Peer Feedback in Synchronous Writing Environments: A Case Study in French
The advent of computer networks has changed the role of the computer in the classroom by enabling collaborative learning strategies. The computer is no longer necessarily an agent that emulates a student-teacher interaction, as commonly found in drill-and-practice programs, but has become a medium that restructures the interaction among participants (Kern, 1996).
Computer-Mediated Communication (CMC) presents a number of advantages over traditional writing environments. For example, Eisenberg & Ely (1993:2) state that the “interaction through networks helps break down communication barriers and inhibitions that often stifle the open exchange of ideas in traditional classrooms”. A study by Kelm (1996), for instance, has shown that in a second language classroom learners are often inhibited from expressing their ideas in the second language. In a computer-networked writing environment, students can participate anonymously in on-line discussions and, to some extent, overcome inhibitions encountered in face-to-face settings.
Networked writing environments also expand student opportunities for social interaction (Harasim et al., 1997). According to Vygotsky’s theory of zone of proximal development, “learners benefit most from social interactions concerning tasks that they cannot do alone but can do in collaboration with more knowledgeable or more experienced individuals” (Vygotsky 1978:86). A second language classroom is a prime example of students with diverse knowledge backgrounds and here CMC can foster collaboration among learners, in particular, among those who are frequently excluded from discussions (Warschauer et. al., 1996; Wang & Hurst, 1997).
Both quantitative and qualitative studies of synchronous writing environments in the foreign language classroom have shown the positive impact of CMC (see also Warschauer, 1997). For instance, an increase in student participation has been found by Chun (1994), Kelm (1992, 1996), and Kern (1996). In addition to quantitative studies of synchronous writing environments, Felix & Lawson (1996), for example, found that students scored higher on the logical linking of ideas when using networked writing environments as opposed to face-to-face instruction. Warschauer’s (1996a) composition class used lexically and syntactically more complex structures and a study by Kelm (1992) suggests advantages to using CMC for developing linguistic accuracy.
The case study presented in this paper investigates the interaction patterns among participants in a synchronous writing environment (see also Cooney, 1998). The focus here is on peer feedback, either in the form of social or cognitive acknowledgement. Feedback is essential to learning in any setting. In a face-to-face learning environment, it is the instructor who commonly evaluates student output, provides praise and/or disapproval, and leads discussion. In contrast, in Computer-Mediated Communication (CMC) the instructor takes on the role of a facilitator or mediator (Wang & Teles, 1998; Berge, 1995) becoming less dominant (Kelm, 1996). The question we pose is whether some students therefore take on the role of the instructor in providing feedback to their peers and leading the discussion. If so, are these students frequent contributors and/or students with a higher than average command of the second language? And finally, how do the relative language skills of each learner influence the overall group dynamic?
In the following, we will first discuss briefly the computer program Aspects, a synchronous writing environment for the Macintosh on a Local Area Network (LAN) which was used for this case study. We will then describe the participants and outline the tasks and methodology taken. Finally we will summarize the results, by providing examples of students’ output during the writing session. We will conclude with further research suggestions.
2.0 Aspects, a Computer-networked Writing Environment
Aspects is a synchronous writing environment for the Macintosh. A moderator initializes the conference and participants connected to the network share the same information on their screens. Users can be anonymous and multiple conferences can be created.
Aspects has three user modes: painting tool, chat box, and mediation. Our study employed the chat box. Participants of the conference can write at the same time and display their messages to the group as a whole. The chat box is well-suited for brainstorming on a particular topic or discussing a document in small groups.
3.0 Case Study
The case study was conducted with 12 second language students enrolled in a 300 level French composition class. One of the students was a francophone who left Québec when she was 10 years old. Her oral skills were excellent and her written skills were above average. The other students were English native speakers studying French as a foreign language. All students in the class were female.
The data were collected during the last one hour on-line session of the semester by which time, the students had had ample opportunity to familiarize themselves with the dynamics of a collaborative writing environment, in particular, with the technology and the different writing modes of Aspects. For class preparation, students had read 4 articles which covered the following topics, respectively:
The task during the on-line writing session was to discuss the points raised by each of the articles through argumentative (rather than expository) writing. The class was divided into four groups at random and each student chose an anonymous name. Each group had to discuss 3 of the 4 topics. Students decided on the order of articles and on the amount of time to spend on each topic. Figure 1 illustrates the three topics chosen by each group. As indicated, each topic was analyzed by three groups.
Table 1. Discussion Topics Chosen by each Group
3.3. Data Analysis
The data were analyzed in three steps. First, we counted the total number of contributions for each student and group during the writing session. Second, we classified student messages with respect to patterns of interaction among four main categories: 1) peer feedback (cognitive and social acknowledgement), 2) comments on content, 3) comments on task, and 4) off-topic. Third, we counted the messages the instructor sent to each group. The following unedited student writing examples illustrate the four categories:
(1) Peer Feedback
(2) Comments on Content
(3) Comments on Task
In counting messages to determine relative contribution, we had first to determine the units of analysis. For the quantitative analysis, we counted the number of words per message for each student and group. As the quantitative unit of interaction we chose the message, that is, one student posting equaled a message regardless of the number of sentences the posting contained.
In analyzing patterns of interaction, we counted the number of sentences rather than messages. This was necessary because the data contained messages with multiple sentences, some belonging to different classes of interaction. For instance, in the following example the first sentence of the message was grouped under Peer Feedback (cognitive acknowledgement), while the subsequent sentence was classified under Comments on Content.
The results of the quantitative analysis of student messages show that the students wrote an average of 55 messages during the one hour session, the lowest being group A with 29 messages, and the highest being group C with 74 messages. Kelm (1996:22) states that “it is not uncommon for our classes to write between 100-130 short messages during a local area network discussion”; however, the messages in this study were relatively long with an average of 18.3 words per message, given in Table 2.
Table 2. Quantitative Analysis for each Group
It is interesting to note that although group A had the least number of messages, the students in group A produced the highest average number of words per message. Group A contained the only francophone in the class. Her messages were very long with an average of 33.3 words per message as illustrated in Table 3 (Student A2). Her postings were also well-focused, expressing the ideas clearly and using precise vocabulary. The following message is an example of her writing:
Table 3. Student Contributions in Group A
The data in Table 3 indicate that the other, non-francophone students in group A also wrote messages which were above average in length. Whether the writing style of the francophone, elaborating on ideas and thus writing long messages, encouraged the other group members remains mere speculation at this point.
Categorizing messages with respect to patterns of interaction, the data in Table (4) show that the average number of sentences for peer feedback was 19.3%, for comments on content 71.8%, about the task 3.1%, and for off-topic5.8%. Group A was focused on the task with no off-topic comments while group B displayed a tendency toward chattiness with a higher number of off-topic comments and a correspondingly lower percentage of content.
Table 4. Interaction Patterns for each Group
With respect to peer feedback, the data in Table 4 show that all groups provided peer feedback with a higher percentage of sentences devoted to cognitive than social acknowledgement in total. However, considering the task and the given learning environment it does not seem surprising that students provided more cognitive acknowledgement. The assignment required students to use a specific genre, in this case, argumentative statements.
Group C, however, stands out with the highest percentage on peer feedback with respect to the total number of sentences (25.5%). Unlike the other three groups, these students also provided slightly more social than cognitive acknowledgement. The students in group C were also very active participants in oral class discussions. This might explain their fewer off-task messages and the fact that they wrote the highest number of total words (see Table 2). It is interesting to note, however, that all groups devoted more sentences to peer feedback than to any other category excepting content.
The data in Table (5) illustrate that students who provided the most social and cognitive acknowledgement also wrote the highest number of total messages in each group(A2, B1, C2, D1). Furthermore, these students also determined when a given topic had been exhausted and which topic of discussion would be chosen next, writing the most sentences about the task. Although comments about the task were few overall, it suggests from the concentration that these students led the discussion.
It is also interesting to note that, although the “discussion leaders”wrote the highest number of total messages, they provided an equal (groups A and B) or only slightly higher (groups C and D) number of total sentences. In addition, they wrote fewer sentences overall on content than other group participants. One possible explanation is that “discussion leaders” tend to follow the general flow of a discussion rather than solely its content. This assumption is confirmed in the data: the “discussion leaders” provided comments on all three areas: content, comments on the task, and social/cognitive acknowledgement. In contrast, other students wrote solely on content, on content and off-topic, or content and social/cognitive acknowledgement only. The “discussion leaders” also wrote the fewest off-topic messages.
There were students (A3 and D3) who did not provide any peer feedback during the group discussion. Student A3, in fact, only wrote comments on content while student D3 composed two off-topic messages in addition to comments on content. Examining the data, the results suggest that the dynamics in groups B and C differed from those of A and D. In groups B and C, although the “discussion leaders” provided the most feedback, the difference is not striking. In both groups, the three students provided feedback to each other. The explanation may lie in group diversity, as discussed below.
Table 5. Interaction Patterns for each Student
We considered the language skills of the individual group members taking into account their written essays throughout the semester and their final course grades. We found that in groups with students of varying degrees of second language knowledge, that is, groups where less knowledgeable students were coupled with more knowledgeable peers, the groups produced fewer off-topic messages. For example, group A with a francophone produced no off-topic messages at all and the instructor also participated less in the discussion in this group. Table (6) shows the number of messages sent by the instructor to each group.
Table 6. Instructor’s Comments
While our data are too limited at present to draw firm conclusions, it is plausible that students with high second language skills lead the group and very little intervention by the instructor is needed. In equally knowledgeable groups (groups B and C), peer feedback is more balanced among the group members and the instructor participates more in the discussion. Finally, while in group A the “discussion leader” had the highest language skills, this was not confirmed with the “discussion leaders” of the other three groups, suggesting that language skill alone does not determine the relative frequency of peer feedback.
In conclusion, our data suggest that students do provide cognitive and social acknowledgement to their peers in synchronous writing environments. The four groups wrote a total of 75 sentences on peer feedback compared with 254 comments on content, group C providing 37 out of the 75 sentences. All groups provided more comments on peer feedback than about the task and off-topic combined. We further found that students who provide the most social and cognitive acknowledgement are also the highest contributors, although their second language skills are not necessarily superior within their group as might be expected. Collaboration in a synchronous writing environment is also fostered when less knowledgeable students work with more knowledgeable peers — the instructor intervenes less, and students produce less off-topic discussion than in equally knowledgeable groups. Intriguing questions remain open, however, or speculative: for example, while all of our participants were female, it would be interesting to investigate to what extent gender influences the group dynamics in this particular synchronous writing environment and what role anonymity might play in such circumstances.