The article extends the study of frames in verbal media discourse. We mobilize insights from linguistic semantics and research in the related fields of cognitive science in order to formulate a frame-semantic understanding of frames as adaptive networks of meaning. It allows us to see frames as flexible scaffoldings whose elements are controlled by contextual configurations. This extension is helpful, we argue, because analyses of public discourse have, to date, mainly operated with a model of frames as fixed ensembles. Understanding frames not as invariant clusters but as adaptive networks has implications for empirical studies, too. Consequently, we outline the applicability of our proposition in an analytical scenario.