Following qualitative researchers’ growing interest in reflexivity, a body of scholarship has emerged that aims to turn informal practices for reflexivity into methods that can be learnt and taught alongside other research practices. This literature, however, has focused on helping researchers become more reflexive toward their situatedness and positionality, rather than toward their use of language and its effects on knowledge production – a process I refer to as ‘linguistic reflexivity’. This article addresses this gap by formalising a method for ‘problematising categories’, an informal approach familiar to qualitative researchers as a promising solution to the analytical and ethical blinders that result from scholars’ unconscious use of language. I proceed in three steps. First, I review the literature to show the analytical, empirical and ethical rationales behind this approach and offer a definition of problematising categories as the practice of making conscious how socio-linguistic units of categorisation unconsciously organise our perception and can represent a problem for knowledge production. This practice, I argue, enables us to decentre ourselves from the taken-for-granted nature of those categories. Second, I develop a three-stage research method for problematising categories: noticing ‘critical junctures' when problematisation is called for, identifying the categorical problem through sensitising questions and reconstructing an alternative. Third, I demonstrate how problematising categories contributes to the research process by applying this method to my experience in problematising the binary pair ‘local’ versus ‘international’ in a research project on the environmental impact of Chinese investment in the Senegalese fishery sector. I show that problematising categories leads to more rigorous empirical findings and nuanced analysis in a way that is feasible within the frame of qualitative research projects. Overall, this article expands the practical tools for linguistic reflexivity and heeds the methodological call to make conscious and explicit choices for every dimension of our research.