Large Language Models and Reflexivity in Qualitative Research: Introducing the Hybrid Reflexivity Framework

Reflexivity is central to qualitative research. It shapes how researchers interpret data, examine positionality, engage with participants, and justify analytic decisions. In this article, I introduce hybrid reflexivity, a framework in which reflective practice is structured through researcher guided interaction with generative large language models (LLMs). Building on FINLAY's (2002) typology, I consider algorithmically mediated prompting as a structured analytic intervention to make inferential movements explicit and open to systematic comparison. Within this framework, LLMs function as provisional interlocutors whose patterned responses can generate counter-readings, surface tacit assumptions, and amplify deficit framings for inspection. Interpretive responsibility remains with the researcher who evaluates generated extensions against primary transcripts and theoretical commitments. Drawing on examples from language teacher identity research, I demonstrate how LLM-mediated exchanges can stabilize, redirect, or intensify discursive cues, prompting closer scrutiny of escalation, affect, and authority. I conclude by outlining methodological and ethical considerations and position hybrid reflexivity as a documented and inspectable extension of reflexive practice to support analytic transparency without displacing human judgment.