統計的フィードバックによる英語学習の促進

全文

(1)Title. 統計的フィードバックによる英語学習の促進. Author(s). マシオン, スチュアート. Citation. 釧路論集 : 北海道教育大学釧路校研究紀要, 第44号: 89-93. Issue Date. 2012-12. URL. http://s-ir.sap.hokkyodai.ac.jp/dspace/handle/123456789/6862. Rights. Hokkaido University of Education.

(2) 釧路論集－北海道教育大学釧路校研究紀要－第44号（平成24年度） Kushiro Ronshu, - Journal of Hokkaido University of Education at Kushiro - No.44(2012):89-93. 統計的フィードバックによる英語学習の促進マシオン、スチュアート北海道教育大学釧路校英語科教育研究室外国人教師. Reinforcing English Learning with Statistical Feedback MASSION, Stuart Department of English Education, Kushiro Campus, Hokkaido University of Education. 概要本研究は、2011年度後期科目の中級実践英語Ⅱを受講した学部生22名を対象に実施された。目的は、提出された英語ライティングに対する統計的分析結果を毎回フィードバックすることにより、英語誤用頻度が低減するか検討することであった。方法は、(1)まず、最初のライティングにおいて受講生の英語誤用を分析した。その結果、冠詞の欠落および動詞の誤用が顕著な誤りとして見いだされた。また、スペリング間違い、語彙や前置詞などの誤用も示された。(2)次に、毎週提出されたライティングは、教師が訂正しコメントも加えて返却された。また、総語数ランキングと英語誤用比率を表す分析結果も配布された。(3)加えて、各授業の最初に、同分析結果に関するディスカッションがなされた。(4)さらに、実験的に1度のみであったが、英語誤用に関するリストづくりが学生によってなされた。以上から得られたデータを分析した結果、受講生の英語誤用の頻度がわずかではあるが低減したことが認められた。 ABSTRACT English writing by EFL second year university students was analyzed to identify typical errors in written English. The predominant errors were found to be omission of determiners and incorrect verb usage. Incorrect spelling, vocabulary, preposition use, plural noun use, capital letters and comma use occurred less frequently. The focus of this paper is whether students can be encouraged to reduce these errors by seeing statistics on the frequency of their errors. Weekly writing assignments over the course of one semester were corrected and returned to the students with teacher commentary along with simple bar charts showing error ratios and word counts. The word count of each student was given and ranked from highest to lowest to encourage more writing output. Error counts were always given for the class group as a whole, although during one week, as an experiment, errors were listed by student. At the beginning of each class, this statistical analysis, along with error samples taken from student writing, were discussed. A slight reduction in the frequency of English errors was observed during the course of the semester. As a native English speaking teacher at a Japanese. students very insecure in their conversational ability is. university, the author has concentrated his first. often cited. Are they more confident in their writing?. year ESL courses on enhancing student English. Japanese students prefer grammatical exercises. conversational ability. Using speaking and listening. to information gap activities, according to Ryan. textbooks, strategies like dialogue practice, information. Richardson in his very interesting comparison of. gap activities and group tasks were used to encourage. Japanese student and teacher perceptions. 1 Has an. English conversation.. emphasis on improving English speaking ability at. It is perhaps not surprising that this conversational. the university perhaps obscured very good abilities in. approach has met with some resistance. The cliche that. English reading and writing? Given the heavy emphasis. six years of the grammar translation method at the. on grammar translation methods prior to arriving at. junior high and high school levels has left Japanese. university, one might expect student writing accuracy to. － 89 －.

(3) MASSION, Stuart exceed their conversational accuracy.. required by the Hokkaido University of Education to. Written conversational dialogues submitted in the. take the spring semester Intermediate Practical English. first year Practical English courses did not bear this. class. There were 42 students in the 2011 spring. out. Students were very good at using the suggested. semester class.. vocabulary and grammar structures, but the rest. The class examined in this paper is the 2011 fall. of their writing revealed many elementary English. semester class, consisting of 22 second year students.. mistakes. How could this happen? Stewart Wachs feels. These were all highly motivated students who had. the grammar translation method usually involves much. decided to take this non-compulsory Intermediate. translation from Japanese to English and vice versa,. Practical English course. At the beginning of this class,. but very little actual communication in written English.. they had already taken three semesters of Practical English courses in addition to two semesters of. It is no secret that most Japanese students spend six. compulsory General English courses and one English. years in junior and senior high English classes hardly. Communications course in their first year. The first two. ever speaking English. What is commonly overlooked. Practical English courses, however, had emphasized. is that most of these students never write in English. speaking and listening skills and, as mentioned above,. either. Instead, they translate words, phrases,. the third course taken in the spring semester of the. sentences, or passages from Japanese to English and. current year had a reading and writing emphasis. sometimes the other way around. By the time these. similar to the course described in this paper.. young people take seats in your writing class they. In that spring semester Intermediate Practical. may have notched another year or two of English. English class, it was observed that students would. study at a university, but chances are they have. repeat the same written errors again and again. rarely, if ever, communicated in written English.. (“fossilization”), despite frequent written feedback. 2. from the instructor and many grammar reviews and Perhaps second year EFL students exhibit poor. exercises. Students seemed almost unaware of any need. written English because of the before-mentioned. to correct their mistakes as the semester progressed.. emphasis on encouraging conversational fluency, with. It must be stressed that no statistical analysis of word. grammatical accuracy often a secondary concern.. count or errors types was given in that course.. Constant correction of speaking errors is usually. This constant recurrence of the same types of errors. considered to be detrimental to building student. in the student writings, despite one semester with a. self-confidence in speaking a foreign language.. 3. heavy written grammar emphasis, prompted the idea. Unfortunately, grammatical errors in writing cannot be. of giving the students a statistical analysis of class. hidden or overlooked as easily as they might be during. total errors by type to see whether this might provide. a conversation.. an incentive to be more careful while writing. This. Is there even a need in this age of telecommunications. weekly statistical analysis might introduce a group. for accurate written English ability? Although audio-. pressure element which, it was hoped, would lead to a. visual telecommunications software like Skype. reduction in common errors as the students could watch. is becoming increasingly popular, users are often. the weekly class errors go up or down on a bar chart.. frustrated by very differing time zones. It’s difficult. This statistical analysis and its effectiveness will be. to “skype” someone who may be asleep in New York. discussed after a few words first about the Word Count. City while one is in the middle of one’s workday. It is. Bar Chart.. therefore very likely that during their professional and personal lives future teachers will still need to. Word Count Bar Chart. communicate with foreign English speakers not just in face-to-face meetings, but via written research papers,. There was a concern that too much error correction. reports, blogs, and emails.. might have a dampening effect on student writing. To address this deficiency, the author decided to. output, so it was also decided to report their essay. design first and second semester Intermediate Practical. word totals to them on a Student Word Count Bar. English courses for second year ESL university students. Chart (an example bar chart for week 2 is shown on. which would concentrate on improving English reading. the next page). Word counts were tallied for each. and writing. All students wishing to teach English are. essay and entered in a spreadsheet next to the essay. － 90 －.

(4) Reinforcing English Learning with Statistical Feedback author’s name. This table was then sorted from highest. weekly assignment 9.. word count to lowest word count. A bar chart was then. The careful reader may have noticed that the author. generated using a popular spreadsheet software.. has omitted plotting the embarrassing result for assignment number 9 (an average word count of only 65). Why was there such a sudden drop-off? After the eighth weekly writing assignment, students were given, in addition to the usual Student Word Count Bar Chart and Error Type Bar Chart, a spreadsheet listing their names and the number of errors of each type they had committed. It is the guess of the author that this caused some embarrassment to the students, and they became very risk-averse in their next writing assignment, writing less and often copying entire English sentences from the textbook, rather than using their own words. To eliminate bias in the statistics, these copied sentences were not included in the word counts. The new spreadsheet reporting student names and the errors they committed was thereafter discontinued, and the average word count jumped back into the 120 to 140 word zone in the next week.. It would appear from the next scatter chart below, which has added a trend line to the average word. Total Class Errors by Type Bar Chart. counts over the semester, that this Student Word Count Bar Chart had its desired effect. The trend line (all line. The weekly statistical analysis had the title Total. fitting in this paper was generated using spreadsheet. Class Errors by Type. Each essay was carefully analyzed. functions on class data) has a positive slope of 0.982.. for errors by type and number. The individual totals were input into a spreadsheet, totaled by error type, and this total was then converted into a ratio of number of errors per 1000 words written because students were sometimes absent from class and the comparability of the weekly statistics had to be maintained. The following figure shows the final Total Class Errors by Type Bar Chart with results for the whole semester.. Initially over the first four weekly assignments, there was a definite upward trend in the class average word count. Average word counts increased as many students appeared to strive to get their names at the top of the Student Word Count Bar Chart. Gradually, the thrill of this competition wore off, and the average word count then began to settle in between 120 and 140 until. － 91 －.

(5) MASSION, Stuart There are eleven bars above each error category. showing them as scatter points connected by lines gives. illustrating the progression of the error ratios over. us the line chart below.. the course of the semester. Whenever the weekly error. At first glance, little progress seems to be occurring. analysis revealed a spike in a type of error (like in week. in any of these error types as the weekly assignments. 5 for determiner errors), the class was given a remedial. progress from left to right. There is a peak in. grammar review on the same topic at the beginning of. determiner errors at weekly assignment 5, a peak for. the next class. About 15 minutes at the start of class. verb tense errors at weekly assignment 7, and the other. were devoted to this review of errors revealed by the. verb errors peak at weekly assignment 10. Determiner. statistical analysis. Teacher and student interaction. errors decrease steadily after weekly assignment 5 until. then continued in the next 45 to 50 minutes of the class,. weekly assignment 9, and then they shoot up again.. using a reading and writing textbook appropriate to. A discussion of reasons for these peaks follows below.. second year university level students. Each unit of this. But first, if we derive trend lines for these plot points,. textbook introduced a new theme, along with needed. we get two negatively sloped trend lines for determiner. sample readings, vocabulary and grammar review.. (-0.5273) and verb tense errors (-.4364) and a positively. Students were given the remaining classroom time. sloped trend line for other verb errors (0.6).. (usually 30 minutes) to write a 100 to 200 word essay in their notebooks. The teacher would walk around the class as the students were writing and provide encouragement and any answers to the questions the students might have had. The journal notebooks were collected at the end of class. Before the meeting of the next class, each student essay was corrected for English mistakes. It was also a welcome opportunity for the teacher to interact with the student on an individual level through commentary on student essay content. Students were instructed to leave a page blank next to their essay for these teacher explanations, examples and comments. All teacher corrections were written in simple English, with occasional Japanese grammar terms sprinkled in, and examples of correct English usage were given. This approach reduced the burden on the students who thereby had no homework (unless they didn’t finish their essay in class). The increased burden fell on the teacher who now had to spend time outside of class on error correction and preparation of the statistical analysis. Copies of the word count and error type bar charts were inserted into the student notebooks and, as stated above, were then returned to the students at the beginning of the next class. Effectiveness of Statistical Reporting to Students This statistical approach made students aware of the type and frequency with which English errors were occurring as the weeks passed. The Total Class Errors by Type Bar Chart at the bottom of the right hand column on the previous page shows the most frequent error types were determiner, verb tense, and other verb errors. Isolating just these three error types and. － 92 －.

(6) Reinforcing English Learning with Statistical Feedback There is no clear explanation for the peak in. tendency to capitalize the first word after a comma as if. determiner errors at weekly assignment 5. A, an, the. a new sentence had begun.. and a host of other appropriate determiners are the bane of all EFL learners whose native languages do. Conclusions. not include this feature. Considerable time in this intermediate class was devoted to emphasizing the. This one semester experiment identified two areas. importance of determiner usage with singular countable. needing special attention in EFL education. Omission. nouns. The students were given handouts with. of determiners and verb usage errors centering. determiner rules and exercises on several occasions.. on improper tense were the most frequent writing. Although the improvement was not dramatic, a slight. mistakes. Providing statistical information on error. improvement in determiner accuracy occurred over the. occurrence in weekly writing assignments seemed to. course of the semester.. exert some downward pressure on the frequency of. Verb tense errors shot up in week 7 because of a. errors. Students generally reacted positively to the. writing assignment which required the students to hear. Word Count Bar Chart which introduced a competitive. a short story in Japanese and then write it in their own. element that increased writing production. However,. words in English. Many students vacillated between the. the students did not react well to having their errors. present and past tenses to tell the story, although they. exposed to the entire class. This led them to write less. had been warned to use the past tense. One student. and to write very conservatively. Assembling the weekly. made an attempt to tell the story in reported speech. statistics was a considerable burden for the teacher.. which is even more difficult to do than direct speech.. Twenty students are about the upper limit for such. The Japanese language has far fewer tenses than. a method which requires extensive teacher time and. English, but it has a very clear past tense which should. input. The frequency of determiner and verb usage. sensitize students to the importance of correct verb. errors reveals two areas where more must be done at. tense.. the junior high, high school, and university levels of. “Other verb errors” were a collection of many non-. English instruction.. tense errors, but the predominant error in this category was the omission or improper use of the third person. Notes and references:. singular “s” in the present tense. Failure to use a verb or using lone participles as a verb occurred less frequently.. 1. This “other verb error” category actually increased over. Views on Effective Language Teaching, OnCUE Journal,. the course of the semester. However, the error number. Spring 2011, Volume 4, Issue 3, page 243. remained relatively consistent between 5 and 15 over. 2. the period until weekly assignment 10 which had a peak. University of Foreign Studies, A Handbook for Teaching. of 21. Replacing this outlier with the average value 8. English at Japanese Colleges and Universities, edited. Ryan Richardson, Comparing Student and Teacher. Stewart Wachs, Associate Professor of English at Kyoto. causes the slope of the “other verb error” trend line to. by Paul Wadden, Oxford Press, 1993, page 73. be reduced considerably from 0.6 to 0.1273.. 3. Other errors shown in the Total Class Errors by Type. Language Classroom, Cambridge University Press, page. Bar Chart are self-explanatory. Spelling errors do not. 103).. Dick Allwright & Kathleen Bailey, Focus on the. include third person singular “s” in the present tense errors which were designated as verb errors. Vocabulary errors were incorrect usage of English words and idioms. This most often occurred when students used words or idioms suggested by their electronic dictionaries. Incorrect, unnecessary or omitted prepositions were also frequent. Reluctance to use plural forms of countable nouns may again be a unique challenge for the Japanese EFL learner because the Japanese language doesn’t emphasize the distinction between singular and plural as much as the English language. Word capitalization and comma errors also occurred. Noteworthy here was a. － 93 －.

(7)