Speaker
Description
This study theorizes how L2 speaking self-efficacy is reconfigured when generative AI functions as a peer within a posthuman framework of distributed agency. Drawing on Bandura’s four sources, a qualitative case study of Japanese pre-entry university students examined how efficacy formation shifted from individual to relational processes in AI-mediated interaction. Findings support a redistribution of efficacy formation, contributing to a posthuman reconceptualization of L2 self-efficacy.
Abstract section 1: Relevance
As generative AI becomes embedded in L2 education, it is typically framed as a pedagogical tool that supports, rather than co-constitutes, learner agency. This anthropocentric view contrasts with posthumanist perspectives, which conceptualize agency as distributed across human–nonhuman relations within sociomaterial assemblages (Barad, 2007). Bandura’s (1997) model of self-efficacy likewise assumes that mastery, modeling, persuasion, and affect regulation arise through human interaction. How these sources are reconfigured when AI functions as a peer remains under-theorized. This study reconceptualizes L2 self-efficacy as a relational construct within AI-mediated activity systems, informing human-AI collaborative L2 pedagogy.
Abstract section 4: Outcomes/results
Findings suggest that AI-mediated peer interaction did not simply increase L2 speaking self-efficacy but reshaped its underlying structure. Mastery shifted from individual accomplishment to co-constructed achievement, and verbal persuasion centered on AI-generated feedback. Learners described their speaking competence as emerging within human-AI interaction rather than residing solely within themselves. Relational co-construction was linked to greater confidence and reduced anxiety. In transitional pre-entry contexts, this shift toward relational efficacy formation may support adaptation to collaborative university learning. These findings indicate a redistribution of efficacy formation, offering pedagogical implications for human-AI collaborative L2 environments.
Abstract section 2: Contribution/research questions
This study explores how L2 speaking self-efficacy is reconfigured within a case of AI-mediated peer interaction. It addresses three questions: (1) How does AI-mediated peer interaction reshape learners’ efficacy-related experiences? (2) In what ways are Bandura’s four sources structurally reconfigured within this activity system? (3) How is perceived co-construction with AI associated with learners’ speaking confidence in this context?
Abstract section 5: References
Bandura, A. (1997). Self-efficacy: The exercise of control. W. H. Freeman and Company.
Barad, K. M. (2007). Meeting the universe halfway: quantum physics and the entanglement of matter and meaning. Duke University Press.
Timmermans, S., & Tavory, I. (2012). Theory Construction in Qualitative Research: From Grounded Theory to Abductive Analysis. Sociological Theory, 30(3), 167–186. https://doi.org/10.1177/0735275112457914
Abstract section 3: Content/method
This qualitative case study examined AI-mediated peer speaking among Japanese pre-entry university students in a transitional classroom context. Data comprised a self-efficacy questionnaire reflecting Bandura’s four sources, a relational co-construction item, and reflective narratives capturing learners’ experiences. An abductive analytic framework (Tavory & Timmermans, 2012) was employed to interpret how efficacy formation was reconfigured within human-AI interaction. Descriptive comparisons and thematic analysis were integrated, and methodological triangulation enhanced analytic rigor.
| Title | Reconfiguring L2 Self-Efficacy in Posthuman AI Learning |
|---|---|
| Teaching Context | College and university education |