily available. In these tasks, the teacher should take the role of the editor, having her learners engage in critical thinking about the learning materials they are using. This is a particularly important process to undertake when evaluating materials which claim to offer collaborative opportunities, but which simply offer shallow cooperative activities in which learners do not need to negotiate the meaning, the task or the process of finding answers.

Determining Individual Learning and Working Styles
One of the questions raised in this investigation was whether personality was a principal influence in determining the degree of collaboration. Personality can be divided into several areas such as learning style(s) but defining learners' learning styles can be difficult, time consuming and non-transferable across disciplines.
For example, a learner who excels in the rote memorization of baseball statistics may not transfer such skills to mathematics or learning vocabulary. Nor may the skills be appropriate to all subjects and situations, such as the learning of creative writing or emergency procedures.
There is also the question of whether a learner's preferred learning style is the most appropriate for learning.
Some learners may say they prefer learning passively through a lecture, but may actually learn better through a simulation. One way in which teachers unconsciously define learning styles is through publication of review phrases such as if you enjoyed this, you may like this. Learners should examine how they learn best, but they should also consider whether their present learning style is efficient and, if not, where it is in need of improvement.
In terms of facilitating collaboration at the computer, it appears that some subjects in this investigation's quasi-experiment were good at determining the collaborative process, others were not so good, and some simply failed to determine the working process altogether. There are arguments for and against strategies training. In some cases, it may be that defining the collaborative process down to the level of assigning roles for each learner defeated the benefits of collaboration _ the opportunities for negotiation of meaning.
However, it is apparent from this investigation that some degree of strategies training about the collaborative process could be useful both for learners and teachers. Such strategies training might include methods for addressing the individual perceptions. For example, strategies to help learners begin a task by explaining what they already know or do not know about the problem (determining the expertise) and deciding what they want to accomplish (determining goals and priorities).

Software Objectives
Explaining what skills each software package attempts to improve is an important task for teachers and learners. Dunkel (1991) asks teachers to consider which kinds of CALL lessons augment development of particular L2 skills such as reading and listening comprehension, oral proficiency, and knowledge of grammar.
One way to do this on a more individual level is for teachers to brainstorm with learners what they think they need to learn in terms of language. This serves as a starting point for deciding the categories and sub-categories in which they may wish to have CALL interaction, and it can be used to make decisions on what kinds of CALL software programs to include in a classroom as well as to create learner contracts for learning. An initial level of categories includes basic skills of reading, writing, speaking and listening as well as the more general category of computer literacy. Sub-categories might include micro-skills and vocabulary related to local and individual needs.
Language teachers generally aim to increase learner-centredness, or locus of control toward the user, in activities in which learners are involved in the curriculum and teaching/learning decision-making process.
Posting a schemata or mental map of a learners' needs along with notes on the software packages that match each need is a good way to allow learners to organize their own learning. This can involve learners in the process, and make them more responsible for their learning, and, in doing so, increase intrinsic motivation.

Making Better Use of Existing Materials
In order for learners to learn, they need to reflect upon their learning in discussion with teachers and peers, in diaries and in reports. In this way, learners begin to examine learning materials and their strategies for approaching them, thus benefiting even when a CALL program does not meet their learning needs. When a CALL program is not suitable, learners and teachers might need to examine ways in which it can be adapted. In some cases, this might involve a learner drawing up a set of questions that will guide another user on how to use the program. The same is true of another possible activity, adding layers of tasks to materials to make them more challenging or more appropriate. For example, learners might create a treasure hunt for key words and concepts within an encyclopedia software program.

Conclusion
Research into computer-based collaboration has mostly focused on individual learners using computers to collaborate over distance with other learners. However, collaboration among pairs or small groups of learners working at a single computer is perhaps a more common phenomenon in the classroom, and is one worthy of further study. This paper aimed to present tools to aid in that study: discourse markers for describing collaborative, non-collaborative and ambiguous strategies as well as practical suggestions for software designers, teachers, and students to enhance collaboration at the computer.