Citation Behaviours of Applied Linguists in Discussion Sections of ‎Research Articles

Introduction

Academic writing is an interaction where knowledge is made based on previous established ideas. Research articles (RAs) are a central and preferred genre in exchanging and disseminating knowledge among academic community members and “an indicator of academic achievement” (Azirah, 2005, p. 4). An academic work is considered successful if it is read and cited by other scientists in the field (Slyder et al., 2011). With the explosion of online international journals and the ease of access to these sources around the globe and considering the importance of citation counts as an indicator of the quality and impact of academic writing and a basis for promotion, citations have become even more vital in the career of those who are cited (Hyland & Jiang, 2017)

While RAs have become the currency of science (Figa`-Talamanca 2007), citations are now "the currency of the scholarly economy" (Hyland & Jiang, 2017, p.1). Referring to the works of the authorities has become a key feature of academic writing since the late medieval period (Taavitsainen, 2002). To become a successful academic work it needs to meet the criteria: the author must demonstrate the novelty of his/her work and situate the work in her/his discourse community (Hyland, 2015). By showing the work's dependence on a disciplinary context, writers demonstrate their familiarity with the discipline and create an intertextual framework for their work. In other words, writers "gradually integrate" their new claims by offering "propositional warrants" and "establishing a context through citation" (Hyland, 2010, p. 118).

Due to its importance in academic discourse and scholarly publications, citation has attracted several scholars in applied linguistics. Citation analysts in Applied Linguistics have focused on several aspects of citations. One aspect is citation count which is concerned with frequency of citations in academic texts (Coffin, 2009; Hyland, 2004; Thompson, 2005). Another line of studies have focused on linguistic features of citations such as reporting verbs and tense (Hyland, 1999; Jallilifar, 2012; Shooshtari & Jalilifar, 2010; Thompson & Ye, 1991). Citation integration (integrative, non-integrative) has also been the focus of some other studies (Hewings, Lillis, & Vladimirou, 2010; Hyland, 2000; Swales, 1990; Thompson, 2005).  The other aspect of citation which has been investigated is functions of citations (Fazel & Shi, 2015; Harwood, 2009; Hu & Wang, 2014; Loan, 2016; Mansourizahed & Ahmad, 2011; Samraj, 2013). Regardless of which aspect of citation has been studied, citation analyses have focused on comparing the use of citations in different disciplines and different languages as well as the citation practice between the novice and expert writers within the same discipline.

Whereas a wealth number of studies on citations on different disciplines have found that citation plays an important role in academic writing (Charles, 2006), differences in citation practices across different disciplines have been observed. Hyland’s (2000, 2004) study on 240 RAs from eight disciplines showed variations in the discourse features, including citations, of these disciplines.  His analysis showed that among the eight disciplines, which included both ‘soft’ and ‘hard’ disciplines (see Becher, 1989), sociologists used the most number and physicists the least number of citations. Hyland concluded that citations in soft disciplines are more frequent than hard disciplines as, because of their nature, they need to situate their studies more firmly in the literature. Hyland (1999) argues that “the imperatives that motivate citation are contextually variable and are related to community norms of effective argument” (p. 362). In this line, other studies have shown Applied Linguistic RA writers (as a soft discipline) overuse citations compared to their counterparts in disciplines such as Computer Science (Posteguillo, 1999) and Medicine (Nwogu, 1997) (as hard disciplines).

Studies on citations have mainly focused on citation practices in published writing, particularly RAs, (Hewings et al., 2010; Hu & Wang, 2014; Martines, 2008), as well as novice writing in the form of doctoral theses (Thompson, 2005; Thompson & Tribble, 2001), master’s dissertations (Jalilifar & Dabbi, 2010; Loan, 2016; Samraj, 2013), articles written by expert and novice writers (Mansourizadeh & Ahmad, 2011), doctoral grant proposals (Fazel & Shi, 2015) and undergraduate students’ writing (Lee, Hitchcock & Casal, 2018). Studies have also focused on comparing citations in the expert writings (research articles) with the writing of novice writers (master dissertations and doctoral theses). Jalilifar (2012) compared citations in the introduction of master’s dissertations and RAs and found differences between the two sets of his corpora. He concluded that the higher use of citation, especially in the integral form, was an indication of the familiarity of the students with the form of citations but a lack of their knowledge of their functions.

The line of inquiry focused on citation from a cross-linguistics aspect (Hewings et al., 2010; Hu & Wang, 2014; Shooshtari & Jalilifar; 2010) have showed that cultural backgrounds can also affect the citation practices of writers. For example, Hu and Wang (2014), in a cross-disciplinary, cross-linguistics study of citation practices in published articles in Chinese and English medium journals in Applied Linguistics and General Medicine found differences in the type and level of citations between these four groups of writers. For example, they found a higher density of citations published in English by applied linguists. They concluded that citation practices can be influenced by the types of information which are communicated, disciplinary and culturally epistemologies and norms of communications. In another study, Shooshtari and Jalilifar (2010) looked at the citation practice of applied linguists in the discussion section of English articles published in local (Iranian) and international journals. Their findings showed an overuse of integral citations by local published articles. They concluded that while local writers were aware of the forms of citations, they ‘underestimated’ the functional aspects of them.

In a more precise recent genre based study, Samraj (2013) looked at the forms and functions of citations used in the discussion section/chapter of ecology, a sub-discipline of biology RAs/master dissertations. Adopting move-step approach (see Swales, 1990), she investigated the rhetorical functions of citations in discussion sections of 16 RAs and 8 master’s dissertations. Previous genre studies on rhetorical structure of discussion sections have identified Reference to Previous Research as one move (communicative segment) which is used to support and/or compare the study with literature in discussion sections. Swales (1990) uses the move of Reference to Previous Research to compare the findings with previous research and/or support the present study. Dudley-Evans’ (1994) model also includes the same move which can be used to compare the findings with literature and/or support the claims and explanations they make. Several other studies (e.g., Holmes, 1997, 2000; Posteguillo, 1999; Peacock, 2002), which have adopted Swales’ or Dudley-Evans’ model, have identified Reference to Previous Research as one of the common communicational purposes in their data to compare the findings with literature. Samraj (2013) identified more precise and detailed functions (eight) for citations in her two corpora among which comparison of results, comprising around two fifth of the whole citations, and interpretation of results, comprising over a quarter of the whole citations, were the most frequent functions in RAs’ discussion section.

Academic discourse is an interactive practice where the writer’s main aim is to persuade the readers and gain discourse community’s acceptance (Hyland, 2010) and discussion section enjoys a crucial role in establishing its importance and creating persuasion. In this section writers go beyond merely presenting their findings objectively and present and argue their own points of view about them in a way that “readers are likely to find persuasive” (Hyland, 2005, p. 176). Besides, a well-written discussion section needs to relate the findings of the study to the relevant literature (Bloomberg & Volpe, 2008; Dörnyei, 2007; Mackey & Gass, 2016; Paltridge & Starfield, 2007; Perry 2005). Therefore, in addition to stating findings and commenting on them, it is also an important stage that writers make connections between existing knowledge and their own study and demonstrate a clear connection between present study and previous theory and research and show how their work “is important in the 'bigger picture'” (Cargill & O'Connor, 2009, p. 10). By critically engaging with literature, the writers contextualize their study and help their audience to understand how the findings are connected to previous work and how they support or challenge earlier studies (Mackey & Gass, 2016). As Petric (2007, p. 246) states, by citing other researchers’ work and indicating similarities and differences of one’s own work and the cited ones, writers foreground their own research and position it in relation to other works and show their contribution to the field. Discussion is also the section that students find the most problematic to write and understand (Bitchener & Basturkmen, 2006; Dudley-Evans, 1994). Citation, a common strategy for situating a work within the disciplinary community, has been identified as one of the important elements of persuasion in academic writing (Hyland, 2004) which has a high density in discussion section (next to introduction section) (Mansourizadeh & Ahmad, 2011).

Genre studies have shown Referring to Literature as an important communicative move in discussion sections in various fields, including applied linguistics. Genre studies, attempting to identify the generic structure of discussion sections of RAs in various fields, have acknowledged reference to literature as an important communicative move. However, they have either limited it to just one move, ‘to compare the findings with literature’ (Swales, 1990; Dudley-Evans, 1994; Peacock, 2002) or even a step under the move of ‘commenting on findings’ (Kanoksilapatham, 2005; Lim, 2010; Yang & Allison, 2003). In spite that, previous research on citation used in discussion sections of master’s dissertations (Samraj, 2008, 2013) and RAs (Samraj, 2013; Kwan & Chan, 2014) have shown that reference to literature is not used only for comparing findings, but serves other functions.

Given the importance of discussion sections, the vital role of citation in academic discourse, particularly discussion section, and students difficulty in writing this section as well their problems in proper use of citations in their writings, studying the rhetorical functions of citations to find out what functions they serve in this sub-genre would yield important information about citation practice of this important sub-genre which can assist EAP developers and newcomers to the discourse community. In order to explore intertextual links, this paper analyzes the discussion sections of applied linguistics RAs for the form (integral, non-integral) and rhetorical functions of citations. In particular the paper plans on answering two questions:

1- What are the forms of the citations used in the discussion section of Applied Linguistics research articles?

2- What are the rhetorical functions of the citations used in the discussion section of Applied Linguistics research articles?

Method

This study is a small scale mixed-method genre-based study and consists of two main parts. Forty-five RAs’ discussions, selected randomly from five high impact journals, are analyzed in terms of formal characteristics of citations (integral, non-integral) and their rhetorical functions.

Data Collection

The corpus consisted of 45 RAs' Discussion sections including nine articles selected randomly from five high impact factor journals in the field of Applied Linguistics published from 2003-2013. The corpus was selected from among a database of over 600 articles which was compiled as part of a large scale project during 2008 to 2015. In selection of the journals it was tried to select the ones that both published qualitative and quantitative articles. Previous research has shown that research design (qualitative/quantitative) might have “rhetorical effects which are reflected in preferred patterns of persuasion” (Hyland, 1999,
p. 81). Hyland’s study on soft and hard disciplines has shown that the differences in epistemology and how these disciplines see the world and what they consider as knowledge influences the way the academics write in them. In Applied Linguistics, research is both done in qualitative and quantitative designs. While quantitative research is more close to the hard side of the continuum, qualitative research is more close to the soft side. In order to prevent any possible influence of the research design, journals such as Modern Language Journal and Language Learning which dealt mostly with quantitative studies were excluded.

The selected journals, based on their accessibility in electronic version for the researchers, included: Applied Linguistics (APP), English for Specific Purposes (ESP), Studies in Second Language Acquisition (SSLA), Language Teaching Research (LTR) and TESOL Quarterly (TESOL). After selecting the journals, two empirical articles (qualitative, quantitative and mixed method) from published issues in each year (2003-2013) of the five journals were selected randomly which gave a total of 100 articles. The focus of this study was on Discussion sections of the articles, therefore all the 100 articles were checked for meeting this criterion. The papers which combined Discussion section with Conclusion, Summary, Findings and Analysis were excluded. This yielded a total of 84 articles which included from 9 to 15 articles from each journal. In order to select equal number of articles from each journal, 9 were selected randomly from each of the five journals which gave a total of 45 articles (45,162 words)

Table 1. Corpus characteristics

Data Analysis

Although the study is mainly a small scale qualitative study, it also adopts some quantitative analysis as well. Data analysis was conducted in two phases: first the forms of the citations were identified and then the rhetorical functions that they served were investigated.

Citation forms. Swales' (1990) categorization of integral and non-integral citations was used to identify the forms of the citations. In integral citations, the name of the cited author is included in the cited sentence which has a grammatical role (as in example 1); in non-integral citations, the name of the cited author appears within parentheses or footnote and plays no grammatical role in the cited sentence (as in example 2).

1) Swales (1990) states that it is the expert or active members of a discourse community that give a specific generic name to a particular communicative event.

2) Based on this definition, genres are not merely types of texts, rather, they are communicative events which are identifiable based on their communicative purposes (Swales, 2004).

In order to identify the forms of the citations, at first all instances of another author’s name regardless of whether it was followed by a year or not as well as third person pronouns whenever referred to a cited author were counted. Following Hyland (2000) and Samraj’s (2013) method, counting the frequency of the citations and assigning a function to them was done independent from being present as a single or multiple reference citation. Based on this criterion, the number of references was recorded first. Next, the frequency of integral and non-integral types and their ration to the whole citations were calculated. Then, the percentages of integral and non-integral were calculated. Finally, the counted citations were normalized at 1,000 words in order to be able to compare with the findings from other studies.

Citation functions. In order to identify the functions of the citations, Samraj’s (2013) typology was used. Her study is one of the few and recent studies that have employed genre-based approach to identifying rhetorical functions of citations in the discussion sections. Thus, it was found a suitable typology to use, given the purpose of the present study and the sub-genre it focused on. Using a genre analysis approach and studying the discussion section of 8 RAs and 8 master’s dissertation’s discussion sections from biology (ecology), Samraj have proposed a typology of citation functions in this sub-genre. The typology includes eight functions of which seven was identified in her RAs corpus:

• · Comparison of results
• · Interpretation of results
• · Explanation of results
• · Evaluation of the study
• · Evaluation of the field
• · Research recommendations
• · Applied Recommendations (identified only in the dissertations)
• · Background

Using a genre analysis approach (Swales, 1990), after identifying the citations, they were studied in their context to find out what purposes they served. A genre (e.g., research articles) and sub-genre (e.g., discussion section of RA) is organized based on a set of communicative purposes which are realized by communicative units. These communicative units are called Move which carry the specific communicative purpose of a particular part of a text. A Move might be realized by one or more subsidiary elements called Step. Thus, Moves and Steps constitute the generic structure of a particular genre.

In order to identify the rhetorical functions of the citations, after they were identified in the first phase, the citations were studied in the context that they were used. It was tried to identify the communicative purpose of the segment that the citation was used in. The identification was not done only based on the sentence that the citations appeared in. In many cases, the wider context of these citations had to be studied in order to find their functions. Adapting Samraj's (2013) framework, only four functions were identified in the corpus of the present study. In her framework, she differentiated between research recommendations and applied recommendations and her analysis showed no use of citation in applied recommendation. As the communicative purpose of both of these is the same, similar to other genre analysis studies, it was decided to label both as Recommendation. After identifying the cases, all of them were examined and double checked carefully to ensure that they all represented the identified function. In the next stage, the overall frequency of each function was counted manually and the percentage of each of the functions was calculated. The overall frequency and percentage of the forms and functions of the citations and preference of some types might provide preliminary understanding of the relative importance of certain forms and functions in this sub-genre.

Findings and Discussion

Citation Forms

As Table 2 shows, a total of 330 citations were identified in the corpus which is an average frequency of about 7.31 citations in each article. The frequencies of the citations ranged from 6 to 12. The density of citations per 1,000 words was 7.31 which was lower than the frequency Jallilifar (2012) identified in introduction sections of Applied Linguistics RAs. This can be explained by findings of other studies (Mansourizadeh & Ahmad, 2011; Hu, et al. 2013) which have shown that citations are densely used in the introduction sections in order to catch the attention of the readers when scanning the papers.

The relative percentage of the two forms of citations, integral and non-integral, showed an overuse of non-integral form (comprising 60.91% of the whole citations) by applied linguists. This indicates that applied linguists, in their discussion section, prefer to focus on the propositions, research actions and findings of other researchers in order to argue and support their findings rather than on the cited authors. Previous studies have also shown a preference for non-integral citations in various disciplines (Hewings et al., 2010; Hyland, 2004; Hyland & Jiang, 2017; Thompson, 2012; Samraj, 2013). The preference of the use of integral and non-integral forms of citations can be related to attribution (transferring the responsibility of statements that are being said by writers). As Thompson (2012) argues, non-integral citations allow writers to concentrate on the proposition and avoid explicitly bringing other researchers in their work. Non-integral citations also help the authors to keep the flow of the argument without interrupting it (Hewings et al., 2010). Finding an increase in the preference of non-integration citation over the fifty years in several disciplines, including Applied Linguistics, Hyland and Jiang (2017) conclude that authors "are moving towards a rhetorical style which gives less prominence to other authors [and allow them to use citations that]… support their ideas but does not foreground the originator" (p.11).

Table 2. Occurrences of Integral and Non-Integral Citations in the Corpus

Citation Functions

Adapting Samraj's (2013) framework, four functions were identified which were used to varying levels in the corpus (see Table 3).

Table 3. FrequencyoftheFunctionsof Citations in the Corpus

As Table 3 indicates, the functions varied in frequency with the comparison of the findings and explanation of results as being the most frequent and interpretation of the results and research recommendations being the least frequent functions.

Comparison of results. Comparison of results with literature is the most dominant function of citations in discussion sections of RAs, the communicative purpose of which is to compare and contrast the findings with relevant studies in literature and connect the findings to the field. Similar observations were also reported in Samraj (2013) and Kwan and Chan (2014). In the data of this study, 42 percent of the whole citations (139 out of 330 instances) were employed for the purpose of comparing the results with literature. The comparisons were made either with a specific or general claim/finding. Connecting the findings to the field and showing the contribution of their studies to the literature was done by showing consistency and inconsistency of the findings with literature. Citation was used to show the ways in which the findings were similar to findings in literature (excerpt 1) and/or a claim or assumption in literature (excerpt 2).

1) There was evidence that repeating a task with well-defined parameters and similar content (i.e. switching roles during a role play) increased the likelihood of its completion, a finding that is consistent with studies that attest to improved proficiency ... (e.g., Gass, Mackey, Alvarez-Torres & Fernandez-Garcia, 1999). [Referring to literature to compare the findings]

2) Our data thus supports Metge and Kinloch’s (1978) contention that for at least some Maori people in large Pakeha organizations ... (see also Metge, 2001).

Excerpt 3, below, illustrate instances that the writers showed the inconsistency of their findings with findings of other studies in literature. In excerpt 4, the writers refer to a model in literature and indicate the conflict between their findings and the model.

3) It is interesting that, in contrast to the study reported in this article, Bitchener et al. did not find any statistically significant effect for direct corrective feedback alone (i.e. without metalinguistic comments).

4) According to Bialystok’s (1993) model, more proficient learners are able to executive attention to target pragmatic features more accurately than less proficient learners because ... However, this was not the case in the context of the present study, suggesting that ...

Comparing findings with literature is a common move in discussion sections that writers use to compare and contrast their findings with relevant studies in literature and connect the findings to the field. It has been emphasized widely in the literature that a well-written Discussion section needs to relate the findings of the study to the relevant literature (e.g., Bloomberg & Volpe, 2008; Dörnyei, 2007; Mackey & Gass, 2016; Paltridge & Starfield, 2007). This has been identified in the generic structure of discussions in various disciplines including applied linguistics (Yang & Allison, 2003). The main purpose of referring to literature is to put the findings of the study in the context of previous research and compare and contrast them with existing findings, claims, assumptions, or theories in literature.

Therefore, in addition to stating findings and commenting on them, it is also an important stage that writers make connections between existing knowledge and their own study and demonstrate a clear relation between present study and previous theory and research, and show how their work is important in the ‘bigger picture’ (Cargill & O'Connor, 2009, p. 10). By critically engaging with literature, the writers contextualize their study and help their audience to understand how the findings are connected to previous work and how they support or challenge earlier studies (Mackey & Gass, 2016). By citing other researchers’ work and indicating similarities and differences of one’s own work and the cited ones, writers foreground their own research and position it in relation to the other works and show their contribution to the field (Petric, 2007). Therefore, by employing this strategy, writers position new findings within the body of existing knowledge and connect it to the wider field.

The analysis showed that referring to literatureto show the inconsistency of the findings to literature was infrequent in the corpus compared to indicating consistency. While consistency/inconsistency of findings was shown by referring to assumption, claim, or contention in literature and/or with findings of studies in literature, when showing inconsistency, the writers indicated inconsistency of their findings with a statement or claim in the literature rather than a specific finding.

Investigating the intentions of writers in citing specific resources, Case and Higgins (2000, pp. 640-1) found “negative citation”, such as reference that “illustrates a perspective or finding that contradicts a perspective or finding” quite rarely in writers’ own study. They suggested that writers apparently prefer citing documents that support their own findings and knowledge claims. However, showing consistency or inconsistency of findings can also be unintentional and can be decided based on the findings. For instance, findings can also be interesting if they are different from literature and challenge it.

Explanationof results.The analysis shows that using source texts to support the writers’ explanations of the findings is the second most frequent rhetorical function of citations (comprising around 32% of the whole citations) in discussion section. In these instances, after the writers went beyond the objective presentation of their findings and presented their own understanding of the results, they referred to literature to support their comments. The references to literature were either in the form of direct or indirect quotations. In Excerpt 5, the findings are presented first. Then, the writer(s) go on and add their explanation of findings. They try to persuade the audience of their explanations' soundness by providing evidence from literature.

5) Many of the distance learners in this sample found the time and created the opportunities to engage in functional practice beyond their course work. During this activity, their main focus was on meaning, with some Incidental Focus on Form.... [Stating Findings]. For intermediate and advanced distance learners, this activity appears to be crucial in providing exposure to ... [Explaining]. White (2003: 114) points out the frequent references in writings on distance language learning to maintaining motivation as a significant factor in learners’ involvement and persistence in their distance-learning experiences. It was ranked as the most significant factor for success by distance language learners in a study that she carried out (White, 1999) [Referring to Literature to Support the Explanation].

In excerpt 6, the first sentence is the writers’ interpretation of the findings and is followed by a direct quotation from literature to support it.

6) This implies that it is difficult for learners to apply the Japanese rule, even if they have the necessary knowledge [Explaining]. The following quotation supports this view:

In fact, I was warned before I came to Japan to be cautious of hai, that it didn’t necessarily mean ‘‘yes, I will do that.’’ It meant ‘‘yes, I understand that.’’ And even with knowing that, I still found difficulties at first in understanding between the people in my company that speak very good English. (JETRO, 1980 cited from Miller, 1991:125) [Referring to Literature to Support the Explanation]

Commenting on findings and providing explanations for them is one of the main functions of discussion section and has been identified in several genre studies (e.g., Yang & Alison, 2003). One of the common ways for writers to comment on their findings is to draw on other studies and ‘borrow’ ideas, concepts, and explanations from literature (Ritchie, Spencer & O’Connor, 2003; White et al., 2003). Therefore, apart from providing their own interpretations and explanations of their findings, writers use interpretations, explanations, and concepts in literature to support the comments they make on their findings. This is a rhetorical function of citation which helps the writers to justify their knowledge claims based on the existing knowledge in the field and promote acceptance from their peers.

Recommendation.Citations were also used to support the recommendations the writers made for further research and practice. Comprising around 13% of the whole instances of citations, it appeared less frequent compared to the other two rhetorical functions discussed above. Referring to literature, either to a research or a theoretical claim, was used to provide evidence for the suggestions that the writers made. In other words, after making suggestions for practice and further research, the writers tried to back them up and provide evidence that justified their statements. In excerpt 7, the writers make a suggestion for further research while positioning a reference to literature at the end of the sentence to support the recommendation.

7) Further research is needed to explore task types that encourage a focus on fluency and meaning, such as interactive tasks and information transfer, as suggested by Robinson (2001). [Making Suggestionsforfurther research and Referring to Literature to Support the Suggestion].

As can be seen in excerpt 8, in the first sentence the writers make a suggestion for practice and then continue with a suggestion for further research which is followed by a quotation by Kramer-Dahl to support and justify these suggestions:

8)  Clearly, a closer connection between top-down implemented language policies and the realities of the classroom in which such policies must be implemented needs to be made and a research priority undertaken concerning the reality of the Singapore classroom [Making Suggestions for Practice]. As Kramer-Dahl (2003: 164) has observed: What has remained grossly under researched and poorly understood, as a result, is the... [Referring to Literature to Support the Suggestion].

By recommendation for practice, writers show the contribution of their study to the contemporary issues in the field and how it can solve the problems. In this sense, the writers promote their works by showing the valuable contribution that they make to the language issues in the real world. Recommending further research not only suggests new lines of inquiry in the field but also “demonstrates to the reader that the author has concluded one phase of the research and has carefully thought about the next phase” (Mackey & Gass, 2016, p. 351). Citations were used as a source to back up the recommendations that the authors suggested. As was mentioned earlier, this was a less frequent rhetorical function of citations. It should also be stated that recommendations were not observed frequently in the corpus and was only present in 18 RAs (40% of the corpus). One of the reasons might be that making recommendations is a move that can be present either in discussion or conclusion sections. This has also been observed by genre studies that have investigated the generic structure of Applied Linguistics RAs (Yang, 2001). This might explain the lesser frequency of this rhetorical function of citations.

Interpretationof results.In the data of this study reference to previous studies in literature was also used to support the deductions that the authors had made, although with much less frequency (identified only 41 times in 45 RAs). The reference to literature was either in the form of direct or indirect quotation. The writers referred to literature to support their deductions and interpretations of the findings to convince the audience that the conclusions they are making are based on and supported by literature. In excerpt 9 below, the writers interpret their findings in the first sentence which is followed by several references to literature to back them up and provide evidence that justifies and supports their statements.

9) Our study also suggests reconsideration of the role of oral participation versus the role of silence in both EAP and mainstream education [Interpreting Findings]. For example, Jaworski and Sachdev (1998) report that Western secondary school students associated silence with productive learning, and Rowe (1974, 1987) and Tobin (1987) demonstrate that ... In addition, as Valdés (1998) demonstrates in a study of an ESL class in the United States, silence may also be used as an important tool for classroom management. Finally, as Zhou et al. (2005, p. 303) point out in their discussion of ... [Referring to Literature to Support the Interpretation].

In excerpt 10, the writers make a deduction in the last sentence while positioning a reference to literature in the first sentence to support the deduction.

10) Goodwin and Goodwin (1992) claim that an important aspect of collaboration and negotiation in constructing and developing an emergent topic involves reaction to the content of the preceding utterances [Referring to Literature to Support the Interpretation]. Such instances of either ‘marked’ topic shift or ‘stepwise’ topic movement described in our analyses thus display characteristics of emergent topical development in conversation [Interpreting the findings].

Interpretations are inferences and logical conclusions which are based on the findings and discussions that are presented previously in the discussion section. Interpreting has been identified in several genre studies as an important element of discussion section, though under different labels such as Generalization, Deduction and Hypothesis, and Claim (Dudley-Evans, 1994; Holmes, 2000; Kanoksilapatham, 2005; Peacock, 2002; Swales, 1990).

Conclusion

Form and rhetorical functions of citations in Discussion sections of applied linguistics RAs were studies in this paper. Studying the form of the citations showed a preference for non-integral forms of citations compared to integral ones which indicates that applied linguists tend to foreground the propositions rather than the authors by using more non-integral citations. Similar results are also observed in previous research on RAs from various disciplines. Comparing the overall frequency of citations and their occurrence per 1000 words in the corpus of the present study and those in literature confirmed that while citation has a vital role in academic writing, it seems that it is used far more in Applied Linguistics than hard disciplines such as Computer Science (Posteguillo, 1999) and Medicine (Nwogu, 1999). This is in line with Hyland’s (1999) finding that writers in soft disciplines, overall, use more citation than their counterparts in hard disciplines. This might suggest that citation practices are discipline-specific and can be ruled by discourse community norms and conventions of proper and effective argument (ibid.).

Several genre studies (e.g., Holmes, 1997, 2000; Lim, 2010; Posteguillo, 1999; Peacock, 2000; Swales, 1990; Yang and Allison, 2003) have identified Reference to Previous Research as one of the common moves in discussion section of various disciplines. Studying the rhetorical functions of citations indicates similarities and differences with these studies. While the findings of this study are in line with those studies in that reference to previous research is mainly used for comparing findings, it challenges them in that reference to previous research in not solely used for comparison purposes but is used with various rhetorical functions throughout the discussion section in order to contextualize the current study within the previous studies. Reference to literature does not simply serve the function of comparison but is employed to support the writers’ explanations, interpretations, as well as their recommendations. While the predominant rhetorical function of citations was to compare and contrast the findings with literature in order to contextualize and relate them to the field, a substantial number of citations were also used to help the writers to support the arguments when making a deduction and/or suggestion, and to back up the comments made on findings. This is in agreement with the observations of several other studies (Kwan & Chan, 2014, Mansourizadeh & Ahmad, 2011; Samraj, 2013).

Academic writers use various types of evidence to back up their claims and arguments and “what counts as suitable evidence to support an argument is governed by the epistemic conventions of a discipline. Epistemic conventions refer to the means of establishing ‘truth’ as based on accepted forms of evidence” (Coffin et al., 2003, p. 27). Citations or ‘Quotative’ mode of knowing (Palmer, 1986, p. 51 cited in Hyland & Jiang, 2017) are one of the important tools in academic argument which are used to achieve persuasion (Gillbert, 1977). While part of persuasion can be achieved through logical argument, “much support for the results and the argument necessarily arises from work already performed and presented to the scientific community” (Gilbert, 1977, p. 116). Therefore, citation as an important element in creating social context of persuasion, is a rhetorical tool that writers use strategically to support their claims by “synthesizing past research and presenting evidence to back up claims or points of view” (Coffin et al., 2003, p. 27). Referring to earlier work not only incorporates the previous research into the new paper but also “provide[s] a measure of persuasive support for the newly announced findings” as referenced work “has already been accepted as 'valid science'” (Gilbert, 1977, p. 116). In his interview-based study of the functions of citation, Harwood (2009) found justifying claims as one of the purposes that computer scientists and socialists used citation for. In fact, citation is and remains "more important than ever as an element of the scholar's rhetorical armoury" and as "part of epistemological and social framework for the acceptance of argument" (Hyland & Jiang, 2017, p. 2).

Writing a research report is a challenging task for novice writers as they need to be familiar with the norms of their discourse community to establish the importance of their research and to show that their study is worthy of attention. In order to be able to negotiate with their discourse community and persuade them to accept their knowledge claims, they need to be able to apply the knowledge of norms and conventions of their discourse community, including citation practices, in their writing. Some of these norms can be acquired by observing the writing of senior members by reading RAs. Studies (e.g., Loan, 2016; Mansourizadeh & Ahmad, 2011) have shown that citation practices cannot be evident from merely reading expert written texts and some of their rhetorical functions are so complex that they can only be acquired by years of experience in the field. Studies on the texts written by graduate students (e.g., Jalilifar, 2012) have also shown that these novice writers are familiar with forms of citations and ignorant of their functions. Besides, students find the function and use of citations difficult to understand (Thompson & Tribble, 2001). The findings of the present study provide some information on the citation behavior of applied linguists particularly in discussion section of RAs. The knowledge on the rhetorical functions and forms of citations can enhance our understanding of the reasons behind choosing citations in various parts of our discussion and can be turned into pedagogical materials and raise rhetorical consciousness through teaching those choices explicitly to ESL students.