Purpose: The purpose of this paper is to highlight some methodological problems concerning the neglect of participants' voices by workplace ethnographers and neglect of the highly interactional and co-constructive nature of research interviewing. The study aims to use discourse analysis, to show the phenomena of workplace learning and expertise to be constituted in participants' talk. Design/methodology/approach: From excerpts of natural talk and research interviews by fish culturists speaking about their learning in a salmon hatchery, discourse analysis is used to analyze how workplace learning and expertise are rhetorically performed. Findings: The paper finds that fish culturists drew on two discursive repertoires/resources--school- and workplace-based learning--to account for their learning and expertise. The main participant affirmed the primacy of interest and practical workplace experience in his job just as he presupposed a weak correlation between school-based (theoretical) and workplace (practical) knowing. However, both kinds of learning were deemed important though articulating this view depended on the social contexts of its production. Research limitations/implications: Discourse analysis does not establish immutable truths about workplace learning and expertise but rather it is used to understand how these are made accountable through talk in real-time, that is, how the phenomenon is "done" by participants. Practical implications: There is increased sensitivity when using ethnographic and interview methods. No method can avoid being theory-laden in its conduct and reporting but discourse analysis perhaps does it better than its alternatives. Originality/value: While some contributors to this journal have also approached workplace learning from a discursive perspective, this paper attempts to understand the phenomenon solely from participants' categories and interpretations.