THE USE OF METADISCOURSE MARKERS IN ESSAYS WRITTEN BY ESP UNIVERSITY STUDENTS

The study deals with metadiscourse markers employed in the essays written by ESP university students of mechanical engineering. The aim was to investigate the frequency of use of metadiscourse markers and to determine their types according to Hyland’s (2005) taxonomy. The corpus consisted of 100 essays dealing with two topics related to mechanical engineering. The research findings indicate that students used interactional metadiscourse categories to a significantly higher extent than interactive metadiscourse categories. The most used markers overall were transitions, followed by engagement markers and hedges. This suggests that students tried to capture the attention of their readers and make them participate in the discussion. Still, some markers were misused or overused, which implies that students need more instruction in the adequate use of markers. It can be concluded that adding explicit instruction in the use of metadiscourse markers to the ESP university curriculum would increase the level of students’ pragmatic competence so as to help them avoid pragmatic failures in writing.


INTRODUCTION
Metadiscourse markers, as devices that primarily contribute to the cohesion and coherence of a text, have been a very popular topic of research in recent years. This is due to the fact that they serve to establish a relationship with the reader and convey the writer's message better, which is especially important when it comes to academic writing. As such, they also play a vital role in the development of pragmatic competence of foreign language learners since metadiscourse "increases the readability of an essay and makes it more likely that the message will be understood" (Intaraprawat-Steffensen, 1995: 254).
The current study deals with the metadiscourse employed in the genre of undergraduate student essays. The aim is to investigate the use of metadiscourse markers in the essays written by first-year students of mechanical engineering within English for Specific Purposes (ESP) courses at the University of Belgrade. More precisely, we seek to identify the types of metadiscourse markers used according to Hyland's interpersonal model of metadiscourse (Hyland, 2005) and obtain an insight into the frequency of their use. The study can be viewed as diagnostic research which serves to assess the students' current skills in using metadiscourse features in writing and identify points where they might need help and clarifications in the context of ESP.
The paper is organised in the following way. In the next section, we reflect on the concept of metadiscourse and its importance in academic writing, as well as on previous studies dealing with the use of metadiscourse markers in English learners' essays. The third section contains a description of the corpus and methods used in the study. The fourth section provides the results of the conducted quantitative and qualitative analyses. The final part discusses the findings and provides pedagogical implications relevant to increasing students' pragmatic competence in ESP.

The concept of metadiscourse
The term 'metadiscourseʼ goes back to 1959 when it was coined by Zellig Harris (Harris, 1959). It soon became appealing to different language scholars, who started inventing their own models and definitions related to metadiscourse. Other terms were created in time, showing similarities with the term 'metadiscourseʼ (metacommunication (Rossiter, 1974), signalling words (Meyer, 1975) or meta-talk (Schiffrin, 1980)).
Almost all scholars agree that metadiscourse is an important segment of both writing and the writer's pragmatic competence; an author who is able to adequately use metadiscourse markers is considered a successful writer. The importance of the appropriate use of metadiscourse for successful academic writing has been frequently emphasised in the literature (e.g. Hyland, 2004;Hyland-Tse, 2004). Metadiscourse has been labelled a crucial element of text meaning, "which helps relate a text to its context, taking readers' needs, understandings, existing knowledge, prior experiences with texts and relative status into account" (Hyland-Tse, 2004: 161). By using metadiscourse features correctly, writers can "highlight certain relationships and aspects of the organisation to accommodate readers' understandings, guide their reading, and make them aware of the writer's preferred interpretations" (Hyland-Tse, 2004: 164). There is a plethora of research articles on metadiscourse and its use in academic writing which highlight its significance for both writers and readers (e.g. Alyousef, 2015;Blagojević, 2008;Hauranen, 1993;Hyland, 1998;2000;2001;2005;Li-Subtirelu, 2015;Vande Kopple, 2012).

Previous studies on metadiscourse in L2 writing
The use of metadiscourse markers plays a very important role in the development of pragmatic competence of foreign language learners, especially in the context of academic target situations, within the language for academic purposes (Hyland, 2006). Pragmatic competence implies "the ability to communicate your intended message with all its nuances in any socio-cultural context and to interpret the message of your interlocutor as it was intended" (Fraser, 2010: 15).
Essay writing is an integral part of developing student L2 skills and it is usually included in the curriculum of foreign language courses at all levels of education, including teaching a foreign language at university. Still, it may be argued that undergraduate students are mostly inexperienced writers when it comes to academic writing, both in their native and second language. The case of L2 writers is emphasised as quite specific in the literature since they are in the process of learning the foreign language, e.g. mastering its morphology, syntax and lexicon (Intaraprawat-Steffensen, 1995: 255), but, on the other hand, are faced with "the task of learning the conventions of an L2 discourse community" (Intaraprawat-Steffensen, 1995: 255). An important issue that is also raised in the literature is the audience that the student writes to in an academic setting. It is usually a teacher, who may pay more attention to lexis and grammar, and less to text coherence and the ideas expressed in the text (Intaraprawat-Steffensen, 1995), thus failing to develop awareness in students about the importance of metadiscourse features necessary for successful writing.
The inability of students to use metadiscourse adequately in university foreign language classes "might be a consequence of an insufficient input of metadiscourse knowledge during their secondary education" (Ho-Li, 2018: 65). The issue of previous knowledge and the extent of knowledge about the correct use of metadiscourse has been mentioned as important in other studies as well (e.g. Intaraprawat-Steffensen, 1995). These authors argue that certain metadiscourse markers show good effects of teaching, such as connectives, coordinating conjunctions and conjunctive adverbs, which are usually included in the ESL textbooks (Intaraprawat-Steffensen, 1995).
Various studies have dealt with the use of metadiscourse in a university setting (e.g. Aull-Lancaster, 2014;Hyland-Tse, 2004;Lee-Subtirelu, 2015;Li-Wharton, 2012;Kobayashi, 2016;, with special attention paid to writing in the second language, especially English. The research was focused on different aspects of metadiscourse, such as establishing the types of metadiscourse markers used, their use in successful and less successful essays, as well as the differences in their use depending on the students' native language and the effects of teaching metadiscourse markers. Some studies investigated the use of specific markers, such as hedges (Crompton, 2012), or specific metadiscourse categories, such as interactional markers (Lee-Deakin, 2016).
It has also been shown that postgraduate writing is characterised by the frequent use of transitions (Hyland-Tse, 2004). The comparison of the use of metadiscourse in successful and less successful university students' essays has shown that successful essays contained more hedges than the less successful ones (Lee-Deakin, 2016), as well as that, regardless of the quality of the essay, boosters appeared less frequently than hedges in student essays. Previous research has found that there are differences in the use of metadiscourse markers in L2 essays depending on the L1 groups (Kobayashi, 2016). Some studies have also revealed gender variations in the use of metadiscourse features (Crismore et al., 1993).

MATERIALS AND METHODS
As stated in the introduction, the aim of this study is to investigate the frequency of use of metadiscourse markers in the genre of student essays and to classify them according to the taxonomy provided in Hyland's interpersonal model of metadiscourse (Hyland, 2005). We used a corpus of 100 essays written by the first-year students of the Faculty of Mechanical Engineering, University of Belgrade (N=100), all native speakers of Serbian. The gender structure of students was 76% of male students and 24% of female. Students' English-language proficiency had not been tested prior to the task of essay writing, and hence the data on their proficiency can only be inferred from the grades they obtained in English in the course they attended at the university (level B2/C1). The largest number of students had the highest grades, 10 and 9 (32% of students obtained the grade 10, and 29% the grade 9), 26% of students had an 8, while only 13% of students had the lowest grades (7% had the grade 6, and 6% obtained the grade 7). Hence, it may be argued that the general proficiency in English of the students who wrote the essays was at an upper-intermediate level towards the advanced level, although this cannot be said with certainty, due to the lack of adequate data.
The students were given the task to write an essay ranging from 150 to 250 words as part of their final English exam, with 45 minutes at their disposal for this activity. They could choose between two topics pertaining to mechanical engineering: Advances in Mechanical Engineering in the 21 st century and Mechanical Engineering -Branches and Opportunities. These issues had previously been discussed during regular classes at the university, but students were not explicitly instructed on how to use metadiscourse markers.
Once the essays had been collected, they were carefully read by both authors. The authors independently identified and classified metadiscourse markers according to the developed coding protocol based on the aforementioned typology provided by Hyland (2005). Since linguistic units may have the role of metadiscourse in some parts of the text and not in some others (Hyland, 2004), special attention was paid to the context in which a specific unit occurred. There were several situations in which the authors' judgements were different, and these cases were solved by mutual discussion, resulting in agreement. The findings of the metadiscourse markers' classification and the results of additional statistical analyses which indicate the relations between the number of metadiscourse markers used and the length of essays, students' gender and their grade in English are provided in the following section.

RESEARCH RESULTS
We previously mentioned that the corpus of student essays comprised 100 essays. The total number of words was 18,650 (i.e. the average length of an essay was 186.5 words). The results of our analysis show that students used 980 metadiscourse markers in total, which is 9.8 per essay (the normalised frequency per 1,000 words was 52.5). The highest number of markers used in an individual essay was twenty-six, while the lowest was one marker, used in three essays. There were 607 interactional markers in total (the normalised frequency was 32.55 per 1,000 words) and 373 interactive markers (the normalised frequency at 20.00 per 1,000 words). The most numerous in the individual categories were transitions (n=281; 15.07 per 1,000 words), which belong to the group of interactive markers, followed by engagement markers (n=254; 13.52 per 1,000 words) and hedges (n=222; 11.90 per 1,000 words), which belong to the group of interactional markers.
The statistical analysis has shown that longer essays, as a rule, contained a larger number of markers (r = .23, p < .05), i.e. the length of essays was significantly and positively correlated with the number of markers. On the other hand, the number of markers in individual essays was not significantly correlated with the students' grades in English (r = .02, p = .87). The difference between male and female students in the number of markers used was not significant either, confirming that both males and females, on the average, used the same number of markers in their essays.
Although the two topics which served as writing prompts for students were considered equally argumentative, the results show that the average number of used metadiscourse markers varied depending on the topic. A larger number of markers was used in the case of the topic Advances in Mechanical Engineering in the 21 st century (11.2 markers per essay on the average) than in the case of the topic Branches of Mechanical Engineering (8.8 markers on the average) and this difference is statistically significant (t (98) = -2,15, p < .05). A fairly similar number of students chose each topic (54% selected the topic Advances in Mechanical Engineering in the 21 st century and 46% opted for the topic Branches of Mechanical Engineering). There were no differences in the average grade of the two groups of students that chose a different topic (t(98) = .10, p = .92).

The use of interactional metadiscourse in the analysed corpus of essays
As stated earlier, the number of interactional markers in the analysed corpus was significantly higher than the number of interactive markers. The distribution of interactional markers according to categories (engagement markers, hedges, boosters, self-mentions, attitude markers) is given in Table 1. Engagement markers were the most used category of interactional metadiscourse (the normalised frequency at 13.62 per 1,000 words). They serve to include the reader in the text and indicate the presence of the reader and the shared knowledge (Hyland, 2005: 53-54). These are some of the examples of engagement markers found in the corpus. 1 (1) Who knows what will happen in future and how new discoveries will look like? (2) You can use air tunel to improve your aircrafts aerodynamics and with that save a galons of fuel.
(3) One of the mechanical engineering branches in which we can see these advances is robotics.
In (1), the whole question is an example of an engagement marker: the writer raises a rhetorical question, which serves to seek the attention of the readership, i.e. engage the readers into an argument. In example (2), the studentwriter explicitly addresses the audience, emphasising their presence in the text by using the pronoun you and the possessive adjective your (cf. Hyland 2005: 53). Furthermore, example (3) shows that inclusive we can also be used to highlight the inclusion of the readers in an argument.
The second most frequently used category of interactional markers in the analysed corpus includes hedges (the normalised frequency was 11.90 per 1,000 words). The purpose of hedges is to express the author's opinion and position rather than facts. Hence, their use in the corpus suggests different degrees of confidence (Hyland, 2005: 52), as shown by the examples below (4-7). In (4) and (6), student writers used the hedge almost, which decreases the strength of the claim and the degree of confidence. This may imply that they wanted to be cautious when it comes to generalisation. In examples (5) and (7), the use of modal hedges could, maybe and perhaps demonstrates that the writers withheld the full commitment towards their proposition.
(4) Almost every process in any kind of industry can be done by mechanical.
(5) Science is improving so fast that we could maybe in future travel through time, go to other planets and live on the, we don't know so we'll have to wait to finde out. (6) Every single branch helps people in almost everything they do though day and, generally speaking, it makes our life easier. (7) Perhaps that one skill is the mechanical engineer's exclusive domain is the ability to analyze and design object and systems with motion.
The frequency of boosters, self-mentions and attitude markers in the analysed corpus of student essays was not very high (with the normalised frequencies at 3.97, 2.68 and 0.38 per 1,000 words, respectively).
Boosters serve to show the author's confidence and certainty regarding the topic and the audience (Hyland, 2005: 52-53). In (8), the use of the booster there is no doubt implies that the student writer wished to emphasise that it was very likely or highly possible that the expressed claim was true. Similarly, the use of the adverb surely in example (9) indicates that the writer had a confident attitude towards the proposition. (8) In the conclusion, I would say that there is no doubt that mechanical engineering has big opportunities and potential to make greater progress in future. (9) One of the biggest advantages of mechanical engineering in 21 st century is a possibility that it will slowly, but surely, for the next couple of decades be in a position to solve global issues.
Self-mentions in (10) and (11) imply the explicit presence of the author(s) in the text, and their use shows openly their relation towards "their arguments, their community and their readers" (Hyland, 2005: 53).
(10) I get a bit sad when I hear that old-fashion pen and ruler are not used any more.
These are only two examples and I could tell you a 100 more.
Attitude markers, indicating the author's affectivity regarding the propositional content (Hyland, 2005: 53), were the least present in the analysed essays. For instance, students used phrases such as it is no wonder that (example 12) not to show surprise or proudly (example 13) to show pride.

(12)
Keeping in mind that engineers have a wide errey of subjects to choose from in colleges and it is no wonder that it is the study of the future.
Thanks to the recent technological advances in mechanical engineering, new generations can proudly say that we as a human kind are closer then ever to realization of an energy free and clean society.
The rather infrequent use of these categories of metadiscourse markers may imply that students are still not aware of their uses and significance, as well as of the effect they can produce in readers. Based on the analysed essays, it may be argued that students lack confidence when it comes to discussing some points in their arguments since they mainly avoid using the personal pronouns I or we, thus opting not to show their attitude towards the topic.

The use of interactive metadiscourse in the analysed corpus of essays
The distribution of interactive markers according to categories (transitions, code glosses, frame markers, evidentials and endophoric markers) is given in Table  2. It should be noted that all other interactive metadiscourse categories apart from transitions were used to a very low extent, while the category of endophoric markers was not found in the analysed corpus at all. The reason for missing endophoric markers in this type of writing can be the shortness of the essays, as well as their main purpose. Namely, endophoric markers are usually used for pointing to other parts of the same text and since the analysed essays can be considered short forms, it is somewhat expected that this type of markers be rarely present or not present at all. Transitions were by far the most frequently occurring category of markers overall (the normalised frequency was 15.07 per 1,000 words), which is in line with some previous studies (e.g. Ho-Li, 2018;Hyland-Tse, 2004). Their function is to link ideas and help the reader to interpret them, implying addition (e.g. and, furthermore), comparison (e.g. similarly, likewise, in contrast, however) or consequence (e.g. thus, therefore, in conclusion, nevertheless, anyway) relations (Hyland, 2005: 50). The examples of various types of transitions identified in the corpus are given below (14-16). In example (14), the use of however and but implies that students sought to compare different views by contrasting them; the transition marker because used in example (15) suggests the reason for the claim stated in the previous clause, while the use of the marker although again implies a contrast between the statements introduced in the subordinate and main clauses (16).

(14)
However, every modul (branch) will bring you some money, but first you have to study, a lot. (15) Mechanical engineers have to work closely with computer scientists, biologists, architects etc. because there is a big distinction between every engineering field. (16) Although it is hard to predict exactly how mechanical engineering will advance in the 21 st century, I think that we can conclude that the new century will bring new advances in the area of robotics.
Code glosses serve to clarify the given information, provide more details or rephrase information (Hyland, 2005: 52). Not many examples of this category were found in the analysed corpus (the normalised frequency at 3.06 per 1,000 words). The code glosses used in the corpus (examples 17 and 18) serve for illustrating instances (for example), as well as for distinguishing one situation from another (particularly).

(17)
For example, mechanical engineering is used in bio-medicine, engineers can fight disieses and do many more.
Particularly, by using various technics, mechanical engineers developed a series of systems which greatly improved possibilities of using renewable energy.
The frequency of frame markers and evidentials is also fairly low in the analysed student essays (the normalised frequencies at 1.61 and 0.27 per 1,000 words, respectively). The function of frame markers is to "provide framing information about elements of the discourse" (Hyland, 2005: 51). In corpus examples (19,20), we can see sequencing, which is introduced so that the readers can clearly see the order of different arguments.

(19)
First of all this is reflected in it's branches. (20) And last but not the least is the 3d Computer-aided design which helps mechanical engineers in creating new inventions with more precision and less time.
Evidentials indicate other sources of information, that is, other authors' stances, and are valuable for arguments (Hyland, 2005: 51). These were also few and far between in the analysed corpus. Illustrative examples (21) and (22) indicate that students wished to substantiate their own claims by relying on the opinion of experts and scientists, thus raising the credibility of their own arguments.

(21)
Experts are of the opinion that in the next ten years for now, every company that operates in mechanical engineering should be more advanced in training and use of computerication.
Scientists believe that is future of the world. The development of mechanical engineering never stops.

Adequacy of the use of metadiscourse markers in the analysed corpus
Looking at the obtained results, it may be argued that students generally recognised the importance of using metadiscourse markers when writing essays (52.5 per 1,000 words). Some students have shown a very high level of proficiency in using these markers within their essays. They were able to properly employ metadiscourse markers for expressing and connecting their thoughts, rephrasing and giving examples, including the readers into the text, balancing between the propositional and non-propositional content of the essay, etc. In contrast, there were students who obviously struggled when it came to this segment of writing; they either used metadiscourse features to a very small extent or used them inadequately.
Some students tended to use as many metadiscourse markers as possible, without paying attention to whether this was appropriate or not. However, students were not informed about the purpose of the research, so this could not have been the reason for using the markers excessively in certain cases. A more plausible explanation for the inadequate use of discourse markers may be students' lack of knowledge about the proper use of these linguistic units, which might be attributed to the fact that they were not explicitly instructed how, why and when to utilise markers at previous levels of education. We may also argue that students used plenty of markers because they considered them as valuable devices to fill in the gaps or write the necessary number of words (approximately 150 words).
Certain markers were repeated a number of times within a single text (e.g. and, but, so, for example, such as). Another instance of the misuse of markers in the analysed corpus regards the overuse of questions (e.g. three long questions in a 206word or 170-word essay). Although the overall frequency of self-mentions was low, in some essays the first-person pronouns I (e.g. I think, I assume, I expect) and we (both exclusive and inclusive: We can make…, we are connected…, we use some kind of…, we drive cars) were used more than three times in a row, which can also be regarded as overuse. Some of the students (seven essays in total), probably wishing to involve the reader in the text as much as possible, used the pronoun you too often in successive sentences.

CONCLUDING REMARKS
The aim of the current study was to investigate the presence of metadiscourse markers in undergraduate student essays written in English with the topic related to their future profession as a prompt. The study has shown that university students frequently use markers although they were not explicitly instructed to do so, at least not in university foreign language classes. Students mostly used interactional metadiscourse (predominantly engagement markers and hedges), which might imply that they readily involved themselves in the text trying to capture the attention of the readers and make them participate in the discussion. This is a difference compared to the findings of previous research, which registered a higher use of interactive metadiscourse (e.g. Crismore et al., 1993;Li-Wharton, 2012). On the other hand, the most frequently used markers overall were transitions, a type of interactive metadiscourse, which may indicate the students' awareness of the importance of connecting ideas into a coherent whole. The high use of transitions has also been recorded in other similar studies (e.g. Hyland-Tse, 2004).
Although the students were not previously introduced to the features of metadiscourse in university foreign language courses, the results of this study undoubtedly show that, in the majority, they were able to use different types of metadiscourse. The pattern of distribution of metadiscourse markers used in the essays shows that the students' primary focus was on engaging the readers, joining different ideas and cautiously stating their own opinions on the given topic. Still, some markers were misused, or the same markers were sometimes overused. Based on the results of this study, we may argue that university students need more instruction in the adequate use of markers. Students should learn about "appropriate ways to convey attitude, mark structure, and engage with readers" (Hyland, 2004: 148) by exploring metadiscourse in their own and in published writing. Further, ESP instruction in writing, e.g. the types of writing tasks in English classes at the university, should target the needs of these students so as to correspond to writing tasks in their future professional life, such as writing project proposals, technical reports for international clients, and ultimately, research articles in English. One of the possibilities for the practical introduction in ESP classes is the usage of authentic corpora (for example, the corpus of research articles, the corpus of student essays) with and without metadiscourse markers given so that the students can see the differences and learn the relevance and purposes of these markers.
The limitation of the current research is the fact that it served only to obtain insight into mechanical engineering students' current proficiency in the use of metadiscourse. Furthermore, we had no data on the input concerning the teaching of metadiscourse features at previous levels of education, so we could not judge whether some students had already been introduced to these markers, which may well have enhanced their awareness of the significance of using these devices when writing essays. Therefore, this research could be extended to, first of all, check the previous knowledge of students in the field of metadiscourse and judge their progress in ESP university courses. Future studies could also include the essays of the students of other majors with the writing prompts related to their own disciplines and compare the frequency of use and types of metadiscourse markers used.
It can be concluded that explicit introduction to the use of metadiscourse markers within the ESP curriculum would help students become more confident and rhetorically aware when presenting their ideas and attitudes so as to avoid pragmatic failures. However, students should be also instructed on how to use these markers adequately in different contexts through carefully devised tasks during English language classes at university. In addition, their attention should be drawn to the right proportion of using metadiscourse markers in writing so as to enhance their overall pragmatic competence in ESP.