Publikation Nr. 5072 - Details
SammelbandOpen Access · CCPeer ReviewEnglish
When the AI gets an F: The affordances of (mediocre) automated writing corrective feedback
Blume, C. C. & Bukmakoswki, L. (2025). When the AI gets an F: The affordances of (mediocre) automated writing corrective feedback. In: Choubsaz, Y., Díez-Arcón, P., Gimeno-Sanz, A., Morgana, V., Murphy, A. C. & Seracini, F. L. (Hrsg.), Advancing CALL: New research agendas – EUROCALL 2025 short papers (S. 579-589). Valencia: Editorial Universitat Politècnica de València.
Abstract
While existing research suggests that automatic written feedback (AWF) is often positively perceived and may contribute to skill development, little is known about the interplay of AWF with generative AI tools. Yet analyses of AWF realized through generative AI need to take place within the context of the given learning situation, the learners’ knowledge and expectations, as well as a tool’s particular affordances – especially when those affordances fall short of expectations surrounding AI. This contribution presents findings from three consecutive studies of an AI-driven application designed specifically for giving formative feedback on various text types in the secondary EFL classroom in Germany. The analysis highlights the benefits as well as the challenges for EFL teachers and learners in light of their knowledge and expectations on the one hand and the tool’s affordances on the other. Triangulated data is derived from quasi-experimental tool analysis, audio recordings of learners’ reception and implementation of the feedback, and participant observation of educators’ responses to the data from the previous two studies. Combined, the studies provide insight into how learners and educators negotiate issues of quality and utility relevant for this group of technologies.

