This study examined the associations among emotional intelligence (EI), grit, motivation, and second language (L2) learning performance, with particular attention to the mediating roles of grit and motivation. A sample of 801 Chinese university students completed validated measures of EI, grit, and motivation, and their English test scores were collected as an indicator of performance. Using structural equation modeling (SEM), EI was positively associated with grit (β = 0.574, p < .001) and motivation (β = 0.426, p < .001), while grit was also positively related to motivation (β = 0.515, p < .001). In relation to performance, significant positive associations were observed with motivation (β = 0.635, p < .001), EI (β = 0.185, p < .001), and grit (β = 0.165, p < .001). Bootstrap analyses confirmed robust indirect associations of EI with performance through grit (β = 0.095), through motivation (β = 0.271), and via the sequential chain EI → grit → motivation → performance (β = 0.188). The model accounted for 33% of the variance in grit, 55% in motivation, and 79% in performance. These findings indicate that EI, grit, and motivation are closely interrelated constructs that jointly contribute to L2 performance, highlighting the importance of socio-emotional and motivational resources in second language learning.

This study investigates the moderating effect of language attitude on the relationships among social media addiction, social pain, and internet trolling among college students. A sample of 891 students from various colleges and universities completed validated measures assessing their levels of social media addiction, social pain, internet trolling, and language attitude. Using a latent variable approach within a multigroup structural equation modeling (SEM) framework, participants were divided into three groups (high, medium, and low language attitude) based on their language attitude scores. The SEM analysis revealed distinct patterns across groups. In the high language attitude group, both social media addiction and social pain significantly predicted internet trolling, with standardized regression coefficients of 0.564 and 0.728, respectively. In the medium language attitude group, the predictive effects remained significant; however, the magnitude of the coefficients decreased markedly (0.264 for social media addiction and 0.562 for social pain). In contrast, in the low language attitude group, neither social media addiction nor social pain emerged as significant predictors of internet trolling. Interestingly, the covariance between social media addiction and social pain remained consistent across the three groups, suggesting a stable interrelationship irrespective of language attitude level. These findings imply that language attitude plays a crucial moderating role in the interplay among social media addiction, social pain, and internet trolling. Specifically, higher levels of language attitude appear to amplify the effects of social media addiction and social pain on internet trolling behavior, while lower levels attenuate these associations. The results underscore the importance of considering individual differences in language attitudes when developing intervention strategies aimed at mitigating problematic online behaviors among college students.

Accurate alignment of items to content standards is critical for valid score interpretation in large‐scale assessments. This study evaluates three automated paradigms for aligning Math items with four domain labels and nineteen skill labels. First, we extracted instruction‐tuned embeddings, trained a suite of classical supervised machine‐learning models, and examined whether principal component analysis (PCA) improved performance. Second, we fine‐tuned a set of encoder‐only language models (11 BERT‐family models) for domain and skill alignment. Third, we explored ensemble learning via majority voting and stacking with multiple meta‐models. Results show that DeBERTa‐v3‐base achieved the best weighted‐average F 1 score for domain alignment (0.950), whereas RoBERTa‐large achieved the best weighted‐average F 1 score for skill alignment (0.869). PCA substantially improved linear classifiers based on embeddings but did not outperform fine‐tuned encoder‐only language models. Ensemble methods did not surpass the best single encoder‐only language model. Overall, the study provides a systematic comparison of embedding‐based classification, fine‐tuned encoder‐only language models, and ensemble strategies for automated item‐to‐standards alignment.