The Impacts of Foreign Language Anxiety on Students in English-Medium Instruction Classes in the Central Region Universities, Vietnam

This study investigates the anxiety frequency of students in English-medium instruction classes and its relationship with gender and English performance. The questionnaire was administered to 300 students in three universities in Central Vietnam. The follow-up interviews and classroom observations were conducted to obtain more insights into the research problems. The results revealed that the EMI students had moderate levels of anxiety. The most provoking causes of students’ anxiety were worries about making mistakes, inconfidence during speaking tasks and fears of negative judgment from peers and teachers. The analyses of the variables related to foreign language anxiety revealed a significant negative correlation between students’ foreign language anxiety levels and their English competence. No statistically significant relationship was found between male and female students. The study provides several pedagogical insights for curriculum designers, educators and teachers to improve the quality of English-medium instruction courses in the Vietnamese context.

findings indicated that those EFL tertiary students considered FLA a severe problem in their English learning, and most expressed a negative attitude towards this trait. Meanwhile, their teachers did not believe in the severity of FLA at the same level.
Another study by Tran (2012) and Tran et al. (2013) investigated both EFL teachers' and students' origins of anxiety and the impacts of FLA on EFL learners. Through the analysis of interviews and autobiographies, it was found that learners themselves, compared to the other three factors, included teachers, external factors and English complexity, were the primary sources of FLA. The result also showed that participants in the study experienced more negative effects of FLA than positive ones, and these types of influences both depended on five elements: "the level of EFLA; the learning attitude, the personality trait, the learning goal, and the awareness of the importance of English." (p.120). From the findings, the researchers called for more attention regarding teachers' awareness and educators' management to improve the English language learning process.
Although FLA has attracted many researchers worldwide, this field of interest needs to be more adequately addressed by Vietnamese researchers. As the review indicated, only two related studies have been conducted by the same research group (Tran et al., , 2013. Moreover, these studies merely emphasized general perceptions of FLA from teachers' and students' sides. None of the revised-above investigated the relationship between their FLA level and English proficiency. Besides, the participants involved in most of those studies were college and tertiary learners studying English as a foreign language, not as a medium of instruction; those in EMI courses in this study seem unexplored. This study was the first attempt to fill this gap.

Research design
Since this study aims to determine the anxiety level from EFL students' perspective and the relationship between FLA level and EFL students' English achievements, an explanatory sequential design is found to be pertinent to this study. Creswell and Creswell (2018) define this type of research design as follows: "The explanatory sequential design is a mixed methods design in which the researcher begins by conducting a quantitative phase and follows up on specific results with a subsequent qualitative phase to help explain the quantitative results. The qualitative phase is implemented to explain the initial results in more depth, and the name of the design -explanatory reflects how the qualitative data help explain the quantitative results." (p.114) Accordingly, the study started with a quantitative survey (FLA scale) and students' English grades. The findings from the quantitative data then helped inform qualitative data collection and analysis through interviews and classroom observations. Then, the theming, connecting and contrasting of this study's quantitative and qualitative findings supported the extension of the understanding of secondary school FLA levels and the relationship between the level and grades.
In this sequential design, the priority was given to qualitative data collection and analysis. According to Creswell and Creswell (2018), in a sequential explanatory design, attention can be given to either a quantitative or qualitative approach, and this may depend on the researcher's interest, the study's audience and what researchers aim to focus on in their research. As the objectives of this study were to seek out EFL secondary school students' FLA level and to explore the potential relationship between students' English scores and their anxiety scale, the weight was put on the qualitative approach, even though it was the second phase of the research process. The quantitative approach allowed the researcher to develop a general understanding of the relationship between the two mentioned variables. The qualitative approach, with students' interviews and teachers' classroom observations, helped explain the quantitative phase results to understand the research problems better.

Participants
The study involved 300 students, 146 males and 154 females, from three universities in Central Vietnam. The students aged from 19 to 21 participated in EMI courses. After the FLA scale was administered, two groups of 12 students (24 in total) were purposefully selected to be observed during their learning process in the classroom and then invited to participate in the focus group interviews later. From the questionnaire analysis, the extreme sampling technique used to select the interviewees and groups to be observed was driven by the survey scores; more specifically, the sampling had to make sure that each group had at least one student from the highest anxiety score group (mean score is above 4.0) and one student from the lowest anxiety score group (mean score is under 2.0).

VIETNAM JOURNAL OF EDUCATION
 106  Horwitz's (1986) FLCA scale was used in the questionnaire to identify the EFL participants' anxiety levels. It consists of 33 statements that measure the frequency of 3 types of anxiety: test anxiety, communication apprehension and fear of negative evaluation. The questionnaire consists of two parts; the first is to collect the students' demographic information, and the second includes 33 statements, constructed on a 5-point Likert scale, from never (1) to always (5). For the statements, the scale was inverted for ease of later calculation. To support the students' comprehension of the questionnaire, the FLCA scale was translated into Vietnamese. The survey took students about 20 minutes to complete. Before completing the FLCA questionnaire, the participants agreed to sign the informed consent for the study.

Research Instruments Foreign Language Classroom Anxiety (FLCA) Scale
This tool was chosen for many reasons: First, it was elaborated by carefully analyzing different aspects of language anxiety in the classroom, which is considered appropriate for this study. Second, it offers numerous advantages, as summarized in He (2018), including (1) assessing the "orthogonal type of anxiety"; (2) describing the "situation of interest thoroughly" for the respondents; and (3) requiring the respondents to refer to specific sources, which then help gain more understanding of anxiety. Furthermore, this scale has been widely used in previous studies, as revised in the literature, and its reliability was confirmed through different tools. Finally, this study chose this scale because, as mentioned by He (2018), "a situation-specific perspective provides the best approach to the research on the relationship between FLA and foreign language learning" (p. 27).

Focus group interviews
The focus group interview was planned after the FLCA results had been identified to deepen data from the FLCA survey. This tool was applied to the student groups. Five student groups were established, and each group consisted of six students as Dornyei (2007) recommended that fewer than six participants in each group would limit the "potential of the 'collective wisdom' whereas too large size makes it difficult for everyone to participate" (p.144). This semi-structured interview gave the researcher more room to deepen the information obtained. The Facebook messenger tool was used in this focus group interview because it was more convenient compared to other types of interviews in many ways, including (1) Time and cost saving; (2) Convenient meeting time arrangement; (3) Participant comfortability; (4) Continuing elicitation.

Classroom observations
The structured observation was employed in the follow-up stage of the study to find out how EMI students behaved during their learning process. The observation was conducted in three periods in three different EMI classes. The researcher kept the field notes and recorded the whole lessons with the permission of the EMI teachers and learners. During the observation, the researcher participated as a research participant; the observation checklist was used to take notes, focusing on how EMI students behaved in class.

Data analysis
The quantitative data were statistically analyzed using the SPSS software. Descriptive and inferential statistics related to students' FLCA scores, the relationship between FLA frequency and gender, and English grades were presented in the finding section for the qualitative data from classroom observations and interviews, which were video-and audio-recorded. The data were classified according to the three themes: FLCA reflections, the relationship between FLCA frequency and gender, and the relationship between FLCA frequency and English grades, and then analyzed and triangulated with the quantitative data.

EMI students' reflections on their anxiety frequency level
To answer the first research question, the FLCA scale by Horwitz (1986) and interviews were employed to collect the data. The quantitative data are presented in Table 1. Table 1 shows the means and standard deviations of students' self-rated anxiety scores. As the questionnaire was designed using the five-point Likert scale, from never (1) to always (5), the higher the mean the students rated, the higher the anxiety frequency was. As seen in Table 1, the average mean value of all 33 statements is 3.27, equivalent to the "sometimes" level, suggesting that these EFL students experienced foreign language anxiety in their learning to a certain degree. This finding is in line with those of many previous studies (Shiwsky, 2001; Kostic-Bobanovic, Liu, 2006;Liu & Jackson, 2008;Trang & Baldauf, 2013;Trang et al. 2013), which indicates that anxiety did exist in EFL students' language learning process. Looking into more details, the frequency varies between the indicators. Concerning test anxiety-related statements, which seem to be the most concerning factor among these secondary students, with the highest value belongs to the statement "I worry about the consequences of my poor English score" (M = 4.15). This result can be explained by the fact that the Vietnam education system still heavily focuses on exams and tests to assess students' achievements and competencies (Trinh & Mai, 2019). That is why the results of tests at all levels put these students under such pressures. Data from the focus group interviews confirmed the finding by adding more responses regarding the pressure those students had to suffer from deriving from their parents' complaints about grades and such a test-oriented education system.

VIETNAM JOURNAL OF EDUCATION
Regarding communication apprehension, a medium frequency was reported by the participants. They seemed to usually be unconfident and nervous when communicating in English with others, which is why they rejected the statements "I feel sure of myself when I speak English" (M = 2.23) and "I feel very self-conscious about speaking English in front of other students" (M = 2.34) and strongly agreed with other speaking-related statements in the survey, mostly with the mean values >3.0 (see Table 1). The interview data also revealed that these students were shy and reluctant when discussing with their peers or teachers. This finding aligns with other studies (Kagwesag, 2012;Doiz et al., 2012;Flowerdew & Miller, 1992).
In addition, students acknowledged the fear of making mistakes, which led to high mean values with the corresponding indicative items, such as "I start to panic when I have to speak without preparation in English classes" (M = 3.71) and "I get nervous when the English teacher asks questions which I haven't prepared in advance" (M = 3.89). Meanwhile, these informants still "sometimes" felt nervous about the English test even though they did prepare for it, with the mean of 3.3 for the statement "The more I study for an English test, the more confused I get". This result is in line with those of Kagwesag, 2012; Al-Bakri, 2013; , who stated that "anxious students are afraid to make mistakes in foreign language learning" (p.130).
Surprisingly, despite the frequency of anxiety in English classes, these students seemed not afraid to join the ranks. Most of the indicative statements were rated with the mean values just around 3.0, such as "I feel like not going to English classes" (mean = 2.42) or "When I am on my way to English class, I feel very sure and relaxed (mean = 3.41), almost falling into the level "usually". The participants "sometimes" felt more pressured in English classes than in others, with the mean of 3.02. The result can be explained by the fact that the students were aware of the benefits EMI courses would bring them, and their participation in these courses was voluntary. Accordingly, they still wanted to participate in these classes despite their anxiety. Besides, the participants showed a sense of competition, demonstrated with the high values of means for the two indicative items namely "I feel that other students speak English better than I do" and "I keep thinking that other students are better at English than I am" with the means of 3.85 and 3.84 respectively. The student's responses to the interview question "Which is more anxietyarousing to you: speaking English in front of your classmates or teachers? Why?" advocated this finding.
As quantitatively reported, EMI students sometimes experienced anxiety in their English classroom. This preliminary finding then paved the path for the interviews and classroom observations. The interview responses and observation field notes revealed several factors leading to the need for more confidence and nervousness in communicating with EMI students. Figure 1 summarizes the responses.
As shown in Figure 1, the fear of making mistakes is the most frequently met source of the students' FLA. Lack of preparation is another common source of FLA among the low anxiety students (N = 13). Meanwhile, the highanxiety students (N = 12) emphasized their lack of understanding of the teacher's requirements because their English competency was too limited. Fear of negative evaluation from teachers and peers were also mentioned with a relatively high frequency, N=12 for low-anxiety students and N = 10 for high-anxiety students). Most low anxiety interviewees reported that peer judgment was more concerning than the teacher's judgment as the teacher could help them correct the mistakes. In contrast, their classmates usually laughed at them and caught them on even minor mistakes. On the other hand, the results from the high-anxiety students in this study show that they feared the teachers' angriness more than their friends' jeers. Data from the class observation notes indicated that the students with high anxiety frequency usually "sat quietly during group work or timorously participated in discussions by looking at her friend's faces".

VIETNAM JOURNAL OF EDUCATION
 109  Figure 1. EMI students' sources of FLA

Difference between FLA frequencies and genders
To identify whether there was any significant difference between genders in their FLA frequencies, statistics from 141 male and 159 female students were processed with Independent T-test. The analysis yielded no significant difference between the total scores of the two genders ( Table 2). The female students in this study scored higher than their counterparts in most statements, but the difference was not significant enough to be considered. .349 Equal variances not assumed However, looking more closely at each statement score in the questionnaire, there are six items (see Table 3) that had its sig value under 0.05, meaning that the differences between the genders' FLA perceptions in these statements were significant.

VIETNAM JOURNAL OF EDUCATION
 110  As seen from Table 3, male participants demonstrated a higher level of anxiety than their counterparts, indicating that they were more anxious in these situations. In other words, the boys in this study had a heightened sense of competition as they were usually afraid of being beaten by their peers (Q7 & 23). Moreover, they rarely felt confident speaking English even though they seemed more comfortable with native speakers (M = 3.79).
To deepen the above results, the follow-up interview responses were analysed. The data revealed that all the interviewees (N = 25) experienced foreign language anxiety to a greater or lesser degree. This qualitative data was in the same line as the above quantitative data. It seems that male students were more competitive than female students as they cared more about others' judgments of their levels of English proficiency. In addition, the boys reported that they were at ease with native speakers as they could talk and use body language without being marked or judged on their mistakes.
HAS2 reported: "When I speak to a native speaker, I can make mistakes, and they usually try to understand me; they have never laughed at me or uncomfortably corrected me, meanwhile with my Vietnamese teacher, I always feel nervous because I am afraid that they will get angry when I make mistakes, they are usually too good to know all types of mistakes I can make".

Relationship between FLCA levels and English scores
Pearson correlation coefficient was used to analyse the quantitative statistics to identify any correlation between EMI students' FLCA and their English grades. Accordingly, the results, which are provided in Table 4, indicated that there was a significant moderate negative relationship between the two compared variables. The correlation is at -3.61, p = .000 (< 001), suggesting that the higher anxiety levels students had, the lower their grades were. This finding is similar to previous studies (Li & Wu, 2004;Chen & Zhang, 2004). The statistical analysis is shown in the following table.  (2-tailed) .000 N 300 300 **. Correlation is significant at the 0.01 level (2-tailed). To clarify the findings, the interview data were used to determine how the anxiety level affected EMI students' English grades. Most interviewees mentioned the difficulty in taking tests as the significant reason for failure in the assessment tasks. The interviews helped reveal that those EMI students could not develop the appropriate ideas and responses to perform well on tests because of their limited receptive and productive vocabulary. The finding aligns with Phuong and Nguyen's (2019) research results, in which many students admitted not understanding the questions asked in the assessment tasks, leading to unexpected results. This is also in accordance with Al Balushib and Rao (2014), Chang (2010), Galloway et al. (2020), Le (2015), and Zurmo and Qasem (2019), who all stated that EMI students would do better if they were assessed in their native tongue.

CONCLUSION
The study set out to explore the frequency of FLA among Vietnamese EMI students and sought answers to the questions if there is any relationship between FLA frequency and gender as well as English performances. Three

VIETNAM JOURNAL OF EDUCATION
 111  main findings were drawn from the study. First, the data analysis found that at a moderate level, foreign language anxiety did exist among the participating EMI students. Based on this finding and compared to Vietnamese studentsrelated studies, it can be concluded that Vietnamese EFL learners suffer from FLA to some extent during their learning process. Furthermore, students with different levels of FLA have different sources of anxiety. More specifically, FLA among the low-anxiety students usually results from non-linguistic factors, including fear of peers and teachers' judgment, lack of preparation and lack of ideas to communicate. Meanwhile, the high-anxiety students often felt unconfident because of their limited English language proficiency. In addition, the study found that even though EMI students had some anxiety in learning EMI courses, they did not prevent themselves from going to English classes. Given the negative impacts of FLA on the learner as indicated in many previous studies, it is recommended that EMI teachers take this phenomenon seriously during their teaching process. They might need to integrate some FLA-diminishing strategies to motivate their students in their learning. Secondly, the result obtained from the study indicates that there is no statistically significant relationship between genders in terms of anxiety frequency. However, a closer look at the interviews reveals that male students are likely to be more anxious as they are more competitive. This finding is similar to the results of previous studies, which were conducted among students around the world. It can be concluded that gender is not a variable to be considered in EMI programs regardless of the geographical and cultural characteristics. Thirdly, similar to many previous studies, the examined EMI students still suffered from FLA, negatively impacting their English proficiency. The test anxiety was found to stem from limited English language competence, especially in terms of vocabulary. Hence, it is suggested that EMI curriculum designers should be aware of the existence of these sources of FLA, and more room should be included in the program to ensure that EMI students gain enough vocabulary size for confident learning. Besides, EMI teachers should focus more on scaffolding to support EMI students with vocabulary knowledge and integrating anxiety-reducing strategies into their teaching activities. Another suggestion is that L1 language should be taken as another language choice in the examination to ensure the reliability and variability in assessing EMI students' performances.
This study, like any other study, is not void of limitations. Firstly, the study's findings are limited to 135 EMI students at three universities in Central Vietnam, which might only be considered representatives for some Vietnamese EMI students. Accordingly, it may only be generalised for some of their population. There is a need for further study to be conducted in different universities located in various geographical areas in Vietnam or a study that will be carried out in a university where students can represent the whole intended population. Evidence from these studies will be more compelling and robust. Secondly, the research tools used in this study might only partially cover some aspects of FLA's impacts on EMI students. Therefore, any generalization of the results from this study should be made with caution. Further studies are suggested to collect data from classroom observations to fully understand how FLA affects EMI students in their learning process. Despite these limitations, the research has contributed to the related literature. It also offers some pedagogical implications for EMI teachers and learners, EMI curriculum designers and educators, and researchers interested in the field.