Metadiscourse in Chinese and American graduate dissertation introductions

Abstract The current study aims to look at differences in the use of metadiscourse in the introductory sections of English and Chinese university students. The corpus of the study comprises 20 “introductions” of university students’ MA theses. This study hinges on Hyland’s model of Metadiscourse (interactive and interactional resources) for its methodological framework and aims at analyzing academic writing of native academic and non-native writers. The findings revealed a statistically significant difference in the use of metadiscourse in theses written by Chinese native speakers (CNS) and English native speakers (NS) university students. The study suggests that metadiscourse needs to become a fundamental component for effective reading and writing instruction for students.


PUBLIC INTEREST STATEMENT
The study hinges on Hyland's (2005) model of Metadiscourse with an ultimate objective of investigating academic writing of native academic and non-native writers.
The entire study is exclusively focused to examine employment of Metadiscourse in English and Chinese university students' "introduction sections" within their dissertations.It is important to mention that the study witnessed a significant statistical difference in the employment of metadiscourse in the dissertations produced by non-native speakers and native speakers among university students.The study revealed that metadiscourse needs to become a fundamental approach for effective reading and writing instruction for students.Insights of Hyland model, once being tactfully incorporated into teaching of academic writing to Chinese students, can be academically more rewarding.

Introduction
Coherence is an ultimate outcome of metadiscourse which can be witnessed in authors' portrayal of ideas and concentrated involvement in the text. Metadiscourse refers to the linguistic mechanism used to communicate meanings with imagined readers, and to express a viewpoint as members of a particular academic community. Metadiscourse markers direct the reader through the text and provide a method for the writer to interact with or influence the reader, (Mohan & Lo, 1985). Hyland (2005) argues that writers project themselves in creating a text. He further states that writers use social and functional aspects of the language in their discourse. Adding he relates, "the term metadiscourse was coined by Zellig Harris in 1959 to offer a way of understanding language in use, representing a writer's or speaker's attempts to guide a receiver's perception of a text" (p. 3). While describing the concept of metadiscourse, Hyland (2005) forms significant categories namely, transition markers, frame markers, endophoric markers, boosters, attitude markers, self-mention and engagement markers. Authors have been observed to make frequent use of all these linguistic categories in their text to make the text more coherent and logical. Sometimes, writers employ a refined academic tone to present their stance effectively. Hammill and Larsen (1996) proposed three key features of master writings, a) writers' ability to use certain conventions of writing e.g., spelling, punctuation, and capitalization; b) linguistic ability focusing on morphological and semantic relations; c) cognitive skills to present preferences in an academic way.
According to Graham and Perin (2007), many adolescents have severe writing difficulties even if their reading is at "benchmark." Among the interactive resources, transitions, such as conjunctions, can be used to distinguish "additive, contrastive, and consequential steps in the discourse" (Afzaal & Xiangyi 2020;Hyland & Tse, 2004). Frame markers, meanwhile, refer to the boundaries of the text and convey information about its structure. Later endophoric markers make reference of another portion of the text. Furthermore, evidentials specify that proceeding or preceding information comes from an outside source. Lastly, code glosses inform the reader that ideational material will be mentioned again, albeit in a different manner. Interactional resources include hedges, which convey the writer's lack of willingness to commit to a proposition while boosters mark the writer's certainty within the given proposition. In addition, attitude markers assert the writer's estimation of a proposition, which may be surprise, obligation, or agreement. Engagement markers address the reader, "selectively focusing their attention or by including them as participants in the text through second person pronouns, imperatives, question forms and asides (Hyland, 2005). Lastly, self-mentions make known the writer's presence in a text through employment of first-person pronouns, such as I and we, or possessives like my, mine, and ours.
This study investigates metadiscourse used in academic writing of the introduction portion of Masters dissertations written in English language by both Chinese and American university students. English is by far the dominant language in academic research in writing. The ability to write not only coherently but also cohesively in English has emerged to be a necessity in academia today. This is never truer in a country such as China, where the educational climate tends to be more competitive in a comparison to native academic scholars. Students use many linguistic markers which enhance the voice of writers within their work e.g., native writers follow conventions of academic writing instinctively for instance, incorporation of clarity, discourse markers, hedges and other cohesive transitions. (Mohan & Lo, 1985). Whereas non-native students practice their reasoning abilities to draft their dissertation. The study focused on the introductory sections of theses because these sections entail the gist of the entire thesis, and students pay much attention to the introductory chapters. Also, it is the first section that the reviewer examines. Students employ all their faculties to make their first impression. An excellently structured introductory chapter entices the reader to read the remainder of the thesis where he can access all the necessary requisites including background of entire problem, objectives of the research endeavour, and the significance of the study.

Interactional metadiscourse in L2 writing theses
Many studies have been conducted on L2 writing with a particular focus on interactional metadiscourse to identify how L1 writing serves as benchmark data, (Qin & Karabacak, 2010;. In addition, the differences between L1 and L2 writing as well as between the two groups encompass the essence of the previous studies about metadiscourse, while the later arouse the attention in pedagogy. As for adult L2 learners, their most familiar type is argumentative writing as it is popular in teaching and testing (Qin & Karabacak, 2010;. In this very context, most of the L2 writers are well acquainted with the process and logic of argumentative writing though the establishment of arguments is often difficult. Compared to L1 learners, Chinese students are less likely to use self-mentions when writing English texts, (Bloch, 2009;Hyland, 2002b). Yoon (2021) identified how learners of English as a foreign language use interactional metadiscourse markers in argumentative essays. Yoon (2021) further indicated that a large-scale quantitative analysis of interactional metadiscourse by focusing on the topics of writings and the background of L1 and the proficiency of L2 is essentially required.
Bal-Gezegin (2016) argues that book review article is viewed as an important academic writing discourse which has many metadiscourse devices which may be employed to express ideas, positions, and arguments. Positive and negative evaluations of the reviewed book are expressed by the reviewers (Bal-Gezegin 2016), implying that book review requires more critical evaluation of the materials and cognitive judgments than research articles (Hyland, 2004), which can be realized by the use of metadiscourses.
Hong and Cao (2014a) examined the use of interactional metadiscourse features in EFL (English as a foreign language) learners' writing texts written by three different L1s (i.e., Chinese, Polish, and Spanish). The authors found that there is no significant difference in the number of hedges among the learner groups, but Chinese learners' essays had more boosters than the other two groups. This affirmed previous findings that Chinese writers are more likely to make assertive claims with reduced hedges and more boosters (Hyland & Milton, 1997). Cheung and Low (2019) examines the interactional metadiscourse patterns across proficiency levels by comparing features of metadiscourses between high quality essays and low-quality essays (e.g., Cheung & Low, 2019;Ho & Li, 2018;Intaraprawat & Steffensen, 1995;. Williams (2017) takes Chinese, Japanese, and Korean speakers as target audience because of the ethnic and linguistic features of the East Asian English learner population (Williams, 2017). Based on the belief that L2 learners are more likely to underuse hedging markers in writings when compared to L1 learners, and L2 learners tend to use more stronger assertions, this study takes the reduced use of hedges as a common feature of all L2 learner groups and indicates that hedging should be a target trait in L2 writing instruction. However, focused instruction can improve the ability to use interactional metadiscourse markers (Abbuhl, 2006Aull & Lancaster, 2014, implying that the effectiveness of instruction for L2 learners' use of interactional metadiscourse extensively and appropriately should be explored.
The study identifies the need of the investigation of interactional metadisocurse features employed in Chinese theses writing in a comparison to native academic scholars' theses. Therefore, this study limits the focus to the introduction sections of theses of native Englishspeaking students and non-native Chinese academic scholars' theses.
Historically, till the end of the Cultural Revolution, English teaching was not up to the mark in China, and a little attention was paid to English language Teaching (ELT). In 1978, English in primary and secondary education and English for non-English majors received more attention.in China, (Lan and Lam, 2020, pp. 12). highlighted that Guangzhou hosted the first international conference on English Language Teaching (ELT) in 1985. Throughout the 1980s, the State's policy directives were met with support in terms of syllabus design and materials creation from ELT specialists in China and some experts from outside. Currently, after a long span of 40-years development, from elementary education to advanced education, English teaching in China has experienced exponential progress, which helps countless Chinese students improve their language skills. English. Compared to 1980s, when students' lack of basic knowledge of English owing to their weak English learning experience at school, contemporary students acquire basic English skills in their secondary education, which requires more flexible and purposeful education in English during their advanced education. Argues that the opportunities for learning English have increased in terms of motivation (at both the national and individual levels) and ELT learning has gained much improvement in China, (p. 12).
In addition, a number of studies have explored English teaching in China at master level. These studies mainly focused on analyzing current teaching situations, especially the academic problems that are required to be dealt with on priority; suggesting different teaching methods to confront those challenges. According to Zhao, Xiao & Han (2017), motivations of learning English varied among graduated students in different grades, including further study abroad, deeply understanding of the world, decent jobs after graduation and requirement of course credits. Besides, compared to English academic reading and writing, more interests were shown to speaking and listening among non-English major graduated students (Zhang & Zhan, 2020).
Admittedly, graduate students that major in different disciplines have a diverse range of demands towards English learning owing to their specific discipline features. Therefore, scholars have also proposed different teaching models of English teaching for students of different majors, (Kanglong & Afzaal, 2020;Wu & Zhang, 2017;Liu & Ma 2018). It goes without saying that diversified and personalized teaching system as well as the compilation of up-to-date textbooks should be intensified to further improve language teaching for contemporary graduate students. Yoon (2021) pointed out that Hyland's (2005) model has attracted great attentions among scholars in various fields, but most of these researches are empirical ones focusing on hedges and boosters and the occurrences of metadiscourse markers.
Previous studies showed that metadiscourse mainly focused on a range of academic genres, including research articles, postgraduate theses, and undergraduate essays (Aull & Lancaster, 2014;Basturkmen & von Randow, 2014;Hyland & Tse, 2004). In China, quite a few of papers concerning this topic are dominated by studies of Chinese students' research articles, including their introductions and abstracts (Hong & Cao, 2014b). In most studies, authors investigate how MD is used by students and then further explore the underlying factors that may lead to the phenomena. Ho and Li (2018) used a survey to assess the way Chinese freshmen construct persuasive arguments in academic writing by investigating their use of MD in argumentative essays. It was found that only a few MD makers were differently used between low-rated essays and high-rated ones, while all the students had problems dealing with constructing convincing persuasive arguments. In this connection, a valuable suggestion solely based on pedagogical reasons stating that metadiscourse should be taught directly and explicitly at high school and at the commencement of university education was put forth in the academic settings which ultimately proved to be very effective, (Xu & Pengyi, 1990a) What's more, researchers try to compare differences of the use of metadiscourse between Chinese students and native English speakers (Hyland & Milton, 1997;Hong & Cao, 2014b;. To determine the differences of the use of metadiscourse, devices in English argumentative essay writing between Chinese college students and American college students, (Hyland and Milton,)compared 15 English argumentative essays written by Chinese college students with essays written American college students. The result revealed that metadiscourse devices were used frequently in all the writings, but the use of textual and interpersonal metadiscourse showed significant differences between these two groups, which could be insightful to improve English academic writing class in China.
Besides, some significant research studies have made a great contribution to enlighten English language teaching in China, especially instructions to English academic writing classes. Mu et al. (2015) pointed out that it was particularly important for teachers who work with Chinese EFL learners to teach novice writers explicitly how to use metadiscourse in their academic writing. (Mu et al., 2015) And by comparing the use of cohesive devices in argumentative writing by Chinese EFL learners at different proficiency levels, Yang and Sun (2012) suggested that problems, like correct use of conjunctions and flexible manipulation of lexical elaboration, are a persistent challenge to EFL learners across different proficiencies, which needs more attention from teachers to help students overcome these academic deficiencies. However, among those studies concerning the use of metadiscourse by Chinese students, far too little attention has been paid to the use of metadiscourse among Chinese master students (Hyland & Tse, 2004).
This study primarily has two major objectives. The first is to explore the use of metadiscourse in the introductory sections of MA theses written by Chinese and American university students. The second objective is to identify the similarities and differences in Chinese and American students' employment of metadiscourse in the academic writing of graduate theses.

Data and method
The study employs both qualitative and quantitative parameters since the data comprises a corpus and the research study exclusively aims at investigating the significant instances of metadisourse. The work entails a comprehensive inquiry of the metadiscourse practices in Chinese and American university students' theses in the fields of English and Social Science. Chinese students have an advantage of writing dissertations either in English or Mandarin. The researchers selected theses from English and social science disciplines to increase the dataset size for better results.
The theses were selected after a careful examination. Representative of Georgia University was approached for an authentically verified data; titles are given in the appendix. These theses were converted into plain text for the analysis. However, only introductory sections were analyzed. Inclusion of theses from social sciences in the major dataset is well justified as all students from social sciences are bound to study academic writing/thesis writing courses by EFL teachers before starting dissertation writing. This study follows Hyland (2004) metadiscourse model which directly relates to analyzing the academic writing of native academic and non-native writers. The theses were further selected from renowned universities of China, and USA, written in English language by native and non-native speakers. Chinese theses are written by non-native speakers whereas theses taken from USA, are written by native American students. Hyland's (2005) description of metadiscourse served as the reference point for the analysis of the current study. The subject model categorizes metadiscourse into two kinds of resources, called interactive and interactional. Interactive resources help to direct the reader during the text. Some types of interactive resources are: transitions, frame markers, endophoric markers, evidentials and code glosses. As the title suggests, interactional resources allow the reader to interact with the text and the argument at hand. The interactional resources can be listed as the following: hedges, boosters, attitude markers, engagement markers and self-mentions. Given below, Table 1 displays Hyland's (2004) model of metadiscourse. Table 1 shows summary of Hyland's metadiscourse features. The model is utilized to analyze metadiscourse characteristics. It comprises two primary categories, each of which has five subcategories. Interactive and interactional are the primary types. The writer will strive to construct the content so that it can fulfill the demands of readers in the interactive classification (Hyland, 2005). The interactive section consists of five subcategories, including transitions (sectional links devices), frame markers (section and act reference devices), endophoric markers (sectional reference devices), evidence (scope devices to refer information) and code glosses (devices to elaborate propositional meaning).
The interactional category of metadiscourse characteristics, on the other hand, is employed to make the writer's message clearer by allowing the receiver to respond to the unfolding text. In other words, this category comprises elements that engage readers and allow them to participate to the debate by alerting them to the author's point of view. This category is divided into four subcategories: hedges (devices used to demonstrate uncertainty), boosters (devices used to demonstrate certainty), Attitude markers (devices used to demonstrate the writer's affective attitude), Self-mentions (devices used to demonstrate the writer's presence in the text), and Engagement markers (devices used to indicate the interaction of the readership).

Corpus construction and analysis
The theses were downloaded in the PDF format and converted into plain text with ant file converter (Window 64-bit 1.2.1) version software. The official websites of Georgia University USA and Shanghai Jiao Tong University China were accessed where the subject data was available in open access. The corpus comprises 20 theses in total in the areas of English and social sciences. Later, introduction sections of all the 20 theses were concentrated upon for an exclusive metadiscoursal inquiry seeking assistance from all the requisite tools of inquiry. Plain text files were and cleaned of headers and were coded manually to identify the specific metadiscourse features. Further, the tokens of metadiscourse characteristics were identified in the English corpus, their frequency (types) in the AntConc was determined using concordance lines. The corpus was manually read, and there were instances (tokens) of interactive metadiscourse characteristics and interactional ones (independent of their frequency).
The metadiscourse traits were then classified using Hyland's classification in order to generate a quantitative database of metadiscourse features in the English corpus. Following that, the corpus was thoroughly reviewed in order to qualitatively assess the texts and discover any indications of alterations in translation. The rationale behind employing Hyland's method is solely based on strong footings as the subject method helps in the entire comparison process to identify the specific function that each metadiscourse entails in the designed sub-corpora. We employed AntConc 3.5.7 (Windows, 2018) software and concordance tool in order to identify the total number of words in the selected articles and to further search each metadiscourse device that was used by these writers. Later, employing a very meticulous approach, the items were also checked manually to review if they really served a specific function as a metadiscourse marker. We followed the normalizing word count principle only after getting all the metadiscourse categories identified.

Research questions
While concentrating on the objectives, the study raises the following questions solely based on researchable dimensions: (1) What different types of interpersonal MD are used in Chinese and American students' MA theses?
(2) What observable differences or similarities between American English students and Chinese English students' employment of MD can be traced employing all the comparative procedures?
(3) What are the significant variations in terms of employment of two types of metadiscourse used by American and Chinese thesis writers?

Results and discussions
The study aimed at investigating the significant instances of correlation witnessed in written theses of Chinese students writing in English and native writers studying in US. Chinese graduate students seem to pay much time and attention in writing in English as compared to the native writers while writing theses.
Quantitative analysis was carried out to investigate 1) MD frequency distribution in the text, and 2) MD density in the text (the amount of MD per sentence). Using metadiscourse density in the analysis allows samples of different sizes to be compared and refers to the amount of MD per sentence. The distribution of MD throughout the text is shown in Table 2, while the results of MD density can be found in Tables 3 and Tables 4.   According to Table 2, native speakers used interactional resources (58% of all MD use) more frequently while non-natives more often utilized interactive ones (64% of all MD use). This result may be slightly unexpected, as it seems plausible that non-native speakers' writing tendencies appear to mirror that of native speakers due to the influence of English as an international language and English as the most frequently used language in academic writing (; Mauranen, 2010).Besides the difference in frequency of use of interactive and interactive resources, the dissimilarities in MD usage continue. Firstly, among interactive resources, the use of transitional markers and evidential varied greatly. NS most frequently used transitional markers (33% of all interactive resources) while Chinese native speakers (CNS) transitional markers account for an extremely small proportion of the occurrences (4% of interactive resources). CNS, on the other hand, most regularly used evidentials (52% of interactive resources) while this number accounts for much less (15% of NS interactive use, see fig 1 & fig 2). Apart from these two noticeable differences, results show similar usage of interactive resources among NS and CNS in regards to frame markers (23% of NS use and 18% of CNS use), endophoric use (15% for NS and 15% for CNS), and code glosses (14% for NS and 10% for CNS). Moving onto interactional resources, considerable disparities can be found between each group's use of selfmentions and hedges. Self-mentions comprise more than 50% of MD usage among NS, however, this number is less than half for CNS. Conversely, CNS speakers were witnessed to be frequently employing hedges in their writing (44% of all interactional resources) whereas this resource encompassed a much small proportion of interactional use by NS (15%). In addition to these two categories, there was no significant variation amongst the other three uses of MD, namely, boosters (10% for NS and 9% for CNS), attitude markers (8% for NS and 8% for CNS), and engagement markers (16% for NS and 12% for CNS).
As mentioned previously, transitional markers accounted for most NS interactive MD use and evidentials comprised the majority of CNS use (33% and 52%, respectively).

Significant instances of inquiry extracted from within theses
The following example shows the qualitative analysis of examples extracted from theses. Example extracted from NS corpus are symbolized as E, and C shows the examples taken from CNS corpus.
Examples: E1: Consequently, while grotesque realism might frequently involve images of destruction and decay, it also often incorporates gestures toward regeneration and renewal. C1: Despite the experimental evidences, scholars (Vande Kopple, 1997) have also asserted that even though the learner output is acceptable, it still shows disparity with the output of native speakers.
Providing Clarification E2: Furthermore, Lamb also recognizes the roles that Garden Hills' remarkable bodies play in the town's socioeconomic regeneration.
C2: Furthermore, the target language output performance of L2 learners largely conforms to the expressing habits of L1 but shows disparity with those of the native speakers.    According to Taymaz (2021) transitional markers such as "also" and "furthermore" aid writers in conveying a connection between clauses, they state, "It is this link that allows writers to explicitly define the key concept and provide a clarification". Several examples from the collected corpora can be seen below.

Centrality claim
E3: This record is a textual construction of time, because these narratives are consciously created by authors entering the textual discourse with Virgil.
C3: The reason for such overlook is partly because of the lack of mature theoretical support for approaching the problem.

Purpose of the study
E4: Dido Queen of Carthage is often considered part of Marlowe's juvenilia, and therefore unimportant to study, but I disagree.
C4: Thus, in the present study written production data will be examined and whether there is an asymmetry between comprehension data and production data can also be investigated.
It is important to mention that the above-mentioned extract entails faulty syntactic structure.

The focus of the study
E5: I focus on their reworking of Virgil because a gap exists in Dantist and Marlovian criticism when it comes to temporal readings of the texts. Because their retellings of the Dido and Aeneas story include alterations of Virgil's space and time, C5: Thirdly, this study focuses on the genre of drama.
Moving onto interactional resources, this study found that self-mentions comprised the majority of NS metadiscourse use and was less than half this number for CNS (51% and 26%, respectively). This data stands in opposition to what was found by Kim and Lim (22013, p. 22) in which they stated, "Selfmentions show very low frequencies across the two sets of introductions". The authors attribute this to the notion that writers prefer to appeal to readers in an impersonal way using the inanimate subject construction. This may reflect that in this genre (academic writing), writers prefer to appeal to readers in an impersonal way using the inanimate subject construction" (Kim and Lim, 2013). A contributing factor may be found in the genre of the corpus selected. The NS thesis introductions mostly consist of English literature major whereas the CNS introductions are mainly from the field of applied linguistics. This finding is consistent with Hyland and Tse (2004) who assert, "Selfmention also plays a far more visible role in the soft disciplines, (p. 22)". Students are often exhorted by style guides and supervisors to students to present their own "voice" and display a personal perspective" and contrast this with writing in the "hard" fields by stating, "the community tends to value competence in research practices rather than those who conduct them, and so a personal voice is subsumed by community knowledge and routines" (Hyland & Tse, 2004, p. 42). Examples of self-mentions include I, we, my, and our. E6: In this paper, I argue that the South provides an ideal location to break down this binary because of the inherent hybridity of monsters.
E7: Through these authors' manipulation of temporality, we learn not only about literature, but also about history as literature.
C6: Besides such salient unidiomatic use of target language, there is another kind of nonnativelikeness, which we term covert nonnative-likeness.
E8: My approach is similar to that of a new breed of medievalists E9: We are readers of these texts not in the time they were written, but in our own time; as such, we should expand our analysis of these texts with our own critical vocabularies to produce new critical readings.
C7: In fact, the increasing research findings call our attention to explore deeper in the aspect of integrating metaphor theory with second language teaching.
Changing the focus to the CNS, evidentials seemed to play the largest role in the interactive category (52% of all CNS interactive resources). Several examples are listed here: E10: As Carolyn Dinshaw elaborates on this type of temporality, "[t]ime is lived; it is full of attachments and desires, histories and futures." C8: Additionally, according to Heltai, translation studies is "an interdisciplinary field" E11: Gary L. Long and Larry W. DeBord note that C9: Just as Lado (21957) stated in Linguistics Across Cultures E12: Accordingly, to Lamb, the town is marked by an "inferno-like atmosphere" C10: Some other scholars like Laviosa (21998) also stress the significance of corpus-based translation studies.
In continuation, hedges make up the majority of interactional resources used by CNS (44%, respectively). Hyland (2004) remarks that this type of interactional resource is used to "withhold writer's full commitment to proposition" resulting in "a relationship with readers to persuade them of interpretations" being founded (Hyland, 2004: 137). Hedges may include might, may, perhaps, possible, and seem. Kim and Lim (2013) suggest this is because, "writers cautiously anticipate readers' reactions when indicating a gap or past research" and writers, "appear to be aware of possible counter-argumentation from the readers, while seeking acceptance for their claims". Examples of Hedges from NS and CNS can be seen below: E13: Consequently, while grotesque realism might frequently involve images of destruction and decay, it also often incorporates gestures toward regeneration and renewal.
E14: This view is perhaps best epitomized by her reading of the novel's closing sequence, C11: most of the previous studies have examined the comprehension data, which may be very different from the production data.
E15: this thesis is an attempt to answer the question posed to Valentino, and a possible solution to the answers that arise when we begin identifying the ways that veterans encounter obstacles in their desire to enter into a culture that seems to close them out.
C12: A possible explanation for such difference is L1 influences E16: the novel seems to chiefly foreground unusual human bodies.
C13: it seems it is too meta-cognitively demanding for the learners to pop up with verb form of "head" in the sentence.
In general, the data presented in this study shows differences in the use of MD by NS and CNS. Finally, regarding Tables 3 and Tables 4 on the density of metadiscourse, NS were found to have a higher average of MD than CNS (1.69 for NS to 0.72 for CNS) to help lead the reader through the text (interactive) and permit them to interact with the author and the subject matter being deliberated (interactional).
Through the analysis of theses and the investigation of metadiscourse, it can be claimed that metadiscourse should become a key component of good writing education for students. The concept of metadiscourse highlights the idea that communication is more than just the sharing of facts and information. It also incorporates the writer's and the reader's personalities, attitudes, and preconceptions. It entails the writer deciding what kind of impression he or she intends to produce on their audience (Hyland, 2005). Therefore, the study highlights that metadiscourse is an important component of coherence that assists students in making their writing successful.

Conclusion
This study analyses 20 introductions of graduate students' theses of Chinese and the US universities to compare relationships of metadiscoursal elements. The analysis of the study found a significant difference in the use of meta-discourse elements proposed by Hyland (2005). The study while examining NS use of markers and elements of metadiscoursal elements suggests a statistical difference in their theses whereas CNS Mandarin speaking students were found deficient in their expressions due to their native language finding it extremely difficult to master metadiscoursal skills in academic writing. By focusing on the differences that are found in the study, the study opens a new vista of inquiry for the future researchers who may further explore many socio-cultural factors affecting the academic writing of CNS students. The study also highlighted socio-cultural factors reflecting the differences of metadiscourse in NS and CNS students.
The differences in quantity of metadiscourse found within the two sets of data collected in this study could have several different explanations. Firstly, it could be due to the difference in length between the two data sets. The non-native speakers' thesis introductions were on average less than half the length of the native speakers. As a result, the demand for metadiscourse was less. Another likely scenario is the influence of socio-cultural factors. Crismore et al. (1993) state that metadiscourse usage is closely connected to socio-cultural contexts. Furthermore, Lee & Deakin (22016b) attributes the lower density of Metadiscourse, specifically in Chinese, as a result of Chinese requiring the reader take on a larger role in deciding what relationship specific sections of a text have within the text in its entirety. That is to say, in Chinese more requirements are placed on a reader to connect the dots and create cohesion in the text. Therefore, being that the CNS in this data are native Chinese-speakers it is plausible that this idea of reader-responsibility in Chinese spills over into the writing of native Chinese speakers when they write academic texts in English.
Nevertheless, upon further examination of the present study's analysis on metadiscourse, it is discovered that there is a wider gap between the usage of interactional resources between CNS and NS (36% of MD versus 58%), whereas among interactive ones CNS seemed more likely to use them than NS (64% of MD versus 42%). In short, NS were more likely to employ interactional resources while CNS appeared to utilize interactives with more regularity. An interesting finding relates to the notion of reader-responsibility found in Chinese prose in which the writer tends to manage the level of a text's personality in order to control the distance between the reader, writer, and text itself. Keeping this very dimension into perspective, it might be expected to find lower levels of engagement and attitude markers within the CNS introductions.
However, a more in-depth examination of the analysis of the data shows a large difference in the type of metadiscourse used by NS and CNS in each of its forms. Namely, the CNS used interactional resources much less frequently than NS (36 % to 58%). This is drastically different than each group's use of interactive forms in which CNS made greater use than NS (66% to 42%). In addition, as mentioned above, both engagement markers and attitude markers were scarcely used by the CNS speakers. The lack of usage in these two categories could explain why the CNS introductions tend to require more work on the part of the reader to interpret the text.
Finally, metadiscourse markers are words or portions of phrases connecting the writer with the reader. The study provides the differences in use of Metadiscourse markers and highlights how these markers assist the reader in organizing the information, interpreting it and evaluating it in a text. Because of this very factor, it seems that the CNS thesis introductions would include fewer interactive resources. On the contrary, however, the results above show quite the opposite. The density of interactive resources is not only higher among type resources employed by CNS, but it is also higher than the NS. These results show strong evidence that Chinese writers do in fact attempt to make their texts more cohesive in order to explicitly establish their position on certain matters. From this, the possibility arises that the notion of placing more responsibility on the reader has had unequal influence on the usage of metadiscourse in academic texts. * cover image-you are able to create a cover page for your article by supplying an image for this purpose, or nominating a figure from your article. If you supply a new image, please obtain relevant permissions to reproduce the image if you do not own the copyright