研究論文
The Efficiency of the Academic Writing System:
Can Prewriting Discussion be Eliminated?
2021 年 5 巻 p. 39-57
詳細
2021 年 5 巻 p. 39-57
Writing is a complex process where learners are required to demonstrate linguistic skills and knowledge by constructing clear, relevant, and logical arguments. To write an essay in a foreign language (English) is not easy for Japanese college students who have insufficient experience in academic writing, sometimes even in their native language. To help those students, a collaborative writing support system has been developed, which allows learners to read English text to analyze information and to write a well-organized academic essay. In a previous study, the effects of the system were observed in the writing performance of EFL Japanese university students; however, a problem―a large amount of time allotted for discussion―was found (Kano, 2018). Therefore, the necessity of discussion was examined utilizing the same system in this study. In order to investigate its effects, a small group discussion session was held for the students in the with-discussion group. Their writing performance was later compared to that of the without-discussion group. The results showed no significant differences between the two groups regarding word choice, grammatical correctness, logical development, and organization of their writing performance. The results suggest that providing time for discussion is not necessary when the system is implemented.
As communication in English becomes more important in society, curricula which aim to foster skills for reading and writing academic texts and daily communication in English have been introduced in colleges and universities. In higher education institutions that are diligently working on English language education, courses focusing on academic reading and writing have been created. As a result, a variety of teaching materials for academic reading and writing have been published.
Writing is a complex process with various factors influencing the quality of the writing performance. It is required for learners not only to have linguistic knowledge and skills in word choice, mechanics, grammar, and syntax, but also to be able to write relevant content with logical development and clarity. In addition, such factors as the purpose of the writing and the characteristics of the audience should be taken into account in order to deliver the message appropriately (Raimes, 1983).
Writing in the writer’s native language is not an easy task, let alone in a foreign language. In college academic writing classes, some activities such as semantic mapping, group discussion, and peer editing are practiced in classrooms to improve students’ writing performance (Zaki & Yunus, 2015). One commonly used activity is discussion, specifically prewriting discussion, which aims to help students generate ideas.
Although it has become common to use prewriting discussions in writing classes, not many research studies have been conducted to investigate its effects on writing performance. In the few studies that have been conducted, conflicting results about the effects of prewriting discussion on performance have been found. Significant effects have been reported in some research studies (Arumugam, Jambulingam, Supramaniam, & Kaur, 2018; Meyers, 1979; Mirzaei & Eslami, 2015; Sweigart, 1991; Xin & Liming, 2005), while no significant effects were found in other studies (Ling, 1998; Mazdayasna & Zaini, 2015; Ping & Maniam, 2015). The author has practiced prewriting group discussions in some writing classes where a considerably large amount of time had to be allotted for discussion at the expense of time for writing; however, its effects on writing performance were not able to be discerned due to the research design (Kano, 2018).
This study examines whether the writing support system enables a learner to write a well-structured, quality essay by himself/herself without dedicating time to prewriting discussion. In order to investigate the effects of discussion, writing performance of the subjects in the with-discussion group and those in the without-discussion group were compared.
According to Gall and Gall (1976), discussion is defined as a method of teaching in which "(1) a group of persons, usually in the roles of moderator - leader and participant, (2) assembles at a designated time and place, (3) to communicate interactively, (4) using speaking, nonverbal, and listening processes, (5) in order to achieve instructional objectives" (pp. 168-169). In the process of discussion, participants tackle an open-ended question by exchanging opinions and developing and modifying their own ideas to reach a variety of conclusions (Nystrand, Wu, Gamoran, Zeiser, & Long, 2003). Hope Davis (2013) argued the necessity of discussion for both affective and cognitive purposes. In the process of discussion, learners become aware of the need for social connection with peers and develop a sense of group affiliation as well as independence. Cognitively, discussion aids learners in gaining a deeper understanding of the content and insight into their lives and society by connecting their experiences to reality.
Gall and Gall (1993) classified discussions into three types: (1) cooperative learning discussions, (2) the subject mastery discussion method, and (3) issues-oriented discussions. In cooperative learning discussion, learners achieve such instructional objectives as academic achievement, positive attitudes toward the subject, higher motivation, higher self-esteem, and collaborative competencies. The subject mastery discussion method is similar to “Learning Thru Discussion” (LTD), developed by William Hill (1978), where learners read an assigned text beforehand to have a discussion in class. Before class, learners are expected to understand related issues in addition to terminology, key concepts, and main ideas. In issues-oriented discussions, which are similar to debates, controversial issues are discussed. The main purpose of these discussions is to provide learners with opportunities to recognize problems, analyze and evaluate opinions, and develop their own ideas. Learners need to obtain information relevant to the problem, but that is of lesser importance than the ability to approach a problem and identify a solution. Hence, in the practice of discussion, cognitive skills such as systematization of information and critical and logical thinking skills are expected to be developed.
2.2 Effects of Prewriting Discussion on Writing PerformanceThe role of discussions in learning has been investigated, and many studies have reported positive effects of prewriting discussion on writing performance. In the 1970s, for instance, Douglas Meyers (1979) compared the writing performance of the students who had prewriting discussion in pairs with that of students who received instruction in grammar, punctuation, and outlining, and he found positive effects of prewriting discussion on the writing performance of the majority of the students. Xin and Liming (2005) reported an improvement in the writing performance of the students who had prewriting discussion, and Mirzaei and Eslami (2015) found positive effects of group discussion on the quality of second language writing in general. Arumugam, Jambulingam, Supramaniam, and Kaur (2018) also found significant effects of small group discussions on the content and organization of academic writing tasks. Additionally, with the development of technology, the use of social networking platforms as a discussion tool is becoming common. Ping and Maniam (2015) investigated the effectiveness of Facebook-focused group discussion on writing performance and reported positive results.
While many studies have reported the positive effects of group discussion, there have been some researchers who found no significant effects. For example, Ling Shi (1998) compared the writing performance of a peer discussion group, a teacher-led prewriting discussion group, and a non-discussion group and found no statistically significant differences among the three groups. Nguyen, Admiraal, Janssen, and Rijlaarsdam (2018) compared two prewriting activities, free writing and group discussion, and found no statistical differences on the text quality the two activities produced while finding significant gains in the productivity of the free writing group.
2.3 Problems of Prewriting DiscussionOne of the problems of practicing prewriting discussion is the time that must be allotted for the discussion (Ping & Maniam, 2015). In the previous study, at least 30 minutes had to be set aside for prewriting discussion due to the students’ introverted personalities. The students needed adequate time to express their ideas and opinions. In the class evaluation administered at the end of the course, five out of fifteen students stated that discussion was a waste of time or that they wanted to work alone (Kano, 2018). According to Neumann and McDonough (2015), six out of twenty-two students reported that they preferred individual work to prewriting group discussion because they had greater concentration and less time pressure. This could be a result of the learners’ interdependent cultures where collective values take precedence over individual voice, which can stifle individual expression (Nguyen, Admiraal, Janssen, & Rijlaarsdam, 2018). Students from those cultures can be reluctant to express their opinions orally, and group discussions are likely to end in failure.
Another problem that participants may face is not grasping the overall picture of information and opinions presented in the process of discussion. Consequently, they may ignore previous arguments and repeat the same ideas and opinions, which can make it difficult to reach a conclusion (Sumi, Nishimoto, & Mase, 1996). One of the solutions to this issue is to visualize the course of discussion. Masafumi Kon (2012) tried to map out a discussion by schoolteachers on a whiteboard in order for them to reach a consensus with ease. In the questionnaire after the discussion, some teachers reported that the visualization of the discussion process had been helpful to generate and develop ideas as well as to activate discussion. Others stated that the degree of relevancy, originality, and importance of each opinion became clear, which served as a foundation for decision making.
With the advance in technology, online discussions have been practiced on various platforms, and there have been attempts to create a guided discussion process. Matsumura, Kato, Osawa, and Ishizuka (2003), for instance, developed and practiced a system that visualized the flow and content of discussion in a discussion map. In the questionnaire after the discussion, the participants reported that the system enabled them to understand the discussion structure in its entirety, and that it aided the participants in quickly grasping important points and arguments. Mochizuki, Hisamatsu, Yaegashi, Nagata, Fujitani, Nakahara, Nishimori, Suzuki, and Kato (2005) developed a software that created a visualization to guide discussion in an electronic conference room, and they evaluated the system through participant observation in addition to self-evaluation data collected by the questionnaire and interview post-experiment. They concluded that creating visual aids to guide discussion helped to promote arguments from various viewpoints, compare and analyze different opinions, and generate new ideas.
In the previous study, a writing support system was developed which visualized the flow and content of discussion, and it was found that the system helped learners organize information on the PC screen in addition to enriching the content of writing through the text reading and discussion (Kano, 2018). Although, the effects of the system as a whole were founds in the previous research, the problem about a significant amount of time spent for discussion remained unsolved; therefore, this study was conducted to examine the efficacy of group discussion in isolation―whether the group discussion could be eliminated from the system while maintaining the same quality of writing performance in word choice, grammatical correctness, logical development, and organization.
The experiment was administered in a required academic writing course for third-year students majoring in the English language. The writing topics covered in the course were animal testing, the death penalty, smoking in public places, hosting the Olympics, corporal punishment, grade-skipping, and a merit pay system. Two class hours were assigned for each of the topics, which included time for reading instructions, writing, revision, and evaluation.
During the first class, the students wrote an essay in 30 minutes on the topic of animal testing responding to the following prompt: Animal testing should be abolished in medical and cosmetic research labs. Do you agree or disagree? Use specific reasons and examples to support your answer. The data from these essays were used as pretest scores. In the final class, the students were required to write an essay responding to the following prompt regarding the employment system: Businesses should hire employees for their entire lives. Do you agree or disagree? Use specific reason and examples to support your answer. The resulting data were used as the posttest scores.
Fifteen university students (7 males, 8 females) who enrolled in the academic writing course in 2015 and twenty students (9 males, 11 females) enrolled in the same course offered in 2016 participated in this study. They were third-year, intermediate level students majoring in the English language at a private university located in the Tokyo metropolitan area. The class offered in 2015 was used as a with-discussion group, where an opportunity for a discussion in a group of three was provided in order to develop arguments and to organize essays using information on the shared monitor. Their TOEIC scores fell between 550 and 690 with the average of 617. On the other hand, the students enrolled in the 2016 course offering were placed in a without-discussion group and completed their essays individually instead. The average TOEIC score of this group was 605, ranging from 550 to 635.
3.3 Development of the Academic Writing Support SystemIn the previous study, a computer-supported collaborative writing system was developed in which learners used post-it cards on the computer screen to summarize, sort, and integrate information to help them develop and organize their ideas. Within the system, the students (1) read English passages on a selected topic, (2) extracted important facts and opinions from the passages and wrote each piece of information on a post-it card, (3) shared the information typed on the post-it cards in a small group, and (4) analyzed and categorized the cards on the computer screen. The support system was developed as an application in Apache+PHP+MySQL.
The system consists of six phases. In Phase 1 (see Figure 1), a short English passage about the selected topics is presented in order to provide students with a background knowledge which leads to a deeper understanding of the topic. The passages in the system were taken directly from a discussion textbook for college students (Ueda, Ueda, Taoka, & Yoneoka, 2014). The students were required to read the passage and jot down any facts on each of the post-it cards on the screen. After completing the post-it cards, they submitted them for use in Phase 3.
In Phase 2 (see Figure 2), students read opinions for or against the selected topic in English passages to expose them to a variety of ideas and expressions. They extracted and summarized pro and con opinions into keywords or key phrases on the post-it cards and submitted them to Phase 3.
Figure 1. Phase 1 Interface
Figure 2. Phase 2 Interface
Figure 3. Phase 3 Interface
In Phase 3 (see Figure 3), the fact and opinion post-it cards completed in Phases 1 and 2 were presented in different colors: Facts in yellow, pro opinions in blue, and con opinions in red. Students could move, sort, integrate, and organize the information by dragging each of the cards to specific areas. Students in the with-discussion group worked face-to-face in groups of three while looking at the shared screen to decide whether they would agree or disagree with the specified topic, and which information they would use to support their claim. Those in the without-discussion group used the information to determine their position on the topic individually.
In Phase 4 (see Figure 4), the fact and pro or con post-it cards the students had selected in Phase 3 were sorted and presented on the screen. They were presented with the option to create new cards if they wanted to add information. The students in the with-discussion group were given time for discussion to decide on any content they wanted to add while the students in the without-discussion group worked independently to research information on the internet.
In Phase 5 (see Figure 5), the students selected the fact and pro or con cards in order to organize their five-paragraph essay, which consisted of an introductory paragraph, three body paragraphs, and a concluding paragraph. The students in the with- discussion group collectively selected cards for the introduction, body, and conclusion paragraphs to organize a logically well-developed essay. Conversely, those in the without-discussion group followed the same procedure individually.
Figure 4. Phase 4 Interface
Figure 5. Phase 5 Interface
The students began writing their five-paragraph essay in Phase 6 (see Figure 6), utilizing the structured card they had completed in Phase 5 to draft their introduction, three body paragraphs, and conclusion. The number of characters and words were displayed at the bottom of the screen. After 30 minutes, the students submitted their essays to be printed out.
Figure 6. Phase 6 Interface
Data were collected from fifteen university students who enrolled in the academic writing course in 2015 (with-discussion group) and twenty students in the same course offered in 2016 (without-discussion group). In the first and the last class, both groups wrote academic essays which were scored by two independent graders in regard to (1) vocabulary choice, (2) grammar, (3) logical development within paragraphs, and (4) organization. Students could earn up to three points for (1) vocabulary choice, which checked whether the students used the appropriate word choice in the given context. Up to three points could be earned for the (2) grammar section, which reviewed word order, verb tenses, subject-verb agreement, articles, and pronouns, where mistakes frequently occur. No instruction on word choice and grammar were given to the students in the class, but it was hypothesized that lexical and grammatical mistakes would be reduced in the process of discussion. Three points were also allotted for (3) logical development, which assesses the consistency of content in a paragraph and whether there is an adequate topic sentence with supporting information. Regarding the (4) “introduction-bodies-conclusion” organization, one point was given if it was clear that the student was trying to write in the introduction-bodies-conclusion structure, even if the student did not complete the essay in the given 30 minutes. It was predicted that logical development and content organization would be refined and elaborated through discussion in which the students were visually manipulating the contents of their writing on the PC.
Two graders scored independently. Since the interrater reliability was considered significant (r = .83, p < .00), the average scores between the graders were used for the data analysis. To assess the difference between pre and posttest, repeated t-tests were administered in terms of (1) vocabulary choice, (2) grammar, (3) logical development in a paragraph, (4) “introduction-bodies-conclusion” organization, and (5) total scores.
Statistical Package for the Social Sciences (SPSS) version 24 was used for the statistical analysis.
In order to compare the differences in the pretest scores of the with-discussion group and the without-discussion group, independent t-tests were introduced. Normality was checked using the Shapiro-Wilk test, which showed that the data followed normal distribution (W(35)=.974, p=.567). Regarding the (2) grammar (t = .986, df = 33, p = .331) and (4) organization (t = .525, df = 33, p = .585) scores, a one-way ANOVA (repeated measure) was administered to examine the effects of group discussion on writing performance since no significant differences were found in the pretest scores of the two groups. On the other hand, as significant differences were found in the (1) word choice (t = 4.455, df = 33, p < .000), (3) logical development (t = 4.226, df = 33, p < .000), and total scores (t = 3.655, df = 33, p = .001), an ANCOVA was applied to control the initial difference as a covariate.
4.1 Word ChoiceThe results of the word choice scoring in the pretest and the posttest for both groups are presented in Table 1 where n=15 represents the with-discussion group and n=20 represents the without-discussion group. The mean score of the word choice in the pretest and the posttest for the with-discussion group was 1.67 (SD = .52) and 2.00 (SD = .46), respectively, with a mean difference of .33. There was a significant difference (p = .019) between the pretest and posttest scores for the with-discussion group. The mean score from the pretest and the posttest for the without-discussion group was 1.00 (SD = .36) and 1.50 (SD = .54), respectively, with a mean difference of .50. The difference between the pretest and posttest in the without-discussion group was also significant (p = .000). The results indicated that both groups’ performance regarding word choice had improved.
| Pretest | posttest | ||
| with-discussion group (n=15) | M SD |
1.67 0.52 |
2.00 0.46 |
| without-discussion group (n=20) | M SD |
1.00 0.36 |
1.50 0.54 |
An independent sample t-test was introduced to examine whether there were significant differences in the pretest scores for both groups. The mean pretest score was 1.67 for the with-discussion group and 1.00 for the without discussion group, which indicated a significant difference between the two groups （t = 4.455, df = 33, p = .000). The results indicated that the pretest would affect the posttest performance, so the correlation between the pretest and posttest was calculated. It was found that r was .636, which was statistically significant with p < .01 (p = .000). Then, an analysis of covariance (ANCOVA) was administered to test for differences in the average scores of the groups, controlling the effect of the pretest as a covariate. In the analysis of covariance, the differences in the mean scores for both groups were not statistically significant (F = 1.190, p = .284), and no significant interaction was determined between time (pretest/posttest) and treatment (with-/without-discussion) (F = .907, p = .348), either.
4.2 GrammarThe grammar scores for the pretest and posttests are displayed in Table 2. Since no significant difference was found in the pretest scores between the two groups, a one-way ANOVA (repeated measure) was utilized to determine the effects of discussion on the posttest scores. While significant differences were found between the pretest and posttest scores (F (1, 33) = 31.498, p = .000), no significant differences were determined between groups (F (1, 33) = 2.870, p =.100), and no significant interaction between time (pretest/posttest) and treatment (with-/without-discussion) (F (1, 33) = .226, p =. 637) was observed, either. The results indicate that the grammar score became higher as a result of the course itself, rather than the opportunity for discussion.
| pretest | posttest | ||
| with-discussion group (n=15) | M SD |
1.60 0.47 |
2.13 0.52 |
| without-discussion group (n=20) | M SD |
1.45 0.43 |
1.90 0.26 |
The results of logical development in the pretest and the posttest for both groups are presented in Table 3. The mean scores of the pretest and posttest for the with-discussion group were 1.40 (SD = .39) and 2.10 (SD = .69), respectively, with a mean difference of .70. The difference between the pretest and posttest in the with-discussion group was significant, with p < .01 (p = .002). The mean score from the pretest and the posttest in the without-discussion group was 0.83 (SD = .41) and 1.48 (SD = .77), respectively, with a mean difference of .65. The difference between scores for the pretest and posttest in the without-discussion group was also significant as p < .01 (p = .001). The results indicated a significant improvement in logical development performance for the students in both groups.
| pretest | posttest | ||
| with-discussion group (n=15) | M SD |
1.40 0.39 |
2.10 0.69 |
| without-discussion group (n=20) | M SD |
0.83 0.41 |
1.48 0.77 |
An independent sample t-test was used to test differences in the pretest scores for both groups. The mean pretest score of the with-discussion group was 1.40 and .83 for the without-discussion group. The difference in the mean scores was .57 and a statistically significant difference was found with p < .01 (p = .001). The correlation between the pretest and posttest was calculated since the results indicated that the pretest would affect the posttest performance, and it was found out that r was 0.4, which was statistically significant with p < .01 (p = .004). An analysis of covariance (ANCOVA) was used to test for differences in the mean scores of each group in order to control the effect of the pretest scores as a covariate. In the covariance analysis, the differences in mean scores for with- and without-discussion groups were not statistically significant (F = 1.058, p = .312), and the interaction between time (pretest/posttest) and treatment (with-/without-discussion) was not significant (F = .522, p = .476), either.
4.4 OrganizationRegarding the “introduction-bodies-conclusion” structure pretest scores, no significant differences were found between the with- and without-discussion groups (see Table 4). Therefore, a one-way ANOVA repeated measure was administered to determine the effects of discussion practice. Despite significant differences being found between the pretest and posttest scores (F (1, 33) = 49.222, p = .000), no significant differences were found between the with- without discussion groups (F (1,33) = .148, p =.708), and no significant interaction between time (pretest/posttest) and treatment (with-/without-discussion) was found (F (1, 33) = 2.372, p = .133), either. Regardless of discussion, all subjects demonstrated the ability to write in an “introduction-bodies-conclusion” structure at the completion of the course.
| pretest | posttest | ||
| with-discussion group (n=15) | M SD |
0.40 0.47 |
0.80 0.25 |
| without-discussion group (n=20) | M SD |
0.33 0.34 |
0.95 0.36 |
The results of the total scores for both groups are presented in Table 5 with fifteen subjects in the with-discussion group and twenty in the without–discussion group. The mean scores of the with-discussion group for the pretest and the posttest was 4.93 (SD = .90) and 7.70 (SD = 1.42), respectively, with a mean difference of 2.14. The difference in scores between the pretest and posttest for the with-discussion group was significant with p < .01 (p = .000). The mean scores for the pretest and the posttest in the without-discussion group was 3.57 (SD = 1.21) and 5.60 (SD = 1.01), respectively, with a mean difference of 2.03. The difference between the pretest and posttest in the without-discussion group was also significant, with p < .01 (p = .000). The results of the overall scores indicated the subjects in both groups significantly improved their writing ability overall by the completion of the course.
| Pretest | posttest | ||
| with-discussion group (n=15) | M SD |
4.93 0.90 |
7.07 1.42 |
| without-discussion group (n=20) | M SD |
3.57 1.21 |
5.60 1.01 |
An independent sample t-test was introduced to examine any significant differences in the pretest scores for both groups. The mean pretest score was 4.93 for the with-discussion group and 3.57 for the without-discussion group. The difference between the mean scores for both groups was 1.36 which indicated a significant difference as p < .01 (p = .000). There was a high correlation between the pretest and posttest (r = .680, p < .000), which indicated that the pretest would affect the posttest performance. In order to control the effect of the pretest as a covariate, an analysis of covariance (ANCOVA) was used to test for differences in mean scores between the groups. In the covariance analysis, the differences between the mean scores for the with-discussion group and without-discussion group were not statistically significant (p = .327).
As a result of this study, no significant effects of discussion were observed regarding vocabulary choice, grammar, logic, and organization, which suggests that the use of the writing support system should enable students to write more natural, accurate, logical, and well-organized academic essays individually.
5.1 Word ChoiceANCOVA was used to assess the differences in appropriate word or phrase selection between the with- and without-discussion groups. The results showed no significant difference on the mean scores of word choice between the with- and without-discussion groups (F = 1.190, p = .284), and no significant interaction was observed between time (pretest/posttest) and treatment (with-/without-discussion) (F = .907, p = .348). The discussion had no effect on the word choice scores as the subjects in both groups scored higher in this category on the posttest. In summation, the opportunity for group discussion did not influence the word choice scores of the students.
5.2 GrammarThe results of the one-way repeated ANOVA demonstrated a significant difference between the pretest and posttest grammar scores (F (1, 33) = 31.498, p = .000), but no significant differences were found between the two groups (F (1,33) = 2.870, p = .100) or interaction (F (1, 33) = .226, p = .637). The opportunity for discussion had no effect on grammar scores as the students in both groups improved their grammatical skills during the semester.
5.3 Logical DevelopmentThe difference in the logical development scores was tested with ANCOVA, which determined that there were no significant differences in the mean scores of the with- and without-discussion groups (F = 1.393, p = .247) and no significant interaction between time (pretest/posttest) and treatment (with-/without-discussion) (F = 2.621, p = .116). The subjects in both groups scored higher on the posttest, indicating that the students in both groups wrote a more logically well-developed essay regardless of whether they had an opportunity for discussion with their peers.
5.4 OrganizationTo determine if the students organized their essay in accordance with the “introduction-bodies-conclusion” structure, a one-way ANOVA repeated measure was administered. The results showed no significant differences between the with- and without-discussion groups (F (1,33) = .148, p = .708) or interaction (F (1, 33) = 2.372, p = .133). However, the difference between the pretest and posttest scores was significant (F (1, 33) = 49.222, p = .000). The results demonstrated that the students became more likely to write their essays in the “introduction-bodies-conclusion” structure at the end of the course whether they engaged in prewriting discussion or not.
5.5 Total ScoreThe results of an independent sample t-test showed significant differences between the pretest and posttest scores for both the with-discussion and without-discussion groups (p < .01), which indicated the subjects in both groups had significantly improved their academic writing. However, the results of ANCOVA showed no significant differences in mean scores between the with-discussion group and the without-discussion group, and no significant interaction was found. The results indicated that it would be possible for an individual student to write an appropriately worded, grammatically correct, and logically well-developed academic essay using the writing support system without an opportunity for prewriting discussion.
The various analyses prove the efficiency of the system; however, some questions remain. For example, while the quality of writing can be appropriately measured by word choice, grammar, logic, and organization, it is a complex process involving many factors, so other scales should be taken into consideration in future research. Additionally, there are some limitations concerning the subjects. The experiment was administered in a third-year university course; therefore, only a specific level of student was able to participate. There is a possibility that different effects of discussion would be observed in different groups of students with different levels of English proficiency. In future research, it will be necessary to use students with different proficiency levels.
This work was supported by JSPS KAKENHI Grant Number 24300286.
