Integrating ChatGPT into University English Teaching: Linguistic Performance, Autonomy, and Learning Experience

Authors

  • Jeany Argueta Universidad Gerardo Barrios, San Miguel, El Salvador
  • Gisela Espinoza Universidad Gerardo Barrios, San Miguel, El Salvador
  • Víctor Sandoval Universidad Gerardo Barrios, San Miguel, El Salvador

DOI:

https://doi.org/10.34190/ecel.25.1.5506

Keywords:

TOEFL ITP, artificial intelligence, higher education, learner autonomy, second language acquisition

Abstract

This study examines how the planned integration of ChatGPT influences English language development, student autonomy, and the overall learning experience in a university English as a Foreign Language (EFL) Intermediate English II course in El Salvador. The aim is to analyze both quantifiable changes in language proficiency and students' interactions with artificial intelligence as a learning support tool, particularly in relation to self-regulation and independent learning processes. Furthermore, the study sought to understand how students perceive the usefulness and limitations of AI in academic contexts. A convergent mixed-methods case study design was implemented with twenty-three undergraduate students enrolled in an Intermediate English II course. Quantitative data were collected through TOEFL ITP simulation tests administered before and after the intervention to assess changes in language proficiency. Qualitative data were collected through a focus group of four participants, where students shared their perceptions, attitudes, and experiences interacting with ChatGPT during the learning process. Both datasets were analyzed separately and then integrated to provide a comprehensive interpretation of the findings. The quantitative results reveal a positive, though not statistically significant, trend in TOEFL ITP scores after the intervention. While quantifiable improvements in test performance were limited, the qualitative findings offer a clearer understanding of the learning process. Students reported using ChatGPT as an accessible and ongoing learning partner to practice language skills, clarify doubts, generate examples, and receive immediate feedback. This consistent interaction contributed to increased confidence, greater autonomy, and the development of self-regulated learning strategies. However, participants also expressed concerns about potential over-reliance on AI and occasional inaccuracies in the generated responses, underscoring the importance of maintaining critical awareness. This standalone study highlights the pedagogical potential of AI tools when intentionally integrated into language teaching and underscores the need to foster critical AI literacy so that such technologies enhance, rather than hinder, meaningful learning.

Author Biographies

Jeany Argueta, Universidad Gerardo Barrios, San Miguel, El Salvador

Master's in Social Research Methodology specializing in English language learning, educational technology, and qualitative research. Research interests explore EFL learning, generative AI in education, learner autonomy, and innovative pedagogical practices, with particular attention to the Salvadoran higher education context.

Gisela Espinoza, Universidad Gerardo Barrios, San Miguel, El Salvador

PhD in Research and Teaching, with a Master's in Web Engineering and a degree in Computer Systems and Network Engineering. Higher education professor and researcher with experience in technological innovation. Her research focuses on gender equity and women in STEM, Smart Campus initiatives, digital solutions, and information systems development.

Víctor Sandoval, Universidad Gerardo Barrios, San Miguel, El Salvador

Mathematician with a Master's in Strategic Management and a PhD in Project Management. Extensive experience as a project manager, scientific researcher, and higher education professor. Advanced skills in data analysis, software development, predictive modeling, and process optimization. Oriented towards academic innovation and the implementation of technological solutions based on business intelligence and operations research.

Downloads

Published

2026-10-07