Motivation has played a fundamental role in second language acquisition (SLA) and has remained a central topic since the 1970s. Among the various theories, Gardner’s (1985, 2007) socio-educational model of motivation and Dörnyei’s (2009) L2 Motivational Self System have received significant attention. However, both have faced criticism in recent years. Specifically, the former has been criticized for its overreliance on integrative and instrumental motivation, with limited grounding in educational psychology or empirical evidence (Oga-Baldwin et al., 2019). The latter model combines three key components: cognition, affect, and motivation. It integrates cognition, affect, and motivation, linking dynamic motivation with L2 learners’ identity. However, the model suffers from serious methodological flaws, primarily due to a lack of basic statistical rigor (Al-Hoorie, 2018). This limitation undermines the model’s ability to reliably predict language proficiency (Nagle, 2021).

To address these shortcomings, the EVT (Eccles and Wigfield, 2020) has recently been widely applied in the field of SLA as a compelling alternative. Originally proposed by Atkinson (1957) and developed by Wigfield and Eccles (2000) and Eccles and Wigfield (2020), this model suggests that an individual’s motivation to perform an academic task rests on two psychological factors: the expectation of success and the values associated with the task (Dong et al., 2022). Expectations of success refer to an individual’s perceived competence in performing various learning tasks in specific domains, extending across past, present, and future contexts. Meanwhile, task values represent the subjective reasons that drive a learner to choose and persist in a particular learning task or activity, and these values consist of four subtypes: achievement value, intrinsic value, utility value, and cost value. Achievement value relates to the importance of performing well on a task or activity and is tied to personal goals; meanwhile, intrinsic value is reflected in the internal gratifications (such as enjoyment and rewards) that the individual gains from engaging in and completing a task or activity. As for utility value, it refers to the individual’s assessment of the usefulness after completing the task or activity, considering current or future goals. Lastly, cost value refers to the individual’s evaluation of the amount of effort, opportunities, and emotional burden required to complete a task or activity.

Several empirical studies (e.g., Loh, 2019; Wan, 2019; Nagle, 2021; Dong et al., 2022) have revealed that this model can be used to explain and predict L2 students’ choices, effort, persistence, and academic performance, establishing a positive relationship between expectations and language proficiency. It should be noted, however, that research on EVT remains limited and has largely focused on the context of English as a foreign language. Thus, this observation needs to be verified with further empirical evidence in the context of other languages (Eccles and Wigfield, 2020).

In addition to motivational constructs, recent research has emphasized affective variables in SLA, particularly language anxiety. Language anxiety refers to “the feeling of tension and apprehension specifically associated with second language contexts, including speaking, listening, and learning” (MacIntyre and Gardner, 1994, p. 284). Rubio (2004) noted that this emotional state can arise from physiological, cognitive, and assertive sources. Given that this specific anxiety primarily occurs in institutional settings, Horwitz et al. (1986) defined language classroom anxiety as “a distinct complex of self-perceptions, beliefs, feelings, and behaviors related to classroom learning arising from the uniqueness of the language learning process” (p. 128). This definition has become the most widely used and well-known theory of language anxiety in SLA research.

These authors classify language anxiety into three specific categories: communication apprehension, test anxiety, and fear of negative evaluation. Subsequent studies have identified additional dimensions that reflect the evolving complexity of language learning environments. Young (1991) emphasized the role of teacher–student interaction, while Gregersen et al. (2014) highlighted classroom competition as an important source of anxiety. Other research has pointed to communicative confidence (Park, 2014) and fear of failure (Toyama and Yamazaki, 2019) as further contributing factors. Moreover, FLCAS (Horwitz et al., 1986) has become the most reliable and widely used instrument for measuring foreign language learners’ anxiety in classroom settings (Teimouri et al., 2018). Building on this scale, Moneypenny and Aldrich (2022) developed a Spanish-as-a-foreign-language anxiety scale by adapting the Arabic Foreign Language Anxiety Questionnaire (Dewaele and Al-Saraj, 2015) for effective use in the ELE context.

Empirical research has consistently shown that language anxiety exerts a negative influence on students’ academic performance in the foreign language classroom (Horwitz, 2017; MacIntyre, 2017). It is also inversely correlated with linguistic proficiency (Aida, 1994; Woodrow, 2006; Shao et al., 2019; Zhang, 2019), as well as other individual differences, such as motivation and enjoyment (Dewaele and Li, 2022; Jiang and Dewaele, 2020; Lou and Noels, 2020).

Complementing cognitive and affective approaches, the social dimension of language learning has gained attention through research on social identity. In SLA, social identity refers to how learners understand their relationship to the world through language. From a sociocultural perspective, social identity has arguably been the most extensively studied subject in SLA (Darvin and Norton, 2015; Almarza et al., 2012). It shapes how learners perceive themselves and their place in society as they acquire a new language. Lambert (1967) suggests that developing linguistic competence in both the first and second languages is associated with changes in the social identity of L2 learners. Building on this foundation, Schumann (1986) introduced the acculturation model, which emphasized social distance as the crucial factor. He argued that reducing this social distance, through greater contact with and identification with the L2 culture, facilitates language acquisition.

Meanwhile, Norton (2013) further advanced the conceptualization of social identity by framing it as the dynamic relationship among the individual, language, and society. She defined it as “how a person understands his or her relationship to the world, how that relationship is structured across time and space, and how the person understands possibilities for the future” (p. 45). In SLA, this means that L2 learners must invest in their second language to understand their social position within the L2 community—a position that may or may not be deemed legitimate by L2 speakers—to subsequently speak, read, and write in that language (Norton Peirce, 1995).

Although early studies focused on immigrant contexts, social identity has been adopted in language education settings (Norton and Gao, 2008). In this context, Luo (2019) introduced the concept of identity change, based on a bilingual orientation (e.g., Gao et al., 2005). This model posits that identity change in L2 learners is a byproduct of the L2 learning process and is reflected in the development of their L1 proficiency. Luo identified four types of identity change: self-confidence (where L2 proficiency enhances learner confidence); subtractive change (where L2 learning diminishes L1 identity); additive change (where L2 strengthens L1 identity); and productive change (where both L1 and L2 identities are mutually enriched). These changes are often reciprocal and may extend to other personal and academic dimensions. This framework is particularly relevant for the current study, as it avoids the immigrant bias of prior theories and captures academic-specific identity shifts (e.g., exam performance pressures). Furthermore, it aligns with China’s utilitarian approach to foreign language learning (Dong et al., 2022).

In recent years, the number of Spanish Philology departments in Chinese universities and the total number of students studying Spanish as a foreign language have reached 105 and 34,823, respectively (Instituto Cervantes, 2024). Compared to the year 2000, these figures reflect a nearly sixfold increase in the number of institutions and more than a twentyfold rise in student enrollment over a 25-year period (Lu, 2023). Despite this significant expansion, there remains a lack of empirical research addressing individual differences among Chinese university students of Spanish. Consequently, it is crucial to analyze the language anxiety, motivation, and social identity of Chinese students of Spanish to explore the potential effects of these factors on their academic performance in Spanish. Therefore, the present study aims to answer the following research questions:

Four instruments were administered via Survey Tencent, adapted for cultural relevance:

The study was conducted in late October 2023, with 421 respondents completing the questionnaire, which consisted of 53 items across 4 dimensions. The questionnaire was administered online through a Chinese survey platform Survey Tencent. After reviewing the data, 317 questionnaires were deemed valid, yielding a response rate of 74.6%.

Data analysis was conducted using SPSS v.26. Several statistical analysis methods were applied, including calculations of internal consistency (Cronbach’s alpha), descriptive statistics for the four study variables (mean and standard deviation), correlation analyses among these variables, and multiple and hierarchical linear regression models. It should be noted that the use of the latter regression analysis requires meeting additional assumptions. Accordingly, following Field et al. (2012), we employed the Durbin–Watson test (p > 0.05), variance inflation factor (VIF < 4), Q–Q plot, and Cook’s distance (< 1) to verify the independence of errors, check for multicollinearity among predictors, assess the normality of residuals, and identify outliers and influential cases, respectively. Unless otherwise indicated, all assumptions were met.

Participation in this study was entirely voluntary and anonymous. No personally identifiable information, such as names, email addresses, or phone numbers, was collected. To further ensure participant privacy, technical identifiers—including IP addresses and device information—were neither recorded nor stored, thereby minimizing the risk of indirect identification. The survey platform was configured to prevent the capture of such metadata. All data were collected and stored in encrypted formats on secure servers, accessible only to the principal investigators. The data are used exclusively for academic research purposes and will not be disclosed to third parties. This study was designed and conducted in accordance with the ethical guidelines of the University of Cordoba and Wenzhou Business College. These measures collectively ensured the confidentiality and integrity of the research process.

As shown in Table 1, the sample consisted of 317 Chinese undergraduate students enrolled in Hispanic Philology programs. The gender distribution was unbalanced, with 248 female students (78.2%) and 69 male students (21.8%). Participants ranged in age from 17 to 25 years, with the largest proportion aged 20 (n = 69, 21.8%), followed by 22 (n = 68, 21.5%) and 21 (n = 59, 18.6%). In terms of ethnicity, the majority identified as Han (n = 307, 96.8%), while a small minority (n = 10, 3.2%) belonged to ethnic minority groups. Regarding the year of study, the most represented group was second-year students (n = 119, 37.5%), followed by third-year (n = 93, 29.3%), fourth-year (n = 67, 21.1%), and first-year students (n = 38, 12%). Regarding students’ place of origin, the distribution was relatively balanced across five categories, with the highest proportion from PSGZ regions (n = 83, 26.2%). With respect to the location of their university, most were studying in provincial capitals (n = 148, 46.7%) or in PSGS pilot zones (n = 102, 32.2%). The average Spanish proficiency grade was 75.6 out of 100 (SD = 12.02), with 60 being the minimum passing score. Grade distribution was centered, with 74.1% of students (n = 235) scoring between 65 and 85, and 25.9% (n = 82) falling outside that range.

Table 2 presents the means, standard deviations, and correlations for the four variables of the study. Regarding language anxiety in ELE, the majority of participants (n = 253, 80.0%) exhibited a low level of anxiety (total scores between 12 and 53), and only 7 students (2.2%) reported a high level of anxiety. Commutatively, nearly 98% of the participants in this study experienced a moderate–low level of language anxiety in the ELE classroom (M = 3.37, SD = 1.36). In regards to identity change, self-confidence (M = 4.62, SD = 1.08) was highest, followed by additive change (M = 4.58, SD = 1.05), productive change (M = 4.45, SD = 1.05), and finally subtractive change (M = 4.16, SD = 0.95). For Expectancy–Value Motivation, all value-related dimensions were above the mean of expectancy (M = 4.63, SD = 1.12). Utility value (M = 5.36, SD = 1.16) was the highest among the four value subtypes (achievement value: M = 5.03, SD = 1.21; intrinsic value: M = 4.98, SD = 1.13; cost value: M = 4.72, SD = 1.30).

As mentioned in the literature on language anxiety in SLA, language anxiety was found to be significantly and negatively related (r = −0.15, p < 0.01) to the average grade. On the other hand, correlation analyses showed that values and expectancy had a significant (p < 0.01) positive correlation with academic performance in Spanish: r = 0.44 with achievement value, r = 0.43 with intrinsic value, r = 0.38 with utility value, r = 0.34 with expectancy, and r = 0.22 with cost value. Similarly, the various types of identity change were positively intercorrelated, with additive change (r = 0.36), self-confidence (r = 0.30), subtractive change (r = 0.27), and productive change (r = 0.24), respectively.

A multiple and hierarchical linear regression analysis was employed to verify the predictable effects of the three individual differences (language anxiety, identity change, and Expectancy–Value Motivation) of the participants on their academic performance in Spanish. To this end, the means of the three types of Expectancy–Value Motivation, the three types of identity change, and language anxiety were used as predictor variables (requiring inter-correlations among them), while the average grade was used as the dependent variable. In the process of this analysis, three models were generated in Table 3.

In the first model (M1), self-confidence, subtractive change, and additive change were used to explain Spanish achievement. The model explained 11% of the variance (R² = 0.11, F = 13.17, p < 0.01), indicating a small to moderate overall effect size. Among these variables, self-confidence was the only significant predictor (β = 0.20; p < 0.01), suggesting that learners who felt more confident in their L2 identity tended to achieve higher grades in Spanish. According to Cohen’s (1988) benchmarks, this represents a moderate effect. Subtractive change and additive change did not reach statistical significance.

In the second model (M2), three EVT dimensions (achievement value, intrinsic value, and cost value) were incorporated as independent variables. This step significantly improved the model fit, increasing the explained variance to 24% (ΔR² = 0.13, F = 15.92, p < 0.01), which constitutes a moderate effect size. Both achievement value (β = 0.26, p < 0.01) and intrinsic value (β = 0.18, p < 0.05) significantly predicted Spanish achievement, indicating that students who highly valued success or derived enjoyment from Spanish learning were more likely to perform well. Self-confidence remained marginally significant (β = 0.11, p = 0.01), suggesting a slight reduction in its effect after introducing motivational factors.

Finally, in the third model (M3), language anxiety was added to examine its additional contribution. In M3, the total variance explained increased slightly to 25% (ΔR² = 0.01, F = 14.60, p < 0.01) and this change represented a small but statistically meaningful improvement. Language anxiety was a significant negative predictor (β = −0.13, p < 0.05), which showed that students with higher levels of Spanish-related anxiety tended to perform worse academically. Although the effect size was small, it held practical significance, as emotional factors are often negatively associated with language learning outcomes. In the same model, achievement value (β = 0.25, p < 0.01) and self-confidence (β = 0.14, p < 0.05) continued to show significant and positive effects on Spanish achievement. These results confirm the meaningful contributions of social identity, EVT, and language anxiety to students’ Spanish achievement.