Foreign language enjoyment and foreign language classroom anxiety among Chinese learners of German

: This study examines both the foreign language enjoyment (FLE) and the foreign language classroom anxiety (FLCA) of 201 Chinese university students of non-German majors who were learning the German language. A questionnaire based on the FLE scale and the FLCA scale was employed to examine the participants’ classroom emotions as well as other learner-and teacher-related variables. Participants reported a higher level of FLE than FLCA, with these two emotions being negatively correlated. Also, regression analysis showed that FLE can be strongly predicted by teacher-related variables, while FLCA is predicted by learner-related variables. Moreover, qualitative analysis of the open-ended questions in the questionnaire showed that learners’ FLE in the German classroom was boosted by their sense of achievement in learning a new language and frequent interaction between the participants and their teachers and classmates. At the same time, the relatively high demands of German language learning and frequent exams resulted in higher FLCA. The findings suggest that teachers could organize diverse classroom activities and integrate more culture-related content to enhance their students’ FLE. Meanwhile, they could also adopt more flexible methods of assessment and provide clearer instructions thereon to help students to be more prepared, thereby ameliorating their FLCA.


IntroductIon
The processes and achievements of foreign language (FL) learning are significantly influenced by learners' emotions, which are in turn affected by both internal and external variables.With increased attention to learners' emotions in the FL classroom, over the past decade an increasing number of studies have focused on both positive and negative emotions (e.g., Dewaele & MacIntyre, 2014;Jiang & Dewaele, 2019).In these studies, foreign language enjoyment (FLE) and foreign language classroom anxiety (FLCA) are seen as the most common emotions in FL learning (Piniel & Albert, 2018), regarded as "the right and left feet of the language learner" (Dewaele & MacIntyre, 2016, p. 215).It also seems that the FLE and FLCA of Asian learners have distinct characteristics compared to those of learners in other parts of the world (e.g., Dewaele & MacIntyre, 2014;Dewaele et al., 2018;Jiang & Dewaele, 2019), but current researchers have mainly focused on learners in English-as-a-foreign-language (EFL) classrooms, especially in the Asian context (MacIntyre & Vincze, 2017;Li et al., 2018).However, as Dewaele and MacIntyre (2022) point out, these studies may give a misleading picture by suggesting that English-based findings are universal.In order to redress this balance, there is a strong need to focus on languages other than English (LOTEs) in terms of classroom emotions.
Therefore, by focusing on Chinese learners in German-as-a-foreign-language (GFL) classrooms, this study aims to examine the characteristics of Chinese LOTE learners' classroom emotions, including the effects of a range of internal and external variables related to the learners' FLE and FLCA as well as the sources of these two emotions.

Learners' emotions in FL classrooms
FL classrooms are rich contexts for various emotions such as enjoyment, anxiety, boredom, and so on (Bown & White, 2010;MacIntyre & Gregersen, 2012).Many empirical studies have confirmed the pattern of learners' emotions in FL classrooms and revealed the importance of emotions in FL learning (e.g., Dewaele & MacIntyre, 2014;Jiang & Dewaele, 2019).The present paper focuses on FLE and FLCA, which have been shown to play a significant role in FL learning.
FLE has been defined as "a complex emotion, capturing interacting dimensions of challenge and perceived ability that reflect the human drive for success in the face of difficult tasks" (Dewaele & MacIntyre, 2016, p. 216).Previous findings have shown that there is a positive correlation between FLE and language achievement, and that FLE is one of the key emotions for predicting language achievement (Jiang & Dewaele, 2019;Li et al., 2020).
As for its negative counterpart, FLCA may be seen as a unique set of self-perceptions, beliefs, or behaviors connected to learning in a classroom setting that are specific to the language learning process (Horwitz et al., 1986).It has been suggested that it can harm FL learning outcomes (MacIntyre, 1995(MacIntyre, , 2017)); over the past four decades negative correlations between FLCA and classroom performance as well as FL achievement have been indicated by various studies across the globe and among different kinds of learners (e.g., MacIntyre, 1999;Horwitz, 2001;Dörnyei & Ryan, 2015).
As the most commonly studied positive and negative emotions in FL classrooms (Mac-Intyre, 2017; Piniel & Albert, 2018), FLE and FLCA were first juxtaposed by Dewaele and MacIntyre (2014), who found a moderate negative correlation between these two emotions.Moreover, they proposed that "FLE and FLCA are different dimensions and not two sides of the same coin", and that these two emotions coexist in FL learners (Dewaele & MacIntyre, 2014, p. 265;2016).Researchers then began to investigate learners' FLE and FLCA in FL classrooms, demonstrating that learners' classroom emotions are affected by learner-internal (e.g., age, gender, language proficiency) and learner-external factors (e.g., teacher's personality, classroom activities) (Dewaele et al., 2016).One intriguing finding is that levels of learners' FLE and FLCA differ to some extent in different cultural environments, highlighting the importance of exploring FL learners' positive and negative emotions across various cultural contexts (e.g., MacIntyre & Vincze, 2017;Jiang & Dewaele, 2019).
By examining the levels of FLE and FLCA among 564 Chinese undergraduate EFL learners, Jiang and Dewaele (2019) found that Chinese EFL learners have a higher level

Monograph VIII
November 2023 of FLCA compared to other learners such as those studied by Dewaele and MacIntyre (2014).The earlier research indicated gender differences in FL learners' FLE and FLCA, characterized by higher FLE and FLCA reported by female learners in comparison with their male peers (Dewaele & MacIntyre, 2014;Dewaele et al., 2016).However, Jiang and Dewaele (2019) pointed out that there were no gender differences in either Chinese FL learners' FLE or their FLCA.Moreover, they found that teachers' unpredictability had no significant effect on Chinese FL learners' FLE, but it was positively correlated with their FLCA (Jiang & Dewaele, 2019).In contrast, teachers' predictability had no effect on Western FL learners' FLCA, while higher levels of predictability were correlated with lower levels of FLE (Dewaele et al., 2018).These findings confirm that FL learners' classroom emotions are influenced by both internal variables and external variables and may vary in different cultural contexts.
Notably, although many studies have been carried out on FLE and FLCA in the context of China, the vast majority of studies have focused on EFL classrooms, "which could create an erroneous view that the findings based on English are universal" (Dewaele & MacIntyre, 2022, p. 157).Given that the international status of English as a global lingua franca has an impact on learners' motivation and investment in learning LOTEs (Dörnyei & Al-Hoorie, 2017;Lu et al., 2020), this dominance may also affect emotions in the LOTE classrooms to some extent.The global currency of English significantly enhances its superior role in cross-cultural communication; Zheng et al. (2019) found that in terms of instrumental value, FL learners are inevitably affected and attracted by English in their decision-making when choosing an FL to learn, especially in non-Anglophone contexts.This is particularly true for FL learners in China, where English is regarded as having the most prioritized symbolic capital, with its status unmatched by other FLs (Zheng et al., 2019;Zheng & Liu, 2021).Moreover, this instrumental value is also closely associated with the acquisition of rich socio-economic and educational resources (Bolton & Graddol, 2012).In comparison, the level of social support granted to LOTEs is typically lower than for English (Dörnyei & Al-Hoorie, 2017); for instance, English is a compulsory subject in secondary and tertiary school curricula in the Chinese educational context (Cheng & Li, 2021), but LOTE learning is generally voluntary or optional (Zheng & Liu, 2021).
As a result of all these factors, LOTE learning in China may not be exactly equivalent to EFL learning.As Dewaele and MacIntyre (2022, p. 158) have argued, "we cannot assume that the emotions for LOTEs mirror those for English, which occupies a unique position among languages and is by far the dominant target language".Based on this review, we believe it is crucial to examine the classroom emotions of Chinese LOTE learners.To this end, we investigated the factors that influence learners' FLE and FLCA in LOTE classrooms in China by focusing on GFL classrooms in Chinese higher education.Two research questions guided our study: RQ1: What are the levels of Chinese GFL learners ' FLE and FLCA, and what is the relationship between these two emotions?RQ2: What factors are associated with Chinese GFL learners ' FLE and FLCA, and what are the sources of these two emotions?

Participants
The 201 participants in the present study were taking a one-year course, College German, at a university in East China.All the participants were non-German major students and were studying different subjects at the university.For some students the College German course was compulsory to obtain a degree, while for others it was not.Hence, the participants in this study were naturally divided into two groups, the compulsory group and the non-compulsory group.They all used the same syllabus, learning materials, and exams, with a similarly high weight of class hours (16 hours per week for the compulsory group and 18 for the non-compulsory group).Of the 201 informants, 134 were male and 67 were female, and 141 were from the compulsory group while 60 were from the non-compulsory group.

VarIaBLe
CaTegory FreQuenCy

Course type Compulsory 141
Non compulsory 60

Questionnaire to examine learners' FLE and FLCA
The questionnaire was written in Mandarin Chinese and started with a section collecting the participants' demographic information (i.e., gender, major, and student number) and asking whether they had learned German before attending the current course.They were then asked to rate their German language skills.Four items related to the four language skills were assessed using a 5-point Likert scale ranging from 1 (very much in need of improvement) to 5 (very satisfactory), including listening (Mean = 1.80,SD = 0.97), speaking (Mean = 2.06, SD = 1.00), reading (Mean = 2.95, SD = 1.14), and writing (Mean = 2.63, SD = 1.11).
In addition, participants were asked to complete a FLE scale and a FLCA scale.Both of these scales were based on Jiang and Dewaele (2019).The FLE scale was composed of ten positively formulated items measuring the participants' enjoyment.The FLCA scale contained six positively formulated items and two negatively formulated items measuring the participants' anxiety.

November 2023
The questionnaire ended with two open-ended questions.In the first question the participants were asked to describe one of their most enjoyable experiences in the GFL classroom.Then they were asked to describe one of their most anxious experiences.

Participants' exam performance
In addition to the questionnaire, we also collected data on participants' exam performance.The participants' scores on the most recent monthly exam (a self-designed diagnostic exam with reference to the National College German Test Band 4), which took place before the distribution of the questionnaires, were recorded.Considering that participants with same score may have different rankings in different classes, presumably affecting the participants' emotions, we also collected the participants' exam rankings in their classes.To reduce the effect of class size, the exam rankings were measured by dividing the participant's rank by the class size (i.e., exam ranking = rank in class / number of students in the class).These two variables (i.e., exam scores and exam rankings) were grouped together as learner-related variables to explore their relationship with the participants' FLE and FLCA.

Data collection procedure
A pilot study was conducted approximately two weeks before the data collection for the main study.Based on the feedback of three participants from the non-compulsory group, we revised our questionnaire to ensure that the content was intelligible and clear to the participants.We then distributed paper-and-pencil questionnaires to all 276 students taking the College German course in early December 2020, and 246 completed questionnaires were collected (a response rate of 89.13%).A final total of 201 valid questionnaires (i.e., where all the required questions were answered) were collected and used for the analyses to follow.

Data analysis
The present study is a mixed-method investigation into both FLE and FLCA.Using the questionnaire results, both quantitative and qualitative data were collected.Since the participant sample size was large (n > 200), we used Q-Q plots to check the normality of their FLE and FLCA scores (cf., Field, 2013).Both of the Q-Q plots shown in Figure 1 demonstrate an approximately normal distribution.Hence, parametric tests were adopted for the data analysis with the aid of SPSS Version 26.Foreign language enjoyment...  2019), the emotion-related descriptions were first coded into three main categories, i.e., FLE/FLCA-self, FLE/FLCA-teacher, and FLE/FLCA-peer.Then specific sources were developed under each category.Specifically, FLE/FLCA-self indicated that the participants themselves were the primary cause of their emotions, while FLE/FLCA-teacher emphasized teachers as the direct cause of enjoyment or anxiety.Finally, FLE/FLCA-peer concerned emotions that were directly caused by the behavior of fellow students or by their interactions.

Levels of FLE, FLCA, and their relationship
The mean scores of the 5-point Likert scales for the participants' FLE and FLCA were 4.03 (SD = 0.62) and 3.42 (SD = 0.82), respectively.A paired t-test revealed that participants perceived a significantly higher FLE than FLCA in their GFL classes (df = 200, t(201) = 7.158, p < .001).Also, Cohen's d was 0.505, suggesting a medium effect size (Cohen, 1988).Moreover, Pearson's correlation analysis showed a significant and negative relationship between the participants' FLE and FLCA (r = -0.416,p < .001).According to Cohen (1988), this correlation coefficient indicates a medium effect size.

Two confounding variables
In the current study there were two possible confounding variables that might have had an impact on learners' classroom emotions, namely participants' gender and course type.Gender has been found to affect learners' FLE and FLCA (e.g., Dewaele & MacIntyre, 2014;Dewaele et al., 2018).Hence, we first examined its effect on learners' emotions.

November 2023
Since Kolmogorov-Smirnov tests showed a non-normal distribution of the FLE scores of both the male and female participants (p < .05)and a normal distribution of their FLCA scores (p > .05),we employed non-parametric tests to compare the difference between the FLE scores for these two groups and parametric tests for their FLCA scores.The results of an independent-samples Mann-Whitney U-test (p > .05)and an independent-samples t-test (p > .05)suggested that the two gender groups showed no significant difference in their FLE or FLCA scores respectively.
With regard to the second confounding variable, as stated in Section 3.1 our participants were naturally divided into two groups according to their course type (i.e., a compulsory group and a non-compulsory group).Kolmogorov-Smirnov tests showed that the FLE and FLCA scores from the compulsory group and the FLCA scores of the non-compulsory group were not normally distributed (p < .05).An independent-samples Mann-Whitney U-test indicated that the differences in the FLE and FLCA scores between students from the two study groups were not statistically significant (p > .05).Thus, these two groups were collapsed together in the subsequent analyses.

Learner-and teacher-related variables
This section focuses on the relationships between the learner-and teacher-related variables and the participants' FLE and FLCA scores, as well as the sources of the two emotions.
A series of correlational analyses were conducted to explore the relationships between the associated factors and the participants' FLE and FLCA scores.The results showed that the learners' self-assessments of the four language skills and their exam scores, as well as the strictness, friendliness, and humor of the teacher, were all significantly positively correlated with FLE (all p < .05),while the learners' exam rankings were significantly negatively correlated with FLE (p < .05).However, there was no significant relationship between the teachers' unpredictability and the learners' FLE (p > .05).
Moreover, the learners' FLCA was significantly negatively correlated with their self-assessments of the four skills, their exam scores, and the humor of the teacher, while the learners' exam rankings and the unpredictability of the teacher were positively correlated with FLCA (all p < .05).There were, however, no significant correlations between the teachers' strictness or friendliness and the learners' FLCA (both p > .05).To further identify the predictors of the participants' FLE and FLCA, we conducted multiple stepwise linear regression analyses.According to Field (2013), the stepwise linear regression method is suitable for exploratory model building.To avoid issues with multicollinearity, only the independent variables that were significantly (p < .05)associated with the participants' FLE and FLCA scores were included in the regression model.A summary of our regression analysis predicting learners' FLE is shown in Table 3.A significant regression equation was found for FLE (F = 31.12,p < .0001,R 2 = 0.444, R 2 Adjusted = 0.43).The adjusted R 2 value suggests that these five independent variables can explain 43% of the variation in the FLE scores.According to the regression models, the strongest predictor of FLE was the teachers' humor, followed by the teachers' friendliness, the learners' self-assessment of reading, the learners' scores in the German exam, and the teachers' strictness.Since the variance inflation factor was well below 10 and tolerance was not lower than 0.20, these independent variables show no clear evidence of multicollinearity (cf., Field, 2013).To investigate how the participants perceived FLE specifically, the sources of FLE were collected and coded (see Table 4).The numbers in brackets stand for the token number of each of the sources.The values in the last row of the table indicate the sum of tokens of the named sources within each main category.One point to note is that the peer interaction that appears under the category FLE-peer was to some degree a result of classroom activities.Thus, as with the quantitative results, the categorization shows that FLE was mainly caused by teacher-related factors.Under the FLE-teacher category, specific teaching activities such as role-playing, drama, singing, and so on were the most frequently mentioned (77 times, accounting for 63.11%).Students' imaginations and creativity were likely to be enhanced by these activities since they were fascinating, interesting and thought-provoking:

Table 2. Summary of correlation coefficients of different variables with FLE and FLCA
The teacher asked us to form sentences in the future tense, i.e., one student completes the sentence given by another student as a condition.Everyone made some very funny sentences, and everyone laughed a lot.(P65) In the FLE-self category, good language performance was most frequently mentioned as a source of enjoyment.Since the one-year German course in this study offered an intensive schedule, the students in these GFL classes had more opportunities to showcase their German skills during activities, such as giving presentations in front of the class and answering questions.Their sense of achievement could thus directly induce FLE: I gave a well-prepared presentation on a family photo.It felt great and was probably one of the highlights of my German class.(P168) In terms of FLE-peer, the most frequently mentioned source of FLE was peer interaction which was mainly facilitated by classroom activities.For instance: I gave a well-prepared presentation on a family photo.It felt great and was probably one of the highlights of my German class.(P168) Both the quantitative and qualitative results showed that FLE was mainly related to teacher-related factors.The same regression analysis was performed for the learners' FLCA (F = 21.46,p < .0001,R 2 = 0.305, R 2 Adjusted = 0.29).A summary of the regression analysis predicting learners' FLCA is shown in Table 5.This analysis found that the strongest predictor of FLCA was the learners' self-assessment of listening, followed by their exam ranking, the unpredictability of their teachers, and the humor of the teachers.The adjusted R 2 value indicates that these four independent variables can explain 29% of the variation in the FLCA scores.There was also no issue of multicollinearity among these independent variables.As with FLE, the self-reported experiences of FLCA were then collected and coded.Table 6 presents the sources of FLCA mentioned by the participants.Echoing the statistical results, the source of FLCA was mainly learner-related factors.In the category of FLCA-self, exams and tests accounted for the largest proportion.German was a completely new language for most of the participants, so exams and tests led to strong feelings of insecurity.For instance: When talking about the monthly exams, I always felt that I had not learned anything.I was very anxious when I took the exam and when the test results were announced.I felt so bad, i.e., overwhelmed by insecurity, guilt, and regret.(P24) Moreover, there was a range of additional self-related sources that did not fit into the other categories, such as being late to class, poor health, etc.In addition, the FLCA-teacher and FLCA-peer categories were also causes of classroom anxiety.In the FLCA-teacher category, questioning from teachers was the most commonly mentioned cause of anxiety: November 2023 When I couldn't complete the exercises and happened to be asked by the teacher to answer the questions, I felt very frightened, very worried, and very anxious.(P114) Finally, in the FLCA-peer category perceived peer pressure was mentioned most often: In the tests that took place every five units my grades were only average, while the other students did much better.I worried a lot about my grades and my future.I was losing my self-confidence.(P175) A Chi-square test showed that each type of source, i.e., self, teacher, and peer, played a significantly different role in FLE and FLCA (Chi 2 = 164.49,df = 2, p < .0000).The self-factors, i.e., FLCA-self and FLE-self, contributed 84.77% to FLCA but only 26.78% to FLE.The teacher factor influenced FLE (51.05%) more than it did FLCA (10.70%), and the peer factor had a bigger effect on FLE (22.18%) than on FLCA (4.53%).Moreover, for the category FLE-peer, the peer interaction was a result of classroom activities which were organized by the teachers.Therefore, both the quantitative and qualitative results indicated that FLE was mainly predicted by teacher-related factors, while FLCA was primarily predicted by learner-related factors.

dIscussIon
The GFL learners in our study reported a significantly higher level of FLE than FLCA, with a moderate negative correlation between them.Our regression models showed that FLE is more likely to be predicted by teacher-related factors, while FLCA is predicted by learner-related variables.This finding was consistent with the learners' own descriptions of their episodes of FLE and FLCA.
The first research question in this study investigated the level and relationship between the FLE and FLCA of Chinese GFL undergraduates.The participants reported significantly more FLE than FLCA.The mean scores of the GFL learners for both emotions in our study were higher than those of EFL learners (Dewaele & MacIntyre, 2014;Jiang & Dewaele, 2019) and the international LOTE sample in Dewaele and MacIntyre (2022).Moreover, these two emotions showed a moderate negative correlation in our Chinese GFL learners, further suggesting that FLE and FLCA are not two poles in the generic classroom emotion dimension (Dewaele & MacIntyre, 2014).
The second research question focused on the factors that are associated with FLE and FLCA.Our multiple regression analyses showed that the predictive power and weight of teacher-related factors always play a dominant role in terms of FLE; teachers' humor and friendliness were the two strongest predictors of FLE.A humorous and friendly teacher contributes to a more relaxed atmosphere, which will help learners feel more comfortable.Meanwhile, teachers' strictness was also a predictor of FLE in our study, suggesting that teachers' supervision and guidance can facilitate learners' mastery of FL knowledge, thereby enhancing their sense of achievement.
The qualitative data confirmed the dominant link between teachers and FLE, since more than half of the enjoyable experiences reported were teacher-related.Within the Chinese classroom there is a teacher-centered tradition, where teachers determine the content, approach, and pace of instruction while students rely on the teachers' feedback, recognition, encouragement, and support.In the present study the enjoyable episodes described by students contained many student-centered activities, but we need to be aware of the indirect influence of the teacher behind these activities (Li et al., 2018).That is, teachers arrange these activities so that students can delve into the language and culture beyond the information in their textbooks, as well as by interacting and collaborating with their peers.
In terms of FLCA, the associated factors were mainly the learners' lack of confidence in their own language skills, their fear of exams, and perceived peer pressure.In our regression model the students' self-assessment of listening showed the best prediction capacity.Among the four skills, students rated their listening the lowest, and a lower level of proficiency can be expected to generate communicative anxiety in classroom interactions with the teachers.Moreover, the learners' awareness of their FL weakness, even just in a particular skill, is closely related to their confidence, which turns out to have a significant impact on FLCA (Jin & Dewaele, 2018).The second strongest predictor of FLCA in the German classes we studied was the students' ranking.The fact that ranking instead of score predicted FLCA also indicates inter-class differences; it was comparison with other students, rather than their scores per se, that caused learners' anxiety.
In accordance with the statistical analyses, the vast majority of the anxiety-provoking episodes reported were related to exam and classroom speaking performance.The former is directly related to the examination-oriented culture in China (Jiang & Dewaele, 2019;Li et al., 2020), while the latter relates to a challenging classroom activity.Learners worry about whether their speaking skills are sufficient and whether they might lose face in front of their classmates and teacher, resulting in a negative impact on their self-image (Arnold, 2011).
The GFL learners in the current study demonstrated a high intensity of both FLE and FLCA in their German classrooms.In terms of the source of these emotions, the percentages of FLE-self and FLE-peer in our sample were also much higher than those of Chinese EFL learners (Jiang & Dewaele, 2019).This unique pattern could be explained in the following ways.First, GFL classes are taught from scratch, which means that students have a stronger experience of novelty and are more likely to enjoy the pleasure of achievement in language learning.Second, the intensive schedule of German language learning contributes to more teacher-student and student-student contact and stronger collective emotional connections, all of which lead to a more positive learning experience (Chen et al., 2021).Specifically, all the participants were majoring in a subject other than German, so as suggested by multiple students in the open-ended questions, they generally regarded the one-year German course as a pleasant change of pace in their studies.
In addition, the global status of English does not necessarily lead to a better learning experience (Dewaele & MacIntyre, 2022).On the contrary, students may consider learning a new FL to be a unique challenge, which leads to a greater sense of achievement.As pointed out by Oxford (2016), the aim of FL learning is not limited to becoming more proficient in the language; it also relates to becoming a happier and wiser person.
As to the high level of FLCA, the uncertainty and insecurity of learning a new target language can also cause increased anxiety.Learning the German language requires more engagement and concentration than relying on or calling on knowledge accumulated over the years, as is the case with College English for these students.Furthermore, the College German course involves many more class hours than the conventional College English course,

Monograph VIII
November 2023 thus imposing a higher learning burden.The monthly written exams that focus on grammar and vocabulary in GFL classrooms increase the students' anxiety, and also have a greater impact on their GPA due to the higher weight of class hours.

conclusIon
The GFL learners in this study reported a significantly higher level of FLE than FLCA, and these two emotions were negatively correlated.FLE was more related to teacher-related variables, while FLCA was more related to learner-internal variables.Within the teacher-centered tradition in Chinese classrooms, teachers are crucial in creating a positive emotional atmosphere.The emphasis on exams and tests in the Chinese educational environment is a major source of FLCA.In addition, our results further reveal the simultaneously private and interpersonal nature of FL classroom emotions (Li et al., 2018); teacher, self and peer are not isolated elements, but interact in dynamic and complex ways.
Compared with previous studies focusing on EFL learners, the participants in our study demonstrated a stronger emotional involvement in their GFL classroom.The status difference between English and LOTEs in the Chinese educational context did not have a direct impact on students' emotions in the classroom; rather, it was the classroom atmosphere and learning experience that played the main role.The novelty and sense of achievement of learning a new FL instead of English and the feeling of solidarity in class may make learning a less global FL more enjoyable.However, the difficulty of learning a new language, the stress of exams and peer pressure all result in a stronger sense of anxiety.
The present study has some limitations.First, the participants were from only one university in China with a distinctive German curriculum, and the findings cannot automatically be generalized to all Chinese GFL learners.Second, the participants came from eight classes with different teachers, and the study did not conduct inter-class comparisons, thus potentially ignoring the effect of teacher differences on classroom emotions.Third, the study did not examine classroom emotions from a temporal perspective, indicating a need for longitudinal research to capture the fluctuations of emotions in GFL learning.
Despite these limitations, however, the study offers two main pedagogical implications for LOTE teaching in Chinese universities, in relation to classroom activities and classroom tests.In the first of these, teachers should be aware of their role in setting the emotional tone in FL classrooms, especially by organizing novel and stimulating classroom activities, adopting a humorous, friendly, and supportive attitude, and being concerned with students' progress and level of study.For example, more culture-related content could be incorporated into the teaching of language skills, and students could be encouraged to use their multilingual skills to engage in cross-linguistic and cross-cultural thinking.Teachers could give students autonomy in shaping class activities to match their own interests (Dewaele & MacIntyre, 2014).For example, based on the students' level of mastery of reading skills, open-ended discussions could be used to allow students to share their observations and ideas on the reading topics.Moreover, teachers could also organize cooperative learning, promote peer collaboration, and foster inter-peer cohesion (Jin & Dewaele, 2018).
In terms of the second pedagogical implication, our Chinese GFL learners viewed the German language as a tool for enhancing cultural understanding and broadening their Foreign language enjoyment... horizons, rather than as a primary communicative tool (Chen et al., 2021).This suggests that traditional written tests that focus on grammar and vocabulary knowledge and the accuracy of language use may not match the students' motivations.As a result, the use of classroom tests and exams may only increase students' anxiety.Hence, teachers could adopt more flexible and motivating assessment formats such as oral presentations, project reports and so on, providing clear instructions to help students to be more prepared and thereby ameliorating their FLCA.

Figure 1 .
Figure 1.Normal Q-Q plots of FLE and FLCA Based on the two open-ended questions in the questionnaire, the qualitative data were collected and coded to complement and validate the quantitative results.Following Jiang and Dewaele (2019), the emotion-related descriptions were first coded into three main categories, i.e., FLE/FLCA-self, FLE/FLCA-teacher, and FLE/FLCA-peer.Then specific sources were developed under each category.Specifically, FLE/FLCA-self indicated that the participants themselves were the primary cause of their emotions, while FLE/FLCA-teacher emphasized teachers as the direct cause of enjoyment or anxiety.Finally, FLE/FLCA-peer concerned emotions that were directly caused by the behavior of fellow students or by their interactions.

Table 3 .
Summary of regression analysis predicting learners' FLE

November 2023 Table 4 .
Sources of FLE

Table 5 .
Summary of regression analysis predicting learners' FLCA

Table 6 .
Source of FLCA