The study focusses on identifying specific systemic factors affecting the potential and, hence, also the realized and current productivity of two Latvian derivational models using corpus linguistic methods. Both models are used to produce action nominals from verbs and roughly correspond to the opposition between event and result nominalizations, although there is some semantic overlap. While being semantically and functionally similar, the two models show markedly different degrees of productivity. The study comprises four steps: 1) extracting all unique derivatives (lemmas) corresponding to both models from the Balanced Corpus of Modern Latvian and linking them to their base verbs, 2) annotating the extracted set of derivatives and base verbs for various potentially significant properties, 3) defining the occurring vs. non-occurring base verbs in terms of minimally sufficient base verb property values and value combinations (precluding factors), and 4) calculating base verb property values and value combinations occurring significantly more vs. less frequently than others (productivity-enhancing and hindering factors).