Framing the Colombian Peace Process: Between Peace and War Journalism

ABSTRACT

This bilingual, cross-national study analyzes stories about the Colombian peace process that were engaged with on social media to understand the use of peace and war framing in news reporting. Using content analysis as a method, this paper operationalized Galtung’s classification of peace journalism and follows framing methodological adjustments and improvements suggested by previous peace journalism scholars. Results show that, even during peace talks, media use war narratives more often than peace frames, and social media users amplify more war than peace-oriented content. Proximity to conflict also was shown to be an important factor, as Colombian media used more war frames than foreign media. These findings are relevant for their implications about how national media consistently emphasized a war frame that social media users amplified, which we argue has implications for how citizens viewed the Colombian peace process, ultimately potentially influencing the decision to vote down the referendum.

KEYWORDS:

• Conflict reporting
• framing
• peace journalism
• proximity
• social media
• peace negotiation
• news sharing

After 50 years of conflict with more than 220,000 deaths and more than 5 million displaced citizens (Centro Nacional de Memoria Histórica 2013), the Colombian government of former president Juan Manuel Santos and the Revolutionary Armed Forces of Colombia (FARC) guerrillas signed a peace agreement on Sept. 26, 2016 (Ríos 2017). While the final agreement addressed fundamental political, social, economic, and legal issues (OACP 2016; Summers 2012), the four-year negotiation process and the final agreement faced strong opposition (Semana 2016). On Oct. 2, 2016, in a referendum election, the agreement was rejected, with 50.2% of Colombian voters opposing the deal (El Espectador 2016). Following the referendum’s failure, the Colombian government and the FARC signed a revised agreement on Nov. 24, 2016, which included some demands from the opposition (The Economist 2016). Congress approved the modified document on Nov. 29, 2016, bringing the world’s longest continuous war to an end (Casey 2016).

Working as a journalist in Colombia, in the middle of this conflict, has been considered a risky activity. From 2016 to 2020, eight journalists were murdered, and 618 threats were reported; statistics that rank Colombia the second deadliest country on the continent after Mexico (FLIP 2020). As a result, Colombian journalists have tended to adopt self-censorship as a defense mechanism to face risks and potential threats even in the post-conflict era (Barrios and Miller 2020). Factors such as dependency on official sources and direct advertising, judicial harassment and editorial pressures trigger self-censorship, particularly in small cities and rural areas (Barrios and Miller 2020). Garcés and Arroyave (2017), who surveyed 751 Colombian journalists, found a positive correlation between violence against news professionals and their autonomy to publish information on government, armed forces, criminal gangs, and structural social problems. Likewise, Hughes et al. (2017) concluded that difficult conditions associated with insecure democracies like the one in Colombia shape journalists’ perceptions about their work environments. However, after the peace agreement was signed, in-depth interviews show that Colombian media and journalists are making efforts to change their practices, and therefore consciously giving more space to peace building, conflict resolution and victims’ voices (Prager and Hameleers 2021).

In this context of violence, opposition and negotiation, media played a crucial role in framing the Colombian peace talks. Drawing on scholarship on peace/war journalism framing, this study explores the news coverage of the Colombia peace process in news articles that were shared, liked, and commented on via social media. One of the main purposes of this paper is to fill three gaps in the peace journalism research tradition and expand scholarship in this field by analyzing media representations of the historical peace negotiation in Colombia. The first gap has to do with the limited amount of works that analyze the relationship between peace journalism and social media. Most research in this area focuses on traditional media content or journalistic practices, but few articles have examined how this theory operates on social media, and how audiences react to it on these networked systems. Moreover, this research aims to understand the relationship between the use of peace or war journalism frames and the number of social media recommendations on Facebook and Twitter, showing to what extent social media audiences amplified or suppressed messages from the press. Second, the current research takes a quantitative content analysis approach to peace journalism reinterpreting Lee and Maslog’s (2005) classification of this theory. The methodological approach not only does support this kind of systematic research but also complements the limited number of quantitative studies in this area. Third, this paper aims to analyze whether peace journalism’s frames and devices, which often fit journalism practices in times of confrontation and war, prevail during peace negotiations.

Origin and Definition of Peace Journalism

Galtung and Ruge proposed the original idea of peace journalism in their seminal essay The Structure of Foreign News (1965), in which they criticized the war reporting style of Norwegian newspapers covering conflicts in Congo, Cuba, and Cyprus. Peace journalism promotes conflict resolution when editors and reporters concentrate on areas of agreement that “tone down political and ideological disparities,” focus on historical context, unveil causes and consequences, include common people sources, and describe the invisible consequences of combat (Knightley 2000, xxii). Conversely, war journalism often uses military language, overemphasizes the visible effects of war (such as human casualties, bloodshed, and material damage), focuses on elite sources, and adopts a superficial narrative with little context or historical assessment (Knightley 2000; Hachten and Scotton 2006; Lynch 2007). As Lynch and McGoldrick (2006) defined it, peace journalism occurs “when editors and reporters make choices — of what to report and how to report it — that create opportunities for society at large to consider and value nonviolent responses to conflict” (5).

From Practice to Theory Under the Umbrella of Framing

News framing — one of the most used theories in communication (D’Angelo and Kuypers 2010) — is a process of selection, exclusion, elaboration, and emphasis that organizes information according to some salient principles that convey a specific story angle to create meaning and convey a specific version of reality (Entman 1993). For example, Gitlin (1980) showed how framing of anti-Vietnam war protests highlighted protesters’ appearances to trivialize them and emphasized the presence of Communists among protesters in order to marginalize and delegitimize protests. By highlighting elements through cultural symbols, language choice, and repetition, news frames can influence citizens’ perceptions and social judgments (Cappella and Jamieson 1997; Chong and Druckman 2007; Iyengar 1991; Reese 2001), shaping how readers understand a news event (Scheufele and Tewksbury 2007), and even their attitudes and behaviors (Bortree et al. 2013; Davis 1995; Hipolito 2011). For example, framing contributes to who readers blame for social problems (Iyengar 1991). Frames’ effectiveness can vary according to cultural resonance and emotional appeals (Ryan 1991; Gamson 1992).

Frames can be issue-specific (focused on a particular episode or individual case) or more generic (thematic, focused on broader trends), such as those focused on human rights, or values (Elliott, Fitzgerald, and Hayward 2009; Iyengar 1991). Generally, previous research has identified five generic media frames: attribution of responsibility, conflict, economic consequences, human interest, and morality (e.g., Matthes 2009; Scheufele 1999). How journalists frame a story depends on individual-level factors (Tuchman 1978), as well as organizational- and social-structural level factors (Shoemaker and Reese 2014).

Scholars recognize that peace journalism, developed as a global training and educational program to influence journalistic practices in times of war (Nohrstedt and Ottosen 2015), relies on framing as fundamental to theorizing conflict reporting (Neumann and Fahmy 2012). The link between framing and peace journalism was evident since the foundation of peace journalism when Galtung (1986) presented key concepts, akin to frames, that help professionals and scholars to differentiate peace from war coverage in international news. McGoldrick and Lynch (2000) extended Galtung’s ideas theoretically when they described peace journalism as a way of framing stories using “conflict analysis” (5). Then, Lee and Maslog (2005) offered an empirical approach to these ideas when they operationalized the “peace journalism model” into a set of evaluative criteria to content analyze the conflict coverage of 10 Asian newspapers. Lee and Maslog (2005) divided their variables into two categories: peace/war journalism approach, and peace/war journalism language.

The Prevalence of War Framing

With some exceptions, most research based on the dichotomy of peace and war framing has found that media and journalists rely on war principles and language to report conflict. For instance, an analysis about the Turkish press showed that journalists reported on the conflict with Syria in terms of “us vs. them,” with loaded language and blame placed on the other side (Ersoy 2016). Similarly, a comparison between the news coverage of local Asian conflicts and the international Iraq war revealed that eight Asian newspapers “relied on war journalism framing to report their local conflicts but used peace journalism frame to cover the Iraq war” (Lee, Maslog, and Kim 2006, 511), showing a relationship between local proximity to the conflict and war framing. An examination of the old territorial dispute in Kashmir showed that newspapers from Pakistan and India continue to adopt a “knee-jerk coverage of conflict, with little consideration for long-term peaceful solutions” (Lee and Maslog 2005, 322). A content analysis of the New York Times and The Washington Post’s coverage of the Kashmir conflict concluded that geopolitical implications were also important when framing stories in terms of war and peace, as these US newspapers often portrayed Pakistan as the enemy (Siraj 2008). A study about the 2010 Israeli-Palestinian related incident of Mavi Marmara concluded that an Israeli newspaper (Haaretz) used significantly more war journalism narratives than the two non-Israeli papers (The Guardian and The New York Times) (Fahmy and Eakin 2014). Exceptions to the dominance of the war frame have been found in media coverage of the Sri Lanka and Philippines internal conflicts, which offered some encouragement to peace journalism during the 2001 ceasefire agreement (Lee and Maslog 2005, 323).

Researchers have also tested the effects of war and peace frames on news users. They have found that people who watched peace journalism frames on TV were “less angry and fearful” and more “hopeful and empathetic” than those who watched war-framed stories as well as less inclined to blame one side of the conflict (Lynch and McGoldrick 2012, 1052). With this in mind, understanding the framing of the Colombian peace process and how audiences engaged with this content becomes even more important for its potential to sway voters to vote for — or against — the peace referendum. The influence of digital technology and social media on the production and distribution of news is “rapidly rewriting the principles and protocols of war and conflict reporting” (Matheson and Allan 2010, 187) and making imperative a new model of peace journalism outside the mainstream media (Aslam 2016).

Even during the coverage of a peace process, news focuses on the ongoing conflict between the two sides, in part because of the media’s constant need for drama (Wolfsfeld 2003). Wolfsfeld (2003) argues that negotiations are “boring events” and as such are rarely seen as newsworthy by media (7). When negotiations break down or violence between the parties erupts on the ground, journalists return to traditional models of reporting, providing “strategic advantages” to those who oppose peace (Wolfsfeld 2003, 7). Therefore, media and journalists could play an important role in intensifying violence, showing an inability to contribute to peace building during periods of calm (Wolfsfeld 2003). Given that previous research has shown consistently the prevalence of the war frame, the following hypothesis and question are presented:

H1: News media will rely more on war frames and language than on peace frames to report the Colombia peace process.

RQ1: What are the most salient indicators of the war/peace journalism frames manifest in the news coverage of the Colombian peace process?

In the competition over media frames, a central idea is that political powers (authorities) have important advantages in influencing news media (Wolfsfeld 1997). However, challengers or antagonists of those in power can also employ the news media to achieve their political goals. Wolfsfeld’s political contest model (1997) theorizes that different forces, including antagonists, compete for news media attention to achieve political influence, and promote their own frames of the conflict to mobilize political support for their cause. This model is useful to understand how news articles that reflect supporters of or antagonists to a peace process interact with the peace and war frames. According to Wolfsfeld, understanding the role that news media play in reporting conflicts could shed light on factors that lead to the success and failure of peace negotiations. Given the peace negotiations context of Colombia, this study formulates the following hypothesis:

H2: News articles that privilege antagonists to the peace process will promote more war frames than peace frames whereas those articles that privilege supporters to the peace process will promote more peace frames than war frames.

Peace Journalism in the Colombian Context

In Colombia, journalists and scholars have adopted peace journalism both as an active tool for reporting and as a theoretical framework to analyze media coverage of conflict. Projects like Medios para la Paz (Media for peace, MPP), created by a group of 80 journalists in 1998 to increase accountable journalism when reporting on the armed conflict, and online media projects like Colombia 2020, from El Espectador newspaper, Pacifista and Verdad Abierta have been relevant to peace building. Studies on peace journalism have also increased. Recently, after the peace agreement, scholars have found that Colombian journalists understand their role as a set of practices that can contribute to peace building and to mobilize audiences towards conflict resolution (Prager and Hameleers 2021). In-depth interviews with reporters have shown that news professionals want to focus on contextualizing the conflict, offering solutions, and giving voice to common people (Prager and Hameleers 2021).

To counter common reporting practices that silence the voices of the victims and rely heavily on official sources (Serrano 2014, 2015; Cortés-Martínez and Thomas 2020), some scholars have proposed immersive journalistic models that use collaborative narratives and storytelling to document lives and experiences of ex-combats, victims and marginalized communities in Colombia (Jukes, Charles, and Fowler-Watt 2020). Moreover, Charles (2021) has argued that a better understanding of trauma could support journalism that aims to build peace in Colombia. Given the large social, cultural, and economic challenges of Colombia due to the long-armed conflict, the journalistic tradition in that country has developed “a special style of advocacy reporting” that “tend to take part in solving community problems”; a practice that does not necessarily fit with normative Western journalism (Arroyave and Barrios 2012, 400).

Proximity and the Peace/War Frames

Proximity, the notion that readers will care more about news that is close to home, can be cultural or geographic (Johnson 1997). Galtung and Ruge (1965), who originated the concept of cultural proximity, suggested that proximity influences whether something will be considered news. Shoemaker et al. (2007) noted that proximity is a multi-dimensional concept. For example, localization refers to the extent to which a foreign event has meaning for a local or domestic audience (Morton and Warren 1992; Shoemaker et al. 2007). While studies differ on the extent to how important proximity is in determining whether something will be covered, research in general posits that proximity influences the quantity and type of coverage, especially when it comes to conflict (Shapiro and Williams 1984).

Scholarship is divided on whether proximity prompts more positive or negative attitudes between different groups. On the one hand, some studies indicate that exposure to and contact between different groups generates positive attitudes toward the other group and can even diminish prejudice (Kinder and Mendelberg 1995; Pettigrew and Tropp 2006). On the other hand, Sherif and Sherif (1953) pointed to proximity causing negative attitudes as groups competed for resources. Likewise, Esses and colleagues (2002) found negative attitudes arise when groups see competition as a zero-sum game. Stein, Post, and Rinden (2000) explained the seemingly contradictory findings by suggesting that casual contact between different groups could prompt more negative inter-group attitudes, while extended contact more positive ones.

Within the peace journalism literature, proximity is an important factor influencing how news is framed. In their study of how U.S. media cover the drug war in Mexico, Lacasse and Forster (2012) found newspapers closer to the conflict used more peace than war-oriented frames, while coverage in those outlets further away from the border was a balance between peace- and war-framed stories. Further, the study showed the more distant newspapers used more demonizing language than the closer outlets, indicating that proximity could perhaps have a positive effect on inter-group attitudes. Still, the study also showed that more proximate newspapers oversimplified the conflict in comparison with more distant newspapers, indicating that peace and war frames are somewhat fluid. As such, more research is needed to understand the interplay between proximity and peace and war framing so we posit the following questions:

RQ2: Are there significant associations between proximity of the country/region of the media outlet and the peace/war frames?

RQ2a: Are there significant mean differences between the salient indicators of the war frame and proximity to the conflict, given the country/region of the media outlet?

RQ2b: Are there significant mean differences between the salient indicators of the peace frame and proximity to the conflict, given the country/region of the media outlet?

Sharing Frames via Social Media

The sophisticated relationship between news media outlets and social media platforms has become a principal component of journalism and communication studies. Social media behavioral research has often examined metrics to better understand this relationship, including assessing the number of page views or “clicks” (Cherubini and Nielsen 2016; Tenenboim and Cohen 2015); sharing (e.g., Kilgo et al. 2016; García-Perdomo et al. 2017; Trilling, Tolochko, and Burscher 2016), commenting (Ksiazek, Peer, and Lessard 2016), and direct reporting of user engagement (Kormelink and Meijer 2017). In this study, we focus specifically on measurable social media interactions (liking, commenting, sharing, retweeting) with news content posted on Facebook and Twitter.

Predictors of these behaviors can be inconsistent, and given the volatility of the digital sphere, continuous changes might be anticipated. In non-experimental examinations of news content and social media interactions, scholars have explored multiple features such as presentation style and framing (e.g., Kilgo et al. 2016; Tenenboim 2017; Valenzuela, Piña, and Ramírez 2017), and news values (e.g., García-Perdomo et al. 2017; Trilling, Tolochko, and Burscher 2016; Weber 2014), though results from one study do not necessarily correlate with the next. Bright’s (2016) analysis shows that conflict and politics are cast aside by Facebook audiences, and instead, affect is a key predictor of sharing. Kilgo, Lough, and Riedl (2017) also discovered emotional aspects of news coverage were linked with increased sharing on Facebook, but not on Twitter. Positive and negative valence of an article can increase audience engagement (Hornik et al. 2015). Kim and Yang’s (2017) assessment of organizational communication showed that likes, shares, and comments were correlated with specific media characteristics. In their analysis of Facebook content, they found liking and sharing were linked more closely with affective characteristics, while commenting was more closely associated with rational features. Specifically, within the context of Latin American countries, researchers have identified several patterns of sharing and press coverage (García-Perdomo et al. 2017; Harlow et al. 2017; Kilgo et al. 2016; Valenzuela, Piña, and Ramírez 2017). Among those, conflict is a consistent characteristic that shifts engagement patterns. However, no article to date has specifically looked at the relationships among the international news coverage of war resolution, and social media audiences. As such, this research explores this relationship in the following way:

RQ3: Do social media audiences interact with and recommend frames of peace and war in different ways on social media?

H3: Negative and positive emotive language will increase social media recommendations of news articles about the Colombian peace process.

Methods

To understand how news media integrated the peace and war frames in their coverage about the Colombian peace process, this study relied on a content analysis of articles about the peace talks published in both English and Spanish by online news media outlets around the world from June 1 to October 15, 2016.¹ NewsWhip, a media tracking and data collection company, was utilized to collect the articles. NewsWhip tracks more than 100,000 publications and media providers around the world and collects all unique URLs from articles hosted and generated by those media outlets’ websites. NewsWhip also associates URLs with public analytic data on Facebook, Twitter and other social media platforms, based on shares, likes, and comments (NewsWhip 2016).

Sample

Keywords were collected using English and Spanish terms such as “FARC,” “Colombian peace process,” “Colombian peace talks,” “peace negotiation,” “proceso de paz en Colombia,” “negociación de paz”, and “proceso de paz colombiano” were utilized to collect the data, resulting in an archive of links that exceeded download capacity. The database prioritized the top 10,000 articles shared at least once on Facebook — the largest social media platform in the world. From this list, an initial random sample of 1200 articles was selected for analysis: 600 for in English and 600 in Spanish. Using a probability sampling method allows researchers to know in advance how likely it is that any element of a database will be selected as part of the sample and to generalize the results (Schutt 2012). During the coding process, irrelevant links were located removed from the sample without replacement (n = 359), resulting in a total of 841 articles (416 in English, 49.5%; and 425 in Spanish 50.5%).

Coding Procedures and Operationalizations

The codebook was initially constructed using Galtung’s (1986, 1998, 2010) classification of peace journalism to determine the presence of peace and war frames in articles about the Colombian peace process that were shared on social media, and reflect enhancements made in more recent journalism studies (Fahmy and Eakin 2014; Lacasse and Forster 2012; Lee and Maslog 2005, 2006). The coding was completed by three coders, all authors in this study. Coders went through extensive training to create and refine the variables of interest in this study. Prior to coding, intercoder reliability testing was conducted using Krippendorff’s as the reliability measure. Final levels of reliability ranged from .78 to 1 and the average pairwise percent agreement ranged from 86% to 98.1%. Each operationalization and alpha level are provided for variables in the final codebook.

Region of Media Outlet

Coders identified the region of the outlet as one of the following: Colombia, Latin America/Caribbean, Europe, United States/Canada, all the rest (Average pairwise percent agreement = 97.1%; α = .91).

Antagonists and Supporters

Coders were asked to read the first three paragraphs of the article to determine what group’s perspective — antagonists or supporters to the peace process — was privileged in the beginning of the story. If the article did not present the view of antagonists or supporters, coders selected neither/other. If the article contained equal points of view from people in favor of and against the peace process, coders marked “both.” (Agreement = 89.6%; α = .80).

Peace and War Frame Approach

Two computed indexes of 9 indicators each were created: One for the war frame², and another one for the peace frame.³ The frequency of the two indexes was compared to determine whether articles contained more war or peace elements. Stories with higher peace than war indicators were considered part of the peace frame whereas articles with higher war indicators than peace scores were classified as part of the war frame.

Causes and Consequences

Coders identified whether (1) or not (0) the article focused on causes and consequences of the peace process and conflict (Agreement: 93.3%; α = .85).

Here and Now

Coders identified whether (1) or not (0) the article focused on the most recent developments of the peace negotiations without providing a broad picture of the conflict (Lynch and McGoldrick 2006) (Agreement = 91.9%; α = .84)

Solutions and Similarities/Problems and Differences

Coders decided if the article provided solutions and similarities, finding common ground between parties that facilitates solutions (1) or, conversely, it focused on differences/problems between the parties that intensified the conflict (2). If both problems and solutions were present, coders selected (3) “equal,” or if not present (4) for neither (Agreement = 93.3%; α = .88).

Invisible Effects of Violence

Coders decided whether (1) or not (0) the article included violence that was less noticeable, such as cultural violence (e.g., hate speech), structural violence (e.g., economic injustice), invisible effects of war (e.g., emotional trauma, mental health) that can damage society (Agreement = 90.8%; α = .82).

Visible Effects of Violence

Whether (1) or not (0) the article focused on visible aspects of violence, such as casualties, displaced people, kidnaped or injured civil populations, mutilated or injured fighters, guerrillas’ attacks or drug trafficking, property damage, or military occupation (Agreement = 5.4%; α = .87).

Common People’s Sources and Voices

Does the article include (yes 1–no 0) the voices of common people when reporting about conflict or the Colombian peace process? (Agreement = 90.5%; α = .76).

Elite Sources and Voices

Does the article present the peace negotiation through elite voices and officials as if they were the sole actors? (yes 1–no 0). (Agreement = 86.2%; α = .75).

Multi-party Conflict Reporting vs. Two-Party Conflict Reporting

The peace frame includes many actors, multiple voices, and parties (Galtung 1998). Coders were asked to count the parties reported in the story. If multiple parties were mentioned in the story, coders selected (1). On the contrary, the war frame reduces the news reporting to two sides in confrontation. Coders were asked to select 0 if the article only presented one or two parties in the conflict (Agreement = 90.7%; α = .78).

Labeling Parties as Good or Bad

Does the article focus on one group or actor to portray it as the evil guys? Or does the article focus on one group or actor to portray it as good? The war frame often divides people in the good and bad, blaming one side of the peace negotiation or conflict (Galtung 1998) (Agreement = 90.8%; α = .81).

Specific Language

Victimizing Language

Coders identified the presence (1) or the absence (0) of victimizing language in the article by portraying citizens, communities and civilians as defenseless, exploited, devastated or demoralized victims and people (Agreement = 90.4%; α = .80).

Unifying Language

Coders identified whether (1) or not (0) the article used unifying language to describe the events and parties involved in the peace process. Articles that reported on how people were coping with conflict or provided an empowering view of those affected by conflict were considered part of the unifying language (Agreement = 92.5%; α = .85).

Demonizing Language

Coders decided whether (1) or not (0) the article used demean language to describe parties involved in conflict, portraying some as evildoers⁴ (Lynch and McGroldrick 2005) (Agreement: 93.1%; α = .86).

Positive Emotive Language

Coders identified the presence (1) or non-presence (0) of positive emotive words in the article (e.g., hope, joy, happiness, optimism, euphoric, amusement, or trust) (Agreement: 98.1% α = .95).

Negative Emotive Language

Coders identified the presence (1) or the absence (0) of negative emotive words when describing actions, situations, and actors (e.g., fear, anger, sadness, anxiety, disgust, rage, disappointment, or hate). In the war frame, strong emotional language is used frequently so that it has the potential to intensifying polarization, violence and hate (Wolfsfeld 2003) (Agreement: 94.4%; α = .86).

Results

From the final sample of 841 stories related to the peace process, 413 belonged to Colombian media (49.1%), 276 news articles were from US/Canadian media (32.8%), 126 came from Europe (15%), 16 from other countries in Latin America (1.9%), and ten from other regions (1.2%). Given the low frequency of Latin American media, that category was collapsed into “other” regions. A descriptive analysis of the data shows that 75% (N = 630) of articles from the sample were from traditional media (e.g., El Espectador, El Tiempo, The New York Times, LA Times, CNN, etc.); 12.7% (N = 107) proved to be well-recognized digital native media (e.g., Huffington Post, Lasillavacía.com, eldiario.es, etc.); 11.7% (N = 98) of articles were published by alternative media (e.g., icrc.org, cfr.org, breitbart.com, etc.), and 0.7% (N = 6) belonged to videostreaming platforms such as YouTube. The sample was also categorized by groups privileged in the news reporting. Frequencies showed that 423 stories favored supporters to the peace process (50.3%), 265 stories favored antagonists (31.5%), and 153 articles (18.2%) did not favor supporters nor antagonists.

The first hypothesis states that news coverage of the Colombian peace process would reflect more war than peace journalism frames. Results supported this hypothesis and showed that, even during the coverage of a peace process, news media more often used the war frame than the peace frame in their coverage [X²= 162.1006, df = 2, p < .001]. From the total sample of 841 stories, 468 (55.6%) contained the war frame compared to 273 stories (32.5%) framed as peace, and 100 news stories (11.93%) were classified as neutral.

The first research question aimed to find significant mean differences between the most salient indicators of the war frame and the peace frame manifest in the news coverage of the Colombian peace process. To answer this question, two samples t-tests were conducted; one for the war frame indicators and another for the peace frame elements. Results showed that the most salient indicators of the war frame were the inclusion of elite sources in the stories (M = .76, SD = .425) [t(840) = 52.229, p < .001], an emphasis on a two-conflict-party reporting (M = .74, SD = .439) [t(840) = 48.844, p < .001], and a focus on the here and now (M = .65, SD = .476) [t(840) = 39.845, p < .001]. Meanwhile, the most salient indicators of the peace frame were not-labeling parties as good or bad (M = .68, SD = .467) [t(839) = 42.139, p < .001], avoidance of victimizing language (M = .61, SD = .489) [t(839) = 35.919, p < .001], and avoidance of emotive language (M = .49 SD = .5) [t(840) = 28.568, p < .001]. Therefore, news articles shared on social media about the Colombian peace process tend to privilege elite-official perspectives and a two-party style reporting while voices of common people and other parties involved are barely heard. Given that independent statistical tests were performed simultaneously, an additional Bonferroni correction test was run to adjust the P value by dividing it by the number of comparisons made (Tables 1 and 2).

Table 1. Mean differences between war frame indicators.

	T	df	Mean difference	SD
Elite sources and voices	52,229	840	.76 ***	.424
Two party conflict reporting	48,844	840	.74 ***	.439
Here and now	39,845	840	.65 ***	.476
Visible effects of violence	36,065	840	.61 ***	.488
Emotive language	29,716	840	.51 ***	.5
Problems & differences	25,218	838	.43 ***	.496
Victimizing language	23,465	840	.40 ***	.489
Labeling as good or bad	19,941	838	.32 ***	.467
Demonizing language	14,321	839	.20 ***	.397

***p < .001.

Table 2. Mean differences between peace frame indicators.

	t	df	Mean difference	SD
Avoiding labeling parties as good or bad	42,138	838	.68 ***	0.467
Non-Victimizing language	35,919	839	.61 ***	.489
Non-Emotive language	28,268	840	.49 ***	.5
Unifying language	25,207	839	.43 ***	.495
Causes and consequences	23,355	840	.41 ***	.497
Solutions & similarities	21,262	838	.35 ***	.477
Multiparty conflict reporting	17,201	839	.27 ***	.439
Common people’s sources	16,614	840	.25 ***	.431
Invisible effects of violence	13,174	840	.17 ***	.377

***p < .001.

The second hypothesis posited that news articles that privileged antagonists to the Colombian peace process would promote more war frames than peace frames whereas those articles that privileged supporters to the peace process would promote more peace frames than war frames (see Table 3). A crosstab test to analyze associations between variables showed that H2 was supported. Results showed that news articles that gave predominance to antagonists incorporated more war frames (45.7%, n = 214) than articles that favored supporters (37.4%, n = 175). Conversely, articles that privileged supporters integrated more peace frames (68.6%, n = 190) than those articles that favored antagonists (12.1%, n = 33) [X²= 107.682, df = 4, p < .001].

Table 3. Comparing group privileged by peace/war frames.

	Frames present
Group privileged	Peace (%, n)	War (%, n)	Neutral (%, n)	Total (%, n)
Antagonists	12.1 (33)	45.7 (214)	18 (18)	31.5 (265)
Supporters	69.6 (190)	37.4 (175)	58 (58)	50.3 (423)
Neither/Other	18.3 (50)	16.9 (79)	24 (24)	18.2 (153)
Total (%, n)	100 (273)	100 (428)	100 (100)	100 (841)

X² = 107.682, df = 4, p < .001.

Answering RQ2, which examined the relationship between proximity and use of peace and war frames, a Chi-square test showed significant differences among regions (see Table 4). Results showed that Colombian media used the war frame (66.6% n = 275) more than North American (47.1% n = 130) and European media (43.7% n = 55) [X²= 46.025; df = 6, p < .001]. Conversely, the peace frame was used significantly less in Colombia (24.2%) than in European (44.4%) and North American media (37.8%) [X²= 46.025, df = 6, p < .001].

Table 4. Comparing country/region of the media outlet by peace/war frames.

	Frames present
Country/Region of the media outlet	Peace (%, n)	War (%, n)	Neutral (%, n)	Total (%, n)
Colombia	24.2 (100)	66.6 (275)	9.2 (38)	100 (413)
Europe	44.4 (56)	43.7 (55)	15 (11.9)	100 (126)
US/Canada	37 (102)	47.1 (130)	15.9 (44)	100 (276)
Other	57.7 (15)	30.8 (8)	11.5 (3)	100 (28)
Total	32.5 (273)	55.6 (468)	11.9 (100)	100 (841)

X² = 46.025, df = 6, p < .001.

RQ2a aimed to find significant mean differences between the indicators of the war frame and proximity to the conflict given the country/region of the media outlet (see Table 5). Results of ANOVA tests showed that there were significant mean differences in proximity of the country/region scores for the following war frame indicators: focusing on here and now [F (3, 837) = 11.538, p < .001], visible effects of violence [F (3, 837) = 36.114, p < .001], labeling parties as good or bad [F (3, 835) = 22.891, p < .001], two-party conflict reporting [F (3, 837) = 3.935, p < .01], stressing problems and differences [F (3, 835) = 59.343, p < .001], and using demonizing [F (3, 836) = 6.038, p < .01] and emotive language [F (3, 837) = 5.566, p < .001].

Table 5. Comparison of mean differences of war frame indicators by country/region of the media outlet.

	Country of the media outlet
War frame indicators	Colombia M (SD)	Europe M (SD)	US/Canada M (SD)	Other M (SD)	F value
Here and now	.71 (.451)^a	.75 (.437)^a	.52 (.51)^b	.61 (.496)^a, b	11.538***
Visible effects of violence	.44 (.497)^a	.78 (.417)^b	.77 (.418)^a, b	.65 (.485)^a, b	36.114***
Elite Sources	.73 (.446)	.81 (.41)	.81 (.394)	.73 (.452)	2.462
Labeling parties as good or bad	.45 (.499)^a	.13 (.334)^b	.2 (.427)^c, b	.15 (.368)^b, c	22.891***
Two party conflict reporting	.79 (.408)^a	.72 (.45)^a	.67 (.469)^b	.73 (.452)^a, b	3.935**
Problems & differences	.64 (.480)^a	.19 (.394)^a	.25 (434)^b	.19 (.402)^a, b	59.343***
Victimizing language	.41 (.491)	.37 (.485)	.41 (.492)	.31 (.471)	.431
Demonizing language	.24 (.429)^a	.063 (.245)^b	.20 (.403)^a	.04 (.196)^a, b	6.038***
Emotive language	.46 (.499)^a	.52 (.502)^a	.60 (.491)^b	.38 (.491)^a, b	4.685**

Notes: Higher means represent the presence of higher war frame indicators. Standard deviations are noted in parentheses. Different superscripts (a, b, c, or d) indicate statistically significant differences in Tukey’s post-hoc tests. Differences in means with the same superscript are not statistically significant. N = 841. **p < .01; ***p < .001.

Post-hoc Tukey HSD tests revealed that Colombian media used significantly more here and now reporting in their peace process coverage (M = .71, SD = .451) than the US/Canadian media (M = .52, SD = .51). Likewise, European media implemented more here and now reporting in their coverage (M = .75, SD = .437) than the US/Canadian media (M = .52, SD = .51). In other words, the US/Canadian media used significantly less here and now reporting than their Colombian and European counterparts. Significant mean differences were also found among the regions when reporting on visible effects of violence such as casualties, destruction, and mutilations. European media (M = .78, SD = .17) focused more on the visible consequences of violence than the Colombian media (M = .44, SD = .497). Conversely, Colombian media focused more on problems and differences (M = .64, SD = .480) and two-party conflict reporting (M = .79, SD = .408) than the US/Canadian media (M = .25, SD = .434) (M = .67, SD = .469). Colombia also labeled more frequently parties as good and bad (M = .45, SD = .499) than the US/Canadian (M = .2, SD = .427) and European media (M = .13, SD = .334). Finally, specific language also showed significant mean differences with European media using less demonizing language (M = .06, SD = .245) than Colombian (M = .24, SD = .429) and US/Canadian media (M = .20, SD = .403), whereas the US/Canadian media (M = .60, SD = .491) incorporated emotive language more than Colombian outlets (M = .46 SD =  .499).

RQ2b asked whether there were significant mean differences between the peace frame indicators and proximity to the conflict given the country/region of the media outlet (see Table 6). Results of ANOVA tests showed significant mean differences in proximity for the following peace frame indicators: focusing on causes and consequences [F (3, 837) = 7.120, p < .001], not-labeling parties as a good or bad [F (3, 835) = 22.982, p < .001], multi-party conflict reporting [F (3, 836) = 4.001, p < .01], focusing on solutions and similarities [F (3, 835) = 30.792, p < .001], unifying language [F (3, 836) = 45.259, p < .001], and avoiding emotive language [F (3, 837) = 4.685, p < .01].

Table 6. Comparison of mean differences of peace frame indicators by country/region of the media outlet.

	Country of the media outlet
Peace frame indicators	Colombia M (SD)	Europe M (SD)	US/Canada M (SD)	Other M (SD)	F value
Causes and consequences	.34 (.474)^a	.36 (.481)^a	.51 (.522)^b	.42 (.54)^a, b	7.120***
Invisible effects of violence	.19 (.390)	.13 (.343)	.18 (.386)	.0 (.0)	2.479
Common people’s sources	.28 (.451)	.24 (.428)	.20 (.4)	.23 (.430)	2.136
Non-Labeling parties as good/bad	.55 (.498)^a	.87 (.334)^b	.76 (.426)^c, b	.85 (.368)^b, c	22.982***
Multi-party conflict reporting	.21 (.408)^a	.28 (.45)^a	.33 (.470)^b	.27 (.452)^a, b	4.001**
Solutions & similarities	.21 (.411)^a	.62 (.488)^b	.41 (.493)^c	.58 (.504)^b, c	30.792***
Non-Victimizing language	.60 (.491)	.63 (.486)	.60 (.491)	.69 (.471)	.396
Unifying Language	.24 (.430)^a	.67 (.473)^b	.59 (.493)^c, b	.58 (.504)^b, c	45.259***
Non-Emotive language	.54 (.499)^a	.48 (.502)^a	.40 (.491)^b	.62 (.496)^a, b	4.685**

Notes: Higher means represent the presence of higher peace frame indicators. Standard deviations are noted in parentheses. Different superscripts (a, b, c, or d) indicate statistically significant differences in Tukey’s post-hoc tests. Differences in means with the same superscript are not statistically significant. N = 841. **p < .001, ***p < .0001.

Post-hoc Tukey HSD tests showed that the US/Canadian media (M = 0.51, SD = .522) reported significantly more causes and consequences of the peace talks than their Colombian (M = .34, SD = .474) and European (M = .36, SD = .481) counterparts. The post-hoc test also revealed that European media incorporated significantly more solutions and similarities (M = .62, SD = .488) than the US/Canadian media (M = .41, SD = .493) and the Colombian news media outlets (M = .21, SD = .411), whereas the Colombian media incorporated less unifying language in their news reporting (M = .24, SD = .430) than European (M = .67, SD = .473) and US/Canadian media (M = .9, SD = .493). The US/Canadian media used more multi-party conflict reporting (M = .33, SD =  .470) than Colombian news providers (M = .21 SD = .408). Finally, results showed that European (M = .84, SD = .364) and the US/Canadian media (M = .72, SD = .452) did not label parties as good or bad more than the Colombian news outlets (M = .48, SD = .501).

The third research question aimed to find how audiences interact with and recommend different frames of peace and war on social media (see Table 7). Results of ANOVA tests showed significant differences when it came to Facebook shares [F (2, 838) = .11.147, p < .001] and Twitter shares [F (2, 838) = 10.402, p < .001]. Post-hoc Tukey HSD tests indicated that stories with the war frame received significantly more Facebook shares and Twitter shares than those stories that contained the peace frame.

Table 7. Comparison of mean number of social media recommendations by peace/war frames.

	Peace M (SD)	War M (SD)	Neutral M (SD)	F value
FB shares	603.53 (1233.76)^a	1073.33 (1613.57)^b	599.90 (1001.23)^a	11.147***
FB likes	1268.22 (2965.18)	1823.03(7393.52)	1046,68 (1915.82)	1.213
FB comments	402.60 (779.04)	524.32 (908.19)	457.08 (836.73)	1.757
TW shares	234.56 (495.76)^a	452.44 (782.62)^b	251.71 (534.78)^a	10.401***

Notes: Higher means represent more social recommendations. Standard deviations are noted in parentheses. Different superscripts (a, b, c, or d) indicate statistically significant differences in Tukey’s post-hoc tests. Differences in means with the same superscript are not statistically significant. N = 841. ***p < .001.

Hypothesis 3 posited that negative and positive emotive language will increase social media sharing of news articles about the Colombian peace process. Results of independent sample T-tests partially supported this hypothesis. In fact, negative emotive language significantly increased Facebook shares (M = 1009.29 SD = 1410.587) [t(839) = 1.978, p < .05], and Twitter shares (M = 522.28 SD = 806.088) [t(839) = 4.846, p < .001]. However, news articles with positive emotive language resulted in less Facebook shares (M = 418.70, SD = 1269,640) [t(839) = 5.176, p < .001], Facebook comments (M = 271.19 SD = 691.065) [t(839) = 4,007, p < .001], and Twitter shares (M = 194.99 SD = 396.342) [t(839) = 4,005, p < .001] than news articles that did not contain positive emotive language (Tables 8 and 9).

Table 8. Comparison of mean number of social media recommendations by negative emotive language.

	Negative emotive language
Social media recommendations	No — M (SD)	Yes — M (SD)	t-value
FB shares	796.825 (1469.77)	1009.291 (1410.58)	1.978*
FB likes	1596.457 (6896.742)	1452.62 (2078.922)	.334
FB comments	482.124 (910.81)	465.46 (743.53)	.261
TWT shares	280.937 (600.90)	522.28 (806.08)	4.845***

*p < .05; ***p < .001.

Table 9. Comparison of mean number of social media recommendations by positive emotive language.

	Positive emotive language
Social media recommendations	No — M (SD)	Yes — M (SD)	t-value
FB shares	1011.03 (1481.19)	418.70 (1269.64)	5.176***
FB likes	1751.69 (6573.17)	938.71 (2147.78)	1.753
FB comments	544.38 (899.74)	271.19 (691.06)	4.007***
TWT shares	411.36 (745.04)	194.99 (396.34)	4.005***

***p < .001.

Discussion

One of the main takeaways from this bilingual, cross-national study is that — in general terms — news media continue to use war frames more often than peace frames, even during the coverage of a peace negotiation. Almost 6 stories in 10 from our sample contained the war frame over the peace frame in stories that were shared, liked, and commented on via social media. Findings from this research suggest news media continue focusing on the ongoing conflict between two sides during peace process coverage in their need to use traditional models of reporting that incorporate drama and oversimplify complex events, as Wolfsfeld (2003) previously noticed during the Oslo negotiations between Israel and Palestine. The most salient indicators of the war frame in the coverage of the Colombian peace process were the tendency of privileging elite sources in the stories and emphasizing on two-party conflict reporting. War framing not only creates favorable conditions for opponents to the peace process but also privileges official voices and places actors on two opposing and irreconcilable sides, often contributing to the breakup of negotiations or escalation of the conflict (Cozma 2015). These salient war framing indicators are opposed to key peace reporting elements, such as including common people’s voices in stories, incorporating the point-of-view of multiple parties involved in the conflict, and providing context and historical background to understand the roots of the struggle (Galtung 1986, 1998; Lee and Maslog 2005; Lynch and McGoldrick 2006; Lynch 2007). Further, the current research found social media audiences amplified stories with war frames over peace frames, indicating that users, and not just journalists, seemed to think about the peace process in terms of conflict more than negotiation. These results are in accordance with previous research that shows violence and conflict often have been identified and used as a news frame or journalistic news value because those elements facilitate storytelling and help both journalists and audiences to make sense of complex social events (Bartholomé, Lecheler, and de Vreese 2017; García-Perdomo 2017). Such findings have implications for all proponents of peace journalism, as in this era of social media, it might not be enough to simply teach journalists how to use peace rather than war frames, as audiences have taken on the role of gatekeepers in diffusing conflict-oriented stories.

Building upon previous scholarship that suggests geographic and cultural proximity of a news event influences how that event will be covered (Galtung and Ruge 1965; Stevenson 1984), especially when it comes to conflict (Shapiro and Williams 1984), the present study advances peace journalism research by explaining the interplay between proximity and peace and war framing in times of negotiations. Like previous peace journalism studies, our paper showed that proximity is an important factor in whether reporters rely on more war or peace frames. This study found that more war frames were used in Colombian media than in the more distant outlets (i.e., Europe and US/Canada), which used more peace frames than the local national media where the conflict originated. Therefore, despite valuable Colombian media’s editorial initiatives such as Colombia 2020, Pacifista and Verdad Abierta, and previous findings of journalists’ noble intentions to contribute to peace building (Prager and Hameleers 2021), this content analysis shows that Colombian media news articles shared on social media contained more elements of war framing than articles shared from foreign media during the coverage of the peace process. These findings contradict Lacasse and Forster’s (2012) work on the Mexico drug war that found newspapers closer to the conflict used more peace than war frames. This perhaps can be explained by the fact that, unlike the Mexico drug war, the case of Colombia is about the end of a conflict and the negotiation of peace. It is also possible that the Colombian case is more directly attached to partisan politics than the Mexican case, and that the armed violence in Colombia has an important political component that the Mexican drug case lacks. It also is worth considering whether the conflict is so ingrained as a news value and as inherent in reporting practices in Colombian media, especially after more than 50 years of reporting on war, that national media approached the peace process in the same way they approached conflict coverage for a long time. In other words, Colombian journalists, unlike the foreign media, reported on the peace process with war frames because that is how they are accustomed to covering the relationship between the government, political opposition, and the FARC.

Looking in-depth at the individual peace- or war frame indicators also helps explain the role of proximity. This study showed that Colombian media focused on problems and differences, adopted two-party conflict reporting, labeled parties as good or bad, and used demonizing language more than media from the US/Canada or Europe. In contrast, the US/Canadian media used less here and now reporting, and European media focused more on visible effects of violence than Colombian news outlets. It follows that national media more than foreign media would be focused on the ins and outs of the obstacles of achieving peace, thus covering the problems and differences, while foreign media perhaps were more interested in a higher-level, more abstract view that would appeal more to a non-Colombian audience.

The Colombian media’s tendency of labeling of parties as good or bad and use of demonizing language also somewhat conflicts with previous research that suggests demonizing language is more common in distant-foreign than in close-national-local media (Galtung and Vincent 1992; Lacasse and Forster 2012). We argue that because this study was focused on a peace process — something foreign media would usually see as positive — it follows that non-Colombian media would use less labeling and demonizing language. Similarly, the foreign media used less here and now reporting than the Colombian media, again pointing to differences in audiences: non-Colombian audiences most likely would want overviews and explanations of the peace process, rather than the breaking news and details of every step of the process that would be more relevant to Colombian readers. More mentions of visible violence in European media also can be attributed to audiences, as non-Colombian readers need basic information about the conflict, including number of dead and displaced people, to understand the peace process, whereas Colombian audiences do not need to be reminded of the historical casualties in every story.

These same patterns are reflected in the findings when it comes to foreign versus Colombian media’s use of peace frame indicators. In other words, generally, the foreign media incorporated more elements of peace frames than did Colombian media, and Colombian media more than foreign media relied on war framing. While previous research suggests that foreign media would take a more negative approach to conflict than national media, this study’s findings indicate that when it comes to the end of a long conflict, there is a role reversal and foreign media more than national media prioritized peace framing.

These findings highlight the role of proximity as a news value and underscore the complexity of proximity when it comes to peace journalism. Further, these findings are useful for helping explain why the world outside of Colombia was surprised when Colombians voted against the peace referendum: the use of peace framing in foreign news coverage perhaps gave the impression than peace was on the horizon, while in Colombia, the preponderance of war framing in news coverage perhaps contributed the no vote. Finding about proximity, however, need some nuance in the Colombian case because there is previous evidence that shows that hyper-local and community media in that country have more context and cultural proximity to the conflict than commercial and urban media, and that community media like radio stations have had an active role in peace building (Rodríguez 2011).

Additionally, for social media recommendations, our results show that social media users on Facebook and Twitter were more likely to amplify the war frame. Our research shows the war frame was shared almost twice as much on these platforms, indicating that social media audiences may have an affinity, even if just by mere attraction, for war narratives. On the one hand, the functions of the war frame are highly susceptible to emotional and sensational aspects of conflict and propaganda. It is possible that the latent emotional appeals of components like victimization and demonization are key factors for sharing. On the other hand, if the war frame is indicative of partisan presentation, and the peace frame removes the powerful emotions and consequences of victimization that are affiliated with war, then it is possible that social media audiences would advance narratives that are more oriented toward war than peace, given the emotional characteristics that conversations in those platforms have. Our results show that positive emotive language was suppressed by social media audiences, and articles without language of hope and reconciliation were amplified. We contend that the longevity of a 50-year war may have created skepticism toward positive narratives, despite advancements in the peace process. Previous research on the effects of war framing has shown audiences who are exposed to the war journalism frames are angrier and more fearful than media consumers exposed to peace frames and who are more inclined to seek solutions and reconciliation (Lynch and McGrodrick 2012). With this in mind, even though it is difficult to draw conclusions about causality from this study, we argue that the war frame reporting and the social media sharing of conflict news could have contributed to the failure of the peace referendum on Oct 2, 2016, and the current difficulties the country is facing to implement peace agreements. Ultimately, our work extends the research on emotions and shareworthiness (e.g., Berger and Milkman 2012; Bright 2016; Kilgo, Lough, and Riedl 2017) by showing that in the context of war, valence of emotional content is important. Negative emotional frames and devices are likely to prompt more sharing on Facebook and Twitter than positive or neutral coverage.

Finally, the current research applies Wolfsfeld’s political contest model (1997) — which theorizes that authorities and antagonists compete for media attention to achieve political influence and promote their frames in times of conflict — to the specific case of the Colombian peace negotiations. The main purpose of adopting this model was to understand the relationship between articles that privilege supporters/contradictors to peace negotiations and the adoption of peace and war frames. As a result, this paper showed that articles that privilege the point of view of antagonists to the Colombia peace process adopted more devices of war journalism reporting than news pieces that favored perspectives that supported the peace talks. Therefore, antagonists’ attempts to promote their frames of the conflict to the news media and mobilize political support for their cause (Wolfsfeld 1997) can be reflected in the use of war journalism reporting while news articles that privileged supporters’ viewpoints of the talks were predisposed to adopt peace journalism elements. Looking at the Colombian context, the Santos’ government and the FARC guerrillas were in favor of the peace process (although with several disagreements about the negotiations) whereas other political forces of the country, like the right-wing Democratic Center Party, were openly against the negotiations and the agreement. These findings could help future peace journalism scholars to understand the role that the news media play framing stories with peaceful or conflicting devices during peace negotiations, promoting the visions of antagonists and supporters, and contributing to maintain or undermine peace talks.

This research is limited in that it analyzes only a sample of news stories shared on social media, so that any conclusions made are restricted to articles that have produced interactions on social networks. Therefore, conclusions are restricted to content shared both in English and Spanish on social media. If the study sample had been extended to other articles and languages, the manuscript would have offered different results. Another important limitation of this study is that it compares articles produced by diverse media organizations from different countries and regions without considering differences in journalistic cultures, professional values, and practices. Although context and culture could affect the frames analyzed and interactions of different audiences, the current research is valuable because it shows key patterns that make results generalizable and valuable in a global digital environment. It also is important to note that peace journalism needs further conceptualization and theorization to make some of its war and peace frames more appropriate to the context of peace negotiations (Shinar 2009). Future research should examine how peace journalism operates in relation to other dimensions of the Colombian conflict such as the current implementation of the peace accord, local vs. national media treatment of victims and ex-combats reincorporation to civil life. Finally, it will be relevant for future researchers in this area to measure the effect of social media frames on voters’ decisions against and in favor of the peace referendum.

Despite these limitations, this study is important because it contributes to the peace journalism scholarship in three main ways. First, by showing that not only is war framing prevalent in conflict reporting, as prior studies have noted, but it also prevails in coverage of peace negotiations. Future research should explicate this further, exploring the journalistic norms and practices that might influence this finding. Second, our research also points to proximity as a key factor in whether coverage will employ a war or peace frame. Unlike previous studies that showed distance as contributing to war-oriented reporting, our study’s findings were more complex, suggesting that local media approached peace negotiations with a war frame more than foreign media, perhaps because for them, the stakes were higher during the long conflict. Lastly, this study advances sharing research by showing that social media audiences engaged with war-oriented stories more than peace-oriented stories. Combined, these findings are noteworthy for their implications about how national media, unlike foreign media, consistently emphasized a war frame that social media users amplified, which we argue has implications for how citizens viewed the Colombian peace process, ultimately potentially influencing the decision to vote down the referendum.

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Notes

1 Significant events occurred during the timespan selected to collect the news articles. First, the Colombian government and the FARC guerrilla announced a permanent cease-fire on June 22nd, 2016. Second, on August 24th both parties reached a final peace agreement in Havana after four years of talks. Third, on September 26, the peace accord was sealed in an official ceremony in Cartagena. Finally, on October 2, 2016, citizens narrowly rejected the peace agreement in a plebiscite convened by the government.

2 The war frame indicators are: Here and now, visible effects of violence, elite sources and voices, two-party conflict reporting, problems and differences, labeling parties as good or bad, victimizing language, demonizing language, and emotive language.

3 The peace frame indicators are: Causes and consequences, invisible effects of violence, common people as sources, multi-party conflict reporting, solutions and similarities, not-labeling parties as good or bad, non-victimizing language, unifying language, and non-emotive language.

4 Coders were trained to identify demonizing words or devices when describing actors, the peace process, and the ongoing conflict between political parties (e.g., backward, uncivilized, threatening, vicious, cruel, brutal, barbaric savage, ruthless, villains, irrational, cruel, despotic, untrusted, heartless, narco-terrorists, criminals, mockers, despotic, traitor, liar, unpatriotic, effeminate, etc.).