21 Novembre 2026

Psicologia Punto One | Psicologia e Psicologi nel Web

Menu

Menu

SCHEDA SCIENTIFICA2026Altra ricerca

Perceived social support, academic buoyancy, and L2 willingness to communicate among Chinese EFL undergraduates: a structural equation modeling study.

Zhu X, Mohd-Said NE, Shamsudin CBM

Frontiers in psychology · 2026

Abstract originale

INTRODUCTION: Willingness to communicate (WTC) stands as a critical factor influencing learners' practical use of a second language (L2). To date, few studies have systematically analyzed the psychological pathways between contextual social support and L2 WTC, with insufficient attention paid to learners' adaptive capabilities. This study explored how perceived social support from teachers, parents, and peers correlates with learners' L2 communicative willingness and examined the mediating function of academic buoyancy in this relationship. METHODS: Using a sample of 369 first-year non-English-major undergraduates in eastern China, structural equation modeling (SEM) alongside 5,000 bootstrap resamples was conducted to analyze the structural relationships among social support sources, academic buoyancy, and L2 WTC. RESULTS: Results indicated that all three forms of social support were positively associated with both academic buoyancy and WTC. Among these, teacher support showed the strongest standardized association with WTC, followed by peer and parent support. Academic buoyancy was found to partially mediate the associations between each source of social support and WTC, with teacher support showing the largest indirect effect (15.32%). DISCUSSION: These findings contribute to WTC literature by positioning academic buoyancy as an important adaptive psychological factor that links contextual support to communicative readiness. The results further highlight the differentiated functions of teacher, peer, and parent support within Chinese EFL contexts. Corresponding pedagogical implications are proposed to establish supportive, resilience-oriented language learning environments.

Parole chiave

EFL undergraduatesacademic buoyancyperceived social supportstructural equation modelingwillingness to communicateStructural Equation ModelingPerceived Social SupportAcademic BuoyancyWillingness To CommunicateEfl Undergraduates
Leggi il testo integrale originale

Testo estratto dalla pubblicazione: sono esclusi immagini, media, supplementi e parti soggette a diritti distinti.

Perceived social support, academic buoyancy, and L2 willingness to communicate among Chinese EFL undergraduates: a structural equation modeling study

Introduction

Willingness to communicate (WTC) stands as a critical factor influencing learners’ practical use of a second language (L2). To date, few studies have systematically analyzed the psychological pathways between contextual social support and L2 WTC, with insufficient attention paid to learners’ adaptive capabilities. This study explored how perceived social support from teachers, parents, and peers correlates with learners’ L2 communicative willingness and examined the mediating function of academic buoyancy in this relationship.

Methods

Using a sample of 369 first-year non-English-major undergraduates in eastern China, structural equation modeling (SEM) alongside 5,000 bootstrap resamples was conducted to analyze the structural relationships among social support sources, academic buoyancy, and L2 WTC.

Results

Results indicated that all three forms of social support were positively associated with both academic buoyancy and WTC. Among these, teacher support showed the strongest standardized association with WTC, followed by peer and parent support. Academic buoyancy was found to partially mediate the associations between each source of social support and WTC, with teacher support showing the largest indirect effect (15.32%).

Discussion

These findings contribute to WTC literature by positioning academic buoyancy as an important adaptive psychological factor that links contextual support to communicative readiness. The results further highlight the differentiated functions of teacher, peer, and parent support within Chinese EFL contexts. Corresponding pedagogical implications are proposed to establish supportive, resilience-oriented language learning environments.

1Introduction

Willingness to communicate (WTC) has been widely regarded as a pivotal construct for understanding why some language learners actively participate in second language (L2) interaction while others remain reluctant, even with comparable linguistic competence (MacIntyre and Legatto, 2011; Peng, 2025). Originating as a relatively stable dispositional trait in first language (L1) communication research (Bensalem et al., 2025; MacIntyre et al., 2019a; Yang and Lian, 2023), WTC was later reconceptualized for L2 acquisition as a situation-responsive readiness to initiate communication with specific interlocutors at a given moment (MacIntyre et al., 1998). As the immediate precursor to actual L2 use, WTC serves as a psychological bridge that enables learners to transform their linguistic knowledge into communicative action (Lee, 2022).

Building on this conceptualization, recent research has increasingly shifted from viewing WTC as a fixed dispositional trait to viewing it as a non-static concept susceptible to both real-time contextual variables and social-cultural interactions (Dewaele, 2019; Shen et al., 2024; Zhang et al., 2024). In EFL classrooms, learners’ WTC emerges from the interplay between personal dispositions and external contextual factors. Relevant studies have shown that perceived communicative competence and foreign language anxiety, two key internal variables, exert notable impacts on learners’ WTC (Lin et al., 2025; MacIntyre et al., 2002; MacIntyre and Gregersen, 2022), alongside various external situational factors (Han and Li, 2025; Wang et al., 2021). Scholars have gradually come to recognize that contextual factors play a tangible role in facilitating or limiting L2 interaction among learners.

Perceived social support is one critical contextual factor that strongly correlates with learners’ emotional states, classroom participation and actual communication performance (Collie et al., 2017; MacIntyre et al., 2001; Wang et al., 2023). Specifically, the variable reflects how students subjectively evaluate the affective, cognitive, and instrumental assistance obtained from instructors, classmates, and family members (Zimet et al., 1988). When learners feel adequately supported, they tend to experience less language learning anxiety, participate more actively, and become more willing to engage in interactive communication (Liu and Hong, 2021; Zarrinabadi and Khodarahmi, 2024). Prior empirical work further indicates that sound interpersonal connections within classrooms can boost learners’ WTC, especially during meaning-focused activities (Li Z. et al., 2025).

Accumulated research evidence has validated the connection between perceived social support and L2 learners’ WTC. Even so, few investigations have clarified the psychological pathways that explain how social support helps learners keep a stable state of communicative willingness. Most existing literature prioritizes affective variables, such as enjoyment and anxiety as mediating variables (Acosta-Manzano et al., 2024; Dewaele et al., 2023; Wang and Hui, 2024; Xiao and Jia, 2025). Recent work has further broadened this line of inquiry by examining how regret about past experiences relates to EFL learners’ motivation and WTC through positive and negative emotions (Almubarak and Alikhani, 2026). Such findings reinforce the relevance of examining psychological mechanisms alongside contextual factors when explaining variation in L2 WTC. Nevertheless, comparatively less attention has been paid to learners’ ability to cope with routine academic difficulties and frustrations (Souzandehfar and Ahmed Abdel-Al Ibrahim, 2023; Yun et al., 2018; Zhai, 2025). In this regard, academic buoyancy offers a viable theoretical framework to address this gap.

Academic buoyancy characterizes learners’ capacity to contend with recurring academic stress (Martin and Marsh, 2008), which differs from resilience, a trait associated primarily with responses to severe adversities. Individuals with higher academic buoyancy can better sustain learning motivation, regulate negative emotions, and keep active participation in classroom activities (Xu and Wang, 2022; Yang and Liang, 2025). Although prior work has validated academic buoyancy as a mediator connecting environmental resources to learning outcomes (Datu and Yang, 2021; Putwain and Wood, 2023), its role in L2 learners’ WTC has not been thoroughly explored (Yun et al., 2018). This research gap is especially obvious in China’s EFL education settings. Local sociocultural features, including hierarchical classroom culture, face concerns and high parental expectations, all interfere with learners’ choices in language communication (Cortazzi and Jin, 1996; Cui and Mahfoodh, 2026; Wen and Clément, 2003). Within such contexts, students’ WTC may rely not merely on their affective states but also on their ability to adapt to daily academic and sociocultural requirements.

Against this background, the present study proposed and tested a structural equation model. This model examined the direct and indirect effects of social support from teachers, parents, and peers on WTC, with academic buoyancy acting as the mediating variable among Chinese tertiary EFL learners. By integrating these constructs within a socially embedded language learning framework, this study delivers three advancements to existing WTC research. First, it extends current theoretical frameworks of WTC by incorporating academic buoyancy, an adaptive capacity construct that has received limited attention in L2 communication research. Second, it identifies adaptive coping capacity as an important psychological pathway linking contextual support to communicative willingness, moving beyond the dominant focus on affective mediators such as anxiety and enjoyment. Third, it differentiates the distinct roles of teacher, parent, and peer support, delivering deeper insight into how multisource social support is associated with learners’ WTC.

2Literature review

2.1Willingness to communicate

Initial L1 communication models treated WTC as a static predisposition, characteristically driving a person’s enduring readiness or reluctance to start conversations (Burgoon, 1976; McCroskey and Baer, 1985). In the field of SLA, MacIntyre et al. (1998) reconceptualized WTC as learners’ situational preparedness to initiate a conversation with a specific individual or group using an L2. This definition marked an important theoretical shift by emphasizing the situational and contextual dimensions of communicative willingness. Subsequent studies have increasingly conceptualized WTC as a dynamic construct responsive to interpersonal relationships, affective states, and situational conditions (MacIntyre et al., 2019b; Yang and Lian, 2023). Often described as a “psychological gatekeeper” to actual language use (Lee, 2022), WTC represents the final volitional stage preceding communication and is closely associated with learners’ opportunities for L2 interaction and their overall learning outcomes (Dewaele, 2019; Zhang et al., 2024).

Empirical studies have identified numerous antecedents of WTC. Research has gradually moved from an early emphasis on internal variables, such as self-esteem and perceived communicative competence (Leeming et al., 2024; MacIntyre, 1994; Zhou et al., 2023), toward more ecologically oriented perspectives that recognize the interaction between individual characteristics and contextual conditions (Peng, 2012; Wang et al., 2021). Within this ecological perspective, two conceptually complementary yet seldom integrated constructs warrant particular attention: perceived social support, which reflects learners’ external contextual resources, and academic buoyancy, which represents learners’ internal adaptive capacity to translate supportive environments into sustained communicative engagement.

2.2WTC and perceived social support

Extensive research on WTC has proven that social environments are strongly associated with learners’ willingness to use an L2 for communication (Li et al., 2021; Peng, 2025). Within this line of inquiry, perceived social support is defined as learners’ subjective assessment of the affective, cognitive, and instrumental support they obtain from key sources such as teachers, peers, and family members (Zimet et al., 1988).

Perceived social support has consistently been linked to higher levels of motivation, engagement, and psychological well-being (Poudel et al., 2020; Wang et al., 2023). In EFL contexts, teacher and peer support have been shown to facilitate WTC by reducing anxiety and creating classroom environments that encourage communicative risk-taking (Ducker, 2021; Liu L. et al., 2024; Yang Y. et al., 2024). For example, Liu and Hong (2021) found that perceived social support positively predicted WTC among Chinese EFL learners, primarily through its role in lowering foreign language anxiety. Likewise, Zarrinabadi and Khodarahmi (2024) demonstrated that teacher support enhanced WTC by promoting both cognitive and affective engagement.

Although substantial evidence supports the relationship between perceived social support and WTC, most existing studies have primarily examined affective mediators, such as anxiety or enjoyment (Lee, 2022; Li C. et al., 2025; Wang and Hui, 2024). Comparatively, less attention has been paid to how perceived social support may enhance learners’ longer-term capacity to manage everyday academic challenges that could impede communication. This limitation highlights the need to consider adaptive psychological resources alongside affective mechanisms when explaining how social support influences communicative willingness.

2.3WTC and academic buoyancy

Academic buoyancy captures students’ capability to handle routine academic challenges, including performance pressure, difficult tasks, and minor setbacks (Martin and Marsh, 2008). It is conceptually and empirically distinct from two related adaptive constructs discussed in SLA research: L2 resilience and L2 grit. Specifically, academic buoyancy primarily focuses on managing everyday and relatively minor academic adversities (Martin and Marsh, 2009; Putwain et al., 2020), which corresponds to the ongoing learning stresses commonly experienced in language classrooms. By contrast, L2 resilience generally concerns learners’ ability to overcome more severe or acute linguistic and academic challenges (Yun et al., 2018). Meanwhile, L2 grit captures a learner’s persistent effort and enduring enthusiasm for long-run language learning goals (Bensalem et al., 2025), highlighting commitment to distant objectives. Recent SLA research has further examined L2 grit alongside other learner characteristics, such as motivation, anxiety, and achievement, particularly in relation to growth-oriented beliefs and mindset interventions (Afsharmehr, 2026; Ansari, 2026). While these lines of research collectively reinforce the broader relevance of adaptive learner characteristics in SLA, they also underscore key differences in their primary focus (Martin and Marsh, 2009). Unlike recovery from major adversity or long-term persistence, academic buoyancy specifically targets immediate responses to daily academic hurdles as they arise. This specific focus renders it uniquely relevant to WTC, which is itself highly situational and context dependent (Lee, 2022; MacIntyre et al., 2019a).

Recent studies in EFL contexts have provided substantial empirical evidence linking academic buoyancy with a range of positive learning-related outcomes. Higher academic buoyancy has been associated with stronger motivational persistence, superior emotion regulation, and elevated classroom engagement (Xu and Zheng, 2025; Xu and Wang, 2022). Additionally, academic buoyancy correlates positively with self-efficacy (Yang S. et al., 2024) and classroom engagement (Liu H. et al., 2024), while exhibiting a negative association with academic burnout (Li X. et al., 2025). These findings suggest that learners who are better equipped to manage routine academic difficulties are more likely to maintain adaptive psychological and behavioral engagement. Emerging evidence further indicates that academic buoyancy may operate as a mediating mechanism through which contextual support promotes positive learning behaviors (Yang and Liang, 2025).

Although direct empirical research connecting academic buoyancy specifically with L2 WTC remains scarce, related evidence concerning other adaptive psychological resources provides a solid foundation for positing such a relationship. For instance, Amini et al. (2025) revealed that personal adaptive resources (resilience and perceived competence) mediated the links between social comparison orientations and L2 WTC among undergraduate EFL learners. While their findings focused on resilience rather than buoyancy, they provide strong theoretical support for the notion that learner-level adaptive resources explain how contextual perceptions relate to communicative willingness. This offers a compelling rationale for examining academic buoyancy as a potential mediator between perceived social support and L2 WTC in this study.

However, few prior studies have simultaneously investigated perceived social support, academic buoyancy, and WTC within a single integrated framework. Most previous research has either treated social support and academic buoyancy as independent predictors (Ibrahim and Basim, 2024) or focused primarily on affective mediators linking perceived social support to WTC (Yang Y. et al., 2024). Consequently, little empirical work has specifically investigated academic buoyancy as a psychological mechanism through which social support may strengthen EFL learners’ WTC.

This study attempts to explore the possible connection among these constructs. Given the multidimensional nature of perceived social support, the present study examines teacher, parent, and peer support separately to investigate their differentiated effects on academic buoyancy and WTC in the EFL context of China.

A structural model of academic buoyancy, WTC, and support from teachers, parents, and peers was proposed based on the aforementioned theoretical reasoning. Figure 1 shows a graphic representation of the model along with its paths. We proposed the following hypotheses:

Hypothesis 1: Teacher support positively predicts WTC.

Hypothesis 2: Parent support positively predicts WTC.

Hypothesis 3: Peer support positively predicts WTC.

Hypothesis 4: Teacher support positively predicts academic buoyancy.

Hypothesis 5: Parent support positively predicts academic buoyancy.

Hypothesis 6: Peer support positively predicts academic buoyancy.

Hypothesis 7: Academic buoyancy positively predicts WTC.

Hypothesis 8: Academic buoyancy mediates the relationship between teacher support and WTC.

Hypothesis 9: Academic buoyancy mediates the relationship between parent support and WTC.

Hypothesis 10: Academic buoyancy mediates the relationship between peer support and WTC.

3Methodology

3.1Participants and the setting

Using convenience sampling, we recruited first-year non-English-major undergraduates from a public university located in eastern China. A total of 374 questionnaires were collected. Five cases were removed because their response times were excessively short, leaving 369 valid cases and a valid response rate of 98.66%. All questionnaire items were mandatory, so the final dataset contained no missing values. The final sample of 369 participants provided an adequate basis for the planned analyses (Kline, 2023). The sample comprised exclusively freshmen enrolled in non-English degree programs, representing students in the initial stage of university study (Owusu-Agyeman and Mugume, 2023; Terenzini et al., 1994).

The final sample comprised 234 female participants (63.4%) and 135 male respondents (36.6%). Most individuals were 18 years old (59.1%) or 19 years old (30.6%). Regarding academic backgrounds, 48.5% of participants were from science-related disciplines, 40.1% from arts programs, and 11.4% from other fields including fine arts and physical education. On average, respondents had roughly 10 years of formal English learning experience. In addition, 99.2% of participants had no earlier experience of studying or living in English-speaking regions, indicating that the sample was predominantly situated in an EFL learning context.

3.2Instruments

Three widely used self-report scales were adopted to measure the core constructs: WTC in English, perceived social support from teachers, parents, and peers, and academic buoyancy. All questionnaire items adopted a 5-point Likert scoring standard, ranging from 1 (strongly disagree) to 5 (strongly agree). To help participants fully understand the items, the questionnaires were presented in Chinese. The WTC and academic buoyancy scales were adapted from English into Chinese by the research team following a forward- and back-translation procedure, with minor discrepancies resolved to ensure semantic equivalence. The MSPSS was administered using an authorized Chinese version (Yang X. et al., 2024); therefore, no additional translation was performed for this scale.

3.2.1WTC scale

WTC in English was assessed using a 10-item scale developed by Peng and Woodrow (2010), designed to evaluate the readiness of learners to participate in various classroom-based oral activities, such as role plays and asking questions. This scale has been applied in EFL research and is considered a reliable indicator of situational WTC. A sample item is “I am willing to give a short self-introduction without notes in English to the class.” In this study, the scale had a Cronbach’s alpha of 0.915.

3.2.2The perceived social support scale

Perceived social support was measured with the authorized Chinese MSPSS (Yang X. et al., 2024). The original MSPSS contains family, friends, and significant-other subscales. Because the significant-other subscale does not specifically assess teacher support, the present authors adapted this subscale into a teacher-support subscale for the Chinese EFL context. This adaptation was based on the rationale that teachers are salient significant others for university EFL learners. Permission to use and adapt the scale was obtained from Yang X. et al. (2024). The adapted scale comprised three 4-item subscales: teacher support (α = 0.887), parent support (α = 0.900), and peer support (α = 0.864). A separate CFA of the 12 MSPSS items supported the three-factor structure in the present sample (χ2 = 127.449, df = 51, CFI = 0.971, TLI = 0.962, RMSEA = 0.064, SRMR = 0.022). Sample items include “My teacher is available when I need help,” “My parents are willing to help me when I have difficulties,” and “My friends are willing to share my joys and sorrows.” Full items and detailed CFA results are provided in Supplementary Tables S1, S2.

3.2.3The academic buoyancy scale

A 4-item scale from Martin and Marsh (2008) was adopted to measure academic buoyancy. The scale evaluates students’ capability to handle routine academic difficulties and minor frustrations. The scale has been widely adopted in EFL and higher education research and has shown satisfactory reliability and validity. A sample item is “I’m good at dealing with setbacks at school.” The internal consistency of this scale was reported to be 0.880.

3.3Data collection

Data collection was conducted through online questionnaires distributed during regular class hours using Wenjuanxing, a widely used online survey application in China. The questionnaire combined the three scales described above. Informed consent was obtained electronically. The first page of the online questionnaire presented the study information sheet, including the study’s purpose, voluntary nature, confidentiality, anonymity, and the right to withdraw at any time without penalty. Participants could proceed to the questionnaire items only after clicking the “I agree/continue” button to indicate their consent. No questionnaire items were accessible to participants who did not provide consent.

3.4Data analysis

SPSS 26.0 and AMOS 26.0 were utilized for data analysis and SEM, respectively. First, the dataset was screened for invalid responses, missing data, and distributional characteristics. Because all questionnaire items were mandatory, no missing values were present in the final dataset. Skewness and kurtosis were examined at both the variable and item levels to assess the suitability of the data for maximum-likelihood estimation. Second, internal consistency was assessed using Cronbach’s alpha. Confirmatory factor analysis (CFA) was then conducted in AMOS 26.0 to evaluate the measurement model. Model fit was evaluated using multiple indices rather than a single criterion. As commonly used guidelines, χ2/df values below 3.00, CFI and TLI values above 0.90, RMSEA and SRMR values below 0.08, and PCLOSE values above 0.05 were considered indicative of acceptable fit (Hu and Bentler, 1999). These values were treated as guidelines rather than rigid decision rules, and model fit was evaluated collectively in relation to model complexity and theoretical plausibility. Convergent validity was assessed using standardized factor loadings, composite reliability (CR), and average variance extracted (AVE) (Hair et al., 2010), while discriminant validity was examined using the Fornell–Larcker criterion and competing CFA models. Third, common method bias (CMB) was assessed using three complementary procedures. Harman’s single-factor test was conducted as an initial diagnostic, and the hypothesized five-factor measurement model was compared with models incorporating fewer factors. Finally, latent SEM was applied to test the proposed mediation model. In this study, teacher support, parent support, and peer support were operationalized as three distinct exogenous latent variables. Academic buoyancy was set as the mediator, and WTC as the dependent variable. Standardized path coefficients, standard errors, critical ratios, p-values, and explained variance (R2) were examined for the structural paths. Indirect effects were tested using 5,000 bootstrap resamples with 95% bias-corrected confidence intervals (Hayes, 2013). The hypothesized partial mediation model was subsequently compared with theoretically motivated alternative structural models, including a full mediation model specification, to examine the relative empirical compatibility of the proposed structural relationships.

4Results

4.1Preliminary analyses

Of the 374 questionnaires collected, five cases were excluded because of excessively short response times, resulting in 369 valid cases. Because the questionnaire items were mandatory, no missing values were present in the final dataset. Descriptive statistics, reliability coefficients, and distributional characteristics were examined before the measurement and structural models were tested. Cronbach’s alpha coefficients ranged from 0.864 to 0.915 across the study constructs, indicating satisfactory internal consistency. The teacher, parent, and peer support yielded coefficients of 0.887, 0.900, and 0.864, respectively. The Academic Buoyancy scale showed an alpha of 0.880, and the WTC scale yielded an alpha of 0.915. Normality was also examined. At the variable level, skewness ranged from −0.329 to 0.012 and kurtosis from −0.166 to 1.382 (Table 1). At the item level, skewness ranged from −0.346 to 0.230 and kurtosis from −0.479 to 0.899 (Table 2). These values indicated no substantial departures from univariate normality and supported the use of maximum-likelihood estimation in the subsequent CFA and SEM analyses.

Table 1Variable-level skewness, kurtosis, means, standard deviations, and correlations among the study variables.

Variable Descriptive statistics Pearson correlations

Skewness Kurtosis M SD 1 2 3 4 5

1. Teacher support 0.012 1.382 3.292 0.657 1

2. Parent support −0.093 −0.166 3.263 0.795 0.395** 1

3. Peer support −0.191 0.582 3.475 0.640 0.487** 0.288** 1

4. Academic buoyancy −0.264 1.137 3.310 0.705 0.505** 0.360** 0.424** 1

5. WTC −0.329 1.296 3.266 0.589 0.567** 0.457** 0.505** 0.482** 1

M, mean; SD, standard deviation; WTC, willingness to communicate in English. Correlations are Pearson’s r values based on composite observed scores. **p < 0.01 (two-tailed).

Table 2Item-level skewness and kurtosis for all questionnaire items.

Item Skewness Kurtosis

WTC1 0.004 0.295

WTC2 0.024 −0.043

WTC3 −0.132 −0.078

WTC4 −0.142 0.414

WTC5 −0.069 −0.106

WTC6 −0.062 0.241

WTC7 −0.251 0.053

WTC8 −0.153 −0.325

WTC9 −0.080 −0.088

WTC10 −0.116 0.105

Teacher support 1 −0.083 0.264

Teacher support 2 −0.031 0.364

Teacher support 5 0.230 0.595

Teacher support 10 −0.167 0.899

Parent support 3 0.061 −0.440

Parent support 4 −0.261 −0.401

Parent support 8 0.055 −0.479

Parent support 11 0.006 0.108

Peer support 6 0.106 0.027

Peer support 7 −0.058 −0.019

Peer support 9 −0.204 −0.150

Peer support 12 −0.346 0.558

Academic buoyancy 1 −0.138 0.296

Academic buoyancy 2 −0.171 0.602

Academic buoyancy 3 −0.131 0.386

Academic buoyancy 4 −0.100 0.443

WTC, willingness to communicate in English. All item-level skewness and kurtosis values fell within the acceptable range for maximum-likelihood estimation.

4.2Measurement model and construct validity

A five-factor CFA model was estimated in AMOS 26.0, with WTC, teacher support, parent support, peer support, and academic buoyancy specified as distinct latent constructs. The results demonstrated an acceptable overall fit to the data: χ2 = 507.587, df = 289, p < 0.001, χ2/df = 1.756, CFI = 0.961, TLI = 0.957, RMSEA = 0.045, SRMR = 0.037, AIC = 631.587, BIC = 874.057 (Table 3). NFI was marginally below 0.90, while all other core fit indices met conventional thresholds, supporting acceptable model fit. All standardized factor loadings reached statistical significance at the level of p < 0.001 and ranged from 0.732 to 0.893. The composite reliability (CR) values ranged from 0.865 to 0.915, while average variance extracted (AVE) values varied from 0.518 to 0.692 (Table 4). These results provided evidence of adequate convergent validity across the five latent constructs.

Table 3Goodness-of-fit indices for the measurement model and structural model.

Fit index Measurement model (CFA) Structural model (SEM) Common reference value

χ2 507.587 507.587 –

df 289 289 –

χ2/df 1.756 1.756 1<χ2/df<3

p 0.000 0.000 –

GFI 0.907 0.907 >0.90

NFI 0.887 0.887 >0.90

IFI 0.962 0.962 >0.90

CFI 0.961 0.961 >0.90

TLI 0.957 0.957 >0.90

RMSEA 0.045 0.045 <0.08

PCLOSE 0.880 0.880 >0.05

SRMR 0.037 0.037 <0.08

AIC 631.587 631.587 –

BIC 874.057 874.057 –

The structural model is a saturated structural model with the same number of free parameters as the CFA model; therefore, global fit indices are identical. NFI = 0.887 was slightly below the conventional 0.90 cutoff; however, the remaining fit indices indicated acceptable model fit. The marginally lower NFI did not materially compromise the overall adequacy of the model.

Table 4Convergent validity, discriminant validity and correlations among constructs.

Construct CR AVE 1 2 3 4 5

WTC 0.915 0.518 0.720

Teacher support 0.888 0.664 0.631 0.815

Parent support 0.900 0.692 0.509 0.444 0.832

Peer support 0.865 0.616 0.569 0.554 0.328 0.785

Academic buoyancy 0.880 0.648 0.539 0.571 0.408 0.488 0.805

Bold values on the diagonal are the square root of the AVE for each construct; off-diagonal values are latent variable correlations from the five-factor CFA model. CR = composite reliability; AVE = average variance extracted; WTC = willingness to communicate in English.

The Fornell–Larcker approach was adopted to examine discriminant validity. As presented in Table 4, the square root of the AVE for each construct exceeded its correlations with the other latent constructs, supporting the empirical distinctiveness of the five constructs. Competing CFA models provided additional evidence for construct distinctiveness (Table 5). The hypothesized five-factor model showed substantially better fit than models combining two or more constructs. For example, the one-factor model showed poor fit, χ2/df = 8.271, CFI = 0.616, TLI = 0.582, RMSEA = 0.141, and SRMR = 0.105. The progressively poorer fit of the four-, three-, two-, and one-factor models was consistent with the specification of WTC, the three sources of social support, and academic buoyancy as distinguishable constructs.

Table 5Competing CFA models for construct distinctiveness and common-method bias assessment.

Model Factor structure χ2/df CFI TLI IFI GFI RMSEA SRMR

Five-factor Model WTC, Teacher Support, Parent Support, Peer Support, Academic Buoyancy 1.756 0.961 0.957 0.962 0.907 0.045 0.037

Four-factor Model WTC + Teacher Support, Parent Support, Peer Support, Academic Buoyancy 3.343 0.879 0.865 0.865 0.797 0.080 0.061

Three-factor Model WTC + Teacher Support + Parent Support, Peer Support, Academic Buoyancy 5.483 0.765 0.742 0.767 0.696 0.110 0.084

Two-factor Model WTC + Teacher Support + Parent Support+ Peer Support, Academic Buoyancy 6.810 0.694 0.666 0.695 0.642 0.126 0.095

One-factor Model WTC + Teacher Support + Parent Support+ Peer Support+ Academic Buoyancy 8.271 0.616 0.582 0.617 0.590 0.141 0.105

CMB Factor Model – 1.700 0.967 0.960 0.968 0.916 0.044 0.036

CMB, common-method bias; WTC, willingness to communicate in English. The five-factor model represents the hypothesized measurement model. The CMB factor model includes an unmeasured latent common-method factor.

4.3Assessment of common method bias

To evaluate potential common method variance, two complementary approaches were applied. First, Harman’s single-factor test indicated that the single factor accounted for 41.55% of the total variance, which was below the commonly used 50% reference point. This result did not suggest that a single common factor accounted for the majority of the observed variance. Then, an unmeasured common latent method factor was incorporated into the measurement model. The resulting model showed χ2/df = 1.700, CFI = 0.967, TLI = 0.960, RMSEA = 0.044, and SRMR = 0.036, compared with χ2/df = 1.756, CFI = 0.961, TLI = 0.957, RMSEA = 0.045, and SRMR = 0.037 for the five-factor model (Table 5). Although the common-method-factor model showed a modest improvement in some fit indices, the improvement was limited. Thus, the analyses did not provide evidence that common method bias substantially altered the measurement model, although its influence cannot be completely ruled out.

4.4Correlation analysis

Correlations among the observed composite scores of teacher support, parent support, peer support, academic buoyancy and WTC were examined using Pearson’s correlation test. All three sources of support were positively correlated with academic buoyancy (Table 1). Specifically, teacher support showed a moderately positive correlation with academic buoyancy (r = 0.505, p < 0.01), followed by peer support (r = 0.424, p < 0.01) and parent support (r = 0.360, p < 0.01).

Similarly, teacher support showed the strongest observed-score correlation with WTC (r = 0.567, p < 0.01), followed by peer support (r = 0.505, p < 0.01) and parent support (r = 0.457, p < 0.01). Academic buoyancy was also positively correlated with WTC (r = 0.482, p < 0.01). These bivariate associations were consistent with the expected direction of relationships and provided a basis for proceeding to the multivariate structural analysis.

4.5Structural model testing

The hypothesized structural relationships were tested using a structural equation modeling approach. We modeled teacher, parent, and peer support as correlated exogenous variables, with academic buoyancy positioned in an intermediary role and WTC designated as the final endogenous variable. Direct structural linkages were estimated from all three support domains toward both academic buoyancy and WTC, complemented by a direct trajectory from academic buoyancy to WTC.

The structural model produced the same global fit statistics as the five-factor CFA model: (χ2 = 507.587, df = 289, χ2/df = 1.756, CFI = 0.961, TLI = 0.957, RMSEA = 0.045, PCLOSE = 0.880, SRMR = 0.037) (Table 3). The identical fit statistics were expected because, with three correlated exogenous support constructs, one mediator, and one outcome, the specified recursive structural relations did not impose additional overidentifying restrictions on the covariance structure relative to the measurement model. Consequently, the global fit indices assess the reproduced covariance structure but do not by themselves establish the directional ordering of the structural paths.

The results of the model path hypothesis test (Figure 2) and Table 6 showed that teacher support was positively associated with both academic buoyancy (β = 0.370, SE = 0.067, CR = 5.406, p < 0.001) and WTC (β = 0.304, SE = 0.057, CR = 4.737, p < 0.001). Parent support was positively associated with academic buoyancy (β = 0.169, SE = 0.047, CR = 3.017, p = 0.003) and WTC (β = 0.231, SE = 0.039, CR = 4.536, p < 0.001). Similarly, peer support was also positively associated with academic buoyancy (β = 0.228, SE = 0.068, CR = 3.591, p < 0.001) and WTC (β = 0.253, SE = 0.058, CR = 4.349, p < 0.001). Furthermore, academic buoyancy was a significant positive predictor of WTC (β = 0.148, SE = 0.053, CR = 2.532, p = 0.011). The model explained 39.1% of the variance in academic buoyancy (R2 = 0.391) and 53.3% of the variance in WTC (R2 = 0.533). Thus, all hypothesized direct paths reached statistical significance in the specified model.

Table 6Standardized path coefficients for the structural model.

Direct path β SE CR p R2

Teacher support → Academic buoyancy 0.370 0.067 5.406 < 0.001 0.391

Parent support → Academic buoyancy 0.169 0.047 3.017 0.003 .

Peer support → Academic buoyancy 0.228 0.068 3.591 < 0.001 .

Teacher support → WTC 0.304 0.057 4.737 < 0.001 0.533

Parent support → WTC 0.231 0.039 4.536 < 0.001 .

Peer support → WTC 0.253 0.058 4.349 < 0.001 .

Academic buoyancy → WTC 0.148 0.053 2.532 0.011 .

The R2 value of 0.391 refers to academic buoyancy as the dependent variable, and the R2 value of 0.533 refers to WTC as the dependent variable.

4.6Mediation analysis and alternative model comparison

To assess the indirect effects of teacher, parent, and peer support on WTC through academic buoyancy, we performed a bootstrapping procedure with 5,000 resamples. Bias-corrected 95% confidence intervals were examined to determine whether the indirect effects differed significantly from zero. As shown in Table 7, all three indirect effects were statistically significant.

Table 7Bootstrap estimates of the total, direct, and indirect effects.

Path Effect type β S.E. 95% CI p % of the total effect

Lower Upper

Teacher support → Academic Buoyancy → WTC Total Effect 0.359 0.084 0.191 0.517 0.000 -

Direct Effect 0.304 0.089 0.126 0.476 0.001 84.68%

Indirect Effect 0.055 0.028 0.010 0.124 0.016 15.32%

Parent support → Academic Buoyancy → WTC Total Effect 0.256 0.062 0.134 0.377 0.000 -

Direct Effect 0.231 0.060 0.114 0.356 0.000 90.23%

Indirect Effect 0.025 0.017 0.002 0.071 0.024 9.77%

Peer support → Academic Buoyancy → WTC Total Effect 0.287 0.073 0.147 0.433 0.000 -

Direct Effect 0.253 0.078 0.102 0.408 0.001 88.11%

Indirect Effect 0.034 0.022 0.004 0.098 0.016 11.89%

Bootstrap resamples = 5,000; WTC, willingness to communicate in English; CI, bias-corrected confidence interval.

For teacher support, the total effect on WTC was β = 0.359 (p < 0.001), the direct effect was β = 0.304 (p < 0.001), and the indirect effect through academic buoyancy was β = 0.055. The indirect effect accounted for 15.32% of the total effect. For peer support, the total effect was β = 0.287 (p < 0.001), the direct effect was β = 0.253, and the indirect effect was β = 0.034, accounting for 11.89% of the total effect. For parent support, the total effect was β = 0.256 (p < 0.001), with a direct effect of β = 0.231 (p < 0.001), and an indirect effect of β = 0.025, accounting for 9.77% of the total effect. Given that the 95% confidence intervals across all three indirect effects did not include zero and the direct effects remained significant, the results were consistent with partial statistical mediation.

To examine whether the hypothesized structural specification provided a better empirical account than a theoretically plausible alternative, we compared the hypothesized partial mediation model with a full mediation model, in which the direct paths from the three sources of social support to WTC were constrained to zero. Model fit was evaluated using χ2, df, CFI, TLI, RMSEA, SRMR, AIC, and BIC, with detailed results reported in Table 8. The full mediation model yielded a significantly worse fit than the partial mediation model, Δχ2 (3) = 107.503, p < 0.001. Furthermore, the partial mediation model demonstrated superior goodness-of-fit indices (CFI = 0.961 vs. 0.943; TLI = 0.957 vs. 0.936; RMSEA = 0.045 vs. 0.055; SRMR = 0.037 vs. 0.061) and lower information criteria (AIC = 631.587 vs. 733.090; BIC = 874.057 vs. 963.827). Although the full mediation model showed a slightly higher R2 for academic buoyancy (0.444 vs. 0.391), the partial mediation model accounted for substantially more variance in the primary outcome, WTC (R2 = 0.533 vs. 0.350).

Table 8Comparison of the hypothesized partial mediation model and the full mediation model.

Fit index Partial mediation model (hypothesized) Full mediation model

χ2 507.587 615.090

df 289 292

χ2/df 1.756 2.106

Δχ2 (Δdf) – 107.503(3)

p 0.000 0.000

GFI 0.907 0.891

NFI 0.887 0.869

IFI 0.962 0.897

CFI 0.961 0.943

TLI 0.957 0.936

RMSEA 0.045 0.055

SRMR 0.037 0.061

AIC 631.587 733.090

BIC 874.057 963.827

R2 (Academic buoyancy) 0.391 0.444

R2 (WTC) 0.533 0.350

R2 values are model-specific, equation-level estimates for each endogenous variable and should be interpreted together with the overall model-fit indices.

Importantly, R2 is model- and outcome-specific and should not be interpreted as an index of overall model adequacy. The two models used different structural specifications, and the R2 values therefore represent model-specific estimates. The higher R2 for academic buoyancy in the full-mediation model (0.444 vs. 0.391) does not indicate better overall model fit or greater explanatory power for the focal outcome. For WTC, the full-mediation model explained 35.0% of the variance, compared with 53.3% in the partial-mediation model, a difference of 18.3 percentage points. This lower explanatory power is consistent with the full mediation model constraining the three direct support-to-WTC paths to zero, whereas these paths were freely estimated in the partial mediation model. Thus, the higher model-specific R2 for academic buoyancy does not contradict the full-mediation model’s lower explanatory power for WTC or its poorer overall fit. Based on the significant chi-square difference and the more favorable fit indices and information criteria, the partial mediation model was retained. Given the cross-sectional design, this model comparison is interpreted as evidence of relative empirical fit to the observed data rather than definitive evidence of causal or temporal ordering.

5Discussion

The study examined how academic buoyancy was related to the associations of teacher, parent, and peer support with WTC among Chinese EFL undergraduates. Using SEM with bootstrapping, the findings provided empirical support for all hypothesized direct and indirect relationships across the three sources of social support. Disaggregated data showed that each type of social support was differentially associated with academic buoyancy and WTC. This outcome helps build a more context-aware interpretation of learners’ WTC.

Overall, the findings suggest that external social resources interact with learners’ internal adaptive traits to jointly predict their WTC. Integrating academic buoyancy into established WTC frameworks enables a more holistic understanding of communicative readiness. Further evidence points to the notion that adaptive coping works in tandem with affective and cognitive factors to foster learners’ willingness to engage in L2 interaction. As such, academic buoyancy appears to act as a vital psychological link that may facilitate the translation of perceived social support into consistent communicative participation. The results also revealed varying strength across different support providers, with teacher support exhibiting the most substantial association, followed by peer and parent support. By clarifying the dynamic relationship between contextual support and academic buoyancy in L2 communication, this research further enriches existing theoretical models in this field.

5.1Three sources of perceived social support as direct predictors of WTC (H1–H3 supported)

As predicted in Hypotheses 1–3, teacher, parent, and peer support were positively associated with WTC. This finding accords with the ecological view of WTC, which regards it not as a static disposition but as a contextually sensitive variable and suggests that social surroundings are related to individual communicative tendencies (Jamalvandi et al., 2020; Khajavy et al., 2018; Liu L. et al., 2024; Peng, 2012; Zhang et al., 2024). Empirically, this finding also resonates with previous SLA research, where supportive classroom environments tend to foster learners’ communicative engagement and WTC (Li C. et al., 2025; Liu H. et al., 2024; Zarrinabadi and Khodarahmi, 2024). Notably, the direct association of perceived social support with WTC still held after academic buoyancy was included as a mediating variable. This may imply that social support acts as a fundamental contextual resource, one that could mitigate psychological barriers in L2 interaction (Tu, 2021). Such function appears particularly relevant for EFL learners in China, a setting widely noted for prevailing face concerns and fear of negative evaluation (Li et al., 2023).

The relatively stronger role of teacher support may be partly understood in relation to the institutional position of teachers in Chinese tertiary EFL classrooms. In local classrooms, teachers generally take charge of instructional activities and largely influence the overall learning atmosphere (Cortazzi and Jin, 1996; Wen and Clément, 2003). Although slightly weaker than teacher support, peer support remains meaningful, likely through mechanisms such as collaborative learning, peer modeling, and the alleviation of interpersonal anxiety during pair or group activities (Ducker, 2021; Liu L. et al., 2024; Zarrinabadi and Khodarahmi, 2024). The descriptively larger coefficient for peer support than parent support may suggest a stronger association with communicative readiness, although the difference between these coefficients was not formally tested.

These findings from previous studies have predominantly emphasized teacher support (e.g., Hejazi et al. (2023); Liu L. et al. (2024); Ma and Xia (2026); MacIntyre et al. (2001); Peng (2020)), demonstrating that not all sources of social support exert the same influence. By disaggregating teacher, peer, and parent support, teacher support showed the strongest associations descriptively with WTC among Chinese EFL learners, reflecting the hierarchical, exam-oriented nature of the educational environment (Wen and Clément, 2003). While parent support yielded a relatively smaller direct effect, it remains a valuable source of encouragement for language learners. Instead of exerting influence via in-class interactions, parent support may function indirectly by shaping learners’ long-term motivation and emotional states, thereby affecting their WTC (MacIntyre et al., 2001).

5.2Three sources of perceived social support as predictors of academic buoyancy (H4–H6 supported)

Hypotheses 4–6 were supported, with teacher, parent, and peer support each positively associated with academic buoyancy. The relatively stronger association observed for teacher support suggests that learners’ perceptions of support within the immediate academic environment may be particularly relevant to their perceived capacity to deal with routine academic difficulties. This interpretation aligns with the principles of the conservation of resources theory (Hobfoll, 1989), which posits that external social resources can be internalized to enhance learners’ psychological capacity to manage everyday academic challenges.

For first-year university students navigating a critical academic transition (Li et al., 2023), perceived support from teachers, peers, and parents serves as an essential external resource that fosters confidence in handling routine EFL learning challenges, including linguistic difficulties in speaking, performance pressure during classroom presentations, and minor setbacks in language assessments (Xu and Wang, 2022). These findings extend educational psychology research on the social antecedents of academic buoyancy to the SLA context (Li et al., 2024; Putwain et al., 2020) and support Yun et al.’s (2018) argument that academic buoyancy in the L2 classroom is context-sensitive rather than purely dispositional. Moreover, the results complement the framework proposed by Zhang et al. (2024) by introducing an adaptive capacity construct shaped by social contextual influences.

5.3Academic buoyancy as a predictor of WTC (H7 supported)

In line with Hypothesis 7, academic buoyancy was positively associated with WTC. This result points to the contribution of adaptive coping capacity, which operates alongside more commonly studied affective variables including enjoyment and anxiety (Li C. et al., 2025; MacIntyre and Gregersen, 2022), as well as cognitive factors such as perceived communicative competence and growth mindset (Hejazi et al., 2023; Zhang et al., 2024). The observed pattern accords with a body of prior research that connects academic buoyancy to favorable learning outcomes in L2 contexts (Bai and Hu, 2025). It also suggests this adaptive capacity can serve as a functional mechanism to sustain learners’ WTC. This study advances extant literature by providing evidence of a direct predictive association between academic buoyancy and WTC. Learners with greater academic buoyancy can better manage negative emotions when facing communication difficulties, reframe difficulties as manageable setbacks, and continue participating orally despite temporary failures (Martin and Marsh, 2008; Yang and Liang, 2025). This finding also addresses the research appeals within SLA scholarship to shift focus away from short-term emotional states. Instead, scholars have urged greater attention to stable ability-based constructs that help learners maintain consistent engagement in L2 communication (Dewaele and MacIntyre, 2024; Souzandehfar and Ahmed Abdel-Al Ibrahim, 2023).

5.4Partial mediating role of academic buoyancy (H8–H10 supported)

Hypotheses 8–10 were supported, as academic buoyancy partially mediated the relationship between all three sources of social support and WTC. The mediation findings are valuable because they suggest that the link between social support and communicative willingness may involve more than the immediate interpersonal environment. Learners’ adaptive responses to routine academic difficulties may represent one psychological pathway through which perceived social resources are aligned with WTC.

These results offer empirical support for extending MacIntyre et al.’s (1998, 2019b) multi-layered WTC framework, which categorizes antecedents into distal individual traits, intermediate psychological factors, and proximal environmental influences. Traditionally, the intermediate tier emphasized immediate affective and cognitive variables, such as perceived communicative competence and anxiety (Lee, 2022; MacIntyre and Gregersen, 2022). This study suggests a plausible extension of this framework by demonstrating that academic buoyancy, a relatively stable adaptive psychological capacity, may also function as a critical intermediate factor, mediating the relationship between proximal environmental resources and WTC.

Notably, while the direct relationships between social support and WTC constituted the primary share of the overall association, the indirect pathways through academic buoyancy also demonstrated statistical significance. The present data also hints that adaptive coping ability may act as a meaningful indirect mechanism connecting social support with learners’ WTC (Acosta-Manzano et al., 2024; Wang and Hui, 2024). In line with observations reported by Yang and Liang (2025), who noted that academic buoyancy mediates links between contextual resources and learning-related outcomes, this study extends this line of reasoning to the domain of communicative willingness.

When viewed alongside the cognitive and affective antecedent model of WTC developed by Zhang et al. (2024), the present findings underline the combined importance of social contextual conditions and personal adaptive resources. The current evidence lends further support to the widely shared view that WTC is associated with multiple interacting forces and is embedded in the social contexts where learning takes place (Dewaele, 2019; Peng, 2025).

Results from this study also suggest that academic buoyancy may be understood as a form of adaptive communicative agency (Yun et al., 2018). Beyond supporting emotional stability, this quality appears to help learners interpret communication-related challenges as routine academic obstacles rather than overwhelming threats (Bai and Hu, 2025). This kind of constructive interpretation may be especially important for sustained classroom participation in EFL contexts strongly shaped by sociocultural expectations, where public oral performance is often exposed to evaluative scrutiny from others. Perceived social support may help establish a secure psychological environment, while academic buoyancy provides the internal resources needed for learners to act confidently within that environment (Lin et al., 2025). Through this combined process, external supportive conditions can be gradually translated into consistent and active communicative engagement.

6Pedagogical implications

The present study offers a few contextually responsive and theoretically informed pedagogical suggestions for supporting WTC among EFL learners. One central implication relates to the role of perceived social support in classroom dynamics. When viewed from a social-ecological lens, language classrooms can be understood not only as spaces for receiving linguistic input, but also as relational environments in which interpersonal connections shape students’ decisions to participate orally (Peng, 2012). For instructors, this means intentionally creating a classroom atmosphere characterized by emotional security and reduced evaluative stress. In culturally sensitive contexts such as China, where learners often express strong concerns over face maintenance, modest yet consistent classroom practices may help reduce unnecessary anxiety. These practices can include supportive responses to student contributions, genuine affirmation of participation regardless of linguistic accuracy, and the framing of errors as natural steps in language development. Over time, such actions may help learners feel safer and more prepared to initiate communication.

The mediating function of academic buoyancy also adds an important layer to instructional design. Learners often need not only external support but also internal psychological resources to manage the everyday challenges of speaking in an L2 (Yun et al., 2018). The broaden-and-build theory states that supportive social environments may help develop such adaptive qualities, which in turn enable learners to sustain engagement when faced with setbacks (Fredrickson, 2004). For practical teaching, this could mean shifting focus away from outcome-oriented evaluation toward process-oriented support. Useful strategies may involve designing communication tasks with gradually increasing difficulty, providing opportunities for repeated practice and reflective thinking, and modeling constructive responses to communication failures. When teachers help learners view struggle as a typical part of language learning and emphasize effort over perfection, students may be more likely to maintain long-term willingness to speak in class.

Differences observed among teacher, parent, and peer support also suggest the value of building a multi-layered support system. Within the physical classroom, structured peer activities, including collaborative tasks, interactive dialogues, and group problem-solving, may strengthen peer relationships while increasing opportunities for meaningful communication (Li Z. et al., 2025). Beyond classroom settings, institutions and teaching staff may consider engaging with families to foster a more nuanced recognition of socio-affective dynamics within the language learning process. Such efforts can help strengthen the broader support system surrounding individual learners.

These implications may also extend to online and blended learning environments, which have become increasingly common in higher education. Digital tools can either support or restrict interaction, largely based on how teachers design and implement learning activities (Wang et al., 2026). Teachers may therefore take deliberate steps to strengthen social presence and emotional connection in virtual spaces (Peng, 2025). Examples include structured online discussions, collaborative digital assignments, and timely personalized feedback. Supporting learners in overcoming common online challenges, such as reduced immediacy or increased self-consciousness, may encourage more active communication across different learning formats.

Overall, the promotion of L2 WTC may benefit from a holistic and integrated approach. Such an approach can combine positive interpersonal classroom dynamics, instructional practices that foster adaptive capacities, and task design adjusted to specific contextual needs. When educators address social and psychological factors alongside linguistic development, learners may gradually develop greater confidence and become more active communicators. In turn, these practices can help advance research on L2 communicative development by reinforcing socially situated and process-based understandings of language learning.

7Limitations

Several limitations should be considered when interpreting the findings of this study. First, the cross-sectional design does not permit causal inferences or establish the temporal ordering of the relationships among perceived social support, academic buoyancy, and WTC. Although the SEM results provide evidence of statistically significant direct and indirect associations within the hypothesized model, they should not be interpreted as evidence of causal effects. Second, the study relied on self-reported questionnaire data collected from the same participants at a single time point. Although several procedures were used to assess potential common method bias, including Harman’s single-factor test and a common latent method factor model, the possibility of method-related influences cannot be completely excluded. Third, the participants were recruited from a single public university in eastern China using convenience sampling. The sample consisted exclusively of first-year non-English-major undergraduates, which may limit the transferability of the findings to students from other universities, educational levels, academic disciplines, or sociocultural contexts. Future research could address these limitations by adopting longitudinal or experimental designs to examine the temporal and causal relationships among the variables. Multi-site sampling and more diverse participant groups could also improve the generalizability of the findings. In addition, mixed-methods research could provide more detailed insight into how learners perceive and experience social support and academic buoyancy in different communicative contexts.

8Conclusion

This study focused on a conceptual model designed to explore connections between perceived social support, derived from teachers, parents, and peers, and WTC among EFL learners. The SEM results showed that all three sources of social support were positively associated with both academic buoyancy and WTC. Academic buoyancy was also positively associated with WTC, and its indirect effects partially mediated the associations between each source of social support and WTC. The findings extend existing WTC research by incorporating academic buoyancy as an adaptive psychological construct within a socially embedded framework. They also provide a differentiated view of social support by examining teacher, parent, and peer support separately rather than treating social support as a single undifferentiated source. Overall, the findings suggest that learners’ communicative willingness is associated with both contextual social resources and adaptive psychological capacities. By positioning academic buoyancy alongside established affective and cognitive antecedents of WTC, this study provides an additional perspective for understanding the psychological and contextual factors related to L2 communicative engagement among EFL learners.

References

Acosta-Manzano I. Barrios E. García-Mata J. (2024 ). Enjoyment, anxiety, and out-of-class foreign language use among Spanish foreign language learners . J. Multiling. Multicult. Dev.
45 , 4332 –4347 . doi: 10.1080/01434632.2022.2159966

Afsharmehr E. (2026 ). Investigating the effectiveness of a growth mindsets intervention for enhancing EFL learners’ L2 grit, motivation and achievement . Lang. Teach. Futures
2 , 56 –76 .

Almubarak H. A. Y. Alikhani H. S. H. M. (2026 ). Regrets about the past as the predictors of Iraqi EFL learner’ motivation and willingness to communicate in L2: the mediating role of positive and negative emotions . Lang. Teach. Futures
2 , 109 –130 .

Amini F. Amirian Z. Tavakoli M. (2025 ). Social comparison orientations as predictors of l2 anxiety, enjoyment, and willingness to communicate: unraveling the mediating roles of perceived competence and resilience [Preprint]. Res. Sq . doi: 10.21203/rs.3.rs-8058746/v1

Ansari Y. (2026 ). Fixed and growth mindsets as predictors of anxiety and grit: do myths about brain functioning play a role?
Lang. Teach. Futures
2 , 3 –29 .

Bai C. Hu J. (2025 ). The effect of foreign language enjoyment on the willingness to communicate of non-English majors: the mediating role of L2 grit and academic buoyancy . Front. Educ.
10 :1614742 . doi: 10.3389/feduc.2025.1614742

Bensalem E. Derakhshan A. Alenazi F. H. Thompson A. S. Harizi R. (2025 ). Modeling the contribution of grit, enjoyment, and boredom to predict English as a foreign language students’ willingness to communicate in a blended learning environment . Percept. Mot. Skills
132 , 144 –168 . doi: 10.1177/00315125241289192 , 39423310

Burgoon J. K. (1976 ). The unwillingness-to-communicate scale: development and validation . Commun. Monogr.
43 , 60 –69 . doi: 10.1080/03637757609375916

Collie R. J. Martin A. J. Bottrell D. Armstrong D. Ungar M. Liebenberg L. (2017 ). Social support, academic adversity and academic buoyancy: a person-centred analysis and implications for academic outcomes . Educ. Psychol.
37 , 550 –564 . doi: 10.1080/01443410.2015.1127330

Cortazzi M. Jin L. (1996 ). “Cultures of learning: language classrooms in China,” in Society and the Language Classroom , ed. Coleman H. (Cambridge: Cambridge University Press), 169 –206 .

Cui H. Mahfoodh O. H. A. (2026 ). An intercultural framework for assessing sociopragmatic competence in English language education: evidence from the Chinese EFL context . Int. J. Appl. Linguist.
36 , 577 –592 . doi: 10.1111/ijal.12800

Datu J. A. D. Yang W. (2021 ). Academic buoyancy, academic motivation, and academic achievement among Filipino high school students . Curr. Psychol.
40 , 3958 –3965 . doi: 10.1007/s12144-019-00358-y

Dewaele J. M. (2019 ). The effect of classroom emotions, attitudes toward English, and teacher behavior on willingness to communicate among English foreign language learners . J. Lang. Soc. Psychol.
38 , 523 –535 . doi: 10.1177/0261927X19864996

Dewaele J. M. Botes E. Greiff S. (2023 ). Sources and effects of foreign language enjoyment, anxiety, and boredom: a structural equation modeling approach . Stud. Second. Lang. Acquis.
45 , 461 –479 . doi: 10.1017/S0272263122000328

Dewaele J. M. MacIntyre P. (2024 ). “You can’t start a fire without a spark”. Enjoyment, anxiety, and the emergence of flow in foreign language classrooms . Appl. Linguist. Rev.
15 , 403 –426 . doi: 10.1515/applirev-2021-0123

Ducker N. (2021 ). Protecting and enhancing willingness to communicate with idiodynamic peer-peer strategy sharing . System
103 :102634 . doi: 10.1016/j.system.2021.102634

Fredrickson B. L. (2004 ). The broaden–and–build theory of positive emotions . Philos. Trans. R. Soc. Lond. Ser. B Biol. Sci.
359 , 1367 –78 . doi: 10.1098/rstb.2004.1512 , 15347528 PMC1693418

Hair J. Black W. Babin B. Anderson R. (2010 ). Multivariate data Analysis: A Global Perspective . Upper Saddle River: Pearson.

Han H. Li B. (2025 ). Perceptions of social interactions and second language willingness to communicate in different activities . Asia-Pac. J. Second Foreign Lang. Educ.
10 :4 . doi: 10.1186/s40862-024-00312-x

Hayes A. F. (2013 ). Introduction to Mediation, Moderation, and Conditional Process Analysis . New York: The Guilford Press.

Hejazi S. Y. Sadoughi M. Peng J.-E. (2023 ). The structural relationship between teacher support and willingness to communicate: the mediation of L2 anxiety and the moderation of growth language mindset . J. Psycholinguist. Res.
52 , 2955 –2978 . doi: 10.1007/s10936-023-10026-9 , 37935808

Hobfoll S. E. (1989 ). Conservation of resources: a new attempt at conceptualizing stress . Am. Psychol.
44 , 513 –524 . doi: 10.1037/0003-066X.44.3.513 , 2648906

Hu L. T. Bentler P. M. (1999 ). Cutoff criteria for fit indexes in covariance structure analysis: conventional criteria versus new alternatives . Struct. Equ. Model.
6 , 1 –55 . doi: 10.1080/10705519909540118

Ibrahim K. A. A. A. Basim N. (2024 ). Applying task supported language learning via Instagram: the effects on EFL learners' self-esteem, academic buoyancy, academic engagement, and language achievement . Asia-Pac. J. Second Foreign Lang. Educ.
9 :2 . doi: 10.1186/s40862-023-00223-3

Jamalvandi B. Jafarigohar M. Jalilifar A. Soleimani H. (2020 ). Mediation and EFL learners’ willingness to communicate: a micro sociocultural study . Appl. Res. Engl. Lang.
9 , 449 –476 . doi: 10.22108/are.2020.117283.1465

Khajavy G. H. MacIntyre P. D. Barabadi E. (2018 ). Role of the emotions and classroom environment in willingness to communicate: applying doubly latent multilevel analysis in second language acquisition research . Stud. Second. Lang. Acquis.
40 , 605 –624 . doi: 10.1017/S0272263117000304

Kline R. B. (2023 ). Principles and Practice of Structural Equation Modeling . New York: Guilford Publications.

Lee J. S. (2022 ). The role of grit and classroom enjoyment in EFL learners’ willingness to communicate . J. Multiling. Multicult. Dev.
43 , 452 –468 . doi: 10.1080/01434632.2020.1746319

Leeming P. Vitta J. P. Hiver P. Hicks D. McLean S. Nicklin C. (2024 ). Willingness to communicate, speaking self-efficacy, and perceived communicative competence as predictors of second language spoken task production . Lang. Learn.
74 , 917 –949 . doi: 10.1111/lang.12640

Li D. Ahmad N. A. Roslan S. (2024 ). Is social support the key to academic buoyancy? A review of the relationship between perceived social support and academic buoyancy . Int. J. Acad. Res. Progr. Educ. Dev.
13 , 1179 –1187 . doi: 10.6007/IJARPED/v13-i2/21520

Li C. Dewaele J.-M. Pawlak M. Kruk M. (2025 ). Classroom environment and willingness to communicate in English: the mediating role of emotions experienced by university students in China . Lang. Teach. Res.
29 , 2140 –2160 . doi: 10.1177/13621688221111623

Li C. Huang J. Li B. (2021 ). The predictive effects of classroom environment and trait emotional intelligence on foreign language enjoyment and anxiety . System
96 :102393 . doi: 10.1016/j.system.2020.102393

Li J. Jia S. Wang L. Zhang M. Chen S. (2023 ). Relationships among inferiority feelings, fear of negative evaluation, and social anxiety in Chinese junior high school students . Front. Psychol.
13 :1015477 . doi: 10.3389/fpsyg.2022.1015477 , 36704691 PMC9872515

Li Z. Li B. Wang X. Zhen L. (2025 ). Clustering perceived learning environment as antecedents of willingness to communicate in a second language . Lang. Teach. Res.
29 , 2777 –2794 . doi: 10.1177/13621688221122611

Li X. Liu H. Xia M. (2025 ). Exploring the role of students’ perceived EFL teacher support in academic burnout: the mediating effect of academic buoyancy . Humanit. Soc. Sci. Commun.
12 :1203 . doi: 10.1057/s41599-025-05589-y

Lin Y. Wang B. Zhang G. Su Y. (2025 ). The relations among foreign language anxiety, academic buoyancy and willingness to communicate in EFL classroom . Front. Psychol.
16 :1634054 . doi: 10.3389/fpsyg.2025.1634054 , 41229600 PMC12602425

Liu H. Elahi Shirvan M. Taherian T. (2024 ). Revisiting the relationship between global and specific levels of foreign language boredom and language engagement: a moderated mediation model of academic buoyancy and emotional engagement . Stud. Second Lang. Learn. Teach.
15 , 13 –39 . doi: 10.14746/ssllt.40195

Liu L. Guan W. J. Qiu Y. Lee J. S. (2024 ). Effects of extramural English activities on willingness to communicate: the role of teacher support for Chinese EFL students . System
124 :103319 . doi: 10.1016/j.system.2024.103319

Liu M. Hong M. (2021 ). English language classroom anxiety and enjoyment in Chinese young learners . SAGE Open
11 :21582440211047550 . doi: 10.1177/21582440211047550

Ma L. Xia J. (2026 ). Teacher-student relationships as a buffer: mitigating anxiety's impact on willingness to communicate across genders . Int. J. Appl. Linguist.
36 , 498 –508 . doi: 10.1111/ijal.12792

MacIntyre P. D. (1994 ). Variables underlying willingness to communicate: a causal analysis . Commun. Res. Rep.
11 , 135 –142 . doi: 10.1080/08824099409359951

MacIntyre P. D. Baker S. C. Clément R. Conrod S. (2001 ). Willingness to communicate, social support, and language-learning orientations of immersion students . Stud. Second. Lang. Acquis.
23 , 369 –388 . doi: 10.1017/S0272263101003035

MacIntyre P. D. Baker S. C. Clément R. Donovan L. A. (2002 ). Sex and age effects on willingness to communicate, anxiety, perceived competence, and L2 motivation among junior high school French immersion students . Lang. Learn.
52 , 537 –564 . doi: 10.1111/1467-9922.00194

MacIntyre P. D. Clément R. Dörnyei Z. Noels K. A. (1998 ). Conceptualizing willingness to communicate in a L2: a situational model of L2 confidence and affiliation . Mod. Lang. J.
82 , 545 –562 . doi: 10.1111/j.1540-4781.1998.tb05543.x

MacIntyre P. D. Gregersen T. (2022 ). The idiodynamic method: willingness to communicate and anxiety processes interacting in real time . Int. Rev. Appl. Linguist. Lang. Teach.
60 , 67 –84 . doi: 10.1515/iral-2021-0024

MacIntyre P. D. Gregersen T. Mercer S. (2019a ). Setting an agenda for positive psychology in SLA: theory, practice, and research . Mod. Lang. J.
103 , 262 –274 . doi: 10.1111/modl.12544

MacIntyre P. D. Legatto J. J. (2011 ). A dynamic system approach to willingness to communicate: developing an idiodynamic method to capture rapidly changing affect . Appl. Linguist.
32 , 149 –171 . doi: 10.1093/applin/amq037

MacIntyre P. D. Ross J. Sparling H. (2019b ). Flow experiences and willingness to communicate: connecting Scottish Gaelic language and traditional music . J. Lang. Soc. Psychol.
38 , 536 –545 . doi: 10.1177/0261927X19867364

Martin A. J. Marsh H. W. (2008 ). Academic buoyancy: towards an understanding of students' everyday academic resilience . J. Sch. Psychol.
46 , 53 –83 . doi: 10.1016/j.jsp.2007.01.002 , 19083351

Martin A. J. Marsh H. W. (2009 ). Academic resilience and academic buoyancy: multidimensional and hierarchical conceptual framing of causes, correlates and cognate constructs . Oxf. Rev. Educ.
35 , 353 –370 . doi: 10.1080/03054980902934639

McCroskey J. C. Baer J. E. (1985 ). Willingness to communicate: the construct and its measurement. Annual Meeting of the Speech Communication Association, (ERIC), p. 11 .

Owusu-Agyeman Y. Mugume T. (2023 ). Academic adjustment of first year students and their transition experiences: the moderating effect of social adjustment . Tert. Educ. Manag.
29 , 189 –209 . doi: 10.1007/s11233-023-09120-3

Peng J. E. (2012 ). Towards an ecological understanding of willingness to communicate in EFL classrooms in China . System
40 , 203 –213 . doi: 10.1016/j.system.2012.02.002

Peng J. E. (2020 ). Teacher interaction strategies and situated willingness to communicate . ELT J.
74 , 307 –317 . doi: 10.1093/elt/ccaa012

Peng J. E. (2025 ). Willingness to Communicate in a Second Language . Cambridge: Cambridge University Press.

Peng J. E. Woodrow L. (2010 ). Willingness to communicate in English: a model in the Chinese EFL classroom context: willingness to communicate in English . Lang. Learn.
60 , 834 –876 . doi: 10.1111/j.1467-9922.2010.00576.x

Poudel A. Gurung B. Khanal G. P. (2020 ). Perceived social support and psychological wellbeing among Nepalese adolescents: the mediating role of self-esteem . BMC Psychol.
8 :43 . doi: 10.1186/s40359-020-00409-1 , 32357919 PMC7195718

Putwain D. W. Gallard D. Beaumont J. (2020 ). Academic buoyancy protects achievement against minor academic adversities . Learn. Individ. Differ.
83-84 :101936 . doi: 10.1016/j.lindif.2020.101936

Putwain D. W. Wood P. (2023 ). Riding the bumps in mathematics learning: relations between academic buoyancy, engagement, and achievement . Learn. Instr.
83 :101691 . doi: 10.1016/j.learninstruc.2022.101691

Shen X. Shen X. Peng J. E. (2024 ). Digital storytelling: a literacy-building tool to promote willingness to communicate in a second language . RELC J.
55 , 819 –826 . doi: 10.1177/00336882231157461

Souzandehfar M. Ahmed Abdel-Al Ibrahim K. (2023 ). Task-supported language instruction in an EFL context: impacts on academic buoyancy, self-esteem, creativity, and language achievement . Asia-Pac. J. Second Foreign Lang. Educ.
8 :43 . doi: 10.1186/s40862-023-00218-0

Terenzini P. T. Rendon L. I. Lee Upcraft M. Millar S. B. Allison K. W. Gregg P. L. . (1994 ). The transition to college: diverse students, diverse stories . Res. High. Educ. 35 , 57 –73 . doi: 10.1007/BF02496662

Tu X. (2021 ). The role of classroom culture and psychological safety in EFL students' engagement . Front. Psychol.
12 :760903 . doi: 10.3389/fpsyg.2021.760903 , 34659072 PMC8511621

Wang X. Hui L. (2024 ). Buoyancy and engagement in online English learning: the mediating roles and complex interactions of anxiety, enjoyment, and boredom . System
125 :103418 . doi: 10.1016/j.system.2024.103418

Wang H. Peng A. Patterson M. M. (2021 ). The roles of class social climate, language mindset, and emotions in predicting willingness to communicate in a foreign language . System
99 :102529 . doi: 10.1016/j.system.2021.102529

Wang C. Teng M. F. Liu S. (2023 ). Psychosocial profiles of university students’ emotional adjustment, perceived social support, self-efficacy belief, and foreign language anxiety during COVID-19 . Educ. Dev. Psychol.
40 , 51 –62 . doi: 10.1080/20590776.2021.2012085

Wang F. Yu X. Pan X. (2026 ). L2 willingness to communicate in the context of online learning environments: a systematic review . Behav. Sci.
16 :229 . doi: 10.3390/bs16020229 , 41750038 PMC12938372

Wen W. P. Clément R. (2003 ). A Chinese conceptualisation of willingness to communicate in ESL . Lang. Cult. Curric.
16 , 18 –38 . doi: 10.1080/07908310308666654

Xiao Z. Jia J. (2025 ). The relationship between college students’ emotional intelligence, foreign language enjoyment, and L2 willingness to communicate: a variable-centered and person-centered perspective . Behav. Sci.
15 :1508 . doi: 10.3390/bs15111508 , 41301310 PMC12649340

Xu X. Wang B. (2022 ). EFL students’ academic buoyancy: does academic motivation and interest matter?
Front. Psychol.
13 :858054 . doi: 10.3389/fpsyg.2022.858054 , 35369193 PMC8968443

Xu J. Zheng Y. (2025 ). Does student assessment literacy matter between motivational constructs and engagement in L2 writing? A survey of Chinese EFL undergraduates . Assess. Writing
64 :100916 . doi: 10.1016/j.asw.2025.100916

Yang S. Azari Noughabi M. Jahedizadeh S. (2024 ). Modelling the contribution of English language learners’ academic buoyancy and self-efficacy to L2 grit: evidence from Iran and China . J. Multiling. Multicult. Dev.
45 , 2701 –2717 . doi: 10.1080/01434632.2022.2062368

Yang Y. Cui Y. Yao S. (2024 ). Teacher support, grit and L2 willingness to communicate: the mediating effect of foreign language enjoyment . BMC Psychol.
12 :383 . doi: 10.1186/s40359-024-01877-5 , 38982544 PMC11232161

Yang H. Lian Z. (2023 ). Ideal L2 self, self-efficacy, and pragmatic production: the mediating role of willingness to communicate in learning English as a foreign language . Behav. Sci.
13 :597 . doi: 10.3390/bs13070597 , 37504043 PMC10375955

Yang Y. Liang S. (2025 ). Classroom social climate and student engagement in English as a foreign language learning: the mediating roles of academic buoyancy and academic emotions . Asia Pac. Educ. Res.
34 , 1123 –1132 . doi: 10.1007/s40299-024-00926-2

Yang X. Xue M. Pauen S. He H. (2024 ). Psychometric properties of the Chinese version of multidimensional scale of perceived social support . Psychol. Res. Behav. Manag.
17 , 2233 –2241 . doi: 10.2147/PRBM.S463245 , 38835653 PMC11149633

Yun S. Hiver P. Al-Hoorie A. H. (2018 ). Academic buoyancy: exploring learners’ everyday resilience in the language classroom . Stud. Second. Lang. Acquis.
40 , 805 –830 . doi: 10.1017/S0272263118000037

Zarrinabadi N. Khodarahmi E. (2024 ). Some antecedents of directed motivational currents in a foreign language . J. Multiling. Multicult. Dev.
45 , 973 –986 . doi: 10.1080/01434632.2021.1931250

Zhai K. (2025 ). Exploring the relationships between academic buoyancy, engagement, and achievement in English reading among EFL learners . SAGE Open
15 :21582440251357156 . doi: 10.1177/21582440251357156

Zhang L. J. Saeedian A. Fathi J. (2024 ). Testing a model of growth mindset, ideal L2 self, boredom, and WTC in an EFL context . J. Multiling. Multicult. Dev.
45 , 3450 –3465 . doi: 10.1080/01434632.2022.2100893

Zhou L. Xi Y. Lochtman K. (2023 ). The relationship between second language competence and willingness to communicate: the moderating effect of foreign language anxiety . J. Multiling. Multicult. Dev.
44 , 129 –143 . doi: 10.1080/01434632.2020.1801697

Zimet G. D. Dahlem N. W. Zimet S. G. Farley G. K. (1988 ). The multidimensional scale of perceived social support . J. Pers. Assess.
52 , 30 –41 . doi: 10.1207/s15327752jpa5201_2 2280326

Come citare questo studio

Zhu X, Mohd-Said NE, Shamsudin CBM. (2026). Perceived social support, academic buoyancy, and L2 willingness to communicate among Chinese EFL undergraduates: a structural equation modeling study.. Frontiers in psychology. DOI: 10.3389/fpsyg.2026.1920700

Consulta la pubblicazione originale ↗