不同語言背景大學生論說文之主位選擇

全文

(1)國立臺灣師範大學英語學系碩. 士. 論. 文. Master’s Thesis Department of English National Taiwan Normal University. 不同語言背景大學生論說文之主位選擇. Thematic Choice in College Students’ Expository Essays of Different Linguistic Backgrounds. 指導教授：張珮青 Advisor: Dr. Pei-Chin Chang 研究生：李孟穎 Meng-Yin Lee. 中華民國一百零五年六月 June 2016.

(2) 摘要. 本研究旨在採用系統功能語言學檢視不同語言背景大學生論說文中的主位選擇。比較對象為台灣及中國大學生、新加坡大學生、母語人士大學生，並以母語專業人士作為參照。240 篇論說文採自亞洲英語學習者國際語料庫(ICNALE, Ishikawa, 2013)，比較的主位為語篇主位、人際主位，及標記主位。本研究分析結果顯示，此三組大學生的主位選擇模式不盡相同。首先，台灣及中國大學生的語篇主位及標記主位使用頻率上顯著多於母語專業人士，且傾向使用外顯的語篇主位銜接語篇。此外，母語人士大學生的語篇主位及人際主位顯著多於母語專業人士，並包括較多的非正式語域。最後，新加坡大學生的主位選擇近似母語專業人士。與專業人士相較，新加坡大學生的三種主位頻率皆未達顯著，傾向用詞彙而非語篇主位來銜接語篇，他們的主觀性人際主位語氣不似母語人士大學生般語氣強烈，且他們的標記主位能銜接文章主題。根據分析結果，本研究提供台灣及中國教科書編者及英語教師設計論說文寫作教材的具體建議，並提出以篇章為主的論說文寫作教學法應有其必要。. 關鍵字：系統功能語言學、主位選擇、大學生論說文寫作. i.

(3) ABSTRACT The study investigates how college students of different linguistic backgrounds organize information in expository essays using NS professionals as a reference. Using a Systemic Functional Linguistic approach, the study compares the thematic choices of 240 controlled essays by Chinese (including essays from China and Taiwan), Singaporean, native college students, and native professionals in the International Corpus Network of Asian Learners of English (ICNALE, Ishikawa, 2013). Three Themes under investigation are textual, interpersonal, and marked Themes. The results show that the students of different linguistic backgrounds have different patterns of thematic choices. First, the Chinese students deploy significantly more textual Themes and marked Themes than the native professionals to achieve cohesion explicit. Second, the native students utilize significantly more textual Themes and interpersonal Themes than the native professionals. The Themes in the native students’ writing tend to contain more informal registers. Finally, the Singaporean students’ writing exhibits patterns of Themes similar to those of the native professionals. None of the Themes by the Singaporean students is significantly more than those of the native professionals. Their cohesion tends to be achieved implicitly through lexical cohesion. Their subjective interpersonal Themes are not strong in tone as those of native students. Their marked Themes continue the topic. Based on the findings, some pedagogical implications are offered for the Chinese-speaking EFL textbook writers as well as English teachers to design expository teaching materials. The results of the study may inform the necessity of a more discourse-level approach to teaching expository writing in the Chinese-speaking EFL context. Key words: Systemic Functional Linguistics, thematic choices, college students’ expository writing ii.

(4) ACKNOWLEDGEMENTS. Stepping into the world of SFL was a complete coincidence. The first encounter with SFL was in a speech held by the English department of NTNU where I learned the fascinating lexical chains, genre, and moves. Then I took Prof. Chang’s Writing Design course, where she devoted much time to SFL and passed down invaluable skills of text analysis. These happened in the first year of my graduate school life. I am grateful to have found my research of interest in the very first year. I would like to give my deepest gratitude to my dear advisor, Dr. Pei-Chin Chang. Her passion and rigor toward research has been pushing me forward, and her gentle and caring personality has driven away much of the pressure from writing the thesis. With her encouraging and inspiring words, every meeting with her became a pleasant and informative experience. This thesis could not have been possible without her guidance. I would also like to give my heartfelt gratitude to my committee members, Dr. Yuh-Show Cheng and Dr. Shih-Chieh Chien, who provided me valuable comments. I would like to thank Chun-Yu for his great patience in assisting me with statistics, and Hui-Ping for her brilliant advice and enduring company. I also own my gratitude to my loving parents and my siblings for their life-long support for every decision I make.. iii.

(5) TABLE OF CONTENTS. CHAPTER ONE. INTRODUCTION ............................................................................... 1. Research Background ................................................................................................. 1 Research Rationale ...................................................................................................... 2 Research Questions ..................................................................................................... 5 Significance of the Study ............................................................................................ 6 Organization of the Thesis .......................................................................................... 6 CHAPTER TWO. LITERATURE REVIEW.................................................................... 9. SFL.............................................................................................................................. 9 The Notion of Stratification ............................................................................... 9 Genre ................................................................................................................ 11 Theme-Rheme ........................................................................................................... 13 Simple Theme .................................................................................................. 14 Multiple Theme ................................................................................................ 15 Unmarked and Marked Theme ........................................................................ 17 Theme Studies of EFL and ESL College Students’ Expository Writing ................... 22 EFL, ESL and NS College Students’ Expository Writing................................ 22 High- and Low-rated EFL and ESL College Students’ Expository Writing .... 24 Thematic Choices and Disciplinary Differences ............................................. 26 Theme-Rheme-based Instruction .............................................................................. 28 CHAPTER THREE METHODOLOGY ....................................................................... 31 The Corpus ................................................................................................................ 31 Unit of Analysis ........................................................................................................ 32 Analytical Framework .............................................................................................. 32 iv.

(6) Procedure .................................................................................................................. 34 Data Analysis ............................................................................................................ 36 CHAPTER FOUR. RESULTS ........................................................................................ 39. Quantitative Differences of Thematic Choices in the Groups .................................. 39 Qualitative Differences of Thematic Choices in the Groups .................................... 44 Informal and Formal Textual Themes .............................................................. 44 Subjective and Objective Interpersonal Themes .............................................. 52 Marked Themes and Topic Continuity............................................................. 57 CHAPTER FIVE DISCUSSION AND CONCLUSION ............................................... 65 Major Findings of the Study ..................................................................................... 65 Pedagogical Implications .......................................................................................... 70 Limitations of the Present Study ............................................................................... 72 Directions for Future Research ................................................................................. 72 REFERENCES ................................................................................................................. 73. v.

(7) LIST OF TABLES Table 1. Social context realization under the SFL framework ......................................... 11 Table 2. Mean and standard deviation of different Themes in the four groups ............... 40 Table 3. Two-way ANOVA result ...................................................................................... 41 Table 4. Multiple comparisons of different Themes in the four groups ........................... 42. vi.

(8) LIST OF FIGURES Figure 1. Stratification and realization path of ideology, context, and language ............ 10 Figure 2 Theme system adopted from Halliday and Matthiessen (2013) ............................ 21 Figure 3. A simplified Theme system in the present study ........................................................ 33 Figure 4. An analysis example ......................................................................................................... 36. vii.

(9) CHAPTER ONE INTRODUCTION. Research Background. Writing proficiency is an important indicator of academic success especially in tertiary education. Students are expected to acquire the academic writing conventions to meet course or disciplinary requirements. Writing is also widely assessed in high-stakes tests. International standardized tests such as Test of English as a Foreign Language (TOEFL) and International English Test System (IELTS) assess English learners’ ability to write expositions. In Taiwan, two college entrance exams, the General Scholastic Ability Test (GSAT) and the Advanced Subjects Test (AST), also examine test takers’ English writing readiness for college education. Important as writing is, most English learners find writing difficult to master. The difficulty may lie in a holistic grasp of content, text organization, genre, authorial stance, and language. Poor writing proficiency often prevents students from performing satisfactorily and compromises their academic performance. For example, in Taiwan, based on the data from the College Entrance Examination Center (CEEC), the writing performance of most test takers has been far from satisfactory in the English subject of GSAT and AST. More than fifteen thousand students receive a zero score in both of the English writing tasks each year. The average scores in GSAT writing were 8.44 out of 20 in 2015. What is worse, examinees that received a zero score soared from 8.31% (12,251 people) in 2014 to 9.24% (13,493 people) in 2015. Apparently, fostering the writing competence of students of English as a foreign language (EFL) is a great challenge to educational practitioners. The challenge may be related to a lack of adequate writing instruction. 1.

(10) Discourse-level writing instruction is usually neglected at schools in many countries (Graham, Gillespie, & McKeown, 2013). Chinese students that study English as a foreign language (EFL) are especially disadvantaged by this negligence because their exposure to English is usually limited in the classroom, and they have relatively little exposure to English outside the classroom compared to native speakers (NS) and students of English as a second language (ESL), who study English in an English-speaking environment. The writing instruction in the EFL classroom is usually limited in sentence level. EFL students are usually required to do sentence-making drills instead of composing structurally sound writing. EFL teachers spend substantial amount of time checking grammar and spelling in students’ writing (Belmonte & McCabe, 1998, 2003; Bohnacker, 2010; Ebrahimi & Ebrahimi, 2012). Little time is devoted to discourse-level writing instruction such as text organization, stance, and genre. Teachers may not be aware that discourse-level writing instruction is crucial to developing students’ communicative skills in writing. Linguists from Systemic Functional Linguistics (SFL) propose that Theme-Rheme construct can be implemented to facilitate discourse-level writing instruction (Fries, 1983; Martin, 1992). If equipped with a Theme-Rheme approach to writing, teachers may be better prepared to teach discourse-level writing.. Research Rationale Theme, the point of departure of a clause, “locates and orients the clause within its context” (Halliday & Matthiessen, 2013, p. 89) under the SFL framework. SFL is a functional model to language that views language as a system for making meaning in the social context (Halliday & Martin, 1993). In this functional framework of language, every clause serves ideational, interpersonal, and textual metafunctions 2.

(11) (Martin & Rose, 2003). Of the metafunctions SFL depicts, the study adopted the textual metafunction, in which Theme, the onset of a clause, and Rheme, the newly expanded information around its Theme, are the main units of analysis (Halliday, 1985; Martin & Rose, 2003). In the Hallidayan tradition, Theme is further classified into topical, textual, and interpersonal Themes based on its metafunctions. The topical Theme is the first ideational component that can be a participant (e.g., I), a process (e.g., Let’s), or a circumstance (e.g., By banning smoking). The topical Theme can be headed by textual Themes (e.g., and, besides) to organize information, or fronted by interpersonal Themes (e.g., maybe, I think) to express the writer’s opinions. The topical Theme can also be unmarked or marked. The topical Theme is unmarked if it is the grammatical subject (e.g., banning smoking). The topical Theme is marked if it is something other than the grammatical subject (e.g., by banning smoking). Theme is an enabling resource that constructs ideational and interpersonal meanings (Halliday & Matthiessen, 2013). Examining thematization allows linguists to scrutinize how writers organize information to structure their essays and interact with readers. Example 1 illustrates how one meaning can be expressed differently through different thematic choices.. Example 1 Two ways of information organization in a sentence1 Theme Textual (1) And so. Rheme Interpersonal I think. (2) Therefore it is believed that. 1. Topical everyone by quitting smoking. can regain health by quitting smoking. everyone can regain health.. The two sentences in Example 1 are contrived by the researcher. 3.

(12) In Example 1, the ideational meaning of the two sentences is similar. That is, quitting smoking helps people regain health. However, the ways of organizing information are quite different. In Sentence (1), the textual Theme (i.e., And so) is less formal. Also, the interpersonal Theme (i.e., I think) involves the writer’s subjective opinion. Finally, the topical Theme does not continue or highlight the topic. The topic (i.e., quitting smoking) stays in Rheme and is less obvious. By contrast, in Sentence (2), the textual Theme (i.e., Therefore) is more formal. The interpersonal Theme (i.e., it is believed that) seems more objective. The topical Theme continues and highlights the topic (i.e., by quitting smoking) in Theme position. Therefore, Sentence (1) is less effective than Sentence (2) for expository essays, which requires a formal, impersonal writing style that highlights key information (Schleppegrell, 2004).. The example also exemplifies how thematic choices account in part for varying degrees of successful writing (Berry, 1995). While professional writers may be experienced in choosing Themes that are characteristic of academic registers, student writers may find it challenging to achieve such mastery. The comparatively ineffective thematic choices may affect student writers’ writing quality and academic performance. Past research concerning thematic choices has found that non-native students deviate from native students in thematic choices (Green, Christopher, & Mei, 2000; Herriman, 2011; Herriman & Boström Aronsson, 2009; Wei, 2013, 201. ang,. am re , Harman, 2011). Usually the studies are under the assumption that non-native speakers are less proficient than native students to manage effective thematic choices. However, the assumption can be questionable in that native students may not always write better than non-native students. For example, Granger (1997) has observed that 4.

(13) NS students produce more dangling participles than EFL students. If native students “do not necessarily provide models that everyone would want to imitate” (Leech, 1998, preface), a reference corpus from native expert writers is needed to make sensible comparisons. As Ä del (2006) notes, professional writing can serve as a norm to be achieved and promoted by advanced learners of English and their teachers. Studies that use professional writing as a reference therefore may be more appropriate for pedagogical applications in the non-native classrooms. In addition to the “norm” issue in the previous studies, some corpora under investigation are very small (Wei, 2013; Yang et al., 2007). Some texts are drawn from different corpora, and some of the writing conditions are not controlled (Green et al., 2000; Herriman, 2011; Herriman & Boström Aronsson, 2009). To bridge the gap, the study compares the thematic choices of 240 controlled essays by EFL, ESL, NS students and NS professionals in the ICNALE. The NS professional writer data serves as reference to further identify patterns of use among the users of different linguistic backgrounds and academic standing.. Research Questions Based on Halliday and Matthiessen’s (2013) Theme system, the current study investigates the quantitative and qualitative differences of thematic choices by EFL, ESL, NS students compared to NS professionals. For qualitative analysis, the study examines the use of informal and formal textual Themes, the use of subjective and objective interpersonal Themes, and the use of marked Themes that continue or discontinue the topic. The two research questions guide the study:. 5.

(14) 1.. What are the quantitative differences of textual, interpersonal, and marked Themes in the EFL, ESL, NS groups compared to NS professionals?. 2.. What are the qualitative differences of thematic choices in the EFL, ESL, NS groups compared to NS professionals? 2-1. What are the different uses of informal and formal textual Themes? 2-2. What are the different uses of subjective and objective interpersonal Themes? 2-3. What are the different uses of marked Themes that continue or discontinue the topic?. Significance of the Study It is expected that a comparison of the thematic choices in the EFL, ESL, and NS students’ expository essays may uncover how student writers of different linguistic backgrounds deviate from professional writers in thematic choices. The deviation may serve as resources for the Chinese-speaking EFL textbook writers as well as English teachers to design expository teaching materials. The results of the study may inform the necessity of a more discourse-level approach to teaching expository writing in the Chinese-speaking EFL context.. Organization of the Thesis The first chapter introduces the research background, rationale, research questions and significance of the thesis. The second chapter reviews studies related to Theme. In the third chapter, the research design, analytical framework, corpus data, procedures, and data analysis are revealed. The fourth chapter reports the quantitative and qualitative results of the analysis. The fifth chapter discusses the 6.

(15) possible factors that influence the differences of thematic choices and provides suggestions for English teachers and textbook writers in the Chinese-speaking EFL context. The limitations of the study are also addressed before concluding the chapter.. 7.

(16) 8.

(17) CHAPTER TWO LITERATURE REVIEW. In this chapter, I will first introduce SFL, genre, and Theme-Rheme, the analytical tools that informed the present study. Next, I will review studies that have implemented Theme- heme analysis in EFL and ESL students’ expository essays as well as Theme-Rheme writing instruction.. SFL SFL is a functional model to language that views language as a semiotic system for making meaning in the social context (Halliday & Matthiessen, 2013). Referred to as a “customer-oriented” theory (Halliday, 1985, p.7) by its founder M.A.K. Halliday, SFL concerns how language users deploy their meaning potential by making choices among alternatives from this system network (Halliday, 1978). In SFL, language is a system of sound, wording, writing, structure, and resource, rather than fragments of rules (Halliday & Matthiessen, 2013), and meaning is achieved through the layers of system or stratification (Halliday & Matthiessen, 2013; Martin, 1992).. The Notion of Stratification Figure 1 below describes the stratification and realization path of ideology, context, and language. The three main components of stratification on the right of Figure 1 are ideology, context, and language. Ideology is realized through context, and context in turn is realized through language content and expression. The three components can be further divided into finer sub-strata.. 9.

(18) Figure 1. Stratification and realization path of ideology, context, and language. According to the realization path, meaning is realized from the innermost phonetics stratum to the outermost ideology stratum in concentric circles. Language, the innermost stratum, is first stratified into language expression and language content, and further divided into four strata: phonetics, phonology, lexico-grammar, and semantics. The interfacing stratum, context, consists of two planes: genre (context of culture) and register (context of situation). Genre deals with the integration of meanings while register constructs genre through three interconnected variables: field (experience or topics), tenor (interlocutor relationships), and mode (medium of communication). Ideology, the outermost stratum, underlies all types of discourse that negotiates meaning, and functions as a contextual and linguistic source of semogenesis. Table 1 shows how the social context is realized under the SFL framework. The 10.

(19) three register variables (i.e., field, tenor, mode) are realized through three metafunctions (i.e., ideational, interpersonal, and textual) and analyzed by Transitivity, Mood, and Theme structure respectively. The construal of reality can be experiential, social, and semiotic.. Table 1. Social context realization under the SFL framework Register Variables Field Tenor Mode. Metafunctions Ideational Interpersonal Textual. Analysis Tools Transitivity Mood Theme. Construal of Reality Experiential Social Semiotic. Of the three metafunctions that SFL depicts, ideational metafunction represents experience, interpersonal metafunction concerns the relationship between people exchanging information, and textual metafunction involves the organization of texts. The present study adopted the textual metafunction. According to Halliday (1985), textual metafunction is realized through three components: thematic structure (Theme and Rheme, including thematic choice and thematic progression), information structure (Given and New information), and cohesion (reference, conjunction, substitution, ellipsis, and lexical cohesion). The study analyzed the thematic choices of texts, in which Theme, the onset of a clause, and Rheme, the newly expanded information around its Theme, are the main units of analysis (Halliday, 1985; Martin & Rose, 2003).. Genre When dealing with text, one indispensible issue is genre. The Sydney School, one branch of SFL in Australia, has contributed much to genre theory and has applied genre theory into the classroom. Instances of such application are curriculum genres (e.g.,. 11.

(20) Fran Christie) and school genres (e.g., Caroline Coffin, Sally Humphrey, Maree Stenglin and Robert Veel). Genre has been theorized as an element of the functional model of language, and defined as a “recurrent configuration of meanings” (Martin & Rose, 2008, p. 6) in a given context of culture. This functional definition is in line with that of Biber’s (1991). Biber views genre as regularly occurring social activities whose types are highly agreed upon by the speech community. Examples of the functional definition of genre are narratives, expositions, lectures, seminars, manuals, recipes, and appointment making. In addition to functions, genre is also “staged, goal-oriented, and social” (Martin & Rose, 2003, p.7). Genre is goal-oriented because the purpose of genre is to communicate, staged because more than one step is required to reach this goal, and social because genre is used to interact with particular readers. Halliday (1978) considers genre to be an aspect of mode. On the contrary, Martin and Rose (2008) suggest that genre entails a particular arrangement of field, tenor and mode, which provides a more holistic and functional interpretation of genre. This interpretation of genre indicates that genre can be about almost any topics (field), audience involved can be distant or close, equal or unequal (tenor), and language form can be spoken or written (mode). In genre-related studies, some commonly seen topics are rhetorical structure (McCarthy & Carter 2014), genre-based pedagogies (Hyland, 2002), and a budding interest in genre and Theme relations (Berry, 1995; Fries 1983; Ghadessy, 1995; Martin 1985). Among the studies, some genres appear more frequently than others because of their applicability in educational contexts. Expositions, for example, are a commonly seen genre in school tests. Martin (1985) categorizes expositions into two types: analytical exposition and hortatory exposition. Analytical exposition persuades readers. 12.

(21) with structured and organized argument whereas hortatory exposition urges readers to do what the thesis of an exposition proposes.. Theme-Rheme The term Theme-Rheme was first coined by Ammann (1928) in the functionalist Prague School. Functional definitions of Theme and Rheme operate at two levels: clausal and discursive. At clausal level, Theme embodies language user’s decisions on order of information prominence within a clause. At discursive level, Theme helps information flows and shapes the interpretation of text. Theme functions as a recurrent point of departure of information that maintains text coherence by connecting the previous discourse and developing the next discourse, and therefore gives continuity to a phase of discourse (Brown & Yule, 1983; Martin & Rose, 2003). The discourse level definition of Theme-Rheme highlights the prominence of Theme position and represents the speaker/writer interpretation in hierarchical value of information (Brown & Yule, 1983; Hoey, 2005; Martin & Rose, 2003; Vasconcello, 1992). For instance, Martin and Rose (2003) indicate that Theme is the most prominent element in a clause. Brown and Yule (1983) suggest that thematic choices affect the interpretation of a hearer/reader in a discourse. The thematic options may drastically influence the way readers interpret a text, and readers expect differently about Theme and Rheme due to their cultural priming (Hoey, 2005). Researchers have different definitions of Theme and Rheme (Halliday, 1985; Hawes, 2001; North, 2005). Yet the most celebrated definition is from Halliday (1985), who defines Theme as 1) the onset of a clause, 2) what the message will express, and 3) the first ideational component that is assigned a Transitivity role. Transitivity refers to the three linguistic representations of the real world: participant, process, and 13.

(22) circumstance. Participants stand for human beings or abstract concepts, process describes actions, and circumstance denotes situations. Following Theme, Rheme is the remainder of a clause. Theme and Rheme constitute a clause conveying a message (Halliday, 1985). Example 2 illustrates the boundary between Theme and Rheme. The left constituent Students who have part-time jobs is Theme because it positions in the beginning of the clause, it predicts what the clause will be about, and it is the first ideational component. Rheme includes the verb and everything after the verb (i.e., usually work better with their degree studies). Theme and Rheme compose a complete clause here.. Example 2 Theme and Rheme boundary2 Theme. Rheme. Students who have part-time jobs. usually work better with their degree studies.. Simple Theme In the Hallidayan tradition, Theme is further classified into three main categories based on its metafunction: topical, textual, and interpersonal Themes. If only the topical Theme occupies the Theme position, it is called a Simple Theme. Topical Theme is the obligatory thematic element because it is a Theme constituent that is assigned a Transitivity role, and once the topical Theme of a clause is determined, the remaining constituents can be consigned to Rheme (Eggins, 2004). Example 3 below demonstrates sentences with a simple topical Theme that is assigned a Transitivity role of participant, process or circumstance respectively.. 2. All examples are drawn from the ICNALE written corpus unless otherwise mentioned. 14.

(23) Example 3 Simple Theme topical Theme (required). Rheme. All college students. should have a part-time job.. Let. the cool people have their smoked dinners.. When my students work with people from. they often come to me with new insights and. another culture and country,. new business ideas which they had never even considered before.. Some scholars (e.g., Berry, 1995; Thompson, 2013) suggest that the term topical Theme is problematic as topical Theme does not necessarily imply topic or what the clause is about, and Fries (1995) argues that the main function of topical Theme is rather to orient the speaker/reader. Therefore, the term experiential Theme should be adopted. However, to stay with the Hallidayan conventions, the study remains the term topical Theme to refer to the first ideational element of a clause.. Multiple Theme The obligatory topical Theme may be headed by textual Themes as cohesive devices, by interpersonal Themes to show authorial stance, or by the combination of the two to exert both functions. Textual and interpersonal Themes are used to orient the reader. Textual Themes do cohesive work by linking the clause to its context with continuatives (e.g., yes), conjunctions (e.g., but), or conjunctive adjuncts (e.g., however). Interpersonal Themes, on the other hand, are any constituent that is assigned a Mood label to establish interaction or express comments. Interpersonal Themes include vocatives (e.g., Sir), modal adjuncts (e.g., probably, surely), and mood-marking devices (i.e., WH-words and finite verb operators in interrogatives, 15.

(24) and Let’s in imperatives). A multiple Theme example is shown in Example 4.. Example 4 Multiple Theme Multiple Theme. Rheme. textual. interpersonal. topical (required). However,. honestly speaking. all college students. should have a part-time job.. The study also adopts the concept of interpersonal grammatical metaphor as interpersonal Theme (Halliday & Matthiessen, 2013). Interpersonal grammatical metaphor is a clause that expresses opinions, evaluation, or judgment. Examples are I think, Do you suppose, and It is likely that. Interpersonal grammatical metaphors can be analyzed from two perspectives: experiential or interpersonal (Thompson, 2013) as shown in Example 5. From the experiential perspective, interpersonal grammatical metaphor is a projecting clause, and should be analyzed as a clause as in version (b) analysis in Example 5. However, from the interpersonal angle, the interpersonal grammatical metaphor is an interpersonal Theme and should be analyzed within the multiple Theme as in version (a). The study adopted version (a). The advantage of version (a) is that it keeps the analysis in the same Theme-Rheme unit and facilitates to uncover how types of thematic choices work in a multiple Theme. It also allows other thematized comments (e.g., it is true that, it is interesting that) to be analyzed in the same way (Thompson, 2013).. Example 5 Analysis of interpersonal grammatical metaphor Theme. Rheme. (a). interpersonal. topical. (b). Theme. Rheme. Theme. Rheme. I. think that. smoking. should be banned in Japan. 16.

(25) Unmarked and Marked Theme Topical Themes can be further divided into unmarked or marked Themes. Topical Themes are usually unmarked unless there is a good reason for choosing otherwise. What is chosen as the topical Theme depends on the Mood (i.e., declarative, interrogative, and imperative) of the clause.. Themes in declaratives. Theme markedness in declarative clauses depends on the mapping of the grammatical subject. Topical Themes that conflate with the subject are unmarked. This type of Theme is termed “unmarked” because it is the most typical form of Theme (Eggins, 2004; Halliday & Matthiessen, 2013). Unmarked Themes can be as simple as a nominal group that includes a pronoun, a common noun, a proper noun, or as complex as a nominalized clause, the examples of which are illustrated in Example 6.. Example 6 Unmarked Theme in declaratives Theme = Subject. Rheme. There. are many excellent reasons to join the work force while still attending school.. Whether or not he will actually go so far. of course will vary from person to person.. as to do so. Compared with unmarked Themes, marked Themes are less typical. Marked Themes refer to thematizing constituents other than the grammatical subject (Eggins, 2004; Halliday & Matthiessen, 2013). Halliday and Mathiessen (2013) suggest a cline of markedness. The most common form of marked Themes is an adjunct (i.e., a adverbial group, a prepositional phrase, a non-finite clause, a subordinate clause). 17.

(26) Marked Themes can also be a complement (i.e., a fronted nominal group other than the subject). Using a complement as marked Themes is least common, so it is the most marked form (Halliday & Matthiessen, 2013). Example 7 demonstrates types of marked Theme examples in declarative clauses. From the top to the bottom are an adverbial phrase, a prepositional phrase, a non-finite clause, a subordinate clause, and a complement.. Example 7 Marked Theme in declaratives Theme. Rheme. Too often. young people depend on others to see to it that their lives are on the right track.. According to Mill's Harm Principle,. we should completely ban smoking.. To be polite to people around you,. smoking should not be allowed in restaurants.. When I packed my job in,. it felt like a load of bricks had been lifted off my shoulders. From there. I derive my two reasons for banning smoking.. Themes in non-declaratives. In WH-interrogative clauses, WH-words are the unmarked choice, and in yes/no interrogatives, the finite verbal operator (i.e., do, have) plus the subject is the unmarked Theme. In imperatives, the unmarked Theme is the predicator that expresses actions (i.e., cry, let’s). In negative imperatives, the unmarked Theme also includes the finite verbal operator before the predicator (i.e., Don’t talk). Although exclamatives are a subtype of declaratives, this small group of clauses resembles the form of WH-interrogatives and is analyzed accordingly, with 18.

(27) WH-words being the unmarked choice. The examples of unmarked Theme in non-declaratives are presented in Example 8.. Example 8 Unmarked Theme in non-declaratives Theme. Rheme. Where. does the infringement on individual liberties and government intervention cease?. Would Toyota's union,. for instance, expose its employees to hazardous chemicals involved in automobile manufacturing?. Let's. take an active part in it!. Don't spend. too much time at part-time job.. What a nice job. you’ve done!. As with declaratives, marked Themes in interrogatives are adjuncts or complements that head unmarked Themes. Marked choices of Themes in imperatives include “you,” and finite operators “Do” to carry an emphatic tone, or adjuncts to explain the reason for the command as shown in Example 9.. Example 9 Marked Theme in non-declaratives Theme. Rheme. And yet, how much control. does the society need to place on its citizens?. However, can we. ban such behaviour?. You, stop. smoking indoors.. Do manage. your time well.. For more clips, please subscribe. to our website.. 19.

(28) Themes in clause complexes. The above examples deal with simple clauses. This section demonstrates how Themes in clause complexes are analyzed. In a clause complex, a dependent clause normally follows the independent clause, and the subject of the independent clause is the unmarked Theme. When the dependent clause precedes the entire clause complex, it is suggested that the whole dependent clause be the marked Theme for the entire clause complex (Halliday & Mathiessen, 2013; Thompson, 2013). Analyses of unmarked and marked clause complexes are illustrated in Example 10 and 11.. Example 10 Unmarked Theme in clause complexes Theme. Rheme. Restaurant themselves. could decide if they want to welcome smokers.. Example 11 Marked Theme in clause complexes Theme. Rheme. If people want to have a more relaxing. they can also go outside or some other. conversation by having a cigarette,. places such as lounge bar.. Predicated Theme. Predicated Themes involve cleft sentence pattern It + be +…. that/who…. Predicated Themes are also a marked choice to carry an emphatic tone. In this sentence pattern, a simple sentence is separated into two clauses. As can be seen from Example 12, the first clause it is from this part-time job serves as a predicated Theme.. 20.

(29) Example 12 A predicated Theme Predicated Theme. Rheme. It is from this part-time job. that I gained both money and a sense of achievement.. One final note of this section is the ellipted Theme. Occasionally a Theme is ellipted and is not analyzed in the Theme analysis. Example 13 shows ellipted Theme examples.. Example 13 Ellipted Themes Theme. Rheme -. A humbling experience I'm sure.. -. Not even in their yard.. The above Theme system can be summarized in Figure 2.. Figure 2 Theme system adopted from Halliday and Matthiessen (2013) 21.

(30) In this section, I have introduced the significance and functions of Theme-Rheme and types of Themes. In the next section, I will review empirical studies regarding Theme analysis in EFL and ESL students’ essays and the application of Theme in the English writing instruction.. Theme Studies of EFL and ESL College Students’ Expository Writing Since Halliday (1985) proposed the Theme-Rheme structure, much interest has been generated in exploring the distinctive features of thematic choices in English learners’ expository essays. To locate these writing’ features, some researchers have compared English learner texts with NS texts (Green et al., 2000; Hawes & Thomas, 1997, 2012; Wei, 2013, 2016). Other researchers have scrutinized EFL and ESL writers’ thematic choice features across proficiency levels (Hawes & Thomas, 2012; Mellos, 2011; Xu, 2000). Still other studies have located effective patterns of thematic choice and progression with a view to enhancing English learners’ text cohesion (Belmonte & McCabe, 2003; Wang, 2007). An increasing interest in this field is to implement Theme-Rheme writing instruction in non-native contexts (Belmonte & McCabe, 2003; Li, 2009; Wang, 2007; Yang, 2009). The following part of this section will review studies related to EFL and ESL students’ themati ation in expository essays and Theme-Rheme writing instruction in the L2 classroom.. EFL, ESL and NS College Students’ Expository Writing In studies that compared EFL or ESL writing with NS writing, researchers have found that EFL or ESL writers tend to use more Themes than NS students because of the transfer of the mother tongue (Green et al., 2000; Herriman & Boström Aronsson, 2009; Wei, 2013, 2016; Xu, 2000). Green et al. (2000) revealed that Chinese college 22.

(31) students tended to overuse marked Themes due to the topic prominent feature of Chinese. In a corpus-based study that compared Hong Kong freshman students' use of marked Themes in expository essays with that of NS, the Hong Kong students tended to overuse conjunctive adjuncts (i.e., moreover, furthermore and besides), which exuded an emphatic tone at best, and affected global coherence and information structure at worst. Similar results were found in Wei’s (2016) comparative study on the Theme choices of 60 essays from Chinese sophomore English majors and American international students. The Chinese students were found to use significantly more textual, interpersonal, and marked Themes, all of which are more likely to be in Rheme. The researcher concluded that Chinese students needed instruction of Theme to order information. Wei (2013) reported that EFL students, irrespective of their native language, tend to use more Themes than NS students. The researcher compared the Theme choices of expository texts from 20 Chinese, 16 Swedish college English majors, and 16 British essays. The two EFL groups were found to use more interpersonal Themes and focused more on humans and temporal concepts, presented vague and general statements, and emitted a sense of spoken English. On the contrary, the NS corpus centered much on inanimate concepts, making the language more academic and formal. Herriman and Boström Aronsson (2009) explored to what extent Swedish advanced learners deviated from NS students in certain thematic choices in argumentative essays. The researchers compared two corpora. One was the Swedish sub-corpus of the International Corpus of Learner English (SWICLE), and the other was the Louvain Corpus of Native English Essays (LOCNESS). The results showed 23.

(32) that the Swedish learners’ essays contained more interpersonal Themes and cleft constructions that expressed personal opinions, both of which resulted in a greater involved style of writing. Xu (2000) investigated the extent to which L1 affected EFL college students’ control of cohesive devices, thematic progression (TP), and cohesion. He compared 141 Chinese sophomore expository essays with NS texts. The results indicated that successful EFL writers applied thematization at discourse level and were better at using cohesive devices in a way similar to NS writers while unsuccessful writers might focus on the local sentence structure in organizing information and were more prone to misuse cohesive devices.. High- and Low-rated EFL and ESL College Students’ Expository Writing Theme structure is also applied to examine L2 students’ thematic errors (Arunsirot, 2013; Belmonte & McCabe, 1998; Bloor & Bloor, 1992; Li, 2009, Mellos, 2011; Wang, 2010). Belmonte and McCabe (1998), Li (2009), and Arunsirot (2013) proved that the Theme-Rheme construct could be a useful tool for L2 teachers to locate students’ errors of themati ation and evaluate their textual cohesion and coherence. Belmonte and McCabe (1998), for example, deducted seven types of problematic thematic errors from 25 compositions from EFL college students in Madrid. Problems related to thematic choices included confusing selection of topical and textual Themes, overuse of there, brand new Theme, and Themes with unclear reference. The researchers concluded that the Theme-Rheme construct could be a valuable marking instrument, a guideline in argumentative effectiveness, and classroom activities in class. Li (2009) investigated 45 argumentative essays of Chinese college students 24.

(33) majoring in English and found that EFL students overused thematized comments such as it is quiet common to state their attitudes in the opening sentences where facts should be introduced instead. Analyzing 114 essays written by 57 Thai English major sophomores, Arunsirot (2013) pointed out that that students’ persistent problems of empty Theme, or ellipted subject, was due to L1 interference, and suggested that explicit and direct learning strategies should be applied to overcome such fossilization. Then the focus or research shifted to how proficiency levels shaped thematization and affected text coherence. Mellos (2011) demonstrated that ways of thematization generated differing degrees of text coherence by comparing four high and four low coherent ESL undergraduate students’ essays. The low coherent essays were found to overuse unmarked Themes of simple nominal groups or pronouns and constant progression in a way that made the texts rather unreadable and lacked development. High coherent essays, on the other hand, used dense and complex nominal groups in topical Themes, applied varied types of textual Themes to comment on abstract concepts, and employed different TP patterns to achieve text coherence. In a study that compared writing papers of high, mid, and low grades in Test for English Majors-4 (TEM-4, a high-stakes test) in China, Wang (2010) found that lower intermediate students were very persistent in constant progression while advanced students applied more multiple Themes, clausal Themes (i.e., noun clause as Theme), and more effective TP patterns (i.e., constant, linear and derived) to make their writing more coherent. The discrepancy of thematization of varying proficiency levels affects not only cohesion and coherence but also writer’s voice and rhetorical effects. Hawes and Thomas (2012) compared the thematic choices of 16 ESL pre-MA students of 25.

(34) journalism in their EAP writing with those of editorials from two British newspapers, The Sun and The Times. The researchers reported that the students departed from expert writers in both voice and rhetorical effects. In terms of voice, the students exhibited polarized subjectivity. Some students avoided taking responsibility for their claims by applying up to four times more pseudo-objective Themes (e.g., there or it) than the newspapers. Others appeared too subjective by employing too many discourse participants (e.g., I). The students also seriously underused non-participant Themes such as it, there, interrogatives, Verb group, and ellipted Themes, which signaled their monotonous sentence structures. The researchers concluded that student writers needed to be trained skills of effective thematic choices so as to present clear arguments and meet the genre requirements.. Thematic Choices and Disciplinary Differences Hewings (2004) and North (2005) investigated disciplinary differences of thematic choice use. Hewings (2004) investigated discipline-specific writing. A total of 1272 clause complexes from geography essays written by first and third year undergraduate students were compared. The third year undergraduates, gaining higher marks, were found to be more able to expand meaning potential by using textual Themes to link, argue, and relate to context and interpersonal Themes to make visible the rhetorical role of the writer. In contrast, first year undergraduates received lower marks. Their writing was more descriptive rather than argumentative due to the plenty unmarked Themes deployed. They also used fewer connectors, which required more reader’s effort to intuit connections. In North’s (2005) study, a total of 1 undergraduate essays, 33 from the arts college and 28 from the science college in Open University in UK were compiled. 26.

(35) North attributed the significantly higher score of the arts students to their obviously more use of textual and interpersonal Themes to connect sentences and their more authoritative stance. The explicit writer intervention by the arts students generated rhetoric in their writing rather than stacks of facts as the science students did. North (2005) concluded that disciplinary backgrounds and prior writing experience shaped writing performance. Conclusions from the above research can be summarized as the following. First, thematic choices are highly related to text cohesion and coherence. Less proficient writers tend to use problematic Themes (e.g., brand new Theme) and overuse marked Theme, which break cohesion and coherence. They also fail to show the writer’s voice in Theme as requested in a target genre. In addition, Thematization is affected by language proficiency, prior academic experience, L1 transfer, and disciplinary backgrounds. To attain cohesive and coherent writing with appropriate voice, EFL and ESL student writers need to be made more aware of the importance and functions of thematic choices. Demonstrating Theme-Rheme analysis or showing EFL or ESL students’ deviation from the target text may enhance their use of thematic choices, and in turn improve their writing quality. With the pedagogical benefits in mind, more and more researchers are urging to integrate Theme-Rheme instruction in EFL contexts (Belmonte & McCabe, 2003; Ebrahimi & Ebrahimi, 2012; Hawes & Thomas 2012; Lin, 2011; Liu & Liu, 2013; Yang, 2009). The following section is devoted to studies related to Theme-Rheme instruction in the L2 writing classroom.. 27.

(36) Theme-Rheme-based Instruction Recent research has applied Theme-Rheme analysis in L2 writing instruction to prove its effectiveness in teaching writing, and some attempts include Belmonte and McCabe (2003) and Yang (2009). Belmonte and McCabe (2003) used TP to help an ESL college student achieve greater cohesion in the revision and meanwhile remained his original intentions. The revised draft proved to flow better. Yet, the study included only one essay. Neither the teaching materials nor the teaching methods were clearly stated, and the criterion to evaluate cohesion was not mentioned. Yang (2009) attempted TP analysis to teach explanation writing in a large class. The researcher demonstrated to an EFL class of 50 Geography major college students in China how linear progression could help writers organize their information sequence from known to unknown. The students were required to explain and analyze the TP of a text and write a proper explanation paragraph in pairs. Some of the students had mastered the analysis and performed better in the writing task. The researcher provided a clear lesson plan in which linear progression was taught to arrange given and new information in a text. However, the intervention was too short, and the learning outcome was not reported in detail. It remained unknown why some students did not acquire TP analysis, what difficulties they encountered, and how they perceived this instruction. After these attempts, more researchers have devoted in detailed procedures of teaching Theme-Rheme or to prolong the duration of intervention (Lin, 2011; Liu & Liu, 2013). Liu and Liu (2013) demonstrated that teaching thematization could improve EFL students’ writing proficiency. Fifty Chinese college students with similar proficiency level were divided into one experimental group and one control group. The intervention lasted ten weeks with two hours’ meeting per week. Three Theme 28.

(37) types (i.e., simple, multiple, clausal) and six TP types (i.e., constant, linear, split Theme, split Rheme, summarized Theme, Rheme-to-Rheme) were taught. The experimental group improved significantly in holistic and coherence scores, and the researchers concluded that successful thematization contributed much to writing coherence and that improvement in coherence raised writing proficiency. Lin (2011) proved that the instruction of cohesive device and TP was feasible for EFL low achievers. Two eleventh graders attended a five-month writing conference during which the researcher guided them to analyze the coherence of some reading passages and applied TP and cohesive devices in their writing. The low achievers improved in holistic and coherence scores and applied linear and constant progression patterns and some cohesive devices (i.e., reference, conjunction, and reiteration) as high achievers would do. These studies support the view that receiving Theme-Rheme instruction is beneficial for L2 students to familiarize themselves with the English thematic structure, to thematize better in the writing process, and to achieve better cohesion and coherence. Given the reported effectiveness of Theme-Rheme in writing instruction, researchers seem to have reached a consensus of implementing Theme-Rheme in the L2 classroom. For example, Belmonte and McCabe (2003), Hawes and Thomas (2012), and Ebrahimi and Ebrahimi (2012) called for more attention to expand students’ inventory in thematic choices and raise their awareness of TP to improve their text cohesion and coherence. The literature reviewed so far in this chapter has compared essays written by native and non-native college students, by high and low proficient college students in EFL or ESL contexts, and by student writers and expert writers. Some of the studies had small corpora (Wei, 2013; Yang et al., 2007). Some texts are drawn from 29.

(38) different corpora, so the writing conditions are not controlled (Green et al., 2000; Herriman, 2011; Herriman & Boström Aronsson, 2009). In addition, most of the research compared EFL or ESL writing with the NS writing without using NS professionals as reference (Green et al., 2000; Herriman & Boström Aronsson, 2009; Wei, 2013, 2016; Xu, 2000). Therefore, the study explored how EFL, ESL, NS college students organize information differently by examining their frequency and use of thematic choices, which was also compared to those thematic choices of the NS professionals.. 30.

(39) CHAPTER THREE METHODOLOGY. The current study explored both the quantitative and qualitative differences of how EFL, ESL, and NS college students organized information in Theme position compared with professional writers. This chapter presents the design of the study. I will introduce the corpus data, the analytical framework, the procedures of the study, and the methods for data analysis.. The Corpus Both the student and the professional essays were obtained from the written data of the International Corpus Network of Asian Learners of English (ICNALE). The ICNALE is a collection of 1.3 million words of controlled second language (L2) essays from ten countries in Asia as well as first language (L1) data compiled by Dr. Shin'ichiro Ishikawa in Kobe University, Japan. The ICNALE controls the writing conditions (e.g. time, length, topic) of the essays and the English proficiency of the student writers, hence making it an appropriate corpus for contrastive interlanguage analysis. The writing task in the ICNALE requires writers to provide arguments or counter-arguments of two topics with reasons and specific examples. The two topics are: Topic A: It is important for college students to have a part-time job. Topic B: Smoking should be completely banned at all the restaurants in the country. The L2 writers’ proficiency in the ICNALE is divided based on their scores on TOEIC, TOEFL, IELTS, or the English vocabulary size test (VST) (Nation & Beglar, 2007) into four levels: A2 (TOEFL 56 and below), B1_1 (TOEFL 57~75), B1_2. 31.

(40) (TOEFL 76~86), and B2+ (TOEFL 87 and above). The study adopted the B2+ level in which the student writers have an equivalent TOEFL score of 87 and above. The study attempted to investigate if the deviation of advanced non-native writing resulted from the different information organization in Theme position. Intermediate and low non-native writing may contain too many grammatical and spelling errors to the extent that the information structure is undermined. Therefore intermediate and low non-native writing may not be suitable for the present study. For the comparison of thematic choices, a selection of approximately 55,900 words of expository essays from four sub-corpora has been used. The student sub-corpora in the study comprised 60 essays from Chinese advanced college students (40 from Taiwan and 20 from China), 60 essays from Singaporean advanced college students, and 60 essays from NS college students. The professional sub-corpus also comprised 60 essays written by NS English teachers. The four sub-corpora were labeled as CHI, SIN, ENS, and PRO respectively.. Unit of Analysis T-unit was chosen as the basic analysis unit. T-unit is referred to as an independent clause with other dependent clauses around it (Fries, 1995). The reason for choosing T-unit is that it is a conventional unit in studies of thematic choices. As Herriman (2011) states, T-unit highlights the importance of initial position and justifies the fact that the Theme structure of a dependent clause is restricted within a larger independent clause.. Analytical Framework The study adopted the Theme system by Halliday and Matthiessen (2013). 32.

(41) mentioned in Chapter Two. The division of Theme and. heme followed Halliday’s. (1985) definition, the first ideational component that is a participant, a process, or a circumstance. The study did not adopt the extended definition of Theme that thematizes the subject as proposed by other scholars (e.g. Berry 1995; Martin & Rose 2003; North, 2005) because Halliday’s (1985) division distinguishes better unmarked Themes (subject Themes) connected with topic continuity and marked Themes (circumstantial Themes) connected with topic discontinuity. The thematic choices under investigation are textual, interpersonal, and marked Themes because they are more marked than the subject Theme to orient readers in sentence-initial position (Ebrahimi & Ebrahimi, 2012). For clarity, a simplified Theme system is illustrated in Figure 3.. Unmarked: subject Topical (required) Marked: in school, if I… Theme. Textual (optional): and, but, besides, however. Interpersonal (optional): maybe, I think. Figure 3. A simplified Theme system in the present study. According to Figure 3, a Theme can be topical, textual, and interpersonal. Topical Themes are required while textual and interpersonal Themes are optional. Textual Themes organize information and include continuatives (e.g., yes, no, well), conjunctions (e.g., and, but, so), and conjunctive adjuncts (e.g., besides, however, therefore). Only coordinating conjunctions are analyzed in the study because subordinating conjunctions (e.g., because, although, after) are a part of dependent. 33.

(42) clauses and would not be analyzed separately as textual Themes. Interpersonal Themes present the writer’s stance. Interpersonal Themes contain vocatives (e.g., David, Sir), modals adjuncts (e.g., probably, usually), interpersonal metaphors (e.g., I think, It is likely that), and mood-marking devices (e.g., What, Do you, Let’s). Topical Themes can be divided into unmarked and marked Themes. Unmarked topical Themes are the grammatical subject. Marked Themes are Themes other than the grammatical subject such as complements (e.g., To the market we go), circumstantial adjuncts (e.g., Fortunately, we escaped the accident.), and dependent clauses (e.g., If I were you, I would spend less.).. Procedure For the procedure of the study, I first divided the 240 essays into T-units and numbered each T-unit. The total T-units were 3,106. Then I divided the 3,106 T-units into Theme and Rheme and categorized the Themes into topical, textual, and interpersonal Themes, and divided the topical Themes into unmarked and marked Themes. I calculated the raw frequency of the textual, interpersonal, unmarked and unmarked Themes before I conducted quantitative and qualitative analysis. What I concerned was how college students organized information with marked choices of Themes, including textual, interpersonal, and marked Themes (Ebrahimi &Ebrahimi, 2012) so I excluded unmarked Themes in the analysis. For quantitative analysis, I conducted a 4X3 two-way mixed design ANOVA of CHI, SIN, ENS, and PRO. For qualitative analysis, I iteratively analyzed the emerging patterns of textual, interpersonal, and marked Themes by the different groups. The procedure of the study is summarized as follows:. 34.

(43) 1. Divide the 240 essays into T-units and number each T-unit. 2. Divide each T-unit into Theme and Rheme. 3. Categorize Theme choices into topical, textual, and interpersonal Themes, and divide topical Themes into unmarked and marked Themes. 4. Calculate the raw frequency of textual, interpersonal, and marked Themes. 5. Conduct a 4X3 two-way mixed design ANOVA of CHI, SIN, ENS, and PRO in terms of textual, interpersonal, and marked Themes. 6. Iteratively analyze the emerging patterns of textual, interpersonal, and marked Themes by the different groups.. Figure 4 below is an example of analysis extracted from a CHI essay. The three columns from the left are the textual, interpersonal, topical Themes and Rhemes of the essay. The three columns from the right are the analysis notes. The “Theme type” column in the middle records the elements of Theme, which can be a topical Theme or a topical Theme fronted by textual and/or interpersonal Themes. The types of textual, interpersonal, and marked Themes are tracked in “TX/IN Theme” and “marked Theme” columns. For example, the marked Theme If we take part-time jobs is a contingency of condition, so it is coded as contingency/condition. Similarly, For example is an appositive device of textual Themes, so it is coded as textual/adjunct/appositive.. 35.

(44) Essay. Analysis Notes. TX/IN Theme topical Theme. Rheme. Theme type. If we take. it is likely for us to. topical. part-time jobs,. distract from our. TX/IN Theme. marked Theme contingency/condition. studying. We. may spend too much. topical. time and energy. For example,. we. may lose time needed. textual^topical. textual/adjunct/appositive. for sleep, rest, study, club activities and recreation.. Figure 4. An analysis example. Data Analysis For quantitative analysis, I conducted a 4X3 Two-Way mixed design ANOVA. The independent variables were the four linguistic backgrounds (i.e., CHI, SIN, ENS, RPO) and the dependent variables were the three thematic choices (i.e., textual, interpersonal, marked). The quantitative analysis was conducted through the IBM SPSS version 22.0. Before ANOVA, the descriptive statistics was run to ensure the accuracy of the data input in SPSS. Also, Box's Test of Equality of Covariance and Levene's Test of Equality of Error Variances were conducted to confirm that the data was suitable for ANOVA. Results in Box's Test of Equality of Covariance (sig. = 0.22) showed that the dependent variables are equal across groups, and Levene's Test of Equality of Error Variances (sig. = 0.16, 0.57, 0.66) showed that the error variance of the dependent variable is equal across groups, both indicating that the data was suitable for ANOVA. 36.

(45) The ANOVA results would discover if there were any significant differences of textual, interpersonal, and marked Themes in the four groups. The significance level was set at 0.05. For post hoc comparisons, Fisher's Least Significant Difference (LSD) was adopted. As to qualitative analysis, I iteratively investigated the emerging patterns of textual, interpersonal, and marked Themes used by the different groups. First, I investigated if the textual Themes used were informal or formal under the Theme system (Halliday & Matthiessen, 2013). Informal textual Themes include continuatives (e.g., yes, no, well) and conjunctions (e.g., and, but, so), both of which are more common in spoken English (Halliday & Matthiessen, 2013). Formal textual Themes are conjunctive adjuncts (e.g., besides, however, therefore), which are more prevalent in written English (Halliday & Matthiessen, 2013). Second, I examined whether interpersonal Themes were subjective or objective using Halliday’s (1994, p. 355) “modal responsibility” scale: from explicit subjective, implicit subjective, implicit objective, to explicit objective. The examples are given below with interpersonal Themes underlined. The four examples have a similar ideation meaning, but the interpersonal meanings are different. In (1), the writer’s opinion is expressed explicitly through the writer’s subjective mental process (i.e., I think). In (2), the writer’s subjective opinion is incorporated in the plural pronoun and the modal verb (i.e., We may think). In (3) and (4), the writer’s point of view is presented as objective judgment rather than opinions. In (3), the judgment is expressed through the modal adjunct (i.e., Clearly) while in (4), the judgment is presented explicitly through a third person impersonal pronoun (i.e., It is important that). (1) explicit subjective: I think part-time jobs are important for college students.. 37.

(46) (2) implicit subjective: We may think that part-time jobs are important. (3) implicit objective: Clearly, part-time jobs are important for college students. (4) explicit objective: It is important that college students have part-time jobs.. Finally, I scrutinized whether marked Themes effectively continued the topic under the dichotomy of topic continuity and topic discontinuity (Halliday, 1985). The examples are given below. In (5), the topic in question (i.e., Smoking should be banned in all restaurants) is continued in the following marked Theme (i.e., Only in this way). By contrast, in (6), the topic (i.e., A non-smoking person) is not continued in the next marked Theme. The marked Theme they refers to smokers rather than non-smokers. (5) marked Theme that continues the topic: Smoking should be banned in all restaurants. Only in this way can we breathe more flesh air in the future! (6) marked Theme that discontinues the topic: A non-smoking person should be able to sit down in a restaurant and not inhale unhealthful smoke while they eat. If they want to smoke they can go outside.. I coded the thematic choices and deducted major themes from the codes. The qualitative differences, if any, would reveal how the students construct meanings in the expository genre and uncover strengths and weaknesses of each group in manipulating Themes.. 38.

(47) CHAPTER FOUR RESULTS. The study investigated how EFL, ESL, and NS college students organize information compared to NS professionals by examining their thematic choices. Generally, it was found that the EFL students deployed significantly more textual and marked Themes than NS professionals to achieve cohesion explicitly. The ESL students utilize patterns of Themes similar to those of the NS professionals. The NS students had significantly more textual and interpersonal Themes than NS professionals, and the textual and interpersonal Themes tend to contain more informal registers.. Quantitative Differences of Thematic Choices in the Groups Research question 1 inquires the quantitative differences of textual, interpersonal, and marked Themes in the EFL, ESL, and NS groups compared to NS professionals. Table 2 below reports the mean and standard deviation of the thematic choices in the four groups. The last row shows the mean and standard deviation of the total textual, interpersonal, and marked Themes. Textual Themes (M=4.71, SD=2.50) are the most frequent type and interpersonal Themes (M=2.62, SD=1.90) are the least frequent type used by the four groups. The far right column shows the mean and standard deviation of the total Themes applied by each group. CHI ranks first in total Themes (M=4.22, SD=1.04) with the highest textual Themes (M=6, SD=2.19) and the highest marked Themes (M=4.1, SD=1.68). ENS ranks second in total Themes (M=3.70, SD=1.44) with the highest interpersonal Themes (M=3.35, SD=2.05). SIN ranks third in total Themes (M=3.14,. 39.

(48) SD=1.03). The mean of SIN’s textual (M=4.17, SD=2.39), interpersonal (M=2.27, SD=1.74), and marked Themes (M=2.98, SD=1.62) is close to those of PRO. PRO has the fewest total Themes (M=2.83, SD=1.03), with the lowest textual (M=3.53, SD=1.99) Themes.. Table 2. Mean and standard deviation of different Themes in the four groups textual. interpersonal. marked. Total Themes. Group. M. SD. M. SD. M. SD. M. SD. CHI. 6. 2.19. 2.55. 2.05. 4.1. 1.68. 4.22. 1.04. SIN. 4.17. 2.39. 2.27. 1.74. 2.98. 1.62. 3.14. 1.03. ENS. 5.13. 2.69. 3.35. 2.05. 2.62. 1.35. 3.70. 1.44. PRO. 3.53. 1.99. 2.32. 1.53. 2.63. 1.73. 2.83. 1.03. Total. 4.71. 2.50. 2.62. 1.90. 3.08. 1.70. 3.47. 1.26. Note: M = Mean, and SD = Standard Deviation.. In order to clarify if the different patterns of thematic choices reach a statistically significant level, a 4X3 two-way mixed design ANOVA is conducted. The significant level is set at the .05. The result of the ANOVA is presented in Table 3. According to Table 3, there is a significant difference in the main effects of Themes (F=77.23, p<.01), indicating that the quantitative differences of the three Themes reach a significant level. There is also a significant difference in the main effects of groups (F=17.19, p<.01). This suggests that the four groups deploy Themes differently, which is also statistically significant. Also, the interaction between Themes and groups reaches significant level (F=6.28, p<.01). The ANOVA result indicates that the frequency of the three Themes is significantly different in the groups. A post-hoc test is needed to locate any significant differences of the three Themes in the four groups.. 40.

(49) Table 3. Two-way ANOVA result. Source. Sum of Squares. df. Mean Square. F. Sig.. Theme. 576.98. 2. 288.49. 77.23. .00. Group. 203.85. 3. 67.95. 17.19. .00. Theme * group. 140.64. 6. 23.44. 6.28. .00. The result of the post-hoc test is reported in Table 4 below. In terms of total Themes, CHI, SIN, and ENS deploy different number of textual, interpersonal, and marked Themes and this difference reaches significant level. CHI has significantly more Themes than SIN (M: 4.22 vs 3.14, p<.01) and ENS (M: 4.22 vs 3.70, p<.05), and ENS has significantly more Themes than SIN (M: 3.70 vs 3.14, p<.05). When compared with PRO, CHI and ENS have significantly more Themes than PRO (M: 4.22 vs 3.70 vs 2.83, p<.01), and there is no significant difference between SIN and PRO (M: 3.14 vs 2.83, p>.05).. 41.

(50) Table 4. Multiple comparisons of different Themes in the four groups Dependent Variable total. textual. interpersonal. marked. (I) group (J) group CHI SIN ENS PRO SIN CHI ENS PRO ENS CHI SIN PRO PRO CHI SIN ENS CHI SIN ENS PRO SIN CHI ENS PRO ENS CHI SIN PRO PRO CHI SIN ENS CHI SIN ENS PRO SIN CHI ENS PRO ENS CHI SIN PRO PRO CHI SIN ENS CHI SIN ENS PRO SIN CHI ENS PRO ENS CHI SIN PRO PRO CHI SIN ENS. Mean Difference (I-J) 1.08** .52* 1.39** -1.08** -.56** .31 -.57* .56** .87** -1.39** -.31 -.87** 1.83** .87* 2.47** -1.83** -.97* .63 -.87* .97* 1.60** -2.47** -.63 -1.60** .28 -.80* .23 -.28 -1.08** -.05 .80* 1.08** 1.03** -.23 .05 -1.03** 1.12** 1.48** 1.47** -1.12** .37 .35 -1.48** -.37 -.02 -1.47** -.35 .02. Std. Error .21 .21 .21 .21 .21 .21 .21 .21 .21 .21 .21 .21 .42 .42 .42 .42 .42 .42 .42 .42 .42 .42 .42 .42 .34 .34 .34 .34 .34 .34 .34 .34 .34 .34 .34 .34 .29 .29 .29 .29 .29 .29 .29 .29 .29 .29 .29 .29. Sig. .00 .01 .00 .00 .01 .14 .01 .01 .00 .00 .14 .00 .00 .04 .00 .00 .02 .14 .04 .02 .00 .00 .14 .00 .40 .02 .49 .40 .00 .88 .02 .00 .00 .49 .88 .00 .00 .00 .00 .00 .21 .23 .00 .21 .96 .00 .23 .96. 95% Confidence Interval Lower Bound Upper Bound .66 1.49 .10 .93 .98 1.80 -1.49 -.66 -.97 -.15 -.10 .72 -.93 -.10 .15 .97 .46 1.29 -1.80 -.98 -.72 .10 -1.29 -.46 1.00 2.67 .03 1.70 1.63 3.30 -2.67 -1.00 -1.80 -.13 -.20 1.47 -1.70 -.03 .13 1.80 .76 2.43 -3.30 -1.63 -1.47 .20 -2.44 -.76 -.38 .95 -1.47 -.13 -.43 .90 -.95 .38 -1.75 -.42 -.72 .62 .13 1.47 .42 1.75 .37 1.70 -.90 .43 -.62 .72 -1.70 -.37 .54 1.69 .91 2.06 .89 2.04 -1.69 -.54 -.21 .94 -.23 .93 -2.06 -.91 -.94 .21 -.59 .56 -2.04 -.89 -.93 .23 -.56 .59. Note: Based on observed means. The error term is Mean Square (Error) 1.32. *. p<.05, **p<.01. 42.

(51) In terms of textual Themes, CHI has significantly more textual Themes than SIN (M: 6 vs 4.17, p<.01) and ENS (M: 6 vs 5.13, p<.05), and ENS has significantly more textual Themes than SIN (M: 5.13 vs 4.17, p<.05). Compared with PRO, CHI and ENS have significantly more textual Themes than PRO (M: 6 vs 5.13 vs 3.53, p<.01) and no significant difference is found between SIN and PRO (M: 4.17 vs 3.53, p>.05). With regard to interpersonal Themes, there is no significant difference between CHI and SIN (M: 2.55 vs 2.27, p>.05) while ENS has significantly more interpersonal Themes than CHI (M: 3.35 vs 2.55, p<.05) and SIN (M: 3.35 vs 2.27, p<.01). When compared with PRO, there is no significant difference between CHI, SIN, and PRO (M: 2.55 vs 2.27 vs 2.32, p>.05), and ENS has significantly more interpersonal Themes than PRO (M: 3.35 vs 2.32, p<.01). As to marked Themes, CHI uses significantly more marked Themes than SIN and ENS (M: 4.1 vs 2.98 vs 2.62, p<.01) while there is no significant difference between SIN, ENS, and PRO (M: 2.98 vs 2.62 vs 2.63, p>.05). In sum, CHI, SIN, and ENS deploy different amounts of Themes and this difference reaches a significant level. When compared with PRO, CHI uses significantly more textual and marked Themes (p<.01). ENS applies significantly more textual and interpersonal Themes (p<.01). There is no significant difference in each of the Themes between SIN and PRO (p>.05). The result suggests that CHI and ENS deviate from PRO in certain Themes and that SIN exhibits patterns of thematic choices similar to those of PRO. This section has reported the quantitative difference of thematic choices in the four groups. The following section will report the results of the qualitative analysis of the Themes in the four groups.. 43.

(52) Qualitative Differences of Thematic Choices in the Groups Research question 2 investigates the qualitative differences of textual, interpersonal, and marked Themes in the EFL, ESL, NS student groups compared to NS professionals. Specifically, the study explores if the textual Themes are informal or formal, if the interpersonal Themes are subjective or objective, and if the marked Themes develop the topic.. Informal and Formal Textual Themes In the previous quantitative analysis, it is found that both CHI and ENS have significantly more textual Themes than PRO while there is no significant difference between SIN and PRO in textual Themes. The qualitative analysis reveals that CHI, SIN, and ENS deploy textual Themes in different manners and that SIN uses textual Themes in a way similar to PRO. An analysis between the textual Themes by CHI and ENS reveals that CHI tends to use formal conjunctive adjuncts while ENS opts for informal conjunctions. In fact, in CHI’s essays, the number of conjunctive adjuncts doubles that of conjunctions. In 43 of the 60 essays by CHI, there exist more conjunctive adjuncts than conjunctions, and 33 of the 43 essays draw on conjunctive adjuncts of enumeration (e.g., first of all, second, finally) to organize information. Example 14 below illustrates the textual Theme pattern in CHI’s essays. In Example 14, seven textual Themes are located in eight T-units. Five of the textual Themes are conjunctive adjuncts of enumeration, addition, and summary (i.e., First, Besides, In addition, Last but not the least, As a result) and two are conjunctions (i.e., and).. 44.