The Use of Metadiscourse: An Analysis of Interactive and Interactional Markers in English Short Stories as a Type of Literary Genre

The present study investigated the frequency and type of metadiscourse markers in short stories as a kind of literary genre and how these markers are used by short story writers to produce persuasive texts. It is a pioneering study, since very few studies in the literature tackled literary genre and no study involved analyzing short stories. The corpus of 88,940 words consisted of 18 short story texts written by the three famous American authors Edgar Allan Poe, Mark Twain and Raymond Carver. To analyze this corpus, Hyland’s (2005) comprehensive model of metadiscourse was used. Results of the study indicated that metadiscourse markers are employed by short story writers to produce coherent texts and to make their stories persuasive. These results agreed with those of previous studies that involved literary texts indicating that metadiscourse markers are used frequently in such texts. The study findings proved that short stories are considered as persuasive texts not only due to non-linguistic factors, such as transportation, but also due to a linguistic one, namely, the use of metadiscourse markers. This finding is the most significant one, since it refutes the opinion that short stories are persuasive texts solely due to transportation and other similar factors.


INTRODUCTION
Metadiscourse is a cover term for words that writers or speakers use for the purpose of indicating the direction and purpose of a text. The term was first used by Harris in 1959 to refer to a way of understanding language in use. In this sense, metadiscourse represents the attempts of the writer or speaker to guide the perception of the text receiver (Hyland, 2005: 3). In other words, metadiscourse enables the writer to "guide, direct and inform" the reader in the way he (the writer or speaker) wants the reader or hearer to respond to the text content. Consequently, it is an important category not only for creating a text, but also for reading it (Crismore, 1989: 64). Adel (2006: 2) argues that metadiscourse refers to "linguistic items that indicate the existence in the text of either the writer and reader (or speaker and hearer) through reference to the text organization or by using other ways to comment on the text. Other writers broadened the concept of metadiscourse through the inclusion of various discoursal features like hedges, connectives and various forms of text commentary to indicate how writers and speakers influence their interlocutor's perception of their texts through their intrusion into them (cf. Williams, 1981;Vande Kopple, 1985 andCrismore, 1989). The most comprehensive definition of metadiscourse is that of Hyland (2005: 14) according to which metadiscourse is The Use of Metadiscourse: An Analysis of Interactive and Interactional Markers in English Short Stories as a Type of Literary Genre 67 ic writing, medical writing, advertising, etc.). However, not much attention has been given to the investigation of the use of metadiscourse markers in literary genres and the role of metadiscourse markers in persuading the readers of literary texts, if any, a task which the present study has set itself to perform through investigating the use of metadiscourse markers in English short stories as a kind of literary text. This is interesting, since studies have proved that persuasion in fictional narratives is achieved through transportation which is based on the idea that when readers read fictional narratives, they are transformed into that world and they start a mental journey into it and that the degree of transportation depends on a number of factors (cf. Persuasion through Fictional Narratives below). However, if it is proved that metadiscourse markers are used by writers of short stories, then it can be argued that persuasion in fictional narratives is achieved not only by transformation, but also through the use of metadiscourse markers. As a genre, the short story exhibits a number of recognizable features or characteristics which are central to it, and which are used in different ways and with different variables by each author and each age (Pasco, 1991: 407).
Concerning the origin of the short story, Pasco (1991: 408) states that we cannot be sure about any place of birth or time, but we all agree that its source goes back to the earliest days of civilization. In other words, the writing of short stories has started rather early.
A short story is "a short, literary prose fiction" (Pasco, 1991: 411). The word fiction is a cover term for both narrative and descriptive stories, and a key feature in stories is that they should be done artistically (Pasco, 1991: 413). In other words, the creation of short stories should be "artistically fashioned, with the apparent intension of making something beautiful" (Pasco, 1991: 414). Other features of a short story are that it is single, general and should communicate a world (Pasco, 1991: 418-420).

Objectives and Questions of the Study
This study aims at: 1. Investigating the types of metadiscourse markers used in the analyzed short stories as a kind of literary genre 2. Identifying the frequency of each type of markers in the corpus 3. Discovering how short story writers manage to make their stories persuasive 4. Relating the results of this study to those of other studies which involved literary and other texts In other words, the present study tries to find answers to the following questions: 1. What are the types of metadiscourse markers used in the short stories analyzed? 2. What is the frequency of each type of markers used? 3. How do short story writers manage to make their stories persuasive? 4. How are the results of this study related to those of other studies which involved literary and other texts?

Approaches to Metadiscourse
The notion of metadiscourse has been tackled by many researchers. However, the approach to handling metadiscourse in these studies differs. Mauranen (1993) identifies two approaches to metadiscourse, namely, the 'integrative approach' and the 'non-integrative approach'. Using these terms, previous studies about metadiscourse can be divided into two groups. The first group involves studies that followed the integrative approach, whereas the second involves studies that followed the non-integrative approach. Studies that employed the integrative approach, or as Adel 2006 calls the 'broad approach', are listed in Table 1 below (as cited in Adel, 2006: 171). In this approach to metadiscourse, the emphasis is on the relationship between the addresser and the addressee rather than on the text itself.
On the other hand, in the non-integrative approach to metadiscourse, or as Adel (2006) calls the 'narrow approach', the emphasis is on the text itself. Thus, the main concern of this approach is the writer/speaker and not the relationship between the addresser and the addressee.
Studies that employed the narrow approach to metadiscourse are listed in Table 2 below (as cited in Adel, 2006: 178).

Mostafavi and tajalli's (2012) study
In this study, the two researchers analyzed literary and medical texts to investigate whether there are any significant differences between those types of texts as far as the frequency and type of metadiscourse markers are concerned. The researches selected 30 literary and medical journal articles and analyzed 90 paragraphs, 45 from each type. The model used for analyzing the data was Vande Kopple's (1985) taxonomy. The analysis yielded that literary and medical texts differ significantly in the amount and type of metadiscourse markers with literary texts using those markers more frequently than medical ones. In addition, the study proved that, in both types of texts, the textual functions of metadiscourse were more frequent than the interpersonal ones.

Sadeghi and esmaili's (2012) study
This study investigated the use of metadiscourse features in two original novels and their simplified counterparts with the aim of comparing their frequency. The two researchers stated that they couldn't find any literature related to the analysis of literary genres in terms of metadisourse resources (Sadeghi and Esmaili's, 2012: 652). The two novels chosen were 'Wuthering Heights' and 'Tess of the D'Urbervilles'. The researchers adopted Hyland and Tse's (2004) model to analyze the corpus. They calculated the frequency of metadiscourse resources per 1000 words and used the technique of Chi-Square to check the difference in the use of these resources between the original novels and their simplified versions. The analysis yielded that there was no significant   difference between the original and the simplified novels as far as the frequency of metadiscourse resources is concerned implying that writers of both texts, the original and the simplified, did their best to present coherent texts.

Ahangari and kazemi's (2014) study
In this study, the two researchers investigated the use of metadiscourse elements in the novel 'Alice in Wonderland' to check how Lewis Carroll made use of Hyland's (2005) metadiscourse markers to produce a persuasive and impressive story. The analysis revealed that there were no meaningful differences between the frequencies of interactive and interactional metadiscourse markers. However, the study proved to have pedagogical implications for teaching English literature. It was found out that all types of metadiscourse items are used more frequently in literary texts than in other genres (Ahangari and Kazemi, 2014: 17).

PERSUASION
Although Miller (1980 as cited in Halmari andVirtanen, 2005:3) believes that all language use can be considered as persuasive, Halmari and Virtanen (2005:3) restrict the meaning of persuasion to "all linguistic behavior that attempts to either change the thinking or behavior of an audience, or to strengthen its beliefs, should the audience already agree". The two authors believe that the audience whether being visible or invisible, actual or implied interlocutors or onlookers play an important role in the process of persuasion.
They adopt the view that this process of persuasion is influenced by the situational and sociocultural context in which it occurs and which it helps to formulate in important ways (2005: 3-4). Halamari and Virtanen (2005: 4) argue that although genres appear, continue, change and fade through time, since they are connected with time and culture (Swales, 1990: 34-37 as cited in Halamari andVirtanen, 2005), the study of persuasion will always be meaningful, because it is a basic part of human interaction, and that learning more about persuasion means learning more about human nature.
Persuasion and genre are related to each other through their communicative purpose in that "genres can be more or less persuasive; persuasion, a communicative purpose, finds its realization through various genres", (Halamari and Virtanen, 2005: 11).

Persuasion through Fictional Narratives
Persuasion through fiction is achieved through the inclusion of fact-related information in fictional narratives. Such information may result in changing the readers' real-world beliefs significantly. According to the models of persuasion through fiction, (e.g. Gerrig, 1993;Green and Borck, 2000 as cited in Appel and Richter, 2007) these effects are the result of the readers being psychologically transported into the world of fictional narrative (Appel and Richter, 2007: 2). Such models also suggest that the persuasive effects of fictional narratives are constant and tend to increase over time.
Thus, persuasion through fictional narratives may have a sleeper effect (Hovland, Lumsdaine and Sheffield, 1949 as cited in Appel and Richter, 2007).
In literature, there is a distinction between fictional and non-fictional texts. Fictional texts are not supposed to contain information about the real world. However, they provide readers with information that could be applied to the real world (cf. Eco, 1994 as cited in Appel and Richter, 2007: 113). A number of studies were able to prove that the fictional narrative does have short-term persuasive effects (see Green, Garst and Brock, 2004 as cited in Appel and Richter, 2007: 114). In their study, Appel and Richter investigated whether or not fictional narratives have long-term effects through altering the readers' beliefs in the long-term. The two authors believe that if the sleeper effect is proved by experiments, this would make the fictional narrative as an effective means of changing our view of the world, and it will prove that the fictional narrative has a more persuasive effect than the non-fictional narrative which proved to have short-lived persuasive effect that decrease quickly by time (see Cook and Flay, 1978;Pratkanis, Greenwald, Leippe and Baumgardner, 1988 as cited in Appel and Richter, 2007).
In their nature, fictional narratives are different from political speeches, advertisements and editorials which continued to be and are still the focus of persuasion research. In them, persuasion is achieved through the use of arguments that are used to persuade the recipient that the factual claims are true, or that the political and ethical claims are appropriate. Fictional narratives, on the other hand, are stories about imaginary characters in an imaginary world. They do not contain arguments and are usually written for the purpose of entertainment rather than persuasion (cf. Bryant and Miron, 2002 as cited in Appel and Richter, 2007: 116). Fictional narrative usually contain a plot and a number of elements (e.g. setting, event, attempt, reaction and consequence, cf. Rumelhart, 1975 as cited in Appel and Richter, 2007).
It is argued by some researchers (Gerrig and Prentice, 1991;Prentice et al., 1997 as cited in Appel andRichter, 2007: 117) that fictional narratives may have a strong influence on the readers' beliefs, and that this influence is the result of some kinds of processes which characterize this type of genre (cf. Green, 2004;Green and Brock, 2000;Slater and Rouner 2002 as cited in Appel and Richter, 2007).
Persuasion in fictional narrative is viewed in terms of transportation which is based on the idea that when readers read fictional narrative, they are transformed into the fictional world of the narrative, and they begin a mental journey into that world. The degree of transportation depends on the degree to which mental images of the events which are described in a text are included in the narrative which, in turn, is affected by a number of factors. The first factor, reading goals, is related to the purpose behind reading a text. The second and third factors have to do with the relation of the readers to that text, i.e. their familiarity and involvement with the events being described in it. The fourth factor is also related to the readers, namely, their imaginary skills and the fifth is related to the narrative, its quality and typicality (for more information about these factors, see Green, 2004; Green and Brock, 2000 as cited in Appel and Richter, 2007: 117).
Thus, it is clear from the above discussion that persuasion in fictional narratives is achieved through what is called transformation. In other words, fictional narratives may have strong as well as continuous persuasive effects and that these effects increase overtime leading to an absolute sleeper effect (Hovland et al., 1949 as cited in Appel and Richter, 2007).
The important question concerning persuasion in fictional narratives has to do with whether the influence of fiction on the recipients' beliefs continues and increase overtime or it declines with time. Well-known models of persuasion, namely, the Elaboration Likelihood Model (ELM, Petty and Cacioppo, 1986;Petty and Wegener, 1999 as cited in Appel and Richter, 2007: 115) and the Heuristic-Systematic Model (HSM), Chaiken, Liberman and Eagly, 1989;Chen and Chaiken, 1999 as cited in Appel and Richter, 2007) suggest that the effect of persuasion through fiction tends to decrease overtime. Those models are based on the idea that for the effect of the persuasive message to continue, it has to be processed in an elaborative or systematic way, and that when the recipient thinks about the content of the message, this will help him/her to form other more beliefs which will function later to support the new beliefs acquired from the message.
The results of Appel and Richter's study proved that readers of fictional narratives are mentally transported into the fictional world of the narrative. Thus, their results are not in line with the ELM and HSM models of persuasion. It was proved that persuasion through narratives depends on processes other than those involved in persuasion through rhetoric texts.

The Appeals of Persuasion
The success or failure of persuasion, according to Aristotle (as cited in Garsten, 2006: 131, andin Larson, 2010: 20) depends on, three types of artistic and inartistic proof. The three appeals of persuasion are the rational (logos), the affective (pathos) and the credible (ethos). Persuaders make use of logical or rational appeals (logos). In this respect, listeners would come to believe something, because they followed and accepted an argument. Persuaders also use emotional appeals (pathos), and in this sense, listeners would come to believe something, because they were moved by an emotion. In addition, persuasion depends on the speaker's credibility (ethos). In this sense, listeners would come to believe something, because they trusted the judgement and goodwill of the speaker (Larson, 2010: 20).
Writers or speakers use various strategies and language devices to persuade their readers. One of those strategies is metadiscourse. According to Hyland (2005: 63), metadiscourse is a rhetorical strategy through which persuasion can be achieved. He (2005: 63) believes that "metadiscourse contributes to the rational, credible and affective appeals which have characterized persuasive discourse since the time of ancient Greece". When investigating the metadiscourse markers found in the genre of company annual reports, Hyland (2005) identified transition, frame markers, endophoric markers and code glosses as the metadiscourse markers of the rational appeal, engagement markers, attitude markers, hedges and pronoun references as the metadiscourse markers of the affective appeal and hedges, boosters, engagement markers and evidentials as the metadiscourse markers of the credibility appeal.
The present research follows the integrative approach to metadiscourse. This is so, since this approach does not only focus on the text organization, but rather takes into consideration the relationship between speaker/writer and listener/ reader. According to this approach, texts indicate the communicative interaction between speakers/writers and listeners/readers. Usually, people, through their speech or writing, want to negotiate some scenarios, and want to influence their listeners or readers. These matters fall within the scope of this study which aims at investigating the persuasive role of metadiscourse markers in fictional narratives.
Thus, the present research will follow Hyland's (2005) model of metadiscourse, since it is the most comprehensive model, and since it considers metadiscourse as interpersonal rather than separating the textual and the interpersonal functions as is the case in much of the metadiscourse literature. This model also takes into consideration the addressors' knowledge, experiences and needs revealing an addressor's understanding of an audience and their expectations. The model is based on the notion of interaction which is based on the link between text and context. Texts usually involve interactions between the addressor and the addressees. The text is viewed as an engagement that is both social and communicative. It provides us with a way of understanding how we present ourselves in a text to succeed in accomplishing our intensions (Hyland, 2005: 14).

Hyland's (2005) Classification
According to Hyland (2005), metadiscourse markers fall into two main categories: 'interactive' and 'interactional'. Interactive metadiscourse performs the function of guiding the addressee through the text (2005: 49). It indicates that the writer is aware of the presence of the audience and how he/she tries to fulfills its 'interests', 'rhetorical expectations' and 'processing abilities'. It also indicates that the addressees expect that the argument will follow conventional text patterns in order for them to perceive the text as 'appropriate ' and 'convincing' (2005: 54). Metadiscourse interactive markers are realized as 'transitions', 'frame markers', 'evidentials', 'endophoric markers' and 'code glosses'. Interactional metadiscourse, on the other hand, is related to how addressers manage interaction. It is considered as the writ- The Use of Metadiscourse: An Analysis of Interactive and Interactional Markers in English Short Stories as a Type of Literary Genre 71 er's personality or 'voice'. The addressers involve the addressees in the text through their arguments, they influence them, direct their attention at something and lead them to interpretations (Hyland, 2005: 52). Since they involve the addressees in the argument, these markers are considered as evaluative and engaging (2005: 54). Interactional markers include 'hedges', 'boosters', 'engagement markers', 'attitude markers' and 'self-mention'. The following Table 3 shows Hyland's (2005) model of Metadiscourse:

Materials
The data for this research consists of 18 short stories written by three famous and influential American short story writers who belong to successive periods of time extending from 1809 till 1988, namely, Edgar Allan Poe (1809-1849), Mark Twain (1835Twain ( -1910 and Raymond Carver . For each writer, six short stories of different lengths were analyzed. The length of the analyzed stories ranges from 501 words to 18,170 as shown in Table 4

IJALEL 8(3):66-77
The choice of these short stories is based on a number of criteria the first of which is that they are written by three of the most famous short story writers, the second is that these texts are among the most popular ones, and the third is that they are available online.

Procedures of data analysis
This study aims at identifying the metadiscourse markers used in short stories as a type of literary genre. To achieve this, the data was analyzed using Hyland's (2005) taxonomy to identify the frequency and type of metadiscourse markers to give an overview of the situation of those markers in the literary genre of short stories. The distribution of each category in the corpus will be presented, and their communicative functions will be discussed. To achieve the aim of the study, the whole eighteen texts will be considered as constituting the corpus of the study which consists of 88,940 words.

The Overall Frequency of Metadiscourse Markers
As was mentioned earlier, the total number of texts analyzed is eighteen short stories written by three of the most wellknown American short story writers (cf. Table 4

above).
The analysis revealed that the total number of metadiscourse markers in the corpus is 6,749 as shown in Table 5 below: The above Table provides the answers to questions No. 1 and No. 2 of the research concerning the types of metadiscourse markers used in the corpus and their frequency. As is clear from the table, in the corpus of 88, 940 words the total number of metadiscourse markers is 6,749. This indicates that the percentage of metadiscourse markers in the corpus is 7.5% which means that for every 100 words in the corpus, 7 words are used with a metadiscursive function and for every 1000 words, 75 words are used metadiscoursally.
As far as the distribution of interactive and interactional metadiscourse markers across the analyzed texts is concerned, it was found out that the interactional markers are used a little more than the interactive ones with 3,463 instances (51.32%) and 3,286 instances (48.68%) respectively.
To find out an answer to question No. 4 of the research which is about how the results of the present study are related to those of other studies which involved literary and other texts, an investigation has been carried out of such studies. This investigation revealed that previous research on metadiscourse markers proved that in written discourses, writers rely more on interactive metadiscourse markers (cf. Tse, 2004 andKhedri, 2014), whereas in spoken dis-courses, speakers rely more on interactional metadiscouse markers (cf. Rui and Xin, 2009;Cavalieri, 2011;Adel, 2012;Yipei and Lingling, 2013;Malmstrom, 2014;Lee and Subtirelu, 2015;Mahmood and Kasim, 2019). When the results of this study are compared to those of previous research, we find that they are a little bit different although the difference is not that significant. Although this study involved analyzing written texts, the frequency and percentage of interactional markers identified are a little higher than those of the interactive markers. This is due to the difference in the type of texts and genre (cf. Adel, 2012: 93). In the previous research which involved written texts, the texts analyzed were academic texts which are different from literary texts. Hyland and Tse's (2004) study investigated metadiscourse markers in postgraduate dissertations and Khedri's (2014) study involved the analysis of experimental research articles. Literary texts are different from academic texts in that in literary texts not all of the interactive metadiscourse markers are used. For example, Evidentials and Endophoric Markers which are two of the interactive markers are characteristics of academic discourse. To cite an example, in Hyland and Tse's (2004) study, out of the 239.8 instances of interactive markers 64.1 were instances of Evidentials and 23.4 instances were Endophoric Markers. It is worth mentioning that in the present study, 0 instances of these two markers were identified in the corpus and this justifies why the frequency of interactive markers is less than the frequency of interactional markers (cf. Interactive Markers below). This justifies why the results of this study which involved written texts are a little bit different from those of other studies which also involved written texts as far as the distribution of interactive and interactional markers is concerned. The reason behind this is that the types of written texts involved are different and thus, the frequency of markers relied upon by writers are different. This fact is supported by the results of Yazdani and Salehi (2016) study which involved investigating the use of metadiscourse markers in English and Persian online headlines. It was found out that interactional markers are used more frequently than interactive ones due to the same reason mentioned above.
It is worth mentioning that when the results of this study are compared to previous research which tackled literary discourse, they are similar in one way or another. Three studies analyzed literary discourse, the first of which is that of Mostafavi and Tajalli (2012) which analyzed metadiscoursal markers in medical and literary texts by using Vande Kopple's taxonomy (1985). The second is that of Sadeghi and Esmaili (2012) which analyzed two novels "Whuthering Heights" and " Tess of the D'Urberbilles" in their original forms and their simplified versions by using Hyland and Tse's (2004) model to see whether there are significant differences between the original novels and their simplified versions concerning the use of metadiscourse resources. In the first study, it was found out that there was a significant difference in the amount and type of metadiscoursal markers in medical and literary texts, and that in both types of texts the textual functions of metadiscourse were more frequent than the interpersonal ones. In the second, it was found out that there was no significant difference concerning the use of metadiscourse markers in the original and simplified versions of the novels analyzed. In these two studies, although the models used to analyze the data are different from the one used in the current study which makes it difficult to compare their results with those of the current one, their results are similar to those of the present study in that metadiscourse markers are employed frequently by writers of literary texts. One point of difference between Mostafavi and Tajalli (2012)'s study and the present study is that, in their study textual metadiscourse functions were used more frequently than the interpersonal ones, whereas in the present study interactional metadiscourse markers were used a little more than the interactive ones. This difference could be attributed to the difference in the models used for analysis in the two studies. The only study that involved analyzing a literary text, namely, the novel "Alice in Wonderland" using the same model used in the current study which is that of Hyland (2005) is Ahangari and Kazemi's (2014) study. According to this study, the frequency mean for interactive metadiscourse is 407.75 and that for interactional metadiscourse is 579.25 which means that the writer relies more on interactional metadiscourse markers than on interactive metadiscourse markers. The researchers applied the T-test on the means of metadiscourse markers identified, and the results proved that there was no meaningful differences between those two types of markers. It is interesting to note that these results are similar to those identified in the current research in that here also interactional markers are employed more than interactive markers and that the difference between them is not that important. Thus, this indicates that short story writers manage to make their stories persuasive through the use of metadiscourse markers and that short stories as literary texts are persuasive due to linguistic factors (metadiscourse markers) and not only non-linguistic factors (transportation). This provides an answer to question No. 3 of the research which deals with the way by which short story writers manage to make their stories persuasive.

Interactive Markers
As was mentioned earlier and displayed in Table 5 above, the number of instances of interactive markers is 3,286. This number is divided unevenly among three of the interactive markers, namely, Transitions, Frame Markers and Code Glosses as shown in Table 6 below: It is evident from the Table above that the most frequently used marker of the interactive ones is Transitions which occurred 2,985 times in the corpus forming 88.10% of the total number of interactive markers. It is worth mentioning that in almost all the metadiscourse studies whether involving spoken or written texts, Transitions is the highest in terms of occurrence among the other interactive markers. This also applies to the studies which analyzed literary texts, namely, those of Ahangari and Kazemi (2014) and Mostafavi and Tajalli (2012). This could be due to the fact that by relying heavily on Transitions, writers of literary texts want their texts to be cohesive and their readers to understand the links between ideas. Also, through the use of transitions which perform the function of connecting steps in an argument, writers of short stories manage to persuade their readers through the rational appeal (logos).
Transition markers are divided into Addition, Comparison and Consequence in terms of function. The most frequently used subcategory of Transition Markers is that of Addition realized mostly through the use of "and". The second subcategory in terms of frequency is that of Comparison realized mainly through the use of "but", and the least subcategory is that of Consequence realized through the use of "thus".
The following are examples of Transitions used in the data: 1. I close my eyes and hold on to the sink. 2. I at once offered to purchase it of the landlord; but this person made no claim to it. 3. By experiment I know that wood swims, and dry leaves, and feathers, and plenty of other things; therefore by all that cumulative evidence you know that a rock will swim;… The second most frequently used marker of the interactive ones as Table 6 above shows is Frame Markers. Those markers occur 381 times in the corpus forming 11.60% of the total use of interactive markers. The function of those markers is to sequence arguments in the text, but not events in time. This is why they are used by the short story writers, since they contribute to making the discourse clear to readers and to persuade readers through the rational appeal (logos). The Frame Marker "then" is the most frequently used one among the other markers followed by "first". Other Frame Markers that appeared in the corpus are: "next", "now" and "finally".
The following are examples from the data: 1. First I drew a box that looked like a hose. It could have been the house I lived in… 2. Finally, I hit upon what I considered a far better expedient than either of these. The least frequently used interactive marker is Code Glosses. They occurred as is shown in Table 6 above only 10 times in the data forming 0.30% of the total use of interactive markers. They function to make the reader understand the writer's intended meaning through providing additional information by means of explanation, elaboration and rephrasing. In addition, through their use, writers manage to persuade their readers by means of the rational appeal (logos) through which readers are made to believe something by following and accepting an argument. "For example" is the highest in terms of occurrence in the corpus followed by "in other words", and finally comes "namely".
The following are examples from the data: 1. A single grain of gimlet-dust, for example, would have been obvious as an apple. 2. For example, an errant simpleton is his opponent, and holding up his closed hand… 3. You will now understand what I meant in suggesting that, had the purloined letter been hidden anywhere within the limits of the Perfect's examination-in other words, had the principle of its concealment been comprehended with the principles of the Perfect-its discovery would have been a matter altogether beyond question. It is worth mentioning that no instances of Endophoric Markers and Evidentials were identified in the corpus. The function of Endophoric Markers is to refer to previous or subsequent parts of the text with the aim of supporting the reader's comprehension of a text, whereas the function of Evidentials is to cite ideas from another source which could be a reference to a reliable source in some genres. In academic writing, they function to support arguments through reference to a related literature (Hyland, 2005). Thus, it is obvious that Endophoric Markers and Evidentials are mainly characteristics of academic discourse and other types of genres. However, this does not mean that they do not function in literary genre. For example, in Ahangari and Kazemi's (2014) study which involved analyzing a novel only two instances of Endophoric Markers were identified and there were no instances of Evidentials. Thus, they do sometimes exist in literary texts, but in a very limited way as is the case in novels.

Interactional Markers
As was mentioned earlier and displayed in Table 5, interactional markers were used 3,463 times in the data forming 51.32% of the total number of metadiscourse markers. As shown in Table 7 below, Self-mention is the most frequently used interactional marker with 2,176 instances forming 63% of the total number of those markers achieved mainly through the use of (I, me, we and mine). Self-mention indicates the presence of the authors in their texts through the use of first person pronouns and possessive adjectives to express their attitudes towards their arguments, community and readers (Hyland, 2005).Through this, writers manage to create persuasion in their texts by stimulating an ethos. In this sense, persuasion is achieved through the writer's credibility and readers are convinced in something, since they trust the writer's judgement and goodwill. Writers of short stories either tell the story themselves or through one of the characters in the story which is usually the main character. Thus, the writer expresses his attitude towards somebody or something either directly being the storyteller or indirectly through one of the characters.
The following are examples from the data: 1. My husband eats with a good appetite. But I don't think he's really hungry. 2. We were five miles from the car. 3. We continued our route in search of the Amontillado.
The second most frequently used interactional marker is Hedges. The total number of Hedges instances in the data as the above table shows is 907 forming 26% of the total use of interactional markers. Writers employ expressions such as the modals (may/might, can/could and would) to express opinions rather than facts and indicate that a statement is based on their reasoning and not on a certain knowledge. In addition to the modals, Epistemic verbs such as (seem and suppose) and Epistemic adverbs such as (probably and whatever) also appeared in the corpus. The use of Hedges can achieve persuasive purposes, since, through their use, writers are trying to influence their readers to convince them to adopt their points of view or opinions or the way they think or behave (cf. Halmari and Vritanen, 2005).
The following are examples from the data: 1. He was gone to Brixton, and might not return before morning. 2. Perhaps you will be good enough to explain to the house why you rise. 3. "Luchresi cannot tell Amontillado from Sherry." The third interactional marker in terms of frequency, as shown in Table 7 above, is Boosters. They occur 206 times in the corpus forming 6% of the total number of interactional markers. Boosters are used by writers in order to reflect their certainty about what they say and they, together with Hedges, indicate the writers' commitment to their texts and their respect to their readers. Instances of (clearly, demonstrate, believe and find) have been identified in the corpus in addition to many other expressions which indicate certainty. Boosters like Hedges are used by writers in order to achieve persuasive purposes through using assured declarations. In this case, persuasion is achieved through the credibility appeal (ethos).
The following are examples from the data: 1. I can only demonstrate that one or the other is a fake, and let a person take his choice. 2. "It is clearly inferred," replied the Perfect. 3. These orders were sufficient, I well knew, to ensure their immediate disappearance, one and all, as soon as my back was turned. Engagement Markers occurred 171 times in the corpus forming 4.9% of the total number of interactional markers. Writers use those markers for the purpose of engaging their readers in the text through talking to them explicitly by the use of pronouns such as (you, we, our, remember, etc.), questions and directives. The most frequently used of the Engagement Markers in the data are directives, namely, obligation modals (should, must and have to).
The following are examples from the data: 1. If he were plain, I should love him. 2. "Please," I say, "I have to go." 3. "I--I don't think It would have done for you to --to --One mustn't --er --public opinion --one has to be so careful --so" Attitude Markers are the least in terms of frequency in that, as Table 7 above shows, they appeared only three times in the corpus forming 0.09% of the total number of interactional markers. Through the use of those markers, writers express surprise, importance, agreement, obligation, etc. rather than being concerned with whether the information is reliable, true or related. They can manage to persuade their readers by the use of those markers through influencing the readers' emotions by means of the emotional appeal (pathos).They are realized by attitude verbs (prefer), sentence adverbs (unfortunately) and adjectives (zero instances).
The following are examples from the data: 1. If you prefer to conduct the inquiry privately, do so. 2. "But if you prefer a public inquiry, then publish this present writing in the local paper…" 3. These fellows are unfortunately a hidden meaning in the Antediluvians.

CONCLUSION
The present study aimed at investigating the frequency and distribution of metadiscourse markers in short stories as a kind of literary genre. Specifically, it aimed at identifying the frequency of each type of markers, discovering how short story writers manage to make their stories persuasive and relating its result to those of other studies which involve literary and other texts. Results of the analysis revealed that writers of short stories use such markers in their stories to produce cohesive and coherent texts. In addition, they employ these markers to make their stories persuasive, since metadiscourse markers form one of the strategies to produce persuasive discourse.
The slight difference in the frequency and percentage of the two types of metadiscourse markers, namely, the interactive and the interactional indicates that they are almost of equal importance in literary texts, and that this little difference is due to the type of text and genre. In addition, the difference in the distribution of the interactive and interactional markers in this study which involved literary texts when compared to previous studies which involved academic texts is due to the same reason mentioned above, since some of the interactive markers are more characteristics of academic rather than literary texts. On the other hand, the fact that the results of this study agree with those of previous studies that analyzed literary texts indicates that metadiscourse markers occur frequently in literary texts, and that both of the interactive and interactional markers play an important role in such texts.
Short story writers use interactive markers to produce coherent texts and to let their readers understand their intended meaning. They also manage, through the use of those markers, to persuade their readers through utilizing rational appeals (logos).
On the other hand, interactional markers are used by short story writers to express their attitudes, opinions and commitment to their texts. Also, they use them for the purpose of engaging their readers in the text. The ultimate goal behind this is to make their texts persuasive, and consequently to persuade their readers of their opinions and attitudes through the use of the credibility appeal (ethos) and the affective appeal (pathos).
The most important conclusion in this study is that persuasion in short stories is achieved through the use of metadiscourse markers which constitute one of the strategies used by persuaders to persuade their audience. This finding is very interesting, since previous studies about persuasion in fictional narratives attributed the persuasion effects of such texts to non-linguistic factors such as transportation, because, according to those studies, fictional narratives are different in their nature from political speeches, editorials and advertisements in which persuasion is achieved through the use of arguments (cf. Persuasion through Fictional Narratives above). Consequently, those studies proved that persuasion through narratives depends on processes other than those involved in persuasion through rhetoric texts.
However, throughout the results of the present study, one can conclude that persuasion in fictional narratives is attributed not only to non-linguistic factors such as transportation, but also to linguistic ones, namely, the use of metadiscourse markers by short stories writers.