中文博士論文摘要中連接性成分之使用

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. 中文博士論文摘要中連接性成分之使用. The Use of Connectives in Chinese Doctoral Dissertation Abstracts. 指導教授：張. 妙. 霞. Advisor: Dr. Miao-Hsia Chang 研究生：李. 懿. 倫. 中華民國一〇三年八月 August, 2014.

(2) 摘要學術摘要是具有高密度邏輯論證的文類，這些邏輯論證常依靠顯性連接成分連接(Swales 1990)。本文旨在探討中文博士論文摘要中，不同領域作者如何使用顯性連接成分表達論證間的邏輯語意關係，以及學科領域差異如何影響作者使用這些連接性成分。本研究共選取九十篇國立大學博士論文摘要作為語料，依領域分成人文、工程、社會科學各三十篇。連接性成分的分類標準依據功能語言學家 Halliday (1994)，分成詳述(Elaboration)、延伸(Extension)、增強(Enhancement)三大類，僅含括非命題性與指涉語篇內部等具有後設論述功能的連接成分。在詳述連接成分方面，統計結果顯示人文領域作者比工程與社會領域作者明顯使用更多連接成分。在詳述連接成分的次類中，闡明類(Clarification)有整體顯著差異。這與學科論證內容有關。人文領域的論證內容較多理論性與概念性的抽象成分，因此作者需要更多的主觀說明闡釋才能清楚表達論證間的邏輯語意關係。在工程領域摘要中未發現總結成分(Summative)，顯示作者較不傾向於使用顯性標記幫助讀者總結論證重點，作者與讀者的關係較疏離。此外，人文領域作者使用較多種類的闡明性連接成分，而工程領域作者使用最少種類以及最少的語言形式，顯示領域差異會影響語言豐富度與篇章修辭。在延伸連接成分方面，變換類(Variation)在人文和工程領域作者間有顯著差異，特別是取代連接成分(Replacive)，人文領域作者使用最多而工程領域作者未使用。這與科學寫作規範有關。在科學工程領域中，作者較無法反駁已知的定理，以及作者較期待自身研究是正面支持前人研究。人文領域則相反，作者透過與前人不同的論述鞏固自己的核心論點，找出與前人的差異才是貢獻所在，因此使用較多取代連接成分。在增強連接成分方面，領域間無差異，顯示增強連接成分是中文學術寫作的必要手段，與學科領域較無關係。增強連接成分有四個次類。時空類(Spatiotemporal)也是學術寫作必要手段，領域間差異不明顯。方式類(Manner)裡則是手段(Means)連接成分使用多於比較(Comparison)連接成分。這是因為學術摘要中，描述研究方法是必要語步(move)(Pho 2008)，因此手段連接成分較多。在工程領域摘要中未發現反面比較連接成分(Negative Comparison)，這與作者期待正面支持前人研究有關，傾向於找到共通點而非相異點。因果條件類(Causalconditional)則是工程領域作者使用最多，人文領域作者最少，尤其在條件 (Conditional)次類，人文領域作者完全未使用。這是因為工程領域必須清楚陳述實驗條件，實驗條件細節對於結果可靠性影響重大。人文領域作者則是較傾向從綜觀的角度，全盤分析各種可能性，因此較無單獨設定條件的需要。此外，工程領域作者明顯依賴因果條件類連接成分多於增強連接的另三個次類，顯示作者使用的語言資源較為單一，豐富性不高。最後則是方面類(Matter)連接成分，領域間無顯著差異，也是學術寫作必要手段，可以幫助作者清楚陳列討論 i.

(3) 面向，使讀者易於掌握。總而言之，人文領域與工程領域間差異較大，與社會科學領域差異較小，連接成分的使用差異與學科知識內容以及學科寫作規範有關。整體而言，人文領域作者使用較多種類的連接成分，展現較豐富的語言形式。工程領域作者則相當依賴少數的類別，使用較不豐富的語言資源進行論證間邏輯語意銜接。關鍵詞: 學術語篇、博士論文摘要、連接性成分、語篇互動標記、領域差異、中文學術寫作. ii.

(4) Abstract Connectives in this study are defined as cohesive devices which guide readers through the text and help writers structure the logico-semantic relations among arguments. They belong to interactive metadiscourse (Thompson 2001) and primarily denote non-propositional and text-internal meanings (Hyland 2005). With a databank comprising ninety doctoral dissertation abstracts chosen from eight national universities in Taiwan, this study investigates the use of connectives in the humanities, engineering, and social sciences. The coding schema follows Halliday (1994), classifying the connectives into Elaboration, Extension, and Enhancement. In the Elaboration category, the humanities writers use connectives significantly more than the engineering and social sciences writers. This reflects the nature of knowledge in the humanities, where writers often deal with highly conceptual and theorized ideas, thus the abstractness of knowledge contents requires writers employ more connectives to explicitly highlight the logico-semantic relations among arguments. Overall speaking, the humanities writers use connectives with a greater variety, whereas the engineering writers exploit connectives with a rather restricted repertoire. This observation confirms Peacock’s (2010) study. In terms of Extension, the humanities writers use the Variation significantly more than the engineering writers, especially in the Replacive subcategory. The Replacive connectives (e.g. 反之 ‘instead’) link a counterclaim with the preceding argument, conveying a tone of denial and refusal. They are not preferred by the engineering writers because the scientific conventions encourage writers to seek common grounds with the previous literature (Becher 1994). On the other hand, the humanities writers show contribution to the academic community by indicating the differences between their interpretation and previous knowledge. Therefore, more Replacive connectives were used in the humanities abstracts. As for the Enhancement, the disciplinary variations were not significant in statistics, suggesting that the Enhancement connectives are deployed by writers for the requirements of Chinese academic writing, instead of showing the contrast between the nature of knowledge and disciplinary writing. In sum, this study supports Swales’s (1990) observation that the use of interactive metadiscourse is a key differentiating feature in the writing of doctoral dissertation abstracts, and disciplinary variations can be unveiled through examining the use of text-oriented connectives in these abstracts. Keywords: academic discourse, doctoral dissertation abstract, connective, interactive metadiscourse, disciplinary variation, Chinese academic writing. iii.

(5) Acknowledgements. There are several people without whom this thesis would never have come to the final stage. First and foremost, I owe my deepest gratitude to Professor Miao-Hsia Chang, who not only supervised my thesis but also gave critical comments on the logicality of argumentative writing. Around two years ago, I had been struggling with the thesis topic, hence I suspended the graduate study for one year and enlisted in the army. During the year of suspension, Professor Chang kept to care about me and always gave me the warmest encouragement. Without her persistent encouragement, this thesis would not have materialized. I also owe thanks to Professor Yung-O Biq, who gave me constructive comments in the proposal hearing. Special gratitude is also extended to the members of my thesis defense committee, Professor I-Wen Su and Professor Hsi-Yao Su, who went through my thesis very carefully, and their meticulous comments were an enormous help to me in the process of revision. Their suggestions were invaluable. Many thanks also go to the Department of English at NTNU, where a pleasant and stimulating linguistic atmosphere was provided. I would like to show the greatest appreciation to my classmates, including Elvis, Sally, Lucy, Chris, Aaron, Caroline, Toro, and Frank. They supported me a lot in the course of graduate study. I would particularly like to thank Iris Chen for her encouragement during suspension. The most heartfelt appreciation must be to my parents. It is they who, in the past years, helped me get through the most melancholic days in my life. On a more personal note, finally, I would like to thank Meng-Ju Lee for her accompanying me through the toughest times and being patient with an ill-tempered child like me. Without her, I would not be capable of going through all the difficulties.. iv.

(6) Table of Contents 摘要................................................................................................................................. i Abstract ....................................................................................................................... iii Acknowledgements ..................................................................................................... iv Table of Contents ......................................................................................................... v List of Tables ..............................................................................................................vii Chapter 1: Introduction .............................................................................................. 1 1.1 Background and motivation ................................................................................. 1 1.2 Objectives of the study......................................................................................... 4 1.3 Definition of connectives ..................................................................................... 8 1.4 Organization of the thesis .................................................................................... 9 Chapter 2: Literature Review ................................................................................... 10 2.1 Defining connectives ......................................................................................... 10 2.1.1 Connectives as metadiscourse .................................................................... 10 2.1.2 Connectives in English grammar ................................................................ 13 2.1.2.1 Quirk et al.’s (1985) conjuncts............................................................. 14 2.1.2.2 Biber et al.’s (1999) linking adverbials................................................ 17 2.1.2.3 Celce-Murcia and Larsen-Freeman’s (1999) logical connectors ......... 20 2.1.2.4 Huddleston and Pullum’s (2002) connective adjuncts......................... 21 2.1.2.5 Carter and McCarthy’s (2006) linking adjuncts .................................. 23 2.1.3 Connectives in Chinese grammar ............................................................... 24 2.2 Connectives and conjunction in discourse ......................................................... 27 2.2.1 Logical semantic relations .......................................................................... 28 2.2.2 Halliday’s (1994) taxonomy ....................................................................... 30 2.3 Connectives in academic settings: Empirical studies ........................................ 33 2.3.1 The use of connectives in different genres ................................................. 35 2.3.1.1 Research articles and research article abstracts ................................... 36 2.3.1.2 Master’s theses ..................................................................................... 37 2.3.1.3 Doctoral dissertations........................................................................... 39 2.3.1.4 Students’ writing .................................................................................. 41 2.3.2 The use of connectives in different disciplines ........................................... 44 2.3.3 The use of connectives in different languages ............................................ 45 2.4 Summary ............................................................................................................ 47 Chapter 3: Methodology............................................................................................ 49 3.1 Coding scheme ................................................................................................... 49 3.1.1 Elaboration .................................................................................................. 50 3.1.2 Extension .................................................................................................... 52. v.

(7) 3.1.3 Enhancement ............................................................................................... 54 3.2 Data collection ................................................................................................... 58 3.2.1 Disciplines .................................................................................................. 58 3.2.2 Doctoral dissertation abstracts .................................................................... 61 3.3 Data treatment and analysis ............................................................................... 63 3.3.1 Level of treatment ....................................................................................... 64 3.3.2 Steps of analysis.......................................................................................... 65 3.4 Summary ............................................................................................................ 68 Chapter 4: Results and Discussion ........................................................................... 70 4.1 General distribution of connectives ................................................................... 70 4.2 Elaboration ......................................................................................................... 73 4.2.1 Quantitative results ..................................................................................... 73 4.2.2 Qualitative discussion ................................................................................. 74 4.2.2.1 Apposition ............................................................................................ 75 4.2.2.2 Clarification ......................................................................................... 78 4.3 Extension............................................................................................................ 83 4.3.1 Quantitative results ..................................................................................... 83 4.3.2 Qualitative discussion ................................................................................. 85 4.3.2.1 Addition ............................................................................................... 86 4.3.2.2 Variation .............................................................................................. 90 4.4 Enhancement ...................................................................................................... 93 4.4.1 Quantitative results ..................................................................................... 93 4.4.2 Qualitative discussion ................................................................................. 96 4.4.2.1 Spatio-temporal .................................................................................... 96 4.4.2.2 Manner ............................................................................................... 100 4.4.2.3 Causal-Conditional ............................................................................ 103 4.4.2.4 Matter ................................................................................................. 107 4.5 Summary and discussion.................................................................................. 109 Chapter 5: Conclusion ............................................................................................. 119 5.1 Recapitulation of research................................................................................ 119 5.2 Summary of findings........................................................................................ 120 5.3 Limitations and recommendations for further research ................................... 124 References ................................................................................................................. 127 Appendix ...................................................................................................................138. vi.

(8) List of Tables Table 2.1 Semantic relations of conjuncts in Quirk et al. (1985:634-639) .................. 17 Table 2.2 Types of elaboration in Halliday’s (1994:324) conjunction system ............ 30 Table 2.3 Types of extension in Halliday’s (1994:324-325) conjunction system ....... 31 Table 2.4 Types of enhancement in Halliday’s (1994:325-326) conjunction system . 33 Table 3.1 Subtypes and examples of elaboration......................................................... 51 Table 3.2 Subtypes and examples of extension ........................................................... 53 Table 3.3 Subtypes and examples of enhancement ..................................................... 56 Table 3.4 Number of abstracts by disciplines .............................................................. 61 Table 3.5 Number of abstracts by institutions ............................................................. 63 Table 3.6 Number of words in the three disciplines .................................................... 63 Table 4.1 Distribution of connectives in the three disciplines ..................................... 71 Table 4.2 Distribution of connectives in the three categories ...................................... 71 Table 4.3 Multiple comparison of Elaboration across disciplines ............................... 71 Table 4.4 Across-discipline comparison of Elaboration .............................................. 73 Table 4.5 Multiple comparison of Clarification across disciplines ............................. 73 Table 4.6 Within-discipline comparison of Elaboration .............................................. 74 Table 4.7 Number and Frequency (per 100 clauses) of Apposition connectives ........ 75 Table 4.8 Number and Frequency (per 100 clauses) of Clarification connectives ...... 79 Table 4.9 Across-discipline comparison of Extension ................................................ 84 Table 4.10 Multiple comparison of Variation across disciplines ................................ 84 Table 4.11 Within-discipline comparison of Extension .............................................. 85 Table 4.12 Number and Frequency (per 100 clauses) of Addition connectives .......... 87 Table 4.13 Number and Frequency (per 100 clauses) of Variation connectives ......... 91 Table 4.14 Across-discipline comparison of Enhancement ......................................... 94 Table 4.15 Within-discipline comparison of Enhancement ......................................... 95 Table 4.16 Multiple comparison of Enhancement within the engineering .................. 95 Table 4.17 Multiple comparison of Enhancement within the social sciences ............. 96 Table 4.18 Number and Frequency (per 100 clauses) of Spatio-temporal connectives ................................................................................................................... 97 Table 4.19 Number and Frequency (per 100 clauses) of Manner connectives .......... 101 Table 4.20 Number and Frequency (per 100 clauses) of Causal-conditional connectives .............................................................................................. 104 Table 4.21 Number and Frequency (per 100 clauses) of Matter connectives ........... 108 Table 4.22 A summary of statistical findings ……………………………………… 110. vii.

(9) Chapter 1 Introduction 1.1 Background and motivation The motivation of the current study has its origin of interest in the extensive study on metadiscourse. Metadiscourse is ‘writing about writing’ (Williams 2003) or ‘discourse about discourse’ (Vande Kopple 1985), concerning the level of discourse at which the writers do not “add propositional material but help the readers organize, classify, interpret, evaluate, and react to such material” (Vande Kopple 1985:83). For example, linguistic elements regarded as metadiscourse can indicate logical relations such as sequencing (e.g. first, second), causality (e.g. therefore), and contrast (e.g. however). These elements not only help writers structure their texts, thus enhancing the readability, but also guide readers through the argumentation with explicit signposts (Booth 1961, Hyland 2005, Vande Kopple 1985, Williams 2003). The purpose of structuring the text and guiding readers through the argumentation is to increase the interaction between the writers and the readers so that the readers can fully understand the message and arguments conveyed by the writers. The reader-writer interactions in the written text can be achieved by two types of metadiscourse (Thompson 2001, Thompson and Thetela 1995). The first type is termed ‘interactive’ metadiscourse, which demonstrates writers’ awareness of readers’ possible reactions and needs and is used to manage the flow of information in the unfolding text (Thompson 2001:59). The second type is termed ‘interactional’ metadiscourse, which concerns the engagement of readers in the text and expresses writers’ voice by commenting or evaluating (Thompson 2001:59). The distinction between ‘interactive’ and ‘interactional’ metadiscourse is later adopted by Hyland (2005), where he incorporates interactive and interactional resources into an interpersonal model of metadiscourse (Hyland 2005:49). The 1.

(10) ‘interactive’ category involves linguistic resources which demonstrate writers’ consideration of organizing discourse, subcategorizing (i) transitions (e.g. in addition), (ii) frame markers (e.g. finally), (iii) endophoric markers (e.g. as noted above), (iv) evidentials (e.g. according to X), and (v) code glosses (e.g. in other words). The ‘interactional’ category concerns the involvement of readers into the discourse and conveys writers’ evaluative judgment toward the unfolding text, including (i) hedges (e.g. perhaps), (ii) boosters (e.g. in fact), (iii) attitude markers (e.g. surprisingly), (iv) self-mentions (e.g. we), and (v) engagement markers (e.g. consider). These explicit signposts which serve as text-structuring devices can be defined as ‘connectives.’ Among the various kinds of metadiscourse proposed by Vande Kopple (2002), text connectives, code glosses, and illocution markers are of primary concern. Text connectives show how different parts of a text are connected and organized (e.g. first, next, consequently). Code glosses help readers grasp the meaning of a previous constituent by either reformulating (e.g. in other words) or exemplifying (e.g. such as). Illocution markers demonstrate the speech or discourse act that the writers perform to the readers (e.g. to sum up, we claim that). Such metadiscoursal elements are important and have in common the characteristic of contributing to the ‘cohesion’ of a text, meeting one of the seven standards of textuality (de Beaugrande and Dressler 1981). In de Beaugrande and Dressler’s terms, the cohesion of a text can be achieved by the use of ‘junctive expressions,’ which further involve conjunction (e.g. and, moreover), disjunction (e.g. either…or), contrajunction (e.g. yet, nevertheless), and subordination (e.g. thus, therefore). The role that the metadiscoursal use of connectives plays in textual cohesion can be seen as a port to the textual study of connectives both in theoretical linguistics and in applied linguistics, which will be expounded on respectively in the 2.

(11) following. Connectives, as an umbrella term for various conjunctive linking devices, are discussed under the heading ‘conjunction,’ which is one of the means to create cohesion, in (Systemic) Functional Grammar (Gregory 1987, Halliday 1994, Halliday and Matthiessen 2004, Halliday and Hasan 1976, Hartnett 1986, Martin and Rose 2003, van Dijk 1980). The logico-semantic relations denoted by connectives may be manifested between a clause (or clausal complex) and its contiguous element, showing that a stretch of text is ‘elaborated/extended/enhanced’ by another (Halliday 1994, Halliday and Matthiessen 2004). The elaborated relation involves two subclasses, viz. apposition (e.g. in other words) and clarification (e.g. in particular). The extended relation subcategorizes addition (e.g. moreover) and variation (e.g. instead). The enhanced relation is more complex, comprising spatio-temporal (e.g. next, secondly, finally), manner (e.g. similarly), causal-conditional (e.g. as a result), and matter (e.g. in that respect). This classification of conjunctions is a finer delineation of cohesive conjunctions proposed earlier in Halliday and Hasan (1976). Connectives are also discussed under the terms such as ‘transitions’ (Bahrami 2012, Hyland 2005), ‘linking adverbials’ (Lei 2012, Peacock 2010, Shaw 2009), and ‘connectors’ (Bondi 2004, Lockman and Swales 2010) in applied fields of language study. Most of the studies from English for Academic Purposes (EAP), English for Specific Purposes (ESP), and English Language Teaching (ELT) put their emphasis on the differences between native and non-native writers in the use of connectives (e.g. Carrió-Pastor 2006, 2013, Lau 2001) and on the differences among various academic genres (e.g. Loi and Lim 2013). The differences in the use of connectives by native and non-native writers are widely investigated, focusing especially on the contrast between English L1 speakers and ESL/EFL students (e.g. Bolton et al. 2002, Granger and Tyson 1996, Milton and 3.

(12) Tsang 1993). Some of the studies argue that English L2 speakers tend to overuse English connectives (e.g. Evenson and Lintermann-Rygh 1988, Field and Yip 1992, Milton and Tsang 1993), while some research concludes that non-native speakers underuse connectives (e.g. Carrió-Pastor 2006, Ventola and Mauranen 1991). The differences of connective use in academic written texts center on the genres such as textbooks (e.g. Crismore 1984, Hyland 2005), research articles (e.g. Khedri et al. 2013), master’s theses (e.g. Burneikaitė 2008, 2009), and doctoral dissertations (e.g. Lei 2012). Among these studies, researchers often investigate a certain aspect of a certain genre. For example, Peacock (2010) examined the use of linking adverbials in research articles across disciplines, and concluded that linking adverbials are more important than previously thought due to their role as cohesive devices and the function of helping writers construct and strengthen their claims. Peacock’s (2010) study has confirmed the importance of connectives as helpful linking devices in academic written texts. When academic writers need to create cohesion or structure their texts in a logical way, they exploit a large number of connectives to help readers grasp the logical connection and the line of argumentation.. 1.2 Objectives of the study Among the academic texts investigated, dissertation, though an important type of academic discourse, has received less attention from applied linguists. The dissertation is not only part of the requirements that the doctoral students need to fulfill before being granted the degree, but also an academic work with a daunting size in length (Swales 1990). The abstract of the dissertation is a succinct summary of condensed information. With rich arguments from the dissertation to be summarized within a short abstract, the arguments in the abstracts may need to be linked together 4.

(13) by employing a plethora of connectives which can explicitly signpost the logicsemantic connections among clausal complexes. Thus, a greater use of metadiscourse, such as connectives, becomes ‘a differentiating aspect of dissertation writing’ (Swales 1990:188). Along the same line, Afful and Nartey (2014) clearly pointed out the importance of studying connectives in abstracts: The place of cohesion in the abstract cannot be marginalized, given the highly informative nature of the abstract as well as its space constraints. Authors, therefore, deploy various salient linguistic and lexical tools to establish ties (a single instance of cohesion) in a terse communicative genre such as the abstract. (Afful and Nartey 2014:95) In spite of the Afful and Nartey’s claim, to date, the abstract of dissertation has not received as much attention as, for example, the introduction and conclusion section of dissertation (e.g. Dudley-Evans 1986, Hopkins and Dudley-Evans 1988). The primary concern of this study is the ‘interactive’ dimension of metadiscourse because ‘interactive’ signals serve to link the propositional information between stretches of discourse. For example, the transition markers “help readers interpret pragmatic connections between steps in an argument” (Hyland 2005:50), signaling three discourse roles suggested by Martin and Rose (2003:127), i.e. (i) addition (e.g. furthermore), (ii) comparison (e.g. similarly), and (iii) consequence (e.g. therefore). In other words, transition markers are text-oriented and function as cohesive devices which signposts the logico-sematic relations among arguments explicitly. Therefore, this study devotes a substantial amount of space to the use of connectives in doctoral dissertation abstracts. Specifically, the following four research questions are proposed: (i). Do writers in the humanities, engineering, and social sciences reveal a general tendency toward the use of connectives in abstract writing?. 5.

(14) (ii) From a cross-discipline of view, to what extent do academic writers show disciplinary differences? (iii) From a within-discipline point of view, do writers have a predilection for certain categories of connectives? (iv) If there are disciplinary differences or categorical preferences, how can we account for in terms of the nature of disciplinary knowledge and the culture of discourse community? The first research question aims to reveal a general tendency in the use of connectives across disciplines. In Liu’s (2001) investigation of Chinese cohesive conjunctives in five written genres, he found that, based on Halliday’s classification, all genres contain the most Enhancement connectives, followed by the Extension and then Elaboration connectives. This suggests that, if a general tendency can be found, the use of certain kind of connectives may not reflect disciplinary variations or differences among genre, but fulfill the general requirements of Chinese writing. In this study, we will examine the use of connectives to see if there is a general tendency among the different disciplines in terms of the use of connectives. The second research question concerns differences across disciplines. According to Koltay (2010), a scientific article generally follows the so-called IMRD (Introduction, Methods, Results, Discussion) structure, which is also followed by the social scientists. Articles in the humanities, however, are not found with the IMRD structure. Koltay (2010) further suggests that these differences may originate from information needs and how these needs are articulated. If so, we can make a hypothesis that the humanities writers may need to exploit more connectives to agglutinate various parts of information together in the transition of moves, while the social sciences and natural sciences writers may need fewer connectives among transitions than the humanities writers. We will examine whether this hypothesis is supported by the data we analyze. 6.

(15) The third research question places its emphasis on the differences within discipline, viz. the categorical preferences in a single disciple. In Halliday’s (1994) taxonomy, conjunctions are broadly categorized into Elaboration, Extension, and Enhancement. For each category, there are a number of subcategories. Writers in a single discipline may not use an equal proportion of connectives in each subcategory, but they may have a preference toward a certain kind of connectives. Therefore it is necessary to delve into the categorical preferences in a single discipline. The fourth research question is to see whether we can provide a sound explanation toward the across- and within-discipline differences as well as their relationships with the nature of disciplinary knowledge and cultures. Academic communities are often viewed as being composed of different ‘tribes’ and individual ‘territories’ (Becher and Trowler 2001), distinguishing from each other with their own language or dialect as well as “a variety of symbolic ways of demonstrating [their] apartness from others” (Bailey 1977:212). Lemke (1998:87) also suggests that the scientific concepts are “semiotic hybrids, simultaneously and essentially verbal, mathematical, visual-graphical, and action-operational.” Thus, the nature of disciplinary knowledge and idiosyncratic community cultures may influence writers’ exploitation of linguistic resources and can shape the line of argumentation into a text that is acceptable by disciplinary conventions. To be brief, this study aims to investigate the use of connectives in Chinese doctoral dissertation abstracts from across- and within-discipline point of view. The discussion will be focusing on how writers in different disciplines bridge the gap between arguments with the help of connectives, reflecting the logical-semantic connection conventions shaped by the nature of knowledge and disciplinary cultures.. 7.

(16) 1.3 Definition of connectives In Section 1.1 we have elaborated on the metadiscoursal functions of connectives. In this section we consider the definition of connectives opined in Hallidayan grammar. Halliday uses the term ‘conjunction’ as the heading in discussing the conjunctive relations among clauses. Conjunction is a type of cohesion, including semantic relations such as elaboration, extension, and enhancement. These relations are realized grammatically by ‘conjunctive Adjunct’ or a small set of ‘conjunctions’ (e.g. and, but, then) which typically occur in the clause-initial position (Halliday 1994:324). The ‘conjunctive (discourse) Adjuncts’ are defined as “those which relate the clause to the preceding text” (Halliday 1994:49). They are often adverbial groups (e.g. briefly, conversely, likewise) or prepositional phrases (e.g. in addition, as a result). In this study, ‘connectives’ include both ‘conjunctive (discourse) Adjuncts’ and ‘conjunctions’ and limit the conjunctions to those which link clausal units. In terms of the definition of connectives in Chinese grammar, Yip and Rimmington (2004:328) distinguish ‘conjunctions’ from ‘conjunctives’. Conjunctions involve linguistic items which link words or phrases (e.g. 和 hé ‘and’, 並 bìng ‘also’) and those which connect clauses (e.g. 因為 yīnwèi ‘because’). On the other hand, conjunctives are monosyllabic referential adverbs which occur toward the beginning of a clause (e.g. 就 jiù ‘then’, 才 cái ‘only then’) (Yip and Rimmington 2004:331). However, connectives in this study are not necessarily monosyllabic. The connectives should cover both ‘clausal conjunctions’ (e.g. 不但…而且 búdàn…érqiě ‘not only…but also’) and ‘clausal conjunctives’ (e.g. 為了 wèile ‘in order to’), which can be monosyllabic (e.g. 即 jí ‘i.e.’) or disyllabic (e.g. 因此 yīncǐ ‘therefore’). In sum, connectives can be defined as an open set of cohesive devices which 8.

(17) have the following features: (i). They function as interactive metadiscourse at the textual level.. (ii) They link clauses or clausal complex, instead of words or phrases. (iii) They are realized by a variety of grammatical guises, including conjunctions, (clausal) conjunctives, adverbial or prepositional phrases, etc. In other words, linguistic resources which can be counted as connectives depend on their textual functions in the discourse. Connectives should cover most of the conjunctive (discourse) Adjuncts in Halliday (1994) and clausal conjunctions and conjunctives in Yip and Rimmington (2004). They are a function-based open set of linguistic resources which operate syntactically on the clausal levels and pragmatically on the textual levels.. 1.4 Organization of the thesis The background, motivation, and objectives of the current study have been presented in Chapter 1, along with a preliminary survey of literature. Chapter 2 goes deep into the previous literature, reviewing studies on metadiscourse, English and Chinese connectives, and empirical research of connectives in academic discourse thoroughly. Chapter 3 describes the analytical framework, i.e. the taxonomy of conjunctions proposed by Halliday (1994) and Halliday and Matthiessen (2004). Chapter 4 presents the results of analysis and discusses these results with a focus on the across- and within-discipline differences, trying to account for the disciplinary choices of connectives with the nature of knowledge and discourse community cultures. Chapter 5 concludes this study.. 9.

(18) Chapter 2 Literature Review 2.1 Defining connectives As a widely investigated topic in theoretical linguistics and applied linguistics, connective is an umbrella term for linguistic units whose primary function is to link different stretches of discourse. Many grammatical categories, such as conjunction and conjunctive adverbials, may serve as connectives (Crystal 1997, Richards et al. 1992), thus the idea of connectives used in this study needs to be further delimited. In this chapter, I will first show how the idea of connectives develops in metadiscourse study. Next, linguistic units which can function as connectives in English grammar are reviewed, and at the same time the inadequacy of metadiscoursal study on Chinese connectives, especially connectives used in academic settings, must be pointed out. In order to solve this problem, a taxonomy provided by functional linguist M. A. K. Halliday (1994) will be introduced, and it is also the theoretical framework adopted in this study. Lastly, previous empirical studies on connectives in academic settings will be discussed.. 2.1.1 Connectives as metadiscourse Since Harris (1959) coined the term “metadiscourse,” the concept of metadiscourse has been developed and interpreted by scholars in rhetoric theory and composition pedagogy (e.g. Crewe 1990, Crismore 1989, Mauranen 1993, Vande Kopple 1985, Williams 2003) and applied linguistics (e.g. Hyland 2005, Ifantidou 2005, Thompson 2001). The most comprehensive work is Hyland (2005), in which metadiscourse is defined as a “cover term for the self-reflective expressions used to negotiate interactional meanings in a text, assisting the writer (or speaker) to express a viewpoint and engage readers as members of a particular community” (Hyland 10.

(19) 2005:37). This definition can be viewed in two parts. One is the linguistic resources used to express the writer’s viewpoint, guiding the reader through the structure of arguments in a discourse. The other is the linguistic resources used to engage readers in an unfolding text, aligning them with the writer. The former linguistic resources with textual functions are of primary concern in the study of connectives. Connectives relate to textual metadiscourse in that they “help readers recognize how our texts are organized and see exactly how different parts of them are connected to each other” (Vande Kopple 1985:83). These “text connectives” (Vande Kopple 1985:83) are words or phrases indicating sequences (e.g. first, next, in the third place) and those indicating logical or temporal relationships (e.g. however, nevertheless, as a consequence, at the same time). Moreover, the linguistic units which remind the reader about previous text (e.g. as I noted in Chapter One), announce the following text (e.g. as we shall see in the next section), and topicalize a new information (e.g. as for, in regard to) also belong to Vande Kopple’s category of text connectives. Based on Vande Kopple’s classification, Crismore et al. (1993) propose a slightly modified model of textual metadiscourse. They use the term “text markers” to include logical connectives, sequencers, reminders, and topicalizers (Crismore et al. 1993:47). Regardless of their labeling of the linguistic units connecting different stretches of discourse, these “text connectives” (Vande Kopple 1985) or “text markers” (Crismore et al. 1993) function mainly as a way of guiding the reader through our text smoothly and “accommodat[ing] its probable knowledge, interests, rhetorical expectations and processing abilities” (Hyland 2005:49). Guiding readers through the text as smoothly as possible is one of the two main interaction types identified in Thompson and Thetela (1995). This function of connectives is termed “interactive” (Widdowson 1984, Hoey 1988, Thompson and Thetela 1995, Thompson 2001), and it “involve[s] the management of the flow of 11.

(20) information and thus serve[s] to guide readers through the content of the text” (Thompson 2001:59). Following Thompson’s (2001) dichotomy between interactive and interactional resources, Hyland’s (2000, 2001, 2004, 2005, 2010) interpersonal model of metadiscourse comprises interactive and interactional category. It is the interactive category that performs organizational function in a text, concerning writer’s endeavor to construct a text in a cohesive and logical way. The interactive category involves five subcategories: transitions, frame markers, endophoric markers, evidentials, and code glosses (Hyland 2005:49-52). Transition markers are mainly conjunctives and adverbials signaling additive, causative, or contrastive relations (e.g. in addition, thus, but). Frame markers are used to schematize arguments in a discourse (e.g. first, next, then, finally, to conclude, my purpose is). Endophoric markers are similar to Vande Kopple’s (1985) reminders and announcements, referring to other parts of the text (e.g. as noted above, see Figure 2, in section 2). Evidentials attribute an argument to a particular resource (e.g. according to X, Y states). Code glosses elaborate, clarify, exemplify, or reformulate what has been said (e.g. in other words, that is, for example, namely, such as). Within the five subcategories, transition markers are the most intuitively obvious subcategory belonging to connectives, but connectives in the broad sense may include members that can fall into the other four subcategories. What is worth noting here is that there is not always a clear-cut distinction between the interactive category and interactional category of metadiscourse. The two aspects of interaction are “essentially the two sides of the same coin” (Thompson 2001:61). Hyland and Tse (2004:161) also suggest that “all metadiscourse is interpersonal in that it takes account of the reader’s knowledge, textual experiences, and processing needs and that it provides writers with an armoury of rhetorical appeals to achieve this.” The following example adopted from Hyland (2004:163) 12.

(21) shows the fuzziness: (1) A parametric estimation technique using global optimization is introduced for the output space partition. But we first discuss the optimization technology in the next section. (EE PhD) The italicized phrase is a frame marker, not only relating a clause to the preceding text, but also expressing writer’s intention of signaling the logical relation between his arguments. Therefore, text connectives not only fulfill their roles as interactive metadiscourse, but also engage the reader in the writer’s thread of thought and argument.. 2.1.2 Connectives in English grammar The classification of connectives in metadiscourse studies has been reviewed in 2.1.1, and it is the metadiscoursal functions of connectives that are of interest here. Although the cover term “connectives” is used in this study, the linguistic units as linking devices acquire their names under many different headings in English grammar. These terms include, among others, “adverbial connectors” (Lenker 2010), “conjunctive adverbials” (Bussmann 1996, Celce-Murcia and Larsen-Freeman 1999, Kaplan 1995), “conjuncts” (Finch 2000, Quirk et al. 1985, Wales 2011), “connective adjuncts” (Huddleston and Pullum 2002), “connective adverbs” (Huddleston and Pullum 2002), “connectives” (Finch 2000, Huddleston and Pullum 2002, Richards 1992, Warner 1985), “connectors” (Chalker 1996), “linking adjuncts” (Carter and McCarthy 2006, Sinclair 1990), “linking adverbials” (Biber et al. 1999), “linking signals” (Leech and Svartvik 1975), “logical connectors” (Celce-Murcia and LarsenFreeman 1999, Milton and Tsang 1993, Quirk and Greenbaum 1973), and “sentence adjuncts” (Sinclair 1990). Despite having multifarious names for calling these linking. 13.

(22) devices, we should bear in mind that their main function is to help writers develop their arguments in a logical and coherent way, and relate one proposition to another overtly. In other words, these linking devices per se “do not encode any kind of propositional meaning, but only work on the interpersonal and textual levels, signalling the connection of two segments of discourse, i.e. their status as connects” (Lenker 2010:46). In the following I will review some of the primary works which are used as the analytical framework by many other scholars. These works include Quirk et al. (1985), Biber et al. (1999), Celce-Murcia and Larsen-Freeman (1999), Huddleston and Pullum (2000), and finally Carter and McCarthy (2006).. 2.1.2.1 Quirk et al.’s (1985) conjuncts In Quirk et al.’s (1985) classification, conjuncts are discussed under the head of a larger category “adverbials,” along with adjuncts, subjuncts, and disjuncts. Conjuncts are used to connect two linguistic units, and the units may vary from “constituents of a phrase realizing a single clause element” (3) to sentences or paragraphs (2). For example: (2) The candidate is a fine teacher, a broadcaster of some experience, and a respected drama critic. In addition, she has written a successful novel. (3) The candidate has written successful, lengthy, popular, and in addition highly original novel. (Quirk et al. 1985:632) In (2) the inter-sentential conjunct in addition is used to conjoin two independent clauses, while in (3) in addition has a focusing role as well as the conjoining one. The inter-sentence relations, such as in addition in (2), thus reveal certain conjunctive meanings between different stretches of discourse. To account for the semantics of conjuncts, there are seven major types of conjunctive relations identified: Listing, summative, appositional, resultive, inferential, contrastive, and transitional type (see 14.

(23) Table 2.1). The function of conjuncts in the listing type is to enumerate or itemize what the writer is going to say. This type comprises “enumerative” (e.g. first, second, third, one, two, three, a, b, c, in the first place, in the second place, on the one hand, on the other hand) and “additive” subtype. The additive subtype is further divided into “equative” (e.g. correspondingly, equally, likewise, in the same way, by the same token) and “reinforcing” (e.g. further, more, moreover, in particular, what is more, in addition). What is worth noting here is that the enumerative conjuncts are not merely assigning numerical labels to items, they “connote relative priority and endow the list with an integral structure, having a beginning and an end” (Quirk et al. 1985:636). Summative and appositional (or appositive) conjuncts are similar in that “they precede an item which is to be looked at in relation to all the items that have gone before” (Quirk et al. 1985:637). The examples of summative conjuncts include altogether, overall, then, therefore, thus, in all, in conclusion, in sum, to conclude, to sum up, and to summarize. The appositive conjuncts mainly include members with “viz” meaning (e.g. namely, thus, in other words) and “e.g. or i.e.” meaning (e.g. for example, for instance, that is, that is to say, specifically). Resultive conjuncts show cause and effect relations among discourse: accordingly, consequently, hence, so, therefore, thus, as a consequence, in consequence, as a result, of course, and of course. The inferential type of conjuncts “indicate a conclusion based on logic and supposition” (Quirk et al. 1985: 638): else, otherwise, then, in other words, in that case. Contrastive conjuncts present a comparison or mark contrastive relations in discourse. The subtypes include “reformulatory” (e.g. better, rather, more accurately, more precisely, alias, alternatively, in other words), “replacive” (e.g. again, 15.

(24) alternatively, rather, better, worse, on the other hand), “antithetic” (e.g. contrariwise, conversely, on the contrary, in contrast, by comparison, on the one hand…on the other hand), and “concessive” (e.g. however, nevertheless, nonetheless, notwithstanding, still, though, yet, in any case, at any rate, in spite of that, after all, admittedly). The last type of conjuncts in Quirk et al’s classification is transitional conjuncts. Transitional conjuncts function more like a topic shifter, shifting reader’s attention to another topic (the “discoursal”, e.g. incidentally, by the way) or to a temporally related event (the “temporal”, e.g. meantime, meanwhile).. 16.

(25) Table 2.1 Semantic relations of conjuncts in Quirk et al. (1985:634-639) Type Function Examples Listing Give a particular structure or orientation to a list Enumerative first(ly), second(ly), in the first/second place, to begin with Additive Equative correspondingly, by the same token Reinforcing again, also, further, furthermore, in particular Summative Introduce an item that in conclusion, to sum up embraces the preceding one Appositional Express the content of the namely, in other words, for preceding item or items in example, that is other terms Resultive Can be tentatively accordingly, consequently, expressed by a coordinate therefore, hence, as a result clause with a vague reason Inferential Indicate a conclusion else, otherwise, then, in based on logic and that case supposition Contrastive Present either contrastive words or contrastive matter in relation to what has preceded Reformulatory more precisely, alias, alternatively, in other words, rather Replacive rather, better, on the other hand Antithetic contrariwise, on the contrary Concessive however, nevertheless, nonetheless, though, yet Transitional Serve to shift attention to another topic or to a temporally related event Discoursal by the way, incidentally Temporal meanwhile, eventually. 2.1.2.2 Biber et al.’s (1999) linking adverbials Although different terms are used, adverbials serving discourse connecting functions are also discussed in Biber et al. (1999:875-892). Biber et al. (1999:762) 17.

(26) distinguish three kinds of adverbials according to their function in the clause. The first kind, called “circumstance adverbials,” is to “add information about the action or state described in the clause, answering questions such as ‘How, When, Where, How much, To what extent?’ and ‘Why?’” (Biber et al. 1999:763). The second kind “convey[s] speakers’ comments on what they are saying (the content of the message) or how they are saying it (the style),” and it is called “stance adverbials” (Biber et al. 1999:764). The third kind of adverbial, viz. “linking adverbial,” plays a more “peripheral role with the rest of the clause than circumstance adverbials typically do,” and serves a discourse connecting function rather than “adding additional information to a clause” (Biber et al. 1999:765). The choice of discussing linking adverbials is based on two reasons. Firstly, according to Biber et al.’s (1999:880) corpus finding, linking adverbial has the highest frequency in academic prose among the four registers investigated (the other three include conversation, fiction, and news). Secondly, an important stylistic feature of academic writing is to present and organize writers’ ideas and arguments through appropriate choice of linking adverbials, and this “not only reflects communicative need but also the characteristic choice of this register to mark the links between ideas overtly, as these arguments are developed” (Biber et al. 1999:880). There are six general semantic categories of linking adverbials identified: Enumeration and addition, summation, apposition, result/inference, contrast/concession, and transition. Each category will be discussed respectively in the following. The enumerative function of linking adverbials makes the writer put pieces of information in a certain order, while the additive function makes the writer either equate or reinforce the weight of an item to a preceding one. Linking adverbials with enumerative function include ordinal numbers (e.g. first, second), adverbs (e.g. firstly, secondly, finally, lastly), and other structures (e.g. in the first/second place, for one 18.

(27) thing, for another thing, first of all, to begin with, next). Additive linking adverbials include examples such as also, by the same token, further, furthermore, likewise, and moreover. Linking adverbial with summative function shows the conclusion or summing-up of information in the preceding discourse. The examples include all in all, in conclusion, overall, to conclude, and to summarize. Together with enumerative and additive adverbials, summative adverbials can “help to structure the information in academic prose, giving readers clear signposts of where they are in the text” (Biber et al. 1999:881). Adverbials of apposition show the second unit of discourse is equivalent to or included in the preceding discourse, and the second unit of discourse is often a restatement, rephrasing, reformulation, or exemplification of the preceding discourse. Examples include which is to say, in other words, that is, i.e., for example, and for instance. The result/inference category of adverbials shows the second unit of discourse is the “result or consequence − either logical or practical − of the preceding discourse” (Biber et al. 1999:877). Typical resultive and inferential linking adverbials include therefore, consequently, thus, as a result, hence, in consequence, and so. Linking adverbials showing contrastive or concessive relationships “mark incompatibility between information in different discourse units” (Biber et al. 1999:878). Adverbials that have a focus on contrast between arguments include conversely, instead, on the contrary, in contrast, and by comparison; those with a focus on concession include anyhow, besides, nevertheless, still, in any case, at any rate, in spite of that, and after all (Biber et al. 1999:879). The last semantic category of linking adverbials identified is transitional. This marks “the insertion of an item that does not follow directly from the previous 19.

(28) discourse,” signaling the relationship among linked discourse is “loosely connected” or “unconnected” (Biber et al. 1999:879). Typical members in this category are by the by, incidentally, and by the way. From the above discussion, we can find that Quirk et al’s (1985:634) conjunctspecific relations overlap with Biber et al’s (1999:875) classification of linking adverbials to a large extent. Both of them include categories such as enumeration, addition, summation, apposition, result/inference, contrast/concession, and transition. Although some of their subcategories may vary from each other, most of the major categories are useful in the classification of connectives. For a better understanding of previous classifications of these discourse linking devices, logical connectors in Celce-Murcia and Larsen-Freeman (1999) will be briefly reviewed in the next section.. 2.1.2.3 Celce-Murcia and Larsen-Freeman’s (1999) logical connectors In Celce-Murcia and Larsen-Freeman (1999:519), logical connectors are defined as “types of cohesive devices, lexical expressions that may add little or no propositional content by themselves but that serve to specify the relationships among sentences in oral or written discourse, thereby leading the listener/reader to the feeling that the sentences ‘hang together’ or make sense.” Typical logical connectors include two kinds: subordinating conjunctions (“adverbial subordinators” in Celce-Murcia and Larsen-Freeman’s term), and conjunctive adverbials. Some adverbial subordinators are simple, usually a single word (e.g. after, although, as, because, before, until, whereas), and some are complex, more like “fixed lexicalized units” (e.g. as long as, as soon as, even if, given that, inasmuch as, in order that, provided that). Unlike adverbial subordinators, conjunctive adverbials are “complete adverbials onto themselves.” They connect two independent clauses rather than subordinate a clause (Celce-Murcia and Larsen-Freeman 1999:522). 20.

(29) Celce-Murcia and Larsen-Freeman further distinguish “truth-conditional connectors” from “inferential connectors” (Celce-Murcia and Larsen-Freeman 1999:528). Truth-conditional connectors (e.g. before, after) “contribute to propositional content and the truth-falsity of the sequences in which they occur.” For example, we cannot change before in Julius Caesar’s career was finished long before Napoleon rose to power in France with after, because the adverbial subordinator before contributes to the truth value of the sentence. Unlike truth-conditional connectors, inferential connectors “do not contribute to truth or falsity” but “clarify the logical relationship the speaker/writer intends” (Celce-Murcia and LarsenFreeman 1999:528). For example, the sentence Caesar was a strong Roman leader; ?thus, Napoleon powerfully ruled France sounds weird because the inferential connector thus is expected to infer certain causal relations between two sentences, and therefore the use of thus in the example is obviously inappropriate. In this thesis, we only take inferential connectors into consideration. As a grammar course book, there is not a clear classification of conjunctive adverbials but a description of usage of individual logical connectors in Celce-Murcia and Larsen-Freeman (1999). In the next section, we continue to discuss Huddleston and Pullum’s distinction of “pure” and “impure” connectives. A property of connective adjuncts in Huddleston and Pullum (2002), like inferential logical connectors in Celce-Murcia and Larsen-Freeman (1999), is that they “do not impose additional truth conditions on their clause” (Huddleston and Pullum 2002:776). That is, they contribute little or no truth value to the proposition of the sentence.. 2.1.2.4 Huddleston and Pullum’s (2002) connective adjuncts There are two properties of typical and central cases of connective adjuncts. First, they do not contribute additional truth-conditional meanings to the clause. 21.

(30) Second, they cannot fall within the scope of negation, interrogation, and focus (Huddleston and Pullum 2002:776). These properties, however, are not unique to connective adjuncts, and they have to be taken into account in conjunction with their crucial role of connecting one clause to its neighboring text. According to their semantic complexity, Huddleston and Pullum (2002:777) draw a distinction between “pure” and “impure” connectives. Pure connectives (e.g. moreover, also) “have no other function than that of connecting their clause to the surrounding text (or context),” while impure connectives, besides the connecting function, have more complex functions such as concession, condition or reason/result. Huddleston and Pullum (2002:778) identify four broad types of pure connectives. The first one is ordering. The ordering function of connective adjuncts is to signal the structure of a piece of discourse, including those that mark the first point (e.g. in the first place, for one thing, on the one hand), a subsequent point (e.g. in the second place, for another thing, on the other hand), and the final point in a series (e.g. last of all, in conclusion, finally). The second type of connective adjuncts is used for addition (e.g. besides, moreover, in addition) and comparison (e.g. likewise, on the contrary). Connective adverbs, such as neither and nor also belong to this category. Expressions which have primary meaning in space and time (e.g. in the same way, at the same time) denote a comparative connective meaning between two stretches of discourse. The third type is elaboration and exemplification. The members in this category are mainly to elaborate or exemplify the preceding discourse (e.g. for example, in other words, more precisely, that is). The fourth type is the markers of informational status. They include members which suggest a change of topic or a turning aside of current concern of discourse (e.g. by the way, incidentally, parenthetically). As for impure connectives, they are adjuncts combining the connective function with concessive function (e.g. nevertheless, nonetheless), 22.

(31) condition (e.g. in that case, otherwise), or reason/result (e.g. accordingly, as a result, in consequence) (Huddleston and Pullum 2002:779). The distinction of pure and impure connectives by Huddleston and Pullum is similar to the simple/complex distinction in Halliday’s (1994) classification of conjunction, which will be discussed in 2.2.2. In the next section, we will introduce the classification of linking adjuncts in Carter and McCarthy (2006).. 2.1.2.5 Carter and McCarthy’s (2006) linking adjuncts Linking adjuncts refer to single words or phrasal items that serve the function of explicitly indicating “the semantic relationship between two clauses or sentences or paragraphs” (Carter and McCarthy 2006:256). They are especially important in academic writing in that the use of linking adjuncts can “give coherence to the text and organize it” (Carter and McCarthy 2006:290). According to the semantic relations signaled by linking adjuncts, Carter and McCarthy classify them into nine categories: (1) additive, (2) resultative, (3) contrastive, (4) time, (5) concessive, (6) inference, (7) summative, (8) listing, and (9) meta-textual linking adjuncts. Additive linking adjuncts show that the second text segment is the addition, specification, or expansion of information in the first text segment (e.g. also, likewise, moreover, correspondingly, in particular). Resultative linking adjuncts present the second text segment as a result or outcome of the event stated in the first (e.g. so, therefore, consequently). Contrastive linking adjuncts denote a contrast between the information in the first and second text segment (e.g. rather, on the contrary). Time linking adjuncts present a relation of temporal sequence between the information in the first and second text segment (e.g. eventually, then, in the meantime). Concessive linking adjuncts indicate that a concession of the following proposition is made, and they often contrast the preceding proposition with the following proposition to some 23.

(32) extent (e.g. admittedly, anyway, though). Inference linking adjuncts suggest that the proposition in the second text segment is inferred or concluded based on the evidence of the preceding text (e.g. in that case). Summative linking adjuncts present a summing-up of the preceding text (e.g. overall, in short, all in all). Listing linking adjuncts make a list of what is going to be discussed in the following discourse (e.g. firstly, lastly, for one thing…for another thing). Meta-textual linking adjuncts indicate that the second text segment is a paraphrasing, explanation, or exemplification of the first text segment (e.g. in other words, that is to say, for example). Meta-textual linking adjuncts can also suggest a shift in topic or a temporary digression from the current topic (e.g. incidentally). The meta-textual linking adjunct in Carter and McCarthy (2006) is a somewhat broad category that encompasses both appositive and transition conjuncts in Quirk et al. (1985), apposition and transitional linking adverbials in Biber et al. (1999), and connective adjuncts of elaboration/exemplification and markers of informational status in Huddleston and Pullum (2002). This reflects the fact that different scholars have different classifications of the linguistic elements which serve connective functions in a text. Furthermore, different scholars may have different labels for these discourse connecting elements, all of the scholars, however, categorize connectives according to their semantic and discourse functions in a text. This is very important and suggests that the judgment of what category a connective belongs to should be based on the function it serves in a particular stretch of discourse, rather than giving it a pre-determined classification without considering its functions.. 2.1.3 Connectives in Chinese grammar Compared to the study of connectives in English, a comprehensive classification of Chinese connectives, especially connectives used in academic texts, remain scarce. 24.

(33) Liao (1986) gives a general description of connectives in Chinese discourse, which is further used as analytical basis by Jin and Liu (2010), Li (2011), and Liu (2001). Cheng and Cui (2004) offer a functional account of Chinese connectives, arguing that connectives do not always indicate or construct the logico-semantic relations between clauses, but facilitate reader’s discourse comprehension and contribute to the textual structure. Some scholars, on the other hand, compare logical connections (e.g. Huang 2007) or adverbial correlatives (e.g. Yuan and Lu 2011) between English and Chinese discourse. Some delve into a specific type of connectives, for example, connectives which are time-related (e.g. Jin and Liu 2010) and connectives used for summation (e.g. Luo 2008). Studies from Chinese metadiscourse (e.g. Duan 2009, Li 2011) are the ones that are closest to the study of connectives used in Chinese academic texts. Duan (2009:87-88) classifies Chinese textual metadiscourse markers into seven categories: (1) Logical markers, which indicate the semantic and structural relations among stretches of discourse, including additive ci wai 此外 ‘in addition’, adversative dan shi 但是 ‘but’, consecutives yinci 因此 ‘hence’, and conclusive zong shang suo shu 綜上所述 ‘to sum up’; (2) Sequencers show the order of specific items (e.g. diyi…dier 第一…第二 ‘first…second’, yi fangmian…ling yi fangmian 一方面…另一方面 ‘on the one hand…on the other hand’); (3) Endophoric markers, which guide readers back to aforementioned information (e.g. zhengru shangwen ti dao de 正如上文提到的 ‘as mentioned above’); (4) Topicalizers, which explicitly mark the topic of current discourse (e.g. zai…fangmian 在…方面 ‘in…respect’, zai…de qingkuang xia 在…的情況下 ‘under the circumstances’, guan yu 關於 ‘with reference to’, zhi yu 至於 ‘as to’); (5) Code glosses, which signal the restatement (e.g. huan ju hua shuo 換句話說 ‘in other words’) or exemplification (e.g. liru 例如 ‘for example’) of information in a previous proposition; (6) Illocutionary, which markers explicitly 25.

(34) indicate the speech act by the writer (e.g. wo xiwang 我希望 ‘I hope’, wo zhichi 我支持 ‘I support’); (7) Announcements, which forecast what the writer is going to argue in the following statements (e.g. xiawen jiang xiangxi lunshu 下文將詳細論述 ‘in the following detailed discussion’). Slightly different from Duan’s (2009) classification, Li (2011:100) categorizes Chinese textual metadiscourse into four types: (1) Topic markers, which indicate choice zhide zhuyi de shi 值得注意的是 ‘it is noteworthy that…’, beginning wo xiang shuo de shi 我想說的是 ‘what I want to say is…’, transition shuodao 說到 ‘speaking of…’, ending jiu zheyang 就這樣 ‘it’s all’, and the order of a series shouxian 首先 ‘first’; (2) Cohesive devices, which mark the connection among different stretches of discourse (e.g. tongshi 同時 ‘at the same time’, yi…wei li 以…為例 ‘take…as an example’, zong er yan zhi 總而言之 ‘in a word’, wulun 無論 ‘whether…or’, weici 為此 ‘to this end’, youci kejian 由此可見 ‘from this it can be seen that…’, fanzhi 反之 ‘on the contrary’, shangmian shuo dao 上面說到 ‘as mentioned’); (3) Evidential markers, which point out the source of certain proposition (e.g. juxi 據悉 ‘it is reported that…’, chengru suo yan 誠如…所言 ‘as someone says’); (4) Code glosses, which restate a certain proposition (e.g. ye jiushi shuo 也就是說 ‘that is’, huan ju hua shuo 換句話說 ‘in other words’). From these examples we can discover that connectives link units which are larger than clauses, and most of them (e.g. shouxian 首先 ‘first’, zong er yan zhi 總而言之 ‘in a word’, huan ju hua shuo 換句話說 ‘in other words’) can be followed by a pause in spoken language or a comma in written text. Next, some connectives (e.g. weici 為此 ‘to this end’, fanzhi 反之 ‘on the contrary’) include anaphoric element, which is one of the means to make a text coherent (Li 2011:138). Thirdly, a connective may be put into different categories in accordance with their different functions in discourse. For example, zuihou 最後 ‘finally’ can either be a sequencer 26.

(35) when it denotes the order of a series of arguments or be a conclusive marker when summarizing aforementioned arguments. This multifunctionality of connectives is the origin of different categorization in different context (Li 2011:138, Liao 1986:63). Finally, Liu’s (2001) discussion of Chinese cohesive conjunctives in five written genres (narrative, argumentation, expressive, news report, and advertisement) is so far the only work that tackles the use of connectives in Chinese argumentative texts. Based on Halliday’s (1994) classification scheme and Liao’s (1986) classification of Chinese connectives, Liu finds that all five written genres have the same preference toward the use of cohesive conjunctives, i.e. enhancement > extension > elaboration, and he concludes that the more argumentative elements a text has, the more cohesive conjunctives a text requires. Although Liu (2001) adopts both Halliday’s (1994) and Liao’s (1986) classification, neither Halliday (1994) nor Liao (1986) centers their study on connectives in academic texts. Therefore, this thesis attempts to uncover how connectives are used in Chinese academic texts. In the next section, we are going to review Halliday’s discussion of conjunction, which is then served as the analytical framework in this thesis.. 2.2 Connectives and conjunction in discourse Halliday’s (1994) framework of conjunction is suitable for the analysis of connectives in this study in that our central concern is the functional-semantic interpretation of clause complex relation at inter-clausal level. The interpretation includes two systemic dimensions. One is the system of interdependency, which consists of parataxis and hypotaxis. The other is the logico-semantic relation among clause complexes, including expansion and projection (Halliday 1994:216). As for the two types of interdependency, paratactic and hypotactic structures are both considered in our analysis, and they are also two types of clause linking discussed in Liu et al. 27.

(36) (1983) as lianhefuju 聯合複句 (paratactic) and pianzhengfuju 偏正複句 (hypotactic). Regarding the logico-semantic relation, expansion is included in our study because it pertains to the three ways (elaboration, extension, and enhancement) by which connectives expand a clause into a clause complex. Projection is excluded because it concerns the role of the secondary clause as a locution or an idea, which is irrelevant to the present study.. 2.2.1 Logical semantic relations Expansion means the secondary clause expands the primary clause through elaborating, extending, or enhancing (Halliday 1994:219). In the elaboration kind, “one clause elaborates on the meaning of another by further specifying or describing it” (Halliday 1994:225). The secondary clause does not add new information to the primary clause; instead, the secondary clause restates, specifies, exemplifies, or clarifies what has been stated in the primary clause. There are three subtypes of elaboration: exposition, exemplification, and clarification. The first two are together under the heading “apposition” (Halliday 1994:226). Exposition is an “i.e.” relationship in which the secondary clause “restates the thesis of the primary clause in different words, to present it from another point of view, or perhaps just to reinforce the message” (Halliday 1994:226). Exemplification is an “e.g.” relationship in which the secondary clause makes the primary clause more specific and explicit by citing an example. Clarification is a “viz.” relationship in which the secondary clause makes the primary clause more precise or clarified by “backing it up with some form of explanation or explanatory comment” (Halliday 1994:226). In the extension kind, “one clause extends the meaning of another by adding something new to it” by means of addition or variation (Halliday 1994:230). Addition means the secondary clause is simply adjoined to the primary clause and there are no 28.