Circle Singing as an Inclusive Educational Practice for English Pronunciation in Higher Education

<title>Abstract</title> <p> <bold>Purpose</bold> This article presents the research design and results of a study examining the effects of Circle Singing on English pronunciation accuracy in higher education. The broader research project investigated Circle Singing as an inclusive and interdisciplinary methodology aimed at developing aural perception skills, English pronunciation, socio-relational competences, and wellbeing. The present article focuses specifically on English pronunciation outcomes, while the remaining dimensions are mentioned only insofar as they help contextualise or interpret the primary linguistic findings. <bold>Design</bold> A quasi-experimental study was conducted at Roma Tre University between October 2025 and May 2026. The sample comprised 188 university students assigned to an Experimental Group (n = 104), which attended a 15-hour Circle Singing laboratory, and a Control Group (n = 84), which did not receive the intervention. English pronunciation was assessed before and after the intervention through SpeechAce, an Automatic Speech Recognition (ASR) platform (Nguyen et al., 2025). Aural perception was assessed through the Mini-PROMS (Profile of Music Perception Skills; Law &amp; Zentner, 2012) and was considered as a complementary variable within the broader research design. Quantitative data were analysed using 2 × 2 repeated-measures mixed ANOVA, with Time (pre-test, post-test) as the within-subject factor and Group (Experimental Group, Control Group) as the between-subject factor. The Time × Group interaction was considered the main inferential test of the intervention effect. Internal consistency and temporal stability were examined through Cronbach’s alpha and the Intraclass Correlation Coefficient (ICC), respectively. Questionnaires and focus groups were used to provide complementary qualitative evidence on participants’ perceptions of engagement, confidence, and socio-relational experience. <bold>Findings</bold> The SpeechAce scale demonstrated excellent internal reliability (α = .965), with high item-total correlations (.900–.925) and satisfactory test-retest stability in the Control Group (ICC = .852, 95% CI [.802, .894]). Descriptive statistics revealed contrasting trends across groups: the Control Group showed a slight decline from pre-test to post-test (M = 78.830 to M = 77.676), whereas the Experimental Group demonstrated a substantial improvement (M = 82.986 to M = 89.498). A preliminary comparison of pre-test scores indicated that the two groups were not fully equivalent at baseline (EG: M = 82.986; CG: M = 78.830), a difference that should be taken into account when interpreting between-group comparisons. The repeated-measures mixed ANOVA revealed a significant main effect of Time, F(1,186) = 58.530, p &lt; .001, η²p = .239, a significant main effect of Group, F(1,186) = 35.070, p &lt; .001, η²p = .159, and — crucially — a significant Group × Time interaction, F(1,186) = 119.872, p &lt; .001, η²p = .392, indicating that the two groups changed differentially over time. Within-group paired comparisons confirmed a significant improvement in the Experimental Group, t(103) = 10.638, p &lt; .001, and a slight but statistically significant decline in the Control Group, t(83) = − 6.441, p &lt; .001. Questionnaire and focus group feedback suggested increased confidence in English pronunciation, high engagement, and positive perceptions of the educational value of the intervention. <bold>Contribution</bold> The study provides empirical evidence that Circle Singing may constitute an effective music-based educational intervention for supporting English pronunciation in higher education. The findings contribute to the growing body of research on the relationship between musical training and language learning while offering a replicable educational model for interdisciplinary language instruction. Main text </p>