Exploring Indonesian university students’ speaking anxiety in online English medium classes (EMI) Eksplorasi kecemasan berbicara mahasiswa Indonesia di perkuliahan daring kelas EMI

This study investigates the speaking anxiety levels of EFL students enrolled in English medium classes (EMI) in a private university. Factors causing students’ speaking anxiety and potential strategies to reduce the speaking anxiety were also examined. Eighty-nine EMI students from ten undergraduate study programs were the respondents of this study. Data were collected by using Horwitz et al.’s (1986) anxiety survey. Findings showed that EMI students exhibited low-level English-speaking anxiety. The cross-tabulation data demonstrated that students perceived comprehension apprehension as the paramount anxiety factor. To reduce their comprehension apprehension, students prefer their lecturers not to overreact their speaking mistakes and degrade them when they make speaking mistakes. These imply that these findings can be used as consideration for preparatory programs for students and lecturers in EMI classes.


Introduction
English medium instruction (EMI) has been underpinning the rise of internationalization in the 21st century as a fundamental strategy for higher education globally, including in Asian countries (Bax, 2010;Council, 2021). Universities in Asia, including Indonesia, are progressively using English to achieve three dominating objectives: internationalizing universities' recognition, attracting international students, and offering an effective language learning opportunity (Kudo et al., 2017;Council, 2021). Since the definition of EMI varies depending on the context in which it is adopted, for this study, EMI is defined as "the use of the English language to teach academic subjects in countries or jurisdictions where the first language (L1) of the majority of the population is not English" (Dearden, 2014:4). In other words, EMI is considered not as a language course, but a course where subject contents are taught using English as a language of instruction (Hellekjaer, 2010). The majority of implementation of EMI in Asian countries adopts the European approach, which was previously known as Content-Based Integrated Learning (CLIL), Content-Based Teaching (CBT), and bilingual education (Vu & Burns, 2014), focusing on the balance of content and English language. Today, EMI is being implemented at both Asian undergraduate and postgraduate levels across all subject areas in programs with international students who are native English speakers and also in courses exclusively offered to home students across Asia (Bax, 2010;Council, 2021).
Although the spread of EMI is becoming prominent in many Asian countries, some notable issues arise, mainly related to EMI students who use English as a foreign language (Floris, 2014;Hendryanti & Kusmayanti, 2018). This is because the practice of EMI in Asian countries such as Japan, South Korea, China, the Philippines, Thailand, Vietnam, and Indonesia often neglects the necessity of minimum requirements of English proficiency and English preparation programs for home students who will enroll in EMI-based study programs (Council, 2021). Hence, this issue must be addressed by institutions even though previous research has implied that when students decide to study in an EMI-based program, they are responsible for their learning of English (Airey, 2012). Rose et al. (2022) affirmed that universities, including home lecturers and academic staff, must also be involved in the design process of EMI curriculum and pedagogy to meet their students' needs.
To anticipate the EMI situation in which a non-language course is taught using English by a lecturer who is not a native speaker of English to students who are also not native speakers of English, EMI lecturers usually use mixed English (L2) and the mother tongue (L1) to bridge the gap in students' English proficiency (Chou, 2018;Chun et al., 2017;Despitasari, 2021;Dewi, 2017;Hendryanti & Kusmayanti, 2018;Kudo et al. 2017;Simbolon, 2021;Suzuki, 2017). One of the reasons given by EMI lecturers in Indonesia for mixing L2 and L1 and implementing the translating and repeating techniques is because they want to ensure that students, especially those with low English skills, comprehend the material presented (Hendryanti & Kusmayanti, 2018;Simbolon, 2021). It resonates with the study by Hung and Lan (2017), where EMI lecturers use Vietnamese more often because they worried that the content delivered in English was not comprehensible to the students. Sercu (2004) emphasizes that when students who are learners of English as a foreign language (EFL) in the EMI class cannot understand what is explained by the lecturer and are not even given the opportunity to interact using adequate English, then students will not feel an increase in their English skills. Moreover, when students with low English proficiency face tasks with linguistic aspects that are too difficult, they are likely to be reluctant to develop and lose motivation to complete their assignments (Ibrahim, 2001;Rolstad et al., 2005). Kudo et al. (2017) emphasized that without adequate language preparation, EMI students would feel anxious, which eventually affects their success in EMI class (Chun et al., 2017;Choi, 2016, Ma et al., 2022Suzuki, 2017). Hashemi (2011) posited that "students' sense of 'self', self-related cognitions, language learning difficulties, differences in learners' and target language cultures, the difference in the social status of the speakers and interlocutors, and the fear of losing self-identity" can cause language anxiety (p. 1811). There is a scarcity of studies investigating English-speaking anxiety in EMI classes in different regional contexts (Cheng & Erben, 2012;Suliman & Tadros, 2011). One of the primary reasons EMI students with lower English proficiency face speaking anxiety in the classroom is their unreadiness to participate in group discussions or presentations with their subject lecturers and other EMI students with more advanced English proficiency (Choi, 2016;He, 2017). Hence, identifying levels and factors of anxiety as well as potential strategies to reduce anxiety are instrumental in increasing EMI students' preparedness to succeed.
In his study, Chou (2018) employed the modified version of Horwitz et al.'s (1986) Foreign Language Classroom Anxiety Scales (FLCAS) and reported that students in the partial EMI (students receiving two language instructions: English and Taiwanese) exhibited higher levels of speaking anxiety compared to students receiving full EMI (intended for students with advanced English proficiency). As EMI students constantly employed rehearsal and paraphrasing strategies, they managed to have their speaking anxiety reduced. Kudo et al. (2017) revealed similar findings on EMI students' high speaking anxiety levels. Their study investigated the levels and factors of Englishspeaking anxiety of Japanese students enrolled in EMI programs. The main factors causing the relatively high speaking anxiety were self-confidence, comprehension apprehension (CA), and fear of negative evaluation (FNE). Kudo et al. (2017) specified that spontaneous oral activities influenced EMI students' speaking confidence since, in EMI courses, students are required a greater amount of group and class discussion and to speak English spontaneously during academic tasks. Due to their low speaking ability, some students were concerned about impromptu speaking activities since they feared having insufficient time for preparation.
Self-confidence, judgment, and self-evaluation of personal value and worth are highly associated with anxiety (Leong & Ahmadi, 2017;Liu, 2006;Toubot et al., 2018). A low self-confidence EMI student will lose the chance to practice their target language because they fear embarrassment and making language mistakes (Leong & Ahmadi, 2017;Liu & Jackson, 2008;Minghe & Yuan, 2013). Hsu and Goldsmith (2021) integrated several teaching strategies in higher education to motivate anxious students, boost their confidence, and lower their speaking anxiety. Lecturers are expected to encourage students to be interested in English-speaking activities and introduce students to more diversified teaching strategies using hands-on and casual language to create a harmonious and light learning atmosphere (Hsu & Goldsmith, 2021). When students display high enthusiasm for using English, their positive attitude toward the target language will develop, eventually influencing their second language or foreign language acquisition (Mohtasham & Farnia, 2017).
According to Aydin (2016), CA is perceived as shyness or social anxiety that could be the primary source of anxiety for low English proficiency students enrolled in EMI classes. It affects EMI students with high English proficiency who feel the benefits if they are in a larger group as anxious learners may remain unnoticed by many people (Toubot et al., 2018). Mohtasham and Farnia (2017) highlighted that CA or self-doubt is the highest provoking anxiety of female EMI students'' speaking anxiety. (McCroskey, 2015) revealed that although they possess sufficient knowledge, anxious learners will underestimate their English-speaking ability because they believe they are incapable of speaking. Even a person who can communicate successfully is affected by CA as he or she fears producing the target language terribly and flawing his or her reputation (Kudo et al., 2017;Toubot et al., 2018). As a result, when EMI students have anxiety in speaking activities, it leads to hesitation to communicate (Ma et al., 2022).
In their study, Kudo et al. (2017) highlighted that EMI students lacked confidence when performing impromptu speeches because they feared negative evaluations from other EMI students. An interesting issue in anxiety is that EMI students with high levels of fear of negative evaluation (FNE) are not always critical of themselves as they tend to avoid interaction by not participating or making succinct contributions (Suzuki, 2017). Therefore, speaking anxiety is directly related to certain situations and affects EMI students regardless of time and place (Kayaoğlu & Sağlamel, 2013). It means that types and anxiety levels can affect student behavior differently (Minghe & Yuan, 2013). Supporting the findings of the speaking anxiety factors in Kudo et al. (2017), Suzuki (2017) underlined that EMI students' fear of negative evaluation from other students might be a factor of high-level English speaking anxiety. Based on Phillips' (1992) observation on foreign language anxiety, Suzuki (2017) emphasized that in an EMI class, an EFL student's lack of English may cause the student's failure to express his or her self-image in English, which eventually causes his or her high level of speaking anxiety. EMI students put effort into reducing their speaking anxiety, including preparation before class. Even though EMI students' speaking anxiety was reduced within three months, the highly anxious students remained apprehensive throughout the semester. Suzuki (2017) recommended a further study to investigate EMI students' preferred strategies to decrease their English-speaking anxiety.
Regarding strategies for reducing anxiety, Frolikova (2014) found that self-reflection in EMI students' efforts to develop speaking skills was very helpful for them because they had the opportunity to self-reflect on their previous learning and experience, then plan their subsequent learning strategies. On the contrary, Mohtasham and Farnia (2017) emphasized that EMI students at the university need wise figures to cultivate the belief that "it is alright to make mistakes when speaking in English". Accordingly, subject lecturers must establish a good rapport with their students, accept them as individuals, tolerate their learning mistakes, and create a supportive and relaxed classroom environment to reduce their speaking anxiety (Mohtasham & Farnia, 2017). If necessary, the subject lecturer must take on the role of a doctor, diagnose their EMI students' anxiety, and reduce the unpleasant effects of anxiety throughout the learning process in the classroom (He, 2017).
Since the pandemic Covid-19 outbreak in early 2020, universities in many countries, including Indonesia, have been closed and shifted to online teaching and learning; there is a call to examine EMI students' speaking anxiety in online classrooms. As reviewed previously, EMI students in offline classes in their home countries experience speaking anxiety to varying degrees, which is primarily because of their limited use of English and other factors affecting their speaking anxiety, including self-confidence, CA, and FNE (Kudo et al., 2017;Suzuki, 2017). When EMI students use less English in an online learning environment, they may experience higher anxiety due to limited interaction and low motivation (O'Doherty et al., 2018). Facing this unexpected learning context, EMI students may encounter difficulties and challenges they have seldom had before, such as technology-provoked anxiety (Bryson & Andres, 2020;Kaisar & Chowdhury, 2020;Valizadeh, 2021). Liu and Yuan (2021) showed that online learning environments strongly influence the anxieties of EMI students and directly affect their speaking proficiency. Moreover, Valizadeh (2021) found that most EMI students in the university felt more anxious in online classrooms due to individual differences and preferences, lack of technological knowledge, and technical difficulties. To these anxious students, the online classroom setting triggers them to feel more suffocated and isolated. In contrast, with the support of synchronous computerized learning, Côté and Gaffney (2021) found that online EMI students experienced lower anxiety than those in face-to-face classrooms. Although commonly subject lecturers dominated online learning and resulted in the lack of learners' presence in online learning, Kaisar and Chowdhury (2020) mentioned that the absence of direct communication and appearance in class compel the EMI students to feel safe. Consequently, EMI students perceive virtual classes to bring flexibility to class schedules and easiness in preparation for language-related activities.
The findings of the present study are expected to inform whether an online EMI learning environment can influence EMI students' high levels of foreign language anxiety (Liu & Yuan, 2021;Valizadeh, 2021) or a factor of low levels of foreign language anxiety (Côté & Gaffney, 2021;Kaisar & Chowdhury, 2020). This study is also intended to confirm if self-confidence (Kudo et al., 2017;Toubot et al., 2018) or FNE (Suzuki, 2017), or CA (Mohtasham & Farnia, 2017) can be the dominant factor that triggers EMI students' English-speaking anxiety. In addition, this study examined EMI students' preferred strategies to reduce their English-speaking anxiety in an online learning environment. As a guide for the study, the following research questions are used: (1) What are the levels of students' speaking anxiety in online EMI classes? (2) What dominant factors contribute to students' speaking anxiety levels? and (3) What are students' preferred strategies to reduce their speaking anxiety?

Method
This study employed a quantitative research method to describe the condition of English-speaking anxiety of EMI students in a prestigious private university in Indonesia. The study was conducted in three stages: the initial stage, data collection stage, and data processing and analysis stage. In the initial stage, the characteristics of samples were determined, and a Foreign Language Speaking Anxiety Scale (FLSAS) designed by Horwitz et al. (1986) was modified and validated. The questionnaire was distributed online to the selected samples in the data collection stage. Finally, descriptive statistics were used to analyze the data in the last stage.

Research participants
The target research population in this study was Indonesian university students enrolled in English medium undergraduate study programs in a private university in Bandung, West Java, Indonesia (henceforth referred to as EMI students). Eighty-nine respondents were selected from 698 students using simple random sampling to participate in this study, meeting the general requirements of a large enough sample (n) ≥30 for a quantitative study (Arikunto, 2019). These respondents have an English Proficiency Test (EPrT) score of 450 and beyond, in which they are categorized into intermediate level or independent users of English according to the Common European Framework of Reference (CEFR). Intermediate learners can understand topics relating to family, work, leisure activities, and travel situations, create simple texts on personal topics, and describe experiences, events, and opinions (The CEFR and EF SET, 2022).

Instruments
This research utilized a modified version of the Foreign Language Speaking Anxiety Scale (FLSAS). The questionnaire was developed using a 5-point Likert scale of 35 items. The first part consists of 11 items covering the profile of the research respondents. The second part of the questionnaire consists of 16 items to measure the speaking anxiety level and identify the dominant factor that provokes speaking anxiety. The third part consists of nine items about strategies to reduce anxiety.
To validate the instruments, this study employed content and construct validity. Content validity was analyzed by two English teaching experts and a psychologist. Questionnaires were distributed to 30 EMI students from 10 study programs to validate the construct and internal consistency reliability. Pearson Moment correlation test was used with a validity coefficient of 0.30 and a significance level of 0.05. The validity test results showed that the lowest validity coefficient of the FLSAS is 0.486, and the highest is 0.930, indicating that the questionnaire is valid. The internal reliability result confirmed by Cronbach Alpha is 0.949, which is satisfactory, with Cronbach Alpha exceeding 0.7.

Data collection procedures
After the questionnaire was validated, it was distributed to 204 respondents, which was 30% of the population using the Google Form application due to ease of use and accessibility. Before the distribution, we contacted the university's assistant manager of the International Class Academic Office (ICAO). After receiving approval from the ICAO, the person in charge (PIC) of the EMI classes in ten undergraduate study programs shared the questionnaire link with the EMI students from the 2018, 2019, 2020, and 2021 cohorts. Eighty-nine respondents completed the questionnaire.

Data analysis techniques
Data analysis was conducted in three phases. In the first phase, the score of the 16 items of FLSAS (item #12-27) was calculated. Since the FLSAS uses a graded Likert scale, the total score ranges from 16-80. One point for each item equals a minimum total score of 16 points, while five points for each item equals 80 as the maximum score. A total score of more than 63 indicates a high level of speaking anxiety, a total score ranging from 48-63 indicates a moderate level of speaking anxiety, and a total score of less than 48 indicates a low level of English-speaking anxiety. The average score was calculated through descriptive statistics to determine the level of English-speaking anxiety of EFL students in the EMI class. Meanwhile, the factors that provoke speaking anxiety are categorized into three dimensions, comprehension apprehension, fear of negative evaluation, and low self-confidence, to answer the second research question. Each category is ranked in the order of scores based on the average total score. The average of each sub-dimension is calculated. Subsequently, the highest mean of each dimension is cross-tabulated with the speaking anxiety levels. As for the third research question, the frequency and percentage of responses for each item in the third section became the basis of the analysis to identify the preferred strategies. The items were ranked from the highest to the lowest frequency.

Results
The results are presented in four main sections. The first three sections depict the overall data to answer three research questions. The first section shows data on English-speaking anxiety levels to answer the first research question. The second section presents results on dominant factors contributing to English-speaking anxiety to answer the second research question. The third section answers the third research question about potential strategies to reduce English speaking anxiety of EMI students. Finally, the fourth section discusses the significant findings of this study.

EMI students' English speaking anxiety levels
The first research question inquired about EMI students' anxiety levels in the EMI classes. Table 1 shows that, in general, EFL students in the EMI classes experience low levels of English-speaking anxiety. Half of the respondents, about 53%, scored below 48, which indicated a low level of speaking anxiety. In addition, the percentage of students who experienced a moderate level of speaking anxiety was 39%, considered relatively high, while only 8% indicated a high level of speaking anxiety. In other words, only a few students seemed to have high speech anxiety levels, while most students experienced low and moderate speech anxiety. It may indicate that the majority of respondents in this study did not experience English-speaking anxiety.

Factors affecting EMI students' English-speaking anxiety
The second research question investigated the factors provoking English-speaking anxiety of EMI students. The mean of comprehension apprehension factors is higher than that of fear of negative evaluation factors. Meanwhile, the mean of low self-confidence factors is lower than that of fear of negative evaluation factors. The factors in Table 2 are sorted in a descending order where the means of each item in each dimension are listed from the highest to the lowest.  Table 2 revealed that the respondents in this study generally perceived "Feeling insecure toward other EMI students' English speaking ability" (M=3.22; SD=1.164) as the highest anxiety-provoking factor in the dimension of comprehension apprehension, followed by "Feeling fearful of making mistakes in EMI subject classes" (M=2.99; SD=1.220) in the dimension of fear of negative evaluation and "Feeling uncomfortable when speaking English in front of other EMI students" (M=3.21; SD=1.039) in the dimension of low self-confidence factor. Furthermore, a cross-tabulation was used to quantitatively analyze the relationship between the factors affecting speaking anxiety and speaking anxiety levels. Table 3 displays the summary data in the cross-tabulation.  Table 3 reports that there are 89 observations with no missing data. The cross-tabulation between the levels of speaking anxiety and dimensions of anxiety factors is shown in Table 4.  Table 4 presents the cross-tabulation of speaking anxiety levels with the dimensions of anxiety factors. The contingency table shows nine cells of data. Column percentages indicate the proportion of respondents with different levels of speaking anxiety that consider each dimension of anxiety factors as the highest anxiety-provoking factor. Of 2.25% of students with high levels of speaking anxiety considered CA as the highest anxiety-provoking factor. It corresponds with 21.35% of students with a moderate level of speaking anxiety and 31.46% with a low level of speaking anxiety. Focusing on FNE reveals that students with moderate speaking anxiety were more likely to consider this dimension as the highest anxiety-provoking factor. It compares to 3.37% of students with a high level of speaking anxiety and 10.11% with a low level of speaking anxiety. At the same time, 11.24% of students with a low level of speaking anxiety consider LSC the highest anxiety-provoking factor, followed by 5.62% of students with a moderate level and 2.25% with a high level of speaking anxiety. Students with high levels of speaking anxiety were strongly influenced by FNE, while students with moderate and low levels of speaking anxiety were strongly influenced by CA. Therefore, this study's dominant factor of EMI students' English-speaking anxiety is CA.

Preferred strategies to reduce EMI students' English-speaking anxiety
To answer the third research question, the answers of 'strongly agree' and 'agree' were combined to show the percentage of EMI students who agreed on each strategy that could reduce their speaking anxiety. Subsequently, the frequency and percentage of each item are calculated and compared. The items in Table 5 are sorted and listed from the highest to the lowest frequencies. The lecturer's manner of correcting is pleasant. For example, when students make mispronounced words, the lecturer provides the correct pronunciation without saying "you're wrong" to the student. 56 62.92% Based on Table 5, the findings show that the majority of EMI students in this study (F>65; P>70%) anticipate their lecturers to implement several teaching strategies to reduce their speaking anxiety in the online classrooms, including not exaggerating students' mistakes, managing all students to have equal speaking opportunities in class, not humiliating the students when making mistakes, allowing students to work in groups or pairs, and being helpful.

Discussion
The first research question examines the level of English-speaking anxiety experienced by EMI students. The findings in the descriptive statistics show that, in general, their anxiety level is at a low level. The results of this study contradicted the findings from previous research by Chou (2018), Kudo et al. (2017), and Suzuki (2017), in which they revealed that EMI students in their home countries experienced high levels of English-speaking anxiety due to their limited English. It is possible because EMI students in their studies experience an offline learning environment which can cause a higher level of speaking anxiety in EMI classes (Liu & Yuan, 2021;Valizadeh, 2021). However, the findings in this study confirmed the findings of Côté and Gaffney (2021) and Kaisar and Chowdhury (2020), who found that EMI students felt significantly less anxious in virtual classes because of the absence of direct communication and appearance in class.
Furthermore, results regarding the factors affecting EMI students' speaking anxiety disclosed that generally, students in this study perceived the dimensions of CA as the highest anxiety-provoking factors, followed by FNE and low self-confidence (LSC). The item (M=3.22 & SD=1.164) can be interpreted as anxiety about their self-doubt in their English-speaking abilities. This finding indicated that the reasons why EMI students in this study feel anxious when speaking English is mainly due to their perception of their ability to speak English because they believe that they are incapable of speaking (McCroskey, 2015;Mohtasham & Farnia, 2017) and as a result, they hesitate to communicate (Ma et al., 2022). Moreover, the other two items (M=2.99; SD: 1.220; M=3.21; SD=1.039) can be interpreted as EMI students' perceptions of other students and their lecturers and their level of confidence when interacting with other students and their lecturers. It can be explained that EMI students in this study have great attention to the evaluations of other students and their lecturers because when they experience concerns about the reactions of other students and their lecturers, this triggers their anxiety in speaking English (Kudo et al., 2017;Suzuki, 2017;Toubot et al., 2018). These findings emphasize that external factors, including lack of lecturer support, insensitive personality from lecturers or classmates, and low personal attention to oneself (Siyli & Kafes, 2015), can trigger anxiety in EMI students. In this case, EMI lecturers must be aware that they should involve their students in online learning interactions instead of dominating them because it may lead to a lack of students' presence leading to a decrease in interaction opportunities and language learning (Kaisar & Chowdhury, 2020). It indicates that EMI lecturers have an important role in reducing the language anxiety of EMI students in this study.
Answering the third research question regarding potential strategies to reduce EMI students' anxiety levels, the findings show that "the teacher's reaction to student errors should not be exaggerated'' is the teaching strategy most chosen by the students in this study. The findings of this study support the results of research by Choi (2016) and Mohtasham and Farnia (2017), where EFL students agree that lecturers do not overreact to their mistakes. Incorrect lecturer behavior can prevent EMI students from expressing their ideas competently and make them anxious when speaking English (Toubot et al., 2018). Concerning the online learning environment, Liu and Yuan (2021) and Valizadeh (2021) emphasized that the online learning environment can influence students' high levels of foreign language anxiety. Therefore, EMI lecturers must pay attention to their students' words, questions, and concerns to increase their self-esteem and self-confidence and to create a comfortable online EMI learning environment.

Conclusions
In general, EMI students in this study experienced low speaking anxiety levels. Confirming the previous study on foreign language anxiety in EMI classes, this study shows that comprehension apprehension, fear of negative evaluation, and low self-confidence raise dominant factors triggering EMI students' English-speaking anxiety. The highest anxiety was associated with comprehension apprehension, in which students felt anxious when they considered other students speaking E were better than them. While, the lowest anxiety was related to low self-confidence, in which students feel unsure when speaking in English. To reduce their anxieties, most students preferred their lecturers not to overreact to their speaking mistakes and embarrass them when they make mistakes in class. The findings of this study suggest that there is a call for a collaboration between language experts and EMI subject lecturers in leveraging EMI students' academic English language confidence and competence. Since the EMI context is relatively new in Indonesia, professional development for EMI subject lecturers in Content and Language Integrated Learning (CLIL) is needed, as well as programs promoting stronger collaboration between EMI subject and English lecturers. EMI subject lecturers need to design bridging activities such as integrating online speaking tasks prior to onsite classroom discussions in their subject classes. In addition, as an alternative way to reduce EMI students' comprehension apprehension, EMI lecturers may prepare lists of key content vocabularies for students to master so they may participate better in classroom activities. Some suggestions should be considered for future research. Important variables, including sample size and the influence of anxiety on the language performance of EMI students, must be investigated further. Future research needs to expand the sample size to generalize the results of the study. Furthermore, future research should evaluate the role of language anxiety in producing communication barriers when using a foreign language.