Artificial Intelligence Use and Foreign Language Anxiety Reduction in University EFL Students at ESPOCH

Abstract

Introduction: This study examined the extent to which the use of artificial intelligence tools, or AI, is associated with a reduction in language anxiety among students at Escuela Superior Politécnica de Chimborazo (ESPOCH), in Riobamba, Ecuador. At this institution English is a compulsory subject in students’ academic training, regardless of their chosen degree program. Based on data collected through the CUIAILE questionnaire, which explores the use of artificial intelligence, language anxiety, and autonomy, the study aimed to identify which variables of the instrument had an independent predictive effect on this perceived reduction. The interest behind this research came from the observation that, in academic contexts, learning English is not only about developing skills in a foreign language, but also about facing anxiety, fear of making mistakes, and insecurity when using the language. Method: A quantitative, descriptive, correlational, and crosssectional study was conducted with a sample of 164 participants, of whom 64.6% were women. The mean age was 22.6 years, with a standard deviation of 9.4. Data were collected using the CUIA-ILE instrument, composed of 58 items distributed across 7 dimensions and structured on a Likert scale from 1 to 5. The questionnaire was administered through a self-developed web application, available at https:// nicolaldesantiago.com/cuiaIle/cuia-ile.html, to ESPOCH English students during the March–July 2026 academic period. Data collection took place between March and April of the same year. The use of a digital application made it possible to collect students’ responses in a direct and organized way, within a learning context where artificial intelligence is becoming increasingly present. One-way ANOVA analyses with Tukey post hoc comparisons, Pearson and Spearman correlations, descriptive statistics, and simple and multiple linear regression models were calculated. In this process, the analysis of frequencies, numbers, associations, and coefficients helped identify patterns that were not immediately visible in the first reading of the data. Language anxiety, therefore, was not treated only as an individual emotion, but also as a process partly shaped by pedagogical, symbolic, and technological conditions. Results: A higher use of AI was associated with lower perceived anxiety among students. The correlation coefficient was 0.812, with a 95% confidence interval between 0.753 and 0.859, R² = 0.660, and p < .001. In the multiple regression model, USOIA presented the highest standardized coefficient, β = +0.687, p < .001. The perception of the teacher’s role, or PEROL, also contributed to the model, β = +0.195, p = .012. The adjusted R² was 0.696. In addition, the frequency of AI use showed significant differences in the scores of perceived anxiety reduction, F = 5.516, df = 3 and 160, p = .001, η² = 0.094. Conclusions: The results show that the use of AI had the strongest predictive weight in the reduction of language anxiety among university students of English as a foreign language. At the same time, the perception of the teacher’s role had a positive contribution to this relationship. This suggests that instructional design, rather than the restriction of technology, may be one of the most useful areas for pedagogical intervention in current language teaching processes. In this context, AI should not be understood only as a simple technical tool, but also as a support space where students can practice, correct themselves, hesitate, and rebuild their voice before facing the uncertainty of academic oral communication.