Inspired by Horwitz et al.’s, (1986) work on Foreign Language Classroom Anxiety (FLCA), there has been a plethora of studies drawing on their theoretical framework and instrument to investigate the anxiety experienced by language learners in various contexts. As Dornyei and Ryan, (2015) observed, language anxiety (LA), which covers both foreign and second language,Footnote 1 “has been in the limelight of L2 research for several decades” (p. 176); it is commonly recognized as one of the most researched individual-differences variables in the field of L2 (second language) acquisition. In the ensuing discussion, to avoid inconsistency of terms, FLA is used to cover both second and foreign languages.

Clarifying the Construct of Foreign Language Anxiety

Horwitz et al., (1986) proposed interpreting foreign language anxiety (FLA) within the framework of situation-specific anxiety and defined FLA as “a distinct complex construct of self-perceptions, beliefs, feelings, and behaviors related to classroom language learning arising from the uniqueness of language learning process” (p. 128, our emphasis). They designed a 33-item generic form of Foreign Language Classroom Anxiety Scale (FLCAS) for measurement, which has been widely applied or adapted in FLA research across different countries and learning contexts. To help L2 scholars and teachers understand the concept of FLA, Horwitz et al., (1986) drew parallels with three related situation-specific performance anxieties: communication apprehension, test anxiety, and fear of negative evaluation. This effort has, however, resulted in considerable misinterpretation of FLA as comprising these three performance anxieties, as Horwitz, (2010, 2016, 2017) remarked. There were also studies attempting to “test” a three-factor structure of the FLCAS (see for example the frequently cited Aida, 1994). Horwitz et al., (1986) actually reminded us that FLA “is not simply the combination of these fears [communication apprehension, test anxiety, and fear of negative evaluation] transferred to foreign language anxiety” (p. 128), and the 33 items on the FLCAS “were not chosen to represent a three-factor model of language anxiety” (Horwitz, 2017, p. 36). Their theoretical argument was supported by Horwitz’s (1986) construct validation study that found small amounts of shared variance between FLCAS and the abovementioned three anxieties as well as trait anxiety, indicating that FLCAS could be discriminated from these anxieties.

Significant Developments in FLA Research

Following Horwitz et al. (1986), FLA research began to enter a “specialized” phase, in which anxieties specifically associated with different learning situations (e.g., in or out of classrooms), language processes (e.g., input, processing, or output), or language skills (e.g., listening, speaking, reading, and writing) were identified, defined, and investigated (MacIntyre, 2017). Studies on FLA have flourished with the development and validation of various situation-specific (or specialized) measurements of anxieties experienced by L2 learners, such as L2 listening anxiety scales (Elkhafaifi, 2005; Kim, 2005), L2 speaking anxiety scales (Woodrow, 2006), L2 reading anxiety scales (Saito et al., 1999; Zoghi, 2012), L2 writing anxiety scales (Cheng, 2004), brief anxiety scales for all four L2 language skills (Cheng, 2017), and scales for anxiety at the input, processing, and output stages of L2 learning (MacIntyre & Gardner, 1994), to name a few. These studies have infused the field of FLA with more fine-grained evidence. A better understanding of the nature and role of FLA has been achieved through analyses of its associations with other learner characteristics (e.g., Dewaele, 2013; Tóth, 2007) and its simultaneous interrelationships with multiple L2-related variables by means of advanced statistical techniques such as structural equation modeling (e.g., Chow et al., 2021; Yamashiro & McLaughlin, 2001). More importantly, the puzzle of contradictory results generated by early research was unraveled: A consistent negative association between FLA and various aspects of L2 performance has been confirmed by a wealth of studies (see the recent meta-analyses of Botes et al., 2020; Li, 2022; Teimouri et al., 2019; Zhang, 2019).

Learners’ FLA in EMI Contexts

The preceding brief summary of studies is by no means exhaustive; it merely serves to highlight some of the important studies. While they have yielded important insights, there are two areas which have not received as much attention as they should have. The first one is FLA in classrooms where English is not only an object of learning but also a medium of instruction, that is, English Medium Instruction (EMI). In this special issue, EMI refers to the use of English as a medium of instruction for content subjects as well as English language learning in contexts where English is not a first language for the majority of the population (Dearden, 2014). In ESL (English as a second language) contexts, EMI is a received practice in English language classrooms; in EFL (English as a foreign language) contexts, however, it is a relatively recent practice. For example, in S. Korea, teaching English through English (TETE) was introduced only in 2001 and certification was introduced by the Korean government in 2009 to enable all English teachers to teach English in English by 2012. The policy was rescinded because of the concern and resistance from teachers, resulting in the use of both L1 and English in English language classrooms (Shin, 2020). Similarly, in Japan, it was as recent as 2013 that teaching English through English was stipulated by the Course of Study for senior high school released by the Ministry of Education, Culture, Sports, Science, and Technology (MEXT). This stipulation has been considered a drastic departure from the predominant use of L1 (Japanese) in teaching English and has aroused heated nation-wide debate (Tsukamoto & Tsujioka, 2013). In Taiwan, since the announcement by the then premier in 2018 to make Taiwan a “bilingual country,” subsequently revised as “bilingual policy,”Footnote 2 plenty of resources has been allocated by the Ministry of Education to the improvement of students’ English proficiency and the professional development of school teachers at primary and secondary levels in the teaching of English and teaching English through English (Yeh and Chern, 2020).

Also in EFL contexts, EMI in content subject classrooms is a very recent phenomenon. EMI has been spreading so rapidly both at school and tertiary levels that English has now become the “lingua academia” or “lingua franca academia,” as a number of scholars have observed. A report released by the British Council and Studyportal, an online platform, showed that during the period 2017–2021, English-taught programs at university level in regions outside of the “Big Four” (i.e., UK, USA, Australia, and Canada) English speaking countries have risen by 77%, out of which 18.9% took place in Chinese region and East Asia (British Council Studyportal, 2021, p. 14). Previously, English-taught programs were mostly offered at doctoral level. However, according to the report, during this period, there was a rise of 75% at master’s level and 84% at bachelor’s level. EMI is indeed “an unstoppable train which has already left the station”, as (Macaro, 2019, p. 232) observed.

The psychological impacts of having to understand and articulate disciplinary knowledge in a foreign language with which one is still struggling cannot be under-estimated. Studies conducted so far have shown high levels of dissatisfaction with EMI content teaching, even among academically able students with high English proficiency levels. For example, Joe and Lee’s (2013) study of over 60 medical students in a Korean university in Seoul, all with high levels of English proficiency, showed that although their comprehension of the lectures was not affected by EMI, nearly half of them indicated having a “negative impression of and excessive anxiety about the English medium course” (p. 205). Fidan Uçar and Soruç’s study (2018) of over 300 Turkish university students studying through EMI and FMI (French medium instruction) found that despite the students’ positive response to sense of achievement and motivation in the survey, the interview data revealed students’ anxieties which were related to teacher, learner, language, and content-oriented factors. University students have also been reported to be struggling linguistically and experiencing a lack of confidence and failure to adapt when transitioning from L1 medium instruction in secondary schooling to EMI in their university studies, resulting in serious retention problems (Staub, 2022). What these psychological impacts are and in what ways and to what extent they have affected students’ attitudes towards the English language, learning processes and approaches, interpersonal relationships, and so on have yet to be better understood. How to address the problems to which they have given rise needs further exploration.

Language and Content Teachers’ FLA in EMI Contexts

The second area is language teachers’ FLA. More than two decades ago, Horwitz (1996) reminded us that language teachers can also be afflicted by FLA and that it is important to recognize and alleviate this. In contexts where L1 used to be the predominant medium of instruction in English lessons, the problem is even more critical. For example, in Japan, the stipulated adoption of EMI in English lessons has left high school English teachers “at a loss as to what to do” (Tsukamoto & Tsujioka, 2013, p. 310). However, to date, studies addressing this issue are scanty. Tum (2015) aptly described this area as the “Foreign Language Anxiety’s Forgotten Study.” The few studies conducted have confirmed Horwitz’s observation, particularly among non-native English teachers. Tum’s study (2015) showed that non-native speaker pre-service English teachers in Turkey experienced significantly high levels of FLA. Liu and Wu’s study (2021) on College English teachers in China showed varying levels of FLA in relation to their age, professional titles, educational level, and overseas experience. Machida’s study (2016) of Japanese elementary English teachers’ FLA showed that out of 133 participants, 104 (77.4%) were anxious about their own English proficiency; they were particularly anxious about being able “to speak English well enough to be a good English teacher” (p. 50). As Murdoch, (1994, p. 254) pointed out, “for non-native English teachers, language proficiency will always represent the bedrock of their professional confidence.” A lack of professional confidence will impact all aspects of their performance in the classroom.

In contexts where EMI in content teaching is becoming ubiquitous, there is a need to better understand the impact of FLA on content subject teachers who are required to teach in a language that they are, or they perceive themselves to be, not entirely competent or confident in using. The impact affects not only their teaching quality, approaches, and processes but also their professional self and identity. While studies on EMI in higher education are burgeoning, not many have examined the impact of FLA in this context. As Richards and Pun (2022) observed, in EMI contexts, teachers’ self-efficacy is mediated by the extent to which they feel they are able to achieve good teaching practice through the medium of English. While some may experience no negative impact on their teaching efficacy, many have reported adverse effects on their teaching quality, including a lack of depth of knowledge and inability to improvise and to respond spontaneously. For example, in S. Korea, since the accelerated implementation of the EMI policy in higher education in mid-2000, the challenge posed to Korean professors has been enormous. Byun et al. (2011) reported professors at Korea University feeling overwhelmed when trying to teach content beyond pre-scripted outlines and phrases in English because of their limited English ability. Kim (2017) recounted a political science professor experiencing difficulty in covering content fully in English, even though he held a US doctoral degree and had taught for ten years. According to this professor, the problem would be even more serious for non-native English speaking professors who earned their degrees in non-English speaking countries. Just as language learners’ self-esteem and self-presentations may be compromised by the limited range of meanings and emotions that they could communicate because of their immature command of a second or foreign language (Horwitz, 2017), a teacher’s professional self and professional identity could be undermined. Recent studies of EMI disciplinary teachers in European universities have yielded supporting evidence. For example, in a study of EMI in a Finnish teacher education program, Hahl et al. (2016) found that not being able to express themselves fully in English negatively impacted the teacher educators’ professional identity. Similarly, Danish EMI teachers reported a sense of embarrassment, even a feeling of “being naked” when they could not find the right English words and their communication faltered (Henriksen et al., 2019, p. 75). As a Danish teacher remarked, “… all the issues are connected to my language skills – all the feelings of embarrassment, and lack of confidence, and my anxiety, and so on. It is connected to my English. I am never nervous in Danish” (Eilert, 2017, p. 46; cited in Richards & Pun, 2022).

With the unstoppable spread of EMI across the globe, more and more teachers and learners will find themselves going through the arduous journey of teaching and learning through a language with which they are still struggling. This special issue highlights the need to attend to an important psychological construct which would have far-reaching effects on teachers and learners, and teaching and learning qualities. We are pleased to present four articles, each of which provides new insights and directions for future research in FLA.

Contributions in the Special Issue

Graham studied the FLA of pre-service content teachers in a teacher education university in Taiwan. When addressing the predictors of pre-service content teachers’ EMI teaching anxiety, it is noteworthy that Graham differentiated teaching anxiety from FLA while, in some studies, the former was confounded with the latter. A new attempt made in Graham’s study was to examine the effects of anxiety about English listening and speaking, two skills relevant to all content teachers in EMI classes, rather than the effect of FLA as a whole. Furthermore, he differentiated L1 (first language, i.e., Mandarin Chinese) teaching anxiety from EMI teaching anxiety. One important finding of the study was that EMI teaching anxiety was not only distinct from L1 teaching anxiety, but also from English listening and speaking anxiety. This finding points to the need to distinguish anxiety about teaching through L2 from anxiety about using L2 in general. Another finding was that English speaking anxiety was a significant predictor of pre-service content teachers’ EMI teaching anxiety at both the planning and execution stages whereas English listening anxiety showed no significant predictability for EMI teaching anxiety. This finding suggests the advantage of investigating skills-specific language anxiety over overall FLA in EMI contexts when a more refined understanding of the role of FLA is desired.

Chou employed structural equation modeling to examine how FLA mediated the relationships of university students’ task goal orientation and perceived English communication competence with willingness to communicate (WTC) in EMI classes. In her study, two dimensions of WTC were distinguished: language-focused (willingness to talk about language, such as the pronunciation and meaning of an English word) and subject-content focused (willingness to communicate about subject content through, for example, discussion or presentation in English). By so doing, Chou was able to reveal that the effects of FLA differed in these two dimensions. When the communication was about subject content, FLA had a significant negative effect on WTC. In addition, it fully mediated the relationship between perceived English competence and WTC, and partially mediated the relationship between task goal orientation and WTC. In contrast, when the communication was about language, FLA did not have a direct effect on WTC, nor did it mediate the relationships of perceived English competence and task goal orientation with WTC. Chou’s study showed that it is important to consider the content of the communication when investigating the influence of FLA in EMI classes. Having said that, since communication in EMI classes at college is largely about subject content, the results suggest that FLA is an issue in college EMI classes that should not be overlooked.

As pointed out previously, most of the existing research on FLA was conducted with students learning English as an L2. Few studies focused on FLA in EMI contexts where L2 English is used to learn academic subjects, let alone in L2 French medium instruction (FMI) contexts. García-Castro and O’Reilly’s study made a valuable contribution to FLA research by comparing the effect of FLA on university students’ online learning engagement in both L2 EMI and L2 FMI contexts in Costa Rica. English and French are of different degrees of linguistic similarity to Spanish, the students’ L1, and of different instrumental value in Costa Rica. This comparison is meaningful because it may shed light on how the impact of FLA differed with the languages used for instruction. Interestingly, this study found that FLA had a consistently positive effect on online learning engagement in both contexts, meaning that students with higher levels of FLA were more engaged in online learning regardless of the medium of instruction, even though students in both contexts requested that the university help them cope with their FLA when learning online. This result contradicts the findings of most of previous research, which showed that FLA had a negative impact on L2 learning in physical classrooms. Given the ubiquity of online learning since the onset of the COVID-19 pandemic, García-Castro and O’Reilly’s study is timely; it shows a promising research area that takes into consideration the potential impacts of contextual and situational factors in moderating the effects of FLA on learning processes.

Different from the preceding three papers, Maher and King adopted a qualitative approach in their investigation of a classroom phenomenon which is prevalent in Asian classrooms, silence, and its relationship with FLA. While silence in the classroom is commonly understood as the absence of verbal contribution, Maher and King proposed a broad characterization of “silence” as encompassing also students’ verbal contributions in L1 when L2 is expected and truncated verbal contributions in L2. The investigation contexts included both EMI English and EMI content classrooms in a foreign studies university in Japan. Based on a cognitive-behavioral model, the authors conducted cycles of interviews with their participants and asked them to recount their participatory behaviors and the associated positive and negative thoughts and emotions in both actual and hypothetical classroom situations. The distinctive contribution made by this study is the bi-directional nature of silence and speaking-related anxiety. While silence in the classroom is often interpreted as a consequence of FLA, the findings of this study showed that “silence,” as defined in this study, was also a trigger of the latter. The very interesting interview data, often missing in quantitative studies, provided a window through which we can gain a better understanding of how psychologically unsettling using a foreign language as a medium of instruction can be and how important it is to ensure that students are properly supported.

Future Research

The significant impact of FLA, within the complex network of factors, not only on language learning and teaching but also in contexts where FL is used as a medium of instruction, cannot be overstated. (Gkonou et al., 2017) observed that language anxiety, alongside motivation, is one of the most important affective factors that second language acquisition scholars consider when discussing individual differences. We suggest that its importance goes beyond teaching and learning in second/foreign language classrooms, and should be a focus of enquiry for all classrooms where content learning is conducted in another language.

Similar to research on other individual-differences variables, a great majority of the studies on FLA to date have been quantitative, correlational, and cross-sectional in nature. This can be seen from the fact that three of the four papers in this special issue are of this nature. Notwithstanding interesting insights they have provided, studies of this kind are less able to illuminate the complex and dynamic nature of human affect, such as FLA (Gregersen, 2020). As Dornyei and Ryan, (2015) pointed out, research into LA “will need to foreground a more dynamic conception of anxiety” (p. 180). The past decade has witnessed an emerging trend of FLA research that investigates learners’ lived experience of anxiety over different timescales, exploring the complex dynamic nature of anxiety (see for example some of the studies collected in Gkonou et al., 2017) and providing enlightening findings based on multiple sources of data over a period of time (e.g., Boudreau et al., 2018; Gregersen et al., 2014; Kruk, 2018). More studies taking this perspective in formulating research design are much desired in order to offer a more sophisticated and in-depth understanding of the impact of FLA in EMI contexts.