A Survey of Computational Framing Analysis Approaches

Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, pages 9335 - 9348 December 7-11, 2022 ©2022 Association for Computational Linguistics A Survey of Computational Framing Analysis Approaches Mohammad Ali College of Information Studies University of Maryland, College Park mali24@umd.edu Naeemul Hassan Philip Merrill College of Journalism College of Information Studies University of Maryland, College Park nhassan@umd.edu Abstract Framing analysis is predominantly qualitative and quantitative, examining a small dataset with manual coding. Easy access to digital data in the last two decades prompts scholars in both computation and social sciences to utilize vari- ous computational methods to explore frames in large-scale datasets. The growing scholar- ship, however, lacks a comprehensive under- standing and resources of computational fram- ing analysis methods. Aiming to address the gap, this article surveys existing computational framing analysis approaches and puts them to- gether. The research is expected to help schol- ars and journalists gain a deeper understanding of how frames are being explored computation- ally, better equip them to analyze frames in large-scale datasets, and, finally, work on ad- vancing methodological approaches. 1 Introduction Vaccine hesitancy has long been recognized as a problem despite research evidence favoring the vac- cine’s effectiveness (Sallam, 2021). Understand- ing how the vaccination is framed by news media might provide a solution to vaccine hesitancy be- cause a frame determines “how [people] evaluate [a problem] and choose to act upon it” (Entman, 1993, p. 54). Like this, exploration of many other problems (e.g., gun violence) warrants analysis of frames, especially in large-scale datasets in this era. Traditionally, researchers explore frames using qualitative and quantitative methods that require manual labor and can handle small amounts of data (D’angelo, 2018; Reese et al., 2001). Production of and easy access to large volumes of digital data in the last two decades prompt scholars to harness the exploration of frames in such big data computa- tionally (Card et al., 2015; Liu et al., 2019; Walter and Ophir, 2019; van Atteveldt and Peng, 2018). Prior studies proposed various computational methods (e.g., topic modeling and neural network). As the scholarship is growing, a scarcity appeared regarding a comprehensive understanding and re- sources of computational framing analysis methods (Nicholls and Culpepper, 2021; Sanfilippo et al., 2008). Researchers might be confused with multi- ple approaches to this analysis, raising questions: how many computational framing analysis methods exist, and which one they should apply? To address the problem and help researchers with such questions, we survey existing computational framing analysis approaches and put the methods and relevant resources together. As such, the survey is guided by the following three research questions: RQ1. What computational methods do researchers use to explore frames in large-scale datasets? RQ2. How do researchers conceptualize a frame in computational framing analysis studies? RQ3. How do researchers use computational meth- ods in exploring frames? The primary contributions of this article are: a) it provides a comprehensive understanding and re- sources of existing computational framing analy- sis methods and puts them together for interested …

Content truncated to 500 words