看圖作文、英語文能力和寫作焦慮對英文寫作之影響

全文

(1)國立臺灣師範大學英語學系碩. 士. 論. 文. Master’s Thesis Graduate Institute of English National Taiwan Normal University. 看圖作文、英語文能力和寫作焦慮對英文寫作之影響. The Effects of Picture Prompts, English Proficiency, and Writing Anxiety on Writing Performance. 指導教授：程. 玉. 秀. Advisor: Dr. Yuh-show Cheng 研究生：郭. 韋. 中華民國一○四年八月 August 2015. 伶.

(2) 摘要. 許多研究探討已知結局的看圖作文，然而，對未知結局讓學生自由發揮寫作能力的研究則佔極少數。本研究主要探討已知結局、未知結局的看圖寫作測驗是否對高中生寫作表現（含流暢度、語法正確度和整體表現）有所影響，並分析上述兩種看圖寫作測驗分別和英語文能力及寫作焦慮的交互作用。受試者為南臺灣 182 位高三學生，先以大考中心 92 年學測英文改編試題成績分成程度相當的兩組，一組依據已知結局的圖示寫作，另一組依據未知結局的圖示寫作。除了看圖寫作外，受試者並填寫背景問卷、外語寫作焦慮量表 (SLWAI)和測後問卷。12 位學生還接受個別訪談。量化分析結果發現：兩種看圖測驗對寫作流暢度、語法準確度、以及整體的寫作表現皆無顯著影響，且看圖測驗的類型與英語文能力或寫作焦慮無顯著交互作用。然而，英語文能力和寫作焦慮各自對作文表現有顯著影響；尤其，英語文能力高或寫作焦慮低者寫作表現較佳。另外，訪談結果發現：高寫作焦慮者偏好寫已知結局之作文，低焦慮者偏好未知結局之作文。測後問卷分析結果則呈現受測者對未知結局抱持著有助於發揮創意的普遍認知。依據本研究發現，我們建議由於已知和未知結局之兩種看圖寫作對寫作表現的影響並無顯著差異，故大型測驗如學測仍可輪流使用。鑑於英語文能力和寫作焦慮各自對作文表現有顯著影響，教師應設法提升學生的英語文能力，並降低其寫作焦慮。關鍵字：寫作測驗、看圖作文、英語文能力、寫作焦慮. i.

(3) Abstract. The present study aims to examine the effect of closed- and open-ended picture prompts on senior high students’ writing performance in terms of overall quality, writing fluency, and accuracy. The interaction between picture prompts and English proficiency or writing anxiety on writing performance was also explored. The sample consists of 180 twelfth graders at a senior high school in southern Taiwan. All participants finished a background questionnaire, an L2 writing anxiety scale (SLWAI), a shortened 2003-GSAT as a placement test, a writing task either with a closed- or an open-ended picture prompt, a post-writing questionnaire. An interview was conducted with 12 participants to probe into their perceptions about the two writing tasks. Statistical analyses showed that picture prompts had no significant effect on students’ writing performance and there was no significant interaction between picture prompts and English proficiency or writing anxiety. On the other hand, English proficiency and L2 writing anxiety each had significant effect on writing performance. The interview results showed that high-anxiety students tended to prefer closed-ended prompts while low-anxiety students favored open-ended prompts. Analysis of post-writing questionnaire showed that there was no significant differences in students’ perceptions toward the two picture prompts, except that the closed-ended prompt was considered to hinder more creativity than the open-ended prompt. The findings suggest that the two picture prompts could still be used alternatively in such large-scale examinations as GSAT and that EFL writing instructors should increase students’ general English proficiency and reduce their writing anxiety. Key words: writing assessment, writing with picture prompts, open-ended prompts, closed-ended prompts, L2 writing anxiety, English proficiency. ii.

(4) Acknowledgements This study would not have been accomplished if it had not been for many people’s assistance. First and foremost gratitude goes to my advisor, Professor Yuh-show Cheng. I would have quit halfway without her instruction and constructive feedback, not to mention her unsurpassed wisdom of research methodology. It is her statistics knowledge that helps me analyze the overwhelming data efficiently. Even after countless revised drafts, Professor Cheng keeps me company and answers my questions extremely patiently in person or by email. Under her tutelage, like a beacon of hope in the dark, I feel a sense of security amidst the migration route, fly back on track, and complete the thesis writing without getting sidetracked. I owe a great deal of debt of gratitude to her. Sincere gratitude extends to my committee members, Professor Hsi-nan Yeh and Professor Wu-chang Chang. Their significant comments nourish my study not only in the thesis proposal stage but in the oral defense; their valuable insights make the thesis even more intact. I would also like to express thankfulness to the two experienced raters in CEEC, Professor Chiou-mei Chen and Professor Jin-lan Wu. Both of them rate the 210 essays, which validates a satisfactory inter-rater reliability in the present study. Special appreciation also forwards to my classmates who offer courage when I encounter frustration among three-year bittersweet learning moments and my colleagues who invite their students to participate in the study and who substitue me while I couldn’t be availble. Sophie, Jennifer, Stacy, and Victor in particular, my sweetest classmates in TESOL program, provide their timely, passionate support helping my thesis proposal and defense proceed more smoothly. On account of their encouragement I could concentrate on my presentation. Last but not least, I hope to dedicate innermost gratitude to my dearest family: my life partner Chun-ting Ku, my parents, and my parents-in-law. Sometimes they shoulder the iii.

(5) responsibilities of doing as many household chores as they can, while sometimes they just grant me comfort and alleviate my plight of thesis writing. Like stars twinkling in the sky, their unconditional love as well as unbelievable tolerance illuminates the difficult but fruitful path of conducting research so that I can realize my ideal in the realm of L2 writing research.. iv.

(6) TABLE OF CONTENTS Page Chapter I: Introduction. 1. Background and Rationale. 1. Research Questions. 7. Significance of the Study. 8. Definition of Key Terms. 9. Chapter II: Literature Review. 10. Writing Assessment. 10. Picture Prompts for Guided Writing. 12. Effects of Picture Prompts on Writing. 12. Attributes of Successful Picture Prompts. 13. The Use of Picture Prompts in the English Writing Assessment in GSAT. 15. Relationships between Language Proficiency and Writing Performance. 17. Research on Anxiety. 19. Anxiety. 19. Foreign Language Anxiety. 19. L1 Writing Anxiety. 20. L2 Writing Anxiety. 21. Chapter III: Method. 24. Participants. 24. Instruments. 24. English GSAT. 25. Questionnaires. 26. Writing Tasks: Picture Writing with Two Types of Pictorial Prompts. 27. Interview. 30 v.

(7) Data Collection Procedure. 30. Scoring Procedures for Writing. 31. Data Analysis Procedures. 33. Chapter IV: Results. 35. Background of the Participants. 35. Results on the Effect of Picture Prompts on Writing Performance. 36. Results on the Interaction Effect of Writing Prompts and Proficiency. 37. Results on the Interaction Effect of Writing Prompts and L2 Writing Anxiety. 40. Results on Students’ Perceptions about the Two Writing Prompts. 43. Post-writing Questionnaire. 43. Interview Results. 45. Chapter V: Discussion and Conclusions. 48. Summary of Major Findings. 49. Discussion. 49. The Effect of Picture Prompts on Writing Performance. 50. The Effect of Proficiency on Writing Performance. 51. The Effect of L2 Writing Anxiety on Writing Performance. 51. Limitations and Suggestions for Future Research. 52. References. 54. Appendices. 64. Appendix A: Adapted 2003 GSAT-English. 64. Appendix B: Pre-writing Questionnaire (Chinese Version). 70. Appendix C: Pre-writing Questionnaire (English Version). 72. Appendix D: Post-writing Questionnaire on the Closed-ended Picture Prompt. 77. Appendix E: Post-writing Questionnaire on the Open-ended Picture Prompt. 78. Appendix F: Picture Writing with Closed- and Open-ended Prompts. 79. vi.

(8) Appendix G: Samples of Writing with the Closed-ended Picture Prompt. 83. Appendix H: Samples of Writing with the Open-ended Picture Prompt. 85. Appendix I: Interview Transcription. 87. vii.

(9) LIST OF TABLES Page Table 1: Topics of English Writing Tests in the GSAT. 3. Table 2: The Percentage of the Examinees Getting Zero. 5. Table 3: Opinions on Better Writing Item Types Collected from College Professors and High School Teachers. 17. Table 4: Scoring Guide of the CEEC Writing Test. 32. Table 5: Follow-up One-way ANOVAs of Writing Performance by Prompt Type. 37. Table 6: Descriptive Statistics of Writing Fluency, Accuracy, and Overall Quality by Prompt Type and Proficiency Level. 38. Table 7: Follow-Up One-Way ANOVAs of Writing Performance by Proficiency Level. 39. Table 8: Estimated Marginal Means of Writing Performance by Proficiency Level. 39. Table 9: Descriptive Statistics of Writing Fluency, Accuracy, and Overall Quality by Prompt Type and Writing Anxiety Level Table 10: Follow-Up One-Way ANOVAs of Writing Performance by Prompt Type. 41 42. Table 11: Follow-Up One-Way ANOVAs of Writing Performance by Writing Anxiety Level. 42. Table 12: Estimated Marginal Means of Writing Performance by Writing Anxiety Level 43 Table 13: T-test Results on Learners’ Perceptions of the Picture Prompts. viii. 44.

(10) LIST OF FIGURES Page Figure 1: Picture Writing with a Closed-ended Prompt. 28. Figure 2: Picture Writing with an Open-ended Prompt. 29. ix.

(11) CHAPTER ONE INTRODUCTION. Background and Rationale Writing has increasingly captured much research attention in second language acquisition domain for the past few decades. Writing not only enables people from different culture backgrounds or from all walks of life to communicate effectively, but contributes to transmission of knowledge into new knowledge from generation to generation. It is also important to students in academic settings across the globe. However, writing involves complex processes, such as generating, organizing and translating ideas. This may account for why learners usually find English writing difficult and challenging, and view writing as a painstaking process. According to Murray (1984), 70% of the time of writing was spent on prewriting (idea-generating) stage. In other words, what students are really assessed in a writing test may not be writing abilities but the idea-generating abilities. To solve this problem, picture prompts are often used to elicit writing. The pictorial stimuli can give students some crucial elements for generating ideas such as introduction of setting, characters, events, development of the plot, resolution, and conclusion, hence freeing them from struggling for the content to be written down (Fang, 1996). Sinatra (1973) indicated that a sequence of pictures helped students compose a piece of unified written passage. These pictures function as concrete stimulus for forming images, which can be recalled more powerfully than words. Hsu (1991) claimed that picture prompts could not only enrich the content of writing but also help students enhance their writing ability as well. Many studies have been conducted to investigate the effects of writing with picture prompts. For example, Funderburk’s (1986) research on English L1 students showed that primary school students 1.

(12) could learn to form imagery effectively from concrete stimuli like pictures. These concrete pictorial stimuli also helped the students generate ideas. Similarly, Yang (2004) suggested that elementary school students benefited from instruction with picture prompts when learning to compose in Chinese L1. In L2 research, Chiang’s (2003) study revealed that pictures could improve the content of English composition for EFL junior high school students. Tsai’s (2009) study on EFL senior high school students found that picture prompts significantly enhanced high achievers’ performance on content, organization, and overall quality of writing, but they did not help the low achievers’ performance. Currently, writing with picture prompts has played a vital role in the majority of writing assessments in Taiwan, an EFL context. It has been adopted in various tests. This can be proved from the writing assessments in the General English Proficiency Test (GEPT) and General Scholastic Ability Test (GSAT) for the past few years. The College Entrance Examination Center conducted a study (Chen, Huang, Lin, Lin, & Li, 1993) on different item types of English Writing Ability test. The result of the study revealed that there are five types appropriate for the Joint College Entrance Examination. The five types are free composition with instructions, composition based on pictures, composition based on a certain situation, composition starting with a topic sentence, and letter writing. Since 2004, the writing test with one picture or a sequence of picture prompts has been the mainstream type in GSAT. Examinees are given one picture or a sequence of pictures and are invited to transfer the visual information into verbal representation according to their understanding and interpretation of the sequential pictures (see Table 1).. 2.

(13) Table 1 Topics of English Writing Tests in the GSAT Year. Item Type. Subject. 2003. Guided Writing. Music Is an Important Part of Our Life. 2004. Picture Writing. A Drunken Man. Closed-ended. 2005. Picture Writing. Dozing Off in an English Class. Closed-ended. 2006. Picture Writing. A Surprise: A Lady and A Gorilla. Closed-ended. 2007. Picture Writing. A Girl and A Cat. Closed-ended. 2008. Guided Writing. An Apology Letter. 2009. Picture Writing. One Single Picture of a Collapsed. Resolution. Closed-ended. Building 2010. Picture Writing. A Man with His Lost Cash. Open-ended. 2011. Picture Writing. To Pursue a Lady. Open-ended. 2012. Guided Writing. A Letter to Give Suggestions. 2013. Picture Writing. Sitting on Priority Seats. Open-ended. 2014. Picture Writing. A Smartphone Addict. Open-ended. The survey conducted by CEEC (1993) also revealed that college professors and senior high school teachers hold picture writing in high esteem. Besides providing story plots for examinees, writing with picture prompts also prohibit the students from memorizing or reproducing samples of essays. This prevention has resulted in the positive washback effect in writing instruction, cultivating students’ writing ability itself instead of the memory skill. A sequence of picture prompts with a constrained, a closed ending has been utilized in English writing assessment in GSAT from 2004 to 2007. This type of picture prompt has. 3.

(14) both advantages and disadvantages. Joshua (2007) suggests that while pictures provide context and background information for student writers, they may hinder creativity. In particular, once students develop language proficiency, pictures may constrain creativity in writing development. Bates’ (1991) study also demonstrated that picture strips could negatively affected the structure of children’s written narratives. When children write without any picture, they use a more developed story narrative than when they are given picture prompts as an initial stimulus. The problem might become even more serious when constraints are imposed upon the children by a series of pictures that do not follow his or her thought or expectations. According to the statistics of the studies by CEEC (Lin, 2005; 2012), the percentage of students who scored zero in the writing section of GSAT was 12.83% in 2004, 11.12% in 2005, and up to 16.1 % in 2009, so the general public might harbor the illusion that it was challenging for examinees to score high in picture writing with a closed-ended prompt. Particularly in 2009, the percentage of students scoring zero reached a record high. However, it should be noted that the picture writing task in 2009-GSAT is much different from the picture writing task in the GSAT of other years. In addition to the use of a single picture instead of a series of pictures, the writing task in 2009-GSAT asked examinees to write what might happen in the past, present, and future based on the given single closed-ended prompt. Therefore, it requires more than simple narration of the succession of events depicted in the pictures. On the other hand, the implementation of writing tests with sequential picture prompts with constrained, closed ending could prohibit the students from merely memorizing the formulaic expression. Therefore, the writing assessment can better probe the students’ writing ability.. 4.

(15) Since 2010, a new type of prompt with open-ended resolution has been born. With open-ended resolution, examinees should develop the narrative and generate a conclusion by themselves based on the three picture prompts provided as well as their originality, life experiences, and background knowledge. In the two consecutive years, picture writing with open-ended prompts was employed as the item type in the GSAT writing section. Somehow, the percentage of the examinees getting zero diminished: 11.1% in 2010 and 9.7% in 2011 (see Table 2).. Table 2 The Percentage of the Examinees Getting Zero Academic. The Percentage of Item Type. Resolution. Year. Students Scoring Zero. 2004. Picture Writing. Closed-ended. 12.83%. 2005. Picture Writing. Closed-ended. 11.12%. 2006. Picture Writing. Closed-ended. 11.03%. 2007. Picture Writing. Closed-ended. 8.67%. 2009. Picture Writing. Closed-ended. 16.1%. 2010. Picture Writing. Open-ended. 11.1%. 2011. Picture Writing. Open-ended. 9.7%. 2013. Picture Writing. Open-ended. 9.2%. 2014. Picture Writing. Open-ended. 8.4%. Picture writing with open-ended prompts seems to have an effect on learners’ writing performance and benefit the low achievers. Thus, the present study aims to probe the effects of picture prompts with closed- and open-ended resolution on senior high school. 5.

(16) students’ writing performance. Another factor considered in this study is language proficiency, the effect of which on L2 writing performance is not yet certain. On the one hand, Cumming (1989), Raimes (1987), and Zamel (1982) reported that learners’ L2 proficiency level had no effect on writing performance. On the other hand, studies by Way, Joiner, and Seaman (2000), Lin (2005), and Tsai and Cheng (2009) showed that higher proficiency students tended to generate higher grades in writing assessment whereas lower proficiency students tended to yield lower grades in composition. More importantly, Tsai’s (2009) finding that picture prompts significantly enhanced writing performance of high achievers but not low achievers suggests that proficiency might moderate the effect of picture prompts. In view of the inconclusive effects of language proficiency on writing performance and the potential moderating role of language proficiency on the effects of picture prompts, the present study also investigates L2 language proficiency. Apart from prompt types and learners’ proficiency level, writers’ characteristics are also believed to have an effect on writers’ composition (McNamara, 1996). Writing anxiety is one of the writers’ characteristics that has received many researchers’ attention. Previous research has recognized writing anxiety as an influential factor in the process of learners’ writing. Moreover, most of the studies indicated that writing anxiety negatively impacted learners’ writing quality (Faigley, Daly, & Witte, 1981; Golen & Lynch, 1999; Lee & Krashen, 2002; Reed, 1992; Tsai & Cheng, 2009). For example, Lee and Krashen (2002) reported that highly apprehensive learners scored lower on standardized writing tests. Similarly, the results of Faigley, Daly, and Witte’s (1981) research confirmed that writing apprehension played a role in writing competency of undergraduates: High apprehensive learners performed worse than low apprehensive peers did on standardized tests and written essays. Their study further showed that writing anxiety had a negative effect on narrative essays, but not on argumentative essays. However, Tsai and Cheng (2009) found 6.

(17) that writing anxiety had a negative effect on both narrative and expository writing tasks. The results of these studies suggest that the effect of writing anxiety on writing performance might differ with the rhetorical tasks. As the effect of writing anxiety seems to differ with rhetorical tasks, it is also possible for the effect of writing anxiety to vary with picture prompts of closed or open endings. For example, low anxiety learners may perform equally well on both kinds of picture prompts. High anxiety learners, however, may perform better on a closed-ended picture prompt than on an open-ended prompt, which demands greater improvisation and is thus cognitively more demanding. Consequently, it is worthwhile to examine how the effects of writing anxiety on writing performance vary with the two picture prompts under study. To sum up, little research has been done on the effects on writing of open-ended and closed-ended picture prompts. Besides, few studies, if any, inspected whether the effects of open-ended and closed-ended picture prompts on writing quality vary with language proficiency or writing anxiety levels. Last but not least, senior high students were seldom recruited as participants in EFL writing research. The aforementioned research gaps inspired the design of the present study. This study targets at examining senior high school students’ writing performance in terms of three factors: the closed- and open-ended picture prompts, writing anxiety, and English L2 proficiency. The foremost concern is the effects of picture prompts.. Research Questions The present study explores the following questions: 1. Is there any significant effect of open- and closed-ended picture prompts on senior high school students’ writing performance, in terms of overall quality, fluency, and accuracy?. 7.

(18) 2. Do the effects of open- and close-ended picture prompts on writing performance vary with learners’ English L2 proficiency, in terms of overall quality, fluency, and accuracy? 3. Do the effects of open- and close-ended picture prompts on writing performance vary with learners’ English L2 writing anxiety, in terms of overall quality, fluency, and accuracy? 4. How do senior high school students think about the two types of picture prompts?. Significance of the Study The present study hopes to contribute to three communities: writing assessment developers, writing instructors, and writing researchers. First of all, this study will add empirical evidence to literature on the role of writing prompts, writing anxiety, and language proficiency in writing performance. Second, by illustrating the effects of the above three factors in writing performance, this study will help writing instructors design learning activities or writing tasks that better address learners’ needs. For example, they may design activities to reduce learners’ writing anxiety or offer more practice on a certain writing prompt if either of them is found to play a significant role in writing. Finally, the results of this study may help testing institutions make informed decisions on designing writing tests. For example, if the two writing prompts make a significant difference in writing performance, it would not be appropriate for testing institutions to use them as parallel tests. If significant interaction between writing prompts and L2 writing anxiety or English L2 proficiency is found and if a particular prompt is found to favor learners of particular characteristics, interpretations of learners’ performance on tests with the writing prompt should be more cautious.. 8.

(19) Definition of Key Terms Closed-ended picture prompts: In this study, they are defined as a single or a sequence of picture prompts for writing with the last prompt of resolution provided. Open-ended picture prompts: In this study, they are defined as a sequence of picture prompts for writing without the last prompt of resolution provided. L2 writing anxiety: It is a type of anxiety arising from writing in L2, which involves distress associated with writing, increased physiological arousal, and maladaptive behaviors.. 9.

(20) CHAPTER TWO LITERATURE REVIEW. The literature review falls into four sections in this chapter: writing assessment, picture prompts for guided writing, language proficiency and writing performance, and research on anxiety.. Writing Assessment Methods of assessing to what extent learners’ writing ability is have been an evolving interest for many researchers, test administration authorities and teachers alike. Regarding writing testing, three major forms of measurement are indirect writing assessment, direct writing assessment, and portfolio assessment (Yancey, 1999). Indirect writing assessment inspects learners’ writing ability through taking multiple-choice proficiency tests, whereas direct writing assessment requires learners to write on topics appointed. Portfolio assessment, as Brossell (1986) described, collects learners’ multiple writing samples on different occasions and in various rhetorical modes. Indirect writing assessment obsesses the merit of better control of statistical reliability. If school faculty needs to predict how learners will perform the skills being assessed in a fair, consistent way, indirect writing assessment can be implemented (Yancey, 1999). Moreover, practically speaking, the implementation of indirect, objective tests is easy, with the lowest cost and the least amount of marking (Hyland, 2003). However, indirect writing assessment also has its demerits. It is criticized for inferential judgment about learners’ writing ability (Cooper, 1984; Hyland, 2003). Instead of judging learners’ writing ability through essay writing, objective tests tends to indirectly measure vocabulary, grammar, usage, reading skills and error-recognition skills.. 10.

(21) The inquiry to validity paves the way for direct writing assessment. According to Hughes and Hyland (1989; 2003), direct writing assessment is more likely to reach face validity and to reflect real-life communication needs. Therefore, ESL professionals prefer assess learners’ writing ability by asking learners to write on a given topic (Hamp-Lyons, 2003). However, it is inefficient, laborious and expensive for raters to complete the marking of essays. In order to enhance reliability, the raters need to receive training and follow a scoring guide before marking. For example, White, Jones, Greenberg, Odell, and Cooper (1993) devised a procedure that enhanced inter-rater reliability through (1) using prompts to direct learners; (2) selecting scoring guides for raters; and (3) devising methods of calculating acceptable agreement. Under direct writing assessment, scoring of essays can be orchestrated either analytically or holistically by more than two raters. By adding up the sum of sub-scores, the analytical scoring procedure assumes that the quality of essays can be judged. While calculating the score of an essay, raters mark the traits separately within one or two minutes without influencing another trait being marked and then finally generate a total score (Hughes, 1989; Spandel & Stiggins, 1980). Nevertheless, some argued that the process of analytical scoring is tedious and time-consuming (Cooper, 1984; Nakamura, 2004). In contrast, holistic scoring takes raters two to three minutes per essay to give marks based on global impression of a writing production, usually with the aid of a scoring guide to endorse reliability of scoring. Thus, holistic scoring is more economical and efficient (Cooper, 1984; Huot, 1990). Besides, the scores yielded by holistic procedures correlate well with those by analytical procedures, which makes holistic scoring prevail over analytic scoring particularly in the circumstances of large-scale writing tests (Charney, 1984; Freedman, 1984).. 11.

(22) Picture Prompts for Guided Writing Nowadays, the use of visual aids has been prevalent when teaching students to learn, let alone to write. Writing with pictures as prompts has long been carried out during in-class teaching as well as assessment. Previous research has shown that visual tools helped learners to transfer thinking to vocabulary and build up their inner thoughts (Canning, 2001; Hyerle 1996). Furthermore, visual stimuli can be used as motivation, general comprehension and conversational spur (Hammerly, 1995). In addition, according to Stevick (1986), picture prompts stimulate ideas and learners’ background knowledge. If there is a situation for learners to write about, learners will find it easier to share their thoughts with readers. As a result, contributions of pictures to the average learners’ literate behavior seem to be great (Fang, 1996). Fang (1996) concluded that pictures played six important roles: Pictures may serve to help (a) establish the setting, (b) define/ develop the characters, (c) extend/ develop the plot, (d) provide a different viewpoint, (e) contribute to the text’s coherence, and (f) reinforce the text. And pictures also provide several benefits, including motivating the reader, promoting creativity, serving as mental scaffolds, fostering aesthetic appreciation, and promoting children’s language and literacy. (p. 134). Effects of Picture Prompts on Writing Well goes the proverb, “A picture is worth a thousand words.” The effects of writing with picture prompts have been investigated throughout recent years. Sinatra (1981; 1986) contended that pictorial stimuli can provide concrete impetuses to form imagery and a sequence of pictures can train learners to develop underlying discourse structure. Pictorial stimuli can serve as a scaffold of not only thoughts for writing but also writing skills for beginners. Søvik and Flem’s (1999) study examined Norwegian elementary school 12.

(23) students, indicating that textual prompts had great effects on persistence on writing time and pauses in writing while pictorial prompts had significant effects on spelling, accuracy, and fluency. In McMaster, Du and Pétursdóttir’s (2009) research, photo prompts provided generic stimuli to elicit wider range of responses whereas the letter prompts offered specific words thus constraining writing performance. Paquette (2007) suggested that pictures demonstrate six skills of effective writing, such as ideas, organization, voice, word choice, sentence fluency, and conventions. Local studies also have been conducted on picture-elicited writing. Chien (2007) classified three-frame picture prompts into the time-based type, the steps-in-a-process type, the topic-development type, the cause-and-effect type with an expected consequence, and the cause-and-effect type with an unexpected consequence. It was found that the third graders in a senior high school had better writing fluency in the last two types. Liu (2008) specifically pointed out in her study that the picture prompts had significant effects on vocabulary richness of third graders of a senior high school. Tsai (2009) claimed that picture prompts had greater effects on content, organization, and overall writing performance of high achievers, compared with low achievers. In addition to improving linguistic skills, learners also change their attitude and motivation by using picture prompts. Chiang’s (2003) study suggested that pictorial stimuli helped learners build a better organization, provide richer content, evoke imagination, and alleviate writing anxiety as well.. Attributes of Successful Picture Prompts Picture prompts can serve as effective visual aids to elevate students’ writing performance; however, not every learner interprets a sequence of pictures in the same way. What’s worse, some may misunderstand the picture and lose the grasp of what the author wants to convey, which might not ensure a successful composition. In this sense, it matters 13.

(24) what essential characteristics of picture prompts are selected in writing assessment. According to Seaman (1993), designing a writing task with picture prompts needs to consider a number of variables: the characteristics of the picture, the native language and culture of the students, and the nature of writing task itself. Whether the picture prompts are comprehensible or not has an influence on the writers’ interpretation. Selecting a picture prompt depends on the picture content (Schweizer, 1999). In particular, the object of picture prompts should be relevant and age-appropriate to the target writer’s background knowledge and experience (Cycowicz, Friedman, Snodgrass & Rothstein, 1994) because writers’ prior knowledge and familiarity can trigger their ideas for writing (Bates, 1991). While selecting pictorial stimuli, complexity should also be taken into consideration: high levels of complexity in terms of interactions among objects in a sequence of pictures elicit more and better writing production than low levels of complexity (Samson & Wescott, 1983). Peterson (1986) further suggests that a balance in level of complexity should be struck. Regarding the type of the prompt, single pictures will generate more descriptive paragraphs, whereas sequences of pictures will generate more narratives (White, 1978). Therefore, it is evident that the picture prompt should be developed with care and preplanning so that a high level of written responses could be created. Hooper, Montgomery, Swartz, Reed, Sandler, Levine, Watson, and Wasileski (1994) define a good visual stimulus as meeting the following criteria: 1. At least two characters, with perhaps one potential protagonist and one potential antagonist although this is not absolutely necessary, depending on the situation depicted. 2. A depiction of some kind of “interesting” or “novel” scene or event (e.g., children lost in a cave).. 14.

(25) 3. Some kind of potential conflict between the antagonist and protagonist that necessitates that the protagonist engage in a goal-based sequence of events to resolve the conflict or a scene in which a main character must resolve a problem using some goal-directed sequence of events. (p.388) Cole, Muenz, Ouhchi, Kaufman, & Kaufman’s (1997) study support the hypothesis of Hooper et al.’s (1994) that a better organized and more cohesive writing can be produced following the criteria. The study examined 50 participants aged 13 to 46 and compared the pictorial stimuli developed based on Hooper et al.’s criteria with conventional line drawing stimuli. The result demonstrates that the Hooper stimuli yielded higher scores than conventional stimuli in terms of structure and cohesiveness of the story; however, the result is not the same in mechanics. Joshua’s (2007) study examined 165 elementary L1 students’ writing activity, finding that younger students benefited from writing with pictures and drawing; however, elder students beyond kindergarten did not positively benefit from the picture paired with the prompt. He thus concluded that though pictures provide context and background information for writers, they may hinder creativity. The deciding factor of a good picture prompts seems to be participants’ background knowledge of language and living experience. Based on Cole et al’s study, Joshua (2007) proposed a criterion for selection of picture prompts in classroom writing tasks: Picture prompts should be provoking and engaging, imaginative, and grade-level appropriate.. The Use of Picture Prompts in the English Writing Assessment in GSAT There are two Joint College Entrance Exams (JCEEs) in Taiwan. The one held in February is called General Scholastic Ability Test (GSAT), while the other held in July is called Department Required Test (DRT). Both entrance exams designate English as a test subject. The English test of both exams involves two sections: multiple-choice questions 15.

(26) and non-multiple-choice questions. The latter comprises two Chinese-English translations and a piece of essay writing. The English writing section of GSAT and DRT shares the same purpose of assessing senior high school graduates’ ability to compose an essay of coherence and cohesion. Each essay is rated independently by more than two raters following a scoring criterion: content, organization, grammar, vocabulary, and mechanics (Lin, 2012). And the full score of an essay on both tests totals 20 points. However, GSAT differs from DRT in test content and format. The GSAT writing section often involves a narrative writing task, while the DRT writing section often involves an expository writing task. Occasionally in several consecutive years, the GSAT writing test furnishes examinees with one or more pictorial prompts to guide them to write a coherent and unified story. Besides, the GSAT writing requires examinees a minimum of 100 words, whereas the DRT writing requires the word count to reach 120 words. Picture writing in the English writing assessment of JCEE first appeared in 1984. However, that was intended for Evening Programs. Starting from 2004, picture writing has become a mainstream writing assessment for senior high graduates. A report released by College Entrance Examination Center (CEEC) in 1993 further indicated that the ideal picture prompt for eliciting writing performance is one single picture with an unambiguous central idea. What’s more, the picture prompt should render some cues for writers to grasp the content without misunderstanding and make scoring unbiased and impartial. Along with the 1993 CEEC report, The Pamphlet of Item Writing Skills and Scoring Guides for the Writing Ability Test on the JCEE was published, in which picture writing is listed as one of the item types of English composition for the JCEE. In addition, one of the surveys reported in the 1993 report pointed out that college professors and high school teachers as well regarded picture writing as a better item type for English composition assessment in JCEE. According to the report (CEEC, p. 5), 44.6% of professors and 37.2% 16.

(27) of teachers favored picture writing, which was ranked as the first and the third favorite writing task among free writing, guided writing, picture writing, question-answering writing, situation writing, writing with provided words and phrases, and others (Table 3).. Table 3 Opinions on Better Writing Item Types by College Professors and High School Teachers Item. Groups. College Professors. High School Teachers. (N = 139). (N= 371). n. %. n. %. Free Writing. 61. 43.9. 162. 43.7. Guided Writing. 58. 41.7. 178. 48. Picture Writing. 62. 44.6. 138. 37.2. Question-Answering. 25. 16.5. 81. 21.8. Situation-Based Writing. 50. 36.0. 192. 52.6. 34. 24.5. 93. 25.1. No Comment. 5. 3.6. 17. 4.6. Others. 11. 7.9. 2. 0.5. Writing With Words and Phrases Offered. Note. More than one item type could be chosen.. Relationships Between Language Proficiency and Writing Performance Whether reading, listening, writing, and speaking are separate skills has aroused heated debate in second language field. Previous research assumed that if the four facilities are independent from one another, writing performance is not related to other skills or overall language proficiency (Cumming, 1989; Raimes, 1987). Raimes (1987) investigated holistic writing quality and the scores on a proficiency test of eight ESL students. The participants completed two tasks: writing a letter and writing an argumentative essay. The result of the study presented a weak correlation between writing evaluation and students’ second language proficiency level. Cumming (1989) analyzed 23 young adult bilingual 17.

(28) program students’ writing performance by three writing tasks: an informal letter, an expository argument, and a summary of a booklet, obtaining a result that writing expertise could be isolated from second language proficiency. Opposite to the abovementioned line of thought, other research has converged on the significant effect of second language proficiency level on quality of essay writing (Lin, 2005; Sasaki & Hirose, 1996; Way, Joiner, & Seaman, 2000). Sasaki and Hirose (1996) analyzed the factor which influenced 70 Japanese college students’ expository writing. The consequence of their study showed that the higher the learners scored on English proficiency tests, the better essay quality they achieved. Similar to Sasaki and Hirose’s finding, Way, Joiner, and Seaman (2000) noted that proficient French learners produced better essays than novice learners. Lin (2005) observed EFL vocational high school students’ writing performance on free writing and guided writing. The study presented a result that whatever the task type was, proficient learners scored higher than less proficient counterparts did. Inconsistent findings of previous research have not been unsnarled so far. More importantly, improvement of research design is needed to enhance the credibility of research findings. First of all, a larger sample is needed because previous studies usually involved a small sample size, which make their findings less convincing. For example, Lin (2005) examined 35 English majors; Cumming (1989) recruited 23 adult participants; Raimes (1987) observed 8 students. In addition, validity or reliability of scoring rubrics and proficiency tests should be ensured. Furthermore, unlike Way, Joiner, and Seaman’s (2000) use of two separate levels of rubrics to measure the writing performance of proficient learners and novice learners, the same scoring rubric should be utilized on different groups of participants to ensure meaningful comparison. In conclusion, the mixed aforementioned findings suggest that more research should be conducted to unsnarl the mysterious relationship between picture prompt types and language proficiency in 18.

(29) affecting writing performance.. Research on Anxiety Anxiety Anxiety, a psychological construct, can be generally divided into three types: trait anxiety, state anxiety, and situation-specific anxiety (MacIntyre, 1999; MacIntyre & Gardner, 1989; Speilberger, 1983). Speilberger (1983) stated that “anxiety is the subjective feeling of tension, apprehension, nervousness, and worry associated with an arousal of the autonomic system.” (p.1) Trait anxiety refers to a “stable predisposition to become nervous in a wide range of situations” (MacIntyre, 1999, p.28). Trait anxiety is connected with an individual’s personality (Speilberger, 1983). People with trait anxiety are inclined to be constantly anxious and lack emotional stability. State anxiety refers to the at-once experience of anxiety (Goldberg, 1993). People with state anxiety would feel anxious about certain event transiently. Situation-specific anxiety occurs under “a single context or situation only” (MacIntyre, 1999, p.28). People with situation-specific anxiety become anxious when giving a speech, taking a test, solving math questions, or using a second language.. Foreign Language Anxiety Previous studies, conducted by Gardner, Horwitz, E. K., Horwitz, M. B., & Cope, and MacIntyre & Gardner (1985; 1986; 1991a; 1991b), found that foreign language anxiety differed from trait anxiety or the anxiety generated from learning other subjects such as math or science. They classified language anxiety into a kind of situation-specific anxiety, which can be provoked when one learns a foreign language. Foreign language anxiety, according to Horwitz, Horwitz, and Cope (1986), consists of three components: communication apprehension, test anxiety, and fear of negative 19.

(30) social evaluation. Communication apprehension appears when learners speak out their ideas. Their anxiety hampers oral delivery of thoughts, and their unhappy experience gives rise to apprehension which hence obstructs future attempts to communicate with others (Sparks & Ganschow, 1991). Test anxiety occurs when learners overestimate the difficulty level of examinations and therefore impose false beliefs on language learning (Horwitz, 1986). Moreover, fear of negative social evaluation comes about when learners are unsure of how to express their opinions. Watson and Friend (1969) defined fear of negative social evaluation as “distress over negative evaluations and the expectation that others would evaluate themselves negatively.” (p.449) A number of studies (Kim, 2000; Price, 1991) indicated that foreign language anxiety has an association with oral achievement. On the other hand, still plenty of studies (Cheng, Horwitz, & Schallert, 1999) further differentiated general foreign language anxiety from skill-specific language anxiety such as reading, listening, speaking and writing anxiety. Their research findings suggested a need to probe into skill-specific anxiety, in particular, writing anxiety.. L1 Writing Anxiety The impacts of writing anxiety on writing performance have received much attention in L1 research. McKain (1991) examined 35 high writing-anxious and 33 low writing-anxious graduate students, finding that the high writing-anxious participants procrastinated more than low writing-anxious participants did. Faigley, Daly and Witte (1981) maintained that writers with high writing anxiety produced essays with tremendously fewer words and with less mature syntactical structure than their peers with low writing anxiety. The researchers also revealed that students with higher w riting anxiety scored lower on the writing tests, which were implemented for evaluation in college writing courses. The findings of Faigley, Daly and Witte (1981) suggest a negative 20.

(31) correlation between writing anxiety and writing performance. On the other hand, a few studies suggested that writing anxiety had no significant effect on high- and low- anxiety students with regard to writing performance (Madigan, Linton, & Johnson, 1996; Richardson, 1981). Richardson (1981) examined college students’ writing performance in terms of holistic writing quality, syntactic structure, and coherence. Participants wrote essays to two groups of readers, a distant audience and a familiar audience. The result manifested no correlation between students with higher anxiety level and those with lower anxiety level. That is, low anxious writers did not compose significantly better than high anxious peers. Likewise, Madigan, Linton, and Johnson (1996) inspected writers with anxiety and those without anxiety, discovering that when given unlimited time on expository essays the two types of writers did not significantly outperform each other in terms of holistic writing quality.. L2 Writing Anxiety In addition to L1 writing anxiety research, a burgeoning number of studies on L2 writing anxiety have recently been conducted to explore effects of anxiety on second language writers. Cheng’s (2004b) research identified the sources of EFL college majors’ writing anxiety as: (1) the instructional practices, (2) personal beliefs about writing and learning to write, (3) self-perceptions, and (4) interpersonal threats. (p.46) Consequently, Cheng (2004b) presented pedagogical implications, hoping to reduce writing anxiety levels of students. First, unfamiliar topics, time constraint, excessive emphasis on language form accuracy, and rigid composition rules all caused students’ writing anxiety. Second, the belief that a good essay was anticipated to be error-free also provoked writers’ apprehension. Third, still another writing anxiety, as Cheng (2004) specified, originated from learners’ low confidence, their self-perceptions of scant language competence in vocabulary and word usage, and their lack of rhetorical conventions, writing experience 21.

(32) and knowledge. Last, students’ fear of mistakes and limited writing proficiency, the threat of negative evaluation, and competitiveness also cause writing anxiety. Even worse, writing anxiety can trigger disappointment among students when producing compositions. Holladay (1981) discovered that one of the reasons that students became frustrated in writing stems from their apprehension of the demand of writing competence. Another study by Elias, Akmaliah, and Mahyuddin (2005) proposed that excessive and improper emphasis on superficial spelling and grammar errors, instead of the content, may lead to writing anxiety and frustration. Among some studies displaying negative correlations between writing anxiety and writing achievement (Gungle & Taylor, 1989; Hassan, 2001; Lee & Krashen, 2002; McKain, 1991), some subtle, contradictory results were observed at times (Masny & Foxall, 1992). For example, Lee and Krashen (2002) found a strong correlation between writing anxiety and writers’ composition grades. Their statistical analysis, multiple regression, claimed that the higher anxiety level the students reached, the lower marks they obtained in the writing tasks. Hassan (2001) further claimed that writing anxiety negatively affected not only the quantity but the quality of EFL university students’ writing performace. In Hassan’s study, the low anxious EFL Egyptian students surpassed the high anxious peers in generating more words with regard to the quantity and in achieving better organization with regard to the writing quality under time constraint. Gungle and Taylor (1989) noted a positive correlation between ESL writing anxiety and attention to form. They found that high anxious students were more concerned about the form whereas low anxious writers concentrated more on the content in their ESL compositions. However, Masny and Foxall (1992) revealed adverse data. In Masny and Foxall’s (1992) study, 28 adult participants aged from 20 to 55 were recruited from ESL intermediate classes in a predominantly French-speaking city in Canada. The findings of Masny and Foxall’s (1992) study confirmed the hypothesis that high achieving writers were less apprehensive, 22.

(33) suggesting that the participants with higher apprehension have a weaker knowledge of writing skills, vocabulary, and limited syntactic resources. The result also indicated that both high and low achieving ESL writers were significantly more concerned about form than content though low achieving writer being more concerned about form than high achievers. Moreover, the results showed that both high and low apprehensive groups more concerned about form than content. The correlational analysis found that apprehension was associated with the participants’ willingness to take advanced writing courses and that despite both groups’ concern for form, they saw the need to write. Although the above studies have enriched the understanding of what may account for the sources of writing anxiety, the mixed findings of previous research reveal room for further exploration of the target population on the effect of writing anxiety on writing with open-ended and close-ended picture prompts.. 23.

(34) CHAPTER THREE METHOD. The present study aimed to examine the effects of two types of picture prompts on English writing performance of EFL senior high school students. Furthermore, the study also probed how the effects of picture prompts vary with English proficiency and writing anxiety and how students of different levels of proficiency and writing anxiety perform on the two writing tasks. The research utilized a quasi-experimental between-subjects design to obtain the goal of the study. Both quantitative and qualitative data were collected. This chapter comprises five parts: participants, instruments, data collection procedures, scoring procedures, and data analysis.. Participants The participants of the current study came from five intact classes of a public co-ed senior high school in southern Taiwan. There were a total of 182 participants; 81 of them were male (45%) and 99 female (55%), with two not reporting gender. These participants were enrolled in their senior year. The five intact classes consisted of boy and girl adolescents, with two classes from the science track and three from the humanity track. As native Mandarin speakers, they had studied English for 9.91 years on average (SD=1.70).. Instruments To gather data, three questionnaires, an English proficiency test, two writing tasks, and interviews were used. An English GSAT was used to measure students’ English proficiency. A pre-writing questionnaire consisting of a background questionnaire and the Second Language Writing Anxiety Inventory (Cheng, 2004) was used to collect background information of the participants and determine the writing anxiety level of each 24.

(35) participant; a post-writing questionnaire was used to understand participants’ thoughts about the two writing tasks. Two types of picture prompts were used to elicit narrative writing in English, the quality of which were graded. Finally, interviews were conducted to understand students’ perceptions of the two types of picture prompts in more depth.. English GSAT The GSAT (General Scholastic Ability Test) is developed by College Entrance Examination Center (CEEC). At the end of the first semester of the third grade in senior high school, approximately 100,000 students in Taiwan take the test to measure their English competence. The GSAT on English is divided into two sections. One section focuses on reading competence, including multiple-choice questions on vocabulary, grammar, and reading comprehension, and the other on writing competence, including two Chinese-English translation questions and a piece of direct writing. Following rigid test development procedures, the reliability and validity of the GSAT is generally recognized in Taiwan. Besides, the vocabulary breadth and depth of the test is suitable for senior high school students. In the present study, the reading competence section of the GSAT in 2003 (see Appendix A) was employed as a basis of dividing students into two groups of equivalent English ability. The major reason for using the 2003 GSAT-English as a placement basis is that it was so old that the participants were very unlikely to have practiced with it. According to Chiang’s (2003) analysis, the reading section of the 2003 GSAT-English consists of 3041 word tokens and 979 word types; the level of syntactic complexity is appropriate for test-takers. In fact, 78 percent of in-service teachers and 66 percent of senior high school students surveyed perceived that the level of difficulty was considerably fair. Moreover, nearly 90% of the test items were of good discrimination index.. 25.

(36) The participants were given 45 minutes as time limit for the placement test. On the background questionnaire, students were asked to indicate if they had taken this test before. Based on their responses, data of those students that had taken this test were removed from the study. Concerning scoring, there were 40 multiple-choice test items, yielding a total score of 70 points. Each correct answer to grammar choice was given one point, and each correct answer to vocabulary and reading comprehension, two points.. Questionnaires The three questionnaires (one pre-writing questionnaire and two versions of post-writing questionnaires) were written in Chinese, the students’ native language. The pre-writing questionnaire consisted of two parts (see Appendices B & C). All of the participants had to complete this pre-writing questionnaire. Part one of this questionnaire probed students’ basic information, including age, gender, class, student ID number, how long they had studied English, whether they studied English somewhere else outside class, whether they had studied abroad, etc. Part two of the pre-writing questionnaire contained the 22-item Second Language Writing Anxiety Inventory (SLWAI; Cheng, 2004), an instrument specifically designed to measure English L2 writing anxiety and has been successfully used with senior high school students in Taiwan (e.g., Tsai & Cheng, 2009; Chiang, 2012). Reliability and validity evidence of the SLWAI has been established in Cheng (2004) through internal reliability analysis, test-retest reliability analysis, factor analysis, and correlation analysis. The SLWAI adopts a 5-point Likert response scale conforming to 1 (strongly disagree), 2 (disagree), 3 (neither agree nor disagree), 4 (agree), and 5 (strongly agree). After reverse scoring the negatively-worded items (Items 1, 4, 7, 17, 18, 21, and 22) and summing up the item score, a higher score on the SLWAI indicates a higher level of writing anxiety.. 26.

(37) There were two versions of the post-writing questionnaire, which were adapted and expanded from the questionnaire used in Chiang’s (2003) study on use of picture-elicited narratives to develop junior high school students’ English writing competence. Each version contained one open-ended question and 12 statements, with which the participants were required to show their agreement on a five-point Likert scale, ranging from 1 (strongly disagree), 2 (disagree), 3 (neither agree nor disagree), 4 (agree), and 5 (strongly agree). One version of the post-writing questionnaire mainly asked about students’ perceptions and attitudes toward closed-ended picture prompts for writing (see Appendix D). This questionnaire was administered to students that were assigned to write English essays according to the closed-ended picture prompt. The other version of the post-writing questionnaire mainly asked students’ perceptions and attitudes toward open-ended picture prompts for writing (see Appendix E). This questionnaire was administered to students that were assigned to write English essays according to the open-ended picture prompt.. Writing Tasks: Picture Writing with Two Types of Pictorial Prompts Based on the criteria defining good visual stimuli proposed by Hooper et al. (1994) and the assessment format of GSAT writing sections, the researcher developed two types of picture prompts: closed-ended and open-ended resolution prompts. As indicated in Hooper et al., the picture prompts of achievement tests should include: (a) at least two characters, (b) an interesting event, and (c) a potential conflict in which a leading character must resolve a problem. Moreover, some qualities were taken into consideration, including eliciting vocabulary that was within the knowledge of most of the participants, a topic familiar to senior high school students, and the clarity of the prompt images. Prior to this study, the researcher confirmed that the theme and prompts had never been practiced in the five target classes before.. 27.

(38) The picture prompt with a closed ending (see Figure 1) required the participants to describe an experience of taking a bus—a male adolescent dozed off on the bus, waking up only to find the driver and all passengers gone and the bus still moving forward, which shocked and perplexed him.. Figure 1 Picture Writing with a Closed-ended Prompt. 28.

(39) On the other hand, the picture prompt with an open ending (see Figure 2) required the participants to describe the same story except for the last frame of prompt, which left an open ending for the participants to stretch their imagination.. Figure 2 Picture Writing with an Open-ended Prompt. 29.

(40) The participants were given 30 minutes to write an English narrative according to the picture prompts (see Appendix F). To ensure that the participants were serious about the writing tasks, the researcher informed that the participants’ scores on the writing test would be counted as part of their English course grade. Besides, the researcher read the instructions from a prepared script, informing the participants of reading the directions carefully, being aware of the time limit, and not using any reference materials or dictionaries. Then 25 minutes after the participants began to write, they received a five-minute-left warning. At the end of 30 minutes, the researcher requested the participants to stop writing and hand in their writing samples.. Interview Interviews were conducted in Chinese to not only clarify the answers obtained from the two post-writing questionnaires but also elicit more in-depth responses from students about their thoughts on both kinds of picture prompts. In addition, the interviews were used as a secondary source of information to supplement the results of statistical analyses. The researcher interviewed a total of 12 participants, with six students (3 high achievers and 3 low achievers) randomly selected from each groups: the closed-ending group and the open-ending group. Based on each student’s responses, the researcher asked follow-up questions to gain an in-depth understanding of the students’ perceptions.. Data Collection Procedure The present study was implemented with five classes of senior high students at a co-educational public school in southern Taiwan. All participants were the third graders and at the end of the final semester in senior high school. Current study was carried out through five phases: (1) distributing an English proficiency test of 2003 GSAT and the pre-writing questionnaire; (2) grouping students into two groups of equal English. 30.

(41) proficiency based on their scores on the 2003 GSAT; (3) administering the English writing tasks to the participants, with one group of the students receiving the picture prompt with a closed ending and the other group, the picture prompt with an open ending; (4) distributing the two versions of post-writing questionnaires to relevant groups; and (5) conducting face-to-face individual interviews. Data collection procedures, except the individual interviews, took two periods of classes. During the first period of class, the participants were given a packet which contained a 2003 GSAT along with a separate answer sheet and the pre-writing questionnaire. The participants had 45 minutes to finish the GSAT, the purpose of which was to assess the participants’ English proficiency. As soon as they completed the GSAT, they were asked to fill out the pre-writing questionnaire. During a second period of class, namely two weeks later, the participants, who were divided into two groups of equal English proficiency based on their grades on the 2003 GSAT, received a writing test packet. One group of the participants were given the picture prompt with a closed-ended resolution and its corresponding post-writing questionnaire, and the other group were given the picture prompt with an open-ended resolution and its corresponding post-writing questionnaire. All participants spent 30 minutes on the writing task. Right after they completed the picture writing task, they were asked to fill out the post-writing questionnaire in 5 minutes. After the writing test, twelve students randomly selected from the two groups of participants were interviewed by the researcher. Six of the interviewees (3 high achievers and 3 low achievers) came from the closed-ending group while the other six interviewees (3 high achievers and 3 low achievers) came from the open-ending group.. Scoring Procedures for Writing The quality of the participants’ writing samples was scored holistically as well as anonymously in order to protect the participants’ privacy and to avoid rater bias. Holistic rating was based on the CEEC scoring scheme (see Table 4). 31.

(42) Table 4 Scoring Guide of the CEEC Writing Test Rank Category. Excellent. The content is clearly connected to the theme; Content concrete, complete supporting details. (5-4 points) The well-organized, coherent article contains introduction, Organization development and conclusion, with transitional words used appropriately. (5-4 points) There are almost no grammatical Grammar & errors, with Sentence versatile sentence Patterns patterns.. Vocabulary & Spelling. (4 points) The diction is accurate and appropriate, with few spelling errors. (4 points) The format, punctuation, and capitalization are almost error-free.. Format. (2 points). Fair. Acceptable. The theme is not clear enough. The relevant statements are incomplete. (3 points) The content is not arranged appropriately. The developing paragraphs are not well proportioned.. The theme is not clear. Most content is irrelevant to the topic. (2-1 points) The emphasis is not clear. The content is inconsistent.. The theme is. (3 points) Some grammatical errors, but do not affect the meaning.. (2-1 points) Many grammatical errors obviously affect the meaning. (2-1 points) Many errors in diction and spelling obviously affect the meaning. (2-1 points). (0 point) Frequent errors are found in grammatical usage, so the content is hardly intelligible. (0 point) Vocabulary is limited to the given topic.. (3 points) Diction is mediocre and repeated. Diction is occasionally inappropriate. (3 points) Some errors are found in the format, punctuation, and capitalization, but do not affect the meaning. (1 point) 32. The basic format, punctuation, and capitalization are disobeyed. (0 point). Incomprehensible. irrelevant to the topic. There is no written language. (0 point) The organization is loose.. (0 point).

(43) The participants’ essays were graded by two raters according to the above scoring rubric. The two raters were teaching in a prestigious university in northern Taiwan and had ten-year rating experience in CEEC. To reach agreement in consistent scoring, a pilot rating session was conducted, in which the two raters score 20 papers after some discussion over the scoring guide. After they finished rating the 20 papers, they discussed their reasons for assigning particular scores to the papers and reached an agreement in rating. After the pilot scoring session, each rater scored 162 pieces of writing independently. Two ratings of each paper were later averaged to calculate a holistic score. The inter-rater reliability estimate of .88 was yielded for the two essay raters by calculating Pearson product-moment correlation, which was higher than .80, a benchmark for reliability suggested by DeVellis (1991). Moreover, fluency and accuracy of writing were measured. The former was determined by counting the number of individual words in each participant’s sample. This measure of fluency has been accredited in previous research (Larsen-Freeman & Storm, 1977; Reid, 1990) and adopted by many EFL writing researchers (e.g., Koda, 1993; Henry, 1996). To measure accuracy, the percentage of error-free T-units in each writing sample was calculated. Initially advocated by Hunt (1965) in L1 settings, the T-unit is defined as a unit that composes a main clause with its subordinate clause.. Data Analysis Procedures In order to answer the first three research questions of the present study, statistical analyses were conducted. The SPSS package (version 22.0) was used to analyze the scores obtained from the 2003 GSAT, the SLWAI, and the writing test. To answer the first question on the effects of open- and closed-ended picture prompts on writing performance, one-way MANOVA was employed, with type of picture prompt serving as the independent variable and the holistic writing score, the fluency score, and 33.

(44) the accuracy score serving as the dependent variables. To answer the second question on whether the effects of closed-ended and open-ended picture prompts on writing performance varied with language proficiency, the participants were classified as high- or low-achievers based on their scores on the GSAT. The cutoff score was the obtained group mean of GSAT. A two-way MANOVA was run, with proficiency group and type of picture prompt serving as the independent variable and the holistic writing score, the fluency score, and the accuracy score serving as the dependent variables. To answer the third question on whether the effects of closed-ended and open-ended picture prompts on writing performance varied with writing anxiety, the participants were classified as high-anxiety or low-anxiety learners based on their scores on the SLWAI. The cutoff score was the obtained group mean of SLWAI. A two-way MANOVA was run, with anxiety group and type of picture prompt serving as the independent variable and the holistic writing score, the fluency score, and the accuracy score serving as the dependent variables. To answer the fourth research question on the participants’ thoughts about and attitudes toward the two types of picture prompts, data collected from the two post-writing questionnaires and interviews were analyzed. Specifically, the participants’ responses to the Likert questions were analyzed via descriptive statistics and compared by t-test; the interview data were transcribed, coded, and categorized.. 34.

(45) CHAPTER FOUR RESULTS. This chapter involves six sections. The first section summarizes results on the participants’ background information. The second to fourth sections present results related to the first three research questions on the effect of prompt types, interaction effect of prompt type and proficiency level, and the interaction effect of prompt type and anxiety level. To answer the fourth research question on students’ perceptions of the two prompts, results gained from the post-writing questionnaire and the interviews are presented in the fifth section. The major findings are discussed in the last section.. Background of the Participants The background questionnaire consisted of 14 items. The first four questions were about the participants’ personal information. Item 5 to Item 9 were about their English learning and English writing experiences. Item 10 and Item 11 were about the participants’ self-perception of their English writing ability and teachers’ evaluation on their writing. Item 13 and Item 14 were about whether they stayed abroad or went on a study tour. Although 182 students participated in this study, only 180 students completed the background questionnaire. Among the 180 participants, 81 participants were male (45%) and 99 female (55%). Their average age was 17.59 years old (SD = .52). They had studied English for 9.91 years on average (SD = 1.70). Around 49% of the respondents did not attend a writing course and 20% of them had attended a writing course for one year. Around 67% of the respondents went to cram school after class. While 40.6% of them thought that they developed a habit of studying English, 83.9% of the respondents spent less than one hour studying English. Most of the respondents had never lived in an 35.

(46) English-speaking country (98.9%) or joined a study tour (96.7 %).. Results on the Effect of Picture Prompts on Writing Performance The participants were divided into two groups of equivalent English ability after taking a placement test in the form of a shortened 2003-GSAT (t = .16, p = .875). The two groups were required to complete a writing task. One group (n = 90) completed writing with a closed-ended picture prompt, while the other (n = 92) completed a writing task with an open-ended picture prompt. Three indexes of writing performance were computed: overall writing quality, fluency of writing, and accuracy of writing. The participants’ overall writing quality was holistically scored, with a full score of 20. Fluency of writing was determined by counting the number of individual words in each participant’s sample, whereas accuracy of writing was measured by the percentage of error-free T-units in each writing sample. In terms of overall writing quality, the open-ending group produced a higher mean score (M = 8.33, SD = 3.52) than the closed-ending group (M = 8.20, SD = 3.26). Similarly, the open-ending group showed greater writing fluency (M = 140.47, SD = 41.94) than the closed-ending group (M = 127.76, SD = 35.28), whereas the closed-ending group showed greater writing accuracy (M = 31.00, SD = 23.17) than the open-ending group (M = 28.51, SD =23.66). One-way MANOVA on the effect of prompt type on the three indexes of writing performance shows a marginal, significant multivariate main effect of prompt type on writing performance (Wilks’ λ = .96, F (3,178) = 2.71, p = .047). Follow-up one-way ANOVAs (see Table 5) show prompt type made a significant difference only in fluency (F (1,180) = 4.89, p = .028), but not in accuracy rate or overall quality. Specifically, the open-ended prompt led students to write a longer essay than the closed-ended prompt. However, when Bonferroni adjustment (dividing the original alpha level by 3) was adopted 36.

(47) to protect against inflated type one error rate due to multiple testing, the effect of prompt type on writing fluency became non-significant, given the adjusted α of .017 (i.e, .05/3).. Table 5 Follow-up One-way ANOVAs of Writing Performance by Prompt Type Partial Eta SS Fluency. Between Within. Accuracy Between Within Overall Quality. Between Within. df. MS. F. 7351.49. 1. 7351.49. 270855.52. 180. 1504.75. 281.77. 1. 281.77. 98718.05. 180. 548.43. .84. 1. .84. 2076.47. 180. 11.54. p. Squared. 4.87. .028. .026. .51. .474. .003. .07. .788. .000. Results on the Interaction Effect of Writing Prompts and Proficiency To inspect if the two types of picture prompts interact with proficiency levels over writing performance (or if the effect of writing prompts is moderated by students’ proficiency level), the participants were first divided into high- and low-proficiency groups. The participants’ scores on the shortened 2003-GSAT were calculated; the results yielded a range of 70 to 11 points and a mean score of 38.67 (SD = 11.88), which was taken as the cutoff score. The participants who scored below the cutoff score were categorized as low achievers, while those who scored above the cutoff score were categorized as high achievers. The low achievers, consisting of 93 students, produced a group mean score of 29.26 and a standard deviation of 6.62 on the shortened 2003-GSAT. The high achievers, made up of 89 students, produced a group mean score of 48.51 and a standard deviation of 7.29 on the same test. Table 6 presents the means and standard deviations of writing fluency, accuracy, and overall quality by prompt type and proficiency level. 37.

(48) Table 6 Descriptive Statistics of Writing Fluency, Accuracy, and Overall Quality by Prompt Type and Proficiency Level Prompt Type Fluency. Closed-ending. Open-ending. Accuracy. Closed-ending. Open-ending. Overall Quality. Closed-ending. Open-ending. Proficiency Level. M. SD. N. High. 141.00. 32.53. 43. Low. 115.64. 33.59. 47. High. 152.89. 39.87. 46. Low. 128.04. 40.65. 46. High. 38.84. 25.31. 43. Low. 23.83. 18.54. 47. High. 36.59. 24.70. 46. Low. 20.44. 19.71. 46. High. 9.81. 3.16. 43. Low. 6.72. 2.60. 47. High. 10.13. 3.08. 46. Low. 6.54. 3.00. 46. A two-way MANOVA was employed to detect whether there was a statistically significant interaction between prompt type and proficiency level on writing performance. The results reveal an insignificant interaction between the prompt type and proficiency level (Wilks’ λ = 1.00, F (3,176) = .23, p = .878). On the other hand, the main effect of prompt type was insignificant, though at a marginal level (Wilks’ λ = .96, F (3,176) = 2.66, p = .050), while the main effect of proficiency level was significant (Wilks’ λ = .75, F (3,176) = 19.19, p = .000). Since there was a significant main effect of proficiency level, a series of one-way. 38.