The study, published in Discover Education, was conducted by Hossein Isaee and Hamed Barjesteh of Islamic Azad University in Iran and Neda Fatehi Rad of Islamic Azad University’s Kerman branch. It examined a question that has received surprisingly little empirical attention: what happens when two of the most talked-about trends in language education, artificial intelligence and positive psychology, are integrated within a single instructional design rather than studied separately? Most previous research, the authors note, has either celebrated the technical affordances of AI or documented the emotional benefits of strengths-based pedagogy, leaving the interaction between the two largely unexplored, especially outside Western and East Asian settings.

To fill that gap, the research team recruited 300 undergraduate English as a Foreign Language learners aged 18 to 24 from twelve universities across Iran, Oman, and Türkiye, with four institutions from each country. Twelve intact classes, each taught by a different instructor, were assigned through cluster randomization to one of three conditions: AI-supported instruction enriched with positive psychology activities, AI-supported instruction alone, or traditional exam-focused teaching. The intervention ran for ten weeks, with two 90-minute sessions per week, and was standardized through a common implementation booklet, a centralized training workshop in Tehran for lead instructors, and follow-up online coordination sessions across all sites.

The AI component relied on ChatGPT, used as a guided pedagogical support tool rather than an autonomous teacher. Learners practiced conversation, vocabulary, and writing with the system, receiving immediate feedback on their language production. In the integrated condition, these tasks were interleaved with positive psychology exercises grounded in Seligman’s PERMA model of well-being, which identifies positive emotion, engagement, relationships, meaning, and accomplishment as pillars of human flourishing. Students kept gratitude journals, wrote reflectively about their strengths, set goals, and discussed their progress in class, while teachers framed feedback not only around linguistic accuracy but around effort, persistence, and resilience. The theoretical logic drew on Fredrickson’s broaden-and-build theory, which holds that positive emotions expand attentional and cognitive resources, making learners more persistent, more willing to take communicative risks, and more engaged.

Quantitatively, the results favored the integrated approach across every major measure. The researchers assessed language proficiency with an institutional examination modeled on TOEFL- and IELTS-style tasks, with productive skills rated by trained raters who were blind to group assignment. Proficiency rose by nearly 20 points in the AI-plus-positive-psychology group, compared with roughly 12 points in the AI-only group and about 7 points in the control group. Analyses of covariance confirmed that the group differences remained significant after controlling for baseline performance, and the AI-only group also outperformed traditional instruction, indicating that the technology itself contributed something real even without the emotional layer.

The affective outcomes told a strikingly parallel story. Learners completed the Foreign Language Enjoyment Scale, the PERMA-Profiler short form for well-being, and the Academic Buoyancy Scale, all carefully translated and back-translated into Persian, Arabic, and Turkish before use. The integrated group showed the largest increases in enjoyment and well-being, and a particularly pronounced advantage in academic buoyancy, the capacity to bounce back from everyday academic setbacks such as exam stress and corrective feedback. A repeated-measures analysis showed that enjoyment grew over time in all groups but grew far more sharply where positive psychology activities were present, suggesting that AI alone delivers only a modest emotional benefit.

Perhaps the most theoretically consequential finding came from the mediation analysis. Foreign language enjoyment partially explained the relationship between instructional condition and language achievement: the integrated program boosted enjoyment, and higher enjoyment in turn predicted stronger proficiency outcomes. Because the direct effect remained significant, emotion did not account for everything, but the pattern supports the idea that enjoyment functions as a proximal affective mechanism linking instructional design to achievement, rather than being a mere byproduct of successful learning. In exam-driven systems where language learning is often associated with pressure and anxiety, that mechanism may matter more than anywhere else.

The qualitative strand of the study, based on interviews with 27 students and all 12 instructors, weekly teacher journals, and classroom observations coded inductively with intercoder agreement above 85 percent, illuminated why the intervention worked. Three themes dominated. First, learners described the AI system as a supportive, patient, and nonjudgmental coach: one Omani student said the system corrected her calmly and that she stopped fearing mistakes, while an instructor in Türkiye reported that normally quiet students began participating once the fear of public correction was lifted. Second, positive psychology activities reframed the emotional climate of the classroom, with learners saying they began noticing small improvements instead of fixating on errors and looking forward to sessions rather than dreading them. Third, cultural adaptation and teacher mediation proved central: students initially found some reflection activities unfamiliar, even questioning whether they counted as serious academic work, and grew receptive only when teachers connected the tasks to locally meaningful values such as perseverance, self-improvement, and communication.

The authors are careful about causal language. Because intact classes rather than individuals were assigned to conditions, and each class had a single instructor who could not be rotated, the design is a cluster-based quasi-experiment, and the findings are framed as evidence of association rather than definitive causal proof. The AI-only condition controlled for exposure to the platform but not for the extra instructional variety and relational attention embedded in the psychology activities, and self-reported affective measures remain more vulnerable to expectancy effects than the blind-rated proficiency scores. The ten-week duration leaves open questions about whether the gains persist, and the sample of twelve universities in general English courses cannot represent the Middle East’s considerable diversity.

Even with those caveats, the study carries a message that is likely to resonate far beyond the three countries where it was conducted. Artificial intelligence in education is usually evaluated on technical sophistication, adaptivity, and efficiency. This research argues that such metrics miss the point: learners experienced the AI system as a relational and emotionally mediated presence, and its effectiveness depended on teacher mediation, cultural responsiveness, and the deliberate cultivation of positive emotion. Language programs, the authors suggest, should move beyond using AI merely for correction and drill, pairing AI tasks with brief reflective or strengths-based activities, and training teachers to interpret AI feedback in emotionally supportive terms. Future research will need active control conditions matched for time-on-task, longer follow-up periods, and possibly AI systems designed from the ground up around affective and motivational principles. But the headline finding is already clear: the smartest educational technology may be the kind that knows when to let human warmth lead.

Subject of Research: Integrating positive psychology principles with AI-supported English as a Foreign Language instruction in higher education

Article Title: A human-centered approach to integrating positive psychology and artificial intelligence in EFL instruction

Article References: Isaee, H., Barjesteh, H., & Rad, N. F. (2026). A human-centered approach to integrating positive psychology and artificial intelligence in EFL instruction. Discover Education, 5(1), Article 955. https://doi.org/10.1007/s44217-026-02151-z

Image Credits: AI Generated

DOI: 10.1007/s44217-026-02151-z

Keywords: artificial intelligence in education, positive psychology, EFL pedagogy, foreign language enjoyment, human-centered AI, technology-enhanced language learning, PERMA model, academic buoyancy, teacher mediation, ChatGPT, language proficiency, learner well-being