Putri Anggraeni, et al / Journal of English Language Teaching 6 (1) (2017) 37 ELT FORUM 6 (1) (2017) Journal of English Language Teaching http://journal.unnes.ac.id/sju/index.php/elt CLUSTERING TECHNIQUE AND PEER ASSESSMENT IN TEACHING WRITING RECOUNT TEXT TO JUNIOR HIGH SCHOOL STUDENTS Eka Yunita Widyawati, Novia Trisanti  English Department, Faculty of Languages and Arts, Universitas Negeri Semarang, Indonesia Article Info ________________ Article History: Received in June 2017 Approved in July 2017 Published in August 2017 ________________ Keywords: clustering; peer assessment; recount text; writing ____________________ Abstract ___________________________________________________________________ This study was conducted under the consideration of the observation in SMPN 39 Semarang that many students faced problems in learning to write a text. This research is aimed at investigating the effectiveness of using clustering technique and peer assessment in teaching writing recount text compared to three-phase technique and the students’ perception of the lesson. A quasi-experimental was used as the research design in this study. The subjects were the eighth graders of SMPN 39 Semarang in the academic year of 2016/2017 with 60 students as the samples. The control group (VIIIC) was treated using three-phase technique and teacher assessment, while the experimental group (VIII D) was taught using clustering technique and peer assessment. The findings revealed that the use of clustering technique and peer assessment significantly improved the students’ achievement in writing recount text. It was proven by the result of t-test in which the t-value (4.31) was higher than t-table (2.00). In addition, based on the result of the questionnaire, 90% of the students agreed that the use of clustering technique and peer assessment improved the students’ writing achievement. All in all, the use of clustering technique and peer assessment is more effective for teaching writing recount text to junior high school students than three-phase technique and teacher assessment. Moreover, most of the students have a positive perception to the lesson. © 2017 Universitas Negeri Semarang  Correspondent Address: ISSN 2252-6706 B3 Building FBS Unnes Sekaran, Gunungpati, Semarang, 50229 E-mail: unnes_english@yahoo.com E. Yunita Widyawati & N. Trisanti / Journal of English Language Teaching 6 (1) (2017) 38 INTRODUCTION As one of the four language skills which should be mastered by junior school students, writing is different since the writers should arrange the words and phrases into some sentences which can be clearly understood by the readers. Meyers (2005) stated that writing is a way to produce language which you do naturally when you speak. A good writing needs a long process that may include planning, drafting, editing (reflecting and revising), and writing final version (Harmer 2004). Some elements including a topic sentence, supporting sentences, concluding sentence, unity, and coherence also should be considered in writing (Oshima and Hogue 2006). According to School-Based Curriculum, any text types in English should be mastered by junior school students; one of them is recount text. Recount text is one of the text types which should be mastered by the eighth-grade students. According to Anderson (2003), recount text is a text that retells past events in the order in which they happen. In constructing a recount text, a writer has to consider some steps; (1) first paragraph that gives background information including the participants, place, and time, (2) a series of paragraphs that retell the events in the order in which they happened, and (3) a concluding paragraph. In addition, the specific language features including the use of proper nouns, descriptive words, past tense, and words that show the order of events are used in writing a recount text. However, based on the observation in SMPN 39 Semarang, most of the students faced any problems in writing, especially in writing a recount text. The main problem was they did not know what they had to write. Also, they felt difficult to organize their idea into a good text. Moreover, most of the students frequently made errors in grammatical use, punctuation, conjunction, mechanics, and spelling. As the solution of these problems, any different techniques should be applied in teaching writing. Clustering is one of many techniques that may be useful in writing. Clustering is one of prewriting techniques that can be used by a writer to produce ideas (Oshima and Hogue 1997). It helps the students to generate any ideas which come to students’ thinking before they start writing. According to Noel (2005), there are a number of steps of using clustering technique. First, the writer can put a word, phrase, or sentence in a circle in the center of a blank page. Then, the writer can put every new idea in smaller circles around the first circle. Lastly, it is ended by drawing a line to another circle to its relationship to a previous idea. Another technique that may be useful in writing is peer assessment. Roberts (2006) defined peer assessment as a process of having students critically reflect upon, and perhaps suggest grades or comments for, the learning of their peers. According to Brown (2004), it offers some benefits, such as direct involvement of students in the learning process, the encouragement of autonomy, and the increase in motivation. However, peer assessment also gives some important drawbacks, such as subjectivity and the ability to discern their own errors. Trisanti (2013) explained some general issues that should be considered in using peer assessment. First, the teacher should explain to the students by giving a copy of form used for assessment and evaluation. Second, the teacher should give an opportunity to the students to practice assessing other group members in which their assessment do not affect project scores. Third, the teacher should provide feedback in order to improve the students’ performance. Based on the background above, this research is aimed at investigating the effectiveness of using clustering technique and peer assessment for teaching writing recount text compared to three- phase technique and teacher assessment and describing the students’ perception to the lesson. In order to achieve the objectives of this research, a number of theories and procedures are used. E. Yunita Widyawati & N. Trisanti / Journal of English Language Teaching 6 (1) (2017) 39 METHODOLOGY OF THE RESEARCH The research design used in the study was Matching-Only Pre-test-Post-test Control Group Design by Fraenkel and Wallen (2008: 271). The design is presented as the following: Group Pre-test Treatment Post-test Experimental O1 X O2 Control O3 C O4 Table 2.1 Pre-test Post-test Control Group Design From table 1, it can be described that the subjects are divided into the experimental and control groups. The basic quality of the subjects is first checked by giving pre-test to both groups. Then, the two groups are taught the same topic, but they are treated by using different techniques of teaching. Post-test is given after the treatment and the result can be computed and analyzed. The subjects of the study were the eighth-grade students of SMPN 39 Semarang in the academic year of 2016/2017 with 274 students. The samples were VIII C, as the control group, consisted of 30 students and VIII D, as the experimental group, consisted of 30 students. The control group was treated by using three-phase technique and teacher assessment, while the experimental group was treated by using clustering technique and peer assessment. In collecting the data, two kinds of instrument were used. The first instrument was writing test which was divided into try-out, pre-test, and post-test. The writing test was given to investigate the effectiveness of using clustering technique and peer assessment for teaching writing recount text to junior high school students compared to three-phase technique and teacher assessment. Another instrument, questionnaire, was distributed to describe the students’ perception of using clustering technique and peer assessment for teaching writing recount text. Questionnaire design used in the study was closed-ended questions consisting of ten statements. A number of steps were done in order to achieve the objectives of the study. Before conducting the research, the try-out test was conducted to see the reliability and validity of the test. It was needed because a good instrument should be reliable and valid. After calculating and analyzing the reliability and validity of the instrument, the pre-test was given to the experimental and control groups. After scoring the pre-test of the students and analyzing them, the same material about recount text was given to both groups. However, they were taught by using the different teaching techniques. The experimental group students were taught by using clustering technique and peer assessment, whereas the control group was treated by using three-phase technique and teacher assessment. The treatments were given in four meetings. After completing the treatments, the post-test was given to both groups. In addition, the questionnaire was distributed to the students of the experimental group to see the students’ perception of using clustering technique and peer assessment for teaching writing recount text. After calculating and analyzing the post-test of the students, the t-test was calculated to see the significant improvement of the students in writing recount text. In addition, the students’ responses to each statement were computed and interpreted to describe the students’ perception of using clustering and peer assessment techniques for teaching writing recount text. In order to find out the validity of the data, this study used methods triangulation. According to Hales (2010:14), methods triangulation is the use of multiple methods to study a situation or phenomenon. Some methods in collecting the data, such as observation, test, and questionnaire were used. These multiple methods were used to make the methods completed each other. E. Yunita Widyawati & N. Trisanti / Journal of English Language Teaching 6 (1) (2017) 40 RESULT AND ANALYSIS In this part, the result and analyses of try-out, pre-test, post-test, and questionnaire are provided. Moreover, the discussion of the research findings is presented clearly at the end of this part. Try-out The try-out was conducted to find out the validity and reliability of the instruments. It was conducted in VIII B with 29 students who took the try-out test. To find out the validity of the test, the writer used content validity. The test was compared with the materials dealing with the curriculum requirement. The School-Based Curriculum states that the-eighth grade students are supposed to write a simple recount text. In reveals that the test is valid since the test is compatible with the curriculum requirement. Another characteristic of a good test is reliable. Since the instrument was writing test, the writer used inter-rater reliability. According to Nunnaly (1960) in Ghozali (2006), an instrument is considered as reliable if the value of Cronbach’s Alpha is higher than 0.6 (>0.6). The computation of inter-rater reliability of the test using SPSS 21.0 program is presented as follows: Reliability Statistics Cronbach's Alpha Cronbach's Alpha Based on Standardized Items N of Items ,770 ,776 2 Table 3.1 The Reliability Analysis of the Test From table 2, it can be seen that the value of Cronbach’s Alpha is 0.77. It means that the test is reliable since the value of Cronbach’s Alpha (0.77) is higher than 0.6. In addition, the reliability of peer assessment rubric used as an instrument in this research was also checked. The calculation of inter-rater reliability of peer assessment rubric used as instrument is presented in the following: Reliability Statistics Cronbach's Alpha Cronbach's Alpha Based on Standardized Items N of Items ,830 ,835 2 Table 2.2 The Reliability Analysis of the Instrument (Peer Assessment Rubric) Table 3 shows that the value of Cronbach’s Alpha is 0.83. It can be concluded that peer assessment rubric used an instrument in this research is reliable since the value of Cronbach’s Alpha is higher than 0.6 (0.83>0.6). From the overall analysis above, the instruments can be used in the real research since they fulfill the requirements of a good instrument Pre-test Pre-test was given to the experimental and control groups in order to know the basic ability of the students in writing recount text. From the pre-test result, the mean scores of the students in both groups were analyzed. The mean score of the control group students was 66.73, while the experimental group students was 66.97. It could be concluded that the experimental and control group students had almost equal achievement in writing recount text before getting the treatment. E. Yunita Widyawati & N. Trisanti / Journal of English Language Teaching 6 (1) (2017) 41 However, the homogeneity and normality of the test should be computed to support the statement. The homogeneity of the pre-test is computed in order to know whether the data of each group are homogeneous. The data are homogeneous if the significance value is higher than the significance level (0.05) (Ghozali 2006). The pre-test homogeneity computation is presented as the following: Test of Homogeneity of Variances Pretest Levene Statistic df1 df2 Sig. ,000 1 58 ,989 Table 3.3 The Homogeneity Test of the Experimental and Control Groups Pre-test From table 4, it can be seen that the significance value (0.99) is higher than 0.05. It reveals that the pre-test data are homogeneous. In other words, the population of the control and experimental groups has the same achievement before receiving the treatment. The normality of the pre-test also should be analyzed to find out whether the data are normal or not. According to Ghozali (2006), the data are considered normally distributed if the value of Asym.Sig.(2-tailed) of Kolmogorov-sminorv test is higher than the significance level (0.05). The computation of normality test using Kolmogorov-sminorv test in SPSS 21.0 program is presented below: One-Sample Kolmogorov-Smirnov Test Experimental Group Pretest Control Group Pretest N 30 30 Normal Parametersa,b Mean 66,97 66,73 Std. Deviation 5,881 6,209 Most Extreme Differences Absolute ,099 ,099 Positive ,099 ,099 Negative -,081 -,090 Kolmogorov-Smirnov Z ,540 ,544 Asymp. Sig. (2-tailed) ,932 ,928 Table 3.4 The Normality Test of the Experimental and Control Groups Pre-test From table 5, it can be seen that the result of One-Sample Kolmogorov-Smirnov Test (Asymp.Sig.2-tailed) of experimental group pre-test is 0.93. It indicates that the data of the experimental group pre-test are normally distributed since the value of Asymp.Sig.(2-tailed) (0.93) is higher than 0.05. In addition, the value of Asymp.Sig.(2-tailed) of control group pre-test is 0.93. It reveals that the data of the control group pre-test are also normal because the Asymp.Sig.(2-tailed) value is higher than 0.05 (0.93>0.05). From the overall analysis above, the treatments can be given to the experimental and control groups since the data of pre-test are normally distributed and homogeneous. Treatments The different treatments were given to the control and experimental groups. The treatment using clustering technique and peer assessment was given to the experimental group, while the E. Yunita Widyawati & N. Trisanti / Journal of English Language Teaching 6 (1) (2017) 42 control group was treated by using three-phase technique and teacher assessment. The treatments were given in four meetings. At the first meeting, the material about recount text was explained to both groups, and they did the same activity and task. At the second meeting, both groups wrote a recount text in groups. At the third meeting, they did the activity in pairs and at the fourth meeting, they did it individually. Post-test Post-test was conducted to both groups after completing all the activities above to measure the students’ achievement after getting the treatment. Based on the post-test result, the mean score of the experimental group was 80.33, while the control group was 74.17. The result indicates that the students’ improvement of the experimental group is higher than the control group. However, the t-test should be computed as a statistical evidence to prove the statement. Before calculating the t-test, the homogeneity and normality of the post-test should be calculated and analyzed first to know whether the data are homogeneous and normal or not. The result of homogeneity post-test computation is presented as follows: Test of Homogeneity of Variances Levene Statistic df1 df2 Sig. ,000 1 58 ,989 Table 3.5 The Homogeneity Test of Post-test From table 6, it can be seen that the significance value (0.99) is higher than 0.05. It reveals that the post-test scores have the same variance. It can be concluded that the population of the control and experimental groups is homogeneous. The normality of post-test also should be computed first in order to see whether the data are normal or not. The normality test of post-test is provided below: One-Sample Kolmogorov-Smirnov Test Experimental Group Posttest Control Group Posttest N 30 30 Normal Parametersa,b Mean 80,33 74,17 Std. Deviation 5,313 5,760 Most Extreme Differences Absolute ,143 ,122 Positive ,126 ,089 Negative -,143 -,122 Kolmogorov-Smirnov Z ,786 ,667 Asymp. Sig. (2-tailed) ,568 ,765 Table 3.6 The Normality Test of the Experimental and Control Groups Post-test Table 5 shows that the value of (Asymp. Sig. 2-tailed) of the experimental group post-test is 0.57. It indicates that the data of the experimental group post-test are normally distributed because the significance value (0.57) is higher than 0.05. Moreover, the Asymp.Sig.(2-tailed) value of the control group post-test (0.77) is also higher than 0.05. It reveals that the data of the control group post-test are also normal. The Mean Scores Difference of Pre-test and Post-test E. Yunita Widyawati & N. Trisanti / Journal of English Language Teaching 6 (1) (2017) 43 The improvement of the experimental and control groups in writing recount text can be seen through the mean scores difference of pre-test and post-test which is presented in the following chart: Diagram 3.1 The Mean Scores Difference of Pre-test and Post-test Chart 1 shows that the mean score of the experimental group pre-test is 66.97. Meanwhile, the mean score of the post-test is 80.33. It reveals that there is significant improvement of the pre-test and the post-test scores achieved by the experimental group students. On the other hand, the writing achievement of the control group students also improves. It is 66.73 for the pre-test and 74.17 for the post-test. In indicates that there is less improvement than the experimental group. The difference of the experimental group mean scores (13.36) is higher than the control group (7.44). It can be concluded that there is a better improvement of the experimental group’s writing achievement after receiving the treatments by using clustering technique and peer assessment. T-test Analysis The purpose of calculating the t-test is to examine the hypotheses of this study. The improvement of the experimental and control groups is crucial to be tested as a quantitative evidence to find out whether the two groups have a significant difference in writing recount text. The result of t-test is presented as the following: Group Statistics Class N Mean Std. Deviation Std. Error Mean Posttest Experimental Group 30 80,33 5,313 ,970 Control Group 30 74,17 5,760 1,052 Levene's Test for Equality of Variances t-test for Equality of Means F Sig. t df Sig. (2- tailed) Mean Difference Std. Error Difference 95% Confidence Interval of the Difference Lower Upper 66.73 66.97 74.17 80.33 0 10 20 30 40 50 60 70 80 90 Control Group Experimental Group The Mean Scores of Pre-test and Post-test Pre-test Post-test E. Yunita Widyawati & N. Trisanti / Journal of English Language Teaching 6 (1) (2017) 44 Posttest Equal variances assumed ,00 ,99 4,31 58 ,00 6,17 1,43 3,30 9,03 Equal variances not assumed 4,31 57,63 ,00 6,17 1,43 3,30 9,03 Table 3.7 Independent Sample T-test of the Experimental and Control Groups Post-test Independent Samples Test Based on the independent sample t-test result above, the test is significant, t(58)=4.31, p<0.05, d=95. It means that the t-value from 58 degree of freedom is 4.31. The probability of obtaining t- value is 0.5 and the effect size is 0.95. The 95% confidence interval for the average percentage of post-test score ranges from 3.30 to 9.03. An examination of the group mean scores indicates that the post-test score of the experimental group (M=80.33, SD=5.31) is significantly higher than the control group (M=74.17, SD=5.76). In sum, after analyzing the result of t-test calculation, the use of clustering technique and peer assessment is more effective for teaching writing recount text to the eighth-grade students of a junior high school than three-phase technique and teacher assessment. The students’ achievement in writing recount text significantly improves after getting the treatment by using clustering technique and peer assessment. Questionnaire Result After conducting post-test, the questionnaire was given to the experimental group. The purpose of giving the questionnaire is to see the students’ perception of using clustering technique and peer assessment for teaching writing recount text. The result of the calculation of the students’ responses to each statement in the questionnaire can be seen in the table below: No Pernyataan Jawaban 1 2 3 4 5 1 Saya menyukai pelajaran bahasa Inggris. 0% 0% 30% 40% 30% 2 Saya menyukai pelajaran menulis dalam bahasa Inggris (writing). 0% 3% 30% 57% 10% 3 Teknik clustering sangat mudah digunakan oleh siswa. 0% 0% 17% 63% 20% 4 Teknik clustering membantu saya untuk mengembangkan ide-ide. 0% 3% 10% 70% 17% 5 Dengan menggunakan teknik clustering, saya dapat mengembangkan ide-ide menjadi sebuah paragraf dengan lebih mudah. 0% 3% 10% 63% 23% 6 Saya menyukai teknik peer assessment dalam pembelajaran menulis teks recount. 0% 10% 33% 53% 3% 7 Suasana kelas menjadi menyenangkan setelah diberikan pembelajaran dengan menggunakan teknik peer assessment. 0% 7% 33% 60% 0% 8 Penggunaan teknik peer assessment dapat meningkatkan pemahaman bagaimana cara 0% 3% 3% 80% 13% E. Yunita Widyawati & N. Trisanti / Journal of English Language Teaching 6 (1) (2017) 45 menulis yang baik. 9 Penggunaan teknik peer assessment membantu saya dalam memperbaiki penulisan teks recount menjadi lebih baik. 0% 7% 7% 67% 20% 10 Penggunaan teknik clustering dan peer assessment dapat meningkatkan kemampuan menulis saya. 0% 7% 3% 57% 33% Table 3.8 The Result of the Questionnaire From the table above, the calculation result of the students’ responses to each statement in the questionnaire can be drawn as the following chart: Diagram 3.2 The Result of the Questionnaire Based on the table 9 and chart 2, the result of the questionnaire can be interpreted. The first statement is intended to know whether the students are interested in English or not. From the result, it indicates that most of the students are interested in English. However, there are some students that do not quite agree about it. The second statement in the questionnaire is to see the students’ interest in writing. From the result, it indicates that most of the students like writing. Nevertheless, there are some students that do not quite agree and few students disagree about the statement. The third statement in the questionnaire is intended to find out whether clustering technique is easy to be applied by the students or not. From the table and chart, it can be concluded that clustering technique is easy to be applied by almost all students. Nonetheless, there are a few students that do not totally understand how to apply clustering technique in writing. The fourth statement in the questionnaire is to see whether clustering technique helps the students to produce and explore their ideas or not. The result shows that most of the students agree that clustering technique helps them so much to produce and explore their ideas. By using clustering technique, they can write any ideas come into their mind before writing a draft. However, there are some students that do not quite agree and disagree that the technique helps them to produce and explore their ideas. The fifth statement in the questionnaire is intended to find out whether clustering technique facilitates the students to develop their ideas into a recount text. From the result, it reveals that most of the students agree that clustering technique helps them to develop their ideas into a recount text 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% Strongly Disagree Disagree Not quite agree Agree Strongly agree Questionnaire Findings Question 1 Question 2 Question 3 Question 4 Question 5 Question 6 Question 7 Question 8 Question 9 Question 10 E. Yunita Widyawati & N. Trisanti / Journal of English Language Teaching 6 (1) (2017) 46 easily. Clustering technique helps them in developing words or phrases into sentences. Then, it helps for developing some sentences into some paragraphs and a text. However, some students argue that clustering technique does not facilitate them to develop their ideas into a recount text easily. The sixth statement in the questionnaire is to describe the students’ opinion of using peer assessment for teaching writing recount text. The result shows that more than a half of the number of the students like using peer assessment technique for teaching writing recount text. Nonetheless, almost a half of the number of the students argue that they do not quite agree and disagree with the statement. They feel that they are less interested in peer assessment technique because it is not easy to do. Also, they sometimes feel reluctant when giving assessment and suggestion to their peers. The seventh statement in the questionnaire is to know the class condition when using peer assessment technique. By seeing the result of the questionnaire, more than a half of the number of the students argue that teaching writing recount text using peer assessment is enjoyable because they have time to discuss with their classmates to evaluate their peers’ work. They can read it and then give assessment and suggestions. However, almost a half of the number of the students do not quite agree and disagree with the statement. They think that using peer assessment in writing recount text is challenging because they do not have good knowledge in writing. The eighth statement in the questionnaire is intended to find out whether peer assessment improves the students’ understanding and knowledge of a good writing or not. The result reveals that more than three-quarter of the students agree that their understanding of a good writing improves after they receive the treatment by using peer assessment. They can learn from the assessment and suggestions given by their friends. Nevertheless, few students are not sure that their understanding of a good writing improves after peer assessment is implemented for teaching writing recount text. The ninth statement in the questionnaire is to find out whether peer assessment technique facilitates the students to revise and edit their own writing. From the calculation result, it indicates that most of the students agree that peer assessment technique facilitates them to revise and edit their writing. They may be aware of the mistakes and errors they make so that they can revise their writing into a better result. Nonetheless, some students do not quite agree and disagree with the statement. The last statement in the questionnaire is to see whether the use of clustering technique and peer assessment improves the students’ achievement in writing recount text or not. From the students’ responses to the item, it indicates that the use of clustering technique and peer assessment significantly improves the students’ ability in writing recount text. Each component of writing improves after receiving the treatment by using clustering technique and peer assessment. However, there are few students argue that the use of clustering technique and peer assessment does not improve their achievement in writing recount text. Discussion of Research Findings The first objective of this study was to investigate the effectiveness of using clustering technique and peer assessment for teaching writing recount to junior high school students compared to three-phase technique and teacher assessment. The research findings revealed that the use of clustering technique and peer assessment was more effective for teaching writing recount text to junior high school students than three-phase technique and teacher assessment. It was proven by the result of pre-test, post-test and t-test. The mean score of the experimental students in pre-test was 66.97, while the control group was 66.73. It indicated that both groups had almost the same achievement in writing recount text before the treatment given. After getting the treatment, the mean scores of the two groups improved. The mean score of the experimental group was 80.33, while the control group was 74.17. The data proved that the experimental group achieved a better result than the control group after receiving the treatment by using clustering technique and peer assessment for writing a recount text. E. Yunita Widyawati & N. Trisanti / Journal of English Language Teaching 6 (1) (2017) 47 Moreover, the t-test was used to prove the statement. Based on independent sample t-test, the t-value was 4.31. It revealed that the t-value (4.31) was higher than t-table (2.00). In addition, the significance value was lower than 0.05. It indicated that there was a significant difference in writing recount text between the experimental and control groups. Hence, the working hypothesis (Ha) which stated that “the use of clustering technique and peer assessment is effective for teaching writing recount text to the eighth-grade students of SMPN 39 Semarang in the academic year of 2016/2017” was accepted and the null hypothesis (Ho) which stated that “the use of clustering technique and peer assessment is not effective for teaching writing recount text to the eighth-grade students of SMPN 39 Semarang in the academic year of 2016/2017” was rejected. Second, the study was aimed at describing the students’ perception of using clustering technique and peer assessment for teaching writing recount text. Based on the result of the questionnaire, most of the students had positive perception to the use of clustering technique and peer assessment for teaching writing recount text to junior high school students. From the calculation, 70% of the students were interested in English and 67% of the students were interested in writing. However, there were few students that stated that they did not quite agree with these statements. In addition, most of the students had a positive response to the use of clustering technique. From the questionnaire finding, 83% of the students agreed that clustering technique was easy to be applied by the students. 87 % of the students agreed that it helped them much more to produce and explore their ideas. Moreover, it facilitated 86 % of the students to develop the ideas into a recount text easily. By using this technique, the content and organization of the text improved. However, there were some students argued that the use of clustering technique did not help the students to explore ideas and develop the ideas into a recount text easily. Moreover, even though some students argued that they were not interested in using peer assessment because it was difficult, 93% of the students agreed that the use of peer assessment improved their understanding and knowledge of a good writing. They could learn from the correction, assessment and suggestions given by their peers so that their understanding of correct grammar, punctuation, capitalization and spelling improved. It also facilitated 87% of the students to revise and edit their own writing. They could rewrite their writing based on the peer assessment rubric so that the result was better than the first writing. In overall, 90% of the students agreed that the use of clustering technique and peer assessment improved the students’ achievement in writing recount text. However, there were still some drawbacks when using clustering technique and peer assessment for teaching writing text to junior high school students. The students needed much time when using clustering technique since they started by writing some words or phrases, and then they organized them into some sentences. Then, they developed the sentences into some paragraphs. Besides, peer assessment could be applied successfully when the students had a good background knowledge so that it also took much time to prepare the students. Subjectivity was also a problem when using peer assessment. Some students felt reluctant when giving low scores to their peers’ work. Triangulation Result The result of the test indicated that the experimental group students, which were treated by using clustering technique and peer assessment, achieved a better improvement than the control group, which was taught by using three-phase technique and teacher assessment. It was supported by the result of questionnaire in which most students agreed that their writing achievement significantly improved after they were taught by using clustering technique and peer assessment. Moreover, when the researcher observed the class activities and situation, most of the students were E. Yunita Widyawati & N. Trisanti / Journal of English Language Teaching 6 (1) (2017) 48 enthusiastic during the lesson since the techniques were easy, helpful, and pleasing. From the overall explanation, it could be concluded that the data were valid, and they supported one another. CONCLUSIONS After scrutinizing the data, the writer concludes that she found 146 errors in all derivational categories. The errors were classified based on the four types of Surface Structure Taxonomy as proposed by Dulay, Burt and Krashen (cited in James, 1998:106). They are omission, addition, misordering and misformation. The research findings indicated that the most frequent errors occurred in misformation with 98 (67,1%) errors. It was followed by omission ( 34 errors or 23,3%), and addition ( 14 errors or 9,6%). Out of the total errors (146), it showed that the majority of errors occurred in deriving noun to noun and verb to adjective with 30 errors. It was followed by 25 errors in deriving verb to verb, 23 errors in deriving adjective to adjective, 11 errors in deriving adjective to adverb, 10 errors in deriving verb to noun, 9 errors in deriving noun to verb, 5 errors in driving noun to adjective and 3 errors in deriving adjective to noun. This is relevant to the questionnaire result which showed the highest percentage (60 % or 15 students) found most of students agreed they found difficulty in deriving noun to noun. It was followed by 12 students or 48% stated that they found difficulty in deriving verb to adjective. It proved that both the test and questionnaire result indicated the most frequent errors occurred in deriving noun to noun. REFERENCES Akmajian, A., Demers, R.A., and Harnish, R.M. (1984). Linguistics : An Introduction to Language and Communication. Cambridge : The MIT Presss. Ellis, Rod. (1997). Second Language Acquisition. Oxford : Oxford University Press. Haspelmath, M. and Sims, A.D. (2010). Understanding Morphology. London : Hodder Education. [Online.] Available at http://morphology.xyz/pdf/HaspelmathSims2010.pdf [accessed 1/10/2016]. Fromkin, V., Rodman, R., and Hyams, N. (2011). An Introduction to Language. Wadsworth : Cengage Learning. James, Carl. (1998). Errors in Language Learning and Use. Exploring Error Analysis. USA : Longman. Katamba, Francis. (1993). Morphology. New York: St. Martin’s Press. Kothari, C. R. (2004). Research Methodology, Methods, and Techniques (Second Revised Edition). New Delhi : New Age International. Priyatmojo, A. S. (2011). Political Discourse : Obama’s Appraisal Attitude. LANGUAGE CIRCLE Journal of Language and Literature Vol. VI/. Available at : http://journal.unnes.ac.id/nju/index.php/LC/article/view/2043 [accessed 08/15/16]