The predictive effect of language achievement on multiple emotions in languages other than English: validating a distal mediation model based on the control-value theory

2.1 The theoretical foundation of emotion

The advance of positive psychology contributed to the birth of the theoretical framework of emotions, further stimulating an increasing body of empirical studies concerning the role of emotions across different language learning contexts. Surrounding emotions, the groundbreaking broaden-and-build theory (Fredrickson 2004) and the control-value theory (Pekrun 2006) provided insights into the pivotal role of emotions in students’ psychological state and academic performance.

The broaden-and-build theory not only differentiated emotions from two valences (positive and negative) but also promoted the merit of positive emotion from an individual’s cognitive, psychological, social, and professional aspects. An exponential body of research has empirically confirmed the benefit of enjoyment and the consequence of boredom and anxiety in language learning (Botes et al. 2022a; Derakhshan 2022; Derakhshan et al. 2022; Li and Han 2022; Wang and Li 2022).

Grounded on but different from the former theory, the CVT emphasizes the complex network among antecedents, achievement emotions, and learning achievement in the academic setting (Pekrun and Perry 2014). Firstly, the CVT diversifies the conceptualization of emotions based on the three-dimensional taxonomy approach, i.e., defining emotions from valence (pleasant/unpleasant), activation (the degree of physiological arousal), and object focus (activity-related/outcome-related) (Pekrun 2006). Following this approach, anxiety could be identified as a negative, activating, and outcome emotion, enjoyment as a positive, activating, and activity emotion, and boredom as a negative, deactivating, and outcome emotion. Secondly, CVT confirms the merit of achievement emotions on activities/outcomes, whereby it affects ‘the cognitive, motivational, and regulatory’ perspectives (Pekrun 2006, p. 326). Of note, achievement emotions are domain-specific which might perform to varying degrees across languages and contexts. Thirdly, the CVT emphasizes a bi-directional network between achievement emotions and language achievement. In other words, emotions and academic performance are interacted and mutually affect each other in a circular loop. Some recent work empirically echoed the complex network between emotions and learning achievement (Li 2021a; Li and Han 2022; Li and Wei 2023). This theory theoretically rationalizes the conceptual framework of the current study.

Prior work, however, converges into one fact that the adoption of the CVT in SLA is largely confined to English language learning, limiting its applicability across languages and contexts. In addition, it remains under-investigated whether the merit of language achievement on a collective of emotions could be endorsed. Thus, this study aims to extend the CVT by scrutinizing the influence of L2 Chinese achievement on enjoyment, boredom, and anxiety in a less-featured Chinese learning context in New Zealand.

2.2 Enjoyment, boredom, and anxiety

The movement of emotions has progressed along with positive psychology in SLA in three stages (Li 2021b). In the embryonic stage, researchers mainly focused on anxiety due to its debilitative role in language learning (Horwitz et al. 1986), in addition to the slow development of theoretical frameworks and instruments in the field. In the later developing stage, some groundbreaking studies shifted attention from pure anxiety to enjoyment or the combination of anxiety and enjoyment in English language learning (MacIntyre and Mercer 2014). For example, Dewaele and MacIntyre (2014, 2016 associated enjoyment with anxiety and regarded them as ‘two faces of Janus’ (2014, p. 237) and the ‘right and left feet’ (2016, p. 215) of language learning. In addition, Dewaele and his co-author developed a 21-item foreign language enjoyment scale and adapted a short version of an 8-item foreign language anxiety scale. Jiang and Dewaele (2019) and Li et al. (2019) further validated the short version of the foreign language anxiety scale and developed an 11-item foreign language enjoyment scale, promoting an exponential body of ensuing emotional studies. In recent years, boredom has gradually emerged in the field (Derakhshan 2022; Derakhshan et al. 2022; Li 2021b; Zhao and Wang 2023a). The integration of enjoyment and boredom/anxiety (Alrabai and Alamer 2022; Dewaele and MacIntyre 2022; Dewaele and Li 2022; Kruk et al. 2022; Zhao and Wang 2023b), and even the combination of enjoyment, boredom, and anxiety (e.g., Alrabai 2022; Dewaele et al. 2022a, 2022b; Shao et al. 2023; Wang and Li 2022) have emerged and continuously attracted the attention of scholars.

The relationship between enjoyment and boredom/anxiety has also been well studied in L2 learning. For instance, anxiety and boredom were negatively associated with enjoyment, and two negative emotions positively correlated with each other in English language learning among 484 Saudi Arabia university learners (Alrabai and Alamer 2022). This relationship has also been validated among 504 (middle school) and 783 (middle and high school) Chinese ethnic minority students (Zhao and Wang 2023a, 2023b). Similarly, this pattern has also been found in a bulk of studies (e.g., Tsang and Dewaele 2023; Wang and Li 2022). Moreover, it is significant to note that this configuration has been stretched and validated in the online education context for Chinese and European learners (Li and Han 2022; Resnik and Dewaele 2023).

Informed by the prior work, it could be inferred that no prior studies have substantiated the contribution of language achievement on a collective emotion in the L2 Chinese learning mechanism, necessitating further investigation.

2.3 The relationship between enjoyment, boredom, anxiety, and language achievement

Inspired by the promotion of CVT, a blast of research concerning the relationship between emotions and language achievement has emerged in recent years. There is a consensus that enjoyment was positively correlated, while anxiety and boredom were negatively correlated with language learning achievement across different settings (Li and Wei 2023; Tsang and Dewaele 2023; Zhao and Wang 2023b).

For instance, Derakhshan et al. (2022) suggested that classroom atmosphere and boredom directly predicated students’ engagement among 287 Iran universities English learners, and the negative effect of boredom has been validated in the online Iranian EFL context (Derakhshan et al. 2021a, 2021b). Botes et al. (2022b) found that anxiety and enjoyment statistically predicted self-perceived foreign language achievement in a polynomial regression for 1,039 online learners. Their further meta-analysis based on 56 studies disclosed the positive predictive effect of enjoyment on WTC and actual and self-perceived language achievement. A longitudinal study by Li and Wei (2023) confirmed the correlation of anxiety, boredom, and enjoyment with English learning achievement and stressed the endurability of enjoyment among 954 secondary students in rural China. Congruently, Zhao and Wang (2023a) affirmed the predictive effect of enjoyment (positive) and boredom (negative) on English language achievement among Chinese ethnic minority learners. The same results have also appeared in Li and Han (2022) as well as Wang and Li (2022), empirically echoing the CVT.

It is intriguing that the significant association between emotion and language achievement may fluctuate if a collective emotion simultaneously enters a sophisticated model. For example, Li and Han (2022) found that enjoyment significantly and positively correlated with both actual and self-perceived online language achievement among 348 Chinese university EFL learners. However, only anxiety predicted actual language achievement, while enjoyment and boredom predicted self-perceived language achievement in the regression model. In the same online learning context, Wang and Li (2022) stated that participants’ actual performance was predicted by enjoyment and anxiety, yet boredom did not function as a significant predictor among 819 Chinese university learners in the stepwise regression analysis. Similarly, Dewaele et al. (2022) identified only anxiety, neither enjoyment nor boredom, significantly predicted English achievement based on 332 international samples in the exploratory structural equation modeling. Thus, the current study will explore their complex relationship by employing an advanced rigid model.

Despite the significant path from emotion to language achievement drawing sufficient attention from researchers, scarce attention has been paid to the reverse loop in terms of the predictive effect of language achievement on emotion in language learning. Dewaele and MacIntyre (2014) confirmed the instigation of self-perceived English achievement on enjoyment among an international sample that those perceived as more capable of English language learning are prone to experience more language enjoyment. Li et al. (2020) investigated the role of actual English achievement on general emotions within the Chinese context. The results indicated that English actual achievement was the main cause and two emotions. Alamer and Lee (2021) found that L2 achievement predicted anxiety but not the opposite through cross-lagged panel analysis among 160 English university learners in Saudi Arabia. Focusing on the domain-specific emotions, Dewaele and Li (2022) discovered that enjoyment was co-predicted by self-rating language competence and the domain reading, grammar, and speaking competence, while anxiety was predicted by speaking, grammar, self-perceived and actual competence among 1,415 middle school Chinese students.

Taken together, it is not hard to find that researchers mainly focus on the one-way pattern, i.e., from emotions to language achievement in English language learning. Very few studies examined the reverse loop from language achievement to emotions in the L2 mechanism. Specifically, no research has been conducted to examine the predictive effect of L2 achievement on collective enjoyment, boredom, and anxiety by considering the mediating paths in a model under the Chinese learning context. Such insufficiency might cause the incomplete understanding concerning the role of emotions in L2 learning. Furthermore, language educators might not provide students with adequate learning strategies and timely emotion regulations. Thus, to offer learners with a cheerful and fruitful L2 learning experience, this study will offer both empirical and theoretical insights by mitigating the gap in L2 Chinese language learning.

2.4 The present study

Rooted in the CVT, there should be a bi-directional reciprocal pattern between emotions and learning achievement in the educational setting. The fact is that only the predictive effect of emotion on language achievement has largely been investigated. The reverse loop, however, remains largely unknown in L2 learning. Given this, the present study proposed the hypothesized model to address the following two research questions (see Figure 1).

Figure 1:

The hypothesized L2 Chinese learning mechanism.

3.1 Participants and contexts

Embracing a convenience sampling approach, 307 pupils from 14 elementary and secondary schools in New Zealand, including 157 (51.1 %) females and 150 (48.9 %) males, participated in the current study. The participants were aged from 9 to 20 (M = 11.48, SD = 2.20). The present study invited participants from three different educational backgrounds to obtain a fuller picture. Specifically, 175 of them (57 %) were elementary school students, 84 (27.4 %) were junior high school students, and the remaining 48 (15.6 %) were senior high school cohorts. In terms of the language profile, most of them (n = 129, 42 %) classified their Chinese competence as intermediate, followed by beginner-intermediate (n = 103, 33.6 %), advanced (n = 43, 14 %), and beginners (n = 32, 10.4 %). Among them, 225 (73.29 %) were New Zealanders, followed by 76 Chinese (24.76 %) and 6 others (1.95 %). It was worth mentioning that over 95 % of participants were Chinese heritage students. Of all the participants, 209 (68.08 %) of them speak Chinese as L1, and the remaining 98 (31.92 %) speak English as L1.

New Zealand is one of the largest recipients of migrants in the Asia-Pacific region. Asians are the third largest ethnic group in New Zealand, and the Chinese account for the majority of the Asian group. It is necessary to point out that Chinese descendants develop as the fastest-growing non-indigenous ethnic group due to the immigrant policy. Two-thirds of the Chinese immigrants in New Zealand reside in Auckland, making Chinese Mandarin one of the most extensively used languages there (Wang and Chik 2022). In view of such a need, Chinese language cram schools are prevalent among Chinese heritage students. Parents with L2 academic demands on their children are likely to choose an appropriate cram school for their children to enhance their language achievement.

However, such a large group of Chinese heritage learners in New Zealand and the cram school learning context have yet to attract attention in the field, necessitating the present investigation to understand the contributing effect of academic achievement on enjoyment, anxiety, and boredom in the current study’s context.

3.2 Instruments

The present study employed a two-section composite questionnaire to obtain both the demographic information (age, gender, nationality, educational background, and self-perceived and actual language proficiency) and the rating items on emotions (anxiety, enjoyment, and boredom) from participants in the New Zealand context. To enhance the accessibility and reliability of the questionnaire, all items were provided bilingually both in Chinese and English. Before distributing the questionnaire, two professors in applied linguistics checked the content validity of the questionnaire (Creswell and Poth 2018). After revising some expressions and wordings, the questionnaire was officially given to the participants.

Considering the Chinese language proficiency and the age profile of participants, the current study has controlled the number of items, that is, extracted and adapted items from previous scales, which complied with the suggestions made by Kenny (1979). The item pool of the questionnaire was selected based on item response theory (IRT) (Embretson and Reise 2000). IRT offers powerful techniques for questionnaire design, including item selection and scale shortening (Fayers 2004, p. 715). To measure three emotional variables, all items were rated on a standard 7-point Likert scale from the lowest 1 (totally disagree) to the highest 7 (totally agree). Reliability analysis, exploratory factor analysis (EFA), and confirmatory factor analysis (CFA) were performed in succession. The factor loading of each item and the discriminant and convergent validity were illustrated in Tables 1 and 2.

L2 enjoyment scale. To assess the perception of enjoyment in learning Chinese, the present study extracted and adapted four items from the Chinese version of the Foreign Language Enjoyment scale by Li et al. (2018). A sample item is “I enjoy learning Chinese”.

L2 boredom scale. Consisting of five items, the L2 boredom scale was adapted from the Foreign Language Learning Boredom Scale by Li et al. (2021). The sample item starts with “I always feel sleepy in the Chinese classes”.

L2 anxiety scale. To measure the anxiety level of the participants, the present study adapted four items of a short version of the Foreign Language Classroom Anxiety Scale (Dewaele and MacIntyre 2014, 2016). A case item is “Even if I am well prepared for the Chinese listening and speaking class, I feel anxious about it.”

Actual L2 Chinese language achievement (ALA). The current study used participants’ midterm test scores to represent their actual language achievement in the first section of the questionnaire survey. To ensure the accuracy of responses, each classroom teacher was invited to double-check the authenticity of examination results.

3.3 Data collection

The present study obtained the official ethical document from the Ethics Committee of the author’s university. Before officially distributing the questionnaire, the authors contacted the principal and each classroom instructor of the target school in New Zealand for collaboration and approval of the study. Upon receiving their permission, the authors informed the aim, procedure, time duration of answering the questionnaire, and their right to withdraw during the data collection. The anonymity and confidentiality of participants were guaranteed, and the data would never be revealed to a third party. Written consent forms from all participants were obtained. As some participants were under the age of 18, we also sent text messages to their parents for their consent before conducting the survey. Finally, a paper-and-pencil form questionnaire was distributed to participants at the end of their formal class from December 15th to 30th, 2022.

3.4 Data analysis

After collecting the questionnaire, five random missing data (n = 5, 1.6 %) of the actual language achievement were identified, and then the imputation technique with the series mean method was embraced to replace the missing data. Subsequently, to ensure the following stages of statistical analysis, the normality of distribution (skewness and kurtosis), the reliability (Cronbach’s alpha) and validity (the exploratory factor analysis/EFA and the confirmatory factor analysis/CFA) of enjoyment, boredom, and anxiety were rigorously examined. Three hundred and seven (N = 307) sample cases were used for the following analysis with SPSS v26 and Amos v26.

The complex relationship between multiple emotions and L2 Chinese learning achievement was investigated using structural equation modeling (SEM) with the bootstrapping technique. SEM can compensate for the shortcomings of traditional regression statistical methods, allowing analysis of the causal relationship among multiple factors in a model (Creswell and Poth 2018).

4.1 Preliminary analysis

Before reporting the results of the research question, we conducted a series of preliminary analyses to establish the assumptions for subsequent analysis. The results of descriptive statistical analysis showed that the skewness and kurtosis of all the items fell within ±2, indicating the normal distribution of the data set (George and Mallery 2020). Further, the EFA presented that each construct has good internal reliability, with Cronbach’s alpha above 0.70 (Hair et al. 2019) and factor loading above 0.702 (see Table 1).

After that, the CFA was performed to fulfill the assumption for SEM analysis. The results of the CFA indicated that all the constructs had good model fit indices (χ ²/df < 3, comparative fit index (CFI) > 0.90; Tucker-Lewis index (TLI) > 0.90; root means square error of approximation (RMSEA) < 0.08; and standardized root means square residual (SRMR) < 0.08) (Kline 2011). In addition, the three emotional constructs of the current study, as shown in Table 2, had good convergent validity with composite reliability (CR) > 0.70 and the average variance extracted (AVE) > 0.50. The constructs also had discriminant validity as the square roots of the AVE were higher than the correlation coefficients between the corresponding factors (Hair et al. 2019).

Table 2 illustrates the level of a constellation of emotions in L2 Chinese learning. The results indicated that New Zealander L2 Chinese participants experienced a high level of enjoyment (M = 4.68, SD = 1.35), a moderate to a high level of anxiety (M = 3.06, SD = 1.48), and a low to moderate level of boredom (M = 2.87, SD = 1.41). In addition, the frequency of descriptive analysis reported that the proportion of participants who indicated a moderate to high level of enjoyment (86.97 %) was twice as students’ percept of boredom (41.04 %) and anxiety (43.97 %). The perception of anxiety was higher than boredom in the moderate to high range in learning Chinese.

4.2 The direct and indirect effects of ALA on emotions

Before examining the direct and indirect paths in the underlying L2 Chinese learning mechanism, the goodness of fit indices of the proposed SEM model was firstly evaluated. The results indicated that the model fitted the sample data excellently (χ ²/df = 2.314, CFI = 0.957; TLI = 0.946; RMSEA = 0.066; and SRMR = 0.050), ensuring the subsequent statistical analysis. Then, the SEM was analyzed, and Figure 2 illustrates the final results.

Figure 2:

The finalized model of L2 Chinese learning mechanism.

Direct effects. This study assessed the predictive effect of actual learning achievement on each emotion and the direct path among different emotions in the SEM with Amos v26. As shown in Table 3, ALA significantly predicted anxiety (H1: β = −0.288, p < 0.001) and enjoyment (H3: β = 0.135, p < 0.01). In addition, anxiety was a significant positive predictor of boredom (H5: β = 0.693, p < 0.001), and boredom was a significant negative predictor of enjoyment (H6: β = −0.710, p < 0.001). To put it another way, participants with a higher level of Chinese language achievement are likely to have a lower perception of anxiety but a higher level of enjoyment. Anxiety was positively correlated with boredom, while boredom, in turn, was negatively related to enjoyment. The two remaining direct paths, H2 (β = −0.091, p > 0.05) and H4 (β = −0.065, p > 0.05), were not significantly identified. However, these findings were important because the insignificant direct effect of ALA on boredom suggested that anxiety fully mediated the association between ALA and boredom, and the insignificant direct effect of anxiety on enjoyment indicated that boredom functioned as a full mediator between anxiety and enjoyment (MacKinnon 2008).

Indirect effects. Regarding the mediating paths in the SEM, the present study found three significant indirect effects (Table 4). Anxiety mediated the relationship between ALA and boredom (β = −0.200, p < 0.001; 95 % CI = −0.304 to 0.112), and boredom mediated the path from anxiety to enjoyment (β = −0.492, p < 0.001; 95 % CI = −0.686 to −0.355). One of the most intriguing findings in the current study was that double mediators of L2 anxiety and boredom mediated the association between ALA and L2 enjoyment (β = 0.142, p < 0.001; 95 % CI = 0.080 to 0.240). Therefore, a distal mediating effect (ALA → Anxiety → Boredom → Enjoyment) was established in the current study. However, both the paths, ALA → Boredom → Enjoyment and ALA → Anxiety → Enjoyment, were not found to be statistically significant.

To determine whether the statistically significant findings are meaningful in practice, we computed the effect size/ES (f ²) of the endogenous variables with the multiple squared correlations/SMC (R ²) as suggested by Cohen (1988, 1992. According to Cohen (1992, p. 157), the small, medium, and large effect sizes (f ²) are 0.02, 0.15, and 0.35, respectively. The results indicated that ALA had a small effect size on anxiety (R ² = 0.09, f ² = 0.10). However, this variable, together with other mediators as shown in Figure 2, had large effect sizes on boredom (R ² = 0.41, f ² = 0.69) and enjoyment (R ² = 0.53, f ² = 1.13).