Fifty Shades of Political Authority: a Large-Scale Study of Reported Speech Framing in Political News

Quotes are a fundamental component of political news. To establish the legitimacy of their narratives, political journalists anchor reporting into relayed facts and discourse coming from pertinent sources within the political landscape. Through repeated quotation, more than mere mentioning, political actors are accredited as relevant and legitimate voices with respect to the context under consideration (Carlson 2009; Fishman 1980; Sigal 1986). Yet, sustained inclusion in news narratives through quotes may not be sufficient to signal political-discursive authority when it is shaded by persisting stereotypes in quote framing (de Courson et al. 2024). Such authority is necessary for political actors to not merely populate political news, but to define it (Hall et al. 1978). This paper contributes to the literature on authority signaling (Chadwick et al. 2018; Ecker-Ehrhardt 2010) by nuancing dynamics of political-discursive authority attribution in the news. We argue that while quotes are important accreditation signals in themselves, reporting verbs are powerful framing devices. In choosing them, journalists may cast different shades of political authoritativeness (or lack thereof) over politicians and their discourse. Political actors may be quoted substantially, but mostly with verbs suggesting unreliability or levity (distancing verbs, or expressions of uncontrolled emotion); or they might be quoted less, but through verbs signaling epistemic and political reliability (drawn from the register of analysis and measured deliberation). Furthermore, both quote volume and framing may vary across issue areas, time, and outlets. The aim of this paper is to analyze how quote volume and framing come together to ascribe specific shades of authority to political actors across the partisan spectrum. We follow a mixed-methods research design. We leverage an innovative method to automatically extract quotes attributed to political personalities from a large-scale dataset of French newspapers, spanning three decades and four newsrooms, totalling more than 2M extracted quotes (Bonutti D’Agostini et al., under review). Our method detects triplets of source, reporting verb, and quoted statement, allowing to examine variations in quote volume, topic, and framing. We also conducted 30 semi-structured interviews with French political journalists to contextualize the signaling value of quotes and reporting verbs, and to identify the semantic categories news practitioners consider most meaningful. This helps make sense of the wide variety of verbs in such an extensive dataset, telling signal from apparent noise. While stylistic imperatives account for some variation of verbs, journalists acknowledge that semantic categories are chosen in accordance with the character of the statement and its source. Through manual and LLM-assisted text classification, we analyze how specific categories of reporting verbs come to be associated with political actors, across issue areas, time, and outlets. We combine this with measures of quote volumes to qualify parties’ media-ascribed political authority. For mainstream actors, differential authority signaling across aligned (opposed) newsrooms mostly translates in relatively higher (lower) quote volumes. For radical parties, differences lie less in volume than in authority attribution or denial through quote framing. Finally, despite a trajectory from “discursive segregation” to generalized quotation, unreliability signals through framing remain higher for radical parties.