‘The Silence Kills Me.’: ‘Silence’ as a Trigger of Speaking-Related Anxiety in the English-Medium Classroom

Classroom interpersonal dynamics play a key role in shaping the interactional profiles of language learners and the development of their speaking skills. However, it can be challenging for learners to navigate social interactions with classmates, especially if they lack confidence in their L2 self. This is especially true for first-year undergraduates in Japan who often struggle to adapt to unfamiliar communicative language lessons taught in the target language. King’s (2013) examination of learner silence established a connection between low oral participation, anxiety, and social inhibition. In some cases, however, silence is not just a non-verbal cue of speaking-related anxiety (Maher & King, 2020); it can also be a contributing factor to avoiding talk. Using King’s (2014) cognitive-behavioural model of silent L2 learners’ anxiety, this paper examines how ‘silence’ can trigger speaking-related anxiety in Japanese university English-medium classrooms. The study used a cognitive-behavioural theory (CBT)-based approach to gain insights into students’ perceptions of their in-class behaviours. Forty-five semi-structured interviews were conducted with 17 participants using an interview schedule based on a CBT formulation to explore students’ silent behaviour, associated thoughts and feelings, and behavioural triggers. Multiple recurrent forms of classroom ‘silence’ reported by the participants included using Japanese when English was expected, finishing speaking turns quickly, and taking on the listener role rather than initiating discourse. The findings reveal ‘silence’ as a trigger of speaking-related anxiety. The discomfort some participants felt during a display of ‘silence’ seemed to trigger a fear of being negatively evaluated by classmates, leading to self-doubt about their language proficiency and social performance. These findings contribute further insights into the causes of speaking-related anxiety and learner ‘silence’. Our study points towards using CBT-based approaches to improve speaking confidence by increasing learners’ awareness of their anxiety triggers.


3
using CBT-based approaches to improve speaking confidence by increasing learners' awareness of their anxiety triggers. Keywords Classroom interpersonal dynamics · Cognitive-behavioural theory · Japanese university language classroom · Language learner anxiety · Student silence 關鍵詞 語言人際互動 · 認知行為理論 · 日本大學語言課堂 · 語言學習 者焦慮 · 學生的沈默 In the ever-growing body of foreign language anxiety (FLA) research, numerous investigations have demonstrated how this negative emotion can inhibit learners' verbal participation in the classroom. This paper builds on the authors' earlier work as part of their exploration of the causes of FLA by placing anxious learners' 'silence' at the heart of the investigation (see King, 2014;Maher & King, 2020). The findings of these studies have established a clear relationship between anxiety and learner 'silence'. This paper sets out to build on these studies by examining how 'silence' can affect anxious language learners. In addition, this paper looks at how 'silence' impacts these learners' social relationships and interactions with peers, especially in an EMI context where they are expected to do everything in the target language.

摘要
Using King's (2014) cognitive-behavioural model of silent L2 learners' anxiety, this paper examines the relationship between the participants' speaking-related anxiety and self-reported in-class silent behaviours in the context of Japanese university EMI classes. In this study, EMI refers to English language and content subject courses taught through the medium of English. In Japan, English language classes are often taught in Japanese in school, as well as in university classes (Aizawa & Rose, 2019;Bradford & Brown, 2017;Sawaki, 2017). We used a CBT-based approach to explore how anxious language learners in EMI contexts perceived inclass silent behaviours-theirs and their peers-and how it affected their anxiety and participatory behaviours.

A Review of 'Silence' and Speaking-Related Anxiety in the Japanese Context
In their study of anxiety in Japanese university foreign language classrooms, Williams and Andrade (2008) found that almost 50% of 243 students expected to feel anxious. While this is a single study, its findings are not isolated. A wave of research examining foreign language anxiety (FLA) within Japanese contexts has highlighted the possible negative impacts on student performance, self-efficacy, and motivation to learn the target language (Effiong, 2016;Nakane, 2007). Toyama and Yamazaki (2018), adapting the Foreign Language Classroom Anxiety Scale (FLCAS) (Horwitz et al., 1986), found that communication apprehension and test anxiety were significant factors that contributed to students' anxiety in foreign language classes. Other studies have also found that issues related to communication apprehension-especially students' anxiety about speaking tasks in the classroom-are often a cause of FLA, leading to silent behaviours in class (Rivers & Ross, 2018;Shachter, 2018).

Silent Behaviours of Anxious Language Learners
The weight placed on participation in the language classroom means that silent behaviour is likely to draw negative attention to a student's in-class performance. A display of 'silence' could reflect poorly on their ability or identity. It may also affect the success of the interaction and their willingness to communicate and potentially influence how their teacher and peers behave towards them (Bao, 2014). Furthermore, students could perceive their own 'silence' as evidence that they are not proficient, affecting their self-esteem as language learners. 'Silence' in this paper builds on Jaworski's (1993) anti-essentialist approach, which extends beyond the definition of a lack of audible sound and is inclusive of the various forms that language learners might display. In this study, 'silence' was viewed as an absence of verbal participation in English, such as when there was an expectation to speak in English from a teacher or peer, or themselves, for example, when a participant wanted to speak but was too anxious. As a result, the participant negatively interpreted the 'silence', and their in-class verbal participation was affected.

Forms of Silent Behaviours
Various forms of 'silence' have been observed in previous research in language learning environments (Bao, 2014;Harumi, 2015;McVeigh, 2002;Nakane, 2007). Short utterances and under-elaboration can be considered a form of 'silence' when the expectation for more verbal content was not fulfilled, for example, a single-word response when a sentence was expected or not fully disclosing requested information (Gilmore, 1985;Jaworski, 1993). This type of 'silence' could be problematic for the student if the teacher or a classmate perceives it as a cue that they want to withdraw from the interaction or cannot manage their speaking turn, resulting in a negative response or the student becoming anxious at being unable to elaborate.
Withdrawal has also been identified as a form of classroom 'silence', for example, if the student has made an active decision not to participate in the learning context of the classroom (Bao, 2014;McVeigh, 2002). In Kiyota's (2022) longitudinal study of a Japanese student in an EMI course, the student became increasingly anxious about the difficulty of the course content. This resulted in her wanting to perform well academically but also feeling unable to attend classes due to stress. Students might also withdraw temporarily during a class. This could be taking on the listener role and avoiding the speaker role in a group speaking activity. For example, a student may feel anxious and unable to contribute due to a lack of linguistic ability or content knowledge or may feel uncomfortable with the group's interpersonal dynamics (Pierce & Gilles, 2022). Nakane (2007) observed this type of 'silence' by Japanese and Chinese students in various participant structures in classrooms overseas.
The use of the L1 can also be 'silence', especially in an EMI context where there is an expectation for learners to do everything in the target language. Examples may include moments when a learner uses their L1 if they cannot produce the desired word in English, giving a short response before switching back into their L1 or returning to the L1 in group work when the teacher was not monitoring the speaking activity. A person may also choose to remain silent about certain topics in a particular language, for example, sensitive topics that might be difficult for them to navigate in the L2 (Tannen, 1985;Vetter et al., 2022). Although students' use of their L1 does not always mean an unwillingness to speak in the target language in class, King's study (2013) suggests that L1 use can reveal a learner's apathy towards using the L2. One participant reported how most students in their class saw speaking tasks with a partner as a chance to chat in Japanese and be off task. Therefore, if students use their L1, their motives are worth investigating, arguably more so in settings with limited opportunities to communicate in the target language outside of the classroom, such as Japan (Nation, 2003).

The Effects of 'Silence' on Anxious Language Learners
The relationship between anxiety and 'silence' in the language classroom is well established (Bao, 2014;Harumi, 2015;Tsui, 1996). King's (2014) cognitive-behavioural model of a silent L2 learner's social anxiety ( Fig. 1) was developed from Clark and Wells' (1995) work on social phobia and demonstrates how language anxiety contributes to silent behaviours. Immediate contextual and higher-level sociocultural factors form part of a complex dynamic system involving the subject content and learning context. These factors can trigger negative thoughts, often involving feared predictions of the consequences of speaking in class and a self-focused image where the learner is overly critical of their speaking performance. As well as fearing the negative consequences of a poor performance for themselves, such as a low grade or feeling embarrassed, some anxious learners may worry about the impact of their performance on their interpersonal relationships with peers. Such worries may involve fears of being negatively evaluated (Bao, 2014;Greer, 2000). If anxious learners believe that their ability is poor, they may feel too inhibited to speak, preferring to limit any potential discomfort.
In addition to negative thoughts about their own performance in the classroom, some anxious learners may feel inhibited by the 'silence' of their peers. If their classmates do not speak, they may feel expected to remain silent too or feel reluctant to act differently. Furthermore, classroom 'silence' resulting from peers' lack of verbal participation may cause anxiety due to uncertainty about what their 'silence'  Clark and Wells' (1995) original model (King, 2014) means and thus how they should respond to it. Silent behaviours are often highly ambiguous and open to misinterpretation (Jaworski, 1993). Theories of social anxiety show that social situations which make a person feel unable to predict or control what happens can trigger negative thoughts and emotions (Hofmann, 2007;Kennerley et al., 2017). Without verbal cues, an anxious learner may struggle to decide how they should behave, generating doubts about their social performance ability, resulting in feared predictions about the situation's outcome. King's (2013) study revealed that some learners experienced discomfort when their peers did not speak, such as not responding to the teacher's prompts or questions. These results demonstrate the low tolerance some learners have for their own 'silence' and for peers' silent behaviours. King and Aono's (2017) study of one-to-one interactions with language learners found that silence can trigger anxiety, often due to uncertainty about how to respond.

Japanese Students' Speaking-Related Anxiety
In Japan, further research into language learner 'silence' and speaking-related anxiety in the classroom comes at a timely moment. The Ministry of Education, Culture, Sports, Science and Technology's (MEXT) English education policy is pushing harder toward a more communicative approach with an emphasis on developing speaking skills through increased amounts of classroom speaking activities such as presentations and debates (MacWhinnie & Mitchell, 2017). This move is a result of MEXT's drive to internationalise campuses with the 'Global 30' plan in 2011 and the 'Global University Project for Global Competitiveness' in 2014 to attract more international students (Terauchi, 2017). As part of this global outlook, EMI has become more dominant in curriculum design, with an increase in content courses delivered in English and a rise in the use of English as a means of instruction in classes. However, despite the government's stance, the reality in the classroom is that students are often reluctant to speak in English. Pedagogical issues related to EMI implementation have raised concerns about whether Japanese teachers are prepared to teach in English. Research has been carried out to investigate whether Japanese students' proficiency is sufficient and what support they need. While research has shown that Japanese students are motivated to participate in EMI classes, they have also reported high levels of anxiety (Kojima, 2021;Terauchi, 2017).
The prominent causes of FLA in EMI classes in Japan have been found to be related to students being unaccustomed to English instruction and struggling with comprehension. The vast majority of students take classes, including English language, at school in Japanese (Osterman, 2014). In EMI, there is the cognitive demand of having to understand the language and the content. If a language student feels unable to comprehend what the teacher says or the class materials, this can lead to FLA and inhibit students from speaking (Horwitz et al., 1986). Kojima (2021) investigated the motivation of Japanese students in EMI preparatory classes and found that they understood less than 50% of the contents and wanted to drop out.
In addition to unfamiliarity with English instruction, another factor is the lack of experience with communicative approaches in English classes at school. Due to the weight given to English skills in high-stakes entrance exams, for high school and university, English classes are mainly designed for exam preparation. Students often work individually and in silence, with little time available for developing speaking skills (Sawaki, 2017). Learning tasks tend to focus on locating parts of speech and explaining vocabulary items, answering a question by finding the correct answer in English texts, and intensive reading of long passages. For first-year undergraduates in Japan, developing their language skills while navigating peer relationships through the medium of English as they adapt to unfamiliar communicative teaching can be a challenging transition from their school experiences (Shachter, 2018).
Japanese students' inexperience in speaking English affects their spoken proficiency and speaking confidence. In studies on FLA and EMI in Japan, students' beliefs about their limited proficiency caused them to feel anxious about speaking English in class. As well as the difficulty level of the content and worries about not having sufficient linguistic knowledge, EMI classes often have speaking tasks that involve spontaneous talk, giving students little time to prepare what they want to say (Suzuki, 2017). Kudo et al. (2017) found that the lack of time to prepare what they wanted to say was frequently mentioned as why students experienced low speaking confidence. A consequence of this, a further contributing factor to their speaking anxiety, was fears of negative evaluation if they could not respond quickly or proficiently. The level of fear was much higher for peers' evaluations than teachers, which Kudo et al. (2017) attributed to the participants being more concerned about how peers perceived them as they had to work with them in class.
As previously mentioned, it is well-established in existing research that anxiety can inhibit a learner, affecting their oral in-class participation. However, it is arguable that the relationship between anxiety and 'silence' is bi-directional, and therefore, silent behaviours should also be examined as a factor which could cause a learner to experience speaking-related anxiety in the classroom. This paper aims to examine how forms of classroom 'silence' trigger language learners' anxiety. By recognising the bi-directional relationship between anxiety and 'silence', this study suggests that focusing solely on verbal participatory behaviours limits understanding of what causes learner anxiety in the language classroom. Moreover, having a wider interpretation of 'silence', including under-elaboration, withdrawal and use of the L1, and how learners perceive their own and peers' displays of 'silence' could give further insights into what causes negative thoughts and emotions related to language learning.

Research Aims and Questions
This study was to find out how in-class silent behaviours affect the oral participation of language learners who experience speaking-related anxiety in an EMI class. The first objective was to examine the bi-directional relationship between 'silence' and anxiety and whether 'silence' contributes to negative thoughts and emotions related to in-class speaking situations. The second objective was to identify factors that triggered speaking-related anxiety. In Clark and Wells' (1995) model of social anxiety, the 'situation' is where an individual's social anxiety is likely to occur. In King's subsequent model of a silent L2 learner's anxiety (2014), the language class is identified as the 'situation'. This investigation aimed to build on King's previous research by asking participants to pinpoint specific moments or factors in the EMI language class that triggered their anxiety. The research questions were: RQ 1: What forms of in-class silent behaviours did the participants describe? RQ 2: How did the participants perceive in-class silent behaviours? RQ 3: What factors did the participants perceive as triggering their speakingrelated anxiety?

Context and Participants
This study took place in a medium-sized foreign studies university located in a metropolitan area in Honshu, Japan's main island. Seventeen participants volunteered to take part in this study. They contacted the authors after seeing our posters on campus advertising a research project for students who were anxious about speaking in class and wanted to increase their confidence. Their English proficiency levels varied from intermediate to advanced based on the institution's placement test. Three participants were first-year students in the Global Studies faculty, taking content and English language skills (EAP writing and speaking) courses delivered in English by Japanese and non-Japanese lecturers. Content courses included economics, business administration, law, political science, and sociology. Fourteen participants were first-and second-year English majors who had EAP skills classes (research presentations, academic writing) taught by expatriate anglophone instructors in English. After participants consented to take part in the study. Two to three interviews were conducted with each participant for an average of 50 min. Eleven participants were interviewed three times, and six participants were interviewed twice. The total number of interviews was 45 and generated a corpus of around 135,000 words.

Cognitive-Behavioural Theory-Based Approach
The study's design used a cognitive-behavioural theory-based (CBT) approach. CBT is a psychological intervention used to investigate the triggers and factors impacting a person's anxiety to help the person develop and implement coping strategies to achieve behavioural goals. The approach centres on the creation of a simplified formulation (see Fig. 2) of the cognitive distortions (negative thoughts) and behaviours that can affect emotional regulation (Kennerley et al., 2017). A CBT formulation is created by asking the individual about recent events when they experienced anxiety. For each event, the person recollects what they were thinking, feeling, and doing and what happened after the event. By creating a 'cycle' of their thoughts, emotions, and behaviours, the person becomes aware of how their thoughts control the other elements of the cycle and any factors that might be maintaining their anxiety (Corrie et al., 2016;Kennerley et al., 2017). By comparing similar situations where the individual did not become anxious and experienced a positive 'cycle', it is possible to gain a deeper insight into contextual factors that might trigger their negative cycle of anxiety.
For example, student 'X' is anxious about making mistakes in class because she might be misunderstood. Last week in class, she experienced a negative cycle of anxiety ( Fig. 2) when she was doing a speaking activity with a partner. This week, she made a mistake in a speaking activity with a different partner but felt more positive and less anxious, experiencing a positive cycle (Fig. 3). After completing the

Social situation
Making a mistake in front of a classmate Fig. 3 Positive CBT formulation cycle of student 'X' (CBT formulation cycle based on Kennerley et al., 2017) formulations and then comparing the contextual factors of each event, student 'X' became aware that she felt more at ease about making a mistake when her partner was a familiar classmate. When she experienced her negative cycle, her partner was a classmate whom she had not worked with much. Also, student 'X' noticed that after she made a mistake, she looked at her partner's face each time. She reported that the classmate she was less familiar with did not react to her mistake, whereas the other partner smiled. From this formulation, the next step would be looking at whether the facial reactions of interlocutors might trigger her anxiety and testing coping strategies to be more objective about people's facial reactions when she speaks. In addition, she could also work on coping strategies to build her confidence when speaking with people she is less familiar.
We felt that CBT would be a suitable approach for investigating the relationship between the emotion of speaking anxiety and the behaviour of 'silence' of learners who had goals to improve their speaking confidence. Furthermore, looking at existing research into language anxiety, a CBT approach represents a data collection technique that has the potential to reveal new insights. For an in-depth but approachable guide to CBT frameworks and formulations, see Kennerley et al. (2017). 1

Interviews
Each participant was interviewed using a CBT approach. We aimed to conduct three interviews with each person; however, six participants could only meet twice due to other commitments. Seventeen participants took part in interviews 1 and 2, and 11 participants took part in interview 3. In the first interview, we conducted an assessment to identify in-class situations where the participants felt unable or hesitant to speak in English (silent behaviours). Table 1 shows a selection of existing instruments used to diagnose social anxiety and foreign language anxiety. We adapted items from these instruments and created a list of nine hypothetical in-class scenarios that might trigger speaking-related anxiety (Table 2). In interview 1, we asked each participant about their experiences and how they felt in each situation. At the end of the interview, we asked participants to choose the scenario(s) that they felt most anxious about and wanted to improve their confidence.
In the second and third interviews, we created CBT formulations (Fig. 2) for the in-class speaking situations that the participants chose. Interviews 2 and 3 were both used to formulate the participants' cycles, but we decided to conduct two sessions to avoid participant fatigue, especially as most participants chose to use English for the interviews. So, for the six participants that were unable to attend the third interview, we scheduled a longer session for interview 2 (with their permission) to complete the formulation. Participants were asked to think of recent examples of each situation and describe what happened and their participatory behaviours. They were then asked to recount what emotions and thoughts they had at that time. The aim was to formulate the CBT cycle of their anxiety and 'silence' to understand how these phenomena were related. Once we had formulated two to three negative cycles, participants were asked to repeat the process for recent examples of positive experiences of speaking in their EMI lessons. By comparing their positive and negative experiences of each situation, we asked participants to reflect on what factors were different and could be possible triggers. Once potential triggers were identified, additional formulations were created using recent experiences and hypothetical examples to test these factors to see if the participants recognised them as something that might be causing their anxiety.

Data Analysis
Before each interview, participants were reminded that they could use Japanese or English, and an interpreter was available if they wanted. All participants chose English for most of the interviews, occasionally using some Japanese. The interviews were transcribed in full, with the Japanese parts transcribed into Japanese before translating them into English. An independent translator was asked to check the transcription before making a technically accurate translation (Richards, (Spielberger et al., 1983) General anxiety Liebowitz Social Anxiety Scale (LSAS) (Liebowitz, 1987) Social Phobia Inventory (SPIN) (Connor et al., 2000) Fear of Negative Evaluation (FNE) (Leary, 1983) Social anxiety, avoidance, and fear  . To analyse the interviews, a qualitative data analysis programme called HyperRESEARCH (Research, Inc., 2021) was used with a detailed codebook. In the initial rounds of coding, 54 codes were created, with 29 provisional descriptive codes based on the names of the parts of King's (2014) model (Fig. 1) and forms of 'silence'. The other 25 codes were structural, based on the research questions, and included CBT terms from the formulation (Figs. 2 and 3). As there were many provisional codes, we used simultaneous coding to leave some flexibility in the coding decision-making process (Neale, 2016). The next stage of coding was to determine themes in the data. The analysis focused on the forms of 'silence', negative thoughts and emotions related to these moments of 'silence', and factors that triggered a negative cycle compared to the factors identified in the positive experience of a similar in-class speaking situation. In the final stage, iterative categorisation (Neale, 2016) was used to summarise the data for each theme to consider the best fit through constant comparison.

Findings and Discussion
In this section, we present a brief overview of the forms of in-class silent behaviours which the participants described. These descriptions provide context to the main findings shown in the next section, which looks at the participants' thoughts and emotions related to in-class silent behaviours and speaking-related anxiety. Pseudonyms have been used in the interview extracts.

Forms of Silent Behaviours
Three recurrent forms of in-class silent behaviours across the majority of the participants were revealed in the data from interviews 2 to 3. Fifteen of the seventeen participants reported using their L1 (Japanese) in class with their peers in both group and pair work activities. We coded this behaviour as 'silence' when the participants explained that they were expected to use English during that speaking situation. 'Silence,' as defined in this paper, is not only acoustic-it may occur also when talk is happening, but participants avoid a specific mode of communication, certain topics, and so on. Thirteen participants described starting a speaking turn in English and then switching back to Japanese. Further examples of this form of 'silence', described by 12 participants, included saying a word in Japanese when they could not remember it in English. Longer examples included using their L1 for an entire activity, which was recounted by three participants. A further example was using Japanese when their peers used it, especially a partner or group member. Eight participants reported this form of 'silence'. Another frequently mentioned form of 'silence' was having a short speaking turn when they felt that more was expected from them by their teacher, peers, or themselves, which was mentioned by seven participants. Previous studies have defined this type of behaviour as a form of 'silence' when a longer turn or further information was expected by the interlocutor (Jaworski, 1993;Nakane, 2007;Tannen, 1985). Five participants described this form of 'silence' as only being able to give one-word responses when they wanted to use a whole sentence. The last frequently reported form was taking the listener role in the class, such as not initiating talk in front of the class or with their peers in a group activity. A point in common among participants' reports of this form is the involvement of peers. Out of the seventeen participants, only two described feeling anxious or unable to speak with their teachers. This finding is similar to Kudo et al.'s (2017) study where participants reported higher levels of fear of negative evaluation from peers than teachers in an EMI class. Table 3 shows the in-class speaking scenarios chosen by participants because they wanted to improve their confidence and discuss more in-depth in the second and third interviews. These results seem to support the findings of Murphey et al. (2014) that suggest that the interpersonal relationships and dynamics of the language classroom can have a significant impact on how anxious learners verbally participate.

Triggers of Speaking-Related Anxiety
Through comparing the participants' positive and negative CBT formulations of various speaking experiences, we created seven categories to represent the factors that appeared to trigger their speaking-related anxiety. Table 4 presents these triggers and the number of participants who described each one (some participants reported multiple factors). Three categories were related to whom the interlocutor was, such as, opposite gender, international student, and non-Japanese instructor. Two categories were connected to the 'setting' of the speaking situation, for example, group work where there was more than one peer to work with and speaking in the classroom which made participants feel pressured to use more academic vocabulary and behave more formally when they spoke. The two remaining categories had to do with behaviours during speaking activities: 'silence' and the facial reactions of peers when a participant was speaking. The data categories revealed 'silence' as a common speaking-related anxiety trigger among our sample and demonstrate that 'silence' can potentially impact verbal participation beyond limiting opportunities to practice speaking.

Impact of 'Silence' on the Participants
For the nine participants whose speaking-related anxiety was triggered by 'silence', their CBT formulations revealed that displays of 'silence'-theirs or their peers'-made them hyper-aware of what was happening around them, which often resulted in negative emotions related to speaking. Also, silent behaviours tended to trigger negative emotions that led to negative thoughts about their language proficiency, fears of receiving unwanted attention due to not speaking when expected to, and feeling pressure to speak because of their peers' 'silence'. Table 5 shows examples of participants' negative thoughts about their silent behaviour and their peers'. The participants' CBT formulations indicate that socio-psychological factors, such as awareness of classmates and interpersonal relationships, are a common theme among the negative thoughts related to 'silence', which supports the  Formal, academic atmosphere of the classroom 2 Peers' reactions during speaking activities 2 Speaking activities with international students 2 Speaking with a non-Japanese instructor 2 Partner's gender 1 findings of similar studies (Effiong, 2016;Greer, 2000;Kudo et al., 2017). The displays of 'silence' caused the participants to worry about how to manage interpersonal relationships with their peers in English, often because they felt unsure about interpreting others' 'silence'. It seems that not only did their 'silence' confirm selfdoubts about their language proficiency, but also doubts about their ability to perform socially within EMI contexts (Kudo et al., 2017). The participants' 'silence' or their peers' 'silence' created a moment of non-talk that they felt compelled to fill. This type of fear is comparable to people with social anxiety who believe they have poor social skills and will be unable to manage an interaction without being rejected (Hofmann, 2007). These show that some learners may have a low tolerance for 'silence' because it makes them feel uncomfortable due to its ambiguity, causing them to feel uncertain about how to interpret and respond to it (King & Aono, 2017).

In-depth Examples of Individual Participants' Negative Thoughts
In this next set of findings, we have used interview excerpts to provide detailed examples of negative thoughts that our participants experienced in their EMI classes. The following excerpt from Chiyuki's (Table 5, excerpt 1) interview, conducted mainly in English, is an example of how her 'silence' contributed to her selfdoubt. She explained how she could not reach the 5-min speaking goal in a conversation activity ( In this case, Chiyuki's interaction with her partner broke down as she felt unable to speak in English. Although the interaction seemed to begin positively in English, 1 3 once she switched to Japanese, she reported feeling anxious about speaking in English again and ended her speaking turn. Shinwa shared a similar experience (Table 5, excerpt 2). After giving a class presentation (Table 2, scenario 3), he was asked a question by his peer and could not answer immediately. Despite feeling calm and prepared for the presentation, this pause triggered his anxious thoughts and doubts about his ability, which distracted him as he was trying to answer. He described how his mind went blank; he forgot the question and was embarrassed to ask his classmate to repeat it.  In Eri's case (Table 5, excerpt 6), she felt nervous and troubled when the teacher asked the class a question, but nobody responded ( The findings of these four cases support those of previous studies that have looked at FLA in EMI contexts regarding students' concerns about their proficiency and speaking without time to prepare (Kudo et al., 2017;Suzuki, 2017). For Chiyuki and Shinwa, their 'silence' confirmed their self-doubts about their proficiency, prompting negative thoughts and emotions, contributing to further silent behaviours, and negatively impacting their speaking confidence. In the cases of Shinwa and Eri, it seems that the pressure of spontaneous talk in English contributed to their speaking anxiety. In Shinwa's example, there is a clear difference between his confidence level during the presentation when he felt prepared and the question-and-answer session when he could not prepare, which led to a difficult silent moment for him in front of his peers. For Eri, being unable to prepare her answer to the teacher's question resulted in her avoiding talk entirely because she did not want to make any mistakes. Another theme in these in-depth cases was the participants' apparent lack of comfort with their peers' 'silence'. Eri expressed discomfort during 'silence', describing a feeling of being pressured to fill it, which triggered her anxiety. For Rina, her partner's 'silence' triggered a negative cycle as she tried to interpret the meaning.
Perhaps due to the ambiguity of 'silence', these anxious learners perceived silent behaviour as a negative cue, questioning themselves and their speaking proficiency. In an EMI setting where there are more demands for verbal participation in English, anxious students are likely to feel more pressure about speaking and, in turn, be more sensitive toward silent behaviours.

Conclusion
Using a CBT-based approach to explore anxious language learners' thoughts, emotions, and behaviours related to silent behaviours in the EMI classroom, the findings of this study have revealed 'silence' as a potential speaking-related anxiety trigger in the Japanese university context. This study demonstrates the bi-directional nature of the relationship between anxiety and 'silence' in the EMI classroom, highlighting that 'silence' is not just a consequence of speaking-related anxiety but can also trigger it.
Our findings revealed how 'silence' could make anxious learners over-sensitive about their surroundings in an EMI context. The ambiguity of a silent moment can generate negative thoughts about themselves, such as self-doubt about their proficiency. For anxious learners, the lack of certainty of the outcome of an in-class speaking situation may trigger their cycle of negative thoughts, emotions, and behaviours. In some cases, this could be a low tolerance of 'silence', such as Eri, who described how the silence 'kills' her.
The findings also suggest that anxious learners not only have to deal with managing classroom interpersonal dynamics through verbal participation, especially when they feel unable to prepare what they want to say before speaking. In an EMI environment where learners are doing everything in the target language, it is understandable that they may become distracted into non-talk by trying to manage their worries about their 'silence' and trying to read the cues of their peers. These findings show that in some cases, a display of 'silence' may have far more significance to an anxious learner in an EMI classroom than a pause or breakdown of an interaction.
A pedagogical implication of this study is that a simplified CBT formulation (see examples in Figs. 2 and 3) can be adapted into a class activity or advising tool (with individual students) to help students become more aware of what affects their speaking confidence and in what type of in-class situation. Students can use the framework to compare negative and positive speaking events in class to determine what factors make them feel nervous and then develop and test coping strategies that focus on those factors. For students who doubt their proficiency, becoming aware that their speaking anxiety and silent behaviours are likely to be situational may help reduce their anxiety by developing more objective thinking and realising that they are not always 'silent' in every class. Furthermore, having students focus on their 'silent' moments can be an effective way to support them in developing their speaking skills. In an EMI class where students are required to frequently speak in English in various settings and tasks, and often with little time to prepare what they want to say, students might lack confidence about keeping up with the demand. So, by having them focus on when they are 'silent', that is, they feel they are not meeting the expectations of others or themselves to speak, teachers can help students set clear goals for improving their speaking ability and confidence.
Finally, CBT-based approaches can be used to develop classroom activities to improve speaking confidence in EMI settings by increasing learners' awareness of their anxiety triggers. Having learners share thoughts and feelings about speaking in class can help promote peer bonding by encouraging them to learn about each other while getting ideas for coping strategies. In our experience, having students realise that they are not alone in feeling anxious about speaking can help alleviate negative thoughts and feelings about speaking in the classroom.
This study did have some limitations. Our research design relied heavily on interviews and so may have benefited from an additional source of data on the participants' in-class performances, such as semi-structured classroom observations. Also, the time constraints when conducting this study only made it possible to conduct two to three interviews with each participant. If it had been possible, conducting several interviews over a longer duration would have produced more insights. Lastly, there were no items in interview 1 that specifically asked about factors unique to EMI classes, such as teacher's instruction in English, spontaneous talk, and learning about a topic in English. This is because we wanted to see what factors organically emerged in the data by asking the participants about general in-class speaking situations. However, future research based on this study could provide more insights into what specific EMI class factors trigger FLA and silent behaviours, building on previous work. With MEXT's global drive, there will likely be an increasing number of EMI courses on offer in Japanese universities, so further understanding of what support students require and providing students with affective language learning strategies alongside linguistic skills is crucial.

Conflict of Interest The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.