Conflicts, Villains, Resolutions: Towards models of Narrative Media Framing

Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics Volume 1: Long Papers, pages 8712–8732 July 9-14, 2023 ©2023 Association for Computational Linguistics Conflicts, Villains, Resolutions: Towards models of Narrative Media Framing Lea Frermann1 Jiatong Li2 Shima Khanehzar1,3 Gosia Mikolajczak4 1School of Computing and Information Systems, The University of Melbourne 2Department of Computing, The Hong Kong Polytechnic University 3CSIRO Data61 4The Global Institute for Women’s Leadership, The Australian National University lfrermann@unimelb.edu.au jiatong.li@connect.polyu.hk shima.khanehzar@data61.csiro.au Gosia.Mikolajczak@anu.edu.au Abstract Despite increasing interest in the automatic de- tection of media frames in NLP, the problem is typically simplified as single-label classifi- cation and adopts a topic-like view on frames, evading modelling the broader document-level narrative. In this work, we revisit a widely used conceptualization of framing from the commu- nication sciences which explicitly captures el- ements of narratives, including conflict and its resolution, and integrate it with the narrative framing of key entities in the story as heroes, victims or villains. We adapt an effective an- notation paradigm that breaks a complex an- notation task into a series of simpler binary questions, and present an annotated data set of English news articles, and a case study on the framing of climate change in articles from news outlets across the political spectrum. Fi- nally, we explore automatic multi-label predic- tion of our frames with supervised and semi- supervised approaches, and present a novel retrieval-based method which is both effective and transparent in its predictions. We con- clude with a discussion of opportunities and challenges for future work on document-level models of narrative framing.1 1 Introduction Media discourse around contested issues is often biased by experiences or interests of the news outlets and different stakeholders they give voice to. News framing by the media has been formal- ized and examined on many levels in communi- cation and social sciences, ranging from selec- tion of information (Levitt, 1981) over discourse- centric (Pan and Kosicki, 1993) and entity focused approaches (Lawlor and Tolley, 2017). While a growing body of work in NLP attempts to auto- matically detect framing in the news or social me- dia, most work adopts well-defined yet oversim- plifying approaches like topic modeling or simple 1Code and data are available at https://github.com/ phenixace/narrative-framing. Resolution: solution/alleviation of the issue Conflict: disagreements between individuals, groups, insti- tutions, countries, etc. Human Interest: emotionalization and dramatization of an issue through the lens of affected individuals Moral: moral or religious references Economic: economic consequences for individuals, groups, institutions, countries, etc. Hero: entity contributing to/responsible for issue resolution Villain: entity contributing to/responsible for issue cause Victim: entity suffering the consequences of an issue Table 1: Five frames (top) and three narrative roles (bottom) considered in this paper. classifiers (see Ali and Hassan (2022) for a recent review); formalize the task as single-label classi- fication ignoring co-existence and interactions of different frames, and focus on localized emphasis frames rather than the full story (Card et al., 2015; Field et al., 2018; Khanehzar et al., 2021). This paper addresses the above …

Content truncated to 500 words