臺灣大學生篇章凝結性分析:質與量的研究

全文

(1)國立台灣師範大學英語學系碩士論文 Mater Thesis Graduate Institute of English National Taiwan Normal University. 台灣大學生篇章凝結性分析:質與量的研究 Textual Cohesion in Taiwanese College Students’ English Writing: A Quantitative and Qualitative Study. 指導教授:張武昌博士 Advisor: Dr. Vincent W. Chang 研究生:曾玉婷. 中華民國九十八年七月 July, 2009.

(2) 摘要過去的研究主要探討篇章凝結及寫作優劣之間的關係，然而研究學者對高低成就寫作者凝結詞的使用有無差異性則有不同看法。此外，少有學者在研究凝結詞時把文體也列為變項，尤其在臺灣，尚無學者比較不同能力的第二語言學習者在不同文體裡凝結詞的使用，因此，此研究目標試著以質與量化的方式，探討臺灣大學生不同能力及寫作不同文體下，篇章凝結性的研究。一共有六十個臺灣師範大學的大學生參與這份研究，所有的人都必須寫敘述文及說明文，並依據寫作的分數來分高成就及低成就組。之後，計算高低組文章所有凝結詞並依據 Halliday &Hasan’s (1976), Stostky (1983)及 Hasan’s (1985)三人凝結詞的架構來做為研究分析的新架構。此外，不適當的凝結詞也會被挑出來進行寫作者錯誤的質性分析。主要研究的發現可從二方面來做摘要—量化及質化。量化方面，首先，高低成就寫作者在凝結詞的總數上有顯著差異，其顯示能力是主要影響凝結性多寡的要素。此外，高成就組裡，指涉詞(reference)比例最多，接著是連接詞(conjunction)、取代詞(substitution)及省略詞(ellipsis)，低成就組裡也可以找到一樣的比例順序。另外，在敘述文或說明文裡，高成就者比低成就者用比較多的同義字(synonym)，第二，文體比較上，高低二組在二文體中凝結詞的總數沒有顯著差異。此外，二組在敘述文裡都使用較多的人稱指涉詞(pronominal reference)，此現象或許顯露出文體可能對作者指涉詞的使用有影響，因為在敘述文裡，角色的闡述說明是不可欠缺的，及故事的發展主要以角色為中心發展。另外，在說明文裡有較多的反義詞 (antonym)被使用，這可能也顯示出文體對反義詞的使用有影響，因為在說明文裡，從不同的角度來說明主題，進而支持論點是很重要的。質化方面，首先，在敘述文及說明文，低成就者比高成就者有更多不適當的凝結詞錯誤，低成就者因基礎文法結構和基本用字拼讀很弱，導致於他們寫出的凝結詞上都帶有錯誤。第二，在敘述文裡，高低成就主要問題在於欠缺及不必要定冠詞的使用，而在說明文裡，高低成就主要問題在說代名詞的使用上出現無法辯認的指 i.

(3) 涉詞。至於詞彙凝結性方面，二組都有用字問題。第三，文體比較方面，文體對高低成就的錯誤沒有影響力。總結來說，程度是主要影響凝結詞總數的指標，而文體差異在寫作者凝結性的使用上沒有很大差異。. ii.

(4) Abstract Previous research mainly concerns the connection cohesion and coherence have with writing quality; yet, researchers of the cohesion analysis have not reached agreement as to whether there are significant differences in high and low proficiency groups’ cohesive devices. Further, few have conducted research into cohesion regarding the written mode as a variable, especially within a Taiwanese context, where none has been focused on the comparison of L2 students’ employment of cohesive devices, especially writers of different levels and in different genres. The aim of this study is thus to explore the employment by Taiwanese college students of cohesive devices in writing in terms of distinct proficiency levels and different genres quantitatively and qualitatively. A total of 60 college students from National Taiwan Normal University participated in the study. All participants were required to compose narrative or expository writing and consequently divided into high and low proficiency groups based on their writing scores. Subsequently, the total number of cohesive devices contained in both high and low proficiency writers’ writing was calculated and analyzed with the combined model of Halliday and Hasan’s (1976), Stotsky’s (1983) and Hasan’s (1985) taxonomy of cohesion. Additionally, any inappropriate use of cohesive devices was identified to conduct the qualitative analysis of the participants’ errors. The major findings of the study can be summarized in respect of two aspects—the quantitative and the qualitative. Quantitatively, first, there were significant differences in high and low proficiency writers’ overall number of cohesive devices both in narration and in exposition, which revealed that proficiency level was the predominant factor influencing the numerous cohesive devices. Also, it was found that in writing from the high proficiency group, reference was the dominant cohesive device, followed by conjunction, substitution and ellipsis in narration and in exposition. The same order of frequency also appeared in low proficiency groups’ writing. Besides, high proficiency iii.

(5) writers used more synonym than low proficiency ones in narration and in exposition. Secondly, there were no significant differences in both high and low proficiency writers’ overall number of cohesive devices between narrative and expository writing. Further, both groups used more pronominal reference in narration than in exposition, which perhaps revealed that the discourse type may influence a writers’ employment of referential cohesive devices since in narrative writing, the elaboration of characters is indispensable and the development of the story mainly revolves around the characters. Additionally, more antonym was found in exposition which probably shows that genres affect the use of antonym because in the expository writing explaining the topic from different sides to support a thesis is essential. Turning to the qualitative aspect, first, in narration and in exposition, low proficiency writers had much more inappropriate use of cohesive devices than their high proficiency counterparts. Low proficiency writers were found to be suffering deficiency in terms of basic grammatical structures and basic vocabulary skills (e.g., spellings, plural inflection) leading them to produce more inappropriate use of cohesive devices. Secondly, in narration, both groups had major problems with the lack/addition of definite articles in the text. Meanwhile in exposition, they both had problems in unclear referent of the pronominal references. With regard to lexical cohesive devices, both groups had major problems with word choice. Thirdly, concerning genre, within the errors made by high proficiency writers there was no difference between narration and exposition and it was also found that genres made no impact on the low proficiency group’s inappropriate use of cohesive devices. To conclude, the proficiency level was the dominant factor influencing the distribution of the cohesive devices and genre difference has no great influence on the writers’ use of cohesive devices.. iv.

(6) Acknowledgements The completion of this thesis constitutes not only an arduous process but also an unforgettable life experience personally. I would never have made it through without the support from my teachers, friends, and my beloved family. First of all, I would like to express the deepest gratitude to my advisor, Prof. Vincent W. Chang, who spent so much time reviewing my thesis and offered me insightful suggestions and feedback during each meeting. Also, I would also like to thank Dr. Hsueh-o Lin, and Dr. Hsueh-ying Yu, two of the oral committee members of this paper. Their valuable comments and encouragement made the completion of this thesis possible. I also have to thank Prof. Jen-i Li, Meilan Lo, Hsi-chin Chu, and Larry Hong-lin Li, who allowed me to collect students writing in their class or encouraged their students to attend my study. In addition, I am greatly indebted to Prof. Ho-ping Feng and Jean Curran, both of whom are experienced writing teachers and kindly offered their help with the grading and correction of the collected student samples. Without their kindness and assistance, the study would not have succeeded. Secondly, this thesis could not have been completed without my friends’ accompany and support. I would like to express my greatest appreciation for my friend, Terrisa Lin, whose assistance with the statistical methods facilitated the analysis of the study. Also, I would like to thank Jeromey, Sandra, Fay, Elain, Fiona and Kelly, who kept pushing me forward when I was working on my thesis and who were able to empathize with the pressure and burden brought about by thesis-writing. Finally, I would like to thank my parents for their love and concern when I was struggling with my thesis. They allowed me to concentrate on my study in a stress-free environment. Therefore, I would like to share the joy from this achievement with my beloved parents.. v.

(7) TABLE OF CONTENTS CHINESE ABSTRACT. i. ENGLISH ABSTRACT ACKNOWLEDGEMENTS TABLE OF CONTENTS LIST OF TABLES. iii v vi ix. 1. INTRODUCTION………………………………………………………… 1.1 Background and Motivation…………………………………………… 1.2 Significance of the Study……………………………………………… 1.3 Definition of Terms……………………………………………………... 1 1 2 4. 1.3.1 Definition of Narration…………………………………………... 1.3.2 Definition of Exposition………………………………………… 1.4 Research Questions……………………………………………..……… 2. LITERATURE REVIEW……………………………………………….…. 4 5 7 8 8. 2.1 Cohesion Theory……………………………………………………….. 2.1.1 Grammatical Cohesive Devices……...…………………………... 2.1.2 Lexical Cohesive Devices……………………...……………….... 2.2 Critique of Cohesion Theory…………………………………………. 2.2.1 Modification of Cohesive Scheme by Stotsky……………………. 9 13 16 19. 2.2.2 Modification of Cohesive Scheme by Hasan……………………. 2.3 Previous Studies on Cohesion in Writing……………………………… 2.3.1 Studies of Writing Involving Different Proficiency Groups…….. 2.3.2 Studies of Writing across Different Genres………………………. 20 22 23 30. 2.4 Summary……………………………………………………………….. 3. METHOD………………………………………………………………….. 39 42. 3.1 Participants…………………………………………..………………… 3.2 Instruments……………………………………………..………………. 42 42. 3.3 Pilot Study…………………………...………………………………… 3.4 Procedures of the Formal Study……………………………………….. 3.4.1 Participants and Data Collection…...……………………………. 3.4.2 Coding Scheme of Cohesive Devices...………………………….. 3.4.3 Data Analysis…………………….……………………………….. 43 46 46 47 51. 4. RESULTS and DISCUSSION……………………………………………… 4.1 Quantitative Analysis…………………………………………………... 4.1.1 High and Low Proficiency Groups in Narrative Writing…………. 53 53 53. 4.1.1.1 High and Low Proficiency Groups’ Use of Grammatical Cohesive Devices………………………………………… 4.1.1.2 High and Low Proficiency Groups’ Use of Lexical. 54. vi.

(8) Cohesive Devices………………………………………… 4.1.2 High and Low Proficiency Groups in Expository Writing……… 4.1.2.1 High and Low Proficiency Group’s Use of Grammatical Cohesive Devices………………………………………… 4.1.2.2 High and Low Proficiency Group’s Use of Lexical Cohesive Devices …………………………………..…….. 59 61. 4.1.3 Summary and Discussion…………………………...…………… 4.1.4 High Proficiency Group in Narrative and Expository Writing…... 4.1.5 Low Proficiency Group in Narrative and Expository Writing…… 4.1.6 Summary and Discussion……………………………………….... 62 65 69 73. 4.2 Qualitative Analysis…………………………………………………… 4.2.1 High and Low Proficiency Groups’ Inappropriate Use of Cohesive Devices in Narrative Writing…………………………... 4.2.1.1 High Proficiency Group’s Inappropriate Use of. 75. Grammatical Cohesive Devices in Narrative Writing……. 4.2.1.2 High Proficiency Group’s Inappropriate Use of Lexical Cohesive Devices in Narrative Writing…………………... 4.2.1.3 Low Proficiency Group’s Inappropriate Use of Grammatical Cohesive Devices in Narrative Writing……. 4.2.1.4 Low Proficiency Group’s Inappropriate Use of Lexical Cohesive Devices in Narrative Writing…………………... 4.2.2 High and Low Proficiency Groups’ Inappropriate Use of Cohesive Devices in Expository Writing…………………………. 76 76 78 80 84 87. 4.2.2.1 High Proficiency Group’s Inappropriate Use of Grammatical Cohesive Devices in Expository Writing….... 87. 4.2.2.2 High Proficiency Group’s Inappropriate Use of Lexical Cohesive Devices in Expository Writing…………………. 91. 4.2.2.3 Low Proficiency Group’s Inappropriate Use of Grammatical Cohesive Devices in Expository Writing…... 4.2.2.4 Low Proficiency Group’s Inappropriate Use of Lexical Cohesive Devices in Expository Writing…………………. 4.2.3 High Proficiency Group’s Inappropriate Use of Cohesive. 5.. 57 58. 92 97. Devices in the Two Genres……………………………………….. 4.2.4 Low Proficiency Group’s Inappropriate Use of Cohesive Devices in the Two Genres…………………………………...…………….. 100. CONCLUSIONS…………………………………………………………. 5.1 Summary of the Quantitative Analysis………………………………… 5.2 Summary of the Qualitative Analysis………………………………....... 106 106 109. vii. 103.

(9) 5.2.1 High and Low Proficiency Groups’ Inappropriate Use of Cohesive Devices in Narrative Writing…………………………... 109. 5.2.2 High and Low Proficiency Groups’ Inappropriate Use of Cohesive Devices in Expository Writing………………………… 5.2.3 High Proficiency Group’s Inappropriate Use of Cohesive Devices in Two Genres……………………..……..…………….... 113. 5.2.4 Low Proficiency Group’s Inappropriate Use of Cohesive Devices in Two Genres……………………………………..……………... 5.3 Pedagogical Implications………………….…………………………… 5.4 Limitations and Suggestions for Future Research……………………... 114 115 122. REFERENCES………………………………………..………………………. Appendix A: TOFEL Scoring Guide…………………………………………... Appendix B: Writing Prompts…………………………………………..……... Appendix C: Scores of Writing Samples………………………………………. Appendix D: Basic Information of the Narrative Samples……………………... 124 132 134 135 136. Appendix E: Basic Information of the Expositive Essays……………………... 137. Appendix F: High and Low Proficiency Level Writers’ Distribution of Grammatical and Lexical Cohesive Devices in Narration………. 138. Appendix G: High and Low Proficiency Level Writers’ Distribution of Grammatical and Lexical Cohesive Devices in Exposition…….. Appendix H: Tables of Quantitative Analysis…………………………………. Appendix I: Samples…………………………………...……………………….. 139 140 148. viii. 111.

(10) LIST OF TABLES Table 1. High and Low Groups’ Use of Cohesive Devices in Narration and Exposition……………………………………………………. Table 2. Comparison of Total Number of Cohesive Devices in Narration between Two Groups……………………………….…………….. Table 3. Distribution. of. Grammatical. Cohesive. Devices. Table 8. 56. Comparison of Numbers of Substitution in the Two Groups in Narration………………………………………………………….. Table 7. 56. Comparison of Numbers of Reference in the Two Groups in Narration………………………………………………………….. Table 6. 55. in. Narration…………………………………………………………. Table 5. 54. Grammatical and Lexical Cohesive Devices in High and Low Groups in Narration……………………………………………..... Table 4. 54. 56. Comparison of Numbers of Conjunction in the Two Groups in Narration…………………………………………………………. 57. Comparison of Numbers of Pronominal Reference in the Two. 140. Groups in Narration……………………………………………… Table 9. Comparison of Numbers of Definite Articles in the Two Groups. 140. in Narration………………………………………………………. Table 10. Comparison of Numbers of Comparative Reference in the Two Groups in Narration………………………………………………. Table 11. Distribution of Sub-forms of Conjunctive Cohesive Devices in Narration…………………………………………………………. Table 12. 140. Comparison of Numbers of Additive Conjunction in the Two Groups in Narration………………………………………………. Table 13. 140. Comparison of Numbers of Adversative Conjunction in the Two. ix. 141.

(11) Groups in Narration……………………………………………... Table 14. Comparison of Numbers of Causal Conjunction in the Two Groups in Narration………………………………………………. Table 15. 141. 141. Comparison of Numbers of Temporal Conjunction in the Two Groups in Narration…………………………………………….... 141. Table 16. Distribution of Lexical Cohesive Devices in Narration…………. 57. Table 17. Comparison of Numbers of Repetition in the Two Groups in. 58. Narration………………………………………………………… Table 18. Comparison of Numbers of Synonym in the Two Groups in Narration…………………………………………………………. Table 19. Comparison of Numbers of Antonym in the Two Groups in Narration…………………………………………………………. Table 20. 58. Comparison of Total Number of Cohesive Devices in Exposition between Two Groups…………………………………………….. Table 22. 58. Comparison of Numbers of Superordinate in the Two Groups in Narration…………………………………………………………. Table 21. 58. 59. Grammatical and Lexical Cohesive Devices in High and Low Groups in Exposition……………………………………………. 59. Table 23. Distribution of Grammatical Cohesive Devices in Exposition…. 60. Table 24. Comparison of Numbers of Reference in the Two Groups in. 60. Exposition……………………………………………………… Table 25. Comparison of Numbers of Substitution in the Two Groups in Exposition………………………………………………………. Table 26. 60. Comparison of Numbers of Ellipsis in the Two Groups in Exposition. 60. x.

(12) Table 27. Comparison of Numbers of Conjunctions in the Two Groups in Exposition………………………………………………………. Table 28. Distribution of Sub-forms of Conjunctive Cohesive Devices in Exposition………………………………………………………. Table 29. 142. Comparison of Numbers of Causal Conjunction in the Two Groups in Exposition……………………………………………. Table 32. 142. Comparison of Numbers of Adversative Conjunction in the Two Groups in Exposition…………………………………………….. Table 31. 142. Comparison of Numbers of Additive Conjunction in the Two Groups in Exposition……………………………………..………. Table 30. 61. 142. Comparison of Numbers of Temporal Conjunction in the Two Groups in Exposition…………………………………………….. 143. Table 33. Distribution of Lexical Cohesive Devices in Exposition…………. 61. Table 34. Comparison of Numbers of Repetition in the Two Groups in Exposition………………………………………………………. Table 35. Comparison of Numbers of Synonym in the Two Groups in Exposition………………………………………………………. Table 36. 62. Comparison of Numbers of Superordiante in the Two Groups in Exposition………………………………………………………. Table 38. 62. Comparison of Numbers of Antonym in the Two Groups in Exposition………………………………………………………. Table 37. 62. 62. High Proficiency Group’s Cohesive Devices in Narrative and Expository Writing……………………..……………………….. 65. Table 39. High Proficiency Group’s Cohesive Devices in the Two Genres. 65. Table 40. High Proficiency Group’s Grammatical and Lexical Cohesive. xi.

(13) Devices in the Two Genres……………………………………… Table 41. 66. High Proficiency Group’s Grammatical Cohesive Devices in the Two Genres………………………………………………………. 67. Table 42. High Proficiency Group’s Reference in the Two Genres………. 67. Table 43. High Proficiency Group’s Substitution in the Two Genres………. 67. Table 44. High Proficiency Group’s Ellipsis in the Two Genres…………. 67. Table 45. High Proficiency Group’s Conjunction in the Two Genres………. 67. Table 46. High Proficiency Group’s Pronominal Reference in the Two Genres…………………………………………………………….. 143. Table 47. High Proficiency Group’s Definite Articles in the Two Genres…. 143. Table 48. High Proficiency Group’s Comparative Reference in the Two Genres……………………………………………………………. Table 49. High Proficiency Group’s Distribution of Sub-forms of Conjunctive Cohesive Devices in the Two Genres………………. Table 50. Table 55. 145. High Proficiency Group’s Temporal Conjunction in the Two Genres……………………………………………………………. Table 54. 144. High Proficiency Group’s Causal Conjunction in the Two Genres............................................................................................. Table 53. 144. High Proficiency Group’s Adversative Conjunction in the Two Genres……………………………………………………………. Table 52. 144. High Proficiency Group’s Additive Conjunction in the Two Genres……………………………………………………………. Table 51. 143. 145. High Proficiency Group’s Lexical Cohesive Devices in the Two Genres……………………………………………………………. 68. High Proficiency Group’s Repetition in the Two Genres………... 68. xii.

(14) Table 56. High Proficiency Group’s Synonym in the Two Genres…………. 68. Table 57. High Proficiency Group’s Antonym in the Two Genres…………. 68. Table 58. High Proficiency Group’s Superordinate in the Two Genres……. 69. Table 59. Low Proficiency Group’s Cohesive Devices in Narration and Exposition……………………………………………..………….. 69. Table 60. Low Proficiency Group’s Cohesive Devices in the Two Genres. 70. Table 61. Low Proficiency Group’s Grammatical and Lexical Cohesive Devices in the Two Genres………………………………………. Table 62. Low Proficiency Group’s Grammatical Cohesive Devices in the Two Genres………………………………………………………. Table 63. 70. 71. Low Proficiency Group’s Reference Cohesive Devices in the Two Genres………………………………………………………. 71. Table 64. Low Proficiency Group’s Substitution in the Two Genres………. 71. Table 65. Low Proficiency Group’s Conjunction in the Two Genres………. 71. Table 66. Low Proficiency Group’s Pronominal Cohesive Devices in the Two Genres………………………………………………………. 145. Table 67. Low Proficiency Group’s Definite Articles in the Two Genres…. 145. Table 68. Low Proficiency Group’s Comparatives in the Two Genres……. 145. Table 69. Low Proficiency Group’s Distribution of Sub-forms of Conjunctive Cohesive Devices in the Two Genres………………. Table 70. Low Proficiency Group’s Additive Conjunction in the Two Genres……………………………………………………………. Table 71. Table 72. 146. 146. Low Proficiency Group’s Adversative Conjunction in the Two Genres……………………………………………………………. 146. Low Proficiency Group’s Causal Conjunction in the Two. 146. xiii.

(15) Genres….. Table 73. Low Proficiency Group’s Temporal Conjunction in the Two Genres……………………………………………………………. Table 74. 147. Low Proficiency Group’s Lexical Cohesive Devices in the Two Genres……………………………………………………………. 72. Table 75. Low Proficiency Group’s Repetition in the Two Genres……….... 72. Table 76. Low Proficiency Group’s Synonym in the Two Genres…………. 72. Table 77. Low Proficiency Group’s Antonym in the Two Genres………….. 72. Table 78. Low Proficiency Group’s Superordinate in the Two Genres…….. 72. Table 79. Number of Inappropriate Use of Grammatical Cohesive Devices between Two Groups……………………………………………... Table 80. Number of Inappropriate Use of Lexical Cohesive Devices between Two Groups……………….……………………………. Table 81. 76. Inappropriate Use of Grammatical Cohesive Devices between the Two Groups……………………………..………………………. Table 82. 76. 87. Number of Inappropriate Use of Grammatical and Lexical Cohesive Devices………………………………………………... 87. Table 83. Number of Inappropriate Use of Lexical Cohesive Devices …….. 87. Table 84. High Proficiency Group’s Inappropriate Use of Grammatical Cohesive Devices ………………………….……………………. Table 85. High Proficiency Group’s Inappropriate Use of Grammatical Cohesive Devices …………………………….…………………. Table 86. 102. High Proficiency Group’s Inappropriate Use of Lexical Cohesive Devices ………………………………………………………….. Table 87. 102. Low Proficiency Group’s Inappropriate Use of Grammatical. xiv. 103.

(16) Cohesive Devices ……………….……………………………… Table 88. Low Proficiency Group’s Inappropriate Use of Grammatical Cohesive Devices …………….…………………………………. Table 89. 105. 105. Low Proficiency Level Writers’ Inappropriate Use of Lexical Cohesive Devices ………………………………………………... xv. 105.

(17) CHAPTER ONE INTRODUCTION 1.1 Background and Motivation Writing is more than just putting down words, constructing a sequence of sentences or paragraphs on the paper because these elements may not be interpreted if readers cannot understand the meaning or ideas expressed within them. Indeed, even cognitive psychologists and psycholinguists who have studied complex processes in four skills agree that producing language is much more difficult than receiving it (Ahring, 1979-1980). Therefore, to edit a piece of good writing is never an easy task. During the process of composing, a writer needs to choose appropriate lexicon, structures, and organizations to make the idea or meaning flow smoothly in the whole discourse. Among the factors that may influence the writing quality, cohesion and coherence play primary roles. Cohesion refers to the semantic relation in a text that makes the text coherent. Through the arrangement of grammatical and lexical devices in the text, the semantic meaning between sentences in the paragraphs is made clear, which further renders the writer’s ideas more cohesive and easier to be interpreted. Coherence depends on the cohesion within the text and the context of the situation. A text without cohesion and coherence is just like fragments with separate ideas dislocated in different places. In writing, it is very important to employ certain devices to glue these separated ideas together and make it a readable text. A text, according to Halliday and Hasan (1976), is a ‘SEMANTIC’ unit, a unit not of form but of meaning. Further, the term ‘texture’ is taken as an essential property to differentiate a text from non-text. In other words, Halliday and Hasan refer to the term ‘texture’ as a kind of text property in place of the well-known term ‘coherence’. Despite the fact that the importance of cohesion and coherence in writing can never be neglected, many EFL writers still don’t have full control of these cohesive devices; 1.

(18) therefore, much incoherent writing can be attributed to the inappropriate use of cohesive items. This phenomenon has motivated me to conduct a study in the area of discourse analysis. Previous research on cohesion and coherence in written discourse mainly concerns the relationship of cohesion and coherence with regard to writing quality, the distribution of these cohesive ties employed by writers of different proficiency levels, across different grade levels or the relationship between cohesion and reading comprehension. In spite of the abundance of resources available in research into cohesion in written discourse, little research has been conducted into cohesion regarding the written mode as a variable which influences the distribution of cohesive devices, especially in the context of Taiwan, where none has been concerned with the comparison of the L2 students’ employment of cohesive devices by writers of different levels and in different genres. This phenomenon inspires me to conduct a study by examining different cohesive devices in EFL students’ writing with different proficiency levels and different genres in order to detect whether learners of different proficiency levels will differ in their usage of cohesive ties in two modes of writing—narration and exposition. Further, cohesive errors of high and low proficiency groups and of different genres will be calculated and examined quantitatively and qualitatively to bring enlightenment and implication for the composition course. 1.2 Significance of the Study The purpose of the study is to investigate the use of cohesion devices in college students’ writing at different proficiency levels and in different genres. Lack of cohesion has been one of the problems preventing writers from developing clear pieces of writing. As a result of this, it is essential to scrutinize how EFL writers try to connect their ideas by the employment of words and phrases—the so-called “cohesive 2.

(19) devices” to make their writing flow smoothly and logically. To be more specific, it is worth examining whether writers of different levels display any disparity in their use of cohesive devices in their writing. Unlike previous studies, this study aims at different proficiency writers' cohesive performance. To highlight difference in cohesion between proficiency groups, writing by English majors is compared with work by non-English majors and participants range across three grades. The occurrence of each cohesive device will be numbered to make a comparison between the cohesive performance of high and low proficiency writers. The frequency count of grammatical and lexical cohesive devices will reveal different writers' ability to use various cohesive devices. Aside from the statistical counting of cohesive items, qualitative analysis of the writing is indispensable since high frequency of cohesive devices alone does not guarantee the quality of a piece of writing. Therefore, qualitative analysis is needed to identify common errors or inappropriate cohesive ties and to distinguish between good writers and poor ones, from the perspective of discourse. In addition, different discourse types are taken into consideration when it comes to the employment of the cohesion device in writing, especially in the context of Taiwan, where a direct comparison of the cohesive patterns between the two genres has not been made. Among the various discourse types, narration and exposition are commonly-seen and familiar to college students as they have being trained to write stories or expositions since senior high school. Therefore, these two genres are chosen to examine how high and low proficiency writers employ cohesive devices differently in composing narration or exposition. Quantitatively, comparison in the number of cohesive devices in each genre is indispensable. Qualitatively, the examination of cohesive errors between genres is necessary to identify writers' problems. Through detailed analysis of college students' writing, we will understand different proficiency 3.

(20) writers' manipulation of cohesive devices and the variation of errors with regard to genre. The discussion of the result will raise writing teachers' awareness of the strengths and weaknesses of different writers and serves as a guideline for helping writers use cohesive devices with accuracy. In addition, the exploration of genres in the study will offer a chance to examine the correlation of discourse types and use of cohesive devices. The patterns of cohesive error types will make teachers aware of common errors students are likely to make and teaching implication will be provided as well to improve EFL writers' writing and shed some light on writing instruction. 1.3 Definition of Terms 1.3.1 Definition of Narration According to Toolan (2001), “narrative typically is a recounting of things spatiotemporally distant” (p. 1); that is, narratives involve the recall of happenings that may be not merely spatially, but, more crucially, temporally remote from the teller and his audience. The framework of narrative structure, according to Labov (1972), consists of six elements. They are abstract, orientation, complicating action, evaluation, resolution, and coda (Labov, 1972, p. 360; Toolan, 2001, p. 148). Evaluation. Resolution. Complicating action. Coda. Orientation. (Abstract) 4.

(21) 1. Abstract: what was this story about? 2. Orientation: who, when, what, where? 3. Complicating action: then what happened? 4. Evaluation: so what? 5. Results or resolution: what finally happened? 6. Coda: That’s it. I’ve finished and am “bridging” back to our present situation. In Labov’s framework of narrative structure, it is optional to have the abstract to summarize or encapsulate the point of the story in the beginning of the narrative. Besides, the abstract normally outlines or advertises the story that a teller intends to follow. Next, as the story unfolds, it is necessary to identify in some way the place, persons, and their activity or the situation, which constitutes the section of orientation in the narration. As for the complicating action, it is the required or obligatory nucleus in a narrative because it tells us what happened in the story (Toolan, 2001). After the narration of the events comes the evaluation stage. The purpose of evaluation is to indicate the point of the narrative. To put it differently, the narrator is required to tell the reader why these events of the narrative are reportable or worth reporting. After the evaluation comes the resolution of the narration; during this stage, the narrator must describe what the outcome of the story is and it may be regarded as the termination of that series of events. Finally, the narrator proceeds to the coda, whose function is to close off the sequence of complicating actions and to signal that the narrative is finished. Coda serves to bridge the gap between the moment of time at the end of the narrative and the present. It will bring the narrator or listeners back to the point at which they entered the narrative. As a result of this, the time reference of the discourse will be shifted from the past to the present tense. 1.3.2 Definition of Exposition Exposition is a genre that differs from narratives (Carroll, 2002). According to Longacre and Levinsohn (1977), time is not relevant in expository writing, though a 5.

(22) variety of expository discourse exists involving explanations of a future stage of events. In addition, it is not uncommon to find argumentation in expository writing since a conceptual or logical development exists within the expository text. Black (1985) defined expository texts as those that convey new information and explain new topics to people. Elis (2004) also provided his definition of exposition as follows. Rhetorical tasks require writers to explain or clarify the topic/subject. In general terms, exposition is entailed in expressing ideas, opinions, or explanation pertaining to a particular piece of knowledge or fact. (p. 28) A rough comparison between the structure of exposition and that of a narrative (recount) is displayed below: Stage. An Exposition Example. Thesis. A good teacher needs to be Orientation. On Tuesday we went on a. understanding to all children.. harbor cruise.. Argument. Stage. He or she must be fair and Events. A Narration Example. in We went underneath the. reasonable. The teacher must Chronological harbor bridge and then work at a sensible pace. The Order. we. went. past. some. teacher also needs to speak with a. submarines. When we got. clear voice so the children can. to Clifton Gardens we. understand.. had a picnic. After we had finished we played on the climbings. Then Mr. Robinson came over and said Mr. Moses was giving. out. frozen. oranges. Then after we finished that we went home. Conclusion. That’s what I think a good Personal teacher should be like.. Comment (optional). Source: Board of Studies, 1988b: 287 6. It was a nice day out..

(23) 1.4 Research Questions This study aims to examine cohesion in Taiwanese college students’ English writing quantitatively and qualitatively by adopting three instruments: (a) TOEFL scoring scheme, (b) a combination model of Halliday and Hansan's (1976), Stotsky’s (1983) and Hasan’s (1985) taxonomy of cohesion and (c) T-test. First, the focus is on comparison between high and low proficiency writers in their use of cohesive devices in narration and followed by exposition. Second, the focus is on comparison of high and low proficiency writers’ use of cohesive devices between narration and exposition. Four research questions are designed as follows. 1. Is there any difference in high and low proficiency level students’ use of cohesive devices in narrative writing? 2. Is there any difference in high and low proficiency level students’ use of cohesive devices in expository writing? 3. Is there any difference in high proficiency level students’ use of cohesive devices between narration and exposition? 4. Is there any difference in low proficiency level students’ use of cohesive devices between narration and exposition?. 7.

(24) CHAPTER TWO LITERATURE REVIEW To explore whether different proficiency writers will show differences in their employment of cohesion devices in different genres, relevant literature is reviewed in this chapter, including cohesion theory, criticism of cohesion theory and previous studies on cohesion in writing. Specifically, Halliday and Hasan’s cohesion theory (1976) is first introduced, followed by Stotsky‘s model (1983) and Hasan’s model (1985). Next, studies of writing on different proficiency groups and studies of writing across genres are presented. 2.1 Cohesion Theory Halliday and Hasan (1976) viewed ‘TEXT’ as a unified whole, a unit of meaning. A text has its ‘TEXTURE’ that distinguishes it from other non-text. To make a text possess the property of texture, certain linguistic features like cohesive devices may be present in the passage to make it a united whole and give it texture; that is, these linguistic features make a sequence of disconnected sentences constitute a cohesive relation. When any two sentences are said to be cohesive in a text, it doesn’t merely mean that these two sentences are related. Rather, it means that what readers presuppose can be fulfilled by relying on the presupposed element in the text. To be brief, the cohesive relation is not merely achieved by the presence of the presupposition. The presupposed element must be present either in the preceding sentence or the following one to satisfy the presupposition. As stated by Halliday and Hasan, cohesion is a semantic unit and it occurs where the INTERPRETATION of the presupposing element in the discourse is dependent on that of another presupposed one. Cohesive relations can be realized either through grammar or through vocabulary, which can be referred to as grammatical cohesion and lexical cohesion 8.

(25) respectively. Take the following sentences for example. (2.1) Wash and core six cooking apples. Put the apples into a fireproof dish. (Halliday & Hasan, 1976, p. 3) At least, two cohesive ties can be detected from the sentences above. One is grammatical and the other a lexical cohesive tie. The grammatical cohesive tie is evidenced by the anaphoric function of reference the, which refers back to the six cooking apples in the previous sentence while lexical cohesion is achieved by the repetition of the noun apples, which produces the cohesive effect between the two lines. 2.1.1 Grammatical Cohesive Devices Both grammatical and lexical cohesion devices can be further categorized. First of all, grammatical cohesion devices can be classified as follows. 1. Reference-The cohesive function of references in discourse can mainly be divided into two types: exophora and endophora. Exphoric references are references that must be made according to the context of the situation, while endophoric references are those that can retrieve information made in the text. In other words, exphoric references are situational references, and endophoric references are textual references and both differ in respect of whether or not the referential meaning is retrieved through the context or the text. Generally speaking, different registers display different amounts of exophoric references and the discourse type of speech normally involves a great deal of expohoric references. According to Halliday and Hasan (1976), only endophoric references contribute to the cohesion of the text because they contribute to the interpretation of the text. Reference classified as endophora can be further specified as anaphora or cataphora reference. The referent of anaphoric references is identified by looking back in the text while that of cataphoric references is identified by looking 9.

(26) forward in the text, as illustrated by the diagram below. Reference. [situational] exophora. [textual] endophora. [to preceding text] anaphora. [to following text] cataphora. As for the types of references, there are three types—personal, demonstrative, and comparative. (1) Personal reference: Personal reference consists of three types: (a) Personal pronouns: I, you, we, she, he, they, it, me, us, him, her, them, one…; (b) Possessive determiners: my, your, his, her, its and their; (c) Possessive pronouns: mine, yours, ours, theirs, his, hers... a. Personal pronoun (2.2) John has moved to a new house. He has built it last year. (Halliday & Hasan, 1976, p. 54) b. Possessive determiners (2.3) John has moved to a new house. His wife must be delighted with it. (Halliday & Hasan, 1976, p. 55) c. Possessive pronoun (2.4) That new house was John’s. I didn’t know it was his. (Halliday & Hasan, 1976, p. 55) (2) Demonstrative reference: The referent of demonstrative reference is established on a scale of proximity in time or in space. And it includes: this, that, these, those, here, there, the…. The following examples are taken from Cohesion in English. (Halliday & Hasan, 1976, p. 63) (2.5) Do you want to know the woman who designed it? That was Mary 10.

(27) Smith. (2.6) Who are those colorful characters?—Those must be the presidential guards. (3) Comparative reference: The comparative reference is indirect reference by means of IDENTITY or SIMILATIRY such as same, identical, similar, equal, other, else, better, less, more…. (2.7) It’s the same act as the one we say yesterday. (Halliday & Hasan, 1976, p. 78) The interpretation of the italic elements (e.g., He, His) in the examples of personal reference and demonstrative references relies on the presupposed element expressed in the preceding sentence, while the interpretation of the italic elements (e.g., same) in the example of comparative reference relies on the following sentence and both consequently, lead to the cohesion of the text. 2. Substitution-Different from reference, substitution is a relation between linguistic items such as words, phrases, or clauses. However, it does share some similarities with reference as it is anaphoric and the substitute item creates a cohesive link with the preceding text. Generally, there are three types of substitution: nominal, verbal, and clausal. (1) Nominal substitution (2.8) These biscuits are stale. Get some fresh ones. (Halliday & Hasan, 1976, p. 92) (2) Verbal substitution (2.9) John is smoking more now than he used to do. (Halliday & Hasan, 1976, p. 115) (3) Clausal substitution (2.10) Is there going to be an earthquake?—It says so. (Halliday & Hasan, 1976, p. 130) The ones in the first sentence anaphorically refers back to the biscuits in the previous sentence and creates a cohesive relation between the two sentences. As for the 11.

(28) cohesive device do in the second example, its specific action is recoverable from the preceding sentence. Finally, different from the nominal and verbal substitution, so in the third sentence substitutes for the whole clause there is going to be an earthquake in the previous sentence and forms a cohesive relation between the two sentences. 3. Ellipsis-Simply speaking, ellipsis is ‘substitution by zero.’ There are three types of ellipsis: nominal ellipsis, verbal ellipsis, and clausal ellipsis. (1) Nominal Ellipsis: Ellipsis within the nominal group is called nominal ellipsis. (2.11) Four other Oysters followed them, and yet another four. (Halliday & Hasan, 1976, p. 148) (2) Verbal Ellipsis: Ellipsis within the verbal group is called verbal ellipsis. (2.12) Have you been swimming?—Yes, I have. (2.13) What have you been doing?—Swimming. (Halliday & Hasan, 1976, p. 167) (3) Clausal Ellipsis: The structure of clause is composed of two elements—modal and propositional element. The modal element includes the subject plus the finite elements in the verbal group while the propositional element is the remainder of the verbal group. The following two examples show the ellipsis of a modal element and the ellipsis of a propositional element. (2.14) What was the Duke going to do?—Plant a row of poplars in the park. (Halliday & Hasan, 1976, p. 197) (2.15) Who was going to plant a row of poplars in the park?—The Duke was. (Halliday & Hasan, 1976, p. 198) 4. Conjunction-The conjunctive elements presuppose the presence of other components in the discourse so they indirectly construct the cohesive relationship inter-sententially. There are four types: additive, adversative, causal, and temporal sequence. (1) Additive (2.16) For the whole day he climbed up the steep mountainside, almost without stopping. And in all this time he met no one. 12.

(29) (Halliday & Hasan, 1976, p. 238) (2) Adversative (2.17) She failed. However, she’s tried her best. (Halliday & Hasan, 1976, p. 252) (3) Causal (2.18) …She felt that there was no time to be lost, as she was shrinking rapidly; so she got to work at once to eat some of the other bit. (Halliday & Hasan, 1976, p. 256) (4) Temporal sequence (2.19) Finally, we should record that the influence of the humanists contributed a good deal towards the final decay of the plainsong tradition. (Halliday & Hasan, 1976, p. 264) 2.1.2 Lexical Cohesive Devices When compared to grammatical cohesion, the effect of lexical cohesion is as not obvious and easy to estimate. Simply put, the cohesive relation each lexical item denotes cannot be detected unless it is by reference to the text. Similarly, each lexical item in the discourse will contribute to the texture of the text and the context or the environment will decide the textual meaning of the lexical item. To make the text cohesive in the discourse, the words in the discourse must be related to a certain degree. In other words, it is the occurrence of the item in the context of related lexical items that provides cohesion and gives to the passage the quality of text (Hallliday & Hasan, 1976). Lexical cohesive devices can be further categorized into two types: reiteration and collocation (Halliday & Hasan, 1976). In Halliday and Hasan’s framework, the category of reiteration includes (a) same word (repetition), (b) synonym (or near-synonym), (c) superordinate, and (d) general word. Repetition of the same lexical item creates a coherent relation in the text. In the following example, to further support the idea that there are children everywhere, the word children is repeated three times. The repetition of children all refer back to the word children in the first sentence and it is used to describe the fact that children are 13.

(30) enjoying various recreational facilities. (2.20) There are children everywhere. There are children on the swing, children on the slides, and children on the marry-go-around. (Halliday & Hasan, 1989, p. 81) Unlike repetition, synonyms are different phonological words which have the same or very similar meanings (Saeed, 1997). In other words, words which are synonyms have identical meanings but this does not mean that their meanings totally overlap with each other (Halliday & Hasan, 1987). Pairs like couch/sofa, boy/lad, lawyer/attorney etc. are synonymous ones. Under the classification of reiteration, it also includes superordinate; superordinate is a relation held between a general class and its sub-classes (Halliday & Hasan, 1987). That is, the word relation is a hierarchical one in that under the general words includes a series of words with a more specific meaning. The hierarchical semantic relation can be visualized by the following taxonomy of human artifacts: (Saeed, 1997, p. 69) tool. hammer. saw. chisel. hacksaw. jigsaw. The last sub-categorization of reiteration is the use of general nouns in the text. Although general nouns are lexical items, they are on the borderline between a lexical item and a grammatical item. Moreover, general nouns are usually accompanied by the reference item the, and the whole combination functions like an anaphoric reference item such as “Henry seems convinced there’s money in dairy farming. I don’t know what gave him that idea.” (Halliday & Hasan, 1976, p. 275) What exactly that idea refers to all depends on the preceding context. From the above example, we can infer that “that idea” refers anaphorically to “there’s money in dairy farming”. In 14.

(31) addition to the anaphoric function of general nouns, another function of general nouns is to express interpersonal attitude, especially those referring to human beings, by adding attitudinal modifiers as shown in the dears, the poor dears, the stupid thing, and the lucky fellow (Halliday & Hasan, 1976, p. 276). These expressions are used to convey familiarity, sympathy or occasionally contempt. Other examples of common general nouns are presented as follows: 1. Human noun: people, person, man, woman, child, boy, girl 2. 3. 4. 5. 6.. Non-human animate noun: creature Inanimate concrete count noun: thing, object Inanimate concrete mass noun: stuff Inanimate abstract noun: business, affair, matter Action noun: move. 7. Place noun: place 8. Fact noun: question, idea (Halliday & Hasan, 1976, p. 274) Aside from using reiteration to achieve cohesion in a text, collocation is the second type of lexical devices that can achieve the cohesive effect. The definition of collocation is “the association of lexical items that regularly co-occurs” (Halliday & Hasan, 1987, p. 284). The co-occurrence of the lexicon enables cohesion to form within the text. Further, as pointed out by Halliday and Hasan (1976), it is the most problematic part of lexical cohesion because in addition to the cover term ‘reiteration’, other lexical items that contribute to the cohesion of the text are subsumed under the category of collocation. Under the categorization of collocation, antonym such as cold/hot and like/hate are lexicon that will contribute to the cohesion of the text. The second type of collocation contains pairs of words drawn from the same ordered series. For instance, days of the week and measurements of currency. Another kind of collocation involves lexical items that are related in terms of the part-to-whole or part-to-part semantic relation. For example, the semantic relation between car/brake and box/lid are part-to-whole relation while mouth/chin, verse/chorus are part-to-part 15.

(32) relation. The last is that pairs of words may be the co-hyponyms of the same superordinate item such as chair/table etc. Even though the suggested framework provided above seems to serve as a panacea to understand the cohesion relation in the discourse, a limitation within this framework continues to exist. As Halliday and Hasan stated, there is no clear criterion for deciding the lexical cohesive relation within the text, for a lexical item is not bound to a particular grammatical category or to a particular morphological form (Halliday & Hasan, 1976, p. 291).Therefore, from time to time seeing clearly enough to determine whether a pair of words such as tooth and dental are counted as the same lexical item but in different morphological forms or are counted as different lexical items is not possible. The only way to solve the puzzle is to refer back to the text itself. Despite this limitation of indeterminacy in its application to actual instances, Halliday and Hasan’s cohesive framework is still essential to help us understand what makes a text cohesive. 2.2 Critique of Cohesion Theory In spite of the fact that Halliday and Hasan’s 1976 framework of cohesion theory was recognized in discourse analysis and was adopted by many researchers to analyze textual cohesion and coherence (McCulley, 1985; Norment, 1982; Neuner, 1987; Tierney & Mosenthal, 1983; Wittey & Faigley, 1981 and so on), critique of the framework continues to occur. Crowhurst in her 1987 study found that twelfth graders in the writing of argumentation had higher incidence of lexical repetition than sixth graders. In her analysis, Crowhurst discovered the employment of lexical repetition caused different situations to arise depending on which of two school grades a student belonged to. While older graders’ use of lexical repetition reflected their maturity to elaborate and summarize arguments; younger graders’ use of lexical repetition reflected an immaturity that caused them to employ limited vocabulary over and over. 16.

(33) From Crowhurst's study, we can learn that quantitative and qualitative analysis of the cohesive devices are both indispensable when attempting to reflect different graders' problems accurately. Additionally, some researchers criticized the incompleteness of cohesive paradigm as it failed to consider cohesive errors such as an inappropriate tie or the failure to provide a needed tie, which could serve as an index of lack of cohesion (Fitzgerald & Spiegel, 1986). Another researcher Stotsky (1983) highlighted several problems with Halliday and Hasan’s categorization of the lexical cohesive devices for they failed to note the text-forming relationships created by the use of derivatives or derivational elements. According to Halliday and Hasan’s theory (1976), a lexical item is not bound to a particular grammatical category or a particular morphological form (p. 291). For instance, talk, talks, talked and talking are all counted as one lexical item. Aside from this, there are “no perfectly clear criteria for deciding just how far this principle can be extended.” Consequently, noun, nominal, nominalize and nominalization are “presumably” all one lexical item. Hence, if any derivatives are deemed as mere repetition of the base word, some crucial information about the textual pattern could be lost (Stotsky, 1983). On the other hand, through the display of derivatives in expository writing, writers can demonstrate their stylistic flexibility and lexical maturity while preserving basic meaning. To be more specific, the occurrence of the derivatives in writing can indirectly imply whether or not the writer is sophisticated. In addition, Stotsky stated that derivational elements appear to be useful in creating an additional cohesive tie aside from the antonymous related items such as idealistic/realistic and result in multiple cohesive ties, which strengthen the cohesive quality of a text and compensate for the increasing density of ideas as writers consolidate phrases into words to achieve a more succinct prose style. Moreover, the employment of derivatives is a predominant feature of expository writing because it 17.

(34) can not only make the meaning more precise but also make the expression more concise. This is why derivatives should be considered a distinct and separate category from repetition, synonymy, or antonym when stylistic differences in ways of achieving lexical cohesion are examined across different kinds of discourse (Stotsky, 1983, p. 434). Thirdly, Stotsky also pointed out that Halliday and Hasan omitted to mention how a cohesive tie created by superordinate plus subordinate would be classified because only the subordinate words which precede the superordinate words are discussed. When it comes to the explanation of superordinate, Halliday and Hasan stated that it refers to any item whose meaning included that of the earlier one. In other words, any item that dominates the earlier one in the lexical taxonomy is a superordinate. Take the following sentence as an example: There is a boy climbing the old elm. That tree isn’t very safe (Halliday & Hasan, 1987, p. 280); in the case of which the tree is the superordinate of elm. As for whether the opposite position of superordinate and subordinate in the sentence could achieve cohesive function and could be counted as another cohesive type or not was not discussed. Therefore, Stotsky doubted that some useful information about text construction may be lost if we failed to classify the use of a superordinate word followed by a subordinate item as one type of cohesion because in essay writing; sometimes it is common to present the general ideas first, followed by detailed discussion or examples. Finally, Stotsky asserted that readers have no clear-cut criteria except for their subjective judgment based on the topic and experience to judge lexical collocation. Therefore, it is the "the most problematical part of lexical cohesion” because any two lexical items that tend to occur in similar contexts and form the cohesive effect is collocation (Halliday & Hasan, 1976, p. 284). Moreover, Stotsky claimed that antonym should be singled out as a separate category like synonyms because there is 18.

(35) no difference in understanding, comprehending or producing these two expressions. Thus, she doubted what linguistic principles separated these two expressions when there are no apparent differences in processing the similar or opposite counterpart. To restate all the points, it is clear that Halliday and Hasan’s scheme needs some modification. The quantitative method of counting the number of cohesive device doesn’t seem to explain all the cohesive patterns between different proficiency writers. Additionally, derivatives include crucial information that could serve as an index of writing proficiency. As for the superordinate, it needs a broader flexibility of the position in the text—that is, the inclusion of subordinate after the superordinate. Last but not least, the judgment of collocation is too subjective, for it is over dependent on the reader’s knowledge of the topic. Also, the entry antonym should be singled out as a separate entry like synonym because there is no difference in the process of producing and understanding these two cohesive devices. Based on the aforementioned drawbacks, some modifications should be made in order to precisely measure writers’ writing proficiency and employment of cohesive devices. 2.2.1 Modification of Cohesive Scheme by Stotsky Due to the flaws of Hallilday and Hasan’s cohesive scheme mentioned above, Stosky modified their scheme and provided a new theoretical framework of her own. This new framework composed two major categories. The first category included a wide assortment of lexical cohesive words possessing a systematic semantic relationship with each other, whether or not they co-occurred frequently. Words with an antonymous relationship, synonymous meaning, subordination or superordination, membership in ordered or unordered sets or derivation were all independent of frequency of contextual association and were classified into the first category of Stotsky’s framework. As for the second category, words which occur in texts with similar topics might continue to be designated as collocationally related words. Unlike 19.

(36) the lexicon in the first category, words classified under the category of collocation had no systematic semantic relationship with each other, and could only be associated with the topic of the text. I.. Stotsky’s Cohesive Framework Semantically related words: a type of cohesion in which one lexical element is. systematically related to a previous one through: 1. Repetition 2. Synonymy or near-synonymy 3. Opposition or contrast 4. Inclusion as a coordinate, superordinate, or subordinate member in an ordered or unordered set (general or specific terms) 5. Derivation or repetition of a derivational element. II. Collocationally related words: a type of cohesion in which one lexical element is related to another only through frequent co-occurrence in similar contexts. In Stotsky’s new cohesive framework, she developed categories such as derivative, and antonym (e.g., opposition or contrast) in order to correctly measure writers’ cohesion in the discourse. However, for the most problematic part of collocation, it appears that she did not provide a clear principle to amend Halliday and Hasan’s classification of collocation. The judgment of frequent co-occurrence of certain lexical items in the case of different text topics still depended much upon the reader’s knowledge of the topic or English proficiency. Therefore, based on the aforementioned reasons, in this study, collocation is not included. Although Stotsky's model seemed to provoke some thoughts on the analysis of cohesive devices, her model was not without flaws for she only included the analysis of lexical cohesive devices in the framework and did not mention grammatical cohesive devices. 2.2.2 Modification of Cohesive Scheme by Hasan In view of the flaws of Halliday and Hasan’s 1976 model, in 1985 Hasan modified the original cohesion scheme in the hope that the new cohesive scheme would provide a better index of discourse coherence. In her new model, three kinds of 20.