Unpacking the Functions of Personal Metadiscourse in Teachers’ Classroom Discourse

: This paper reports on a corpus-driven study on teachers’ use of three personal metadiscourse markers, namely, engaging you , inclusive- we , and self-mentioning I , in teachers’ classroom discourse. The analysis is based on eight sessions of teacher contributions to classroom discourse from four native English-speaking English for Academic Purposes (EAP) teachers in the UK. A quantitative analysis shows that teachers unanimously attach great importance to actively engaging students in classroom instructions. The qualitative analysis identiﬁed four types of metadiscourse functions in relation to teacher–student(s) interactions, including managing comprehension, managing students’ responses, imagining scenarios, and managing students’ discipline, which were further divided into nine sub-categories. This study extends the scope of existing research on personal metadiscourse functions and enriches metadiscourse research in formal instructional settings, thereby enhancing our understanding of personal metadiscourse use in spoken academic genres. Moreover, it may also provide insights for researchers and practitioners in EAP teaching by providing a ﬁne-tuned functional categorisation of personal metadiscourse markers.


Introduction
The term metadiscourse was first proposed by Zalling Harris [1]. It generally refers to the self-reflective expressions used to negotiate interactional meanings in a text, assisting the writer/speaker to express a viewpoint, and engage with readers/listeners as members of a particular community [2]. However, it was not until the 1980s that this notion began to attract the attention of researchers in linguistics. For example, a few pioneering scholars [3][4][5][6][7] strived to clarify the term, including its definitions, characteristics, and classifications. Yet, there were still relatively few empirical studies on metadiscourse until that time.
Over the past two decades or so, metadiscourse has witnessed a robust development, being studied in a variety of genres, such as news article genres [8][9][10], and business and commercial genres [11][12][13][14][15][16][17][18][19]. In addition, metadiscourse has been found to be an effective way of engaging listeners or readers in preaching or religious texts [20]. Moreover, it is also said to contribute to the argumentative and interactional effect of editorials [21][22][23]. Taken together, these studies can be classified under the general rubric of non-academic genres. Yet, they account for a relatively small proportion of the overall research into metadiscourse, when compared with academic genres.
Metadiscourse research in academic genres is the most prevalent area being investigated. A survey of existing literature reveals that most metadiscourse research in academic genres has focused on written discourse, whereas those in spoken academic genres are comparatively scarcely investigated despite the increasing attention in spoken genres in recent years [24][25][26][27][28]. Nevertheless, metadiscourse is claimed to play a far more significant part in spoken discourse than in written discourse, as there is a greater "need to manage Sustainability 2022, 14, 13502 2 of 16 spoken interaction in real time" [29] (p. 37). This makes it all the more interesting and worthwhile to explore metadiscourse in spoken academic genres.
As one of the key aspects and components of spoken academic discourse, teachers' classroom discourse plays an important part in the classroom teaching and learning process. It is not only the means of organising instructions, but also an important source of input for students, especially in English as an SL/FL context [30]. Moreover, teachers' classroom discourse may also provide L2 learners with "linguistic models of how to interact meaningfully and appropriately in communicative situations, although students at this point in their development may only notice a few at a time" [31] (p. 89). Irrespective of the significance of teacher discourse in instructional settings, the most extensively investigated aspects of classroom discourse are those related to teacher-student(s) interaction, such as corrective feedback [32], teachers' questioning strategies [33], power relationship [34,35], and identity construction [36]. However, the extent to which teachers' use of metadiscourse may contribute to the organisation of and interaction in classroom teaching is still underexplored. Nevertheless, multiple studies have noted the facilitating role of metadiscourse, in particular devices explicitly signalling text structure, in students' classroom comprehension [37][38][39][40][41][42][43].
Furthermore, recent years have witnessed increasing scholarly attention to metadiscourse research in classroom teaching contexts [29,[44][45][46][47][48][49][50]. For example, Pérez-Llantada [48] quantified the frequency variation of metadiscourse use across primarily monologic and primarily dialogic speech events. She indicated that textual metadiscourse occurs less frequently in dialogic speech events but more frequently in monologic ones. In a comparison study between teachers' use of metadiscourse in EAP lessons and academic lectures, Lee and Subtirelu [47] found that teachers' use of metadiscourse in the classroom has a close bearing on the pedagogical context, i.e., whether it is in a dialogical seminar or a monologic lecture. They also revealed that EAP teachers make more frequent use of linguistic expressions to explicitly engender student engagement, while university lecturers are more concerned with establishing relationships between ideas to assist students gaining disciplinary knowledge. Adopting Ädel's [24] reflexive model of metadiscourse, as will be elaborated in the next section of this article, Zare and Talakoli [49] compared the functions of personal metadiscourse in academic monologic and dialogic speech, which were represented by classroom lectures and discussions. Findings show that speakers make more frequent use of discourse organisation metadiscourse in monologic academic lectures, but use more audience-interaction metadiscourse in dialogic speech. In a more recent study, Bernad-Mechó [25] made a comparative analysis of the structuring segments and content sequences in lectures in Humanities from the Yale University OpenCourseWare from a multimodal perspective. The results revealed that structuring segments seems to be more modally dense than content sequences.
Taken together, the four studies reviewed above show that interactional metadiscourse is commonly employed in instructional settings. Such research has probed into metadiscourse use in different aspects of either EAP seminars or lectures from a variety of perspectives. The findings have increased our understanding of metadiscourse in classroom contexts. However, in-depth research into the use of certain metadiscourse markers are rare compared with the general research into metadiscourse categories. Scholars has noticed this gap and have begun to explore some frequently used metadiscourse, such as an investigation of the metadiscursive you across different genres by Ädel [51]. Thus, this study intends to fill this gap by focusing specifically on the functions of three relatively frequently used personal metadiscourse markers, namely the engaging you, inclusive-we, and self-mentioning I, used by native English speaking EAP teachers in the UK. Specifically, this study first investigates the frequencies and ranges of three personal metadiscourse, and then their functions in relation to teacher-student(s) interactions in classroom teaching.
These three metadiscourse markers are chosen for further analysis primarily for two reasons [52]. The first relates to their relatively higher frequencies among all metadiscourse markers. Secondly, the analysis of salient metadiscourse markers by using the collo-These three metadiscourse markers are chosen for further analysis primarily for two reasons [52]. The first relates to their relatively higher frequencies among all metadiscourse markers. Secondly, the analysis of salient metadiscourse markers by using the collocates tool in AntConc indicates that these three metadiscourse markers are more inclined to co-occur with other metadiscourse categories. Therefore, taking these three metadiscourse markers as the focal items may facilitate their function analysis together with their co-occurring metadiscourse markers. Moreover, the function of personal metadiscourse markers was also emphasised by Ädel [24], which will be illustrated in the follow section.

Overview of Research on Personal Metadiscourse
Personal metadiscourse was proposed by Ädel [53], as distinct from impersonal metadiscourse within the reflexive metadiscourse framework. It refers to the metadiscursive items that make direct reference to the interlocutors of the current text, either by pronouns (personal I, we, you and their oblique and possessive forms) or nouns (such as the writer, dear reader). In her influential monograph, Ädel [53] identified 16 discourse functions, which were then refined and expanded to 23 types of discourse functions [24,54]. These discourse functions were further framed within four major categories, including references to the audience, discourse organisation, metalinguistic comments, and speech act labels. Ultimately, the four major categories were incorporated in two broad categories of audience interaction and metatext (see Figure 1 below). Ädel's [24,54] reflexive model of metadiscourse was adopted later by a number of researchers to make a series of comparative research in both spoken and written genres. Ädel's [24,54] reflexive model of metadiscourse was adopted later by a number of researchers to make a series of comparative research in both spoken and written genres. For example, Ädel [54] compared the use of metadiscourse across three genresadvanced student writing, published academic prose, and spoken lectures. Other studies such as Salas [55] compared the use of reflexive metadiscourse in research articles from three disciplines (medicine, economics, and linguistics) written in Spanish. Hasselgård [56] compared the texts in two disciplines (linguistics versus business) written in English by three novice groups of Norwegian learners, British, and American native students. Zare [49] made a comparison between monologic and dialogic academic speeches. Li and Xu [57] compared the reflexive metadiscourse in Chinese and English sociology research article introductions and discussions.
This paper mainly focuses on the functional category of reference to audience, which was reframed as teacher-student(s) interactions in this study. Notably, based on a thorough observation of the dataset for the current research, this study combined the two functions of managing the massage and anticipating audience's response as one category of managing students' response. We are aware of the fact that there are occasions when students can also manage the classroom teaching message or anticipating the teachers' or other students' response, but this is not within the scope of this research as it focuses exclusively on an exploration of teachers' classroom metadiscourse functions.
Following this line of thought, the present study explored four types of discourse functions within the category of references to audience or audience interaction. The category audience interaction was reframed into teacher-student(s) interactions in the present study, as the audience specifically refers to the teacher and students in this context. Among them, managing comprehension/channel is used to check "participants' understanding and uptake in relation to the channel" [54] (p. 115). Data from this study indicates that it is employed to manage students' as well as teacher's comprehension, therefore being renamed as managing comprehension. Managing audience discipline includes "cases in which the audience is directly addressed and typically reprimanded or complimented for their metalinguistic behaviour" [54] (p. 116), which mainly refers to managing students' discipline in this study. Anticipating the audience's response, paying special attention to "predict[ing] the audience's reaction to what is said" [54] (p. 116), was reframed as managing students' responses. The final category, imagining scenarios, asks the audience, i.e., the students in this case, to see something from a specific perspective [54] (p. 117). These four types of discourse functions will be explored and exemplified in detail in this research to uncover the functions of personal metadiscourse in classroom teaching contexts.
Irrespective of the prevalence of its application as noted above, Ädel's [24] taxonomy of metadiscourse functions provides a general but relatively abstract categorisation of the discourse functions of metadiscourse. Concomitant studies also largely followed such a categorisation in a quite rigid way [49]. In view of the few studies of metadiscursive use in spoken academic genres, there is a need "for the analysis of metadiscursive acts and their wording to be packaged pedagogically, especially for the benefit of non-native speakers of English" [24] (p. 94). A preliminary survey of the dataset in this research revealed that two major pedagogical functions, namely teacher-students' interactions and teacher discourse organisation, were more prominent in classroom teaching. As such, this study probes into these two functions and addresses the following two research questions:

1.
What are the frequencies and ranges of the personal metadiscourse markers, we, you and I, used by the native English speaking EAP teachers in the UK? 2.
What are the functions of the three personal metadiscourse markers I, we and you in these EAP teachers' classroom discourse?

Context and Participants
This research was situated in the EAP teaching classrooms of universities in the British context, where English was the medium of instruction for classroom teaching. The courses were delivered mainly to international students who learned English as their second language in this context. These courses were designed to develop students' "situated language use" [58] (p. 8) and academic skills such as writing academic essays and listening to academic lectures, in order to undertake academic work as university students [47] (p. 52).
Specifically, four EAP teachers from three universities were selected for the present study. The three universities were all well-known research-intensive universities in the UK. Their names were eliminated due to ethical reasons. Concerning the EAP teachers, all of them were female, with teaching experiences ranging from 10 to 21 years. These teachers were selected based on the criterion that they are native speakers of English. By native, here it means that they were born and grew up in the UK, and speak English at home and school [59], despite that they all had some teaching experience in foreign countries. As for their qualification, they all have a Royal Society of Arts (RSA) diploma in teaching English as a foreign language. That is to say, they all have the qualification of being an EAP tutor. In relation to the composition of their students in these EAP classrooms, due to the increasing number of Chinese students entering the UK for furthering their postgraduate study, a significant proportion of course places are filled by them. These students were in the pre-sessional or in-sessional EAP courses. There were 16 to 20 students in each class, the layout of the classes was arranged in a U-shape, and the class sizes ranged from 5 to 20 students.

Data Collection and Treatment
The data collection began with recordings of EAP teachers' classroom teaching after a systematic process of recruiting related teacher participants. These recordings were transcribed verbatim and then coded with reference to their metadiscursive functions. It is noteworthy that Ädel's [53] four selection criteria of metadiscourse were employed to help identify the metadiscourse items of the personal metadiscourse. These four criteria involve referring to the world of discourse, current discourse, referring to the speaker, and the listener of the interactional situation. Instances such as you when referring to someone in a generic way were eliminated from the metadiscourse markers of this study. Moreover, instances of dysfluency, such as you, when used for twice or more times, were only counted once.
Ultimately, each of the four EAP teachers recorded two sessions, each of which was around 1 h long. In total, the word count was 32,860 words for the transcribed teacher discourse, and 36,276 for the classroom discourse involving both teachers and students. Based on this dataset, we built a self-compiled corpus. In order to calculate the frequency of the teachers' metadiscourse, a separate version, which removed the students' contribution and incorporated only the teachers' classroom discourse, was created. However, it was the corpus that involves both teachers' and students' classroom discourse that were referred to when examining the contexts and functions of metadiscourse markers. This could situate teachers' classroom discourse in the broader classroom interactional contexts, thus facilitating subsequent qualitative analysis of the functions of metadiscourse markers in the related classroom language instructional process.

Data Analysis
This research adopted mixed methods to analyse the results. First, a quantitative approach was employed to present the frequencies and ranges of personal metadiscourse. It presented an overview of the personal metadiscourse use by the four EAP teachers. This was followed by a qualitative analysis of the metadiscourse functions of three personal metadiscourse markers. Notably, this research does not intend to compare individual teachers' metadiscourse use, but to explore the various metadiscursive functions available for teachers in classrooms. This may reveal the variety of the functions of personal metadiscourse markers, thereby providing more insights for researchers and practitioners in this field.
The procedure of data analysis was followed on the basis of research questions. For the first research question, the raw frequencies of the three personal metadiscourse items were generated by importing the coded texts into AntConc and inputting the metadiscourse markers into the search term tool. In order to make all the personal metadiscourse items comparable to each other, the normalised frequency of each item was calculated against per 1000 words.
For the second research question, concordance lines were run to investigate the metadiscourse markers in their immediate contexts. The personal metadiscourse markers were explored for their pedagogical functions in reference to 100 words on both sides of its concordance line. For those items whose pedagogical functions could not be judged by its immediate context, the extended text was referred to for a thorough analysis. Notably, the identification of the functional categories of these personal metadiscourse markers were classified not based on individual metadiscourse markers, but based on a complete pedagogical function as classified by Ädel [24], which was sometimes realised by clusters of you, I, and we. To ensure that the functional categories were accurately recognised, the authors first coded the data separately and then discussed the discrepancies together until all the occurrences were fully agreed upon. In the following three sections, we will first present the quantitative and qualitative results, respectively, and then discuss the major findings as they relate to the existing literature. Table 1 shows the overall frequencies of the three personal metadiscourse markers by individual teachers. These three personal metadiscourse instances were widely used by each of the four EAP teachers in the corpus. Among the three metadiscourse markers, the second person pronoun you was employed with the highest frequency by all of the teachers. This was followed by the self-mentioning I and then inclusive-we by all the other three teachers (T1, T2, and T3), except that of T4, who used inclusive-we with a slightly higher frequency. This may suggest that the three personal metadiscourse markers are common metadiscourse devices used by native EAP teachers to engage international students in their classroom teaching. This result may suggest that EAP teachers recognise the presence of their students, pull them along with their instructions, focus their attention, acknowledge their uncertainties, and involve them in the on-going classroom pedagogical tasks [2]. Such a convergent nature may corroborate previous research and reveal the highly interactive and involving nature of formal instructional settings [2,47,60,61].

Metadiscourse Functions of You, We, and I in EAP Teachers' Discourse
This section illustrates in detail the four major metadiscourse functions of the three personal metadiscourse markers with examples.

Managing Comprehension
The overall function of managing comprehension is to ensure that the teacher and students are comprehensible to each other, or they are "on the same page" [54] (p. 115). An analysis of data revealed that it primarily consisted of two reciprocal activities in this case. The first involved teachers' effort in checking students' understanding (e.g., Does that make sense to you?). The second encompassed teachers' expression of their own understanding of students' utterances, either using a statement to directly indicate understanding or not, or asking for repetition or seeking clarification (e.g., Can you repeat . . . ? Do you mean . . . ?). Compared with checking students' understanding, teachers' expression of their own understanding displays more diversified forms of linguistic representations. This may be due to the fact that teachers sometimes employ some indirect ways to check students' understanding, such as asking students to do some exercises, as is shown in the following extracts (See Appendix A for an illustration of the transcription conventions). Teacher: Ah, okay. You, like a conclusion to the chapter, like a summary and conclusion to the chapter that brings everything together.

Student: yeah.
In Extract 1, the teacher overtly indicated that she had already figured out the students' responses to the pertinent tasks by the statement "I think I know what your answers are". Moreover, the teacher in Extract 2 employed a series of questions to seek clarifications and confirmations from the students to achieve a thorough comprehension. Furthermore, in Extract 3, the teacher endeavoured to express her understanding of the student's response by reformulating and extending the student's contribution. In a sense, such a reformulation, in particular the extension, of students' contribution could also help to clarify learners' thoughts.

Managing Students' Responses
As demonstrated from the dataset, managing responses mainly entails three types of sub-functions, namely: eliciting students' responses, commenting on students' responses, and sometimes anticipating students' responses. Furthermore, eliciting students' responses entails two subcategories: eliciting students' verbal responses and eliciting students' actions. Therefore, managing responses in this research will be investigated from four types of sub-functions: eliciting students' verbal response, directing students' action, anticipating students' response, and commenting on students' response.
Firstly, eliciting students' verbal responses is typically realised by teachers' questions, including display questions and referential questions [62,63]. Eliciting answers from students is an important engagement strategy in classrooms. Data from this study evinced that eliciting answers from students primarily involved three types of functions, namely eliciting answers from an individual student, from a group of students, and from the students as a whole class. This may be affected by specific types of pedagogical tasks, or the locations of the students in a classroom. For example, some group work may lead to the teacher's elicitation of answers from certain groups of students or from a representative student of that group. These functional types of eliciting answers can be illustrated by the following extracts. In Extract 4, the teacher was talking about the use of non-integral citations. She attempted to encourage the nominated student to provide the answers to what follows "the name of the person". Next in Extract 5, the teacher was leading the students in a task of completing the given sentence sections. After allowing them some time to do the exercise, she attempted to elicit answers from the female students sitting together on one side of the classroom as opposed to the male students on the other side.
The second category relates to directing students' actions. It conveys what the teacher wishes the student(s) to do or experience based on the discourse, or refers to cases in which the teacher explicitly comments on the desired uptake. This type of directing students' action accompanied by a personal pronoun is generally indirective and softening in illocutionary force. For instance, Extract 7 presented two directives that the teacher delivers to the students by using the sentence pattern "What I would like you to . . . ". The first one was to require the students to be explicit of which of the two samples is more coherent in structure, whereas the second directive was to ask them to pick out related transitional expressions. Similarly, in Extract 8, after finishing pertinent language points instruction, the teacher asked the students to apply these language points to their previous writings. The ways of expressing directives in these two extracts were delivered in a relatively mitigating manner compared with imperatives, which may help build a relaxed classroom atmosphere and allow space for the students to opt for alternative actions. However, there were also cases in which their speech may have a fairly strong degree of commanding tone. In Extract 9 below, the teacher was talking about the conclusion in writing. She explicitly informed the students of the indispensability of the conclusion by saying "you shouldn't forget about the conclusion". This was followed by her further emphasis of its significance with the attitude marker important. Then in the stage of evaluating the statement, the teacher expressed her prompts strengthened by the modal auxiliary verb "must" to inform the students that they should pay special attention to her directive statement. It was noteworthy that both of the above instances were used to inform students of what is likely to be forgotten. Thus, the teacher adopted an intensifying tone to attract the students' special attention. The third type, anticipating students' responses, typically predicts students' reactions to what is being said, most often by attributing statements to students as potential objections or questions. Let us take the following Extract 9 as an example. The teacher was talking about ways of numbering tables in thesis writing. Then she explained the reasons for adopting a certain method, which may be much easier. She then predicted that this would be something tedious that the student would be reluctant to undertake. This anticipation of potential negative reactions enhanced her argument to adopt the numbering way, which included chapter numbers. In addition, anticipating students' response also refers to predicting some possible questions that students may raise. The fourth category of managing responses concerns commenting on students' responses, involving two aspects in this research. The first relates to commenting on students' answers. This is the typical way of commenting on students' responses, as noted by the traditional initiation-response-feedback (IRF) exchange pattern between the teacher and students. For example, in Extract 10, the teacher was working with the students on an exercise that gave the first letter and asked the students to provide the whole word that went with it. One student offered an answer, but the teacher rejected it as incorrect. Then she explained the potential meaning in its specific context if this answer was accepted. The second aspect referred to commenting on student-initiated questions. In Extract 11, the teacher responded to the student's question with a positive confirmation followed by calling the student's name, a specific form of referring you. She then elaborated her confirmation with an example. The above two examples may reveal that, apart from the simple confirmation marker such as "good" or "well done", the teachers' comments on students' responses are sometimes followed by convincing illustrations of the reasons for these comments. Teacher: Critical, um-, well, in the sense it would be good, but it doesn't fit the sentence. So critical with the real world, doesn't it↑ but saying that, you might say that you want the discussion section to be more critical. Overall, this section illustrated the pedagogical function of representative personal engagement markers in their co-occurrences with other metadiscourse units in managing responses. Overall, four types of pedagogical functions are involved, namely eliciting students' verbal responses, directing their actions, anticipating, and then commenting on their responses. In the following section, attention will be drawn to the functional category of the imagining scenarios.

Imagining Scenarios
Similar to some of the above analysis, imagining scenarios also includes two features. First, it guides the students to see something from a specific perspective, often in a vivid and engaging fashion. It is a strategy of engaging students and can be regarded as a mutual thought experiment between the teacher and the students. In Extract 12 below, which took place at the end of the class, the teacher intended to wrap up the lesson with a metaphor of what they had covered in that session. She employed the imaginary scenario of a Russian Doll to explain the patterns of development in relation to writing. This might facilitate students' understanding of the sequence of structure development related to writing. Second, it also incorporates imagining possible occasions in which students may be involved, as shown in Extract 13. The teacher in this extract presupposed two possibilities related to paraphrasing or summarising and then states the possible actions students may take on each of these two occasions.

Extract 12 (T4)
Teacher: Do you know the Russian Doll↑ . . . this is a metaphor that you can make a concrete understanding of how . . . the supporting details move from the BIGGEST. . . to the smaller, to the smallest one. . .

Extract 13 (T2)
Teacher: . . . if you're paraphrasing and summarising, you don't have to use the page number. You can if you want to, but you don't have to.

Managing Students' Discipline
Managing students discipline refers to cases in which the learners are directly addressed and typically instructed to behave in a certain manner. Extract 14 is a student presentation after some discussion. The student had already begun to answer the question posed by the teacher. However, there was still discussion from other students. The teacher noticed this and thus she tried to draw other the students' attention to listen to this student's utterance. This management of student discipline was fulfilled by the utterance "[right↓] every one of you listening↑ yeah↑" shortly after the student started speaking, marked by the overlap sign of utterance "[ ]". Notably, very few instances of this type were observed in the data. This may be due to the fact that the current research is being undertaken at a tertiary level, which normally needs less effort to manage students' discipline as opposed to those at lower levels (e.g., primary and secondary levels). To sum up, a qualitative analysis of the functions of the three personal metadiscourse makers uncovered four major categories of functions, including managing comprehension, managing students' responses, imagining scenarios, and managing students' discipline. In-depth analysis of the dataset revealed subtler subtypes of functions that these EAP teachers employed in their classroom teaching. Specifically, managing comprehension consisted of a reciprocal process of expressing teachers' understanding of students' utterances or checking students' comprehension of teacher talk or intention. Managing students' responses involved the teachers' management of both verbal and behavioural responses. Imagining scenarios were employed to facilitate understanding and problem solving by engaging teachers and students in a mutual thought experiment. There are also occasions when teachers manage students' discipline, but in very few instances.

Discussion
This research has examined the frequencies and ranges of metadiscourse markers used by four native English-speaking EAP teaches in the UK, and the pedagogical functions of three salient personal metadiscourse markers, I, we, and you used by these teachers in relation to their co-occurrences with other metadiscourse items in context.
In relation to the frequencies of metadiscourse use by EAP teachers, engaging you was employed with the highest frequency by all of the teachers, followed by inclusive-we and self-mentioning I, except for one of the four teachers, who reported using slightly higher frequency of self-mentioning I than the inclusive-we. This observation lends further support to previous research into spoken academic genres [47,61] in claiming that formal instructional contexts in classroom teaching are characterised by "high levels of involvement and interactivity" [60] (p. 102). It echoes previous research in that the second person pronoun you is extensively used by EAP teachers to 'orientate listeners to the discourse and focus students' attention on the topic' [60] (p. 107), or to 'set up pedagogical tasks' [47] (p. 16). More specifically, the finding of this research supports Lee and Subtirelu [47], who found that the second person pronoun you occurs nearly four times as commonly as inclusive-we. Lee [64] also notes that using inclusive-we and engaging you is one of the central ways of establishing and maintaining high levels of student involvement. Additionally, this result also diverges slightly from previous studies that claim that inclusive-we is the major pronoun in spoken academic discourse [65,66]. Taken together, the high levels of interactivity and involvement of EAP teachers' classroom teaching process may also be due to the highly interactive teaching nature of these EAP courses [25], as stated explicitly to her students by one EAP teacher (T3) in her classroom teaching: "This is not a lecture, but more like a seminar".
This study also examined the metadiscourse functions of the three personal metadiscourse markers. Taken together, this research identified four major categories of functions performed by personal metadiscourse markers in their co-occurrence with other linguistic expressions. The first function is managing comprehension, which aims to ensure that the teacher and students are "on the same page" [54] (p.115) in the classroom teaching activities. This study observed two types of managing comprehension. One related to the teachers' effort in checking the students' understanding, and the other concerned the teachers' expression of their own understanding of the students' utterances. The second category involves managing the students' responses, which involves eliciting the students' verbal and behavioural responses, commenting on the students' responses, and sometimes anticipating the students' responses. Among them, eliciting the students' verbal responses was generally realised by the teachers' display or referential questions. Eliciting the students' behavioural response or directing the students' actions was mostly indirective and softening in illocutionary force, but also showed cases with a rather strong degree of commanding tone. Anticipating the students' responses involved occasions of predicting possible objections or questions from students. Managing responses comprised instances of commenting students' answers or student-initiated questions. The third broad category related to imagining scenarios, which may be seen as a mutual thought experiment between the teacher and students. It mainly served to guide students to see something from a specific perspective or imagining possible occasions in which students may be engaged. The last category, managing students' discipline, was observed, but it was rather rare in the present study, which may reveal that the teachers in tertiary level EAP courses require less effort in managing students' discipline compared with those in primary or secondary levels.
Taken together, this research presents a systematic exploration of metadiscourse features of personal metadiscourse markers in classroom contexts. It has enriched current research into personal metadiscourse, extended its application to the classroom teaching contexts, thus adding to our knowledge of the booming research in formal instructional contexts [67][68][69][70][71][72]. Comparing with existing research into the general metadiscourse features of classroom metadiscourse, as mentioned in the preceding literature review section [47,49,61], this study makes a more profound exploration of these features by focusing specifically on personal metadiscourse markers. Such an in-depth analysis can reveal the subtler functions of personal metadiscourse that have not been observed before, such as the subfunctions within the four major functional categories revealed from the qualitative functional analysis. Moreover, it may build up our knowledge of the working mechanism of classroom discourse, and may provide insight to researchers and practitioner in EAP teaching in raising their classroom language awareness and offer affordances to language teacher education programmes.
Furthermore, based on Ädel [24,54] and the dataset in the current research, this study also proposed a fine-tuned functional framework for personal metadiscourse in formal instructional contexts. It focused specifically on the broad category of audience interaction, or what we call teacher-student(s) interactions in this case, allowing for a thorough analysis and illustration of the metadiscourse functions of three salient personal metadiscourse markers. In this study, four major types of metadiscourse functions of these personal metadiscourse markers were identified, which were further divided into nine sub-categories in this current research. This adapted framework of the functions of personal metadiscourse in teachers' classroom teaching can be shown in Table 2 below. The present research to some extent echoes Ädel's [24] call for a refined pedagogical functional analysis of personal metadiscourse markers and enriches the existing research of metadiscourse in educational contexts. In this sense, this study develops the application of a reflexive model of metadiscourse ( [49,[55][56][57]73] by reframing it pedagogically in formal instructional genres.

Conclusions
In a nutshell, based on a self-compiled corpus of eight sessions of EAP teaching, the present study investigated the functions of three commonly used personal metadiscourse markers, i.e., engaging you, inclusive-we, and self-mentioning I. It fulfils its aim of exploring the functions of personal metadiscourse markers in the classroom teaching contexts. The findings indicate that the three personal metadiscourse markers were used pervasively by all the EAP teachers. Among them, the second person pronoun you was employed with the highest frequency, followed generally by the self-mentioning I and then inclusive-we. This distribution corroborates previous research in similar areas [47,61].
More importantly, the current study also identified four broad types of metadiscourse functions of EAP teachers' personal metadiscourse markers, including managing comprehension, managing students' responses, imagining scenarios, and managing students' discipline. It also probed into the scenarios and elaborated the nine sub-categories of metadiscourse functions of three representative personal metadiscourse markers in this specific genre. Observations of this research have both theoretical and pedagogical implications. First, it extends previous research of personal metadiscourse to the classroom teaching context, thereby enriching existing research in classroom discourse. Second, it may build up our knowledge of metadiscourse use in spoken academic genres, in particular in formal instructional contexts. Moreover, this research may also provide insights for EAP-related researchers and practitioners by raising their language awareness, thereby promoting their professional development. It may also offer affordances to language teacher education programmes, and more importantly by providing a fine-tuned functional categorisation of personal metadiscourse markers.
Having said that, the current research is by no means intended to be exhaustive and cover the pedagogical functions of personal metadiscourse in a comprehensive way. Rather, it endeavours to unfold the relatively prominent functions as is observed from the available dataset for the current research. This leads to another issue that relates to the generalisability or transferability of the current research. Due to the comparatively smaller dataset in this study, as compared to, say, those based on the BASE or MICASE corpus [24,49], this research is also not intended to be generalizable to the overall spoken academic genres in EAP instructional contexts. However, it presents a relatively detailed analytical procedure and clear functional framework, which may facilitate interested researchers to make a replication study based on their own specific research contexts and datasets, to judge the transferability of these pedagogical functions in accordance with their particular contexts. Future studies may adopt larger corpora to enhance the generalisability of this type of research and further our understanding of the metadiscourse phenomenon. Institutional Review Board Statement: The study was conducted in accordance with the Declaration of Helsinki, and approved by the Committee for Research Ethics and Governance in Arts, Social Sciences and Business at the University of Aberdeen (15 August 2022).

Informed Consent Statement:
Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available due to signed consent of the research participants.

Conflicts of Interest:
The authors declare no conflict of interest.

Appendix A. Transcription Conventions
The transcription conventions employed in this research were mainly adapted from Jefferson [74] and Sert [75].

Symbol Meaning
(( )) Doubled parentheses shows the transcriber's additional descriptions or comments, for instance, about the features of the contextual information.

↓↑
Upward or downward arrows used preceding a syllable to indicate there is a rising or falling of intonation in it.
[ ] Square brackets around portions of utterances show that these portions overlap with a portion of another speaker's utterance. The left bracket indicates the point of overlap onset, while the right bracket indicates the point at which the two overlapping utterances end.

=
Equal signs indicate no break or gap. A single equal sign indicates no break in an ongoing piece of talk, where one might otherwise expect it, e.g., after a completed sentence. A pair of equal signs, one at the end of one line and one at the beginning of the next, indicate no break between the two lines of talk. : A colon after a vowel or a word is used to signal that the sound is extended. The number of colons shows the length of the extension.

CAPITAL
Capital letters show that the speaker spoke the capitalised portion of the utterance at a higher volume than the speaker's normal volume.