Unwilling Coalitions? Coverage and Framing of Government Conflict in German Newspaper

Coalition governments are standard in many parliamentary systems. While parties forming a government coalition act together as one political institution, party competition doesn’t simply end with the adoption of the coalition agreement (Korte et al., 2006). Working within a coalition government is therefore a continual balancing act between differing political ideologies, party interests, and the public’s expectation of a unified institutional voice. Simultaneously, potential conflicts within a government are highly newsworthy and can therefore expect a great deal of media attention (Harcup & O’Neill, 2017). Consequently, media outlets may choose to highlight the presence of conflicts by emphasizing “conflict between individuals, groups, or institutions as a means of capturing audience interest” (Semetko & Valkenburg, 2000, p. 95), termed conflict frames (see for example Bartholomé et al., 2015, 2018; Schuck et al., 2016; Van Der Goot et al., 2024) when reporting on intra-governmental conflict. In the German private and print media system, media pluralism is conceived mainly in terms of intermedia rather than intramedia diversity. Consequently, quality newspapers often reflect specific ideological leanings (Eilders, 2002), potentially resulting in a slanted portrayal of the current government. Media bias can be conceptualized as systematic differences in reporting based on ideological slant and can be comprised of visibility bias, describing disproportionate levels of coverage devoted to a given topic, and tonality bias, referring to the slanted framing or evaluation of that topic (Bartholomé et al., 2015, 2018; Schuck et al., 2016; Van Der Goot et al., 2024). Applied to conflict framing in government coverage, this can manifest either in an overrepresentation of conflict frames or in an unequal tone in the evaluation of governmental conflicts. This study therefore investigates: Do German newspaper exhibit visibility and tonality bias when reporting on intragovernmental conflicts, depending on the ideological positioning of the government? To investigate these research questions, we collected a total of around 1.5 million articles from six different German news outlets ranging from 1990 to 2024, among them three conservative and two left-liberal newspapers and a public broadcaster. We filtered our sample using Named Entity Recognition, validating entities against curated lists of German cities and high-ranking politicians. Due to the complexity of the task, Large Language Models were used to parse the text logic for conflict frames and their actors involved, followed by a BERT- based sentiment analysis to assess tonality bias against media outlets and ideologies of the coalition. Preliminary results, tested on a sample of around 30.000 articles, suggest differences between the outlets regarding visibility bias. Especially BILD, a conservative tabloid newspaper, reports significantly more on intra governmental conflicts within left leaning cabinets. We cannot identify specific differences in tonality, however, due to the specific language used in news reporting, improvements are being considered to our methodological approach in order to better capture such nuances. However, given that the usage of conflict frames might be biased, distorted representation of intra governmental conflicts may emerge based on the consumption of certain media outlets and formats.