Spoken Metadiscourse in Malaysian ESL Job Interviews

Various job interview skills training have been carried out to prepare students for employment. However, the outcome of job interviews is questionable, as one of the reasons for unemployment among Malaysian graduates is their inability to communicate competently in job interviews. The Malaysia Education Blueprint 2015-2025 (Higher Education) stated that unemployed graduates were also lacking in a strong command of English. Since job interviews are placed heavily on verbal exchanges between the interviewer and job candidate, the role that language plays in interaction is worth exploring. This paper investigates spoken metadiscourse in 16 actual first-stage technical and non-technical Malaysian ESL job interviews. Corpus linguistics was employed for data analysis. An adapted framework of spoken metadiscourse is employed to analyse the types and frequencies of textual and interpersonal metadiscourse in the corpus. The findings revealed that both textual and interpersonal metadiscourse were used, but there are variations in the distribution and composition of metadiscourse in the two categories across disciplines. Excessive use of metadiscourse was found to impede communication flow instead of assisting speakers to be persuasive in their speech. This implies the need for targeted instruction on metadiscourse, specifically among language learners in higher learning institutions to facilitate appropriate usage of metadiscourse in speech.


INTRODUCTION
Malaysia aspires to be a high income nation by the year 2020.To achieve this, the nation's human capital skills and capacity need to be enhanced.However, from a considerable amount of research, it was found that there is a mismatch between employers' requirements of entrylevel employees and the actual skills of the graduates (Krishnamoorthy, 2007;Isarji et al., 2013).This also means that a lot is expected of the higher education system to improve graduates' employability.According to a report by Pillai et al. (2012), more training on the communication skills in English language was needed for university students.This report was based on self-evaluation surveys distributed to public university students and their employers while they undergo industrial training.It was suggested that more training is needed with re-gards to the future graduate's communication skills, an area that has been identified as being a major setback to the Malaysian graduates' employment.
For the fresh graduates, job interviews are their first step to enter the job market.A job interview is a challenging task for job candidates to accomplish and the challenge is greater when it needs to be done in a language different from the candidate's mother tongue.In the Malaysian job interview context, English is the language most often used at multinational companies and government-linked companies (GLCs) job interviews.These organisations generally seek for candidates who are able to communicate competently in English, apart from their technical knowledge on the job.Moreover, Azizan and Lee (2011) mentioned that verbal and written communication skills in English is identified as the most sought-after attribute in prospective employees.Unfortunately, a national study of higher learning institutions commissioned by the Ministry of Education reported that the Malaysian graduates are unable to present ideas and explain issues in writing, verbally, and in group discussions; to write reports, project papers, proposals and minutes of meeting; and to negotiate and convey ideas in planned and impromptu situations (Isarji et al., 2008;Ainol Madziah, 2011).According to Ainol Madziah et al. (2011), the productive skills (speaking and writing) are directly assessed in the recruitment process through essay writing, face-to-face interviews, written and spoken prompts, impromptu speech and oral presentations.Lack of proficiency in these areas may cause difficulty in getting jobs (Malaysian Employers Federation, 2016).
Realising the need to address these issues, the present study seeks to explore language aspects of the Malaysian English as A Second Language (ESL) job interviews, with a focus on metadiscourse.Metadiscourse refers to the linguistic items which serve the textual and interpersonal functions of language (Crismore, Markannen & Steffensen, 1993).Textual functions are employed by speakers to make connections and organise the text, in order to make it explicit and logical to the audience.Meanwhile, interpersonal functions assist speakers to convey their attitude, personalities and evaluation towards the content in the text.Consequently, two categories of metadiscourse are identified, namely textual and interpersonal metadiscourse, both serve the functions aforementioned, respectively.

LITERATURE REVIEW
Metadiscourse researchers (Ädel, 2006;Crismore & Abdollehzadeh, 2010;Dafouz-Milne, 2008;Hyland, 2010) noted that successful communication comes with social impact in engaging with the audience, shaping their propositions to create convincing texts by making language choices in social contexts.The management of such interaction is deemed challenging for second language learners (Gao, 2007;Hu, 2005).However, studies on metadiscourse have been largely focused on written language despite it being acknowledged as an indispensable part of the language (Ädel, 2006;Mauranen, 2010).Considering that there is a strong element of uncertainty and unpredictability in job interviews, speakers should be able to overcome this through employing language strategies such as metadiscourse to enhance clarity and explicitness in speech.When speakers employ metadiscourse in speech, they guide the hearer in interpreting the discourse.This way, meaning in speech is negotiated better.
Metadiscourse refers to how writers or speakers project themselves in their texts to interact with their receivers (Hyland, 2005).Li and Wharton (2012) argued that language users with broad repertoire of metadiscourse may benefit from it since the linguistic resources permit an intentional manifestation of stance in text.In the job interview discourse, metadiscourse may assist speakers to explicitly express their attitude to what is said, for example -by employing expressions signalling doubt (I suppose), or certainty (I'm convinced that).On the other hand, those without such a repertoire may be faced with the constraints of making a stance they would not intentionally have chosen, which may lead to misunderstanding in speech.In communication, both speakers and listeners share the expectation that the understanding of meaning in speech is assisted by the speaker's use of linguistic device, such as metadiscourse, to highlight the relative importance of ideas using 'emphatics' such as certainly, or signal cohesive links between ideas through the use of 'transitional markers' such as so, or but.When these markers are missing, listeners may experience difficulty in understanding the message delivered, and communication problems may arise.
The present study intends to explore spoken metadiscourse in two disciplines, namely the technical (science-based professions) and the non-technical (social science and humanities professions) (Becher, 1994).Previous metadiscourse studies on written language have claimed that writing in different disciplines varies systematically (Hyland, 2000).This means that the language used in a specific discipline will vary from one to another and will reflect the discipline's norms, values and expectations of that discipline because academic disciplines are communities of users whose language vary in their practices.Therefore, text produced by members of disciplinary communities are the concrete realisation of those varied practices.In metadiscourse interdisciplinary studies, Hyland (2005) mentioned that when presenting an argument, the soft-knowledge fields (marketing, philosophy, sociology and applied linguistics) do so more cautiously to promote explicit engagement through the use of hedges such as it seems to me that… or we tentatively suggest… preceding an argument.As opposed to that, the hard sciences (engineering and science) used language in a concrete, impersonal, and value free way to give an objective viewpoint of the facts, so there was less frequent use of hedges in their presentation of arguments.
While these arguments are made on the basis of metadiscourse on written language, such information is yet to be known for the spoken language, since studies of metadiscourse in the spoken language has remained limited in number.Previous studies on spoken metadiscourse have been done on academic discourse, such as lectures and classroom talks.Metadiscourse was found to assist academic lecture comprehension (Pérez & Macià, 2002) and have a positive effect on oral communication (Rui & Xin, 2009).However, the use of metadiscourse in non-academic discourse such as job interviews is yet to be known.Ahmad Tajuddin (2015) noted that having substantial lexical and syntax knowledge to be used in communication helps to contribute to meaningful and effective interaction.It can be implied that metadiscourse is worth-exploring in the context of job interviews, considering that the stakes are high for the job candidates to impress their future employers in the job interviews.Metadiscourse is expected to occur in job interviews since there is a need for careful attention to the verbal exchanges between the candidate and interviewers in order to ensure effective communication between the speakers.

RESEARCH QUESTIONS
The research questions that guide the present study are: 1. What are the similarities and differences of metadiscourse features used in the technical job interviews as compared to the non-technical job interviews? 2. How does the metadiscourse resources function in job interview communication?
Taken together, these questions will allow for a comparison between two types of metadiscourse categories namely textual and interpersonal metadiscourse, as well as shedding some light on the variation of their use across two disciplines (science-based professiontechnical jobs; and social science and humanities profession -non-technical jobs).

METHODOLOGY A CORPUS BASED APPROACH
This study takes a corpus based approach and it involves a specialised corpus.The use of a carefully and strategically compiled specialised corpus makes it possible to overcome one of the criticism corpus linguistics has been faced with, that 'a corpus presents language out of its contexts' (Hunston, 2002:23).With a specialised corpus, a broader explanation of linguistic features and their intended meanings becomes possible since there is a focus on the communicative goals, participants and environment, which are not easily detected in large corpora.Some examples of specialised corpora are the Hong Kong Corpus of Spoken English (HKCSE) developed by the English Department at Hong Kong Polytechnic University, containing approximately 900,000 words across four sub-corpora which includes the areas of academia, business, conversation and public; or Bolton, Nelson and Hung (2002) who study the writing of university students in Hong Kong and the UK.Although such corpora do not provide a basis for generalised claims about language use, they allow for more in-depth investigation of qualitative features with its controllable and manipulatable data size (Flowerdew, 2004).In other words, specialised corpora allow for language patterns that may be specific to the contexts researched to be discovered.
The present study uses a self-compiled corpus, namely the Malaysian ESL Job Interview Corpus, which involves two sub-corpora to be compared -the technical and nontechnical job interviews.The data comes from 16 actual first-stage job interviews, whereby ten job interviews are from the technical discipline (i.e.science-based profession) for the post of Plant Operation Engineer in a multinational company based in Malaysia; and six nontechnical job interviews (i.e.social science and humanities job) for the post of English language lecturers in a public university in Malaysia.A total of 16 candidates and four interviewers were involved.The job interviews were transcribed, ten from the technical discipline and six from the non-technical discipline.Although the number of interviews in each discipline is different, the corpora are comparable in terms of size since the total number of words in each sub-corpora is almost similar.The numerical description of the data is shown in the table below: The technical job interviews are conducted on a one-on-one basis, with one interviewer and one candidate in one interview session, while the non-technical job interviews are done via panel interview, with one job candidate and three lecturers as the interviewers.In the corpus, the names of the speakers are indicated by the abbreviations according to their respective disciplines.The candidates in the technical discipline are assigned with the code 'T' (for Technical), while the candidates in the non-technical discipline are given the code 'NT' (for Non-technical), followed by the number assigned for the candidates; T1 refers to Candidate 1 in the technical job interview, and NT2 is Candidate 2 in the non-technical job interview.Meanwhile, the interviewers are assigned with the code 'IV'.For the technical job interview, 'IV' is the code used.Meanwhile for the 3 interviewers in the non-technical job interviews, they are assigned with the codes 'IV' followed by the numbers; IV1, IV2 and IV3.
The technical job interviews are obtained from the 'on-campus job interview' conducted by Company A, a multinational company based in Malaysia.Only the first-stage job interviews are conducted on campus, and the permission to collect the audio-data was granted by the Human Resource Manager on the day the job interview was conducted.The nontechnical job interviews are interviews for a full time position as English lecturers in University A, a Malaysian public university.Both the technical and non-technical job interviews are first-stage, entry level job interviews conducted at University A. Since these interviews are actual job interviews, the interview sessions strictly followed the pre-determined questions set by the two organisations involved in the study.

DEVELOPING METADISCOURSE FRAMEWORK FOR ANALYSIS
The present study referred to the trends in metadiscourse studies which are important in selecting and developing appropriate model for analysis.The metadiscourse frameworks introduced by Vande Kopple (1997), Crismore, Markannen and Steffensen (1993), as well as Hyland (2005) were used as a guide to specify the textual and interpersonal functions of metadiscourse, as can be seen in Table 2 below.Previous researchers such as Vande Kopple (1997) and Hyland (2005) have frequently accentuate that a metadiscourse function should only be identified in context, emphasising that 'there are no simple linguistic criteria for identifying metadiscourse'.For example, Hyland (2005) listed linguistic items that functions as transitions, but emphasized that the items may not always function as such, and that the metadiscourse function of transition could be realised by an item not in his list.Working from this, the present study embarked on an exploration of metadiscourse in spoken texts with the expectation that other language forms (other than those which have been identified in previous metadiscourse studies), and with the awareness that ESL speakers in job interviews may use metadiscourse inappropriately.
According to Suau-Jiménez ( 2016), studies on metadiscourse on academic genre have challenged the previously established metadiscourse frameworks, as well as argued for flexibility and evolution of the existing frameworks.Non-academic genres generated by professional communities and non-academic disciplines also shed new, potentially enriching perspectives on the metadiscourse framework.Therefore, Suau-Jiménez (2016) proposed that a reframing or at least an adaptation would be needed to cater for new genres in a variety of genres, domains or even language.Hence, the work of the present study has made some minor modification to the previous metadiscourse framework to cater to the spoken discourse understudied.More specifically, the subtypes of commentaries and attitude markers under interpersonal metadiscourse are merged, unlike the previous frameworks (Hyland, 2005;Vande Kopple, 1985;Crismore et al., 1993) which have categorised them separately.
Vande Kopple (1997: 8) mentioned that commentary draw readers into an implicit dialogue by 'commenting on the reader's actual or hoped for stance' through the use of phrases such as you might not agree that or you might want to skip to the next chapter.Meanwhile, attitude markers are the way writers display their attitude towards the text (certainly, surprisingly).It has been discovered that in spoken discourse, the speakers' attitude can be expressed through direct comments made by the participants in the interaction (Aguilar, 2008).In the present study, commentary is understood as a category under the interpersonal metadiscourse resources in which the speakers comment on the importance, relevance or difficulty of an assumption (evaluative), express the speaker's mood or feelings towards the content or situation (attitudinal) and as a way to appeal directly to the hearer (dialogic).In the following table, the spoken metadiscourse framework that is used to analyse the corpus in the present study is presented.As illustrated in Table 3, the purpose of textual metadiscourse as linguistic resources is to help navigate the listeners through the text, which allows for a more cohesive and coherent text.Similarly, textual metadiscourse paves the ways for speakers to better present their ideas.The use of metadiscourse addresses the means of organising spoken discourse to meet the expectations of the speakers in a conversation.In communication, speakers and listeners share an expectation that listeners are aided in the interpretation of the message by the use of the speakers' contextualisation markers, such as the textual metadiscourse subtypes as seen in Table 3.Under the category of textual metadiscourse, there are several subcategories such as text connectives (transitional markers (additives, causatives and contrastive), topicalisers, sequencers, reminders and illocutionary intents), code glosses, and relators.The second type of metadiscourse is the interpersonal metadiscourse, which is used to convey emotions or reactions to the propositional content and to build a personal rapport with the listeners or readers.To use this interpersonal linguistic device, speakers in job interviews may use hedges, emphatics, evidential and commentary.As mentioned in the earlier section, the categorisation of textual and interpersonal metadiscourse in the framework used in the present study is being based on previous established metadiscourse frameworks such as Vande Kopple (1985), Crismore et al. (1993) and Hyland (2005).Since these frameworks are based on written text, the examples were adjusted to suit the spoken discourse examined in the present study.

DATA ANALYSIS PROCEDURES
The first step in the data analysis procedure is the metadiscourse analysis.A metadiscourse can range from a word or short phrase -usually with conjunctions (just now) and modal adjuncts (certainly, maybe) -to a complete clause or sentence.For example, sequencers can be identified from word sequence (first, next) or statements that serve as a reminder (as I mentioned earlier).Therefore, a wordlist was created using the WordLister, a tool for analysis in the corpus software; WordSmith Tools (Scott, 1998).The metadiscourse framework in Table 3 is used as a guide in the identification of metadiscourse in the text.The corpus was searched electronically based on the suggested words in each of the metadiscourse subtypes of both textual and interpersonal metadiscourse.WordSmith's Concord tool locates the metadiscourse resources in the corpus and displays them in standard concordance lines: the search word is centered (the node) and its surrounding line.The concordance search makes it possible to reduce misinterpretation of the searched results.Next, the instances of the metadiscourse items are manually examined, in order to determine whether they function as metadiscourse or not.Items that did not function as metadiscourse were then excluded from the concordance list in WordSmith Tools 7.0.This combination of automatic and manual analysis has also been employed in the works of Dafouz-Milne (2003) and Ädel (2006) among others.Following the quantitative exploration, the corpus is analysed in detail to explore the functions of metadiscourse.This means that metadiscourse resources are analysed from the quantitative and qualitative approaches.

SIMILARITIES AND DIFFERENCES OF METADISCOURSE BETWEEN DISCIPLINES
In this section, comparisons between the two disciplines in their use of metadiscourse are presented.Table 4 shows that the speakers in the technical job interviews used metadiscourse more frequently than the speakers in the non-technical job interviews.Both types of disciplines used more textual metadiscourse than interpersonal metadiscourse, but there is a difference in the respective proportions.The technical discipline's use of textual metadiscourse accounts for 78.25% use of total metadiscourse, as compared to 75.07% in the non-technical texts.The use of interpersonal metadiscourse in the technical discipline is 21.75% and 24.63% in the non-technical texts.This suggests that the non-technical discipline employs more interpersonal metadiscourse than the technical discipline.
It was found that out of the ten subtypes in the textual metadiscourse, the mean frequencies of eight subtypes, namely Causatives, Additives, Contrastive, Toplicalisers, Illocutionary intent, Sequencers, Code gloss -Exemplifiers and Code gloss -Explanation were higher in the technical disciplines.However, only four subtypes were statistically different between the two disciplines, namely Causatives, Sequencers, Reminders and Illocutionary Intents.Reminders is the only subtype which is statistically different and with higher frequencies in the non-technical discipline.Meanwhile, in the interpersonal metadiscourse, five out of eight subtypes (Emphatics, Evidential -Personal belief, Evidential -Deduction markers, Commentary -Evaluative and Commentary -Attitudinal) are with higher frequencies in the technical job interviews.Out of the eight subtypes, only two subtypes show statistically significant differences between the two disciplines, namely Hedges and Commentary -Dialogic.The mean frequencies of these two subtypes were higher in the non-technical job interviews as compared to the technical ones.In both disciplines, transitional markers (additives, causatives and contrastive) are the most commonly employed textual metadiscourse in the data, with a smaller proportion of total metadiscourse in the non-technical discipline than in the technical discipline as seen in Table 4.This indicates that the speakers in the technical job interviews used more transitional markers (so, but and and) per 1000 words than their non-technical counterpart.The high frequency use of these transitional markers is found to show that there is a pattern of 'surface binding' (Latawiec, 2012) by the speakers in job interviews.This results in a greater cohesion in their talk since ideas are arranged and managed through signaling between clauses.Transition marker -additive is found to be the most used transitional marker among the three.However, a detailed observation on the specific transition markers which have the tendency to be chosen more often shows that the speakers in the technical discipline tend to use the marker so and as transition resources, though may not be using them appropriately.Some uses of so in the data function as 'pseudo-bridges' rather than true causals or additives, which is used for chaining or simply adding new information when describing their personal information and working experience, and is found to be common among English language learners (Latawiec, 2012), as seen in the following excerpt (addressed metadiscourse subtype is underlined): Excerpt 1: As seen in Excerpt 1, so is identified as pseudo-bridges when the job candidate (T5) reverts to using causatives as a strategy to 'cover-up'; masking his imperfection of speech as he presents information and communicates ideas (lines 2, 5 and 8).When used inappropriately, so in this excerpt made the candidate's utterance seem incoherent.This carries important implications in job interview speech, whereby candidates should be cautious in their choice of words to ensure clarity in their speech.
Meanwhile, there is a striking difference in the use of sequencers mainly between the two disciplines, which occurs 52.64 times per 1000 words in the technical texts as compared to only 6.01 occurrences per 1000 words in the non-technical texts.In the corpus, sequencers are typically marked by numerical sequencers such as first, second and adverbial sequencers such as next and after that.The speakers in the technical job interviews apply sequencers in their speech significantly more, which means that they have the tendency to construct their speech in a sequential manner in explaining their arguments and ideas as compared to the speakers in the non-technical job interviews.When speakers in the technical job interviews adopt sequencers, it appears that they expect the listeners to accept and follow the argument in a sequential manner.Additionally, the data shows that sequencers are less employed in the non-technical job interviews, which implies that there is a higher expectation of listenerspeaker interaction in the non-technical job interviews, considering that they have a higher level of proficiency in using English language in communication.
Another notable difference in the use of textual metadiscourse between the technical and non-technical texts is in the use of reminders.Reminders, which is used to relate an idea in the ongoing explanation to the some ideas that were discussed earlier using phrases like like/as I said earlier and you mentioned/ said that or words with temporal meaning such as just now, are used often in the non-technical texts, and rarely occurs in the technical texts as seen in the excerpt below: Excerpt 2: In Excerpt 2, the interviewers ask Candidate NT2 to clarify if it is okay for her to be placed in other campuses other than the one she applied for.Candidate NT2 mentioned in lines 6-7 that "even in Kedah I'm to go, what more in Selangor".When Interviewer IV1 responds by saying "So you can go anywhere" (line 12), Candidate NT2 agreed by saying "Yes" in line 13.However, in her subsequent utterances, she said that "I'm hoping to have one settled place" which then Interviewer IV1 interjects at line 12, reminding her of her earlier statement that she did not mind being placed anywhere (line 16).
It is also observed that the conversation in the non-technical texts are less linear compared to the technical ones, in the sense that there is a reference to different parts of topics discussed throughout the interview, hence creating a path for better negotiation of meaning between the speakers.An example is shown in the following excerpt: Excerpt 3: Yes the main campus is in Kepong.They changed from PJ to Kepong.

IV2:
Yes it used to be near my house @@@ 4 IV1: @@@ But [College 2] is a nice place… 5 NT3: [College 2] is a nice place… yes… but as I said… 6 IV2: It doesn't help you grow… yeah… (Non-technical Job Interview -3) Excerpt 3 illustrates an example of reminder used by Candidate NT3, using the phrase "as I said" (line 5) in the attempt of reminding the interviewers of what she had mentioned earlier.Interviewer IV2 understood her intention and completed her utterance on behalf of her by repeating her exact sentence in the earlier part of the interview, "It doesn't help you grow" (line 6).This example also illustrates the nature of job interview conversation in which collaborative completions and co-construction of speech normally occur (Kerekes, 2007;Lipovsky, 2010).
In the technical job interviews, the conversation is more linear.The findings revealed that after a topic has been discussed, the speakers move on to another topic, and it is very unlikely that the previous topic is revisited.As opposed to that, the speakers in the nontechnical job interviews made reference to other parts of the job interviews which have been previously discussed.Therefore, the use of reminders is more common the non-technical discipline.

INTERPERSONAL METADISCOURSE RESOURCES
In the interpersonal metadiscourse, the subtypes of hedges and emphatics perhaps are the most interesting to be discussed.While the non-technical texts were found to use more hedges (18.59) as compared to only 11.52 in the technical texts, the use of emphatics appears more in the technical texts with 39.18 occurrences per 1000 words as opposed to a mere 17.23 occurrences in the non-technical texts.With regards to credible appeal of the speakers, the use of hedges by the non-technical speakers have assisted them in building a modest and considerate profile, while their technical counterparts are more likely to express certainty show to their stated propositions to build up a convincing profile to their roles as job candidates and to the interviewer through the use of emphatics.
Speakers in the non-technical job interviews employed hedges through the use of modal auxiliaries such as can, could and would to mitigate their claims, as seen in the following ecerpt: Excerpt 4: In Excerpt 4, Candidate NT5 is describing her personal attributes whereby she acknowledges herself as being emotional (lines 3-4).The use of various modal auxiliaries such as can, could and would refers to matters of personal beliefs, that is, on how she perceives herself as an individual, which also serves as a basis for Candidate NT5 to express judgments about how she handles students in the institution she is currently teaching at.At the end of her utterance, Candidate NT5 opted for hedges would to tone down her statement when she mentioned that she is "the person who best fit that particular position" (line 10).Similar examples can also be seen in the interviewers' utterance, especially when discussing the nontechnical candidates' job placement, for example, "you might probably be placed in other campuses other than you have applied for", and "we could have a place in another campus".The use of hedges in these excerpts demonstrate that the speakers are showing detachments to their claims, that is, the final decision of candidate's placement is not decided by the interviewers.
Other than that, hedges are also used by speakers in the ESL job interviews through the use of probability adverbs such as maybe.The use of hedges is common in the candidates' self-description, for it appears that the candidates are attempting to be strategically polite in their responses.For example, when asked "How do you see yourself in five years?" -this was the candidate's response: Excerpt 5: (Candidate T1, Technical Job Interview) Meanwhile, the technical speakers employ more Emphatics in their speech.They are able to display their commitment to the stated propositions, by showing that they are fully responsible to what they say, through the use of high commitment modals, such as must, should, need (to) and have (to); emphatic adverbs which includes definitely and obviously; and emphatics which mark solidarity with the audience, such as as we know.An example of how emphatics are used in job interview speech is presented in the following excerpt: In five-years-time… I see myself as an experienced engineer… where I have the skills and also maybe responsible to monitor new engineers.And also… maybe… I can venture into R & D… research and development.
Excerpt 6: In Excerpt 6, need to as used by the speakers reflects a sense of strength in their claim because the topic addressed in this example requires improvement on the part of the candidates.Meanwhile, must is used to show that there is a need to undertake such an action, namely "displaying set of skills" (line 6).The use of must also revealed the speaker's desire to attract the candidate's attention to the interviewer's stance or opinion, serving as a guide to the candidate in improving herself for the next interview.
These findings suggest that the speakers in the technical job interviews are more content-oriented.This is also reflected through the high use of sequencers such as first, next, and after that (52.64 occurrences per thousand words, as compared to only 6.01 in the nontechnical discipline), which displays the speakers' preference to arrange information in speech step-by-step.It implies that speakers in the technical discipline expect the listeners to accept and follow their argument in steps.Therefore, it is not surprising that the language used by speakers in the technical discipline lacks the use of hedges (11.52 occurrences per thousand words as compared to 18.59 in the non-technical discipline), for they rely on being direct in speech.Meanwhile, speakers in the non-technical job interviews are assumed as expert users of English as a second language, thus, their familiarity with the language assisted them to engage in dialogic exchanges which promotes engagement between the speakers.By showing uncertainty of one's claim, speakers are able to build up a cautious and humble image, while at the same time maintain the cooperativeness between speakers in the job interview conversation.This is evidenced with the use of hedges in the following excerpt: Excerpt 7: (Candidate NT1 and Interviewer-2, Non-technical Job Interview) In Excerpt 7, the interviewer comments on the pair of shoes that Candidate NT2 put on for the interview.The use of hedges in this example has assisted the speakers to promote positive attitude towards the content.The use of quite, would and probably in this excerpt has led to Candidate NT2's acceptance that "first impression is very important" (line 8-9).The collaborative completion of sentence, "Sometimes the type of shoes will reflect the …" (line 7) and Candidate NT2's interjection in line 8 "I believe so" illustrates that the speakers understood the necessity of discussing the issue and the near-criticism was taken positively by Candidate NT2.
1 2 IV1: We were quite taken aback… 'Eh?She's wearing sneakers?' @@@ And this is an interview… 3 NT2: No no no.It's very formal actually.It's a formal shoes.Although Hyland's (2005) research is on metadiscourse in academic research articles, his findings revealed that the soft knowledge fields express arguments more cautiously, since hedges is strongly represented in their writing.The findings of the present study concur with this, for the qualitative analysis revealed that speakers in the non-technical job interviews employed a wider range of items and functions of interpersonal metadiscourse in job interviews, as compared to their technical counterparts (Table 4).
In the present study, Commentary is employed by speakers in job interviews to display their involvement in the assertive claims that they made.It is further categorised into three subtypes namely evaluative, attitudinal and dialogic.It was discovered that these specific subtypes revealed power differences between the interviewers and candidates, in which the use of commentaries are found more in the interviewers' speech than in the candidates'.Commentary -Evaluative in the present study is mostly used by the interviewer, in the form of making remarks such as 'it seems like you understood the process very well', or giving specific information such as 'must be a mistake there, yeah?'.This interpersonal metadiscourse often occurred when the interviewers provided feedback to the candidates based on their responses to the questions posed.Commentary -Evaluative is used the most in both technical and non-technical disciplines.
There is a striking difference in the mean frequency of Commentary -Dialogic, with a higher occurrence in the non-technical discipline.Commentary -Dialogic is used by speakers in three ways: i) by means of markers like you know; ii) by means of direct questions (am I right?I think you know that, right?); iii) by means of imperatives (Wait).In the present study, Commentary -Dialogic is found to occur in instances when speakers are giving advice or making self-justifications, and in general reflects the shared knowledge between the speakers.The higher frequency of Commentary -Dialogic in the non-technical job interviews seems quite justifiable considering that there are differences in the two disciplines in making direct appeals to their audience.In the non-technical job interviews, the speakers project their expression of their ideas through dialogic engagement with their audience.Commentary -Dialogic is also used in reference to the speakers' parity in terms of shared knowledge which the speakers assume the audience may share (Aguilar, 2008), (you know that; I know that you've explained; I think you know that too, right?).By signaling that they are aware of their audience's knowledge, the interviewers are able to maintain politeness while giving criticism and self-justifying their statements.Hyland (2005: 145) argued that the softknowledge fields rely more on a dialogic engagement and more explicit recognition of alternative voices than the hard sciences.This seems to support the findings of the present study, whereby the speakers in the non-technical discipline are found to be making direct appeals to their audience as they deliver their ideas.

HELPFUL AND IMPEDING USES OF SPOKEN METADISCOURSE
In this section, excerpts from the corpus are closely analysed in order to explicate the uses of metadiscourse in the job interview speech.The following excerpt is taken from a stretch of talk in the probing stage of the job interview, whereby the candidate was asked to justify her reasons for joining University A after deciding to leave her position in a corporate organisation.As seen in Excerpt 8, Interviewer IV1 uses textual metadiscourse in the form of reminders such as 'you said that', and commentary -dialogic 'I know you've explained that', which contributes to building coherence in speech as the earlier parts of the discussion is revisited.By signaling that they are aware of their audience's knowledge, the interviewer was able to maintain politeness by acknowledging that the candidate has mentioned about the addressed issue, while giving criticism and self-justifying her statements.However, there are instances where metadiscourse only adds to the wordiness and repetitiveness of talk, and therefore, im-pedes rather than helps the flow of propositional meanings, as can be seen in line 6, with the repetitive use of sequencer 'and then'.In Candidate NT1's turn (line 11), she begins with the topicaliser 'well', which according to Vande Kopple (1997), focussed attention on the phrase that indicates the shift, reintroduction, changes in topic, or is brought up 'to set a particular contrast in stark relief' (p.2).Her initial position in line 13 'And I would say that, yes' to express favour of interviewer IV1's assumption.This is then followed by her own justification signalled by the contrastive 'but' and additive 'and', which helps shape the discourse and reasoning.It is evident that the flow of propositional meaning is evidently facilitated by the textual metadiscourse.The use of interpersonal metadiscourse in Candidate NT1's turn includes a few instances of hedging, as in 'I would say that' in line 17 as evidence that Candidate NT1 is cautiously expressing her thoughts to avoid from being perceived as lacking in handling workplace stress.In this example, the use of both textual and interpersonal metadiscourse contributed to speakers handling the balance between objectivity and subjectivity in speech.
In the job interview corpus, speakers mostly use transitional markers to add items to their list of propositions through the use of and or also.Additionally, in an attempt to maintain the discourse flow in the temporal or sequential rhetorical pattern, speakers in both technical and non-technical job interviews employed the metadiscourse item 'so' to structure their speech, as seen in the following excerpt: In Excerpt 9, Candidate T5 uses so for any additive relation, in fact, some of the uses of so are in place of and, and in result binds the stretch of talk in a more additive-like manner, instead of functioning as a causative device.The sequence signals (such as those of globally binding sequencers first, then, next) seem to lose their potency, as the listeners realise their additive or even coma-like role.It was also found that and then is used in a similar way, which leads to their 'de-ranking from more globally binding sequencers to less-global as inter-clausal binding' (Latawiec, 2012, p. 128).This also means that the sequencers were functioning only at surface level instead of building coherence with other parts of their speech.The phenomenon illustrated in Excerpt 9 has been found in studies of pseudo-bridging 'so' uses (devoid of causative function) or in studies of narrative cohering in African-American writing or speech with frequent and followed by then (McCarthey, 2002).
The findings revealed that the use of metadiscourse in the corpus of Malaysian ESL job interview vary in terms of the metadiscourse categories, namely textual and interpersonal metadiscourse across the two types of job interviews (technical and non-technical).In general, metadiscourse is employed more frequently in the technical job interviews than in the non-technical job interviews.Additionally, the high frequency of transitional markers (additives, causatives) and sequencers under the textual metadiscourse category bear persuasive force, since the speakers in the corpus employ these metadiscourse subtypes to guide the listeners to follow the logic of the ongoing discourse.The findings illustrate the nature of job interview discourse which involves narrative and persuasive verbal strategies (Kerekes, 2007).Meanwhile, in expressing arguments, speakers in the non-technical discipline employ interpersonal metadiscourse such as hedges which allow speakers to express ideas more cautiously.As opposed to that, the speakers in the technical job interviews use emphatics to display their commitment to the stated propositions, instead of using hedges.These findings support Hyland's (2005) claim that metadiscourse is used differently in different disciplines.It was also found that excessive use of metadiscourse is evident in the speech of the Malaysian ESL speakers in job interviews.Although not affecting the text comprehensibility, excessive use of metadiscourse is seen as impeding propositional flow.

CONCLUSION
The findings on metadiscourse items identified in the corpus revealed that the proportion of the textual metadiscourse used is higher than the interpersonal metadiscourse across disciplines, with interpersonal metadiscourse usage being higher in the non-technical disciplines than technical ones.Within textual metadiscourse, transitional markers are consistently the most frequently used.However, the high occurrence of transitional markers is also associated with its use as pseudo-bridges.Meanwhile, the use of interpersonal metadiscourse is rather limited as compared to the frequency of occurrence of the textual metadiscourse.Therefore, some possible pedagogic implications can be applied for training the Malaysian ESL speakers to make effective use of metadiscourse, both the textual and interpersonal, when speaking in English in job interviews.
The present study has contributed to the specific areas in job interviews training.Since the Malaysian ESL speakers are found to be relatively comfortable in using textual metadiscourse, they may benefit from the teaching of interpersonal metadiscourse in spoken discourse.The corpus used in the present study serves as a good resource for the actual uses of metadiscourse in speech.Additionally, examples of metadiscourse in the corpus can be utilised in awareness-raising activities as an initial step to help learners grasp the significance of metadiscourse use.
While it is acknowledged that English for professional communication courses offered in higher learning institutions have emphasized on various language and communication skills for employment, the present study suggests that metadiscourse can be considered an area of focus to prepare future graduates for job interviews.Metadiscourse can be taught while students are introduced to language in job interviews, as the audio or video data of job interviews may accompany the concordance lines and students can be made aware of metadiscourse uses in contexts to observe the specific stages and topics in the job interviews setting, so that it would be easier for them to internalise this language device.
The spoken metadiscourse framework used in the present study is found to be useful in the exploration of spoken metadiscourse in the Malaysian ESL job interviews.The findings revealed that some of the more detailed categories such as commentary has revealed not only similarities and differences between the two disciplines (technical speakers had a narrower range of resources to realise this subtype), but also power differences in the use of this particular subtype (it is used more in the interviewers' speech than in the job candidates').Therefore, the adaptation and alteration of widely accepted metadiscourse frameworks seems to be valuable to account for specific sets of data (Li & Wharton, 2012).Since studies on metadiscourse in spoken texts have been limited, more studies on spoken metadiscourse related to workplace communication should be conducted in preparing future graduates for employment.
. We just want to highlight this… just in case… you know… you get into the second level of the interview… you would probably have to walk in the room with your pair of shoes.Sometimes the type of shoes will reflect the… 8 9 NT2: <=> I believe so.I believe that first impression is very important.Yes.
during the industrial training… aaa… I was in-charge… to assist the plant manager to manage the plant operation.So… we are… the company is… aaa… provide aaa… the site… construction site with the mix… mix… [sic] ready-mix concrete.So… aaa… the production is based on the daily order.So I assist in the plant operation by… by doing the control system… aaa… to mix all the ingredients… aaa… such as (XXX) GGBS… OGB… and other ingredients including the chemicals… to aaa… to get the… aaa… concrete as satisfied by the customer.Besides that… I also do… I also did safety… safety… plant safety… checklist every day.So… it is included safety on the equipment… and also… plant truck… mixing truck.(CandidateT5, Technical Job Interview)

TABLE 3 .
Framework of Spoken Metadiscourse Framework in the Malaysian ESL Job Interviews

TABLE 4 .
Metadiscourse use in the technical and non-technical job interviews I can… I'm emotional.I can feel for people.And that's where… I can connect with these weak students.You see when weak… weak students they don't like to talk much especially in English class… especially Chinese students.They're very quiet and all.So when we could approach them… talk to them… that's where we can see that they actually need help in certain parts.That's why my boss said that I would actually… I would be the person who best fit that particular position… right now.
(Candidate NT5, have a high level of confidence… you are able to speak in English… and I can see that you are a focused person.But… in interviews… especially you as a fresh graduate… you need to display your set of skills.You must show it clearly.This part… you were not able to tell me clearly. And I would say that, yes… being in the corporate world is more glamor… they pay you more… plus the bonus and everything.But the stress is there.And for me to cope with the stress level… with my condition right now that I'm married… with commitments… I would say that the stress… I bring the stress to home.I don't want that to happen… I mean… prolong.(CandidateNT1 and Interviewer 1, Non-technical Job Interview)