Teachers need to learn complex skills in higher education, such as diagnostic argumentation. We suggest that relations between the argumentation facets justification, disconfirmation, and transparency are a relevant indicator for the quality of diagnostic argumentation. In an experimental study, we investigated whether automatic adaptive feedback – based on natural language processing – compared to static feedback facilitates relations between the argumentation facets in preservice teachers' diagnostic argumentation when learning with case-based simulations. A sample of N = 60 preservice teachers received adaptive or static feedback on their written explanations concerning simulated cases of pupils having behavioral or reading and writing problems. Using Epistemic Network Analysis, we analyzed learners' written explanations and found that adaptive feedback compared to static feedback facilitates relations between justification, disconfirmation, and transparency in preservice teachers' diagnostic argumentation. The results confirm that adaptivity is an important feature of effective feedback, which can be automated by methods of natural language processing.