Weakly Supervised Learning of Nuanced Frames for Analyzing Polarization in News Media

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing, pages 7698–7716, November 16–20, 2020. c⃝2020 Association for Computational Linguistics 7698 Weakly Supervised Learning of Nuanced Frames for Analyzing Polarization in News Media Shamik Roy Purdue University roy98@purdue.edu Dan Goldwasser Purdue University dgoldwas@purdue.edu Abstract In this paper, we suggest a minimally- supervised approach for identifying nuanced frames in news article coverage of politically divisive topics. We suggest to break the broad policy frames suggested by Boydstun et al., 2014 into fine-grained subframes which can capture differences in political ideology in a better way. We evaluate the suggested subframes and their embedding, learned us- ing minimal supervision, over three topics, namely, immigration, gun-control, and abor- tion. We demonstrate the ability of the sub- frames to capture ideological differences and analyze political discourse in news media. 1 Introduction As the political climate and the news media in the United States become increasingly polar- ized (Prior, 2013; Pew Research Center, 2018), it is important to understand the perspectives underlying the political divisions and analyze their differences. Political framing, studied by political scientists (Entman, 1993; Chong and Druckman, 2007), provides the means to study these perspectives. It is a nuanced political strategy, used to bias the discussion on an issue towards a specific stance by emphasizing specific aspects that prime the reader to accept that stance. To help clarify this definition, consider two articles on the highly polarized immigration issue. Example 1: Different Perspectives on Immigration Adapted from Alternet (Left) Employees-many of whom are undocumented immigrants from Mexico, Ecuador and elsewhere-toil seven days a week for less than minimum wage, with no overtime pay. Adapted from Breitbart (Right) Mass immigration has come at the expense of America’s working and middle class, which suffered from poor job growth, stagnant wages, and increased public costs. The two articles capture opposite political per- spectives, liberal (top) and conservative (bottom). They do not directly contradict each other, instead they focus the discussion on different aspects help- ing them argue their case. The first emphasiz- ing the deprivation of minimum wage for immi- grants, and the second emphasizing implication on wages for U.S. workers. This process is known as framing. Our goal is to define, and automatically identify, relevant framing dimensions in politically- motivated coverage of news events to the extent they can capture and explain the differences in per- spectives across the conservative-liberal ideologi- cal divide (Ellis and Stimson, 2012; Preot¸iuc-Pietro et al., 2017). We focus on three divisive topics – immigration, gun-control and abortion. Previous work by Boydstun et al. (2014) stud- ied policy issue framing on news media and sug- gested 15 broad frames to analyze how issues are framed, which include economic, morality and se- curity, among others. These framing dimensions can help capture ideological splits (Johnson and Goldwasser, 2016; Johnson et al., 2017b). For ex- ample, by framing the immigration issue using the morality frame or using the security frame, the reader is primed to accept the liberal or conserva- tive perspectives, …

Content truncated to 500 words