Artificial intelligence and teacher mediation in English listening instruction: an integrative review

Abstract

The incorporation of artificial intelligence into English listening instruction created a need to examine not only the tools but also the pedagogical intervention that organized their use. This integrative review synthesized evidence on listening outcomes and teacher mediation. Twenty-four full-text files retrieved through structured searches and citation chaining through July 28, 2026, were examined. After two duplicate files, one from each of two pairs, were removed, twenty-two unique reports were assessed: eleven listening-outcome studies formed the core corpus, four provided contextual evidence on mediation, five reviews supported interpretation, and two reports were excluded. Core-study quality was evaluated with design-specific criteria, and findings were integrated through thematic comparison. Several structured programs reported improvements in listening comprehension, aural decoding, or affective variables; however, two comparisons found no significant between-group listening advantage, and heterogeneity prevented attribution of the effects solely to technology. Five teacher functions were interpreted: orchestration, scaffolding, distribution of responsibilities, affective-relational support, and verification. Mediation appeared as a recurrent pedagogical condition, although its independent causal effect has not yet been established.