Understanding How Images and Text Are Used to Frame Cable News Content

• In Event: Journalistic Norms and Practices

Abstract

News organizations increasingly use images and text to help frame political discussions. In some instances, the images and text both tell a similar story, while in other instances juxtapositions are used to highlight some portions of text at the expense of others. This study uses a large collection of television transcripts and footage beginning in 2007 to understand how images and text shape the way topics are interpreted. More specifically, using text and image data from CNN, Fox News and MSNBC, this study answers three basic questions: (1) how are images and text used to frame certain topics? (2) do these frames change over time? And, (3) do these general patterns vary across networks? Collectively, the answers to these questions will not only increase our understanding of how images and text are used to frame political discussions, but by looking at both mediums simultaneously, we will also speak to broader issues within political communication, like media bias and gatekeeping effects.

“Gatekeeping” is often described as “selecting, writing, editing, positioning, scheduling, repeating and otherwise massaging information to become news” (Reese, Vos, & Shoemaker, 2009, 73). Given television news’ important role in facilitating communication among three branches as well as between the government and the public (Cook, 1998, 2006; Farnsworth & Lichter 2011a, 2011b; Sparrow, 1999), understanding such interpretations and frames is essential to democratic governance (Farnsworth & Lichter, 2011b). However, as CNN, Fox News and MSNBC become more prominent, news has also become increasingly focused on entertainment (Farnsworth & Lichter, 2011b), leading some to become more concerned over potential bias (Barkin, 2003; Farnsworth & Lichter, 2011b; Salwen, Garrison, & Driscoll, 2005). This study will consider both of these relationships using all available cable news transcripts and associated images beginning in 2007.

Ultimately, we expect to find images frame text by either reenforcing the textual theme or by deliberately standing in juxtaposition. The variance between these general relationships is expected to be correlated with the topic and network. Some topics, like economic policy, are often overly complicated and images may be used to make these issues more relatable. Conversely, images can also be used to intentionally lead viewers to interpret the text in different ways to achieve some story objective. For example, an American flag may be shown when discussing censorship in authoritarian regimes to emphasize how those actions undermine American values. Regardless, the use of images will vary by network depending on topic, something that would be consistent with media bias. We will then test the effects of these various image and text permutations using an original experimental design. At a minimum, we expect to find issue and frame recall effects, but we also expect that in some instances vote choice will also be impacted.

In addition to the substantive implications of our study, we also think our study will speak to a growing literature which explores how images are used for issue framing (Brantner, Lobinger, & Irmgard, 2011; Corrigall-Brown & Wilkes, 2012; Gitlin, 1980; Powell et al., 2015). Not only is such framing important in and of itself, but other studies have found it can also influence voters’ assessment on candidates and voting in general (Banducci, Karp, Thrasher, & Rallings, 2008; Barrett & Barrington, 2005; Coleman & Wu, 2015; Grabe & Bucy, 2009; Rosenberg, Bohan, McCafferty, & Harris, 1986; Joo, Steen, & Zhu, 2015). Moreover, by looking at when and where text and images intersect, we will also speak to the well-established text-as-data literature (for review, see Hollinbaugh, 2019; Grimmer and Stewart, 2013), especially the many studies that have used quantitative text analysis to study issue framing (Buscher et. al, 2016; Stecula & Merkley, 2019; Walter & Ophir, 2019).

Authors

• Bryce Dietrich, University of Iowa

• Hyein Ko, University of Iowa