Second Language Acquisition - Theory and Pedagogy: Proceedings of the 6th Annual JALT Pan-SIG Conference.
May. 12 - 13, 2007. Sendai, Japan: Tohoku Bunka Gakuen University. (pp. 42 - 55)

Correlations between Japanese high school students' language learning beliefs, self-reported attitudes towards group dynamics, and performance in two in-house EFL tests
J Title
by Yo Hamada (Temple University Japan), Setsuko Oda (International Christian University),
and Kazuya Kito (Temple University Japan)


The value of introducing insights from group dynamics is beginning to attract attention in the field of second language education, and recent studies suggest that it is important for language teachers to have an increased awareness about group dynamics. In particular, three aspects of group dynamics are considered significant in language teaching: (1) the building of student-teacher rapport, (2) promoting positive peer language learning role modeling, and (3) fostering group work interactions. Research indicates that these three points might influence learner attitudes and motivation. However, no studies have yet to investigate how group dynamics impact learners' performance. In the present study, we examine the correlation between a group of 59 Japanese high school students beliefs about three aspects of group dynamics with their performance in L2 learning in a Japanese secondary school context. The extent that self-reported beliefs about group dynamics and student midterm and final test scores correlate was explored systematically. Results revealed that there is a weak but positive correlation (r = .22) between students' beliefs about group dynamics and their test performance.


Classroom group dynamics, student rapport, language learning role models, group/pair work

Although effective group dynamics is widely considered as essential in fostering learning, it has not attracted attention among researchers until recent years. Indeed, language teachers are more often than not considered to be experts on language, not psychology or sociology. In reality, however, many non-linguistic issues must be taken into account in the process of language teaching. This paper highlights how avowed attitudes towards group dynamic processes correlate with test performance.
This decade has seen a surge in interest in the study of group dynamics among teachers and researchers. Many now regard classes as communities of learning individuals with different characteristics and backgrounds rather than mere, temporary groups of language learners (Ryder & Wilson, 1996; Senior, 2002; Dornyei & Murphey, 2004). Dornyei and Murphey (2004, p. 5-6) have espoused the concept of "group dynamics" from the field of social psychology as follows:
In a 'good' group, the L2 classroom can turn out to be such a pleasant and inspiring environment that the time spent there is a constant source of success and satisfaction for teachers and learners alike. And even if someone's commitment should flag, his or her peers are likely to 'pull the person along' by providing the necessary motivation to persist. (p. 5 - 6)

[ p. 42 ]

In other words, teachers, as central figures in groups, bear responsibility to make optimal use of group dynamics for effective learning.
". . . teachers, as central figures in groups, bear responsibility to make optimal use of group dynamics for effective learning."

Although it is easy to assume that having increased motivation will also enhance students' progress, no studies to our knowledge has attempted to find out if there is a correlation between an observed group dynamics and students' performance on some tests.
In order to investigate the dynamics of a group, it is important to investigate what is actually happening in a group as objectively as possible. Also, it is necessary to ask insiders what they think is happening in their group in order to better understand its dynamics. As can be seen from a line of studies on leaner beliefs about language learning, what kind of beliefs the students have seems to have a great impact on their success in learning (Kern, 1995; Horwitz, 1985, 1987). If what students believe about language learning can influence their language learning outcomes, how students perceive group dynamics should also have an impact on their learning.
The present study seeks to explore how students' perception of three aspects of group dynamics can influence their language learning. The three aspects this study focuses on are student-teacher rapport, peer role modeling, and group work interaction. These have been studied previously by Murphey, 1997, 1999a, 1999b; Murphey and Arao, 2001; Dornyei and Murphey, 2004, Senior, 2002; and Fushino, 2006. First, we investigated the beliefs of 59 high school students about those three aspects of group dynamics and then saw how this correlated with their achievement in two school exams. The data was then quantitatively analyzed with a limited descriptive analysis (Trochim, 2006). We conclude by discussing some of the pedagogical implications.

Issues in group dynamics

This section first provides an overview of how group dynamics is situated in the field of L2 learning. Next, we introduce three aspects of group dynamics considered to be especially important in language learning. Finally some findings from the past belief studies are mentioned.

What is group dynamics? How is it important in language learning?

Group dynamics, first developed by the German-American social psychologist Kurt Lewin, is an inter-disciplinary study which evolved in the 1940s from diverse fields of sociology and psychology. It has attracted much interest from such fields as business, psychotherapy and politics. In educational settings, however, research has been relatively scarce. One reason is due to the complicated factors involved in what constitutes a "group" (for an extensive review on this issue, see Dornyei & Murphey, 2004, p. 10 - 33.)
Group dynamics in L2 learning contexts focuses on the dynamics of groups of learners, which develop and evolve as time goes by. It consists of not only a domain of knowledge, but also the attitudes and interactions one has with a general group. (Dornyei & Murphey, 2004, p. 2).
Senior (2002, p. 402) found that learners learn most effectively when language classes pull together as unified groups. He also maintains that teachers should understand that all language classes are made of individuals, and encourages teachers to set up learning tasks to accommodate the social needs of their students as well as the learning objectives. Dornyei and Murphy (2004, p. 5) also mention that a pleasant and inspiring environment in which teachers and students enjoy spending time together enhances classroom success and the level of satisfaction among teachers and learners.

[ p. 43 ]

Student-teacher rapport

Few teachers would doubt the value of having a good, friendly relationship with their students. Brown (2001, p. 212) describes that constructing rapport with students means establishing relationships with openness and respect. This way, teachers can help students feel that they are capable, competent, and creative.
Needless to say, the teacher is a central figure in most learner groups, and he/she should make every effort to help establish a nice, relaxing atmosphere among students in order to reduce student anxiety at the initial stage of group-forming. Good teacher leadership may be a factor promoting good group interaction (Alger, 2005). A nice, friendly environment may reduce the anxiety of many students, and for this reason establishing good student-teacher rapport is of utmost importance (Dornyei & Murphy, 2004).
"Good teacher leadership may be a factor promoting good group interaction."

Language learning role modeling

As long ago as 1967, Polanyi recognized that modeling within a community of practice is a first step in the learning process (Polanyi, 1967). The concept of role modeling usually includes both the teacher role model and peer role model, but the present study focuses on the peer role model.
Murphey (1997, p. 1) strongly advocates the use of near peer language learning model by stating that "structuring classroom experiences to enhance near peer role modeling may be one of the most powerful ways teachers can enhance learning." One activity he introduces is publishing students' language learning histories in order to take advantage of peers experience, using Vygotsky's zone of proximal development (ZPD) theory.
Following the same near peer modeling approach, Murphey and Arao (2001) made an intriguing experiment in which they showed a video to change students' beliefs about language learning.
Murphey (1999, p. 1) insists that the near peer model is more effective than other role models, by stating "the greater impact comes from our peers and not our teachers and parents."

[ p. 44 ]

Student beliefs

A number of studies have been conducted on learners' beliefs, starting from Horwitz (1985, 1988) and Kern (1995). In Japan, quite a few studies have also been conducted such as those of Sakui and Gaies (1999); Ueki (2002); Nakayama (2004); Suzuki and Wada (2004); Suzuki (2004); Suzuki and Kumazawa (2006).
Sakui and Gaies (1999) studied Japanese learners' beliefs about language learning and found out that learners hope to experience a feeling of joy from studying English. For example, most learners agreed "English conversation class should be enjoyable (p. 482)." Analyzing their beliefs, it seems that investigating learners' beliefs about language learning can lead to more effective instructional planning and implementation (p. 487).
Suzuki and Kumazawa (2006) conclude their research by stating that it is necessary for teachers to make sure what beliefs students have in order to construct effective classes. Also, student beliefs reflect the social environments that students are immersed in, and are sensitive to entrance examination pressures and school rankings (p. 221). Teachers have a duty to become familiar with students' learning environments as well as their beliefs, so as to tailor lessons appropriately. This includes sensitivity to student preferences and moods.

Research Question

The main research question addressed in this study is: How do the self-reported learner beliefs about three aspects of group dynamics correlate to their language learning performance as ascertained by two in house tests? The null hypothesis is that there is no statistically significant correlation between the questionnaire items and test performance. The alternative hypothesis is that students who expressed a higher interest in group dynamics also tended to get higher test scores.



Participants for this study were 59 first-year students in a private high school in central Tokyo, aged 15-16. 19 of them are male students and 40 are female students. The English proficiency level of most students is below the National average for the Japanese high school students. According to Dictionary of Entrance Exams in High Schools in Tokyo, the average standardized rank score (hensachi in Japanese) of this school is 46, which means the English proficiency level of most students is below the national average by 4 points. Most respondents in this study had standardized rank scores somewhere below that.
The 59 participants are all enrolled in Oral Communication (OC) classes taught by the first author. Although about 100 students volunteered to participate in the study, only 59 completed both instruments used in this study.

[ p. 45 ]


Three instruments were used in this study: (1) a 5 item survey in Japanese and English using a 5-point Likert scale, (2) a 50 item mid-term test, and (3) a 49 item final test. Each of these items will be briefly described.

(1) Survey Questionnaire

The questionnaire consisted of three questions in Japanese and two in English about the overall group dynamics of the class. An English translation of the original questionnaire, which is in Appendix A, would look something like this -

	1) How do you like your oral communication class?          5   4   3   2   1
	2) How do you like your teacher?                           5   4   3   2   1
	3) How do you like your class?                             5   4   3   2   1
	4) How do you like pair work or group work?                5   4   3   2   1
	5) Have you been influenced by a role model in this class? 5   4   3   2   1

Questions 1 and 2 are postulated to measure how the students perceive their student-teacher rapport. Questions 3 and 4 thought to investigate their views about group work, and Questions 5 is conjectured to indicate how students feel about peer role modeling.

(2) Midterm test

The mid-term was composed of seven listening sections and three writing sections. Five of the listening sections were directly related to the contents taught in class. For the other two listening sections included more challenging material. The three writing sections were similarly formatted. The questions for the two sections were directly from the class lessons. The last section was slightly modified to test the practical use of their knowledge. Some of the items which appeared in the midterm test are listed in Appendix B.

(3) Final test

The final test was composed of eight listening sections and three writing sections. five of the listening sections were directly related to the contents taught in class. For the other two listening sections included more challenging material. The three writing sections were similarly formatted. Some of the items which appeared in the midterm test are listed in Appendix C. Table 1 summarizes the differences between the midterm and final test.

Table 1. Descriptive information about mid-term and final-test used in this research.

Mid-term test Final test
Listening Sections 7 8
Writing sections 3 3
Total Sections 10 11
Test Type criteria-referenced paper-and-pencil test criteria-referenced paper-and-pencil test
Total number of items 50 49
Time Length 50 minutes 50 minutes
Max Possible Score 100 98

[ p. 46 ]


The mid-term was administered in November 2006 and the final was in February 2007. Students were given 50 minutes to finish each test.

Data analysis

First, the participants were categorized into three levels, based on their average scores of the two tests: Level 1 for those who scored 35 points or less (n=14), Level 2 for those with mid-range scores of 36-65 points (n=37), and Level 3 for those whose average score on the two tests was 66 points or higher (n=8). The test scores were then correlated with the questionnaire survey responses using a one-way ANOVA for each item with SPSS.


The descriptive statistics for both tests are summarized in Tables 1 and 2.

Table 2 Descriptive statistics for the two exams
(n=59) Mean Max Min SD Skewness Kurtosis Range Variance
mid-term 50.1 86 10 19.24 -0.22 -0.72 76 370.09
final 45.92 89 8 19.8 0.05 -0.78 81 372.04
combined 48.01 85 10 18.46 -0.06 -0.736 75 340.81

The data of Table 2 illustrates how the mid-term and the final test results were similar. The mean, max, min, SD, skewness, and kurtosis for the two exams suggest that the two exams were nearly equal in difficulty.
Now let us take a look at the survey results, which are summarized in Table 3.

Table 3 Student questionnaire results (n=59)
Survey Question Number Mean SD Strongly disagree
Strongly agree
1. How do you like OC? 2.56 1.35 30.50%
2. How do you like your teacher? 3.51 1.13 6.78%
3. How do you like your class? 3.41 1.26 10.17%
4. How do you like pair/ group work? 3.00 1.20 13.56%
5. Have you been influenced by a role model in this class? 2.83 0.90 8.47%

[ p. 47 ]

In Table 3, the means and standard deviations for each item on the questionnaire survey are presented. The items with higher mean values indicate that more respondents strongly agreed with that item. The responses to Question 2 and 3 suggest that most participants generally liked their teacher and class. On the other hand, the responses to Question 1 suggest that a large number of participants have a negative attitude toward OC.
Now let us see how well this Likert scale correlated with the average test performance. To calculate that, a one-way ANOVA was conducted between proficiency or test score and each of the five survey questions. The results are indicated in Table 4.

Table 4 A one-way ANOVA of each question and test performance. (n=59)
Survey Question Source Sum of Squares Degrees of freedom Mean Score F-distribution variable Level of significance
1. How do you like OC? Composite
Test Score
9.28 2 4.64 2.85 0.07
2. How do you like your teacher? Composite
Test Score
2.60 2 1.30 1.21 0.31
3. How do you like your class? Composite
Test Score
15.13 2 7.56 6.31 0.00
4. How do you like pair/ group work? Composite
Test Score
14.02 2 7.01 6.33 0.00
5. Have you been influenced by a role model
   in this class?
Test Score
7.47 2 3.74 6.38 0.00

Table 4 suggests that three of the five variables were statistically significant: liking the class, enjoying pair or group work, and being influenced by an in-class role model.

Now let us further explore the relation between proficiency groups and class rating. You will remember that we divided this class into three proficiency levels: (1) low, (2) medium, and (3) high. The color-coded responses of each student to survey question 3 appear in Figure 1.

Figure 1. Differences in the class favorability rating among three proficiency groups.
Figure 1

[ p. 48 ]

What Figure 1 suggests is that students who indicated that they enjoyed the class also tended to score higher on their mid-term and final exams. Since Figure 1 does not imply causality, however, further investigation is needed on this point.
Now let us consider how test performance relates to attitudes about pair/group work. Figure 2 correlates composite test scores with Likert-scale responses to the question of how much the respondents liked pair/group work.

Figure 2. Differences in pair/group work favorability ratings among three proficiency groups.
Figure 2

What Figure 2 suggests is students who indicated that they like pair/group work tended to score higher on their mid-term and final exams. However, since causality cannot be ascertained, further investigation is due.
Now let us consider how test performance might relate to having in-class role models. Figure 3 correlates composite test scores with Likert-scale responses to the question of whether the respondents had a peer role model for English.

Figure 3. The relationship between proficiency group and how strongly respondents thought that they had an in-class role model.
Figure 3

Figure 3 indicates that students who had an in-class role model tended to score higher on their mid-term and final exams. Again, since it would be premature to purport causality, further research is warranted.

[ p. 49 ]


Results show that for the students in the Level 1 (the low-scoring group), a statistical significance between three of the five variables examined in this study existed. First, the data suggest that the students who indicated that they did not like the class tended to achieve lower test scores. Moreover, students who said they did not enjoy group work interactions tended to also score lower on their tests than those who indicated otherwise. This suggests that student perceptions about group work interaction are somehow related to test scores.
This study supports past research by Dorneyi and Murphey (2004). Although a significant relation was not found between how much students said they liked their teacher and their performance on the mid-term and final tests, the authors maintain that it should be teachers' responsibility to keep the classroom atmosphere as pleasant as possible in order to promote good group work interaction among students.
One other significant variable in this study was about the extent that students indicated they were influenced by a peer model. This item will be discounted from this study because there is not enough information to really shed light on ways that peers influence each other. Further studies are needed with more in-depth questions to explore this issue.
So far we have been focusing on low-performing students. Among students in the mid- and high-performing groups, it should be noted that no significant differences were noted between test scores and any of the variables in this study, which was analyzed using Tukey and Scheffe tests. How should this be interpreted? One possible interpretation is that students with relatively high proficiency are not as likely to be affected by group issues as students with relatively low proficiency. However, before suggesting any causal relationship, further investigation is needed.
Dornyei (2001, p. 42-46) explains how group dynamics focuses on how groups behave and develop, and suggests how positive group dynamics can be cultivated. For example, giving learners chances to get to know each other better, taking advantage of group legends (co-constructed stories), and establishing constructive group norms can help cultivate a classroom esprit de corps.
". . . an important first step is for teachers to recognize group issues as important, so as to create a comfortable environment in which students can enjoy, relax, and at the same time learn."

This paper has emphasized that an important first step is for teachers to recognize group issues as important, so as to create a comfortable environment in which students can enjoy, relax, and at the same time learn. As Lightbown and Spada suggest (1999, p. 40-41), it is essential to get to know more about learners and, in Krashen's terms, to lower their 'affective filters.'
Although most high school teachers tend to focus heavily on teaching only the linguistic elements needed for the university entrance examinations, efforts to improve the classroom atmosphere should also be valued more because, as this study suggests, it might also affect students' performance. In light of the significance of group interaction, another thing teachers should keep in mind is making optimal use of pair/group work.

[ p. 50 ]

Next, the concept of language learning role models should be introduced to students explicitly. In Japanese high school contexts, this notion is relatively new. By introducing this concept, students may better understand language learning role models and why this can trigger their willingness to find an appropriate role model. Obviously the same role model is unlikely to suit everyone. Each student is different in regards to proficiency level, learning style, and overall personality. In considering these differences, it is ideal for each student to have his or her own language learning role model. Integrating Murphey's (1997) near peer model based on the concept of Vygotsky's ZPD is an area for further research about Japanese high school English education as evidenced by this study.


This study has investigated the correlation between 59 Japanese high school students' self-reported beliefs about group dynamics in relation to their performance on two in-house tests. The results suggest a weak, but positive correlation between three aspects pertaining to group dynamics and scores on the mid-term and final tests among 14 students who did not perform well on those tests.
In mentioning these results, it is important to acknowledge some of the limitations of this study. First of all, as Manfredo and Shelby (1988, pp. 731-743) has pointed out, self-reported attitudes may not actually reflect actual attitudes. It is quite likely that at least some students may have idealized their responses rather than reporting their actual attitudes. For this reason self-reports should be corroborated with other measures. Second, since the student survey consisted of only five questions, there was no doubt considerable construct under-representation: to accurately detect how students felt about social dynamics a broader range of questions (and probably response types) is needed. Finally, we should not forget the fact that only 59 of the original 100 volunteers in this study did not complete all three instruments and were subsequently discounted from this study. Over 40% of the original volunteers represent a unknown sub-set of the original survey population whose characteristics cannot be surmised.
To further strengthen and generalize these results, these suggestions are made:
  1. This research question should be investigated with a larger number of participants,
  2. Studying more motivated learners of high proficiency levels,
  3. The survey questions need to be worded in such a way that there would be less expectancy bias effect. As Garson (2006, 2007) suggests, the way the surveys are worded can have a significant impact on response outcomes.
  4. More sophisticated data analyses using G-theory should be conducted.
Despite its limitations, we hope that the result of this study represents an early step toward understanding the benefits and values of group dynamics and that this study will trigger further investigations on group dynamics in the field.


Alger, G. (2005, June 23). Literacy teachers' interactions with instructional leaders: Students reap the benefits. Current Issues in Education 8 (13). Retrieved November 1st, 2007 from

Brown, D. H. (2001). Teaching by principles: An interactive approach to language pedagogy. N.Y: Pearson Education.

Brown, J. D. (1998). Understanding research in Second Language Learning: A teacher's guide to statistics and research design. Cambridge: Cambridge University Press.

Dictionary of entrance exam in high schools in Tokyo [Tokyo-to Juken Jiten], retrieved August 31, 2007, from

Dornyei, Z. (2001). Creating the basic motivational conditions: Motivational strategies in the language classroom (p. 31-49). Cambridge: Cambridge university press.

Dornyei, Z., & Murphey, T. (2004). Group dynamics in the language classroom. Cambridge: Cambridge University Press.

[ p. 51 ]

Garson, G. D. (2006, 2007). Survey Research. Retrieved December 4, 2007 at

Fushino, K. (2006). Course choices and attitude toward group work. In K. Bradford-Watts, C. Ikeguchi, & M. Swanson (Eds.) JALT 2005 Conference Proceedings. (pp. 195-205). Tokyo: JALT.

Horwitz, E. K. (1985). Using student beliefs about language learning and teaching in the foreign language methods course. Foreign Language Annals, 18 (4), 333-343.

Horwitz, E. K. (1988). The beliefs about language learning of beginning foreign language students. Modern Language Journal, 72 (3), 283-294.

Kern. G. R. (1995) Students' and teachers' beliefs about language learning. Foreign Language Annals (28), 71-92.

Lightbown, M., P., & Spada, N. (1999). Theoretical approaches to explaining second language learning: How languages are learned (p.32-48). Oxford: Oxford University Press.

Manfredo, M. J., Shelby, B. (1988). The effect of using self-report measures in tests of attitude-behavior relationships. Journal of Social Psychology, 128 (6) 731-743.

Murphey, T. (1998). Motivating with near peer role models. In B. Visgatis (Ed.) JALT 1997 Conference Proceedings: Trends and Transitions. (pp. 201-205). Retrieved March 9, 2007, from

Murphey, T. (1999a). Publishing students'language learning histories: For them, their peers, and their teachers. Between the Keys: Newsletter of the JALT Material Writers SIG 7 (2), 8-11. Retrieved March 9, 2007, from

Murphey, T. (1999b). The Rapport Hypothesis: Reinterpreting the Peers vs. Parents Debate. Nagoya Avenues, 8-9. Retrieved March 9, 2007, from

Murphey, T., & Arao, H. (2001). Reported belief changes through near peer role modeling. TEFL-EJ 5 (3), Retrieved March 9th, 2007, from

Polanyi, M. (1967). The tacit dimension. New York: Anchor Books.

Ryder, M. & Wilson, B. (1996). Dynamic learning communities: an alternative to designed instructional systems. Retrieved September 29, 2007 from

Sakui, K., & Gaies, S. J. (1999). Investigating Japanese learners' beliefs about language learning, System, 27, 473-492.

Senior, R. (2002). A class-centred approach to language teaching. ELT Journal, 56, 397-403

Suzuki, S. (2004, November). Japanese teachers' beliefs about learning and teaching. Paper presented at the second international conference of ASIA TEFL, Seoul, South Korea.

Suzuki, S., & Kumazawa, T. (2006). Constructing a Japanese secondary school students' beliefs model. Step Bulletin, 18, 215-223.

Trochim, W. M. K. (2006). Descriptive Statistics. Retrieved November 2, 2007 from

Ueki, R. (2002). Structure of high-school students' beliefs about learning. Journal of Japanese Educational Psychology, 50 (3), 301-310.

Main Article Appendix A Appendix B Appendix C

2007 Pan SIG-Proceedings: Topic Index Author Index Page Index Title Index Main Index
Complete Pan SIG-Proceedings: Topic Index Author Index Page Index Title Index Main Index

Last Main Next

[ p. 52 ]