Keywords

1 Introduction

Writing an academic text is more than just putting together one’s thoughts and applying the correct grammatical rules in the target language. In addition to using language to refer to the experimental world in an academic paper, one should be aware of how to use language to organize the text, guide the reader, and create an engaging and interactive piece of work. The latter applications of language fall within the scope of metadiscourse. Metadiscourse is a term which has been defined and used differently by scholars. Vande Kopple (1985) considers it as a non-propositional linguistic element which signals the presence of the writers in the text as they help their readers to “organize, classify, interpret, evaluate, and react” towards what is written about the subject matter (p. 83). According to this view, metadiscourse is mainly being used for organizational, interpretive, and evaluative purposes in a text. However, there are other researchers who consider metadiscourse as having a wider scope. Hyland (2005), for example, describes metadiscourse as “the self reflective expressions used to negotiate interactional meanings in a text, assisting the writer (or speaker) to express a viewpoint and engage with readers as members of a particular community” (p. 37). The viewpoint adopted by Hyland stretches the boundaries of metadiscourse significantly. It not only includes the organizational function of metadiscourse in a text, it also considers the interactive, commentary, and attitudinal functions of this linguistic device in academic discourse.

The different applications of metadiscouse in academic and non-academic discourse are discussed extensively in the literature These include rhetorical (e.g., establishing coherence and logic), social (e.g., interaction between the writer and reader and making bonds between participants), organizational (e.g., walking the reader through the text), and pedagogical (e.g., enhancing reading/listening comprehension and recall) functions (Chaudron & Richards, 1985; Crismore & Vande Kopple, 1997; Hyland, 2004; Mauranen, 1993; Vande Kopple, 1988). Despite the fact that metadiscourse has received a lot of attention in the past, it is not still fully explored (Ädel, 2006). Hyland (2017) has also emphasized that “conceptions of metadiscourse, and individual studies themselves, are more usefully seen as contributing different aspects to our understanding of discourse” (p. 19). More research is still needed to underpin different functions of metadiscourse, especially in less studied languages. This chapter aims at exploring the conventions of metadiscourse in Persian academic discourse. This paper, more specifically, investigates the distribution of forms and functions of metadiscourse across three academic disciplines (i.e., Sociology, Education, and Medicine) in Persian. To pursue this goal, we use Ädel’s (2006) reflexive model of metadiscourse which mainly considers “guidance” and “interaction” as the primary functions of metadiscourse. The current research is among the pioneering studies on Persian which explores the variation of metadiscourse in this language using a reflexive model. We hope that the findings of this paper will provide some guidelines for language educationists and language policy makers as to how improve academic Persian in order to communicate more effectively and establish its position in a larger academic community. The outline of this chapter is as follows. The second section introduces academic language in general. The different approaches to metadiscourse followed by the reflexive model used in this study are presented in the next two sections. The following section will review studies related to metadiscourse in Persian. The methodology used in this study makes the next section of this chapter. The results and findings followed by general discussion and conclusion will make up the last two parts of this paper.

2 Academic Language

The term academic language started to be used in the past 40 years when Cummins (1979) made a distinction between basic interpersonal communication skills (BICS ) and cognitive academic language proficiency (CALP) . He found that (English) language learners spend little time on acquiring basic conversational skills while a different register of the same language (i.e., academic) proved to be quite challenging and time consuming for the same language learners. He suggested that language proficiency is not a uniform construct. This means that a unidimensional model of general or global language proficiency cannot account for all aspects of language use or performance. According to him, academic language is what people do with language rather than grammatical features used in the text. Cummins stated that “the essential aspect of academic language proficiency is the ability to make complex meanings explicit in either oral or written modalities by means of language itself rather than by means of contextual or paralinguistic cues (e.g. gestures, intonation, etc.)” (Cummins, 2000, p. 69, emphasis in the original text). Cummins’ approach to academic language was not specifying the linguistic features used in this register . Therefore, it is hard to apply his view in language classrooms where educators need to know specific features of academic language before they start teaching it (See Ranney, 2012 for further discussion).

In more recent years, scholars have started to shift their attention from BICS /CALP distinction to academic language per se. The first motives behind this shift was the abundance of rhetorical features shared between social and academic registers (Schleppegrell, 2001). The earliest studies which focused on academic language mainly investigated the vocabulary of academic discourse. These studies simply focused on the occurrence of lexical forms in academic language aiming at providing an account of distributional frequencies of the lexical items in the target language. This was done at the cost of missing sentential and discoursal dimensions of academic language.

The second reason for academic language receiving a lot of attention was the advances made in the fields like composition studies, second language writing, and contrastive rhetoric. Contrastive rhetoric started by Kaplan (1966) and developed later as an approach to examine the discourse and rhetoric. Kaplan assumed that each language and culture has rhetorical patterns and categories which are unique to themselves. He suggested that the differences in writing could reflect cultural and educational trainings. Since that time, this approach has had a major influence on areas such as EFL /ESL and academic language teaching in university settings (Connor et al., 2008). One of the contributions of this approach to the field of academic writing is that it opened new research topics where academic discourse could be compared across different languages and disciplines. For example, the functional categories (e.g., hedges, boosters, attitude markers, and engagement markers) and different components (e.g., moves, steps) of different genres (e.g., research articles, lab reports, or grant proposals) have been the major topics of research in the last two decades or so (Connor & Mauranen, 1999; Myers, 1989). One of these rhetorical features which has been widely discussed and researched in the field of academic discourse is metadiscourse. The following section presents different approaches to this rhetorical device in academic language.

3 Different Approaches to Metadiscourse

Due to the fuzzy nature of metadiscourse, there is a wide spectrum of perspectives towards metadiscoursal studies.Footnote 1 This could range from a narrow text-centered view in one end to a broad interpersonal view in the other end (Hyland, 2017). The simplest approach to metadiscourse views it as metatext which includes discoursal expressions refereeing only to the internal structure of the text and its purpose (Mauranen, 1993). Sentence (1) provides an example:

(1) jɒftehɒje in motɒleʔe neʃɒn nædɒd ke tæd͡ʒvize tizɒnidin piʃ æz æmæl bɒʔese kɒheʃe dærd pæs æz æmæle septoplɒsti dær bimɒrɒn miʃævæd.

The findings of this study did not show that preoperative tizanidine administration reduces postoperative pain in septoplasty in patients. (MED_3, S131)

In sentence (1), the author is explicitly referring to the whole text by using the term this study. This illustrates using a metadiscourse marker by the writer as a signpost to guide the reader with the text. On the other end of the continuum, there are scholars who took an “integrative” approach where metadiscourse not only refers to guiding the readers throughout the text and its organization, but also it “involves the personalities, attitudes and assumptions of those who are communicating” (Hyland, 2005, p. 3). The integrative approach adopts Halliday’s three levels of linguistic (meta)-function in its model, namely the ideational, the interpersonal and the textual levels. In the Systemic Functional Grammar (SFG) proposed by Halliday (1973), metadiscoursal items have both interpersonal and textual functions. As for the interpersonal function, the writer makes himself/herself visible in the text through expressing his/her personal attitudes and feelings or starting a dialogic conversation with the reader. The textual function is fulfilled by providing landmarks and signposts throughout the text to organize the text and guide the reader. Ädel (2006) criticizes that the SFG-inspired model uses the original terminology used in Halliday’s SFG (i.e., “interpersonal” and “textual”) in a differentway which could be a source of confusion. Moreover, she states that contrary to the researcher’s expectation that consider the interpersonal and the textual functions as the “twin main functions” of metadiscourse, these two are not at the same level in the SFG-inspired model.Footnote 2 She takes a “reflexive” approach and develops a new model for metadiscourse which adds personal discourse functions into the model. The following section presents Ädel’s (2006) model which is also adopted in the current study.

4 Ädel’s Reflexive Model of Metadiscourse

There are competing models for metadiscourse in the field of applied linguistics. One of the reasons for such a diversity is that the existing approaches draw on different linguistic theories to develop their models. In contrast to integrative approach discussed above, the reflexive model initially started by Mauranen (1992, 1993) and further developed by Ädel (2006) is mainly based on Jakobson’s (1998) three functions of language: the expressive , the directive, and the metalinguistic. The corresponding component of these language functions in the speech event are the writer, the reader and the text/code. According to Ädel, the reflexive metadiscourse includes at least the following three aspects: (1) how scholarly writers refer to themselves, (2) how they relate and speak to their readers, and (3) how they refer to their own texts. As for the first aspect, research has shown that scientific disciplines vary from each other in terms of how authors use first person singular I or exclusive first person plural we to refer to themselves. There are some fields in English which favor using self-reference to refer to the author of the paper while there are other fields which mainly stick to impersonal style (Hyland, 2005). In addition to disciplines, there are some studies which have shown different tendencies of languages (e.g., English, Finnish, Spanish) for using expressions referring to the author (Mauranen, 1993; Salas, 2015; Williams, 2012).

The second aspect is related to creating a dialogue and establishing relationship with the readership. This could be performed either through using directives or inclusive pronoun we. Similar to exclusive pronouns, research has shown that the extent and functions of inclusive we vary both across disciplines and languages (Harwood, 2005; Taki & Jafarpour, 2012). The last aspect of reflexive metadiscourse refers to the textual features or metatext which talk about the text itself. In fact, this is the most basic function of metadiscourse which includes items in discourse which refer to the internal structure of the text, its organization, and purpose.

In the reflexive model of Ädel, metadiscourse is all interpersonal and divided into two main categories: “metatext” and “writer-reader interaction”. Metatext is “described as metadiscourse that guides the reader through the text or comments on the use of language in the text. ...‘Writer-reader interaction’, on the other hand, is described as metadiscourse that is used by the current writer to interact with her imagined reader in ways that create and maintain a relationship with the reader” (Ädel, 2006, pp. 183–184). Both “metatext” and “writer-reader interaction” are further divided into Personal and Impersonal categories. Figure 1 below presents Personal and Impersonal configurations of ‘metatext’ and ‘writer-reader interaction’ in Ädel’s (2006) reflexive model.

Fig. 1
figure 1

Reflexive model of metadiscourse. (Adapted from Ädel, 2006, p. 38)

Ädel’s (2006) reflexive model extends the concept of metadiscourse from the text to the writer of the text and its imagined reader. She argues that the reflexive model as a functional model exhibits more consistencies and is more precise compared to the reflective model. One of the main advantages of this model is that it includes the writer and reader in their contextualized roles as writer and reader. Ädel (2006: 182) emphasizes that “by including both the writer and the reader, we can draw a distinction between primarily writer-oriented and primarily reader-oriented material”. The other advantage of reflexive metadiscourse model is establishing criteria for identifying metadiscourse units. These include explicitness or self-awareness of text, contextuality, current text, and writer/reader qua writer/reader. Ädel’s non-integrative approach allows a precise identification of micro-level discourse functions. This provides the researcher with a more accurate picture of the metadiscourse phenomenon compared to other broader perspectives which include stance and evaluation in their models. Toumi (2009) has made an attempt to modify Ädel’s (2006) model to render it more applicable to research article genre. He uses a different classification for reflexive metadiscourse categories by including two subcategories of high versus low explicit reflexivity in his model. These two subcategories still contain instances which are identical to the original model. Moreover, the second difference in Toumi’s model is that it does not “consider personality as a metadiscursive category rather it regards it as a characteristic of the metadiscourse unit” (p. 72). This means that if one of the elements in the unit is classified as personal, the whole unit is categorized under reflexive personality. The changes proposed by Toumi to the model are minor and not very substantial. In the current study, the original model developed by Ädel is adopted due to its wider application which allows a cross-studies comparison.

5 Academic Persian and Metadiscourse

The majority of studies investigating the metadiscourse strategies in Persian have mainly examined this linguistic device either cross-linguistically (e.g., Persian vs. English) or have explored its application by Persian native speakers using English as L2 and compared it with English native speakers across different disciplines (Abdi, 2009; Ariannejad et al., 2019; Falahati, 2004, 2007; Mozayan et al., 2018; Rahimpour & Faghih, 2009; Salar & Ghonsooly, 2016; Shokouhi & Baghsiahi, 2009; Taki & jafarpour, 2012; Zarei & Mansoori, 2011). In one of the earliest studies on this topic, Falahati (2004, 2007) investigated the distribution of forms and functions of hedging in academic research articles in Persian and English across three disciplines (i.e., psychology, chemistry, and medicine) to see how writers use this device differently across languages and fields. The findings of this study showed that the English writers use hedges almost 61% more than Persian writers. The English psychology and Persian medicine research articles were found to be the most heavily hedged disciplines. The results also showed that the discussion sections of research articles (RAs, henceforth), in general, favor more hedges than the introduction section. The author used both epistemological and interpersonal significance of hedging in academic discourse to account for the difference in the frequency of hedges across the two languages. Rahimpour and Faghih (2009), in another study, examined metadiscourse in the discussion section of ninety Persian and English research articles in applied linguistics. The English articles were written both by native and non-native speakers. They examined a subset of metadiscourse categories proposed by Hyland (2004) which included transitions, frame markers, endophoric markers, evidentials, code glosses, hedges, boosters, attitude markers, engagement markers, and self-mentions. The first five items in this list are interactive metadiscourse and the rest are classified as interactional metadiscourse. Their results showed that the authors in the two languages used interactive metadiscoursal factors significantly more than interactional ones. Moreover, English authors employed interactional metadiscourse more than Persian writers while frame markers and code glosses were used more by Persian native speakers. In another study, Ariannejad et al. (2019) investigated a number of interactional metadiscourse markers, namely hedges, boosters, and attitude markers in 100 research articles (50 in Persian and 50 in English) in the field of architecture. The general findings of their study showed that the English-language writers used more metadiscourse markers compared to Persian-language writers. The former group used hedges and boosters significantly more than Persian authors while attitude markers were used in Persian articles more than English articles. The different writing styles across the two groups is explained in terms of different nature of the two languages as being either writer-responsible or reader-responsible. They explained that the higher application of the markers and signposts in English articles is for guiding readers in the text and helping them understand the authors’ interpretations while readers in Persian, as a reader-responsible language, are expected to disclose the intended meanings of the author and discover the relationship between different units of the text which results in lower frequency of interactional metadiscoursal markers.

In a similar study, Jalilifar (2011) used Hyland’s (2004) model and investigated two subtypes of metadiscourse (i.e., hedges and boosters) in the discussion section across psychology and applied linguistics RAs in Persian and English as L1 and L2. The results showed that the authors used these two pragmatic devices differently in terms of their frequency, type and function across languages and disciplines. The English native writers used hedges almost two times more than Persian writers where the hedges used by the former group were mainly reader oriented. The boosters were reversely used more by Persian authors compared to English native authors. The two disciplines showed close correspondence in terms of using the two rhetorical devices due to both representing soft fields. In order to explain the existing differences across Persian and English rhetorical systems, the author states that “while in Persian writing, a reader-responsible language, writers use a less hedged discussion and readers are assumed to infer much from the text, English texts, writer responsible, allow more hedges in discussion and guide readers through the text” (p. 184). The reviewed literature shows that metadiscourse has received good amount of attention in Persian; however, these studies have mainly focused on this rhetorical feature across both English and Persian. Moreover, they have primarily applied a subset of Hyland’s (2004) metadiscourse model in their studies. To the best of our knowledge, there is no study using a reflexive model of metadiscourse to analyze research articles in Persian across distinct academic disciplines. To this end, this research uses a reflexive metadiscourse model to investigate the employment of metadiscursive markers for establishing a relationship between the writer, the reader, and the text across three academic disciplines in Persian (i.e., Sociology, Education, and Medicine). In the current study, we try to address the following three research questions:

  • Q1: What are the lexical and grammatical markers (i.e., forms) which signal the presence of metadiscourse in academic Persian discourse?

  • Q2: What are the functions of lexical and grammatical markers which signal the presence of metadiscourse in academic Persian discourse?

  • Q3: Are there any differences between the three disciplines (i.e., Sociology, Education, and Medicine) in terms of the frequency of metadiscursive markers and their functions?

In order to address the questions in the study, we used the methodology which is presented in the next section.

6 Methodology

6.1 Data Selection Criteria

The research articles used in this study come from three disciplines: Sociology, Education, and Medicine. This decision was made in order to make sure that the selected articles represent different disciplines across the academy. Becher’s (1989) classification was used for choosing the disciplines. According to this taxonomy, disciplines are divided into hard and soft fields. Hard fields include sciences and engineering while soft sciences include humanities and social sciences. After selecting the disciplines, the next step was to choose the journals from which articles were supposed to be selected for the analysis. A few experts in each field were consulted and were asked to nominate highly ranked journals in their disciplines. Moreover, we considered the rankings of the journals from which we selected the articles. These journals were mainly ranked as"علمی" scientific by the Iranian Ministry of Science, Research, and Technology (MSRT)Footnote 3 which is a top ranking for academic journals. Twelve articles were selected in each discipline, making 36 in total (12 articles * 3 disciplines = 36).

The articles for the analysis were chosen based on different criteria. First, only empirical papers with Swales’ (1990) Introduction, Method, Result and Discussion (IMRD) rhetorical sections were selected. In the current study, we only analyzed the metadiscursive expressions in the introduction and discussion sections of research articles. This is due to the fact that it is these two sections which are the most rhetorical parts in research articles (Hyland, 2000; Mauranen, 1993; Vassileva, 2001). In sociology articles, the introduction section was decided to be any parts appearing before the method section. In this field the introduction section is divided into subsections such as parts providing theoretical and empirical reviews of previous studies. All abstracts, footnotes, long quotations, endnotes, and reference lists in the RAs were deleted before analysis. In the current paper, no attempt is made to compare the metadiscourse markers across introduction and discussion sections.

The second criterion for selecting the articles was the date of publication. The articles used in the corpus were all limited to those published within the last ten years. It is assumed that time influences the style of the writers and we tried to take this variable into account (See Appendix A for the complete list of articles). Table 1 presents the total number of articles, words, as well as the mean number of words per articles across the three disciplines.

Table 1 Corpus description

6.2 Procedures

The main goal of this study is to identify and classify the linguistic units which act as metadiscoursal expressions. In order to follow this goal, the introduction and discussion sections of all research articles were read carefully and all the metadiscoursal expressions were identified, annotated and then registered both in the pdf files and an Excel file for quantitative and qualitative analyses. The reflexive model of metadiscourse proposed by Ädel (2006) was used in this study. This model divides metadiscourse into two main categories of “metatext” and “writer-reader interaction”. Metatext is divided further into four subcategories: text oriented, reader-oriented, writer-oriented, and participant-oriented. Text-oriented metadiscoursal expressions are further divided into four groups: reference to the text/code, phoric markers, discourse labels, and code glosses. Sentences (2)–(4) are provided as the representative for writer-oriented, participant-oriented, and text-oriented categories, respectively.

(2) dær jek tæhɢiɢi ke dærbɒreje æbʔɒde ed͡ʒtemɒʔije rævɒnʃenɒxtije kotulegi ænd͡ʒɒm ʃode bud jeki æz ʃerkætkonændehɒ tæd͡ʒrobei ke dær modæte d͡ʒostod͡ʒuje kɒr dɒʃte rɒ bɒzgu kærde ke dær ind͡ʒa eʃɒre mikonim.

In a study of the social-psychological dimensions of dwarfism, one participant recounted an experience he/she had while looking for a job, which we refer to here. (SOCIO_11, S641)

(3) eʔmɒle in ʃive tænhɒ zæmɒni movæd͡ʒæh væ moʔtæbær xɒhæd bud ke dær jek doreje zæmɒnije moæjæn bɒ goruhi æz ʃɒerɒn jɒ ædibɒn movɒd͡ʒeh bɒʃim ke ʃɒʔer jɒ nevisændei bozorg rɒ be ostɒdi jɒ be mænzæleje olguje ædæbije xod pæzirofte væ sonæte ædæbije monsæd͡ʒem væ tæɢribæn jekdæsti rɒ ʃekl dɒde bɒʃænd.

The implication of this method will be justified and valid only when in a certain period of time we encounter a group of poets or writers who have accepted a great poet or writer as a master or as their literary model and have formed a coherent and almost uniform literary tradition. (EDU_11, S755)

(4) dær edɒme, nætɒyed͡ʒe bærxi æz in pæʒuheʃhɒ rɒ be tore moxtæsær morur mikonim.

In the following, we review briefly the results of some of these studies. (EDU_12, S771)

After identifying both the form and function of the metadiscourse markers, the raw frequencies of the tokens representing them in the three academic disciplines were counted separately. Since the number of words was not evenly distributed in the three sub-corpora, we also calculated the relative frequencies of metadiscourse markers per 1000 words. The fact that metadiscourse is a pragmatic category means that the same item could function as metadiscourse or not. In order to ensure that the tokens were coded reliably, all items were read and examined in their sentential contexts to make sure that they are functioning as metadiscourse. The second author of this paper coded all the tokens in this study. The challenging units (almost 5% of the total tokens) were highlighted in an excel file and were examined further by the first author of this paper later. In order to determine the number of metadiscourse markers/units in our corpus, we followed Ädel’s (2006) method. This included counting the smallest linguistic units which signaled the presence of metadiscourse. Each grammatical sentence could contain more than one metadiscourse marker. Sentence (5), for example, contains two tokens each representing specific subcategory of text-oriented metadiscourse (i.e., Discourse Label and Reference to Text).

(5) hædæf æz in pæʒuheʃ tæhlile ʃekɒfe kejfiæte xædæmɒte ɒmuzeʃe mæd͡ʒɒzi væ hozuri æz didgɒhe dɒneʃd͡ʒujɒn bud.

The purpose of this study was to analyze the gap between the quality of virtual and face-to-face education services from the students’ perspective. (EDU_1, S157)

The following section presents the results and findings of the study.

7 Results and Findings

In this section, the results of lexical and grammatical markers (i.e., forms) which signal the presence of metadiscourse in academic Persian are presented along with their functions. These results are given across the three disciplines (i.e., Sociology, Education, and Medicine) in order to highlight the differences across academic fields. In most of the tables, the raw frequency and relative frequency (i.e., frequency per 1000 words) are presented together. This is because the size of corpora across the three disciplines is different. Moreover, the relative frequency allows one to have cross-studies comparison. Table 2 shows the total distribution of metadiscourse markers across the three disciplines. According to this table, the relative frequencies of metadiscourse markers in Sociology and Education RAs are 10.7 (n = 482) and 10 (n = 384), respectively. The rate of application of metadiscourse markers in Medicine RAs is 7.7 (n = 135). This result shows that the number of metadiscursive devices used by sociologists and educationists similarly is greater than medical specialists. Such a pattern could be explained by considering the nature of both education and sociology disciplines as soft sciences. Salas (2015) in her study on research articles in Spanish has reported that the total frequencies of metadiscourse markers in linguistics, economics, and medicine RAs are 11, 7.71, and 7.75, respectively. The results of our study remarkably mirror the ones presented by Salas once we divide the disciplines based on their soft or hard nature. Hyland (1998) has also reported that the density of metadiscourse in marketing articles is 20% more than biology, astrophysics, and applied linguistics.

Table 2 Raw and relative frequency of metadiscourse markers across the three disciplines

Table 3 presents the raw frequency and relative frequency (per 1000 words) of Personal and Impersonal metadiscourse across the three disciplines. The density of Personal metadiscourse used by sociologist is the highest (1.18, n = 53) while Medicine RAs use the lowest rate of this category (0.23, n = 4) and Education RAs fall in between (0.81, n = 31). As for the Impersonal metadiscourse, the authors in Sociology (9.55, n = 429) and Education (9.25, n = 353) use the highest rate of Impersonal metadiscourse while writers of Medicine use the lowest rate (7.46, n = 131) across the three disciplines. The occurrence of Impersonal metadiscourse in the three academic disciplines is very similar to the total metadiscourse presented in Table 2. This means that the two disciplines of Sociology and Education show similar pattern in the density of Impersonal metadiscourse which make them distinct from RAs in Medicine. The Personal metadiscourse, on the other hand, shows a considerable variability across the three disciplines. It is this category which is employed by authors very differently across the three disciplines.

Table 3 Raw and relative frequency of Personal and Impersonal metadiscourse across three disciplines

As for the ratio of Impersonal to Personal metadiscourse markers, Persian writers use Impersonal metadiscourse markers much more than Personal ones. Sociology RAS show the lowest ratio (almost 8 times) while Medicine RAs have the highest ratio (almost 32 times).

Table 4 below presents the results of Personal and Impersonal metadiscourse taken from two other studies. The first set of results come from English RAs of biology, astrophysics, applied linguistics, and marketing (Hyland, 1998) and the second set are Spanish results from RAs in linguistics, economics, and medicine (Salas, 2015). Due to a different taxonomy used in the English study, its results are not directly comparable to the results of the current study. The English results are presented here to provide a cross-disciplinary comparison for using Personal versus Impersonal metadiscourse markers and their relevant subcategories.Footnote 4 The noticeable higher rate of metadiscourse in this language is due to additional categories such as hedges , emphatics, and attitude markers existing in the taxonomy used in the study. Please note that the numbers presented in Table 4, similar to the current study, are frequency per 1000 words.

Table 4 Frequency of Personal and Impersonal metadiscourse (per 1000 words) in English and Spanish

In general, Table 4 shows that both English and Spanish authors, similar to Persian authors, use Impersonal metadiscourse markers more than Personal ones. However, the variability across these two categories in the same discipline is much greater in Persian compared to Spanish and English. In fact, the total ratio of using Impersonal to Personal metadiscourse markers in Persian is greater than 10 whereas this ratio for Spanish and English is 1.5 and 1.3, respectively. There is more balance between the employment of Personal and Impersonal metadiscourse markers by both English and Spanish authors compared to Persian authors. Persian writers use personal metadiscourse markers considerably much less than their English and Spanish colleagues. Another noticeable difference here is that the rankings for the density of using Personal and Impersonal metadiscourse markers in English in the same discipline changes while this ranking stays the same in Persian. This means that while Persian sociologists used the highest rate of both Personal and Impersonal metadiscourse and Persian medical specialists used these two categories the least, the English biologists and astrophysicists used the lowest rates of Personal metadiscourse while they used the highest rates regarding Impersonal metadiscourse. From this perspective, the Persian authors show similar pattern to Spanish authors; however, the density of Personal metadiscourse markers compared to Impersonal ones in the same discipline in the two languages is remarkably different.

Further investigation of Personal metadiscourse in Persian shows that there are two major functional categories used by the authors in the three disciplines. The first category is self-mentions (i.e., referring to the writer/author) and the second is reference to the participants (i.e., both the writer and the reader). Education RAs contain writer-mentions almost two times more than the other two disciplines. The writers in all disciplines did not use any personal pronouns such as "من" I or "ما" we in the subject position, rather the self-mentions were only realized through using words such as "محققان" researchers or "پژوهشگر" researcher. The exclusive personal pronoun "ما" we was only used in genitive structures accompanying other words such as "در تحقیق ما" in our research. The majority of self-mentions (i.e., almost 75%) are made by attached verbal suffixes. Persian is a pro-drop language which allows the subject of the sentence to be dropped without losing its reference (See Salas, 2015, for a similar case in Spanish). Sentence (6) provides an example for the self-mention realized through verbal ending.

(6) hæmɒngune ke gofte ʃod hædæf æz ænd͡ʒɒme in tæhɢiɢ ɒn bud ke be fæhme biʃtæri æz mæʔluliæt væ tæd͡ʒɒrobe æfrɒde dɒrɒje mæʔluliæte d͡ʒesmɒni dæst jɒbim.

As mentioned, the purpose of this study was to gain a better understanding of disability and the experiences of people with physical disabilities. (SOCIO_11, S695)

The limited number of self-mentions in the corpus of current study suggests that Persian writers do not show their presence explicitly and they are mainly invisible in the text. These writers mainly tend to employ a strictly impersonal style.

The micro-level analysis revealed that when the writer was in focus, the Persian authors used specific discourse functions. Table 5 below presents the frequency of different discourse functions related to the writer’s presence across the three disciplines. The total results show that sociologists show the highest rate of writer-oriented metadiscourse realization in their text (047, n = 21) followed by educationists (0.21, n = 8) and medical specialists (0.17, n = 3). According to this table, the discourse functions at work mainly included Introducing Topic, Saying, arguing, Clarifying, and Contextualizing. The Persian writers mainly use this category when they want to introduce what is going to come in their articles or bringing up the topics which are important for the readers. Ädel (2006) has also mentioned that Introducing Topic is a very common function of personal metadiscourse in her academic English corpus.Footnote 5 Moreover, she has mentioned that the English authors in her study employed a wide range of discourse functions including Reminding, Exemplifying, and Focusing. These discourse functions were absent in the RAs written by Persian authors. These writers used only a subset of discourse strategies available in the academic discourse when compared to the English authors.

Table 5 Distribution of the discourse functions of writer-oriented metadiscourse across three disciplines

Salas (2015) has reported the frequency of writer-oriented metadiscourse in her study for linguistics, economics, and medicine as 2.58, 2.02, and 1.20, respectively. Hyland (1998) has also reported that for the category Person MarkersFootnote 6 in his study, the RAs in biology, astrophysics, applied linguistics, and marketing show the rate of 2.4, 5.3, 2.9, and 4.4, respectively. This confirms that Persian writers in the three disciplines have less tendency to present themselves in their text compared to English and Spanish authors. English and Spanish authors are noticeably more visible in their texts compared to Persian authors. This makes the English and Spanish academic discourse more interactive and engaging than the Persian academic discourse which could lead to a stronger relationship and tighter bonding between the writer and the reader in both English and Spanish texts compared to Persian texts.

The second functional category under Personal metadiscourse found in this study was participant-oriented metadiscourse. Table 6 shows the frequency of various functions of this category across the three disciplines. According to this table, Sociology RAs contains the biggest number of this category (0.71, n = 32), followed by Education (0.60, n = 23), and Medicine (0.06, n = 1). According to this table, the most frequent discourse functions when the writer brings the reader into the dialogic scene are Aligning Perspective followed by Arguing and Contextualizing. This ranking is strikingly different from the one reported by Ädel (2006). She has reported Appealing to the Reader, and Anticipating Reader’s Reaction as the top two discourse functions used by American native writers in her corpus. The results of our study show that the functional category Anticipating Reader’s Reaction even has not been used by the Persian writers. Crismore (1989) has pointed out that anticipating the reader’s reaction is a central function in metadiscourse. The considerate writer should always foresee the reaction of the reader to their texts and the probable objections raised by them (Ädel, 2006). It seems that Persian authors do not pay special attention to the imagined reader and do not plan to address the objections or counterarguments raised by the reader regarding the writer’s claims in the text.

Table 6 Distribution of the discourse functions of participant-oriented metadiscurse across three disciplines

The Persian sociologists and educationists have predominately used Aligning Perspective as the main discourse function in their RAs. They have mainly used attached verbal suffixes corresponding to inclusive we in order to make the reader involved in their text and fulfil the function. According to Ädel (2006), the primary goal of Aligning Perspective function is to have the reader take the writer’s perspective and agrees with his/her arguments regarding some issues. The writers of RAs in Persian have usually used this function in conditional sentences. Moreover, the topics which are discussed in such sentences are usually non-controversial so that the chance of being accepted gets higher. Sentences (7) and (8) show that the writers are inviting the readers to share with them the same perspective regarding a topic which is not very controversial.

(7) lezɒ ægær bexɒhim nomreje honærd͡ʒujɒn rɒ be dɒneʃ væ tævɒnɒʔije ɒnhɒ dær dærse mæzkur nesbæt dæhim mitævɒn goft in honærd͡ʒujɒn dær dærshɒje mæhɒræti væ kɒrgɒhi nomerɒti behtær kæsb kærdeænd, jæʔni nesbæt be dærshɒje næzæri movæfæɢijæte biʃtæri dɒʃteænd.

Therefore, if we want to attribute the students’ score to their knowledge and ability in the mentioned course, we can say that these students have obtained better scores in skill courses and workshops, that is, they have been more successful than theoretical courses. (EDU_6, S482)

(8) emruze ʃɒhedim ke dær besjɒri æz zæminehɒ kenɒr gozɒʃtæn væ be hɒʃije rɒndæne in æfrɒd tæʔæd͡ʒob bærængiz næbude væ be næhve besjɒr gostærdei suræt migiræd.

Today we see that in many areas it is not surprising to exclude and marginalize these people and it is done very widely. (SOCIO_11, S643)

The other discourse functions used in participant-oriented metadiscurse were Arguing, Contextualizing, Appealing to the Reader, and Imagining Scenarios. The fact that Persian authors use participant-oriented metadiscourse when arguing for or against something is unexpected. Sentence (9) provides an example to illustrate this function.

(9) bænɒbærin kæm budæne mizɒne hæmbæstegi mijɒne bærxi æz moælefehɒje huʃe kælɒmie kudækɒne piʃdæbestɒni bɒ nomreje roʃde zæbɒne ɒnhɒ rɒ mitævɒn ingune tod͡ʒih nemud ke entezɒr mirævæd bɒ æfzɒjeʃe sene in goruh æz noɒmuzɒn hæmbæstegie biʃtæri mijɒne nomreje huʃe kælɒmi væ roʃde zæbɒnie ɒnhɒ rɒ ʃɒhed bɒʃim.

Therefore, the low level of correlation between some components of verbal intelligence of preschool children with their language development score can be explained by the fact that as they age, we expect to see more correlation between the score of verbal intelligence and their language development. (EDU_9, S575)

Salas (2015) has reported that participant-oriented metadiscourse has been used differently by the Spanish authors. The linguists used the highest rate of this category (i.e., 1.13) while medical specialists and economists used it at the lower rate of 0.43 and 0.36, respectively.Footnote 7 Hyland (1998) has reported that the rates of occurring Relational MarkersFootnote 8 in English biology, astrophysics, applied linguistics, and marketing RAs are 0.7, 1.4, 2.5, and 3.3, respectively. These results indicate that both Spanish and English authors on average make more attempts to establish relationships and interact with their audience compared to Persian writers. Ädel (2006) has also stated that the relationship between the writer and the reader is emphasized in the English texts, especially by the discourse function Appealing to the Reader. This metadiscourse function is ranked average-low in the Persian RAs while it is ranked very high in the argumentative essays written in American English reported by Ädel (2006).

Table 7 presents the results for Impersonal metadiscourse markers. The total results show that sociologists use the highest rate of Impersonal metadiscourse (9.55, n = 429) while medical specialists use the lowest rate (7.46, n = 131). The Spanish linguists, economists and medical specialist are reported to use this category 7.06, 4.77, and 4.69, respectively (Salas, 2015). These results show that soft sciences such as sociology, education, and linguistics are more dense in terms of Impersonal metadiscourse markers compared to hard sciences like medicine. Further examination of Impersonal metadiscourse markers showed that there are four major functional categories as Reference to Text, Code Glosses , Discourse Labels, and Phorics existing under this category. The highest rate of functional category across the three disciplines was Discourse Labels (n = 448) while the lowest rate was Code Glosses (n = 90). The medical specialists used the highest rate of Reference to Text (3.52) and educationists employed the most Discourse Labels (4.74) across the three disciplines. English and Spanish writers use the functional category Phorics among the top two in the list, while this category is ranked the second from below in the Persian RAs (Ädel, 2006; Salas, 2015). Ädel describes Phorics as the road signs which point to different portions in the current text at different times. Hyland (1998) states that this functional category “play[s] an important role in making additional ideational material salient and therefore available to the reader in aiding the recovery of the writer’s argumentative intentions”. (p. 443). The fact that Persian authors make use of this functional category less than English and Spanish authors suggest that unveiling the argumentations made in the text may not be the primary goal of the Persian writers.

Table 7 Distribution of different categories of Impersonal metadiscourse across three disciplines

Further investigation of the four major functional categories of Impersonal metadiscourse revealed that each has some subcategories. Table 8 presents the discourse functions which are employed under each subcategory. The densities of subcategories Adding, Enumeration, In/Direct Code Glosses , and Whole Text are highest in all subcategories. In general, the results in this section showed that the distribution of metadiscousre in Persian is very specific and does not follow the existing patterns in Spanish and English. While the density between Personal and Impersonal metadiscourse in English and Spanish was relatively balanced, Persian RAs were quite skewed in terms of the distribution of these two categories. This means that Persian authors are less visible in their texts compared to English and Spanish writers. As a result, there is less interaction and probably less guidance provided to Persian readers. Our findings also showed that sociologists and educationists use Impersonal metadiscsourse markers similarly, but they get separate from each other when it comes to using Personal metadiscsourse. The results indicated that Persian authors use only a subset of metadiscursive features available in academic discourse. The following section presents the discussion of this study.

Table 8 Distribution of different subcategories of Impersonal metadiscourse across three disciplines

8 Discussion and Conclusion

The spread of a given language or its registers could happen either by social or communicative needs. The social needs are created as a result of receiving socioeconomic benefits or achieving political integration. The communicative needs, on the other hand, are created because the newly generated knowledge should be transmitted via effective and persuasive mediums (Garcia, 2012, p. 2). This means that in order for a variety of language to occupy the position of academic register , it needs to have specific features to meet the needs of the larger target academic community. The findings of the current study showed that the academic register of Persian does not fully employ rhetorical resources to achieve communicative objectives. The results showed that the application of Personal metadiscourse compared to Impersonal metadiscourse in Persian texts is remarkably lower than that of English and Spanish. This means that explicit reference to both the writer and the reader in the Persian texts is not enough and this makes the academic discourse in this language less interactive and more impersonal. Moreover, the results of our research showed that Persian authors use functional categories of Impersonal metadiscourse like phorics less than English and Spanish authors. This means Persian authors provide less signs to the reader for their navigation through the text. Hyland (2017) emphasizes that “metadiscourse refers to how we use language out of consideration for our readers or hearers based on our estimation of how best we can help them process and comprehend what we are saying” (p. 17). This implies that writers are responsible towards their readers when more clarification, guidance, and interaction is needed. In order to account for the unexpected lack of both interaction and the presence of the writer/reader in academic Persian texts, one could argue that this is due to the nature of this language defined as a reader-responsible rather than a writer-responsible language. Hinds (1987), in his seminal work on the typology of languages, found that in some languages like English it is the writer who is primarily responsible for effective communication while in some other languages like Japanese this responsibility is on the side of the reader. More recent studies have shown that both Spanish (Mur Dueñas, 2011; Salas, 2015) and Persian (Jalilifar, 2011; Pishghadam & Attaran, 2012), similar to Japanese, are reader-responsible languages. This means that writers in these languages tend to leave the responsibility to the readers to interpret the content and to make relationships between different parts of the texts. This could also result in using less metadiscourse markers by the Persian as well as Spanish authors.

The results of our study showed that Persian, a reader-responsible language, does not show the same distribution of metadiscursive markers as Spanish, which is also classified as a reader-responsible language. While both these two languages show lesser density for metadiscursive devices which makes them a reader-responsible language versus English, a writer-responsible language with higher density, both Persian and Spanish diverge from each other as to how metadiscourse markers are distributed. Our results showed that the ratio of using Personal to Impersonal metadiscourse in Persian was one to ten whereas this ratio was one to two for Spanish. This means that Spanish is a more interactive language compared to Persian despite the fact that both of these languages are classified as reader-responsible languages. This indicates that terms such as reader-responsible and writer-responsible languages are very loose terms which cannot reflect the actual rhetorical practices performed by a particular academic community. The classification of languages categorically either as reader-responsible or writer-responsible language conceals the rhetorical habits and activities practiced by the academic community. The findings of the current study showed that the academic register of Persian lacks participant interaction. This means that Persian academic writers and language policy makers need to pay special attention to this important rhetorical feature lacking in the actual practices among the target discourse community.

Despite such shortcoming, the current status of academic Persian and the extent of rhetorical features used in this register could still satisfy the primary needs of the smaller and particular discourse community. But if academic Persian is to establish its position in a larger discourse community among competitive Middle Eastern languages, it needs to provide researchers with a rich strain of rhetorical strategies and choices. The findings of this study showed that academic Persian texts used in this study lack interpersonal resources in terms of the writer and the reader involvement. The literature has emphasized that in order to win the community’s acceptance and create a powerful and persuasive text, keeping a good balance between objective information, subjective evaluation and interpersonal negotiation as a powerful convincing factor in social construction of knowledge is needed (Abdollahzadeh, 2011; Bazerman, 1988; Kuhn, 1972). The results of our analysis showed that the Persian authors were rarely visible in their texts and the readers did not receive enough references. For promoting the existing status of academic Persian, therefore, we suggest that the Persian academic community should aim for pushing the current position of academic Persian on the continuum of writer-reader-responsibility towards a writer-responsible language. This means that the participant interaction and the involvement of both the writer and the reader in Persian texts should be increased. This could be achieved by implementing linguistic policies which direct the academic register of Persian towards such a goal. The enforcement of such policies will provide more chance for academic Persian to establish itself as strong medium of communication among a larger academic community including both native and L2 users.

We need to mention that the data used in this study included only the introduction and discussion sections of the RAs. This may suggest that the peculiar distribution of metadiscourse markers found in this study is due to the nature of corpus. Since the density of rhetorical devices in the introduction and discussion sections is highest (Hyland, 2000; Mauranen, 1993; Vassileva, 2001), it is very unlikely that including the other two sections (i.e., methodology and result) of the RAs will change the distributional patterns found in this study. To sum up, we tried to find the features and strategies which could promote the status of Persian as an academic language in this chapter. We showed that interpersonal aspects and greater involvement of the writer and the reader in the text are the boundaries which need to be extended in academic Persian.