Traditionally, social science has been an âobstinately verbalâ field (Fyfe & Law, 1988, p. 4) in which scholars have shown a clear preference for the analysis of words over visuals. Visuals were considered âintellectual[ly] lightweightâ (Grabe & Bucy, 2009, p. 67), âpandering to [the] low tastesâ (Becker, 1995, p. 9) of âthe subhuman, the savage, the âdumbâ animal, the child, the woman [sic], the massesâ (Mitchell, 1994, p. 24; Postman, 1986). Other reasons for dismissing visuals included their ubiquity in tabloids and their capacity to stir emotions and to manipulate (impressionable) audiences (see Bergstroem, 2008; Habermas, 1989).
Given this pervasive âword biasâ1 (Grabe & Bucy, 2009, p. 6), researchers who shut their eyes to visuals did not have to justify their position for many years. This changed because of the devotion of a few researchersâparticularly Doris Graber (1980, 1986, 1987, 1988)âto visual analyses. For these scholars, visualsâ omnipresence, their exploitation for strategic purposes, and their accessibility to those less educated made them more interesting for research, not less (see also Grabe & Bucy, 2009).
Acknowledging this has led to a growing interest in âall things visualâ (Hughes, 2012, p. xxix). This interest is clear in the discipline-wide praise of studies focusing on the visual channel of communication. Image bite politicsâGrabe and Bucyâs 2009 monographâwas surely one of the landmark publications of recent years to strengthen the case for studying visuals. Even so, scholars must acknowledge that visual-only studies, just like verbal-only studies, are incomplete (see Coleman, 2010; Graber, 1987). Audiences are not exposed to either the verbal or the visual components of a message but rather to both words and visuals. Thus, at a time when the hurdles associated with researching visuals have been progressively reduced, our discipline must move toward âintegrative workâ (Coleman, 2010, p. 235) that analyzes both words and visuals. In this book, I investigate the feasibility of integrative analyses by focusing on framing theory.
Various definitions of frames and framing circulate in the academe (see Coleman, 2010; Entman, 1993; Entman, Matthes, & Pellicano, 2008; Reese, 2001). In my understanding, frames are rather enduring verbal or visual interpretations of issues or people. Frames seem natural to those whose ideas, norms, or values they reference. They organize social reality, simplifying complex matters by emphasizing certain aspects and disregarding others. Full-fledged frames define a state of affairs as problematic, suggest moral evaluations, identify causes, and propose ways to alleviate the problem. The treatment recommendations, especially, give frames their functional character by making certain policies, attitudes, or behaviors seem like the natural response to the problem. Visuals can convey frames by themselves, and these frames can be similar to or different from those articulated verbally in the same message. Visual frames fulfill the four framing functions outlined by Entman (1993) through the content of the visual and, possibly, in less obvious ways (e.g., through variations in nonverbal behavior, camera angles, and camera distances). Frames are available in culture, in communication (written and spoken words; still and moving images), and in peopleâs minds. Framing is the act of developing and promoting frames. In this way, frames pertain to issues or people. The latter group, frames about people, is known as âcharacter frames.â Grabe and Bucy (2009) introduced the term âcharacter framesâ in reference to politicians running for office. Through character frames, some aspects of peopleâs personalities and lives are selected and highlighted in verbal or visual communication, while others are disregarded or downplayed. Aspects may be chosen for emphasis to explain why a certain individual got into the state described or portrayed, to suggest a moral evaluation, or to propose ways to deal with that individualâs current state. Character frames organize social reality and make certain policies, attitudes, or behaviors seem like the natural response to the problem described or portrayed. Investigating character frames is worthwhile as the way in which journalists and other actors communicate about those most directly affected by an issue can impact audiencesâ perception of the issue as a whole more than baseline statistics.
Colemanâs (2010) call for integrative framing analyses set in motion the train of thought for this book. The point of departure for my investigation into their feasibility was my expectation that they would be twice as laborious as verbal-only or visual-only studies. Nonetheless, I expected the effort to be manageableâgiven the wide availability of methodological advice on both verbal and visual framing analyses (e.g., Coleman, 2010; David, Atun, Fille, & Monterola, 2011; Matthes & Kohring, 2008; Messaris & Abraham, 2001; Tankard, 2001; Van Gorp, 2010)âand, more important, worth the effort. But it soon became evident that mastering the body of knowledge associated with verbal and visual framing analyses, respectively, was only the first step in coming to grips with integrative framing analyses. Many questions about how words and visuals can relate to each other unfolded in the process. This reinforced my belief that the lack of precise, hands-on, methodological guidance on the integration of words and visuals was responsible for the small number of integrative framing analyses to date. In this book, I hope to blaze the trail for those considering embarking on such studies themselves. I propose a step-by-step approach to integrative framing analysis and implement it in an empirical study to demonstrate that it is feasible and effective.
The remainder of this book is divided into six chapters. Chapter 2 testifies to the need for integrative framing analyses by offering three arguments based on previous studies. One argument is that most messages consist of words and visuals (Kress, 2010) and that, on many occasions, visuals are the dominant element of such messages. A second line of reasoning draws on a vast body of experimental research that strongly suggested that visuals were awarded more attention and preferential processing than words (e.g., Holsanova, Rahm, & Holmqvist, 2006; Knobloch, Hastall, Zillmann, & Callison, 2003; Zillmann, Knobloch, & Yu, 2001). Finally, I review another corpus of studies investigating memory and recall for information conveyed verbally and/or visually. Here, it becomes apparent that messages conveyed visually are better at attracting attention than words are (e.g., Bucher & Schumacher, 2006; Donsbach, 1991; Garcia & Stark, 1991). But even in the unlikely event that audiences pay equal attention to words and visuals in a given message, there is no reason to focus solely on the verbal component of that message. As discussed in Chapter 2 in depth, when words and visuals do not convey the same information, which is frequently the case, audiences are much more likely to retain the information conveyed through visuals than through words (e.g., Bucy & Newhagen, 1999; Grimes, 1991; Lang, 1995).
In Chapter 3, I describe the difficulties associated with integrative framing analyses. I argue that a first challenge is posed by the wide array of approaches to verbal framing analyses and visual framing analyses, respectively (Coleman, 2010; Matthes & Kohring, 2008; Rodriguez & Dimitrova, 2011). The difficulty lies in deciding which of the approaches available makes the most sense for addressing oneâs particular research questions and hypotheses. As discussed in depth in this chapter, this decision can be particularly troublesome for visual framing, given the lack of clarity concerning the way to extract meaning from visuals. Scholars differ greatly in this regard. Drawing on the âmethodological trail of breadcrumbsâ provided by Coleman (2010) with the intention âto help framing scholars feel less lost in the visual forestâ (p. 235); on Grabe and Bucyâs (2009) empirical study on visual character frames; and on Rodriguez and Dimitrovaâs (2011) systematization of visual framing studies, and assisted by insights from Geise, Lobinger, and Brantner (2013), this chapter presents the five different foci chosen by framing scholars who include visuals in their analyses. A detailed review reveals thatâregardless of their focus on the verbal or the visual modalityâsome approaches have more weaknesses than others, but they resonate with researchers nonetheless. Thus, I pose that scholarsâ choice of one approach or the other depends not just on an evaluation of strengths and weaknesses, but also on its suitability for answering oneâs research questions, on oneâs understanding of frames, oneâs methodological preferences, cost-benefit calculations, and, finally, on efforts toward methodological triangulation.
Then, after the presentation of approaches to verbal and visual framing analyses when conducted individually, the approaches to integrative framing analyses identified in previous studies are reviewed. Here, I argue that one of the two approaches currently circulating in the literature has a clear advantage over the other. Specifically, in keeping with the literature reviewed in Chapter 2, I suggest that collecting data separately from words and visuals is much more valuable than attempting to do this simultaneously. Nonetheless, with this field of research still in its infancy, it becomes apparent that the sequence of steps to be followed in such analyses is far from clear. The question of precisely how to conduct integrative analyses when collecting data separately from words and visuals arises at the end of Chapter 3. This question is addressed in Chapter 5.
But first, in Chapter 4, I turn to the factors influencing the acts of selection and emphasis inherent to framing. Why do journalists and other communicators use the frames that they do? Based on previous research on frame building, however scarce, I explain how framing can sometimes occur unconsciously, because of the resonance of certain interpretations with the underlying culture (e.g., Gamson, 1989). I then move to a second explanation, which seems more plausible to me: More often, a communicatorâs interests and goals motivate that communicatorâs use of frames (e.g., Hallahan, 2015). I pose that framing analyses are incomplete without an investigation into how frames serve certain interests. Here, I present two lines of research, one attempting to link the prevalence of news sources in a story to news frames (e.g., Dimitrova & StrömbĂ€ck, 2012), the other contrasting news frames to frames conveyed by actors external to the media (e.g., HĂ€nggli & Kriesi, 2012). The latter are known as advocacy frames. While previous research identified other factors that impact frame building at the macro-level and meso-level (e.g., Dimitrova & Kostadinova, 2013), they are not at the core of this book.
In Chapter 5, I propose a clearly defined sequence of six steps to be followed, which I demonstrate in the next chapter. They concern the preparation of the material for the analysis, the data collection, and the data analysis. The focus is placed on identifying verbal frames in written text, identifying visual frames in still images, and also on assessing the interplay between them. For the assessment of the interplay, a formula is introduced. It allows the computation of what I call the verbal-visual frame congruence ratio (CR Frames). Applying this formula to the data results in a new interval-level variable that can be used in subsequent analyses.
In Chapter 6, I execute each of these six steps while conducting a study of the frames conveyed through written words and still images about people living with HIV/AIDS (PLWHA) in news, special interest publications (SIPs), and public service announcements (PSAs). This chapter begins with a literature review on the framing of PLWHA and continues with a presentation of hypotheses and research questions. The last sections of the chapter are devoted to the presentation and discussion of the results of the integrative framing analysis of PLWHA. I hope that journalists and communication practitioners can benefit from reading especially this section of the book, as it deals with a variety of factors potentially influencing the framing of PLWHA. Some of the factors whose influence on framing was tested in this study are the communication context (news, SIPs, PSAs), the sourcing practices (news sources, photo sources), the characteristics of the audience (aka community structure; i.e., political views, urban-rural ratio, religiosity, and HIV/AIDS prevalence), and the goals of communication. Finally, several influences on variations in the degree to which verbal frames and visual frames conveyed the same interpretation were tested.
The theoretical and methodological contributions of this book are addressed in Chapter 7. Despite the topic selected for the empirical test of the methodological approach proposed in Chapter 5, this approach can be applied to the study of any topic. I hope that this book can reveal bridges across our discipline as envisioned by Waisbord (2015), tear down the wall between researchers self-identified as either words-people or visuals-people (see Fahmy, Bock, & Wanta, 2014), and lead to an increased number of integrative framing analyses.
Note
- Hofmann (2008) used, therefore, the harsher term of âverbal snobberyâ (p. 270, my translation); see also Gazzaniga (1998).
As explained in the Introduction, visual-only studies, just like word-only studies, are necessarily fragmentary (see also Coleman, 2010; Graber, 1987). This chapter presents three arguments in favor of analyzing the contribution of both words and visuals to the framing of people or issues across communication contexts. They concern the ubiquity of multimodal messages and the ways in which visuals outplay words when it comes to attention, processing, memory, and recall.
Multimodal Messages and the Predominance of Visuals
Nowadays, purely verbal messages are typical of little more than radio broadcasting; purely visual messages are equally scarce (see Mitchell, 2005; MĂŒller, 2007). Monomodal messages are artificial by nature. In the words of Kress (2010), multimodal messages (i.e., those in which several modes of communication are used) are âthe normal state of human communicationâ (p. 1).
Multimodality is typical of news accounts (e.g., Kress & van Leeuwen, 1998; Macken-Horarik, 2003; Roosvall, 2016), not just on television (Gunter, 2015; Machin & Niblock, 2006), but also in newspapers (VobiÄ & TomaniÄ TrivundĆŸa, 2015). VobiÄ & TomaniÄ TrivundĆŸa (2015) showed how print journalists engaged in a âdesperate search for âany imageâ,â something they termed the âtyranny of the empty frameâ (p. 502). Furthermore, messages in magazines and advertising (whether social or commercial) are typically multimodal (Griffin, 2004; Newton, 2001). According to recent studies, visuals became the dominant, and sometimes the only, component of print ads (e.g., Brier, 2009; Gold, Cohen, & Shumate, 2008; Scalvini, 2010)âsomething that led Lobinger (2012) to the observation that photos are âadvertisingâs darlingsâ (p. 124). Finally, photos are now a crucial component of magazines (Griffin, 2004; Hull, Smith, & Schmittel, 2015).
More Attention and Preferential Processing for Visuals over Words
In environments glutted with messages, people can only process a small fraction (Findlay & Gilchrist, 2003; Graber & Holyk, 2012). The selection of the very few stimuli to be further processed is guided by what draws peopleâs attention (Eder, 2002; Hickethier, 2002; Holmqvist, Holsanova, Barthelson, & Lundqvist, 2003; Liu, 2005; Stark Adam, Edmonds, & Quinn, 2007). The extent to which a message manages to attract attention is important, as attention is prerequisite to further effects (Lobinger, 2012).
Previous research clearly suggested that visuals outperform words when it comes to attracting peopleâs attention (Bucher & Schumacher, 2006; Donsbach, 1991; Garcia & Stark, 1991; Holsanova, Rahm, & Holmqvist, 2006; Mendelson & Thorson, 2004; Pfau et al., 2006; Zillmann, Knobloch, & Yu, 2001). Photos showing faces and those that are emotionally involving seem particularly effective (Kanwisher, McDermott, & Chun, 1997; Knobloch, Hastall, Zillmann, & Callison, 2003; Levy, Hasson, & Malach, 2004; OâToole, 2005; Parker, 2003; Zillmann et al., 2001). Scholars estimate that about 80% of newspaper readers notice visuals (Blum & Bucher, 1998), but only about 10%â20% acknowledge the text (Kroeber-Riel, 1996).
An array of eye-tracking studies substantiates this argument. In these studies, scholars use a special device to measure where people look and for how long, when exposed to a certain stimulus. The results obtained from this line of research are quite consistent. Peopleâs scanning of stimulus material is consistently interrupted when the eye encounters photos, especially, but also headlines (Barthelson, 2002; Bucher, 2008; Bucher & Schumacher, 2006; Holsanova et al., 2006; KĂŒpper, 1990; McLellan & Steele, 2001; Stenfors, MorĂ©n, & Balkenius, 2003; Wolf & Grotta, 1985). Photos and headlines are among the most reliable âentry pointsâ (Garcia & Stark, 1991, p. 1) for multimodal messages.
Visuals impact more than peopleâs selection of few stimuli out of those demanding attention. They also have an effect on information processing in that visuals are processed before and faster than words (Gibson & Zillmann, 2000; Pfau et al., 2006). Scholars argue that visual information is processed about 6,000 times faster than words (Barry, 2005; Gazzaniga, 1992; Newell, 1990). At least four theoriesâthe Dual Coding Theory (DCT) (Paivio, 1986), the Heuristic Systematic Model (HSM) (Chen & Chaiken, 1999), the Elaboration Likelihood Model (ELM) (Petty & Cacioppo, 1984), and the Limited Capacity Model (LCM) (Lang, 2000)âspeak to this phenomenon.
First, Dual Coding Theory (Paivio, 1986) postulates that words and visuals are processed in interconnected subsystems dedicated to words or visuals, respectively. The verbal subsystem processes words one by one, logically and consciously (Gazzaniga, 1998; Rodriguez & Dimitrova, 2011), whereas the visual system processes visual cues immediately and unconsciously (Busselle & Bhandzic, 2011; Gordon, 2004; MĂŒller, 2003; Paivio, 1979, 1986). Thus, the DCT proposes that the processing of verbal information is necessarily much slower than that of visual information.
Additionally, the Heuristic Systematic Model (Chen & Chaiken, 1999) and the Elaboration Likelihood Model (Petty & Cacioppo, 1984) shed some light on why visuals are processed before and faster than words. Though there are differences between these theories (see for a recent review Coleman & Wu, 2015), delving into them is beyond the...