Digital technology has significantly disrupted media systems, forcing media organizations to adapt to a changing business environment and audience behavior. In particular, the political coverage of traditional news organizations, online news, and political talk online are increasingly intertwined and competing for attention. This emerging assemblage of diverse sources, potentially diverging coverage and publication logics, and different temporal rhythms is known as the “hybrid media system”. Still, whether in this process the internet is challenging, reinforcing, or solely mirroring existing power structures remains unclear.We will examine which of these claims is true by empirically analyzing the dynamics of communicative power in political discourse. We see communicative power as the ability to influence political discourse by determining which frames—argumentative figures identifying political or social problems, sources, and remedies—are given relative prominence in public discourse. Thus, the analysis of the emergence, competition, and dominance of frames in political coverage on different platforms will provide evidence on the structure and dynamics of communicative power in the hybrid media system.Over the course of two years, we will continuously and automatically collect political coverage and political talk online on selected news sites and platforms in four countries with varying media systems and political cultures—Brazil, Germany, South Korea, and the USA. In parallel, we will work on automatically identifying topics and frames in the coverage of selected events by using sophisticated extensions of Latent Dirichlet Allocation models, keyword-based vector models, and syntax-aware techniques employing language agnostic methods whenever possible. We will qualitatively evaluate these models by comparing our results to human coded manually labeled frames. This automated identification of frames, validated through human coders, will let us answer if new sources of political coverage and talk online indeed diverge from political coverage on traditional news sites in the relative prominence of frames. The use of time series analysis techniques will further allow us to identify temporal interdependencies between the prominence of frames across platforms. Our comparative design will ultimately allow us to assess the robustness of our findings given different contextual factors.Our contributions will fall in three areas: (1) providing comparative evidence on communicative power in contemporary media systems; (2) providing new evidence on the impact of digital technology by empirically testing theory-driven expectations during a new stage of the internet’s development and adoption; (3) testing and developing advanced methods in automatic content analysis across languages.This project will show if and how the internet has influenced or disrupted communicative power and if it has done so to different degrees in different contexts. Our findings will be highly relevant to policy makers, journalists, researchers, and the general public. The nature of this proposal requires the combination of conceptual, methodological, and interpretative skills of social and computer scientists and contextual knowledge of the cultural and political backgrounds of the selected countries. This makes it an excellent fit for interdisciplinary research and a promising project to showcase the potential of computational social science.