Designing roles, scripts, and prompts to support CSCL in gStudy

Computers in Human Behavior 26 (2010) 815–824

Contents lists available at ScienceDirect

Computers in Human Behavior
journal homepage: www.elsevier.com/locate/comphumbeh

Designing roles, scripts, and prompts to support CSCL in gStudy q
R. Morris a, A.F. Hadwin a,*, C.L.Z. Gress b, M. Miller a, M. Fior a, H. Church a, P.H. Winne b
a
b

University of Victoria, Dept. of Educational Psychology, 3800 Finnerty Road, Victoria, BC, Canada V8W 2Y2
Simon Fraser University, Burnaby, BC, Canada

a r t i c l e

i n f o

Article history:
Available online 7 February 2009
Keywords:
Computer-supported collaborative learning
Roles
Prompts
Scripts
Scaffold
gStudy

a b s t r a c t
This paper addresses the paucity of computer supported collaborative learning (CSCL) tools and research

that focus on actual computer embedded supports, guides, and scaffolds to effectively support the collaborative process. This paper: (a) explores the potential of support in the form of roles, scripts, and prompts
to scaffold collaborative engagement in computer-based learning environments, (b) explores ways these
supports might be implemented in a CSCL learning environment, namely gStudy, (c) describes how collaborative supports in gStudy might enhance opportunities for students to learn to self-regulate collaborative activity, and (d) uses examples from our research to propose ways these types of support tools
might advance research in CSCL.
Ó 2008 Elsevier Ltd. All rights reserved.

1. Introduction
Learning is an active and social process of constructing knowledge rather than simply acquiring information (Vygotsky, 1978).
With peer support, learners can overcome obstacles they could
not master if working alone and can increase their learning by
working towards a common goal (McMaster & Fuchs, 2002). Thus,
collaborative learning is pervasive in the education system today.
Collaborative learning has been used successfully in all academic disciplines, and the benefits of cooperative techniques and
collaborative learning for reading and reading comprehension, in
particular, have been well documented by researchers (Liang &
Dole, 2006). Accordingly, small collaborative groups are used in
classrooms (Johnson, Johnson, & Holubec, 1993), and more
recently, in online settings (O’Donnell, Hmelo-Silver, & Erkens,
2005), to ensure that students maximize their own and other’s academic potential (Jenkins & O’Connor, 2003).
For a variety of reasons, collaboration does not always lead to
increased performance. Although results show that students who
elaborate and actively participate benefit from peer collaboration
(Cohen, 1994), groups typically lack a diffusion of engagement
and responsibility towards completing the task. Lower achieving
students are frequently ignored or are off task (Mulryan, 1992)
and higher achieving students often dominate the group (Cohen,
1994). In addition, students with high status, often associated with
ability, gender, race, or social standing within the class, contribute
q
Portions of this paper were presented at the Annual Meeting of the Canadian

Society for the Study of Education, York University, Toronto, ON, May 26–30, 2006.
* Corresponding author. Tel.: +1 250 721 6347.
E-mail address: (A.F. Hadwin).

0747-5632/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.chb.2008.12.001

more when compared to students with low status within the group
(Webb, 1992). Effective collaboration requires an environment that
promotes positive interdependence, or in other words, an environment where the outcome cannot be achieved without the
contribution of each group member (Johnson & Johnson, 1989).
Even if positive interdependence exists, however, the group
may not possess the necessary cognitive skills to complete the task
(Cohen, 1994). Incorporating instructional practices that structure
group tasks can remedy this problem by distributing responsibility,
ensuring that all members evenly contribute, and by providing
scaffolds that help guide the cognitive process (O’Donnell, 1999).
Methods of structuring interaction can include scripting interaction, having students assume a role, providing prompts to fulfill
roles, giving specific task instructions, modeling, and providing
instruction on specific discourse skills (King, 1999). Most commonly, roles, facilitated by scripts and prompts, are used to create
structure.
In order for computer supported collaborative learning (CSCL)
to enhance productivity, teamwork, and learning in computer
and online environments, there is a need for software tools to
incorporate some of the instructional practices that support the
collaborative process as well as the collaborative product. These
tools aim to assist students during collaboration activities by providing opportunities to share and collaboratively edit electronic
documents, engage in group discussions and forums, and provide
and receive peer feedback (Hadwin, Winne, & Nesbit, 2005). CSCL
tools generally fall into two approximate categories: external support tools and internal support tools. External tools support the

learner or collaborator with assistance from outside the CSCL software environment or task context. These include software applets
such as online help directories, email web forms for submitting


816

R. Morris et al. / Computers in Human Behavior 26 (2010) 815–824

questions, and help menus. External tools are similar to external
resources and guides focusing heavily on task products or technical support and typically omit information on the collaborative
process. They usually take learners or collaborators away from
the current tasks and activities to another program, tool, or
source.
Internal tools, such as chat, discussion boards, white boards,
and file sharing applications facilitate collaboration by promoting
interaction in real time from remote locations. To support the collaborative process some of these tools incorporate additional supportive features such as assignment to groups, roles, scripts, and
prompts to facilitate productive collaborative discussion, encourage collaborative learning, and support self- as well as group regulation. For example, scaffolding the process by which students
provide feedback during collaboration might teach individuals
how to give feedback, why it is important, how to receive it, and
what to do with it (Carroll, Neale, Isenhour, Rosson, & McCrickard,
2003). These types of tools are considered ‘‘internal” to the learning environment because the support is anchored in the immediate
task and internal to the collaborative environment, rather than
sending a user ‘out for help.’
The cooperative/collaborative literature in face-to-face (nonCSCL) environments provides empirical support for multiple effective internal supportive techniques to strengthen the collaborative
processes (e.g., King & Rosenshine, 1993; O’Donnell and King,
1999; Slavin, 1995). This type of embedded, internal support, however, has not been fully examined in computer supported collaborative learning contexts. For example, Hadwin, Gress, and Page’s
(2005) review of supportive CSCL tools suggests that external support tools constitute approximately 80–90% of the kind of CSCL
support described in the literature. Few tools appeared to utilize
guided chats, discussions, or engagement to guide or support the
collaborative process. Furthermore, Hadwin et al. (2005) reported

that tools described in the CSCL literature were typically not examined for their effectiveness. In summary, the authors, found that
the most common convention in the CSCL literature appeared to
be providing CSCL collaborators with synchronous and asynchronous chat tools without guidance or scaffolding for using those
tools to enhance the collaborative process.
To address the need for guided and scaffolded support for productive collaboration in the CSCL literature, this paper: (a) explores
the potential of support in the form of roles scripts and prompts to
scaffold collaborative engagement computer-based learning environments, (b) explores ways those supports might be implemented
in a CSCL learning environment namely gStudy, (c) describes how
collaborative supports in gStudy might enhance opportunities for
students to learn to self-regulate collaborative activity, and (d)
uses examples from our research to propose ways these types of
support tools might advance computer supported collaborative
learning (CSCL) research.
2. Roles, prompts, and scripts
2.1. Roles defined
Roles provide the structure to facilitate collaboration and task
completion. When structure is provided through roles, students perceive a sense of security and, therefore, are able to concentrate on the
task. Roles can be defined as prescribed functions that guide individual behaviour and group collaboration (Slavin, 1995). They may also
be viewed as a scaffold in the learning process where the goal of collaboration is to acquire new knowledge, including cognitive and collaborative skills. Assigning roles may foster interdependence while
concurrently requiring individual accountability (Slavin, 1995).
Roles can further be classified as procedural/functional roles and
cognitive/intellectual roles (Palincsar & Herrenkohl, 2002).

Functional roles focus on how to carry out a task by classifying
and assigning particular kinds of task completion functions such as
data collector, recorder/notetaker, or editor. Classic functional roles
include: a recorder responsible for note taking and recording information and a materials manager responsible for establishing the
list of resources required for the task (Slavin, 1995). For example,
Strijbos, Martens, Jochems, and Broers (2004) implemented the
data collector role, where students were responsible for both an

inventory of the literature database and the gathering of additional
information. Tasks were very action oriented, lending themselves
nicely to a functional role.
However, academic learning tasks often involve more than just
completing a task. Academic work involves moving toward a cognitive outcome such as remembering, understanding, analyzing,
applying, evaluating, or generating new ideas and concepts. Academic tasks are about thinking processes directed toward thinking
products (Doyle, 1983). This perspective puts learning ahead of
doing. Cognitive roles focus on supporting engagement in academic work by classifying and assigning relevant types of thinking,
processing, and cognitive engagement into designated roles in a
collaborative context. Rather than scaffolding the ‘‘doing” of a task,
cognitive roles scaffold the ‘‘thinking” of the task (O’Donnell,
Hmelo-Silver, & Erkens, 2005).
2.2. Roles in the CSCL literature
While the use of roles as collaborative supports in CSCL environments is still in its infancy, recent work suggests that CSCL
environments have potential to provide innovative opportunities
for learners to adopt and experiment with roles. They may also
be useful contexts for researching aspects of role taking on collaborative processes and outcomes. For example, Robertson and Good
(2003) examined the qualitative and quantitative effects of the
commercial computer game, Ghostwriter, on the characterization
in children’s imaginative writing. Through team collaborative
role-play, each student controlled a character in an adventure
story. Players moved the characters around the virtual world by
collaborating through typed messages. This encouraged students
to engage with the story and, through discourse, aided planning
and development of story content. The researchers found that, as
in the classroom, team collaborative role-play encouraged children
to understand characterization in stories through empathizing
with their character. Students’ stories were far more descriptive
with deeper development of the characters than the stories written
by students in normal classroom circumstances. Ghostwriter creates game-based opportunities for players to assume a cognitive

role that interacts with the environment and other players. Students of all abilities appear to have benefited from using Ghostwriter. In fact, the authors purported that the virtual role-play
environment was particularly successful for students with low literacy levels and behavioural problems.
Another innovative application of roles in a CSCL environment
comes from work reported by Chou, Lin, and Chan (2002). Building
on reciprocal tutoring where two students take turns playing the
cognitive roles of tutor and tutee, Chou et al. (2002) embedded
scaffolding tools in a virtual ‘learning companion’ that was
designed to support and facilitate reciprocal tutoring. In this computer-based learning environment (CBLE), the learning companion
supported the learner by fulfilling one of the collaborative roles:
the tutor or the tutee. Collaboration in this instance was limited
to one actual learner and a highly skilled artificial tutor. While
some might argue this is not ‘‘real” collaboration, this type of reciprocal tutoring system’s value lies in its: (a) accessibility for students who can collaborate with the virtual learning companion
at their convenience, and (b) adaptability to a learner’s specific
tutoring strategy needs.


R. Morris et al. / Computers in Human Behavior 26 (2010) 815–824

These two studies are excellent examples of the innovative
internal support tools emerging in the CSCL literature. The design
of these tools, however, is only the first step. An important question that remains unanswered is whether students ‘‘learn” to collaborate better when these types of computer-based support are
incorporated into the collaborative environments. In other words,
do students become more adept in working with peers to complete
cognitively based tasks after collaborative scaffolds are removed?
Can CSCL tools be used to temporarily scaffold or support the
development of collaborative skills and strategies that are then
transferred by learners to future collaborative contexts? A second
line of questioning with respect to these types of computer-based
supports for learning might extend beyond examining the products
or outcomes of these types of CSCL environments toward the ways

supports change the quality or process of collaboration. When students are provided with roles to use during collaboration, does it
change the quality and quantity of engagement in the collaborative
task?
We posit that these two kinds of questions are central to
advancing theory and practice in this area. While outcome or product research dominates the field of CSCL, there are promising
examples of research that target changes to collaborative processes. For example, Hsieh, O’Neil, Harold, and Rossier (2002)
examined collaborative problem-solving processes in a computer-based knowledge-mapping environment, where students
were assigned two roles: leader or searcher. They discovered that
the nature of the task is an important determinant for the relationship of group involvement and group productivity. When the goal
is group productivity, designing roles that can fulfill task interdependency and participation requirements is important for collaborative learning.
Besides focusing on process versus product, researchers have
begun to answer questions about role type in CSCL. For example,
Strijbos et al. (2004) examined the effect of functional roles on perceived group efficiency during CSCL. They discovered that groups
in the role condition appeared to be more aware of their efficiency
compared to groups in the non-role condition, regardless of performance. They also found that roles increased group coordination.
However, findings did not support the notion that roles affected
group performance in terms of grades. Future research in this area
might extend beyond functional roles to examine whether cognitive roles yield different results.
Where functional roles clarify group efficiencies, cognitive roles
can affect relationships during collaboration. Van der Puil, Andriessen, and Kanselaar’s (2004) qualitative analysis explored the
dependency of effective collaborative argumentation on interpersonal relational aspects that develop during synchronous interaction. They found that discourse argumentation appears to affect
the relations between participants in a negative way. However,
they found that role change shifted the focus in order to create distance from the argument. Thus long-term, predefined roles appear
to be too rigid.
Some research findings suggest that students borrow from
other roles they have taken while enacting new roles (see Soller,
2001 for further discussion). While preliminary in nature, these
findings point to the possibilities of computer supported collaborative role taking for promoting transfer in the form of changes to the
way learners contribute and engage in collaborative discussions
after roles and support structures have been removed. To our

knowledge, transfer effects of role taking have not been empirically
examined. Yet, computer-based learning environments provide
optimal opportunities to examine the transfer effects associated
with shifting roles or removing role structures from the environment. This is an important avenue for CSCL research because it targets the potential of these environments to change students’
knowledge and behavior of collaboration beyond our CSCL tools.

817

Using empirical findings to weigh the advantages of structured
roles in online collaborative learning contexts, such as reporting a
high degree of engagement in the collaboration (Soller, 2001),
against the disadvantages, such as feeling much less control over
the collaborative process (Soller, 2001), is important for making
design decisions. The challenge for the field is to target psychological process and outcome variables associated with learning and
transfer in research about the effectiveness of computer supported
cognitive role taking for collaboration. Switching roles, having a
computer simulate a student in role, or using software to guide
roles all provide opportunities to extend our understanding about
how CSCL contributes to changes in learning processes and outcomes. Can roles in CSCL contexts help students learn to regulate,
co-regulate or share in the regulation of collaborative activity at
cognitive, metacognitive, motivational, and metacognitive levels?
Can students carry that learning beyond a structured collaborative
environment to new collaborative contexts? Do students gain
declarative knowledge about the nature or purpose of particular
roles in collaborative tasks? Do they develop procedural knowledge about how to enact specific roles? And, perhaps more importantly, do students who have worked in CSCL environments
develop conditional knowledge regarding when to use a certain
role, once scaffolds are removed? If students are not able to apply
what they learned to new situations then this type of scaffold may
not be useful for helping students learn how to collaborate more
effectively. In addition, it is important to investigate the influence

of roles on learning in authentic educational settings.
2.3. Scripts and prompts
Simply providing students with roles to aid collaboration may
not be sufficient to change collaborative processes and outcomes.
Students may not know how to carry out the role or may need
additional structure to feel confident to collaborate or complete a
task. The benefits of structure in collaboration are especially salient
when tasks are demanding and may potentially result in cognitive
overload. To improve efficiency, students may require additional
scaffolds to help deal with task completion and collaboration
(Rummel & Spada, 2005). Providing structure or information about
how to carry out a role does not have to involve human guidance in
the CSCL environment (Ge & Land, 2004). Computer-based collaborative environments offer potential in that support for collaboration can be provided automatically through scripts and prompts.
Scripts describe how to fulfill a role step by step, whereas prompts
guide what to do or say within that role. Both scripts and prompts
can be embedded software environments as forms of collaborative
support. Scripts and prompts are both structures designed to support collaborators to instantiate specific roles.
2.4. Scripts defined
Scripts consist of instructions regarding how group members
should collaborate and complete tasks through their respective
roles. Scripts are essentially tip sheets or recipes for acting out a
particular role or coordinating multiple roles. Cooperation scripts
specify and sequence collaboration through complex instructions
(Makitalo, Weinberger, Hakkinen, Jarvela, & Fischer, 2005). Scripts
can further be subdivided into social or epistemic scripts. Social
scripts describe how to structure and sequence discourse and collaborative activities, whereas epistemic cooperative scripts
describe the cognitive processes and strategies to be used in solving tasks (Weinberger, Ertl, Fischer, & Mandl, 2005). Providing
structure through scripts has the potential to reduce uncertainty
at both social and cognitive levels. As online learners lack visual
feedback from other students and are often deprived of knowledge

about the quality of their contributions, they remain uncertain as


818

R. Morris et al. / Computers in Human Behavior 26 (2010) 815–824

how to proceed (Kester & Paas, 2005; Makitalo et al., 2005). Past
research has found that social scripts motivate learners to refine
their beliefs and actions because they focus attention on the fact
that contributions will be reviewed by peers (Weinberger et al.,
2005). When students are given explicit instructions on how to
review a peer’s work, they are following a social script.
Epistemic scripts that facilitate cognitive processes may be
more task-specific and focused on the domain studied (Palincsar
& Herrenkohl, 2002). In one CSCL study, students used epistemic
scripts to structure engagement in a problem-oriented peer-discussion environment and a peer-tutoring environment (Weinberger et al., 2005).
2.5. Prompts defined
Prompts provide hints, suggestions and reminders for enacting
a role in the moment (GE & Land, 2004). Prompts, also known as
sentence openers or question stems, facilitate role taking and
enacting of scripts (Weinberger et al., 2005). Question prompts
are one way to direct the group to effectively collaborate, while
providing a balance between structure and flexibility (GE & Land,
2004). Prompts encompass procedural prompts, reflection
prompts, and elaboration prompts (GE & Land, 2004). Procedural
prompts help students complete specific tasks and learn cognitive
strategies. ‘Another reason that is good. . .’ is an example of this type
of prompt (GE & Land, 2004). Elaboration prompts help learners
articulate thoughts and elicit explanations (GE & Land, 2004).

‘Why is it important. . .’ or ‘what is a new example of. . .’ are examples
of elaboration prompts (King & Rosenshine, 1993). Finally, reflection prompts can encourage reflection on a metacognitive level.
An example of this type of prompt could be: ‘to do a good job we
need to. . .’ (Davis & Linn, 2000). Often the research examines all
three types of prompts, without distinguishing between these different classifications.
2.6. Review of the literature on prompts and scripts
Prompts and scripts are not always clearly defined in the
research literature and results on their effectiveness vary. A review
of the literature indicates mixed findings. Some studies report that
the use of scripts and prompts promotes collaboration and cognition during CSCL, whereas others report less conclusive findings
on the effectiveness of scripts and prompts. A salient theme in
the literature seems to be finding a balance between the optimal
amount of structure for supporting on task collaborative dialogue,
and the optimal level of structure for promoting motivated engagement. Scripts have been found, in some cases, to increase learning
and discourse. Rummel and Spada (2005) developed an instructional approach to improve CSCL by giving students the opportunity to learn from scripted collaborative problem-solving. This
mixed-methods design found that collaborating with a script had
positive effects on process and outcome: Discourse levels were
low with the unscripted group compared to the group that utilized
scripts. Scripts were also found to be particularly important for
supporting novice learners to coordinate dialogue in the first few
collaborative meetings. In comparison, unscripted conditions were
not effective for novice students. Despite these positive findings in
terms of increased discourse levels, qualitative findings indicated
that dyads in the scripted condition had initial motivational issues
as they found the scripts too rigid. Overall, findings may suggest
that complex cognitive skills benefit from scripting early in the collaborative process but may need to be phased out as students
develop competency in order to maintain optimal motivation for
engagement.
Beers, Boshuizen, Kirschner, and Gijselaers (2005) found that
when individuals worked in multidisciplinary teams to solve

complex problems, coercive scripts that constrained and directed
discussion were associated with negotiating common ground or a
common problem space. However, as coercion decreased, the participants communicated more and about a wider range of topics.
Findings such as this point to the importance of continued research
examining the optimal role of computer-based collaborative
scripts in enhancing collaborative processes and products. For
example, when script coercion is reduced over time and after practice, does it continue to have positive effects on negotiating common ground, while also increasing communication and
exploration of topics in a solution?
Comparing types of scripts is another avenue for future
research. Makitalo et al. (2005) claim that epistemic scripts have
potential to reduce uncertainty resulting in more discourse and
less information seeking in collaborative activities. Their findings
corroborated these hypotheses. Collaborators who used epistemic
scripts engaged in more active discussions and information seeking
behaviours, but did not achieve better collaborative outcomes.
Makitalo et al. (2005) interpreted these results as indicating that
learning environments should provide some degree of uncertainty
to necessitate beneficial interaction patterns, such as information
seeking. For instance, in Makitalo et al.’s (2005) research, the
unscripted group sought information in a direct and successful
manner, while the scripted group sought information more indirectly and less successfully. It may be that the epistemic script
restricted the learners too much in the sense that its prompts used
closed questions and, therefore, did not facilitate elaborative processes. Future research might examine the extent to which interaction should be structured on an epistemic level in order to support
the way learners cope with the uncertain situation of online learning. It appears that learning environments should provide some
degree of uncertainty to necessitate beneficial interaction patterns
including such things as information seeking.
Weinberger et al. (2005) extended Makitalo et al.’s (2005) work
by comparing epistemic scripts to social scripts. Prompts were provided in both cases to support the scripts. Findings indicated that
social scripts were beneficial with respect to prompting elaboration and supporting individual acquisition of knowledge while epistemic scripts were not. Weinberger et al. (2005) interpreted these

results as evidence that providing learners with an epistemic script
may not always result in individual knowledge acquisition, and it
may be important to augment epistemic scripts with social scripts.
Prompts, combined with scripts, were found to encourage students
to explore and discuss alternative viewpoints. However, too much
structuring through prompts and scripts may further impede interaction of learners when the script divides labour into tasks that can
be worked on by each learner individually. The authors purported
that, instead of receiving a task strategy, learners should be
prompted to construct a conceptual model themselves. Therefore,
scripts may sometimes need to make tasks more difficult for
learners.
In contrast, in Kester and Paas’ (2005) discussion of instructional interventions in CSCL environments, the authors purport
that scripts sometimes enhance social and cognitive processes.
However, instead of suggesting that tasks need to be more difficult,
Kester and Paas (2005) used cognitive load theory to suggest that
collaborative tasks that are too scripted may lead to cognitive overload thereby hindering learning. They further suggest that cognitive overload may be avoided by fading script support as
expertise increases giving more group control over support.
Prompts are less structured than scripts, yet these types of support are often used in conjunction. Providing prompts can aid complex discourse. Ge and Land (2004) reviewed the effects of
scaffolding ill-structured problem-solving processes using question prompts during CSCL. They proposed that students do not
always engage in high-level discourse unless prompted to do so.


R. Morris et al. / Computers in Human Behavior 26 (2010) 815–824

However, even when provided with question prompts, students
sometimes fail to use the prompts, thereby failing to attend to
important aspects of the task. As such, Ge and Land (2004) made
a number of suggestions to help guide future research about
prompts. For instance, Ge and Land (2004) suggest that prompts
should consist of specific questions targeting cognitive and metacognitive responses. Second, these responses should be stored in

a database for review and re-examination by learners. Finally, Ge
and Land (2004) stressed the need to investigate the relationship
between the effects of different scaffolding techniques and the
level of learner knowledge and experience. According to the
authors, it is possible that question prompts are more useful at
novice stages rather than later stages of content learning. At later
stages, examining how experts solve problems may become more
effective. Thus, there may be an interaction effect between type
of question prompt and amount of prior knowledge and experience. For example, reflection prompts may be more useful than
procedural prompts for learners with more knowledge, since procedural prompts may be redundant. Lazonder, Wilhelm, and Ootes
(2003) conducted one of the few published studies to examine
transfer effects of prompts. They too reported mixed findings
regarding the effectiveness of prompts in a CSCL context. Prompts
or sentence openers were implemented in a semi-structured chat
tool that allowed students to compose messages in a free chat area
or via sentence openers. In a second study, the tool was used to
explore the students’ unprompted use of sentence openers to
examine transfer effects. The authors examined alternative ways
to use sentence openers more effectively. The sentence openers
they used occurred from natural dialogue, where typical opening
phrases were taken from students’ chat history files. The sentence
openers were brief thus increasing the chance students would use
their own sentence openers. Also, the openers were not imposed
and instead were implemented in a semi-structured chat tool.
Results indicated that students hardly used sentence openers and
were skeptical of their usefulness. These results are surprising as
early studies show that students responded positively to sentence
openers. One possible explanation of these findings is that studies
were based on synchronous interactions when CSCL, and particularly electronic chat discussions, were relatively new activities
for learners. A recent proliferation of chat tool use for personal

and work-based communication means that students have already
developed natural chat patterns, styles, and symbols. These
advanced chat users may feel constrained by chat tools that offer
scripted prompts and resist their use. On the other hand, advanced
chat users send chat responses so rapidly, they often do not reflect
on the content. Perhaps the use of prompts offers promise in that it
slows down interaction, affording more opportunities for reflection. This finding is supported when examining asynchronous writing where the researchers found that students were more reflective
and thoughtful. Nevertheless, differences in chat response rates,
reflection activities, and use of prompts across users with varying
degrees of text-based chat experience warrants further
investigation.
In summary, the limited number of studies examining the effectiveness of scripts and prompts in CSCL environments report mixed
findings regarding the psychological and social advantages and disadvantages. It appears that too much structure constrains dialogue
to the scripted role. Too little structure, however, provides little
support or guidance and results in a lot of talk that may not be
on task. Epistemic scripts increase discourse and decrease information seeking, yet do not seem to affect product outcome, whereas
social scripts appear to have some influence on group and individual product. Furthermore, findings indicate that epistemic scripts
may need to be paired with social scripts particularly for novice
learners. Some researchers have found prompts support scripts in
many important ways; however, they appear to be most useful

819

for novices and, like all scaffolds, should be faded out. The transfer
effects of prompts and scripts have yet to be explored thoroughly.
Some researchers feel prompts and scripts should actually make
tasks more difficult to promote cognitive growth, whereas others
suggest that prompts and scripts introduce increased cognitive
overload. To compound the problem, online educators may not
have the time nor the resources to adequately monitor roles supported by scripts and prompts.

To examine these issues more systematically, researchers need
to be able to draw on tools that allow them to manipulate types
of support, levels of support, and frequency of support while
simultaneously collecting data about collaborative dialogue,
behavior, beliefs, and regulation. A challenge for conducting
research about the effectiveness of different types of roles, scripts,
and prompts across tasks of varying difficulty and for learners of
varying levels of content, collaboration, and chat expertise is that
the field lacks flexible chat tools for manipulating this range of
experimental conditions.
2.7. Introduction to cognitive tools for collaboration in gStudy
gStudy (Winne, Hadwin, Nesbit, Kumar, & Beaudoin, 2005) is a
powerful tool for supporting and researching solo and collaborative regulation of learning. As a learning tool, gStudy provides users
with a range of tools and prompts such as notetaking templates
that guide the use of a range of notetaking and processing strategies, a strategy library for looking up information about strategies,
a glossary for keeping track of new terms and concepts, and a chat
tool with embedded prompts for supporting and guiding synchronous text-based chat discussions. As an instructional tool, gStudy
provides a shell for embedding instructional texts, activities, performance measures, images and organizing that information for
learners. As a research tool, gStudy: (a) provides a flexible arena
for comparing different instructional and experimental conditions,
and (b) records everything the learner does in terms of keyboard
and mouse clicks, text and menu selections, and text inputs in a
detailed time-stamped logfile. These precise traces of student
interaction within a learning tool provide data upon which to
examine, among other events, the interaction between chat-based
discussions and students’ actual learning and navigating activities.
Following, we describe how aspects of gStudy can be used to further our understandings of roles, scripts, and prompts for enhancing
computer supported collaborative learning processes and outcomes
across a range of learner, role, script, and prompt conditions.
2.8. Operationalizing roles, scripts, and prompts in gStudy

gStudy supports the use of roles, scripts, and prompts in collaboration through its sophisticated text-based chat tool (gChat). As
with other chat tools, such as MSN and Skype, gChat (Hadwin,
Gress, Winne, & Jordanov, 2006) enables multiple users to engage
in synchronous, or ‘‘real time,” chat. Participants can click on the
gChat icon in gStudy’s interface and be instantly connected to other
participants online. Furthermore, gChat supports one-on-one chat,
as well as multi-user chat in which multiple participants can chat
simultaneously. Messages in gChat are viewed in a split screen text
box comprised of: (a) an upper box displaying the ongoing chat
between the individuals (Fig. 1), and (b) a lower box displaying
what the user is typing prior to be submitted to the online discussion (Fig. 1). Thus, participants have a private area to compose their
text-based response and a public area to share and view the chat.
Unlike conventional chat tools, gChat supports the use of roles,
scripts, and prompts in collaboration. In other words, it provides
both structure and cues to scaffold collaborative work. The use of
roles and prompts is supported directly within gChat, while the
use of scripts is facilitated by a number of different gStudy tools


820

R. Morris et al. / Computers in Human Behavior 26 (2010) 815–824

Fig. 1. gChat tool in gStudy with task manager role prompts highlighted.

(see Winne, Hadwin, & Gress (this issue) for a full description of
gStudy tools). Additionally, roles, scripts, and prompts within gStudy and gChat are configurable on the fly by the researcher or
instructor affording opportunities to investigate and facilitate a
wide range of collaborative issues.
Roles in gStudy are enabled through gChat. When users enter a

chat they can choose amongst a set of pre-stocked collaborative roles
that provide structure and cues to scaffold collaborative work. A
drop-down menu in the bottom right corner of the chat window provides a drop-down list of roles defined by a researcher or instructor
(Fig. 1). Selecting a role populates a list of sentence starters or
prompts in a drop-down menu to the left of the role (Fig. 1). When
a user selects a prompt, it automatically populates the text entry
window where it can be edited, augmented and/or sent to the chat.
For example, in Fior (2008) and Morris (2008), grade 10 secondary
school students in classroom environment utilized gChat to engage
in collaborative discussion of a difficult reading task. gChat enabled
these participants to choose between four reciprocal teaching roles
to facilitate their discussion. Roles were based on Palincsar and
Brown’s (1984) reciprocal teaching roles: (a) summarizer, (b) questioner, (c) clarifier, and (d) predictor. After reading and marking up a
very challenging text in an independent study session, students
came together in groups of four to discuss the text. The role of the
summarizer was to prompt the group to synthesize text information.
The questioner guided the group to ask questions regarding the text.
The clarifier guided the group to clarify and simplify terms or concepts that were unclear in the text. Finally, the predictor helped
the group to hypothesize about the kinds of things they would need
to know for an upcoming comprehension test. Each role was accompanied with specific guidance for enacting that role through both
scripts and prompts described below.
Scripts in gStudy. As participants may be unclear about how to
carry out a particular role, gStudy can further reduce uncertainty
and scaffold collaboration by providing users with detailed scripts,
or instructions, for each role. gStudy supports the use of scripts in
at least two ways. First, scripts for each role can be stocked in a
multi-media kit in gStudy. When participants select a script in
the kit, they gain access to specific multi-media instructions for
carrying out various collaborative roles and the purpose of each
role. As gStudy supports text, images, video and audio, script

instructions can include multi-media examples and models of each
script being instantiated. Furthermore, participants can keep script
information in gStudy visible while they participate in a collaborative discussion. As such, each member of a group can monitor their
chat activities to ensure they are adequately fulfilling their role in
the collaborative task.

In Fior (2008) and Morris (2008), scripts for all for reciprocal
teaching roles were stored in a collaboration kit within gStudy.
The scripts for each role were presented in this kit in much the
same way that a textbook passage would be presented in a webpage. Fig. 2 illustrates a script presented for a student assigned
the role of questioner in a collaborative discussion. The script
includes information about the role of a questioner as well as when
and how to enact the questioning role in a collaborative discussion.
While this example script is limited to text and images, it is also
possible to augment the script with audio or video snippets.
In gStudy, learners are able to access script information via a
table of contents in gStudy prior to engaging in their chat activity.
Learners can highlight scripts, make their own crib notes from the
scripts, and create their own audio-, text-, or image-based examples of scripts. In this way, the script becomes customizable and
revisable by a student, rather than being a static document to be
pulled up as a reference. Furthermore, scripts can be kept visible
throughout collaborative discussions so that learners can use them
as a guide, add to them, or make links in the chat to specific
instructions in the script.
Other tools in gStudy, such as chat logs, notes, labels and glossaries, can also help students to review, reflect upon and evaluate
their use of scripts. For instance, the contents of every gChat are
recorded and saved as a chat log. Chat logs can be uploaded in gStudy much like any other content document. This affords opportunities for students to annotate, index, or graphically map connections
between chats. For example, students may be assigned a task of
selecting strings of text in gChat and making links that correspond
to various steps or activities outlined in the collaborative script.

Students can use this information to make data-based evaluations
of their own strengths and weaknesses in contributing to the collaborative discussion through that particular role. In addition,
when chat logs are uploaded to gStudy, they become information
objects just like any other text or note. Thus, chat logs can be
hyperlinked to reflection notes, to do notes, annotation labels, glossary definitions, and other chat logs to facilitate future collaborative activity or ongoing collaborative project work. Finally, after
students have gained experience using roles, scripts, and prompts,
script templates can be used to scaffold learners in generating their
own collaborative scripts. For example, building on the script
example provided in Fig. 2, learners may be asked to fill in a script
note such as the one presented in Fig. 3.
Prompts in gStudy. gStudy provides additional structure for role
engagement by facilitating the use of prompts in gChat. In this
environment, prompts are quick sentence starters and statements
students can use to engage with their assigned chat role. When


R. Morris et al. / Computers in Human Behavior 26 (2010) 815–824

821

B. Questioner Script
What will I be doing?
Yo u have already read a very difficult text on crystal methamphetamine. To assist your group in
understanding that text, you will be discussing it collaboratively. To help you do this, each
person in your group has been assigned a different role (summarizer, questioner, clarifier, and
predictor). Please do the following things:
1. Read about your role below
2. In your group, take turns (explained below) using your assigned roles to guide the
group discussion.
3. Use the prompts that you were shown in gChat (listed below) to help you with this

task.
4. Try to answer any questions other group members ask.
What is my role?
Yo u are the Questioner: As the questioner, you will get your group to generate questions. For
the main ideas that you read, have your group write down a question that the main idea will
answer. Good questions should include words like “who, “where”, “when”, “why”, and “what”.
For example, if you are reading an article about the extinction of the dinosaurs, you might find
the following main idea: “Most scientists now believe that the extinction of dinosaurs was caused
by a large meteor striking the earth.” Yo u could then write this question: “What event do most
scientists now believe caused the mass extinction of the dinosaurs?” Also, have your group think
about the types of questions other people (including themselves) may have about the text.
When do I ask questions?
Section 1: neurological effects
• As the questioner you will ask your question, after the summarizer’s question has been
answered.
• Yo u will continue your group’s discussion by getting your group to create a question that
relates to the first paragraph. Yo u may want to use one of the prompts in gChat to help
you get started.
• After the question has been created, the clarifier and predictor will take turns asking their
questions that relate to the first paragraph. Try to answer at least one of the questions
your group asks.
• When each of you has had a turn asking a question, move on to the second paragraph and
after the summarizer’s question has been answered, continue your group’s discussion by
getting them to create another question that relates to the second paragraph.
• Repeat this process for the last paragraph.
Section 2: social issues
• Now that you have some experience helping your group create questions, decide when it
makes sense to do this for the social issues text.
• Don’t forget to answer questions and use the prompts provided.
How do I ask questions?

Use the Prompts in gChat:
•
•
•
•
•

Did you have any questions about...
What are you curious about?
What might your mom or dad ask about this?
What might a friend ask about this?
What question will the main idea answer?
Fig. 2. Example of a questioner script presented in gStudy (Fior, 2008; Morris, 2008).

students participating in multi-user chat select a particular role,
they gain access to role-specific prompts situated in a drop-down
menu in the gChat interface (see Fig. 1). When a user selects a
prompt, it immediately appears as text in the send field. The user
can augment, edit, or simply use the prompt in order to facilitate
their participation in the discussion. As previously discussed,
prompts elaborate the meaning of the role by providing a series
of statements that illustrate what someone in that role might say
to the group. Second, prompts provide a quick way for users to
contribute to the chat discussion.
In Fior (2008) and Morris (2008), each collaborative role was
stocked with a specific set of prompts to facilitate role

engagement. Participants gained access to prompts through a
drop-down menu in the chat window. As students engaged in
a collaborative activity to increase comprehension of a difficult

text, participants were guided to use these prompts to gain a
better understanding of the function of their role, to provide
cues about the type of contributions a person in his or her role
might make to the discussion, and to quickly add statements or
text to the chat (Table 1).
In addition to supporting the use of roles, scripts, and prompts
in collaboration, gStudy also enables systematic investigation of
this type of support by recording time-stamped detailed records
of both chat content and participants’ instantiation of roles and


822

R. Morris et al. / Computers in Human Behavior 26 (2010) 815–824

Fig. 2 (continued)

Fig. 3. Script note template for creating new scripts.

prompts. For instance, each time a chat participant utilizes a
prompt within a particular role, gStudy provides a time-stamped
record of its use. Furthermore, gStudy also records fine-grained
trace data of students’ interaction with scripts in the gStudy browser by recording activities such as mouse clicks, scrolling, and
engagement with tools, such as labelling and notes.

As such, the roles, scripts, and prompts embedded in gStudy
enable the investigation of a number of pertinent questions in this
field, such as the type and degree of support students require in
different types of collaborative tasks. For instance, Fior (2008) utilized gStudy roles, scripts, and prompts in a grade 10 classroom
environment to investigate whether students who receive more

823

R. Morris et al. / Computers in Human Behavior 26 (2010) 815–824
Table 1
Examples of roles, scripts, and prompts.
Roles

Scripts

Prompts

Predictor

You are the Predictor: As the predictor, your role will be to help your group see how one part of the text is related to
or predicts other parts of the text. You will also help your group hypothesize what the consequences may be for an
action or event described in the text. For instance, if you read about the introduction of a foreign species to a fragile
ecosystem, you may want to ask your group about the consequences of that action

Summarizer

You are the Summarizer: As the summarizer, you will ask your group questions to help them sum up the
information in the text. The article can be summarized across sentences, across paragraphs, and across the article as
a whole. Stop after each paragraph or major section of the passage. Ask your group to construct a sentence that sums
up only the most important idea(s) that appear in that part. Good summary sentences include key concepts or
events but leave out less important details

-

Questioner

You are the Questioner: As the questioner, you will get your group to generate questions. For the main ideas that
you read, have your group write down a question that the main idea will answer. Good questions should include
words like ‘‘who, ‘‘where”, ‘‘when”, ‘‘why”, and ‘‘what”. For example, if you are reading an article about the
extinction of the dinosaurs, you might find the following main idea: ‘‘Most scientists now believe that the extinction
of dinosaurs was caused by a large meteor striking the earth.” You could then write this question: ‘‘What event do
most scientists now believe caused the mass extinction of the dinosaurs?” Also, have your group think about the
types of questions other people (including themselves) may have about the text

-

Clarifier

You are the Clarifier: As the clarifier, your role will be to get your group to clarify anything that is unclear.
Sometimes in your reading you will run into words, phrases, or whole sentences that really don’t make sense. Here
are some ways that you can get your group to clarify the meaning:Unknown words. If your group comes across a
word whose meaning they do not know, suggest they read the sentences before and after to see if they give clues to
the word’s meaning.Unclear phrases or sentences. Suggest your group rereads the phrase or sentence carefully and try
to understand it. Get them to think of other interpretations or examples

structure (in the form of roles, scripts, and prompts) in a collaborative reading comprehension discussion participated more in group
discussions. Fior (2008) found that the relationship between participation and self-efficacy is different when students are given
roles, scripts, and prompts to guide their discussion as opposed
to when they just engage in open text-based chat about an
assigned reading. Specifically, she found that students who were
given roles, scripts, and prompts were less likely to have participation rates that were highly correlated with their initial self-efficacy
for collaboration. In comparison, students who were not supported
in their collaboration through roles, scripts, and prompts had participation rates that were highly correlated with their entering selfefficacy.
Other pertinent avenues of research enabled by gStudy’s facilitation of roles, scripts, and prompts in collaboration include examination of the effectiveness of these tools in supporting students

to engage in both individual and shared regulation of collaborative
processes as well as the co-regulation of group collaborative processes. For instance, as roles, scripts, and prompts are configurable
by the researcher, they can be modified to target different types of
collaboration. The scripts included in Fior (2008) and Morris’
(2008) studies targeted individual self-regulation of collaboration
by scaffolding students’ engagement in different types of comprehension roles.
3. Future directions: Using roles, scripts, and prompts in gStudy
to support and research self-regulated learning
In our continued work examining ways to support students in
self-regulating learning we plan to use these tools to compare different types of roles, scripts, and prompts such as those targeting
cognitive processing versus those targeting self-regulatory processes to explore the effectiveness of these roles on the development of strategy knowledge, strategy use, monitoring activity,
and learning outcomes. One set of roles, scripts, and prompts will
target cognitive processing of information such as selecting important ideas and concepts, monitoring understanding and recall,

-

What might we need to know?
What might happen if...
How might that affect...
How might that relate to...
Are there consequences to...
What is this part about?
What was the main point?
What did you get from this?
Can you put that in your own
words?
Can you give a general summary of. . .. . .?
Did you have any questions
about...
What are you curious about?

What might your mom or dad
ask about this?
What might a friend ask about
this?
What question will the main
idea answer?
Can anyone explain...
What do you think that means?
Are
there
any
other
interpretations?
Can you think of an example?
How can we make sense of
that?

assembling or making connections between ideas, rehearsing to
be learned concepts, and translating concepts into a group’s own
words and graphical, textual, or pictorial representation. A second
set of roles, scripts, and prompts will target the self-regulatory
cycle proposed by Winne and Hadwin (1998): defining or coming
to a common understanding of the assigned task, articulating goals
and plans for completing the task, enacting the task by drawing on
a repertoire of strategies and techniques for task completion, and
continually evaluating and adapting self-, task-, group-, and concept knowledge and processes. Harnessing the configurability of
roles, scripts, and prompts in gStudy and gChat, these studies facilitate addressing questions such as: What kinds of roles, scripts, and
prompts promote the development of strategy use and strategy
knowledge in collaborative settings? What kinds of roles, scripts,
and prompts result in better task understanding and performance?

A second line of research will compare roles, scripts, and
prompts that focus on modeling and explaining processes to
peers in a group to those focusing on questioning and prompting
members of a group to engage, and to a third set of roles,
scripts, and prompts that emphasize co-constructing meaning
with fellow group members. Specifically, we examine three
questions: (1) How do students use collaborative scripting tools
designed to guide small groups in sharing the regulation of
learning? (2) Are these collaborative scripting tools effective
for promoting socially shared regulation of learning (e.g., common task perceptions, shared goals, collective efficacy, and collective strategy knowledge)? (3) To share in the regulation of
learning, do students require well-developed self-regulatory
knowledge and skills?
Acknowledgements
Support for this work was provided by grants to Philip H. Winne
from the Social Sciences and Humanities Research Council of Canada (410-2002-1787; 512-2003-1012, R. Azevedo, A. F. Hadwin, S.
Lajoie, J. Nesbit, & V. Kumar, -Co-Investigator), the Canada Research
Chair program, and Simon Fraser University; and to Allyson Had-


824

R. Morris et al. / Computers in Human Behavior 26 (2010) 815–824

win from the Social Sciences and Humanities Research Council of
Canada (410-2001-1263).
References
Beers, P. J., Boshuizen, H. P. A., Kirschner, P. A., & Gijselaers, W. H. (2005). Computer
support for knowledge construction in collaborative learning environments.
Computers in Human Behavior, 21, 623–643.
Carroll, J. M., Neale, D. C., Isenhour, P. L., Rosson, M. B., & McCrickard, D. S. (2003).

Notification and awareness: Synchronizing task-oriented collaborative activity.
International Journal of Human-Computer Studies, 58, 605–633.
Chou, C., Lin, C., & Chan, T. (2002). An approach to developing computational
supports for reciprocal tutoring. Knowledge-Based Systems, 15, 407–412.
Cohen, E. G. (1994). Restructuring the classroom: Conditions for productive small
groups. Review of Educational Research, 64, 1–35.
Davis, E. A., & Linn, M. (2000). Scaffolding students’ knowledge integration: Prompts
for reflection in KIE. International Journal of Science Education, 22, 819–837.
Doyle, W. (1983). Academic work. Review of Educational Research, 53, 159–199.
Fior, M. N. (2008). Self and collective efficacy as correlates of group participation: A
comparison of structured and unstructured computer-supported collaborative
learning conditions. Unpublished master’s thesis, University of Victoria.
GE, X., & Land, S. M. (2004). A conceptual framework for scaffolding ill-structured
problem-solving processes using question prompts and peer interactions.
Educational Technology Research and Development, 52, 5–22.
Hadwin, A. F., Gress, C. L. Z., Page, J., & Ross, S. (2005). Computer Supported
Collaborative Work: A review of the research 1999–2004. Poster presented at the
Annual Meeting of the Canadian Society for the Study of Education, London, ON,
Canada.
Hadwin, A. F., Gress, C., Winne, P. H., & Jordanov, M. (2006). GChat: A chat interface
with scaffolds to enhance collaborative effectiveness (version 1.0) [computer
program]. Burnaby, BC: Simon Fraser University.
Hadwin, A. F., Winne, P. H., & Nesbit, J. C. (2005). Annual review: Roles for software
technologies in advancing research and theory in educational psychology
[Invited Annual Review Article]. British Journal of Educational Psychology, 75,
1–24.
Hsieh, I. G., O’Neil, J. R., Harold, F., & Rossier, A. B. (2002). Types of feedback in a
computer-based collaborative problem-solving group task. Computers in Human
Behavior, 18, 699–715.
Jenkins, J. R., & O’Connor, R. E. (2003). Cooperative learning for students with

learning disabilities: Evidence from experiments, observations, and interviews.
In S. Graham, K. Harris, & L. Swanson (Eds.), Handbook of Learning Disabilities
(pp. 417–430). New York: Guildford.
Johnson, D. W., & Johnson, R. T. (1989). Cooperation and competition: Theory and
research. Edina, MN: Interaction Book Co.
Johnson, D. W., Johnson, R. T., & Holubec, E. J. (1993). Circles of Learning: Cooperation
in the Classroom (4th ed.). Edina, Minn.: Interaction Book Company.
Kester, L., & Paas, F. (2005). Instructional interventions to enhance collaboration in
powerful learning environments. Computers in Human Behavior, 21, 689–696.
King, A. (1999). Discourse patterns for mediating peer learning. In A. M. O’Donnell &
A. King (Eds.), Cognitive perspectives on peer learning (pp. 87–117). Mahwah, NJ:
Lawrence Erlbaum Associates.
King, A., & Rosenshine, B. (1993). Effects of guided cooperative questioning on
children’s knowledge construction. Journal of Experimental Education, 61,
127–148.
Lazonder, A. W., Wilhelm, P., & Ootes, S. A. W. (2003). Using sentence openers to
foster student interaction computer-mediated learning environments.
Computers & Education, 41, 291–308.
Liang, L. A., & Dole, J. A. (2006). Help with teaching reading comprehension:
Comprehension instructional frameworks. The Reading Teacher, 59, 742–753.

Makitalo, K., Weinberger, A., Hakkinen, P., Jarvela, S., & Fischer, F. (2005). Epistemic
cooperation scripts in online learning environments: Fostering learning by
reducing uncertainty in discourse? Computers in Human Behavior, 21, 603–622.
McMaster, K. N., & Fuchs, D. (2002). Effects of cooperative learning on the academic
achievement of students with learning disabilities: An update of TateyamaSniezek’s review. Learning Disabilities Research & Practice, 17, 107–117.
Morris, R. (2008). The effectiveness of roles, scripts, and prompts in promoting reading
comprehension during computer-supported collaborative learning. Unpublished
master’s thesis, University of Victoria.
Mulryan, C. M. (1992). Student passivity during cooperative small groups in

mathematics. Journal of Educational Research, 85, 261–273.
O’Donnell, A. M. (1999). Structuring dyadic interaction through scripted
cooperation. In A. M. O’Donnell & A. King (Eds.), Cognitive perspectives on peer
learning (pp. 179–196). Mahwah, NJ: Lawrence Erlbaum Associates.
O’Donnell, A. M., Hmelo-Silver, C. E., & Erkens, G. (Eds.). (2005). Collaborative
learning reasoning and technology (1st ed.). Mahwah, New Jersey: Lawrence
Erlbaum Associates, Inc.
O’Donnell, A. M., & King, A. (1999). Cognitive perspectives on peer learning. Mahwah,
New Jersey: Lawrence Erlbaum.
Palincsar, A. S., & Brown, A. L. (1984). Reciprocal teaching of comprehensionfostering and comprehension-monitoring activities. Cognition and Instruction, 1,
117–175.
Palincsar, A. S., & Herrenkohl, L. R. (2002). Designing collaborative learning contexts.
Theory into Practice, 41, 27–32.
Robertson, J., & Good, J. (2003). Using a collaborative virtual role-play environment
to foster characterisation in stories. Journal of Interactive Learning Research, 14,
5–29.
Rummel, N., & Spada, H. (2005). Learning to collaborate: An instructional approach
to promoting collaborative problem solving in computer-mediated settings. The
Journal of the Learning Sciences, 14, 201–241.
Slavin, R. E. (1995). A model of effective instruction. Educational Forum, 59, 166–176.
Soller, A. (2001). Supporting social interaction in an intelligent collaborative
learning system. International Journal of Artificial Intelligence in Education, 12,
166–187.
Strijbos, J., Martens, R. L., Jochems, Wim M. G., & Broers, N. J. (2004). The effect of
functional roles on group efficiency using multilevel modeling and content
analysis to investigate computer-supported collaboration in small groups. Small
Group Research, 35, 195–229.
Van der Puil, C., Andriessen, J., & Kanselaar, G. (2004). Exploring relational
regulation in computer-mediated (collaborative) learning interaction: A
developmental perspective. Cyberpsychology & Behavior, 7, 183–195.

Vygotsky, L. S. (1978). Mind in society: The development of higher psychological
processes. Cambridge, MA: Harvard University Press.
Webb, N. M. (1992). Testing a theoretical model of student interaction and learning
in small groups. In R. Hertz-Lazarowitz & N. Miller (Eds.), Interaction in
cooperative groups: The theoretical anatomy of group learning (pp. 102–119). New
York: Cambridge University Press.
Weinberger, A., Ertl, B., Fischer, F., & Mandl, H. (2005). Epistemic and social
scripts in computer-supported collaborative learning. Instructional Science, 33,
1–30.
Winne, P. H., & Hadwin, A. F. (1998). Studying as self-regulated engagement in
learning. In D. Hacker, J. Dunlosky, & A. Graesser (Eds.), Metacognition in
Educational Theory and Practice (pp. 277–304). Hillsdale, NJ: Lawrence Erlbaum.
Winne, P. H., Hadwin, A. F., & Gress, C. L. Z. (2009). The learning Kit project: Software
tools for supporting and researching regulation of collaborative learning.
Computers and Human Behavior, this issue. doi:10.1016/j.chb.2007.09.009.
Winne, P. H., Hadwin, A. F., Nesbit, J. C., Kumar, V., & Beaudoin, L. (2005). GStudy: A
toolkit for developing computer-supported tutorials and researching learning
strategies and instruction (version 2.0) [computer program]. Burnaby, British
Columbia, Canada: Simon Fraser University.