-
Title
-
Measured, Not Heard: AI-Generated Feedback and Year 6 Girls’ Self-Perception as Public Speakers
-
Author
-
James Porter
-
Year Published
-
2026
-
Description
-
This action research study explored how Year 6 girls interpret and respond to AI-generated feedback on public speaking, and what impact this had on their positive self-perception as speakers. Conducted over a thirteen-week autumn term in a London girls’ junior school, the inquiry was embedded within an existing oracy curriculum and centred on three speaking tasks supported by Microsoft Speaker Progress. A mixed-methods design combined three-timepoint self-perception questionnaires, AI-generated metric reports and stored recordings, think-aloud protocols during rehearsal, focus groups conducted before, during, and after the intervention, teacher field notes, and reflexive journalling. Rather than producing a straightforward narrative of confidence gain or loss, the findings present a more complex and ethically significant picture. AI feedback did not operate as a discrete intervention acting uniformly upon pupils; instead, it was encountered within a wider feedback ecology in which it interacted with teacher feedback, peer responses, prior experience, emotional safety, and pupils’ own developing self-judgements. Within this ecology, AI feedback sometimes supported performance, sometimes reduced emotional risk, and sometimes introduced uncertainty, meaning its influence on self-perception was partial, conditional, and relational. Importantly, the quantitative and qualitative strands did not converge neatly: questionnaire items linked to understanding, preparation, and knowing how to improve showed clearer movement than affective items relating to confidence and nervousness. Read as analytically productive, this divergence helps explain why clarity and competence did not reliably translate into confidence, and why shifts in self-perception were not always captured by quantitative measures alone.
-
Tags
-
Technology, Teaching and Learning
