You are on page 1of 17

Learning and Instruction 15 (2005) 571e587 www.elsevier.

com/locate/learninstruc

Tutor scaffolding styles of dilemma solving in network-based role-play


Kai Pata a,*, Tago Sarapuu a, Erno Lehtinen b
b a Science Didactics Department, University of Tartu, Tartu, Estonia Centre for Learning Research and Department of Teacher Education, University of Turku, Turku, Finland

Abstract This study investigated tutoring during collaborative role-play dealing with environmental dilemmas in a synchronous network environment. The relationships of different tutor scaffolding styles with students discourse acts and their decision-making variables were studied. A role-play with a jigsaw design was developed using the model of real-life environmental negotiations. Tutor scaffolding supported the process of making judgements. Five role-plays consisting of 31 decision-making discussions were carried out with 14e17-year-old students (N Z 62) from Estonian secondary schools. The ndings indicated that the activeness of tutor scaffolding was related to the higher frequency of students task-related discourse acts, as well as their ability to generate a mutually accepted ownership of problem representations in teams. 2005 Elsevier Ltd. All rights reserved.
Keywords: Verbal scaffolding styles; Dealing with environmental dilemmas; Synchronous learning environment

1. Introduction In this study, we analyse the role of verbal tutor scaffolding during the process of solving environmental dilemmas in a role-play in a network-based synchronous chat environment. The chat room is a comparatively new medium for carrying out learning activities. Resembling

* Corresponding author. Centre of Science Didactics, University of Tartu, 46 Vanemuise Street, 51014 Tartu, Estonia. Tel.: C372 7 375809; fax: C372 7 375082. E-mail address: kpata@ut.ee (K. Pata). 0959-4752/$ - see front matter 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.learninstruc.2005.08.002

572

K. Pata et al. / Learning and Instruction 15 (2005) 571e587

face-to-face discussions, chat rooms are more suitable for various decision-making activities (Newman, 1996) than asynchronous forums, but the interactions of participants in chat rooms appear to be less controllable compared with classroom situations. As learners are physically located at a distance from each other, the role of the tutor in facilitating the task-focused activities is of great importance (Hakkarainen & Lipponen, 1998). In order to understand how the tutor inuences this learning setting it was necessary to nd out what the role of tutor-scaffolding activities is during chat discussions, and how the application of different tutoring styles could be related to the effectiveness of team decision-making. This study was aimed at investigating the occurrence and relationships of tutors and students scaffolding and task-related discourse acts in different discussion groups. 1.1. Scaffolding the network-based dilemma management It has been demonstrated that spontaneous explorations and group discussions about complex problems are not sufcient for higher-order learning (Brown & Campione, 1996; Brown & Palinscar, 1989). The metaphor of scaffolding (Wood, Bruner, & Ross, 1976) is used to illustrate the effective supportive processes that enable the learners to solve a problem, carry out a task, or achieve a goal that would be beyond their unassisted efforts. Scaffolding means providing support at the right level of the current skill while a student is carrying out the task and then gradually fading out of assistance (Jarvela, 1995) or providing assistance to students on an as-needed basis with fading out of assistance as the competence increases (Pressley, Hogan, Wharton-McDonald, Mistretta, & Ettenberger, 1996). According to the concept of the zone of proximal development (Vygotsky, 1978), scaffolding helps to decrease the distance between students actual developmental level, as determined by independent problem solving, and the level of their potential development as determined through problem solving under adult or peer guidance. Scaffolding involves recruiting learners interest, reducing their choices, maintaining their goal orientation, highlighting critical aspects of the task, controlling their frustration, and demonstrating activity paths to them (Wood et al., 1976). It has been shown that learners can be assisted in building new knowledge structures, both with the help of the teacher who models the desired learning strategy or task and then gradually shifts responsibility to the students, or with the help of more advanced peers who help the less sophisticated students in the learning process (Pressley et al., 1996). In network-based learning environments, different kinds of supportive interventions are necessary. Students need help in familiarising themselves with the technological possibilities of the learning interface; they must be guided to follow their path within the learning activity, while becoming aware of their own learning processes; and social interaction has to be facilitated (Berge & Collins, 1995; Collis, Andernach, & VanDiepen, 1997). Scaffolding is used for helping students to generate content or perform tasks by suggesting metastrategies such as instructioning, questioning, and cognitive structuring, which then enable students to organize their learning activities (Tharp & Gallimore, 1988). Guzdial, Carlson, and Turns (1995) distinguish between scaffolding for collaboration and scaffolding for problem solving. The latter is dependent on the type of task and consists of process scaffolding (instructions and feedback) and of content scaffolding (hinting, prompting, summarising, etc.). The scaffolding types discussed in the current study are process- and content-related scaffolding. The studies of VanLehn, Siler, Murray, Yamauchi, and Baggett (2003) have shown that the tutors moves in response to similar student actions are not consistent, and there does not appear to be any relationship between the use of a specic tutoring tactic and learning. This suggests

K. Pata et al. / Learning and Instruction 15 (2005) 571e587

573

that the tutors scaffolding style is not consistent. Many studies, on the other hand, have reported that tutors tend to use certain styles in supporting learners: some tutors value students learning process, while others concentrate on the learning atmosphere and the facilitators role (Pinto, Rendas, & Gamboa, 2001); tutors may focus either on promoting students understanding by being explanation-oriented or by facilitating their factual-knowledge-acquiring processes (Hakkarainen, Lipponen, & Jarvela, 2002); they may act in the role of Controllers, taking more initiatives than students or as Guides, allowing students to prevail (Rasku-Puttonen, Etelapelto, Hakkinen, & Arvaja, 2002). Student groups also tend to appreciate different styles of tutoring (Dolmans, Wolfhagen, & Snellen-Ballendong, 1994). Several studies (Hakkarainen et al., 2002; Rasku-Puttonen, Etelapelto, Arvaja, & Hakkinen, 2003) have shown that the teachers scaffolding style has an inuence on students performance. Hakkarainen et al. (2002) showed that the explanation-oriented teacher who used more content-related prompts obtained deeper knowledge-building patterns in students than the factual-knowledge-oriented teacher, whose interventions were not explicitly focused on explanation. Rasku-Puttonen et al. (2003) demonstrated that the Controller-teacher and his students negotiated more than the Guide-teachers students about the meaning of the task and the subject in hand. The Controller-teacher was able to enhance students self-regulation skills, whereas the Guide-teachers students asked more for help and support from the teacher at the beginning of the activity, but by the end of the project they appeared to be more focused on problem solving than the Controller-teachers students (Rasku-Puttonen et al., 2003). Many of these studies have investigated the long-term effects of tutors style on students performance. The current study focuses on the effects of tutors interactions on students performance during short real-time synchronous network-based discussions. 1.2. Evaluating the process of managing dilemmas In their study on scaffolding, Chi, Siler, Jeong, Yamauchi, and Hausmann (2001, pp. 489e 491) classied the tutors and students actions separately during collaborative face-to-face activity. The tutors utterances during scaffolding were: giving explanations, answering the students questions, answering content questions, scaffolding with generic and content prompts, giving direct feedback and asking comprehension-gauging questions. The tutor also used reading text sentences aloud and making self-monitoring comments. Since the dialogues were tutor-dominated, fewer students action categories were classied: asking questions, answering questions, responding to the tutors scaffolding prompts, reecting, reading sentences aloud, and spontaneously giving unprompted self-explanations. Chi et al. (2001) were looking at the scaffolding interactions as tutor-centred. Although the tutor is the only actor deliberately using scaffolding during joint decision-making discussions, it is obvious that more knowledgeable peers will also inuence the learning situation by peer scaffolding. Wu, Farrell, and Singley (2002) have found that student tutors are performing problem-solving facilitation techniques similar to those of adult tutors described by Graesser, Person, and Magliano (1995). On the basis of the above-mentioned studies, we aimed at developing a categorizing system, which would cover both tutors and peers discourse acts related to process and content scaffolding (task management), as well as decision-making acts (decisions, arguments, positive and negative feedback), and those dealing with social level relations. In order to assess the impact of different scaffolding discourse acts on collaborative judgement making, the group-negotiation process and the groups nal decisions have to be evaluated. While dealing with the dilemmas, the members of different interest groups need to reach

574

K. Pata et al. / Learning and Instruction 15 (2005) 571e587

common ground in the shared problem space, thus obtaining coherent presuppositions of their mutual knowledge (Clark & Schaefer, 1989). Reaching common ground is one of the variables characterising effective discussions in teams (Clark & Schaefer, 1989; Lehtinen, 2003). The more similar the nal problem representations of the individuals are at the end of the discussion, the more likely it is that they have reached common ground, necessary achievement for completing the tasks in their team. The coherency of the individuals nal problem representations can, therefore, be a variable expressing the extent of achievement of common ground in the team. One aspect of reaching common ground is the willingness and openness of the negotiators to change their problem representations towards the other team members viewpoints, as well as taking ownership of shared problem representation. The term ownership refers to our ability to take responsibility for negotiating meaning and, thus, to make some meanings our own (Wenger, 1998). The team members mutual willingness to take ownership of shared problem representations can be evaluated by comparing their nal individual problem representations with the nal team decisions. The teams with less willingness to take ownership might use a single persons viewpoints as the basis of the nal decision, whereas teams that are more open to each others opinions will sum up each others viewpoints, accepting the peers opinions, or will make a joint decision based on common ground after the discussion, sharing the ownership fully. For each dilemma, several problem representation models could be found, because of the different actors involved in the situation. Environmental dilemmas are characterised by the opposition of economic, ecological, ethical and legal viewpoints (Liebrand & Messick, 1996; Zandbergen & Petersen, 1995). Common ways of solving environmental problems are strategies directed towards avoiding the causes of the problem and those dealing with the consequences of the problem. By combining the dimensions of economic, ecological, ethical and legal viewpoints with two strategies of problem-solving, different problem representation models can be derived for the counterparts of the dilemma. In rational decisions for different role groups, presumably only specic aspects of the problem should be emphasised. Above all, each role group should consider its own aims and choose optimal solutions according to these. In order to make competent expert decisions that take into account the needs of other actors in the dilemma, the nal decision should comprise the whole range of problem aspects. Although there are studies on the role of tutoring in problem solving settings that deal with reaching particular solutions, the role of scaffolding tactics in more open situations dealing with dilemmas is still unknown. The objective of the current study was to investigate the role of tutor scaffolding during dilemma-solving activity in a synchronous chat room. The relationships between tutors scaffolding discourse acts and students scaffolding and task-related discourse acts in different discussion groups were examined. The inuence of the tutors scaffolding style on the students task-related discourse acts, the group-negotiation processes and their nal decisions were analysed. The following research questions were addressed: (i) How are the tutors and students separate discourse acts related to each other? (ii) What are the characteristics of different general level scaffolding styles used by tutors during decision-making role-play in a synchronous environment? (iii) What kind of relationships can be found between the tutors scaffolding styles and the students discourse acts? (iv) What kind of relationships can be found between the tutors scaffolding styles and the students group decision-making characteristics?

K. Pata et al. / Learning and Instruction 15 (2005) 571e587

575

2. Method 2.1. Participants The study focused on the data of 31 decision-making discussions from 20 role groups and 11 expert groups collected during the two phases of the role-play. These data were saved by the system during ve role-plays carried out in a synchronous network-based environment. The participants were 62 14-year-old to 17-year-old students from ve Estonian secondary schools. In each role-play, 10e12 pupils participated. The number of persons in the role groups ranged from 2 to 5, in expert groups from 4 to 7. The students participated in the activity during science lessons in the school computer labs. Science teachers provided the students with technical assistance. Two online tutors, a male and a female teacher, with a similar educational background in teaching environmental subjects and experience in online tutoring, supported the activity. The male tutor scaffolded 6 role groups and 3 expert groups, while the female tutor supported 14 role groups and 8 expert groups. The tutors were in a separate location from the students.

2.2. Procedure The experimental data were recorded during the jigsaw type of environmental role-plays on the theme Stray dogs (Pata & Sarapuu, 2003). The activity comprised the following steps: making individual decisions, discussion and negotiations in different role groups, and, nally, decision-making discussion in the expert groups. The total length of each role-play varied from 45 to 60 min. A java-based synchronous learning environment Collaborative Virtual Workplace 4.0 (CVW) (http://cvw.mitre.org) offered a graphical interface with several virtual rooms and oors that could be used for working in separate teams. The rooms were named according to the role groups (Dog-owners Club, Stray Dogs Asylum, etc.), and the user names for participants were chosen according to their future roles in the game (citizen1, animal protector2, etc.). This created an anonymous atmosphere simulating the real situation aimed at supporting decision-making and creativity in virtual reality (see Connolly, Jessup, & Valarich, 1990; Ingram, Hathorn, & Evans, 2000). The role and expert discussions ended with the drawing up of decision documents. Students could use their individual- and role-decision documents (Notes) while moving from one virtual room to another, and could also share these with team members. The rooms were equipped with discussion recorders. It was possible to provide learners with adequate instructional support by establishing appropriate links to the web pages in the role-group rooms. Each role-group room was provided with the documents of Problem introduction, Role description and Additional material. During all the steps of the decision-making process, verbal tutor scaffolding directed towards the initiation of peer scaffolding was offered. The tutors had to use a scaffolding script with an agenda and scaffolding prompts, but they could also modify it exibly if needed. They were instructed to act as instructors and not as equal members of the decision-making teams and to avoid certain interactions like suggesting his or her personal viewpoints and decisions. The forms of tutor scaffolding were: (i) process scaffolding, for giving technical help and guiding students through the different stages of the activity, as well as keeping them on-task; (ii) content scaffolding aimed at facilitating brainstorming of different solutions with question prompts and suggesting ways for evaluating the decisions; and (iii) scaffolding for collaboration, maintaining a social

576

K. Pata et al. / Learning and Instruction 15 (2005) 571e587

atmosphere. The tutors were instructed to exibly regulate the extent of scaffolding utterances in teams depending on the task-related activeness of students. 2.3. Data analysis Discussions that took place during the role-plays were saved in the system in written format and used as the corpora for discourse analysis. On the basis of the categorization schemes of Baker, Quignard, Lund, and Sejourne (2003), Bales (1999), and Chi et al. (2001), a common discourse act category system was elaborated to describe the tutors and the students communicative acts on the same bases (see Tables 1, 2). The tutors and the peers supportive discourse acts were classied into process and content scaffolding. In addition, the feedback/help acts, decisions, arguments and content-related conrmations from the task-focused area, and the social level discourse acts from the non-task-focused area were extracted. An emotional subscale was applied to distinguish agreements (agr) and disagreements (dis) in conrmations, and satisfaction (sat) or anticipation (ant) in social level discourse acts. All the discourse acts from the decision-making discussions of the activity were classied according to these categories by the rst author of this paper. To test the reliability of the categorization, another researcher independently categorised the data within a six-group interaction sequence (N Z 404 acts), and the inter-rater agreement was measured. The k-value of 0.77 was considered satisfactory. In order to analyse whether common ground (Gr) was reached in the role and expert groups, students content-related statements during the discussion were extracted from it and used as
Table 1 Tutors discourse act categories Category Instruction e TI Description Tutors functional and process-related information, instructions, suggestions, directions, hints for orientation Tutors questions asking for process-related feedback, information, conrmation about what students are doing Tutors task-related judgements, opinions, decisions Tutors questions asking for task-related opinions, analysis, evaluation Tutors task-related conrmations of opinions, showing agreement or disagreement Tutors talk related to social atmosphere, showing content or discontent, satisfaction, anticipation Scaffolding type Scaffolding the process Example Please consider the team-members opinions when making the decision! Its time to make the decision document visible to everybody! Are you ready to move to the next room? How far are you with composing the joint decision document? We can regulate the functioning of the law. What is bothering you most about neglected dogs? Is it possible to implement decisions like these in reality? Yes, its a good idea! No, its impossible!

Feedback e TF

Scaffolding the process

Judgement e TJ Evaluation e TE

Scaffolding the content Scaffolding the content

Conrmation e TC (agr, dis)

Scaffolding the content

Social talk e TS (sat, ant)

Scaffolding the collaboration

Hello! It is nice to work together.

K. Pata et al. / Learning and Instruction 15 (2005) 571e587 Table 2 Students discourse act categories Category Instruction e SI Description Students functional and process-related instructions, directions Students questions asking for process-related help, directions, orientation, suggestions, possible ways of action Students task-related judgements, opinions, decisions Students task-related arguments, clarications, conrmations Students questions asking for task-related opinions, evaluation, analysis Students task-related conrmations, showing agreement or disagreement Students social-related talk, showing content or discontent, satisfaction or anticipation Objectives Scaffolding the process Asking for help in order to carry on the activity Performing the task Performing the task Scaffolding the content Performing the task Maintaining the social atmosphere Example

577

Help e SH

Write down your opinions! We must make decisions! Make the decision visible! What is going on? Where and how should I write? Tell me what you are discussing? The dogs should wear leashes! We must earn money! But it is easier to recognise the lost dogs this way! Who will pay these people? Are there any better ideas? Yes I agree with you! I disagree with the decision! Hello! Its nice to be here! Dont you have anything sensible to say! Nonsense!

Judgement e SJ Argument e SA Evaluation e SE

Conrmation e SC (agr, dis) Social talk e SS (sat, ant)

indicators of their problem representations. The coherency of the team members problem representation was measured on the scale of: 1 e No grounding; common ground was not reached, the students did not form external problem representations, or the students nal problem representations were not coherent. 2 e Common ground was reached in the group, the students nal problem representations were coherent. To evaluate how the groups took mutual ownership (Wenger, 1998) over shared problem representations (Ow), the nal group-decision documents were compared with the individual external problem representations, as well as with the discussions. All the discussions of role and expert groups were evaluated according to the following scale: 1 e No ownership or individual ownership; the nal decision was not made or few negotiations were carried out, the teams decision was derived from a single individual external problem representation or from a single role-decision document. 2 e Mutually accepted ownership; few negotiations were carried out, the group members individual problem representations or role-decision documents were automatically summed up into the nal decision. 3 e Mutually generated ownership or mutually generated and evaluated ownership, the nal decision document was composed as a result of the group discussion, or the nal initialdecision document was composed during the group discussion; it was then evaluated, and modied into the second decision document.

578

K. Pata et al. / Learning and Instruction 15 (2005) 571e587

For an evaluation of the competency of the role and expert decisions (Ev), the model of ideal environmental problem representation was developed. According to this model, solving the dilemma requires consideration of both the strategies used to prevent the causes of the problem, and elimination of its consequences, in terms of ecological, economic, ethical and legislative aspects. It was predicted that in competent role group decisions, mainly one aspect of the problem would be emphasized in terms of removing possible hazards and dealing with dangerous consequences, but that, in the competent expert group judgements, the complete ideal decision model would have to be achieved. This presumption formed the basis for the evaluation scale of environmental decisions: 1 e No decision documents were made, or the important aspects of the particular role were not considered in the decision document. 2 e Important aspects of the particular role were considered in the decision document but only from the point of view of preventing the causes or eliminating the consequences of the problem. 3 e All the important aspects were considered in the decision document, from the point of view of both preventing the causes of the problem and eliminating the consequences of the problem. Therefore, each group was characterised by the frequency of each discourse act category, the measure of common ground in groups (Gr), the form of ownership over the problem representations (Ow), and the competency of decisions (Ev). In order to measure the reliability of the ratings, two independent raters carried out the categorization of decision variables in all the role-plays. The values of Cohen k emerged as kGr Z 0.78 (Gr), kOw Z 0.95 (Ow), kEv Z 0.70 (Ev), and these were considered satisfactory. The statistical analysis was carried out with SPSS 0.11.0. In order to characterise the tutor and peer scaffolding and the students decision-making during the role-play, frequencies for each discourse act category were counted for the 31 groups. The means of the frequencies of the different tutors and students discourse acts during the discussion phases of the role and those of the expert groups were compared with independent samples t-test. The relationships between the tutors and students scaffolding acts, as well as their relationships with the rest of the students discourse acts were analyzed by Spearman correlation analysis using the data on the frequencies of discourse acts in each group. The tutors discourse acts which were not used or that occurred with low frequency were excluded from the further analysis. In order to explore the groups of inter-related discourse act variables, the factor analysis was performed as a principal component analysis with nine discourse acts related to tutors and students scaffolding and students task performance. The KMO and the Bartlett test were used to test the sampling adequacy and the equal variance across samples. The solutions were evaluated using eigenvalues, screen plots and the amount of variance. Because this paper focuses on the inuence of scaffolding, only scaffolding-related acts and the students task-related acts were included in the factor analysis. To characterize the tutoring behaviour, K-Means clustering with all the tutors discourse acts was performed, searching for two to four clusters. The best tting cluster model was selected by taking into account the greatest distance between the nal cluster centres. ANOVA with discourse acts as factors made it possible to distinguish the tutors acts that differed signicantly in their frequency in two clusters. To nd out which was the behaviour of each of two tutors, the clusters were also related to the tutor information. The differences in students discourse acts in groups with different scaffolding were analysed

K. Pata et al. / Learning and Instruction 15 (2005) 571e587

579

by independent samples t-test. To nd out how tutoring styles were related to the groups decision-making variables of common ground (Gr), form of ownership over problem representations (Ow), and competency of decisions (Ev), cross-tabulation and Chi-square analysis were performed. 3. Results 3.1. The relationship between tutors scaffolding acts and students discourse acts On the basis of discourse act frequencies in the teams, it was found that, during the discussions, all the different students discourse act types were used, and the students performed both decision-making and scaffolding, whereas the tutors activity was mainly oriented towards process and content scaffolding instructions, and not towards conrmations or social level discourse acts within the team (see Table 3). It was expected that the nature of discussions in the role and expert groups might differ, as in role groups the students might be mainly brainstorming to generate various decisions, whereas in expert groups they have to concentrate more on defending earlier role-group judgements or evaluating the decisions of other groups. This supposition about the difference in frequencies of certain discourse acts was not supported by the independent samples t-test: the means of tutors and students discourse act frequencies during the discussion phases of the role and expert groups did not differ signicantly. This enabled us to perform the following analyses without separating the data from the role and expert groups. The Spearman correlations between the frequencies of tutors and peers scaffolding and students decision-making discourse acts (see Table 4) indicated that tutors instructions (TI) and evaluative prompts (TE) and peers evaluative prompts (SE) were highly signicantly ( p ! 0.001) related, while the peers instructions (SI) were signicantly ( p ! 0.01) related to students judgements (SJ). The tutors instructions (TI) were highly signicantly ( p ! 0.001) related to the students arguments (SA), whereas the other scaffolding discourse acts by the tutor and the peers (TE, SI, SE) had a signicant ( p ! 0.01) relationship with arguments. All the tutors and peers scaffolding acts were highly signicantly ( p ! 0.001) related to students negative content-related conrmations (SCdis) in teams. There was a signicant ( p ! 0.01) relationship between peers instructions and both the tutors and peers evaluative prompts, and the content-related agreements (SCagr). It was found that tutors scaffolding instructions (TI, TE) were not signicantly related to the occurrence of students process- (SI) and task-related (SE) scaffolding discourse acts in teams. The tutors content scaffolding prompts (TE) were signicantly ( p ! 0.01) related to students positive (SSsat) but not to students negative social talk (SSant) in teams. The peers process and content scaffolding events appeared to be highly signicantly ( p ! 0.001) related to the teams negative social atmosphere (SSant). A highly signicant ( p ! 0.001) relationship was found between students
Table 3 The mean frequencies of tutors and students discourse acts in 31 teams Mean frequency of discourse acts in groups TI/SI Tutor Students 9.16 2.3 TE/SE 2.45 2.9 TF/SH 1.45 4.0 TJ/SJ 0.13 15.7 e/SA 5.77 e TCagr/SCagr 0.29 7.3 TCdis/SCdis 0.06 2.5 TSsat/SSsat 0.45 3.5 TSant/SSant e 2.0

580

K. Pata et al. / Learning and Instruction 15 (2005) 571e587

Table 4 Tutor and peer scaffolding discourse acts inter-correlations with the other discourse acts in teams (N Z 31) TE TI TE SI SE 0.502** SI 0.252 0.216 SE 0.326 0.199 0.443* SJ 0.591** 0.519** 0.393* 0.586** SA 0.520** 0.387* 0.386* 0.423* SCagr 0.064 0.358* 0.414* 0.375* SCdis 0.462** 0.529** 0.506** 0.515** SH 0.378* 0.307 0.535** 0.446* SSsat 0.287 0.400* 0.281 0.241 SSant 0.090 0.176 0.465** 0.464**

Note. Sig. (1-tailed) *p ! 0.01, **p ! 0.001.

help questions and peers process scaffolding (SI) acts. There was also a signicant relationship ( p ! 0.01) between students help questions (SH) and tutors process scaffolding (TI). It was supposed that different discourse acts of the tutor and the students form an inter-related system during the discussions. Thus, the principal component factor analysis (see Table 5) was carried out with both tutors and students scaffolding- and task-related discourse act frequencies in teams. Although the number of cases (N Z 31) compared with tested discourse acts (N Z 9) was small, the results of the KMO test Z 0.768 for sampling adequacy and Bartletts test for equality of variances c2(36) Z 146.26, p ! 0.001 indicated that the principal component factor analysis was applicable. The analysis differentiated the discourse acts into two factors. It appeared that the tutor scaffolding acts, instructions (TI) and content-related prompts (TE), were related to the students judgements (SJ), arguments (SA) and content-related agreements and disagreements (SCagr/dis) forming the Tutor scaffolded decision-making factor (51% of common variance, eigenvalue Z 4.63). The second component, Peer scaffolded help-requesting factor (13% of common variance, eigenvalue Z 1.19), combined the discourse acts of peer scaffolding (SI, SE), students help questions (SH) and their content-related disagreements (SCdis). The two factors accounted for 64% of the total variance of discourse acts. 3.2. The characteristics of tutor scaffolding styles The K-Means clustering (see Fig. 1), performed with all the tutors discourse acts showed that the two different tutoring styles, passive and active tutoring, were used with different
Table 5 Rotated component matrix of the factors relating the tutors and students scaffolding and students task-related discourse acts in teams (N Z 31) Discourse acts Component Tutor scaffolded decision-making factor SJ TE TI SCagr SA SH SI SCdis SE 0.815 0.782 0.737 0.735 0.675 0.870 0.673 0.663 0.589 Peer scaffolded help-requesting factor

0.621

Note. KMO Z 0.768 and Bartlett test c2(36) Z 146.29, p ! 0.001. Tutor scaffolded decision-making factor: eigenvalue Z 4.63, 51% common variance; Peer scaffolded help-requesting factor: eigenvalue Z 0.19, 13% common variance. Two factor solution Z 64% total variance.

K. Pata et al. / Learning and Instruction 15 (2005) 571e587

581

Tutoring styles
16 14 12 10 8 6 4 2 0 TSsat TCagr TI TJ TF TE TCdis TSant

means

interactions

passive active

Fig. 1. The nal centres of variables of the two tutoring style clusters. Discourse acts TI and TE differ in tutoring clusters at signicance level p ! 0.001.

groups. The tutoring styles differed from each other highly signicantly ( p ! 0.001) in their frequency of the usage of process (TI) (F(1,18) Z 44.98) and content (TE) scaffolding (F(1,18) Z 31.96) acts. In the passive scaffolding cluster, process (TI) and content (TE) scaffolding was performed less frequently than in the active scaffolding cluster. The passive tutoring style appeared in 22 groups, all guided by Tutor 1, whereas the active style appeared in nine groups and characterised the behaviour of Tutor 2. The differentiation of tutoring styles by K-Means clustering was based on the discourse act frequencies in different groups that did not reveal the actual sequence of communicative turns. Therefore, the discourse patterns were studied to get a more detailed characterization of the two tutors behaviour. Although an in-depth discourse analysis was not performed within the framework of the current study, some tendencies common to the interaction of tutors using different styles with the students could be described. The following Example 1 describes how the passive tutor interacted with the students. In many cases, the passive tutor took little initiative in the discussions and expected that the students would generate self-regulation patterns by themselves after the tutors preliminary instructions (TI). The students in the teams guided by the passive tutor interacted mainly with their peers, often addressing each other by user names (recorded in brackets). The discussions in the passive tutors groups varied in length and intensity. Although it seemed to be effective in some cases for the tutor to wait for the students to take the initiative, this scaffolding strategy also resulted in students minimal interaction in teams and low task-related performance. Example 1. The passive scaffolding style (translated from Estonian): Tutor: You as animal protectors must decide what your suggestions are to improve the situation of neglected dogs. TI Protector6: Hmmm! SSsat Protector6: Animal protectors, send your opinions! SI Protector6: Send them to me! SI Protector7 [to Protector6]: We need to produce adverts so that each dog could nd a caring owner! SJ Protector4: I dont believe that all the dogs will nd one. SCdis Protector6: Protector4, do the dogs need to nd the owners by themselves? SE Tutor: Can everybody have a dog? TE

582

K. Pata et al. / Learning and Instruction 15 (2005) 571e587

Protector6: Make a new suggestion! SI Protector6: Only people who can take care of them, I suppose. SJ Protector6: It means feeding them properly, walking with them. SJ Protector7 [to Protector6]: Not everyone can take care of animals, often the kids wishes are fullled, but later they dont know how to take care of the dog and they let them wander around all alone! SA The active scaffolding style could be characterised by Example 2. It was common in many teams that the tutors scaffolding prompts were responded to by students judgements and arguments directed mainly towards the tutor. It was noticeable that by active usage of content scaffolding prompts the tutor was able to maintain the initiative in teams. The students were forced to follow the discussion path selected by the tutor. This type of active facilitation and coconstruction seemed to be chosen because of the application of the synchronous distant learning environment. By being active, the tutor could always control the students task-related performance. Nevertheless, the lack of fading out of tutors assistance might also have had a negative inuence on students self-initiative in interpreting their decision-making task. Example 2. The active scaffolding style (translated from Estonian): Tutor: You as animal protectors, are you satised with the situation that a lot of hungry neglected dogs are walking around in the city? TE Protector1: No, we dont like it, especially that dogs are taken home and then abandoned. SCdis, SJ Tutor: Can the animal protectors inuence the owners in any ways? TE Protector1: Actually, yes, we can, but the owners will not listen to us anyway. SCagr, SJ, SA Protector1: We may say that if animals are treated badly, punishment will follow. SJ Tutor: What to do then e there must be some ways to inuence? TE Protector3: Of course there are e rstly, we should affect their conscience. SCagr, SJ Tutor: How? TE Protector1: This may not work, they dont care, we should distribute informative leaets. SA, SJ Protector3: We should show what the actual situation in dogs asylums is. SJ Tutor: Yes, thats a good idea! TCagr

3.3. The relationship of the tutor scaffolding style and students decision-making discourse acts In order to clarify whether there was a relationship between tutors scaffolding styles and the students discourse acts, the independent samples t-test was carried out (see Table 6). The signicant differences between tutoring clusters appeared to be in the frequency of students judgements (SJ) ( p ! 0.05) and arguments (SA) ( p ! 0.01), and almost signicant ( p ! 0.1) differences were found between the scaffolding style and students task-related disagreements (SCdis) and social-related talk showing satisfaction (SSsat). The active tutoring style of Tutor 2 with a higher rate of process (TI) and content (TE) scaffolding was related to the more actively performing groups. In comparison, the passive tutoring style of Tutor 1, where less scaffolding was used, resulted in fewer judgements, arguments and task-related conrmations being made by students in the teams.

K. Pata et al. / Learning and Instruction 15 (2005) 571e587 Table 6 The t-test values of the tutor scaffolding clusters in groups (N Z 31) Discourse acts SI SH SJ SA SE SCagr SCdis SSsat SSant Tutoring-cluster passive active passive active passive active passive active passive active passive active passive active passive active passive active N 22 9 22 9 22 9 22 9 22 9 22 9 22 9 22 9 22 9 mean 1.81 3.44 3.72 4.77 9.63 30.44 4.00 10.11 2.40 4.11 5.77 11.00 1.54 4.77 2.77 5.11 2.27 1.44 df 29 9.68 29 28.21 29 10.96 29 11.85 29 14.95 29 8.88 29 10.01 29 11.29 29 24.86 t-value 1.340 0.529 3.067 4.093 1.404 1.252 2.000 1.893 0.444

583

Sig. (2-tailed) 0.211 0.601 0.012* 0.003* 0.171 0.243 0.073 0.068 0.660

Note. Sig. (2-tailed) *p ! 0.05.

3.4. The relationship of the tutor scaffolding style and the students group decision-making variables In order to investigate the relationship between tutoring styles and the groups decision-making variables of grounding (Gr), ownership of shared problem representations (Ow), and quality of decisions (Ev), cross-tabulation and Chi-square analysis were performed. No signicant relationship appeared with the variables of common ground or the quality of decisions in teams, but there was a signicant relationship between the tutoring style and the form of ownership of problem representations (c2(2) Z 6.032, p ! 0.05). The active tutoring style resulted more often in reaching mutually generated ownership. Ownership of problem representations appeared to signicantly (c2(4) Z 9.814, p ! 0.05) affect the quality of nal team decisions (Ev): if no ownership or individual ownership was reached, the teams decisions appeared to be incompetent. Mutually generated ownership, on the other hand, helped the teams to consider all the important aspects in their decision documents. 4. Discussion Our analysis was based on the supposition that the tutors and the students discourse acts can be described concurrently within the framework of scaffolding and decision-making, using similar discourse act categories for the tutor and the students (see Baker et al., 2003; Bales, 1999). This made it possible to nd the peer scaffolding acts that inuenced the learning activity, in addition to the tutors scaffolding, and to give a more complete overview of the collaborative decision-making situation. The data analysis was based on the frequencies of discourse acts that enabled us to relate different discourse act types in a meaningful way, but did not permit us to determine their causal relationships.

584

K. Pata et al. / Learning and Instruction 15 (2005) 571e587

The correlation analysis between the tutors and peers discourse acts made it possible to relate the usage of certain scaffolding acts to the task- or social-related acts. Tutors scaffolding instructions were positively related to students help questions. A similar, but stronger relationship was found between peers process scaffolding instructions and students need for support. In addition, the peers content scaffolding prompts were signicantly related to the students discourse acts asking for help. Thus, peer scaffolding acts tended to be more related to students help requests than tutor scaffolding acts. This indicated that the tutors and peer-tutors functions in teams might differ. On the other hand, both the tutors and the peers scaffolding acts appeared to be signicantly related to the occurrence of students judgements, arguments and content-related feedback interactions. Thus, it can be assumed that not only the tutors but also the peers scaffolding could play an important role in teams decision-making. The factor analysis was performed in addition to the correlation analysis because it was assumed that communication is a complex system with certain sets of discourse acts favouring the occurrence of mutually related discourse acts. The ndings about the inter-related sets of variables, which could be named as the Tutor scaffolded decision-making factor and the Peer scaffolded help-requesting factor, indicated different applications of tutors and peers scaffolding acts when supporting the decision-making. The students decision-making functions were related more to the tutors scaffolding instructions and prompts, whereas the peer scaffolding acts appeared more frequently in teams with a higher rate of uncertainty about the task (students frequent help questions) and disagreements about each others arguments and judgements. These ndings supported the results of the correlation analysis about the different roles of the tutors and the students supportive acts in the decision-making process, but also indicated the possibility that, in collaborative situations, different scaffolding agents (tutor, students) might be performing scaffolding partly independently. For instance, they may tend to direct their scaffolding acts towards different aims, not supported by the other. The two factors can also be explained by the existence of two types of scaffolding patterns in teams. The peers might be less active in using instructions and prompts in the tutor scaffolded pattern of decision-making because the tutor is taking the initiative. When the tutor is actively fullling the support function, the peers self-regulative acts might be suppressed. In the peer-scaffolding pattern, on the other hand, the students may be trying to compensate for the missing tutors scaffolding initiative by their regulative acts. These preliminary results can serve as the motivation and starting point for future studies about the relationships and functions of tutors and peers scaffolding acts in teams. It could be expected that the tutoring style is dependent on students discourse acts and needs in teams. Nevertheless, different studies have reported that tutors apply person-specic tutoring styles (Hakkarainen et al., 2002; Pinto et al., 2001; Rasku-Puttonen et al., 2002, 2003). The ndings of the current research support the notion of person-related scaffolding styles. In spite of the application of exible tutoring scripts that should have resulted in the two tutors practising various scaffolding techniques with the students from different teams, it appeared that both of them had used more or less consistent styles in supporting the different groups. The passive tutorial behaviour consisted of less frequent process and content scaffolding acts with the tutor not in the dominating role. In the case of active scaffolding, the tutor used frequent scaffolding acts and led the decision-making process by keeping the initiative. Chi (1996) has distinguished two types of tutor actions: (i) prompting can be generated without knowing the content domain and it facilitates students self-construction, and (ii) scaffolding that includes some aspects of the content and is a joint activity, where the tutor co-constructs knowledge together with the students. The passive style of supporting the decision-making found in the current study appeared to resemble more the tutor behaviour dened by

K. Pata et al. / Learning and Instruction 15 (2005) 571e587

585

Chi (1996) as prompting for self-construction. In many cases, the students of the passive tutor were expected to initiate both the task-regulation and the knowledge-generation processes by themselves as a result of the tutors scaffolding. The active scaffolding style in this study could be related to the tutoring behaviour which Chi (1996) calls scaffolding in her study. The students were interacting mainly with the tutor but not with their peers, and the co-construction of knowledge occurred between the tutor and the students. The design of this study did not permit controlling the causal relations between the tutors scaffolding style and certain students discourse acts. The effect of different tutoring behaviours on students interactions was described in qualitative terms. In the active style tutoring groups, the students applied more active decision-making behaviour. It appeared that active tutoring was signicantly related to the greater activeness in students teams in producing more task-related discourse acts, judgements, arguments and disagreeing conrmations, whereas the passive style less frequently produced such acts among students. These data tend to support the results of VanLehn et al. (2003). According to them, the key to improved student learning is not the specic discourse acts initiated by the tutor but the mutual interaction process between the tutor and students, where the tutors prompts make students think harder and use the knowledge that they already possess. Frequent turn taking, characteristic of the active style, has emerged as the one of the indicators of effective learning and co-construction of the tutor and the students on the problem-solving task in many studies (Chi, 1996; Rasku-Puttonen et al., 2003; VanLehn et al., 2003). The more active tutoring style appeared to be related to the generation of mutually accepted ownership of problem representations in teams; joint decisions were made by summing up all the individuals decisions, or they were constructed in the discussions based on common ground employing the full ownership of team-members viewpoints. This also indicated the co-construction of knowledge analogous to the tutors actions dened by Chi (1996) as scaffolding. On the basis of their experiments in problem-solving settings, VanLehn et al. (2003) have suggested that the key to effective tutoring might not depend on the content of tutorial explanations but on the tutorial behaviour that makes students think, such as generating the opportunities for impasses or giving zero-content prompts. This study was based on analyzing the frequencies of tutors and students discourse acts. Neither the causal relations between the discourse acts of the tutor and the students nor the different interaction patterns were investigated within the frames of the study. Nevertheless, several tendencies as to how different discourse acts and interaction styles might be related were found. The tutoring behaviour in synchronous collaborative learning environments is a very complex issue, and different tutoring styles might occur in response to students interactions. It is obvious that these tutoring styles have different effects on students learning. Therefore, more systematic knowledge of the effects of tutoring styles is needed. Further experimental studies and in-depth pattern analysis should be carried out in order to nd the causal relationships between the tutors and the students discourse acts.

Acknowledgements This research was supported by grant no. 5996 from the Estonian Science Foundation and MER funding 0182542s03. We express our appreciation for the co-operation and assistance of the Estonian science teachers from the in-service course of Computers in Science Education 2000e2001.

586

K. Pata et al. / Learning and Instruction 15 (2005) 571e587

References
Baker, M. J., Quignard, M., Lund, K., & Sejourne, A. (2003). Computer-supported collaborative learning in the space of debate. In B. Wasson, S. Ludvigsen, & U. Hoppe (Eds.), Designing for change in networked learning environments. Proceedings of the CSCL 2003 (pp. 11e20). Dordrecht: Kluwer Academic Publishers. Bales, R. F. (1999). Social interaction systems. New Brunswick, NJ: Transaction Publishers. Berge, Z. L., & Collins, M. P. (Eds.). (1995). Distance learning. Computer mediated communication and the online classroom, Vol. III. Cresskill, NJ: Hampton Press. Brown, A. L., & Campione, J. C. (1996). Psychological theory and the design of innovative learning environments: on procedures, principles and systems. In L. Schauble, & R. Glaser (Eds.), Innovations in learning: New environments for education (pp. 289e325). Mahwah, NJ: Erlbaum. Brown, A. L., & Palinscar, A. (1989). Guided, cooperative learning and individualized knowledge acquisition. In L. Resnick (Ed.), Knowing, learning, and instruction (pp. 393e451). Hillsdale, NJ: Lawrence Erlbaum Associates. Chi, M. T. H. (1996). Constructing self-explanation and scaffolding explanations in tutoring. Applied Cognitive Psychology, 10, 33e49. Chi, M. T. H., Siler, S. A., Jeong, H., Yamauchi, T., & Hausmann, R. (2001). Learning from human tutoring. Cognitive Science, 25, 471e533. Clark, H. H., & Schaefer, E. F. (1989). Contributing to discourse. Cognitive Science, 13, 259e294. Collis, B., Andernach, T., & VanDiepen, N. (1997). Web environments for group-based project work in higher education. International Journal of Educational Telecommunications, 3(2/3), 109e130. Connolly, T., Jessup, L. M., & Valarich, J. S. (1990). Effects of anonymity and evaluate tool on idea generation in computer-mediated groups. Management Science, 36, 6. Dolmans, D. H., Wolfhagen, I. H., & Snellen-Ballendong, H. A. (1994). Improving the effectiveness of tutors in problem-based learning. Medical Teacher, 16(4), 368e377. Graesser, A. C., Person, N. K., & Magliano, J. P. (1995). Collaborative dialogue patterns in naturalistic one-on-one tutoring. Applied Cognitive Psychology, 9, 495e522. Guzdial, M., Carlson, D., & Turns, J. (1995). Facilitating learning design with software-realized scaffolding for collaboration. In D. Budny (Ed.), Frontiers in education 1995. (Atlanta, GA). Hakkarainen, K. & Lipponen, L. (1998). Epistemology of inquiry and computer supported collaborative learning. A paper presented at the American Educational Research Association (AERA) Annual Meeting. San Diego, April 13e17, 1998. Hakkarainen, K., Lipponen, L., & Jarvela, S. (2002). Epistemology of inquiry and computer-supported collaborative learning. In T. Koschman, R. Hall, & N. Miake (Eds.), CSCL 2: Carrying forward the conversation (pp. 129e 155). Mahwah, NJ: Lawrence Erlbaum Associates. Ingram, A. L., Hathorn, L. G., & Evans, A. D. (2000). Beyond chat on the Internet. Computers & Education, 35(1), 21e35. Jarvela, S. (1995). The cognitive apprenticeship model in a technologically rich learning environment: interpreting the learning interaction. Learning and Instruction, 5, 237e259. Lehtinen, E. (2003). Computer-supported collaborative learning: an approach to powerful learning environments. In E. De Corte, L. Verschaffel, N. Entwistle, & J. Van Merrieboer (Eds.), Unravelling basic componets and dimensions of powerful learning environments (pp. 35e53). Amsterdam: Elsevier. Liebrand, W. B. G., & Messick, D. M. (1996). Frontiers in social dilemmas research. Berlin: Springer Verlag. Newman, D. R. (1996). How can WWW-based groupware better support critical thinking in CSCL? In U. Busbach, D. Kerr, & K. Sikkel (Eds.), CSCW and the web e proceedings of the fth ERCIM/W4G workshop. Arbeitspapiere der GMD 984, Sankt Augustin: GMD. Available from. http://orgwis.t.fraunhofer.de/projects/W4G/proceedings/ wwwcscl.html. Accessed April 1996. Pata, K., & Sarapuu, T. (2003). Developing students mental models of environmental problems by decision-making role-play in synchronous network-based environment. In J. Lewis, A. Magro, & L. Simonneaux (Eds.), Biology education for the real world. Proceedings of the IV ERIDOB Conference (pp. 335e348), (Paragraphic/Groupe Lienhart). Pinto, R. P., Rendas, A., & Gamboa, T. (2001). Tutors performance evaluation: a feedback tool for the PBL learning process. Medical Teacher, 23(3), 289e294. Pressley, M., Hogan, K., Wharton-McDonald, R., Mistretta, J., & Ettenberger, S. (1996). The challenges of instruction that supports student thinking. Learning Disabilities Research and Practice, 11(3), 138e146.

K. Pata et al. / Learning and Instruction 15 (2005) 571e587

587

Rasku-Puttonen, H., Etelapelto, A., Arvaja, M., & Hakkinen, P. (2003). Is successful scaffolding an illusion? Shifting patterns of responsibility and control in teacher-student interaction during a long-term learning project. Instructional Science, 31, 337e393. Rasku-Puttonen, H., Etelapelto, A., Hakkinen, P., & Arvaja, M. (2002). Teachers instructional scaffolding in an inno vative ICT-based history-learning environment. Journal of Teacher Development, 6, 269e288. Tharp, R. G., & Gallimore, R. (1988). Rousing minds to life. Cambridge: Cambridge University Press. VanLehn, K., Siler, S., Murray, C., Yamauchi, T., & Baggett, W. B. (2003). Human tutoring: why do only some events cause learning? Cognition and Instruction, 21, 209e249. Vygotsky, L. S. (1978). Mind in society: The development of higher psychological processes. Cambridge: Harvard University Press. Wenger, E. (1998). Communities of practice: Learning, meaning, and identity. Cambridge, UK/New York, NY: Cambridge University Press. Wood, D. J., Bruner, J. S., & Ross, G. (1976). The role of tutoring in problem solving. Journal of Child Psychology and Psychiatry, 17, 89e100. Wu, A. S., Farrell, R., & Singley, M. K. (2002). Scaffolding group learning in a collaborative network environment. In: Conference proceedings of CSCL 2002 (pp. 245e253), (Boulder, Colorado). Zandbergen, P., & Petersen, F. (1995). The role of scientic information in policy and decision-making. The Lower Fraser Basin in transition: A symposium and workshop. Surrey, BC, Canada: Kwantlen College.

You might also like