Educational Online Discussion Forums (EODFs) can be successful in enhancing collaborative learning by attracting students to participate and interact (Swan et al., 2000). This asynchronous discussion provides an opportunity for L2 learners and teachers to engage in potentially dialogic interactions. As far as oral speech lies at the core of face-to-face classroom interaction (Singh, Hawkins, & Whymark, 2007; Wells, 1999), electronic discourse can be a venue for the sort of written conversation characteristic of online interaction (Davis & Brewer, 1997). According to Saadé and Huang (2009, p. 89), “This online interaction is central to the development of the instructional process in asynchronous computer discussion. The significance of discourse in the learning process is anchored in the theories that view the development of thought to be mediated by social discourse (Vygotsky, 1962; Wertsch, 1998).”

To fully understand the discourse that occurs in ODFs in the teaching-learning context, a methodology to measure and analyze data for analytical and holistic perspectives must be used (Schrire, 2006). In this research, the purpose was to describe the discourse that occurred during one semester within a specific online learning context. The theoretical framework for the research was based on Hyland’s (2004) model. It was applied to the investigation of an EODF used as one learning component of a general English course for male and female university students of engineering. By adopting a corpus analytic approach to the students’ comments and interactions within the context of the course, it was possible to describe the preference they manifested in using metadiscourse markers.

REVIEW OF LITERATURE

Discourse Markers: A Generic Perspective

Genres represent a community's culture and hence focus on social actions embedded within its practices; they are a form of situated cognition implanted in a community's culture (Sapir, 1993). Some familiar examples of genres would include business reports, research articles, and textbooks, while e-mail, blogs, and discussion forum postings are examples of more recent media-based genres. A genre is a reflection of conventions observed by the professional or academic community that have a communicative purpose in common. It incorporates a language (discourse) defined in terms of what people do with it.

The Web forum genre combines commentary (personal opinions on a current topic), community (Web pages containing framework for the content supplied by users such as forums), Usenet portals with user-generated content, and conversation forum/guestbook, which includes forums where people talk about random events. In fact, the forum involves many-to-many communication, often about a certain topic. It may be more accurate to rename these discussion forums. “Thematic forums,” “free conversational forums,” and “professional forums” seem true genres. The thematic forum/newsgroup involves interactive discussion on a specific topic.

Within genre analysis, discourse markers refer to are words or short “lexicalized phrases” that structure texts. This structuring is achieved by showing “how the speaker intends the basic message that follows to relate to the prior discourse” (Schiffrin, 2001, p. 59). Discourse markers function to produce cohesion and coherence in a given text creating a connection between propositions existing within the text. Some of the connections noted by Schiffrin are causal (Therefore), conditional (If X, then Y), temporal (then he...), adversative (However) and additive (Furthermore). Discourse markers also occur when speakers shift their orientation to information. In this case, the marker alerts the listener that something within the speaker has changed. Schiffrin (1999) uses “oh” as an example of this type of discourse marker.

Discourse markers are found in various grammatical categories such as conjunctions, interjections, and adverbs. Schiffrin (2001) argues that discourse markers can “connect utterances on either a single plane or across different planes” (p. 57). This allows a discourse marker to establish coherence by connecting units of dynamic meaning and to enhance cohesion as the feature of the surface structure of the text. On the list of discourse markers, “then” is used in two different examples. This reveals the potentiality of a single to demonstrate various kinds of relationships between clauses. As Fraser (1988) argues, the absence of the discourse marker does not render a sentence ungrammatical and/or unintelligible. It does, however, remove a powerful clue about what commitment the speaker makes regarding the relationship between the current utterance and the prior discourse.

Metadiscourse: Basic Features and Types

Metadiscourse is a ubiquitous feature of the way writers portray themselves and their comments. Hyland (2000, as cited in Hyland & Tse, 2004) describes metadiscourse as the linguistic resources employed to organize a discourse or the writer’s attitude toward its content or the reader. Crismore (1984) believes that the aim of metadiscourse is to “direct rather than inform the readers” (p. 280). In his book Genre Analysis, Swales (1990) suggests that metadiscourse is “writing about the evolving text rather than referring to the subject matter” (p. 188). Hyland and Tse (2004) believe that writing is viewed as an engagement between writer and reader which possesses a social and communicative basis; and metadiscourse is employed by writers so that they can signal their attitude toward both the text content and audience through projecting themselves into their discourse.

Hyland (2005) suggests that every research article, book review, student essay, grant proposal, language class, and conference presentation can only succeed if speakers and writers deploy metadiscourse appropriately to convey a credible persona and relate to an audience in ways that seem familiar and engaging. Ultimately, we are convinced by an argument that seems to describe the world in a way that makes sense to us. It follows that metadiscourse plays a critical role in bringing us to this point. As Simons (1980) has eloquently expressed it, “although the scientific donkey may have been pinned with an unbecoming rhetorical tail, it is still capable of carrying a heavy load” (p. 127).

A central aspect of metadiscourse is its context-dependency, the close relationship it has with the norms and expectations of those who use it in particular settings (Hyland 2000, as cited in Hyland & Tse, 2004). This contextual specificity is particularly apparent in the ways in which metadiscourse is distributed across different genres, assisting writers and responding to and constructing the contexts in which language is used. Hyland (2005) explains that as metadiscourse functions to represent the social purposes of writers, it can be considered as a social act rather than a simple string of language items, and this means that its use will vary enormously depending on the audience, the purpose, and other aspects of the social context. In turn, studying this variation reveals the diversity in patterns of use and helps us understand the ways individuals use language to orient to and interpret routine communicative situations.

Metadiscourse markers are considered as forms which make textual and interpersonal relations. According to Halliday (1973), the interpersonal function concerns the relationship between addresser and addressee, that is, the role of the speaker and the role assigned to the hearer. Interpersonal markers are comprised of the two interactive and interactional dimensions (Hyland, 2001), The former concerns the ways the writer seeks to accommodate its knowledge, interests, rhetorical expectations, and processing abilities while the latter concerns the ways the writer conducts interaction by intruding and commenting on his or her message.

Popular writings are most obviously distinctive in their use of interactional metadiscourse, particularly in the ways writers make their attitudes to material explicit. Epistemic devices, which allow the writer to comment on the status of propositions, are key features of good writings. Hedges and boosters carry the writer’s degree of confidence in the truth of a proposition, displaying an appropriate balance between scientific caution and assurance, but they also present an attitude to the audience (Hyland, 2005). Knowledge claims have to be carefully handled, so writers must invest a convincing degree of assurance in their propositions while avoiding overstating their case and risking inviting the rejection of their arguments.

Hyland (2004) developed a two-componential taxonomy of metadiscourse markers which is portrayed below:

(1) Interactive Resources: These devices allow the writer manage the flow of information in order to propose his/her preferred interpretations. These resources, according to Hyland, encompass the following:

• Transitions: These devices mainly indicate additive, contrastive, and consequential steps in the discourse. Some examples are in addition, but, thus, and.
• Frame markers: They indicate text boundaries or elements of schematic text structure, such as my purpose here is to, to conclude.
• Endophoric markers: They refer to information in other parts of the text and make the additional material available for the readers. They include in section 2, noted above, etc.
• Evidentials: They refer to sources of information from texts other than the current one, such as Z states, according to X.
• Code glosses: These devices show the restatements of ideational information, such as in other words, e.g.

(2) Interactional resources: They “focus on the participants of the interaction and seek to display the writer’s persona and a tenor consistent with the norms of the disciplinary community” (Hyland 2004: 139).

The interactional resources include:

• Hedges: Theyindicate the writer’s unwillingness to present propositional information categorically and include downtoners which reduce force of statements (such as about, perhaps, fairly, almost, partly),frequency adverbs which make statements indefinite (such as usually, sometimes), and hedges which decrease responsibility for truth (such as probably, perhaps, may).
• Boosters: These devices reinforce truth value through expressions of certainty and emphasis. Some examples are it is clear that, definitely, certainly, really. They also include amplifying adverbs, such as totally, always.
• Attitude markers: They indicate the writer’s appraisal of propositional information. Some examples are I agree, surprisingly.
• Engagement markers: They address readers explicitly or make a relationship with the reader. Some examples are you can see that, note that, consider, second-person pronouns (address reader directly) you, your, yourself rhetorical questions (speak directly to reader) necessity modals (direct reader to action or thought) must, should presupposition markers (assume sharedness) of course, obviously.
• Self-mentions: They refer to the extent of author presence in terms of first person pronouns and possessives (direct involvement of writer). Some examples are I, we, our, my, me, mine.

Mulholland (1999) found out that when expected interactional features are missing and important affective and interpersonal elements are omitted, writers damaged the message and risked spoiling harmony and cooperation. Elsewhere, Ohta (1991) and Scollon (1994) found out that Asian students are often believed to favor collectivist ways of expressing identity or opinion, avoiding self-mention to disguise the direct involvement and views of the writer.

Globalization, as Hyland (2005) argues, has increased intercultural and interlingual contacts and raised questions concerning whether writers socialized into a non-English writing culture learn rhetorical habits that affect their writing in English. Researchers have, therefore, started to explore metadiscourse in various languages and the way speakers of those languages use it in English. The fact that many of these studies have focused on academic texts is unsurprising given the internationalization of this field for both students and professional scholars. According to Lantolf (1999), cultural factors help shape our background understandings, or schema knowledge, and are likely to have a considerable impact on what we write and how we organize what we write, and our responses to different communicative contexts. Culture is seen as inextricably bound up with language (Kramsch, 1993). Cultural factors have the potential to influence perception, language, learning, and communication, particularly the use of metadiscourse.

PURPOSE OF THE STUDY

In the Persian context, metadiscourse markers have been investigated in EFL learners’ discourse in two ways. Some studies focused on the learners’ use of metadiscourse markers while producing a text in English; some others sought to describe the learners’ use of metadiscourse markers from the perspective of contrastive/cross-linguistic genre analysis. However, the use of metadiscourse markers in the dialogic discussion genre, particularly in an online context, has received scant attention. As a result, the present study aimed to investigate the use of interactional metadiscourse markers in English comments written by university students in an online discussion forum. To this end, the following questions were raised:

(1) What types of interactional metadiscourse markers are employed by university EFL students in an educational online discussion forum?

(2) Is there any significant difference between male and female university EFL students in their use of interactional metadiscourse markers written in an online discussion forum?

METHOD

Corpus

The corpus of this study included 168 comments tagged to an online discussion forum used in a general academic English course for the students of engineering. The comments were on six different topics, such as globalization and corporal punishment. The comments were written by 28 university students of engineering enrolled in a general English course. They consisted of 14 males and 14 females, and their age range was between 18 and 20.

Data Collection Procedure

The students were asked to write their comments on six different topics throughout the semester which lasted for 12 weeks. Every other week, the teacher introduced a topic which was, by and large, controversial, so that it would motivate the students to comment on. The students were asked to think about the topic and tag their comments to an online discussion forum set up by their instructor for this purpose. By the end of the semester, the students’ comments on six topics had been tagged to the forum. All in all, there were 168 comments to be analyzed for the investigation of the students’ use of interactional metadiscourse markers.

Framework for Data Analysis

In order to compare and analyze probable differences between metadiscoursal characteristics of the comments tagged to the discussion forum, the metadiscourse taxonomy of Hyland (2004) and the Concordance software were employed. The Hyland model includes five categories of interactional metadiscourse markers: 96 hedges (e.g. about, perhaps), 64 boosters (e.g. certainly, definitely, think, believe, it’s clear that), 65 attitude markers (e.g. I agree, important, surprisingly), 76 engagement markers (e.g. you can see that, note that, consider, do not), and 11 self-mention (e.g. I, we, our, my).

The corpus, including 168 comments of students in the discussion forum, was fed into the Concordance software in order to find the categories and numbers of interactional metadiscourse markers. Since the 1,000-word approach is the usual method adopted by researchers to analyze discourse markers (e.g. Crismore, Markkanen & Steffensen, 1993), a corpus of every 1,000 words was selected in order to make male and female students’ corpora comparable in terms of the proportion of metadiscourse markers’ frequency to the total number of words in the comments. After calculating the frequency of each category of metadiscourse marker per 1,000 words, the chi-square test was run to investigate whether there were any gender differences in the use of the markers.

RESULTS

EFL Learners’ Use of Interactional Metadiscourse Markers

In order to compare the type and number of interactional metadiscourse employed by male and female university students in their use of interactional metadiscourse in a discussion-based e-forum, first we calculated the frequency of different types of interaction metadiscourse using the Concordance software. Table 1 shows the total frequency of interaction metadiscourse used by EFL learners.

Table 1: Interactional metadiscourse and its categories used in the online discussion forum

Different metadiscourse markers were not utilized in the same rate. Among 174 discourse markers used per 100 words, engagement markers appeared most frequently in the corpus (M=65.10). By contrast, hedges were the least favored metadiscourse markers (M=18.23). Figure 1 depicts variation in the use of metadiscourse markers in the total corpus.

Figure 1:  The number of interactional metadiscourse and its categories used in the total corpus

To illustrate the use of various metadiscourse markers, an extract from the corpus written by a female participant tagged to the discussion forum is given below:

I(self-mention) guess the trend in most of the societies seems to be to abandon physical punishment in favor of other disciplinary tools. Some parents use this method: Isolating a child for a certain amount of time or supporting him or her from watching his or her favorite TV program when s/he gets a little unruly, gives him/her a chance to cool down. Other parents respond and reward desired behavior combined with ignoring undesirable behavior. Some parents sit the children to discuss and explain the right behavior while others simply scold or yell at the kids upon a wrongdoing. Many people in our (self-mention)society may (hedge) refute these methods as being too mild or permissive. When parents are extremely permissive, children become spoiled. On the other hand, parents are extremely strict when they expect immediate obedience, give no explanation for demands and use physical punishment frequently. Children of too strict parents are timid. Effective discipline is helping and teaching. The purpose of discipline is to help children (or students at uni:)learn to do what is regarded right . This is an important (attitude marker) cornerstone of disciplining: When you (engagement marker) react, you (engagement marker)are acting quickly, and that usually (hedge) means emotionally. The biggest danger of reacting is that you (engagement marker) may (hedge) hurt the other person, either emotionally or physically. Generally speaking, when you (engagement marker) respond in any situation, you (engagement marker) should (hedge, engagement marker) take some time to think (booster), so that you (engagement marker)can be more logical. Now I (engagement marker) prefer (attitude marker) you (engagement markers), my (self-mention) friends take a side, agree (attitude marker) or disagree (attitude marker), for or against? (engagement marker).

Gender Differences in the Use of Interactional Metadiscourse Markers

To investigate gender differences in the use of metadiscourse markers, the first step was to measure these differences in the use of total discourse markers. As the findings in Table 2 show, the corpus contained 3,426 interaction markers in total, of which 1,606 were used by males and 1,820 were used by females.

Table 2: Interactional metadiscourse and its categories used by males and females in the online discussion forum

In fact, males used about 166 metadisccourse markers and females used about 181 of all interaction metadisccourse markers in every 1,000 words. The findings show that females incorporated more discourse markers in their online discussions. The difference is graphically displayed in Figure 2.

Figure 2: Interactional metadiscourse markers used in English in 1000 words in the online discussion forum

To measure the significance of the gender difference in the use of interactional metadiscourse, chi-square was employed. The results of the chi-square analysis showed that the value of observed chi-square was not significant (x²=.56, df=1, p<.05), indicating that there was not a significant difference between males and females in their use of total interactional metadiscourse markers.

After investigating the impact of gender on the use of metadiscourse markers in general, gender differences in the use of the five components of metadiscourse markers were addressed. The five sections below describe the results of the investigation.

Attitude Markers

Different types of interaction metadiscourse were used differently by participants. The use of Attitude markers is displayed in Table 3. As the table shows, 412 attitude markers were used by the whole population, of which 153 were used by 14 males and 259 were used by 14 females. In other words, males used 10% while females used 18 % attitude markers per person. Among attitude markers, important (e.g. “This is an important cornerstone of disciplining”), agree (e.g.“I totally agree with Mojgan”), even (e.g. “Maybe I need to repeat my sentences to him and tease myself twice, three times or even more”), unfortunately (e.g. “It is very useful, but unfortunately it isn't proposed in our country”), interesting (e.g. “This subject is way more interesting than the last one”), and prefer (e.g. “Some people prefer staying single throughout their lives”) have more frequency.

Table 3:Attitude markers used in the online discussion forum

As the frequency per 1,000 words shows, female learners used about 25 attitude markers, while male learners used about 15. This result is displayed clearly in Figure 3.

Figure 3: Attitude markers used in 1000 words in the online discussion forum

To calculate the gender difference in the use of interactional metadiscourse, chi-square was employed. The results revealed that the value of observed chi-square was not significant (x²=2.38, df=1, p<.05). This shows that there was not a significant difference between male and females in their use of attitude markers.

Self-mentions

The number of self-mentions used by this population is shown in Table 4. This table shows that the participants used 906 self-mentions, of which 440 were used by males and 466 were used by females. It can be said 45% and 46% self-mentions were used by males and females, respectively. The self-mentions used most frequently were words such as I (e.g. “I don't believe in age gap”), we (e.g. “We are a part of this big puzzle”), my (e.g. “I have my own ideas about the basic questions I've mentioned above”), our (e.g. “Failing to grant material needs would injures our body, promptly”), me (e.g. “I agree with them but life and love are really imprecise for me, especially love”), and us (e.g. “We can see many examples of them around us”).

Table 4: Self-mentions used in the online discussion forum

In every 1,000 words, females used about 46 and males used about 45 self-mentions. As a result, there was a marginal difference between males and females, with the latter drawing on more self-mentions in the writing. Figure 4 illustrates the difference.

Figure 4: Self-mentions used in 1,000 words in the online discussion forum

The significance of gender differences in the use of self-mentions was measured through chi-square. Based on the results, the value of observed chi-square (x²=.00) was not significant at the significance level of p<.05, with 1 degree of freedom (df=1). This shows that the difference between males and females in their use of self-mentions was not significant.

Boosters

Descriptive statistics about the use of boosters are shown in Table 5. The table shows 469 boosters were used by both males and females: 225 by males and 244 by females. The words think (e.g.“I think the disadvantages are more than advantages”), believe (e.g.“I believe in destiny and I think no event occur in the real world”), and know (e.g. “Because when we don't know our fate, we must try to make it better”) had more frequency among boosters.

Table 5: Boosters used in the online discussion forum

In 1,000 words, female participants (M=24) used boosters more frequently than males (M=23), albeit marginally. Figure 5 shows the difference graphically.

Figure 5: Boosters used in 1,000 words in the online discussion forum

The chi-square test was employed to calculate gender differences in the use of boosters. The chi-square results showed that the value of observed chi-square was not significant (x²=.021, df=1, p<.05). As a result, the finding does not support any significant difference between males and females in their use of boosters.

Engagement Markers

Table 6 shows the frequency of engagement markers. As the results in the table show, 1,281 engagement markers were used, with 593 of which used by males and 688 by females. Engagement markers such as should (e.g. “If you have faith in God and the Day of Judgment, you should be responsible for your own works”), consider (e.g. “I believe in men's free will and consider it one of the results of God's overwhelming knowledge and power”), must (e.g. “I must admit that nowadays money has become one of the most important concerns of the young”), do not (e.g. “We humans do not often understand the value of something unless we loose them”) and question mark (e.g. “What does that have to do with anything?”) were used more frequently than other engagement markers.

Table 6: Engagement markers used in the online discussion forum

Females used 68, while males used 61 engagement markers per 1,000 words. As Figure 6 depicts, females outperformed males in the use of engagement markers.

Figure 6: Engagement markers used in 1,000 words in the online discussion forum

To calculate gender differences in the use of interactional metadiscourse, chi-square was used. The findings showed the value of observed chi-square (x²=.492) was not significant at p<=.05, with 1 degree of freedom (df=1). Based on the observed chi-square value, it can be concluded that there was not a significant gender difference in the use of engagement markers.

Hedges

The use of hedges is depicted in Table 7. As the table shows, 359 hedges were used by both genders. Male participants used 196 hedges, and females 163 hedges. The hedges such as may (e.g. “It may have a special speech or meaning”), seems (e.g. “It seems comedians and comic movie makers have a great mission in their life”), would (e.g. “I know it would be a hard thing to do”), probably (e.g. “But the discussion probably won't take more than an hour”), usually (e.g. “It has usually no good effect but has inverse result”), sometimes (e.g.“Sometimes going to a psychologist can help us”), and might (e.g. “This approach might take a little work and time”) were used more frequency than the other hedges.

Table 7: Hedges used in the online discussion forum

As shown in Figure 7, unlike the other discourse markers, more hedges in 1000 words were used by males than females.

Figure 7: Hedges used in 1,000 words in the online discussion forum

The chi-square test was run to determine the significance of gender difference. The value of observed chi-square (x²=.444) at p<.05, with 1 degree of freedom (df=1), did not exceed the critical value. Therefore, gender did not result in any significant difference in the use of hedges.

As you can see from the above findings, there was not a significant difference between the use of hedges in two groups of males and females. The whole picture of the use of all categories of interactional metadiscourse markers is offered in Figure 8, which shows the proportion of each discourse marker to all discourse markers in percentages and across genders.

Figure 8: The percentages of interactional metadiscourse and its components used across genders

DISCUSSION

On the basis of the above findings, it can be concluded that English texts written by university students contained a total of 3,426 interactional metadiscourse markers; males used 1,820 (46.88%) and females used 1,606 (53.12%).  In addition, the whole population used 412 attitude markers, of which 153 (4.47%) were used by males and 259 (7.56 %) by females. Moreover, 906 self-mentions were used in total, with males using 440 (12.84%) and females using 466 (13.60%). The participants used 469 boosters: 225 (6.57%) by males and 244 (7.12%) by females. The most frequent markers were engagement markers, which were used 1,281 times: 593 (17.32%) by males and 688 (20.08%) by females. By contrast, the least frequent markers were hedges, which were used 359 times (5.72%) by males and 163 times (4.76%) by females.

The analysis of the corpus in the present study indicates that interactional metadiscourse markers constituted 16.88% of all the words the university students used in their online discussion forum. However, they employed the five categories of metadiscourse markers in various proportions, in the descending order of engagement markers (6%), self-mentions (4.5%), boosters (2.4%), hedges (2%), and attitude markers (1.7%).

Findings from this research clearly demonstrate that there was a difference between the type and amount of interactional metadiscourse employed by university students in an online discussion forum. Many reasons can account for the difference. One strong reason for variation in use, particularly the preference given to engagement markers, is related to the nature of the forum. Since the genre of this forum was discussion blog and participants discussed their opinions about different topics, engagement markers and self-mentions were used more frequently than others. This shows that participants relied more on their own personal opinions. This preference for certain metadiscourse markers suggests that their use is influenced by generic features.

The second reason is founded on the nature of participants’ major. They were engineering students, who tend to favor arguments as other hard sciences students do. An important aspect of a positivist-empirical epistemology favored in hard sciences is that the authority of the individual is secondary to the authority of the text and facts should be allowed to “speak for themselves.” Writers tend to use linguistic objectivity, as Hyland (2005) argues, to disguise their interpretive responsibilities as well as their rhetorical identities. The less frequent use of hedges, boosters, and attitude markers compared with engagement markers is one way of minimizing the writer’s role in arguing and appealing to readers. By contrast, hedges and boosters tend to be more common in the humanities and social science papers. This is mainly because the soft-knowledge fields are typically more interpretive than the hard sciences and their forms of argument rely more on a dialogic engagement and more explicit recognition of alternative voices.

Finally, preference in the use of certain metadiscourse markers may be related to the participants’ corresponding preference in L1. As there are a great many differences in the patterns of using metadiscourse markers between foreign language learners and native speakers (Ädel, 2006), the preference for certain metadiscourse markers by the participants in this study might be related to the corresponding patterns in their L1, Persian. As there is no particular research describing the use of EFL learners’ metadiscourse markers in academic discourse realized in online discussion genre, this reason presently lacks experimental support.

The findings related to gender differences showed no significant difference in the use of interactional discourse markers, except for self-mentions. This runs counter to some previous studies which showed the gender of the writer could influence how much or what type of metadiscourse is used. Crismore et al. (1993), for example, found gender and cultural differences between Finnish and American male and female writers. Finnish females used hedges the most and US males the least. Moreover, some research shows that males draw on emphatics more than females and manifest a more confident writing style (Francis, Robson & Read, 2001; Tse & Hyland, 2008).

CONCLUSIONS AND IMPLICATIONS

Metadiscourse markers are an important device for structuring the text. While interactive metadiscourse markers contribute to textual cohesion, interactional metadiscourse markers shape the interaction between the writer and the reader. Due to the dialogic nature of the discussion forum, participants in this study employed different interactional metadiscourse markers. The findings documented the use of all categories of metadiscourse markers, suggesting that metadiscourse markers are inherent to the online discussion as a highly dialogic type of genre.

Since metadiscourse markers play crucial roles in mediating the relationship between what writers intend to argue and their discourse communities, the results of the present study have obvious importance in increasing students’ awareness of the way they organize their writings. Metadiscourse is a valuable tool which provides rhetorical effects in the text, such as providing logic and reliance in the text, so the instruction of metadiscourse markers is a useful means for the teachers to help students enrich their writing practices for effective communication. In view of the significance of metadiscourse markers, descriptive studies about the frequency of their use should be followed by interventionist methods to teach EFL learners to enhance their metadiscoursal proficiency and to use metadiscourse markers more effectively. Moving toward the instruction of metadiscourse markers requires an investigation of variables affecting the acquisition of discourse markers, such as the explicit and implicit teaching of them, the effect of input enhancement and output tasks on their acquisition, and the role corrective feedback can have on EFL learners’ metadiscoursal knowledge. Further, EFL learners’ awareness should be raised as to effective use of proper discourse markers in terms of the characteristics of a particular genre.  

  Moreover, teachers have to be aware of all differences in the use of metadiscourse when they teach student to write by giving serious consideration not only to the topic and purpose of writing but also top the genre of writing, i.e. discussion, and the medium of communication, i.e. online communication. Therefore, pedagogically speaking, teachers need to teach all types of metadiscourse rhetorically, and future metadiscourse studies in terms of various genres are highly expected to add our knowledge of effective rhetorical strategies for this job.

As regards gender, this study showed female EFL learners used more metadiscourse markers  than males did, though the differences were minor and hence not significant. As gender and its role in the use of metadiscourse markers have received scant attention, future research should reveal more about possible gender-specific metadiscourse practices. In addition, this study was limited to a particular type of discourse, i.e. discussion, via an Internet-bases medium. Considering the possible impact of discourse type and communication medium, more research is needed to address the use of metadiscourse markers in various genres, such as narratives and conversational interactions used by male and females EFL learners.