Introduction
Integrating positive psychology into applied linguistics and language acquisition marked a significant shift in examining language learning emotions (MacIntyre & Mercer, Reference MacIntyre and Mercer2014). Positive psychology holds a holistic view that values both positive and negative emotions, recognizing that managing negative emotions is essential for language learning (Li & Wei, Reference Li and Wei2023; Tsang & Lee, Reference Tsang and Lee2023). Extensive studies have implied the feasibility of addressing foreign language boredom (FLB), a negative emotion that hinders language achievement through reducing motivation and engagement, within the framework of positive psychology (Borgonovi et al., Reference Borgonovi, Pokropek and Pokropek2023; Dewaele et al., Reference Dewaele, Botes and Greiff2023; Dewaele et al., Reference Dewaele, Albakistani and Ahmed2024). Notably, there has been a remarkable surge of interest in researching FLB in recent years, positioning it as a prominent focal point in foreign language emotion research (Li et al., Reference Li, Dewaele and Hu2023; Pawlak et al., Reference Pawlak, Kruk, Zawodniak and Pasikowski2020). Boredom constitutes a prevalent experience during FL learning and potentially detrimentally influences learners’ learning behavior and academic achievement (Dewaele et al., Reference Dewaele, Botes and Greiff2023). Investigating how boredom affects FL achievement can provide valuable insights into the specific mechanisms where FLB impedes language learning progress, which is crucial for educators, informing the design of engaging and effective language learning environments.
Numerous studies have identified different effect sizes of FLB on language achievement (Dewaele et al., Reference Dewaele, Botes and Meftah2023; Pawlak et al., Reference Pawlak, Kruk, Zawodniak and Pasikowski2022), calling for synthesizing analysis across different individual studies. One method to deepen the understanding of this issue is to conduct a meta-analysis, an approach that aggregates data from multiple studies, providing a comprehensive overview and identifying patterns or inconsistencies that individual studies might miss (Borenstein et al., Reference Borenstein, Cooper, Hedges and Valentine2009). By systematically synthesizing existing research, meta-analysis enhances statistical power, increases the generalizability of findings, and helps uncover underlying factors influencing the relationship between FLB and language achievement (Jak, Reference Jak2015). This study intends to explore the magnitude and direction of the effect of FLB on learning achievement, as well as potential moderating factors through multilevel meta-analysis. By critically reviewing relevant research in this domain, this study provides an empirical foundation for understanding and enhancing emotional experiences throughout the foreign language learning process while offering practical suggestions for future research endeavors.
Literature review
Foreign language boredom
Boredom is a multifaceted psychological experience encompassing feelings of disinterest and lack of excitement toward the current activity or environment, indicating a state of mental disengagement or resistance (Mikulas & Vodanovich, Reference Mikulas and Vodanovich1993). In foreign language learning, boredom can be a commonly encountered subjective experience characterized by a lack of interest, engagement, or stimulation for FL learning tasks and activities (Pawlak et al., Reference Pawlak, Zawodniak and Kruk2020). This phenomenon, known as foreign language boredom (FLB), represents a negative emotional state that extends from conventional boredom and manifests as a diminished drive toward learning goals and a negative emotional experience while acquiring a foreign language (Pawlak et al., Reference Pawlak, Kruk, Zawodniak and Pasikowski2022). Positive psychology emphasizes enhancing positive emotions and reducing negative ones to improve the overall learning experience, which fosters a balanced emotional state essential for effective learning (Dewaele et al., Reference Dewaele, Botes and Greiff2023; MacIntyre & Mercer, Reference MacIntyre and Mercer2014). FLB is found to correlate with levels of enjoyment and anxiety experienced during FL study (Dewaele et al., Reference Dewaele, Botes and Greiff2023; Botes et al., Reference Botes, Dewaele, Greiff and Goetz2024). As evidenced by the research of Dewaele et al. (Reference Dewaele, Botes and Greiff2023), FLB is intricately linked to teacher-related factors, exerting a significant influence on learners’ FL attitudes and achievement due to its influence on motivation, engagement, and psychological states (Pawlak et al., Reference Pawlak, Kruk, Csizer and Zawodniak2023; Tsang & Dewaele, Reference Tsang and Dewaele2023). FLB led to decreased motivation, loss of attention, and lack of attention to foreign language learning, thus affecting the depth and breadth of learning and negatively predicting foreign language achievement (Li et al., Reference Li, Dewaele and Hu2023). Understanding and mitigating boredom is vital for improving learning experiences and outcomes, aligning with the goals of positive psychology to promote overall student well-being and academic achievement.
Measurements for assessing FLB consist of laboratory-based physiological indices and self-report measurements. The former employs electrothermal activity, heart rate variability, eye-tracking techniques, and neuroimaging techniques to capture boredom-related physiological responses during foreign language learning tasks (Betella et al., Reference Betella, Zucca, Cetnarski, Greco, Lanatà, Mazzei and Verschure2014; Gawda et al., Reference Gawda, Szepietowska, Soluch and Wolak2017). The artificiality of controlled conditions may restrict the ecological validity of the findings, and the reliance on specialized devices for data collection in the laboratory poses practical challenges, impeding the feasibility of conducting large-scale studies (Pavlenko, Reference Pavlenko2005). Self-report measurements, including questionnaires and surveys, have been extensively utilized to assess FLB (Zhao et al., Reference Zhao, Lan and Chen2023; Zhao & Wang, Reference Zhao and Wang2023). The Achievement Emotions Questionnaire (AEQ) was developed to evaluate various achievement emotions experienced by students in academic contexts (Pekrun et al., Reference Pekrun, Goetz, Frenzel, Barchfeld and Perry2011). The AEQ comprises 24 scales designed to measure nine emotions experienced during class, studying, and exams including enjoyment, hope, pride, relief, anger, anxiety, hopelessness, shame, and boredom. However, AEQ needs more contextualization and conceptualization tailored explicitly to foreign language learning (Li et al., Reference Li, Wei and Lu2023).
Currently, the development of FLB scales has been addressed by Pawlak et al. (Reference Pawlak, Kruk, Zawodniak and Pasikowski2020) and Li et al. (Reference Li, Dewaele and Hu2023). The former devised the Boredom in Practical English Language Classes Questionnaire (BPELCQ) based on data from 111 Polish English majors, while the latter developed the Foreign Language Learning Boredom Scale (FLLBS) using data from over 3,000 non-English major college students in China. The reliability and validity of the former questionnaire require further verification, whereas the total scale and subscales of the latter have attained desirable levels of reliability and validity. Subsequently, researchers have embarked on studies centered around FLB using these scales to investigate boredom and its relationship with antecedent and sequent variables (Apridayani & Waluyo, Reference Apridayani and Waluyo2022; Kruk et al., Reference Kruk, Pawlak, Shirvan and Shahnama2022). While some studies on antecedents focus on the influence of individual cognitive styles, teaching modes, and teacher’s enthusiasm on FLB (Apridayani & Waluyo, Reference Apridayani and Waluyo2022; Dewaele & Li, Reference Dewaele and Li2021), other empirical studies have explored the relationship between FLB and language achievement (Kruk, Reference Kruk2022; Wang et al., Reference Wang, Wang and Li2023). Among them, a significant portion has focused on the extent to which boredom can influence and predict foreign language achievement (Li & Wei, Reference Li and Wei2023; Wang et al., Reference Wang, Wang and Li2023; Zhao & Wang, Reference Zhao and Wang2023).
Foreign language achievement
Foreign language achievement encompasses an individual’s proficiency and success in acquiring and using a non-native language, involving an evaluation of the learner’s language knowledge (e.g., grammar, vocabulary, and pragmatics), abilities in comprehending, speaking, reading, and writing and skills in cross-cultural communication in the target language (Cizek, Reference Cizek1996; Durán, Reference Durán2008; Neufeld, Reference Neufeld1979). Many factors influence foreign language achievement, including learner’s motivation, learning emotions, and access to the target language (Madigan & Curran, Reference Madigan and Curran2021; Peng & Kievit, Reference Peng and Kievit2020). It is crucial to recognize that foreign language acquisition is a dynamic process wherein learners can consistently develop and enhance their language abilities through ongoing practice and exposure (Anderman & Patrick, Reference Anderman and Patrick2012).
Foreign language achievement measurements not only include objective scores such as course test scores and standardized language test scores but also reflect the language level and mastery of students in foreign language learning through students’ self-reported scores and self-perceived ability (Botes et al., Reference Botes, Dewaele and Greiff2020; Li, Reference Li2020; Zhao & Yang, Reference Zhao and Yang2023). Quantifiable and standardized measurements provide consistent evaluations of a learner’s language proficiency and concrete feedback on the mastery of language skills and content knowledge within a structured educational setting, offering a reliable means of comparison among students and across different contexts (Bachman & Adrian, Reference Bachman and Adrian2022; Spolsky, Reference Spolsky2000). In addition, learners’ perceived language achievement, which refers to the subjective perception or belief an individual holds regarding their level of proficiency and success in acquiring a foreign/second language, has been increasingly used to assess learners’ learning achievement (Dewaele et al., Reference Dewaele, Witney, Saito and Dewaele2018; de Saint Léger, Reference de Saint Léger2009). It is based on the learner’s assessment and subjective evaluation of their language learning progress, skills, and knowledge. Several researchers have emphasized that there may be a lack of perfect alignment between perceived language learning achievement and actual proficiency levels (Babaii et al., Reference Babaii, Taghaddomi and Pashmforoosh2016; Yoon & Lee, Reference Yoon and Lee2013). It becomes imperative for research studies to incorporate measurements of both perceived learning achievement and objective test scores to gain a comprehensive understanding of learners’ language abilities and progress.
Foreign language boredom and achievement
Prolonged experiences of boredom during language learning can generate negative attitudes that lead to disengagement in language learning, which may impact learners’ language performance and achievements (Wang & Li, Reference Li2022; Dewaele et al., Reference Dewaele, Botes and Greiff2023). A growing body of research has investigated the relationship between FLB and language learning achievement, consistently revealing a negative correlation (Li & Wei, Reference Li and Wei2023; Tsang & Dewaele, Reference Tsang and Dewaele2023; Zhao & Wang, Reference Zhao and Wang2023). This correlation extends beyond overall language performance, as FLB has been shown to negatively predict learners’ achievement in various contexts. FLB was found to negatively predict learners’ performance of English as a foreign language for specific purposes research based on aviation courses (N = 198) (Dinçer & Atay, Reference Dinçer and Atay2022). FLB negatively affected foreign language writing test scores and self-perceived writing ability (Jiang, Reference Jiang2023; Li et al., Reference Li, Dewaele and Hu2023). Such a negative correlation may be consistent across national and educational contexts (Li & Wei, Reference Li and Wei2023; Tsang & Dewaele, Reference Tsang and Dewaele2023).
However, some studies also point out that FLB may not directly predict foreign language achievement, and it is not the only factor that directly leads to low performance; thus, the interference of other factors should be considered (Li & Lu, Reference Li and Lu2022). Based on the study of 954 English learners at a rural middle school, Li and Wei (Reference Li and Wei2023) found that FLB can independently predict foreign language scores in three time points, but after entering the combined emotion model (enjoyment, anxiety, and boredom), boredom loses its adequate predictive power for academic scores. Researchers also found differences in the effect of FLB while using different measurements for learning achievement under different teaching modes (Apridayani & Waluyo, Reference Apridayani and Waluyo2022; Wang & Li, Reference Li2022). FLB could significantly predict learners’ self-perceived academic achievement in online classrooms for English as a foreign language, but it does not significantly predict learners’ actual reading and writing performance (Wang & Li, Reference Wang and Li2022; Huang & Zhang, Reference Huang and Zhang2024).
Previous studies have identified several issues that warrant attention. First, various FLB measurements (such as AEQ and FLLBS) have been used in previous studies. Second, different language achievement measurements were employed, including course grades, language tests, and self-rated scores. Third, although most studies support the negative relationship between FLB and learning achievement, the exact strength of this relationship still needs to be resolved. Finally, a meta-analysis that considers potential moderators is needed to provide insight into the relationship between boredom and achievement. Therefore, these problems may bring potential risks, leading to the need for more necessary systematic and comprehensive research on FLB and the inability to draw relatively accurate, generalizable conclusions.
Moderators
Meta-analysis provides a unique opportunity to delve into potential moderating factors, such as research characteristics, which could elucidate systematic variations in effect sizes observed across studies, offering insights into the intricate relationship between FLB and foreign language achievement. The moderation analysis within meta-analysis enhances the robustness of the analysis and facilitates a deeper understanding of the complexities underlying the impact of FLB on language achievement.
Given the multifaceted nature of FLB, different scales can provide standardized and systematic ways to measure boredom in foreign language learning, capturing its varying intensity and performance (Dewaele & MacIntyre, Reference Dewaele and MacIntyre2016). By distinguishing different levels of boredom, ranging from mild disinterest to profound disengagement, researchers can examine the relationship between boredom and learning achievement, including the potential threshold at which boredom becomes detrimental to foreign language learning performance (Li & Wei, Reference Li and Wei2023). Incorporating these scales enables researchers to better understand the relationship between boredom and learning achievement and identify moderating factors.
It is essential to acknowledge that the specific types of foreign language achievement measures employed in studies may moderate the relationship between boredom and learning achievement. Utilizing various measurement approaches allows for a more comprehensive evaluation of learning outcomes, as different measures may yield distinct outcomes and capture different dimensions of language achievement (Bachman & Adrian, Reference Bachman and Adrian2022; Spolsky, Reference Spolsky2000). By employing a range of measurements, researchers can delve deeper into the intricate dynamics between boredom and learning achievement, identifying specific areas of vulnerability to the detrimental effects of boredom and exploring the factors that influence this relationship (Bachman & Adrian, Reference Bachman and Adrian2022). This multifaceted approach to measurement contributes to a more nuanced understanding of the complexities underlying the interplay between boredom and foreign language learning outcomes.
Specific language skills may also moderate the relationship between boredom and achievement in foreign language learning (Jiang, Reference Jiang2023; Liao & Dong, Reference Liao and Dong2018). The impact of boredom may vary across different language skills based on their cognitive demands and nature (Li et al., Reference Li, Dewaele and Hu2023). Specific skills require continuous attention, concentration, and engagement, rendering them more susceptible to the negative effects of boredom (Jiang, Reference Jiang2023; Liao & Dong, Reference Liao and Dong2018). Analyzing the relationship between boredom and specific language skills allows researchers to identify the skills most prone to boredom and ascertain their influence on academic performance. Different language skills may elicit varying levels of learner participation and motivation, resulting in differential effects of boredom on learning achievement.
Furthermore, educational stages correspond to distinct developmental milestones and cognitive abilities (Gajda, Reference Gajda2016; Lee et al., Reference Lee, Xie and Lee2021), influencing the relationship between boredom and foreign language performance. Boredom may affect academic performance differently across different stages of education due to variations in attention span, engagement levels, and cognitive and socioemotional development (Lee et al., Reference Lee, Xie and Lee2021; MacIntyre, Reference MacIntyre1992). Younger learners, such as those in secondary school, may exhibit different attention spans and levels of engagement compared to older learners in higher education (Bell & Wolfe, Reference Bell and Wolfe2004). The effects of boredom on academic performance may vary depending on the educational stage due to these developmental disparities.
Additionally, different instructional modes, online and offline modes, diverge in terms of interactivity and engagement (Derakhshan et al., Reference Derakhshan, Kruk, Mehdizadeh and Pawlak2021), influencing the relationship between boredom and language learning performance (Derakhshan, Kruk et al., Reference Derakhshan, Kruk, Mehdizadeh and Pawlak2022). Online teaching typically incorporates digital platforms, multimedia resources, and interactive activities, while offline teaching relies on face-to-face interactions, physical materials, and classroom instruction (Kruk et al., Reference Kruk2022). The level of interaction and engagement in each mode affects learners’ interest, motivation, and susceptibility to boredom (Shimray & Wangdi, Reference Shimray and Wangdi2023). Online teaching modes, characterized by greater flexibility and learner control over the learning process, can enhance learner autonomy and motivation, reducing the likelihood of boredom (Shimray & Wangdi, Reference Shimray and Wangdi2023). Conversely, offline teaching with more structured and rigid schedules may lead to higher levels of boredom (Derakhshan et al., Reference Derakhshan, Fathi, Pawlak and Kruk2022). The degree of flexibility and control embedded within each teaching mode can moderate the relationship between boredom and learning performance.
Current study
To address the limitations of individual studies, as illustrated in the literature review, a meta-analysis is crucial for gaining deep insights into how boredom affects FL learning achievement. The present study aims to advance and integrate existing empirical research and previous reviews by addressing three primary research objectives. First, a comprehensive and systematic literature review will examine the relationship between boredom and learning achievement in foreign language learning. Second, the direction (positive or negative) and the effect size of FLB on foreign language achievement should be examined as documented in the existing literature. Last, to broaden the scope of the meta-analysis by investigating potential moderators of the relationship between FLB and learning achievement, including contextual factors and language-related variables.
Based on these research objectives, the study seeks to address the following two research questions (RQs):
RQ1: How does FLB impact foreign language achievement? What is the direction and magnitude of this impact?
RQ2: How do different language achievement measurements, FLB measurements, educational stages, and teaching modes influence the relationship between FLB and language learning achievement?
Methodology
This study utilized a multilevel meta-analysis, an analytical approach designed to account for dependencies between effect sizes and effectively model the hierarchical structure of the data (López-López et al., Reference López-López, Page, Lipsey and Higgins2018). By capturing the complexity inherent in nested data, the multilevel meta-analysis enhances the accuracy of estimates and mitigates issues related to data dependencies, allowing for a clearer interpretation of results, especially in fields where multiple effect sizes may be derived from a single study, thereby reflecting the true dynamics of the research context. (Assink, & Wibbelink Reference Assink and Wibbelink2016; Cheung, Reference Cheung2019; Norouzian & Bui, Reference Norouzian and Bui2024).
Literature retrieval
Multiple methods have been employed in this study to retrieve relevant literature on FLB (Vuogan & Li, Reference Vuogan and Li2023). Both published and unpublished studies were considered eligible for inclusion, minimizing the potential impact of publication bias on the statistical outcomes (Vuogan & Li, Reference Vuogan and Li2023).
Several online databases were initially searched, including Web of Science, Google Scholar, Education Resources Information Clearinghouse (ERIC), China National Knowledge Infrastructure (CNKI), and Linguistics and Language Behavior Abstracts (LLBA). The search utilized various keywords, including boredom, language, foreign language, second language, classroom, emotion, language competence, language ability, language proficiency, and foreign language learning achievement. Building on references from previous research, manual searches were conducted. Selection criteria were based on frequently cited journals in previous studies, ensuring consistent and comparable literature reviews (Teimouri et al., Reference Teimouri, Goetze and Plonsky2019; Vuogan & Li, Reference Vuogan and Li2023). These influential and authoritative journals have published many groundbreaking studies in SLA and applied linguistics and are widely used in different reviews. These journals include Studies in Second Language Acquisition, Applied Linguistics, Journal of Applied Linguistics, Foreign Language Annals, Language Learning, Canadian Modern Language Review, International Review of Applied Linguistics, Language Teaching Research, Language Testing, the Modern Language Journal, Second Language Research, System, and TESOL Quarterly.
Furthermore, the backward and forward citation searches were conducted based on influential and highly cited literature. As a result, 529 relevant studies on FLB were retrieved.
Inclusion/exclusion criteria
This study applied carefully defined inclusion and exclusion criteria to determine the final research sample. The following eligibility criteria were considered.
First, the included studies must be empirical, with a focus on quantitative research. Studies employing mixed methods (both quantitative and qualitative) are also eligible for inclusion. However, systematic reviews, scoping reviews, and purely qualitative studies are not considered for inclusion. A meta-analysis that provides effect sizes can be included. Second, the selected studies must include at least one measure of FLB and one measure of foreign language achievement.
Third, the selected studies should report a relationship between FLB and foreign language achievement. This relationship should be expressed as a correlation coefficient or other statistics that can be translated into a correlation (e.g., t or F). Studies that analyze correlation matrices using multiple regression or structural equation model analysis can also be included.
In addition, the current study would include all forms of study design if a correlation coefficient can be retrieved from it. In the current meta-analysis, pretest data would be used when experimental studies with pretest and posttest data are included to avoid confounding effects and enhance the accuracy and reliability of effect estimates. This approach ensures comparability of participants’ baseline characteristics, controlling for initial differences and minimizing external variables. The uniform collection of pretest data enhances consistency and reduces the impact of heterogeneity factors—such as variations in intervention implementation, assessment tools, and time spans—on the meta-analysis results (Vuogan & Li, Reference Vuogan and Li2023). In cases where the study examines differences between groups, different groups will be entered into the database and labeled separately.
Last, several potential moderating variables, such as teaching modes and educational stage, were identified. However, the presence or absence of a moderating variable did not exclude any studies. Figure 1 illustrates the workflow for document inclusion and exclusion in this study.
Database search flowchart.
The Newcastle-Ottawa Scale (NOS) was adapted to check methodological transparency and quality of individual studies included in the meta-analysis. By assessing selection, comparability, and outcome, the NOS standardizes study ratings, which facilitates the inclusion of high-quality research and improves the robustness of meta-analysis conclusions (Stang, Reference Stang2010). The scoring index of methodological transparency is presented in Table 1 (See the elaboration of the index in Appendix). Each item is scored on a scale of 0 and 1, with selection scoring up to 3 points, comparability up to 2 points, and outcome up to 4 points, totaling a maximum score of 9 points. Studies scoring over 5 are suitable for meta-analyses and over 7 are considered high quality (Stang, Reference Stang2010). After scoring, the studies included in the study are suitable for meta-analysis (score above 5), with the mean score equal to 7.06 and SD equal to 1.51.
Scoring index of methodological transparency
Coding procedure and coding system
To comprehensively capture various aspects of each study in the sample, this study integrated a comprehensive literature coding system tailored to the research purpose, research questions, and previous studies in the field (Vuogan, Reference Vuogan and Li2023). It consists of five parts, including 39 coding indices: background information, research methods, participants/sample information, instrument of boredom, and instrument of achievement.
The coding scheme was adaptively revised to characterize the diverse research features in this area, as shown in Table 2. This study also explores some of these features as potential moderators of the overall correlation effect size. Two coders, both doctoral students in applied linguistics, participated in the coding process. The intercoder agreement reached 98%, indicating a high level of agreement and ensuring the reliability of subsequent analyses (Marsden et al., Reference Marsden, Mackey, Plonsky, Mackey and Marsden2015).
Coding system
Data analysis
The data analysis process adheres to the steps outlined in Figure 2 and is conducted in accordance with established guidelines and recommendations for multilevel meta-analysis (Assink & Wibbelink, Reference Assink and Wibbelink2016; Cheung, Reference Cheung2019; Cui et al. Reference Cui, Ding, Yu, Zhang and Li2024; Norouzian & Bui, Reference Norouzian and Bui2024), comprising the following six steps, effect size calculation, overall effect size estimation, outlier detection and influence analysis, heterogeneity analysis, moderator analysis, and publication bias assessment. All statistical analyses were conducted using the “metafor” package (Viechtbauer, Reference Viechtbauer2010) in R (Version 4.1.2; R Core Team, 2021).
Flow chart of data analysis.
Effect size calculation
The effect size was calculated using Fisher’s z transform of the correlation coefficients. Fisher’s z was chosen because it has the properties of stable variance and producing a normal distribution indicator, which facilitates more reliable meta-analysis inferences (Jak, Reference Jak2015). The conversion to Fisher’s z helps to handle studies with different sample sizes and different effect sizes, while ensuring that the variance of the effect size is properly estimated (Jak, Reference Jak2015). The variance of Fisher’s z transform effect size is calculated according to the respective formulas for consistency and accuracy. The calculation of the effect size was initially performed in Excel, and the results were then converted back to the correlation coefficient (r) for easy interpretation.
Overall effect size estimation
A multilevel random model implemented using the “metafor” package in R (Viechtbauer, Reference Viechtbauer2010) was employed to estimate the overall effect size, considering the variance at three levels: within-study variance, between-studies variance, and overall variance. This modeling strategy allows the variance to be partitioned into different levels, thereby addressing the dependency of effect sizes within studies. By incorporating random effects at both the study and effect size levels, the model accurately reflects the nested structure of the data, providing a more robust estimate of the overall effect size (Assink, & Wibbelink Reference Assink and Wibbelink2016; Cheung, Reference Cheung2019).
Outlier and influence analysis
Outlier and influence analyses involving calculating Cook’s distance, DFBETAS, and leverage (Hat) values for each study were conducted to identify studies or effect sizes that exerted disproportionate influence on the overall results (Viechtbauer & Cheung, Reference Viechtbauer and Cheung2010). The influence of each effect size was evaluated within the multilevel model, allowing for a more nuanced understanding of how individual studies contribute to the overall model. Outliers and influential studies were further investigated using influence index plots and leave-one-out analyses to assess their impact on the overall effect size estimate.
Heterogeneity analysis
Heterogeneity was quantified using the total Q-statistic and I2 indices, which describe the proportion of variance attributable to different levels. A one-sided log-likelihood ratio test was performed to examine the significance of variance components at each level. The results from these tests provide evidence on whether substantial heterogeneity exists within or between studies, validating the necessity of the multilevel model (Huedo-Medina et al., Reference Huedo-Medina, Sánchez-Meca, Marín-Martínez and Botella2006). Low heterogeneity suggests that the model has successfully accounted for the variability in effect sizes.
Moderator analysis
To explore potential sources of heterogeneity, moderator analysis was performed using a multilevel mixed-effects model, allowing for the inclusion of moderator variables to explain variability in effect sizes across and within studies (Norouzian & Bui, Reference Norouzian and Bui2024). The mixed-effects model considers both fixed effects (moderator variables) and random effects (study and effect size levels), providing a comprehensive understanding of how moderators influence the overall effect (Cui et al., Reference Cui, Ding, Yu, Zhang and Li2024). The inclusion of moderators enables the examination of contextual or methodological factors that may impact the observed effect sizes.
Publication bias assessment
Studies reporting relatively high effect sizes were more likely to be published than those reporting low effects, and published papers were more likely to be included in the meta-analysis, so the combined effect sizes might be overestimated, known as publication bias (Borenstein et al., Reference Borenstein, Cooper, Hedges and Valentine2009). This study adopted the Funnel Plot, Egger’s regression test, and the trim-and-fill method to assess publication bias. The Funnel Plot graphically shows the presence or absence of publication bias and a simple scatter plot for each study and effect size. When the funnel plot appears roughly symmetrical, the study may not have serious publication bias (Sterne & Egger, Reference Sterne and Egger2001). Egger’s regression test was employed to statistically test for funnel plot asymmetry at both the two-level and multilevel model structures (Egger et al., Reference Egger, Smith, Schneider and Minder1997). Additionally, the trim-and-fill method was applied to estimate the number of missing studies due to potential publication bias (Duval & Tweedie, Reference Duval and Tweedie2000).
Results
Results of the methodological transparency analysis
After conducting a methodological transparency analysis using the NOS scale and considering the background of independent studies, all included studies achieved an acceptable level, with total scores above five. However, some issues were identified, examining the three dimensions of methodological transparency. In the selection dimension, while most studies had good sample representativeness, 42.4% (K = 14) did not report response rates and questionnaire return rates. In the comparability dimension, only 24.2% (K = 8) of the independent studies controlled for confounding variables, whereas the majority did not. Only 51.2% (K = 17) of the studies in the outcome assessment dimension reported handling missing and omitted values. In addition, this study identified a lack of transparent reporting on language proficiency across independent studies. Specifically, 54.5% of the studies (K = 18) did not report participants’ language proficiency, while two included participants with mixed proficiency levels. Among the 13 studies that did provide proficiency information, there was significant variability and a lack of clarity with classifications ranging from beginner to intermediate, intermediate to advanced, and others that were often based on differing or unspecified criteria.
Results of the overall effect and outlier analysis
This study included 33 independent studies with a total of 47 effect sizes. The effect sizes and confidence intervals for each sample are displayed in Figure 3.
Effect sizes and confidence intervals for each sample.
The multilevel random model showed that the overall effect size was significant, with an effect size of r = -.240, 95% confidence interval (CI) = [-.303, -.187], df = 46, t = -8.504, p < .001. The two-level random model illustrated that the overall effect size was significant, with r = -.250, 95% CI = [-.310, -.202], df = 46, t = -9.507, p < .001. The Heterogeneity analysis indicated significant heterogeneity, Q (46) = 553757.301, p < .001. To assess the influence of individual studies on the outcomes, an influence analysis was conducted. A study is considered influential if its removal significantly alters the model’s results (Viechtbauer & Cheung, Reference Viechtbauer and Cheung2010). The outlier detection process revealed no significant outlier distribution, as illustrated in Figure 4.
Results of outlier detection.
Results of heterogeneity analysis
The heterogeneity test for the multilevel model did not yield significant results with Q (46) = 768.013, p < .001). Then, a comparison of model fit between the two-level and multilevel models was conducted, as presented in Table 3. When the within-study variance was constrained to zero, the model fit significantly worsened, as indicated by the log-likelihood ratio test, indicating heterogeneity at the within-study level. In contrast, fixing the between-study variance at zero did not result in a significant decline in model fit, indicating that between-study variance has a minimal impact on the observed heterogeneity.
Model comparison.
Results of moderator analysis
To address RQ2, this study conducted a multigroup analysis on seven moderate factors: educational stage, language achievement measurements, domain-specific skills, foreign language boredom measurements, teaching modes, learning contexts, and methodology transparency level.
First, the included literature was divided into subgroups based on educational stages: elementary school, middle school, high school, university, and a mixed group (referring to studies that did not specify the educational stage of the language learners or included learners from different educational stages). Second, to investigate the differential impact of FLB on foreign language achievement measurements, the included literature was categorized into four groups: course score, self-perceived achievement, self-reported score, and standardized test score (e.g., International English Language Testing System [IELTS]/Test of English as a Foreign Language [TOEFL], College English Test [CET]4/6, and Hanyu Shuiping Kashi [HSK]). One study that did not report the assessment form was labeled as NA. Third, domain-specific skills were categorized into five groups: overall language proficiency, reading, writing, speaking, and reading and writing. Studies that combined reading and writing scores without distinguishing between the two were classified into the reading and writing skills group.
Additionally, two studies did not provide specific descriptions of language proficiency measurement and were marked as NA. Fourth, this study included four scales—AEQ, BPELC, FLLBS, and PSBEC—to explore the effects of FLB measurements on language achievement. Fifth, as the impact of FLB on foreign language achievement may vary under different instructional modes (in-person and online), this study conducted a moderation analysis of instructional modes. Finally, the moderator analysis of the FL/SL learning background and methodology transparency level was conducted, respectively. Table 4 illustrates the combined effect sizes, confidence intervals, and p-values for each subgroup.
The moderator analysis.
Note: **= p < .05; *** = p < .001
Results of publication bias
Funnel plot, Egger’s regression test, and trim-and-fill were employed to assess publication bias. The funnel plot is a graphical tool used to detect publication bias in research. It is depicted as a scatter plot where the horizontal axis represents the effect size of each study, and the vertical axis denotes the standard error. Each circle on the plot corresponds to specific data. The funnel plot was asymmetrical (Figure 5), suggesting that publication bias is not a significant concern (Sterne & Egger, Reference Sterne and Egger2001).
Funnel plot of effect size.
The two-level Egger’s regression test showed a low risk of publication bias, z = .057, p = .955. The multilevel Egger test with the modified predictor was nonsignificant, z = 1.206, p = .996. The trim-and-fill showed that 0 effect sizes were missing to the right side of the mean effect size (SE = 1.414). Based on the above results, this meta-analysis appears to have a low risk of publication bias.
Discussion
Methodological transparency and overall effect size
The results of methodological transparency analysis highlight several weaknesses in the research field. First, the omission of reporting response rate can lead to nonresponse bias, where the opinions of non-respondents may significantly differ from those of respondents, thereby skewing the results; second, the absence of controlling confounding variables may raise the research risk, as extraneous variables may influence the results, leading to erroneous conclusions about the relationships under investigation; and third, inadequately addressing missing data can result in biased estimates and reduced statistical power, compromising the validity of the findings in this field (DeKeyser, Reference DeKeyser2007; Little & Rubin, Reference Little and Rubin2019). These deficiencies urge rigorous methodological practices with improved reporting standards, enhanced control of confounding variables, and robust handling of missing data to advance the quality of emotion research in SLA. Furthermore, methodological, and interpretive challenges were aroused due to the lack of clear reporting on language proficiency across independent studies, which impedes the ability to draw meaningful conclusions, as proficiency is a critical covariate in language learning research (DeKeyser, Reference DeKeyser2007). Additionally, using differing and ambiguous criteria for language proficiency may create inconsistencies and complicate comparisons across studies (DeKeyser, Reference DeKeyser2007). The transparency issues identified in this study are consistent with recent research in applied linguistics, which emphasizes the importance of transparent practice as part of a broader quality framework (Isbell et al., Reference Isbell, Brown, Chen, Derrick, Ghanem, Arvizu and Plonsky2022; Plonsky et al., Reference Plonsky, Marsden, Crowther, Gass and Spinner2020). These issues are not isolated; rather, many of the challenges observed align with the broader research agenda in SLA (De Costa, Reference De Costa2016). As proposed by Plonsky (Reference Plonsky2024), transparency is a foundational element within the study quality framework, which is essential for achieving a more robust evaluation of research practices and ensuring that findings can be synthesized more reliably (Isbell et al., Reference Isbell, Brown, Chen, Derrick, Ghanem, Arvizu and Plonsky2022). These results yearn for transparent measurements to enhance the comparability of studies, facilitate synthesizing research, and improve the overall quality and reliability of research in language learning.
The results revealed the negative effect of FLB on foreign language achievement (r = -.250, p < .001). According to Plonsky and Oswald (Reference Plonsky and Oswald2014), an effect size close to .25 is considered small, around .40 is moderate, and above .60 is large in SLA. Therefore, the results indicate a small effect of FLB on language achievement, consistent with previous studies (Li et al., Reference Li, Dewaele and Hu2023; Li & Lu, Reference Li and Lu2022; Tze et al., Reference Tze, Daniels and Klassen2016). Boredom can hinder the consolidation of new knowledge and skills, hinder language practice, and negatively affect long-term memory (Pawlak et al., Reference Pawlak, Kruk, Zawodniak and Pasikowski2020). When learners are bored, they may be less likely to actively engage in learning activities, retain information, or effectively develop language skills (Li, Reference Li2022). As a result, foreign language performance may suffer due to the reduced effort, motivation, and overall learning engagement caused by boredom (Kruk & Zawodniak, Reference Kruk, Zawodniak and Szymański2018).
Moderation effects
The subgroup analysis of different educational stages revealed inconsistent impact effect size of FLB on foreign language achievement. The effect size of different educational stages approximately follows the order of middle school group > the elementary school group > the high school group > university group, indicating that the influence of FLB on foreign language achievement tends to weaken as the educational stage advances, which may be attributed to learners’ cognitive maturity and changes in language learning content. As the educational stage rises, learners’ cognitive abilities, including attention, autonomous learning skills, memory, and critical thinking abilities, further develop, which enables learners to handle negative feedback more effectively during the learning process promptly and engage in positive emotion regulation (Bell & Wolfe, Reference Bell and Wolfe2004), reducing the adverse impact of negative emotions on foreign language learning achievement (Ozfidan & Burlbaw, Reference Ozfidan and Burlbaw2019). Additionally, as the educational stage progresses, the content of foreign language learning changes. In the initial stages of language learning, the emphasis is often on basic vocabulary and grammar, which can be repetitive and unstimulating (Langacker et al., Reference Langacker, von Stutterheim, Carroll, Lambert, Behrens, Strauss, Pavlenko, Paribakht, Wesche and Rinner2006). As learners progress to higher stages and their proficiency in the foreign language improves, they begin to encounter more diverse and interesting language content, which may evoke enjoyment and alleviate the negative impact of FLB (Pawlak et al., Reference Pawlak, Kruk, Csizer and Zawodniak2023; Zhao & Wang, Reference Zhao and Wang2023).
However, the middle school group exhibited the highest effect size, indicating that the achievement of middle school learners is most affected by FLB, which may be related to the unique characteristics of the middle school stage. Middle school learners are transitioning from concrete operational thinking to abstract thinking (Gajda, Reference Gajda2016), which may challenge them to grasp and persevere in learning abstract and complex language aspects. Middle school often marks a shift from a more nurturing and playful learning environment in elementary school to a more structured and academically demanding environment, which may make some students feel less adapted and potentially affect their engagement and interest in foreign language learning (Madjar & Chohat, Reference Madjar and Chohat2017; Papi & Hiver, Reference Papi and Hiver2020). Moreover, it was found that the effect of boredom on language achievement was slightly greater in FL contexts compared to SL contexts, which may arise from varying levels of exposure and immersion. In FL contexts, learners may experience more boredom due to structured classroom settings and limited real-life application, whereas in SL contexts, learners can benefit from practical use (Pawlak et al., Reference Pawlak, Kruk, Zawodniak and Pasikowski2020). Additionally, learners may face more pressure and less intrinsic motivation in FL than in the SL context due to cultural and educational differences (Chen et al., Reference Chen, Sun and Yang2022).
The results of FLB’s effect on different language achievement measurements illustrated significant between-group differences with inconsistent effect size. This finding is consistent with the meta-analysis results by Teimouri et al. (Reference Teimouri, Goetze and Plonsky2019) on foreign language anxiety, which implies that the use of different foreign language achievement measurements (subjective measurement and objective measurement) requires more research attention (Li & Zhang, Reference Li and Zhang2021). Both self-perceived foreign language achievement and self-reported course grades involve subjective measurement issues, especially self-perceived foreign achievement, which is essentially subjective and influenced by learners’ emotions, attitudes, and self-concepts (Babaii et al., Reference Babaii, Taghaddomi and Pashmforoosh2016; Ding & Zhao, Reference Ding and Zhao2020). On the contrary, course grades and standardized test scores are objective criteria for measuring learners’ foreign language performance, aiming to evaluate specific knowledge and skills (Durán, Reference Durán2008). The results of the two categories of foreign language achievement measurements suggest that future researchers must be cautious about relying solely on self-reported academic achievement indicators, as it may introduce potential research risks (Botes et al., Reference Botes, Dewaele and Greiff2020; Cizek, Reference Cizek1996). To reduce the risks associated with self-report data, future studies could consider using triangulation methods that combine self-report data with objective measurements (such as exam scores or grades) and other sources (such as teacher evaluations or observational data), which provides a more comprehensive and accurate reflection of foreign language learners’ performance and learning experiences.
The results show that foreign language boredom has the highest impact on writing, followed by writing and reading, overall foreign language proficiency, speaking, and reading (p>.05) that did not reach statistical significance. These findings partially support previous studies (Teimouri et al., Reference Teimouri, Goetze and Plonsky2019), indicating that writing skills can be significantly influenced by negative emotions (boredom and anxiety), followed by speaking and reading. This pattern may be attributed to the distinctive characteristics associated with different foreign language skills. Learners may experience negative emotions when facing the challenge of production in a foreign language, which requires a deep understanding of grammar, vocabulary, and syntax, as well as the ability to organize thoughts and ideas coherently (Garner, Reference Garner2016). When individuals feel bored, their cognitive resources could be more optimally utilized, making it challenging to generate original ideas, organize thoughts coherently, and employ appropriate language structures (Jiang, Reference Jiang2023).
Furthermore, the results illustrate that the effect size of FLB on learning achievement varies when different boredom measurements are used. When using the achievement emotions questionnaire (AEQ) and foreign language learning boredom scale (FLLBS), significant effects of cross-independent studies of FLB on learning achievement were observed, suggesting that these two scales demonstrate high measurement stability and reliability. However, AEQ is not a specialized scale focusing on foreign language learning emotions despite the good reliability, which may lead to a lack of context-specific content and requires more precise measurement dimensions of foreign language emotions (Botes et al., Reference Botes, Dewaele and Greiff2021). Therefore, more attention is needed to avoid potential risks when using AEQ in a foreign language context. The cross-language background stability of the FLLBS has been confirmed, which has been supported by studies on foreign language learning with different native language backgrounds (Derakhshan et al., Reference Derakhshan, Fathi, Pawlak and Kruk2022; Dewaele et al., Reference Dewaele, Albakistani and Ahmed2024). Boredom in the personal English language classroom scale (BPELC) and the Perceived Second/Foreign Language Boredom Emotion Checklist (PSBEC) did not reach statistical significance. BPELC lacks rigorous reliability and validity tests and had a small sample size during initial development, resulting in decreased reliability (Li et al., Reference Li, Dewaele and Hu2023). The PSBEC is based on a precursor scale measuring boredom in the context of foreign language learning in Iran (Rad et al., Reference Rad, Roohani and Mirzaei2022), which lacks cross-language and cross-cultural reliability and validity. Given the confirmed stability of the FLLBS across individual studies, the current study suggests that future investigations consider the FLLBS as a reliable instrument for assessing boredom in foreign language learning contexts as the reliability of the FLLBS has been validated across diverse native language backgrounds, rendering it a more accurate and context-specific measure.
Moreover, this study found that the impact of FLB on learning achievement in the online environment is slightly greater than that in the offline environment, suggesting that there are differences in learners’ emotional experiences in different classroom environments, and online classroom environments may have a more substantial influence on shaping learners’ foreign language emotions (Zhang & Huang, Reference Zhang and Huang2023). Online learning often lacks the same level of social interaction and real-time feedback as offline learning, which can exacerbate feelings of boredom and disengagement, hindering the improvement of language performance (Derakhshan et al., Reference Derakhshan, Fathi, Pawlak and Kruk2022). Without the variety of face-to-face interactions and other offline activities, learners may quickly lose interest and become bored with language learning, resulting in lower achievement in online learning than in offline study (Chen et al., Reference Chen, Sun and Yang2022; Jongsma et al., Reference Jongsma, Scholten, van Muijlwijk-Koezen and Meeter2023).
Conclusion
This study employed the meta-analysis to examine the relationship between FLB and foreign language achievement, drawing on a comprehensive analysis of 33 empirical studies (N = 27,838) with 47 effect sizes and potential moderating factors. The findings indicate that FLB has an overall small negative effect on foreign language achievement. By integrating the findings of previous studies on FLB, this study presents conclusive evidence supporting the significance of boredom as a substantial negative predictor of foreign language achievement. The study also reveals that the negative effect sizes of FLB on achievement vary depending on educational stages, FL/SL learning contexts, domain-specific skills, measurements of boredom and achievement, and teaching modes.
Implications for future studies
This study underscores the need for future research to report on methodological practices with enhanced reporting standards, improved control of confounding variables, and effective handling of missing data to advance the quality and reliability of research in SLA. Moreover, the study emphasizes the importance of considering the reliability of FLB measurements and the distinct characteristics of both subjective and objective foreign language achievement measurements to avoid potential research risks. To mitigate the risks inherent in self-report data, future research is recommended to consider employing triangulation methods that integrate self-report data with objective measures (such as exam scores or grades) and additional sources (such as teacher evaluations or observational data). Considering the demonstrated stability of the FLLBS across multiple individual studies, the present research recommends that future investigations adopt the FLLBS, an effective instrument for accurately measuring boredom in diverse educational settings, as a reliable tool for assessing FLB. In addition, it is suggested that future researchers explore measurement corrections for these relationships and employ nonlinear statistical measurements to uncover potential nonlinear associations between the two constructs. This study also identified substantive gaps in the research field of foreign language boredom, particularly regarding boredom induced by specific foreign language skills. Previous studies primarily focus on writing, while the relationship between boredom and other foreign language skills remains insufficiently explored. Notably, research investigating the relationship between FLB and foreign language listening needs to be included, underscoring the need for further investigation.
In addition, this study highlights the theoretical advantage of multilevel meta-analysis in adequately representing the underlying processes by which the data were generated (Gucciardi et al., Reference Gucciardi, Lines and Ntoumanis2022). Multilevel meta-analysis facilitates the accurate capture of variance across different levels and mitigates issues arising from nested data, providing a more authentic representation of the underlying research processes (Norouzian & Bui, Reference Norouzian and Bui2024). Therefore, this study advocates for the adoption of multilevel meta-analysis in future research within the field of SLA, where scholars increasingly employ multiple measurements for exploring emotions and their relationships with various variables (performance or different psychological and cognitive variables.
Pedagogical implications
The meta-analysis indicates that FLB negatively predicts FL achievement, highlighting the need to address boredom in language education. It is suggested that Educators use engaging and diverse materials in various task types (e.g., discussions, group work, games) to cater to students’ interests and reduce boredom. Providing timely and constructive feedback can motivate students by helping them see their progress, reducing boredom. Multiple assessment methods (e.g., formative assessment, peer assessment, self-assessment) can make the evaluation process more engaging. By integrating these strategies, educators can effectively mitigate the negative impact of boredom on foreign language achievement, enhancing learners’ engagement and learning outcomes.
Limitations
However, this study relies on synthesizing existing literature, which introduces the possibility of incomplete literature searches and potential biases. The findings may be influenced by methodological variations, sample characteristics, and potential publication bias. This study did not include studies published in other languages, such as Polish, Turkish, Arabic, and Persian, which are indeed significant in boredom research, and their exclusion may limit the generalizability of our findings. Therefore, this study strongly recommends that future research endeavors strive to incorporate studies published in diverse languages to provide a more holistic and nuanced understanding of FL boredom. Furthermore, more potential moderators that may influence the relationship between boredom and achievement in foreign language learning need to be explored. Factors such as language acquisition order (L2, LX), the presence of multiple foreign languages, and changes in the language learning environment may all impact the strength of this relationship.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/S0272263125000142.
Open access
