Helping university students develop productive views of what characterizes good scientific research and scientific knowledge is an important objective of science education. However, many studies show that students’ views of the nature of science (NOS) do not become more informed or even become less informed after learning activities that engage students in scientific practice. This could mean that the way we teach students to conduct scientific research might unintentionally strengthen or bring about uninformed views of NOS. In this article, we argue that scientific discourse (how language is used in science) might play a role in this relation between learning scientific inquiry and uninformed views of NOS. We argue that there could be implicit notions relating to NOS underlying scientific discourse that uninformed readers might translate to uninformed views of NOS. We call these implicit notions underlying scientific language use “epistemological notions.” In this article, we further define this construct of epistemological notions, contrast it with explicit views of NOS and other related constructs, explain how we think epistemological notions might affect explicit views of NOS, and present a framework we developed to characterize them in scientific language use. It is a descriptive and interpretative analysis framework which combines, optimizes, and extends several text analysis methods, discourse analysis, and reflexive thematic analysis. We provide a guide to use the framework and point out quality criteria. We finish by advocating the framework for educational researchers interested in developing instructional interventions during which learning about science is combined with explicit reflection on NOS. Identifying epistemological notions in scientific language use could provide starting points for these activities by making explicit the translation from NOS understanding to scientific practice and vice versa.