SaveDonbassPeople: Twitter, Propaganda, and Conflict in Eastern Ukraine

This is a post-print version of the article published in ​The Communication Review ​and accessible via the following link: https://www.tandfonline.com/doi/abs/10.1080/10714421.2015.1085776 To cite this article: Mykola Makhortykh & Yehor Lyebyedyev (2015) #SaveDonbassPeople: Twitter, Propaganda, and Conflict in Eastern Ukraine, The Communication Review, 18:4, 239-270, DOI: 10.1080/10714421.2015.1085776 #SaveDonbassPeople: Twitter, Propaganda, and Conflict in Eastern Ukraine Mykola Makhortykh and Yehor Lyebyedyev ABSTRACT In this article we explore the use of the #SaveDonbassPeople hashtag in the context of an online protest campaign against the military operation in Eastern Ukraine. The campaign was initiated by anti-government activists, but soon became contested by supporters of the Ukrainian government, turning Twitter into an online battleground. Our findings suggest that in the course of the campaign, Twitter was predominantly used as a propaganda outlet to broadcast opposing views on the ongoing conflict. INTRODUCTION In the second half of May 2014, the official site of the Dnieper-Donetsk Workers Union (DDWU) asked its Russian audience to take photos of children and share them on Twitter alongside the #SaveDonbassPeople hashtag in order to stop “the aggression of the bloody Kiev junta” (“#SaveDonbassPeople!,” 2014). The announcement was followed by practical recommendations, including ways of finding photogenic children and making them look miserable, as well as instructions on how to make one’s Twitter profile seem more Donbass-like. However, the true purpose of this propaganda guide was spoiled by its appearance on the website of a fictional Ukrainian political party, members of which had already achieved prominence as a result of their online subterfuges. Despite all these clues, a number of web users took the announcement seriously, citing it as evidence of the “information war” (vita_linka, 2014) against Ukraine. This joking announcement of the DDWU, together with the rather serious reaction from the Internet audience, provides insight into the deep connection between Twitter and the Ukrainian crisis. Twitter is a popular microblogging service that in recent years has become associated with political protests and civil campaigns. Commonly viewed as a tool of organization and mobilization, Twitter has been used by public activists in various geographical and social contexts such as Moldova, Iran, and the United States. In all these cases, Twitter mainly served as a platform for protesters, which led a number of journalists and academics (e.g., Breuer, Landman, & Farquhar, 2014; Eltantawy & Wiest, 2011; Friedman, 2014; Sullivan, 2009) to identify Twitter as an influential example of the democratizing impact of digital media on the political sphere. This optimistic assumption, however, does not necessarily hold true when set against the empirical reality of online protest research. A number of studies (Aday et al., 2010; Lysenko & Desouza, 2012; Morozov, 2009a; Vicari, 2013) suggest that Twitter’s impact on protest campaigns is easily overestimated, as the platform is more often used not as an organizational mechanism, but as an information outlet. Furthermore, even information campaigns on Twitter are not necessarily free from the shortcomings of mainstream media such as the disproportionate impact of a few influential information brokers (Lin, Keegan, Margolin, & Lazer, 2014) or the subversion of modern technology by authoritarian regimes for propaganda purposes (Morozov, 2009b). In this article we explore the use of Twitter by activists and bystanders during the conflict in Eastern Ukraine. In order to do so, we focus on the #SaveDonbassPeople hashtag, which was used during an online campaign that started in the second half of May 2014 and continued until the end of June. The campaign was initiated as a call to protect people from Eastern Ukraine from the Ukrainian government, but soon came to be contested by pro-government advocates, who attempted to provide alternative coverage of the ongoing events. The resulting tensions turned Twitter into an online battleground where each side tried to mobilize public support and promote its interpretation of the conflict in Eastern Ukraine. The article begins with a background section that introduces the context of the conflict in Eastern Ukraine. It is followed by a discussion of previous studies on the use of Twitter during public protests as well as methods of data collection and analysis used in our research. It then provides an overview of our findings, starting with an exploration of the use of the #SaveDonbassPeople hashtag during the campaign and a discussion of the kinds of actors involved, and ending with an investigation of strategies used by pro- and anti-government advocates in the course of the campaign and an examination of frames employed for presenting and interpreting the conflict. The article concludes by discussing our findings and the potential shortcomings of this research, as well as the possibilities for further study. BACKGROUND The Ukrainian crisis of 2013–2014 started with a series of protests against the suspension of signing the Association Agreement between Ukraine and the European Union in November 2013. After the brutal dispersal of protesters in Kiev on November 30, the scale of protests significantly increased and calls for European integration became overshadowed by demands for the resignation of President Yanukovych. Public unrest in Ukraine reached its peak in February 2014, when violent clashes between protesters and pro-government forces led to the flight of Yanukovych and the formation of a new interim government. The crisis, however, continued as the center of unrest shifted to Eastern and Southern Ukraine, which were traditional bases of support for Yanukovych and his Party of Regions. During the Euromaidan protests a number of pro-government rallies (known as anti-Maidans) took place in those regions, often leading to clashes between pro- and anti-government activists. The crisis of legitimacy of the new Ukrainian government as well as a negative perception of anti-Yanukovych protests and fear of disruption of existing ties with Russia, which were historically strong in Eastern and Southern Ukraine, led to a new wave of unrest in February and March 2014. The declaration of independence of Crimea on March 11, which was followed by the accession treaty between Crimea and the Russian Federation, amplified anti-government1 protests in mainland Ukraine and, especially, in the Donetsk and Luhansk oblasts, which both belong to the historical Donbass region. On April 6 protesters seized the buildings of regional state administrations in Donetsk and Luhansk and proclaimed the formation of the Donetsk and Luhansk People’s Republics, which were later united into a self-proclaimed confederation of Novorossiya. On April 7 the acting President of Ukraine, Oleksandr Turchynov, announced that “anti-terrorist measures” (“Turchinov objavil,” 2014) will be taken against armed insurgents, who are seizing administrative buildings in Donetsk and Luhansk. As the situation in Eastern Ukraine continued to deteriorate, on April 14 Tyrchynov authorized an antiterrorist operation (ATO), which led to a number of clashes between the Ukrainian army and anti-government groups.22 The latter, however, continued to expand the territory of People’s Republics, capturing administrative buildings in Sloviansk, Kramatorsk, Antratsyt, and several other cities in Eastern Ukraine. After a number of bloody confrontations between pro-Kiev and pro- Novorossiya supporters—particularly, in Odessa and Mariupol—tensions in Ukraine continued to escalate. The unrecognized referendums that took place in the People’s Republics on May 11 resulted in the majority of votes for independence from Kiev and the subsequent demands to withdraw pro-Kiev troops from the territory of Novorossiya. The Ukrainian government rejected these demands that led to intensification of fighting after the Ukrainian presidential elections on May 25, when the Ukrainian army managed to push back the Novorossiyan forces and recapture several cities in Eastern Ukraine. The transformation of sporadic clashes into a full-fledged military conflict caused an outcry in social media, including Twitter. Although Twitter remains less popular among Ukrainians than other social networking sites, such as Vkontakte or Odnoklasniki, its audience totaled more than 430,000 in April 2014 (Yandex, 2014). Close associations between the Euromaidan protests and Twitter contributed to the rapid growth of its Ukrainian audience in December–January (Yandex, 2014) and the propagation of the image of Twitter as a key platform behind the protest (Bohdanova, 2013). These reasons explain the decision of Novorossiya supporters to initiate a campaign on Twitter against the use of military force in Eastern Ukraine. The campaign, however, was almost immediately contested by pro-government activists, leading to an online confrontation that is examined in this article. . LITERATURE REVIEW Various studies have established links between Twitter and protest campaigns in different parts of the world. Since 2009, when Twitter was extensively used by anti-government As the Ukrainian government changed from pro-Russian to pro-Western in February 2014, government proponents and opponents also changed their roles. Henceforth, the “anti-government” term refers to pro-Russian opponents of post-Euromaidan Ukrainian government. 2 These pro-Novorossiya groups included a number of organizations, such as the Russian Orthodox Army, the “Oplot” group, the “Vostok” battalion, and several companies of Don Cossacks. 1 activists in Moldova (Mungiu-Pippidi & Munteanu, 2009) and Iran (Burns & Eltham, 2009), the platform has been associated with protest activities. In the years since, Twitter has played an important role during protests in Tunisia (Breuer et al., 2014), Egypt (Attia, Aziz, Friedman, & Elhusseiny, 2011), Russia (Nikiporets-Takigawa, 2013), Turkey (Genç, 2014), and Venezuela (Aguado, 2014). In addition to these “Twitter revolutions” (Morozov, 2009a), the platform has also been used by activists involved in social (Tremayne, 2014) and environmental movements (Segerberg & Bennett, 2011), which further emphasizes the growing role of Twitter in the public sphere. Although it is recognized that Twitter impacted all these protest campaigns, the exact nature of this impact is evaluated differently. For instance, Tufekci and Wilson (2012) argue that Twitter, together with other social media platforms, played a central role in public mobilization during protests in Egypt. In their study of the protests in Tunisia, Breuer, Landman, and Farquhar also claim that digital media “provided the basis to construct a national collective identity supportive of protest action” (2014, p. 30). The same conclusions are drawn by García, Chauveau, Ledezma, and Pinto (2013) in their study of Chilean student movements, in which they point to the direct relationship between Twitter and activity on the streets. These optimistic claims about the significant impact of Twitter on protest movements are often countered by more skeptical assessments. It is hardly debatable that Twitter streams can serve as “cross-cutting transmission belts” (Segerberg & Bennett, 2011, p. 203) that transcend personal networks and connect multiple actors, which have different ideological and organizational identities. Yet, while Twitter allows activists to expand their social networks quickly, these newly established ties tend to be weak and not so useful for public mobilization (Gladwell, 2010). A number of studies of protest campaigns show that these weak ties limit the potential of Twitter to marshal public support. According to Lysenko and Desouza (2012), Twitter was not used during the first stage of the Moldova revolution because, for their mobilization efforts, activists relied not on an online campaign, but on a network of offline youth organizations. Similar observations are made by Morozov (2011), who argues that social networking sites had a limited impact during the Arab Spring when compared to offline activist networks. Finally, in their research on protest movements in the United States, Spain, and Greece, Theocharis, Lowe, Van Deth, and Albacete (2013) demonstrate that only a few tweets called on people to take part in protests, and these calls came from a small group of committed activists. These divergent assessments of Twitter’s role in protest campaigns highlight the need for additional empirical research on the use of the platform by protest movements. A number of studies (Freedman, 2014; Gruzd & Tsyganova, 2014; Onuch, 2015; Szostek, 2014) suggest that social media have had a significant impact on the Ukrainian crisis; yet in order to assess how social media, in particular Twitter, were used by protest movements in the course of the crisis, we need to explore concrete cases, such as the #SaveDonbassPeople campaign. Furthermore, we suggest that not every #SaveDonbassPeople tweet was produced by activists as a part of the campaign and a number of bystanders could also use the hashtag for their own purposes. Thus, the first research question we would like to formulate is: How was the #SaveDonbassPeople hashtag used by activists and bystanders in the course of the campaign? One peculiar feature of the #SaveDonbassPeople campaign was a confrontation between activist groups that pursued different goals in the context of the conflict in Eastern Ukraine. Although a few studies (Lynch, Freelon, & Aday, 2014; Nikiporets-Takigawa, 2013; Radchenko, Pisarevskaya, & Ksenofontova, 2012) discuss consequences of such conflicts of interests, the use of Twitter by competing activist groups seems to be underproblematized. In their study on the use of Twitter in the Syrian conflict, Lynch et al. (2014) argue that the presence of activist groups with conflicting purposes led to over- or underrepresentation of certain points of view on the conflict as advocates of different sides attempted to mobilize public support. Similar observations come from the studies of Russian protests (Nikiporets-Takigawa, 2013; Radchenko et al., 2012), which examined different strategies used by opposing activist groups for discrediting their opponents. Thus, the second research question we would like to formulate is: What strategies were used by proand anti-government advocates to mobilize public support and/or discredit their opponents? The last aspect of the #SaveDonbassPeople campaign that we examine is its use for framing the conflict in Eastern Ukraine. According to Noakes and Johnston (2006, p. 2), frames are essential for social movements, because they “indicate what is going on and why it’s important,” and, thus, allow activists to explain relevance of collective action and motivate individuals to act. A number of studies (Lysenko & Desouza, 2012; Meraz & Papacharissi, 2013; Vicari, 2013) demonstrate how Twitter is used by activists for disseminating information about protests and broadcasting their identity to the world. A few scholars (Fisher, 2010; Morozov, 2009a), however, problematize the use of Twitter for framing protests by questioning the reliability of activists’ interpretations. Similarly, Lynch et al. (2014) argue that instead of providing a comprehensive view on events, social media can support particular narratives that are curated by small groups of activists. Based on these divergent assessments, we suggest that although Twitter can be used to bypass the gatekeepers of traditional media and provide an alternative view on events, it can also be used as a propaganda outlet by activists in their struggle against government or other activist groups. Thus, our third research question is: How did supporters of opposing camps frame the conflict in Eastern Ukraine through the #SaveDonbassPeople campaign? METHODOLOGY Data Collection For this study we collected tweets that included the hashtag #SaveDonbassPeople as well as its derivatives, which constitute all words that begin with the keyword (for instance, #savedonbasspeoplefrom). The data set includes 81,404 tweets, which were collected in real-time mode through the Streaming Twitter API between May 28 and June 12 2014. We decided to use the Streaming Twitter API for two reasons: first, it provides sufficient3 data on 3 The Streaming Twitter API returns around 1% of all the tweets produced at a given time (Morstatter, Pfeffer, Liu, & Carley, 2013), yet because of a relatively small number of #SaveDonbassPeople tweets, we assume that it was sufficient for collecting the majority of tweets on this topic. the use of the #SaveDonbassPeople hashtag on Twitter; second, it is available free of charge. Although it also has certain shortcomings (e.g., the impossibility of acquiring data that appeared before the beginning of data collection and the necessity to gather data in real time), we assumed that these disadvantages are compensated by its benefits. The starting date for data collection corresponds to the first calls for resistance to the “information war,” which appeared on Internet resources that supported the antiterrorist operation in Eastern Ukraine (Information Resistance, 2014b). These calls labeled videos published with the hashtag #SaveDonbassPeople as a “new technique for manipulating human consciousness” and urged audiences to create “alternative videos” (Sloviansk, 2014). Based on these appeals, we identified the beginning of confrontation around the #SaveDonbassPeople hashtag and started collecting tweets in real-time mode. Two brief moments when the process of collection was interrupted for technical reasons were marked as breaches of the line graph in Figure 1, which shows Twitter activity for the observed period. We also used data provided by the Kribrum company4 to determine that the #SaveDonbassPeople hashtag was first used on May 26, in connection with a discussion of Ukraine’s potential federalization. Data Analysis THE USE OF THE #SAVEDONBASSPEOPLE HASHTAG We employed a combination of methods to explore how the #SaveDonbassPeople hashtag was used during the period of study. We started with a temporal analysis of the #Save Donbass People activity, using six-hour periods as units of time for a breakdown of our data set. In addition to counting the overall number of tweets per period of time, we plotted the number of retweets per each period and for those cases, when more than 1,000 retweets originated from one source, identified its political affiliation (i.e., ATO or Novorossiya). We chose 1,000 retweets as a threshold value, because before it the number of retweets decreased exponentially, whereas afterward the decrease became linear. The analysis of temporal patterns was followed by the identification of the most active and the most influential users who produced messages with the #SaveDonbassPeople hashtag. As an indicator of a user’s activity we used the sum of sent tweets and received retweets. Based on this metric, we identified the top 50 users, who collectively produced 46,998 tweets, which constituted around 58% of our data set. Then, we classified these users using a simplified version of the classification schema that was developed by Lotan et al. (2011) for the study of protests in Tunisia and Egypt. 4 A commercial company that collects and analyzes Twitter feeds in Russian. More information is available on the company’s website, ​http://www.kribrum.ru/ Our schema included the following types of users: (a) activists: individuals who either identify themselves as activists or tweet purely about activist topics (such as the conflict in Eastern Ukraine or anti-government protests); (b) bloggers: individuals who either identify themselves as bloggers or tweet on a variety of topics, including nonactivist ones; (c) bots: accounts that produce automated posts on Twitter (often for commercial purposes); (d) celebrities: individuals who are famous for reasons unrelated to activism (e.g., artists or politicians); (e) mainstream media organizations (MMOs): news and media organizations; (f) think tanks: research organizations that are focused on Eastern Europe. In addition to determining users’ types, we also classified users by their affiliation—that is, whether or not a user supports the ATO, Novorossiya, or tries to stay neutral. Both classifications were produced by two independent coders, who studied users’ profiles, their recent activity, and sites linked in their profiles. Krippendorff’s alpha was counted for both classifications to ensure inter-coder reliability; resulting rates for these classifications (as well as the following ones) are listed in Table 1. In the case of discrepancies, two original coders discussed and consensus-coded them; the same procedure was used for all other classifications in this article. For examining the most influential users in terms of the number of followers, we selected all users who had more than 100,000 followers before the beginning of the #SaveDonbassPeople campaign. This threshold value was chosen because before it the number of followers decreased exponentially, whereas afterward the decrease became linear. We selected the number of followers instead of other indicators of influence (e.g., the number of mentions or replies), because we were interested in a user’s ability to communicate the message to the large number of followers instead of his/her visibility inside our data set. We used the same classification schemata as for the most active users; both classifications were again produced by two coders, who used the same sources of data. TABLE 1 Krippendorff’s Alpha Rates for Inter-Coder Reliability Classification Rate Active users—users’ political affiliation 0.82 Active users—users’ types 0.84 Influential users—users’ political affiliation 0.86 Influential users—users’ types 0.93 Content analysis—tweets’ language 0.94 Content analysis—tweets’ political affiliation 0.89 Content analysis—tweets’ types 0.87 Content analysis—external resources’ types 0.91 Hashtag co-occurence—hashtags’ language 0.98 Hashtag co-occurence—hashtags’ function 0.96 External materials—materials’ types 0.81 Finally, we employed content analysis for examining tweets with the #SaveDonbassPeople hashtag. We used a random sample of 1,024 tweets for achieving the confidence level of 99% with the error level of 4%. Then, two coders classified these tweets with a number of schemata, which identified the language of a tweet, its political affiliation, its type, and the type of an external resource referenced (if any). The language classification schema included the following options: (a) English; (b) Russian; (c) Ukrainian; (d) other languages. For the classification of tweets’ affiliations we employed the same schema that was used earlier for classifying actors’ affiliations: (a) ATO; (b) Novorossiya; (c) neutral. Our classification of tweets’ types included the following options: (a) comments: tweets that express users’ personal views and often contain emotional statements; (b) conversations: tweets that refer to other users and include their usernames prefixed by the “@” symbol; (c) hashtag only: tweets that consist only of hashtags; (d) news: tweets that share updates without providing personal comments; (e) retweets: tweets that are written by other users and reposted without changes; (f) spam: nonsensical messages that are probably generated automatically. The last classification schema was used because of the large number of tweets with links to external resources (864 out of 1,024). The schema included the following types of resources: (a) news sites: platforms for publishing and sharing news (Novosti Donbassa, Russia Today); (b) social networking sites: platforms for building social networks (Vkontakte, Twitter); (c) video hosts: platforms for sharing and distributing video materials (YouTube, Smotri-tube); (d) other: websites that do not fit either of previous categories (GlobalResearch, Change.org). STRATEGIES OF PRO- AND ANTI-GOVERNMENT ADVOCATES We explored two strategies used by pro- and anti-government advocates during the #SaveDonbassPeople campaign: the use of auxiliary hashtags and the addition of external content. In order to examine the first strategy, two coders classified all the hashtags that appeared in more than 10 tweets, according to their language and function. We chose the threshold value of 10 tweets, because below it the number of auxiliary hashtags increased in a geometric progression; furthermore, many hashtags that were used in nine or fewer cases were misspellings of other hashtags. The language classification schema included the following options: (a) English; (b) Russian; (c) Ukrainian; (d) other languages; (e) undefined.5 The function classification schema included the following types of auxiliary hashtags: (a) infiltrating: pro-ATO hashtags that diminished the impact of the #SaveDonbassPeople hashtag; (b) informative: nonaffiliated hashtags that served an informative purpose; (c) reinforcing: pro-Novorossiya hashtags that enhanced the impact of the #SaveDonbassPeople hashtag. 5 These hashtags cannot be attributed to a particular language, because they are written in the same way in different languages. Examples include abbreviations such as #днр (the abbreviation for the Donetsk People’s Republic, which is the same in Ukrainian and Russian) and single words such as #майдан (the word for “maidan” in Ukrainian and Russian). For examining the second strategy we used the same random sample of 1,024 tweets as before; however, at this time we classified content that was added to tweets through external links to overcome the limitation of 140 characters. Two coders classified external content into the following categories: (a) amateur footage: videos produced by nonofficial monitors and direct witnesses; (b) demotivational posters: images accompanied by verbal texts that comment on the images’ content; (c) news reports: video reports produced by professional journalists; (d) photos: real-world images without a verbal commentary; (e) selfies: self-portrait photos accompanied with the #SaveDonbassPeople slogan; (f) texts: verbal records of various kinds (e.g., posts, articles); (g) other: types of content that do not fit any other category; (h) deleted: content that was deleted. FRAMING OF THE EASTERN UKRAINIAN CONFLICT In order to examine how the #SaveDonbassPeople hashtag was used for framing the conflict in Eastern Ukraine we employed qualitative content analysis. Based on a close reading of tweets from the random sample of 1,024 messages that were used in other sections of our study, we explored how advocates of Novorossiya and ATO used selected patterns of presentation and interpretation to promote their views on the conflict in Eastern Ukraine. FINDINGS The Use of the #Savedonbasspeople Hashtag TEMPORAL DYNAMICS Outside its context, the #SaveDonbassPeople hashtag has positive emotional connotations, because it emphasizes the value of people’s lives and calls for peace. However, temporal analysis showed that peaks in its usage corresponded to active phases in the confrontation between the Ukrainian army and the Novorossiyan militia. The highest peaks in the hashtag’s use (peaks 1 and 7, Figure 1) coincided with the battle in Alexandrovka (“Antiterroristicheskaja Operacija,” 2014), and the capture of Krasnyi Liman by the Ukrainian army (“Operacija po Antiterroru,” 2014). The latter event marked the end of the large-scale offensive of the ATO forces in the first half of June, and the beginning of a series of counter-operations of the Novorossiyan militia. The decrease in the use of the hashtag after peak 7 can be attributed to the end of the government’s offensive and a shift toward local skirmishes. One particular feature of the hashtag’s use is its dependence on retweeting. Almost 90% of #SaveDonbassPeople tweets were retweets, unlike observed patterns of Twitter activity that usually assume a much smaller proportion of retweeted content (Boyd, Golder, & Lotan, 2010). Even in the case of natural disasters, when retweeted content accounts for a significant proportion of messages, retweets usually constitute only around 40%–50% of all messages (Bruns, Burgess, Crawford, & Shaw, 2012). The main peaks in the use of the #SaveDonbassPeople hashtag were related to messages from several users, who received the large number of retweets in turns. This pattern of taking turns is particularly noticeable in the period between peaks 3 and 7, when three pro-Novorossiya users received the large number of retweets one after another within a short period of time. One significant exception to this “athletic relay” pattern is represented by the pro-government user @euromaidanpr, whose activity was consistent during the whole period of study. Unlike anti-government advocates, whose activity peaked during the offensive operations of the Ukrainian army, peaks in the @euromaidanpr activity corresponded to the battle around the office of the Ukrainian border guard service in Luhansk (“Antiterroristicheskaja Operacija,” 2014), and the crash of a Ukrainian An-26 transport plane near Sloviansk (“Operacija po Antiterroru,” 2014). The relationship between peaks and influential sources of retweets indicates that the #SaveDonbassPeople campaign was propelled by a few individual users from pro- and anti-government camps. Based on differences in the use of the hashtag by ATO and Novorossiya advocates, we can identify two different approaches to the campaign. In the case of ATO supporters, we observed an organized campaign, which involved a group(s) of users, as in the case of the Information Resistance project.6 This campaign was accentuated through a single Twitter account—@euromaidanpr—that was registered before the beginning of the campaign and probably managed by several users who were engaged in continuous online activity during the whole period of study. In contrast, the activity of Novorossiya supporters involved several individual users who seemingly did not coordinate their actions with each other. In several cases peaks of retweets of messages from prominent Novorossiya advocates overlapped; furthermore, none of these users was able to sustain a steady stream of retweets like @euromaidanpr. Instead, after achieving one or two significant peaks of retweets, pro-Novorossiya users usually lost their impact on the campaign, whereas in the case of a centralized campaign we would expect sustainable promotion of influential accounts. TABLE 2 The Number of Followers of the Most Influential Users Account name Number of followers (May 28, 2014) Number of followers (June 12, 2014) euromaidanpr 34,359 34,502 krobzadrot 16 16 lowmaintainlife 623 3,989 donbasspeople 0 122 nash_slavyansk 1,459 2,696 ruredaktor 423,674 423,570 newsbalkan 4,249 4,302 This activist-driven project aims to “counteract external threats to the informational space of Ukraine” (Information Resistance, 2014a), and serves as one of the main information sources about the ongoing conflict in Eastern Ukraine. 6 novorussia2015 4,385 5,419 These different approaches toward the #SaveDonbassPeople campaign also influenced the way in which pro- and anti-government activists approached their audience. Based on the data about the number of followers of the most influential users in the beginning and in the end of data collection (Table 2), we suggest that ATO supporters mainly targeted the existing audience because the number of followers of @euromaidanpr remained stable through the whole campaign. Although some of the pro-Novorossiya users (i.e., @ruredaktor) were also focused on the existing audience, the majority of them experienced rapid growth in the number of followers. Unlike ATO supporters, who relied on the preexisting audience, many Novorossiya advocates were relatively inactive on Twitter before the campaign, and thus were forced to engage potential followers more actively. ACTORS The classification of the most active users (Table 3) suggests that activists—not media organizations or celebrities—played the leading role in the #SaveDonbassPeople activity. Similar to the protests in Egypt and Tunisia (Lotan et al., 2011), two types of users—bloggers and activists—were the most active in sending the message across Twitter; however, in the case of the #SaveDonbassPeople campaign the level of interaction between these two user groups seems to be minimal. TABLE 3 Types and Affiliations of Active Users Activists Bloggers Bots Celebrities MMOs Think-tanks ATO 0 0 0 0 1 0 Neutral 0 0 1 0 1 1 Novorossiya 25 17 4 0 0 0 One important reason behind this lack of cooperation is that in our case bloggers and activists operated in different ways. Many bloggers were English- and Spanish-speaking individuals who openly published their names and/or links to their other projects. In contrast, the majority of activists were Russian-speaking individuals who preferred to stay anonymous or semi-anonymous. Even while activists were busy tweeting and retweeting messages in English, their audience remained largely Russophone, as shown by the discussions in activists’ profiles. Although anti-government activists were particularly vigorous in using the #SaveDonbassPeople hashtag, it was also used by other types of actors, some of whom were only partially affiliated with one of opposing camps or showed no political affiliations. The case of two MMOs (@vladtime and @spilnotvenglish) is particularly illustrative: even while one—@spilnotvenglish—was obviously sympathetic to the Ukrainian army, both of them used the hashtag mostly for sharing recent updates about the conflict instead of persuading their audience to support a particular side. This observation—together with the similar use of the hashtag by a foreign think tank (@geopolitics_by)—indicates that some of #SaveDonbassPeople messages were not affiliated with the online campaign and that the hashtag was used not only by advocates of ATO and Novorossiya, but also bystanders, who probably employed it to disseminate information about the conflict in Eastern Ukraine. The classification showed that the activity of Novorossiya advocates was more noticeable, because only one actor among the most active ones was supportive to the ATO cause. Other pro-ATO users published fewer numbers of tweets, but—as in the case of @euromaidanpr—this does not necessarily mean that they were less influential. It is worth noting that only two pro-Novorossiya users—@lowmaintainlife and @donbasspeople—from those behind the peaks of #SaveDonbassPeople activity (see Figure 1) belonged to the 50 most active users. This discrepancy can be explained by the difference between more active users, who publish the largest numbers of tweets, and more influential users, who receive the largest number of retweets, which is a common pattern in Twitter activity (Bruns & Stieglitz, 2013). However, this observation casts a measure of doubt over the view of Twitter as a platform that empowers grassroots activists: even while some of them published hundreds of messages during the #SaveDonbassPeople campaign, this does not necessarily mean that these messages were read by other users or spread across the platform. The discrepancy between users’ activity and influence explains why both camps tried not only to produce large numbers of messages, but also to involve popular Twitter accounts in the campaign, including those users whose everyday activity was distant from the campaign’s subject. As shown in Table 4, eight out of 23 classified accounts participated in the campaign on the Novorossiya side. The majority of them belonged to Russian public figures, such as Nikolai Valuev, Oleg Gazmanov, and Ivan Okhlobystin. The most recognizable accounts that were sympathetic to the ATO cause belonged to Ukrainian activist groups (@appleip3) and media organizations (@5channel and @ukrpravda_news); however, similar to @spilnotvenglish, the latter accounts used the #SaveDonbassPeople hashtag mainly for reporting events in Eastern Ukraine instead of counteracting the anti-government campaign. Accounts that were not related to Eastern Europe as well as those accounts that were used to publish prepaid advertisements were identified as neutral. An important feature of messages published by neutral accounts was the presence of links that usually lead either to YouTube videos or profiles of other Twitter users. TABLE 4 Types and Affiliations of Influential Users Activists Bloggers Bots Celebrities MMOs Think-tanks ATO 2 0 0 0 2 0 Neutral 0 10 0 0 0 0 Novorossiya 0 4 0 5 0 0 These observations allow us to suggest that the #SaveDonbassPeople campaign could involve at least three potential Twitter audiences: Russophone users (through the accounts of Russian celebrities and individual activists), Ukrainophone users who supported the Euromaidan protests and the new Ukrainian government (through the accounts of pro-Western media organizations and activist groups), and Anglophone Twitter users (through the accounts of popular bloggers). Based on the number and types of users involved in the campaign, we suggest that in the first and the third cases the main goal of the campaign was to mobilize public support on behalf of Novorossiya by attracting attention to the conflict in Eastern Ukraine, whereas in the second case the main purpose was to provide alternative coverage of the conflict. CONTENT Based on affiliation and language classifications (Table 5), we suggest that the #SaveDonbassPeople campaign was focused on an Anglophone and, to a lesser extent, a Russophone audience. The majority of messages (62%) supporting either of the two sides were written in English, whereas tweets in Russian represented only 27% (ATO) and 36% (Novorossiya). The opposite picture was observed among neutral messages, where the majority of tweets (58%) were written in Russian, while tweets in English accounted for only 38% of messages. The small number of messages in Ukrainian can be explained by several reasons. First, it can reflect the linguistic competences of people who used the #SaveDonbassPeople hashtag; however, this explanation seems more relevant for Novorossiya supporters than for ATO ones, among whom several Ukrainophone media organizations were present. Second, it can indicate that both Novorossiya and ATO advocates were less concerned with convincing the relatively monolinguistic population of Western Ukraine, where Ukrainian is a preferred language, and concentrated on communicating their views to the population of Eastern and Southern Ukraine, where Russian is a preferred language, and central regions, where both Ukrainian and Russian are commonly spoken (Khmelko, 2004). TABLE 5 Languages and Affiliations of Messages From the Random Sample ATO Neutral Novorossiya Total English 102 32 482 616 Russian 44 49 284 377 Ukrainian 13 0 1 14 Other 3 3 11 17 Together 162 84 778 1,024 The classification of tweets’ language and affiliation suggests that both anti- and pro-government advocates primarily used Twitter to disseminate information about the conflict among Anglophone Twitter users instead of mobilizing Ukrainian citizens to express their dissatisfaction with the ATO and/or coordinating anti-government activities; otherwise we would expect the higher number of tweets in Ukrainian and Russian. Furthermore, we found that only a few tweets from the sample called for action—for example, asked users to participate in online flashmobs—or provided practical guidelines—for example, asked users to add certain hashtags or post more photos. Thus, we suggest that similar to the Moldova revolution (Lysenko & Desouza, 2012), the main goal for Novorossiya and ATO supporters on Twitter was to influence the Twittersphere’s perception of the conflict and convince other users in the rightfulness of their cause. The classification of messages’ types (Table 6) indicated that both sides pursued these goals mainly through retweets, which accounted for the majority of messages that supported Novorossiya and ATO (79% and 84%). Although retweeting is viewed as “a rather lightweight form of participation” (Poell & Bora, 2012, p. 704), it is an effective means of online campaigning. A number of studies (Lotan et al., 2011; Meraz & Papacharissi, 2013) suggest that retweets proved to be particularly effective for generating information cascades (Bikhchandani, Hirshleifer, & Welch, 1992), which enabled the rapid propagation of news and interpretations across personal networks during the Arab Spring. However, retweeting is also a less personal means of engaging with a topic, one that requires less energy and thought than writing a personal comment or sharing a piece of news. Two latter types of messages were more common among neutral tweets, where there were fewer retweets but a greater number of comments and conversations. TABLE 6 Types and Affiliations of Messages from the Random Sample Comment Conversation Hashtag only News Retweet Spam Total ATO 12 2 6 6 136 0 162 Neutral 11 4 2 16 44 7 84 Norovossiya 58 32 41 30 616 1 778 Total 81 38 49 52 796 8 1,024 Our analysis also showed that the content of retweets depended on their political affiliation. Neutral retweets usually concerned recent news and latest updates; in contrast, the ATO/Novorossiya retweets more frequently delivered emotional messages. Instead of spreading the latest updates across Twitter, as in other instances of public unrest such as natural disasters (Bruns et al., 2012) or parliamentary elections (Bruns & Burgess, 2011), advocates of Novorossiya and ATO tended to retweet straight-out slogans (e.g., “Stop Kiev junta!” or “Stop Russian terrorists!”) and impersonal indicators of support for their side (e.g., posters with words “#SaveDonbassPeople” or generic images of sad children). Such content was made available through embedded hyperlinks, which explains the large number of tweets with URLs in our sample (84%). The classification of external resources that were referenced in #SaveDonbassPeople tweets (Table 7) summarizes differences between the neutral messages and the ATO/Novorossiya-related ones. Although the percentage of hyperlinked tweets is large among all three categories, advocates of ATO and Novorossiya referenced external resources more frequently. ATO supporters were particularly keen on linking their messages to external sources, as shown by the 95% of pro-government messages containing hyperlinks. Even though in terms of absolute numbers Novorossiya advocates were able to produce more hyperlinked messages, ATO supporters were more persistent in promoting a small selection of materials by continuously referencing them in their tweets. As both pro-ATO and pro-Novorossiya campaigns were informative in their nature, we suggest that this persistence can be viewed as an additional evidence of a more organized campaign that used Twitter for accentuating several information resources. TABLE 7 Types of URLs and Affiliations of Messages From the Random Sample Social networks Video hosts News sites Other Total ATO 132 7 10 1 150 Neutral 11 23 17 6 57 Norovossiya 504 107 36 9 656 Total 647 137 63 16 863 The lack of links to the mass media—for example, mainstream news agencies—and the predominance of references to the social networking sites (e.g., Twitter and Vkontakte) in #SaveDonbassPeople tweets can be attributed to several factors. First, it can be viewed as the continuation of a long-term trend among activists to turn to social media as their preferred platform of communication, driven by their inability to attract the attention of the mainstream press (Couldry & Curran, 2003; Lievrouw, 2011). Under such circumstances, social networking sites can, in theory, become platforms for alternative journalism, or at least provide a different interpretation of events, thus challenging mainstream protest reporting (Poell & Bora, 2012). Furthermore, the rapid and chaotic deterioration of the situation in Eastern Ukraine that hampered the work of mainstream media outlets could increase the value of citizen media, including filming and live streams by non-professionals available through YouTube and Twitter. Second, unlike mainstream media organizations, online-only information agencies and individual users are less susceptible to reputational damage if they publish unverified or biased information. In their study of the Tunisian and Egyptian uprisings, Lotan et al. (2011) found that individuals shared information more liberally than media organizations, and were therefore more likely to spread news before the news was vetted or verified. Similarly, we assume that individuals were more liberal in the sense of producing and sharing emotional statements with obvious pro- or anti-government affiliations, which were used both by ATO and Novorossiya advocates to win the support of their audiences. Together, all these observations suggest that the #SaveDonbassPeople hashtag was mainly used in the context of the online campaign of the same name. This campaign was promoted by a number of anti-government activists and bloggers, whose efforts were counteracted by a few pro-government activist groups and media organizations. Both pro- and anti-government advocates were particularly active at those moments, when their side suffered losses in the course of the conflict; while doing so, activists from both sides produced and retweeted affective messages in order to provoke an emotional reaction from their target audience, which mainly consisted of Anglophone and, to a lesser degree, Russophone Twitter users. However, not all messages with the #SaveDonbassPeople hashtag were part of the #SaveDonbassPeople campaign; instead, the hashtag was also used by bystanders, which included media organizations, think tanks, and individual users. Unlike messages, which were produced by ATO and Novorossiya supporters, these neutral tweets were mainly written in Russian and more often included references to recent updates than emotional statements. These observations together with the predominance of comments and discussions among the neutral messages suggest that the #SaveDonbassPeople hashtag was used not only for campaign purposes, but also for passing on relevant information and discussing the conflict in Eastern Ukraine. Competing Strategies AUXILIARY HASHTAGS Hashtags are keywords that are used to organize and facilitate communication on Twitter. By adding a hashtag to a message, users make their tweets more visible and engage with other users tweeting on the same subject. However, hashtags can be used not only for organizational purposes, but also for information dissemination. On Twitter, the top hashtags appear in the “trending topic” area of the user’s profile, attracting his or her attention to a particular issue. Thus, hashtags often serve as a backbone for online campaigns, capable of organizing followers and attracting public attention. The co-occurence of multiple hashtags in the same message is one aspect of Twitter studies that tends to be overlooked by scholars. Many data sets are established around one hashtag (like, for instance, the #SaveDonbassPeople hashtag); however, this does not necessarily mean that collected messages include only that particular hashtag. Instead, tweets can include two, three, or even more hashtags, which mark key topics addressed in these messages. In this section we examined two strategies that involved the use of multiple hashtags and were employed by anti- and pro-government advocates in the course of the #SaveDonbassPeople Campaign. The first strategy was based on using auxiliary hashtags as crosscutting networking mechanisms that allowed users to embed their messages into several hashtag streams simultaneously and, thus, address audiences across (and beyond) the community that has built up around the #SaveDonbassPeople hashtag. Our data suggest that this strategy was frequently used in the course of the campaign because almost 63% (51,408 out of 81,404 tweets) of messages from our data set were supplemented with additional hashtags. The incorporation of messages about the conflict in Eastern Ukraine into large-scale flows of information through the addition of hashtags such as #ukraine or #war was a common practice. Another example involves the addition of hashtags in other languages such as #ucrania (Spanish), ‫( אוקראינה‬Hebrew)​, or #ukrayna (Turkish). Sending information across a wider network is not, however, the only function of auxiliary hashtags. Our study suggests that auxiliary hashtags were also used as independent messages, which either enhanced the impact of the main hashtag (reinforcing) or changed it radically (infiltrating). For instance, the inclusion of the hashtag #supportfromX, where X is replaced by a country name (e.g., #supportfromireland), reinforced the #SaveDonbassPeople hashtag by showing international recognition of the suffering of the population of Eastern Ukraine. The addition of hashtags that pointed to the enemy from whom the people of Donbass should be saved can be viewed as another case of reinforcing. Examples of such hashtags included both extended versions of the main hashtag (#savedonbasspeoplefromukrarmy) and new keywords that pointed both to internal (#nokievnazi) and external enemies (#stopnato). Attempts to infiltrate the #SaveDonbassPeople stream evolved around shifting the blame for the suffering in the Donbass region to Russia and to separatist movements. The inclusion of auxiliary hashtags after #SaveDonbassPeople, such as #fromdnr or #fromrussianterrorists, to produce a new message is one example of how the initial meaning of the online campaign was twisted. The production of new hashtags based on the ones used by opponents is another example of how protests were framed in a different way by pro-government advocates. Examples of such “new–old” hashtags included #russianpropagandakillsourguys and #savedonbasspeoplefromputin. Based on our classification of auxiliary hashtags (Table 8), we suggest that both ATO and Novorossiya advocates were mainly focused on Anglophone Twitter streams rather than Russophone or Ukrainophone ones. This suggestion is based on two observations: first, English hashtags constituted the majority of auxiliary hashtags (69%) and were used most frequently (73% of messages with auxiliary hashtags). Second, although the majority of auxiliary hashtags in all three languages were informative in function, the largest proportion of reinforcing/infiltrating hashtags was found among Anglophone hashtags. Thus, the Anglophone Twittersphere became the main battleground for anti- and pro-government activists who tried to promote their view on the conflict. In contrast to the English ones, the Russian and Ukrainian auxiliary hashtags that were not purely informative in nature focused either on reinforcing (Russian ones) or infiltrating (Ukrainian ones) the #SaveDonbassPeople campaign. This observation suggests that advocates of ATO and Novorossiya were less interested in propagating their views among people who were viewed as supporters of the opposing side (Russian speakers in the case of ATO supporters and Ukrainian speakers in the Novorossiya’s case). Instead, both sides competed for the support of the Anglophone audience, whereas hashtags in local languages were mainly used for incorporating information about the conflict into other information streams. TABLE 8 Languages and Functions of Auxiliary Hashtags English Infiltrating Informative Reinforcing Total 30 334 203 567 Russian 3 130 26 159 Ukrainian 3 9 0 12 Other 0 12 0 12 Unidentified 2 69 2 73 Total 38 554 231 823 EXTERNAL CONTENT Another strategy that was employed during the #SaveDonbassPeople campaign involved the use of external content that was referenced in tweets through hyperlinks. Our earlier classification of external resources used in #SaveDonbassPeople tweets indicated that social networking sites served as a major source of references both for ATO and Novorossiya supporters. The classification of external content (Table 9) indicated that these references usually led to images supplemented with verbal texts; two common genres among these images were demotivational posters and selfies. It is worth noting, however, that the images that were used in the course of the campaign differed from the conventional representations of both genres. For instance, while demotivational posters are usually humorous and funny, in the context of the #SaveDonbassPeople campaign they portrayed scenes of death and destruction, accompanied with somber calls for ending the ongoing conflict. As for the second genre, although the term “selfie” was still the best description for those types of images, not all of them satisfied its conventions (for instance, some photos featured passport pages instead of faces, or small children, who probably did not take photos themselves). Although both ATO and Novorossiya activists employed images from the same genres, the content of pictures varied depending on one’s allegiance. Demotivational posters made by Novorossiya advocates usually showed gory images from Eastern Ukraine or old Soviet posters; both of these were supplemented with a few lines of text accusing the Ukrainian government of genocide and/or Nazism. In contrast, ATO supporters used posters with Ukrainian flags emblazoned with slogans calling for national unity and fighting terrorism. In general, pro-ATO demotivational posters avoided references to the Ukrainian past, whereas Novorossiya advocates constantly tried to place current events in some historical context and establish a sense of continuity with older conflicts such as the Second World War. In order to do so, they often used historical metaphors (“Sloviansk is a new Stalingrad”) and recognizable symbols (e.g., photoshopping swastikas into the photos of Ukrainian officials and adding red stars to the images of Novorossiyan leaders). TABLE 9 Types of External Materials from the Random Sample Amateur footage Deleted Demot. posters News reports Other Photos Selfies Texts Total ATO 6 7 32 1 3 4 86 11 150 Neutral 10 2 2 14 3 2 2 22 57 Norovoss iya 25 72 135 35 16 87 248 38 656 Total 41 81 169 50 22 93 336 71 863 Unlike demotivational posters, where there were significant differences between ATO and Novorossiya supporters, both sides used similar techniques to produce selfies. The most common type of selfie was a photo of a person holding a piece of paper with a hashtag written on it. Usually, it was the #SaveDonbassPeople hashtag, but occasionally auxiliary hashtags, such as #savedonbasspeoplefromputin or #nokievnazi, were also used. Variations of selfies included photos of passport pages with a sheet of paper with a hashtag located between pages and photos of children with a sheet of paper in their hands. The major difference between the two camps was the more extensive use of photos of children by Novorossiya advocates, who presumably wanted to evoke compassion from the potential audience by using sentimental images. Another peculiar feature of #SaveDonbassPeople activity was the limited use of amateur footage. Although it is assumed that digital media facilitates the emergence of ambient journalism (Hermida, 2010), the amateur evidences of the ongoing conflict attracted relatively meager attention in the course of the #SaveDonbassPeople campaign. This discrepancy can be explained by two reasons: First, the large number of references to news reports (mainly of Russian TV channels) in pro-Novorossiya tweets can be interpreted as an attempt to add credibility to the claims of anti-government activists. Second, because the majority of links to deleted external materials led to YouTube, we assume that this category mainly consisted of amateur footage, which was quickly removed from YouTube because of its offensive or gory nature. Although our classification suggests that ATO supporters were more willing to disseminate links to amateur video than news reports, it is worth noting that all references from the former category led to one video, which showed the aftermath of the killing of a Novorossiyan official. Based on these observations, we suggest that anti- and pro-government advocates competed with each other for control of information streams related to the conflict in Eastern Ukraine. Both examined strategies dealt with the dissemination of information about the conflict; however, instead of providing updates about the latest developments, advocates of Novorossiya and ATO propagated emotional statements that dehumanized the opposite side. Because both sides focused on the Anglophone audience, they tended to use less language-dependent materials such as images or short statements in the form of hashtags. Framing the Conflict in Eastern Ukraine In the last part of our analysis we explored how ATO and Novorossiya advocates used selected patterns of presentation and interpretation—also known as frames—to promote their view on the conflict in Eastern Ukraine. Frames determine both the information that is presented to an audience and the method of presentation, which affects the way an audience perceives an issue (Iyengar, 1991). As a result, frames define how individuals evaluate ongoing events, which makes their use particularly important for protest movements broadcasting their identity to the world (Meraz & Papacharissi, 2013). We were able to identify five types of frames that were used during the #SaveDonbassPeople campaign: historical, geographical, religious, ethnic, and political. Historical frames were the most common and concerned references to the past, in particular the Soviet period of Ukrainian history. Such frames were mainly used by anti-government advocates, who framed the current conflict through Second World War memory, which serves as a major dividing line for collective identities in Eastern and Western Ukraine (Kappeler, 2009; Marples, 2007; Portnov, 2013). Novorossiya advocates positioned themselves as fighters against fascism and portrayed supporters of the Ukrainian government as the successors to Nazi Germany. The equation of Ukrainian officials with Nazi leaders was one of the recurrent motifs of the #SaveDonbassPeople campaign, which argued that the aim of the “Ukrainian Nazis” was the extermination of the local Russian population, either through mass killings or the banishment of the Russian language and culture. ATO supporters used different historical frames: instead of referring to the Second World War, they portrayed their opponents as living Soviet anachronisms, using such derogative terms as “sovki” and “vatniki.” Both words refer to individuals who hold positive views of the Soviet period and/or certain “Soviet values,” and support the restoration of the Soviet Union. According to ATO advocates, people who assessed the Soviet period in a positive way were also apologists for Soviet crimes in the Ukraine (e.g., the Great Ukrainian Famine) and thus should be viewed as successors to the apparatus of Soviet repression. Geographical frames were based on the contraposition of Western Ukraine (together with Kiev, which was “occupied” by Western Ukrainians) against Eastern Ukraine. According to Novorossiya advocates, Western Ukraine was the source of radical nationalism that caused the Ukrainian crisis, whereas ATO supporters referred to Donbass people as pro-Soviet/proRussian collaborators. The origins of these frames can be connected to the recent history of Ukrainian elections, where pro-EU and pro-nationalist parties usually received greater support in Western Ukraine, while pro-Russian and anti-nationalist forces secured the majority of votes in Eastern Ukraine (CEC, 2012a; CEC, 2012b). However, the roots of geography-driven dissent could also be related to the historical division of Ukrainian territories between different countries, given that Ukraine’s contemporary borders were established only after the Second World War.7 The significance of this East-West divide in the context of the #SaveDonbassPeople campaign is emphasized by the lack of references to Central, Southern, or Northern Ukraine; instead, both ATO and Novorossiya supporters were focused on setting Galicia against Donbass, which became two symbolic markers of intra-Ukrainian dissent. 7 From the end of the 18th century, Western Ukraine belonged to the Austro-Hungarian Empire, while the rest of the country was part of the Russian Empire. After the First World War, the majority of Ukrainian territories became part of the Soviet Union, while Western Ukraine was integrated into Poland. Religious frames were based on religious differences between Western and Eastern Ukraine, in particular various branches of Christianity that exist in Ukraine (Bociurkiw, 1995; Plokhy & Sysyn, 2003; Wasyliw, 2014). Although the majority of the Ukrainian population practices Eastern Orthodoxy, the western regions of Ukraine were influenced by the Catholic Church, resulting in the establishment of the Ukrainian Greek Catholic Church (UGCC) at the end of the 16th century. Currently, more than 90% of the UGCC communities are concentrated in Western Ukraine, and in some western regions they are dominant (Razumkov Centre, 2011, p. 16). Although a recent study (Razumkov Centre, 2011, p. 47) claims that this religious divide has limited impact on everyday life because of the high level of religious tolerance in Ukraine, anti-government advocates used it for framing the ongoing conflict in religious terms. Throughout the #SaveDonbassPeople campaign, Novorossiya supporters positioned themselves as defenders of Eastern Orthodoxy from Catholicism and the UGCC, which presumably wanted to exterminate Orthodox believers in Eastern Ukraine. In contrast to Novorossiya advocates, who at certain points adopted an almost fundamentalist Orthodox stance, ATO supporters avoided using religious frames altogether. Ethnic frames were based on ethnic differences between Ukrainian regions, in particular the high percent of ethnic Russian population in Eastern Ukraine (Khmelko, 2004). Although Weller (2002) argued that the likelihood of ethnic conflict at the regional level in Ukraine is insignificant because of the low perceptions of ethnic distance between Ukrainians and Russian, both anti- and pro-government advocates positioned the conflict in ethnic terms by labeling the actions of opponents as ethnic cleansing and/or genocide. ATO advocates stressed that protests in Eastern Ukraine were initiated by “Russian terrorists” whose ultimate goal was to expel or exterminate local Ukrainians. In contrast, Novorossiya supporters argued that the nationalistic Ukrainian government, which is secretly ruled from the United States, is trying to wipe out the ethnic Russians, who constitute the majority in Eastern Ukraine. At the same time, anti-government advocates differentiated between “bad” Ukrainian nationalists and “good” ethnic Ukrainians, mirroring the earlier Soviet tendency of distinguishing “good” nations and “bad” nationalism (Scherbak, 2013). Finally, political frames were based on juxtaposing the need to protect a state’s territorial integrity with the right to self-determination. ATO advocates invoked the inviolability of Ukrainian state borders, while Novorossiya supporters argued that the Donbass region could claim its independence from a failed Ukrainian state. Unlike earlier types of frames, which were based on intra-Ukrainian differences, political frames were used less frequently and referred to external precedents (in particular Yugoslavia). For instance, ATO supporters accentuated the Croatian experience of defending their country’s territorial integrity against Serbian separatists, whereas Novorossiya advocates pointed to the case of Kosovo as the justification for People’s Republic’s claims. Our observations suggest that similar to the Egyptian protests (Meraz & Papacharissi, 2013), frames used in the course of the #SaveDonbassPeople campaign were influenced by strong emotions. These emotionally charged patterns of interpretation were largely focused on the opposing side, so instead of broadcasting their own identity, both anti- and pro-government advocates mainly constructed the identity of their opponents. The process of establishing the Other’s image by juxtaposing collective identities often referred to cultural, religious, and ethnic differences between different parts of Ukraine. With the help of digital media these differences were exaggerated and inflated in a way that not only allowed ATO and Novorossiya supporters to establish essential differences in their self-identification, but also to foment the ongoing conflict by positioning it as a clash of identities. CONCLUSIONS In our study we explored how the #SaveDonbassPeople hashtag was used by activists and bystanders in the course of the online campaign on Twitter. Although the campaign was initially focused on protecting the human rights of the Eastern Ukrainian population and condemning the use of force by the Ukrainian government, it soon came to be contested by different activist groups. Both pro- and anti-government activists tried to use Twitter to propagate their view on the conflict in Eastern Ukraine, which resulted in a heated confrontation around the #SaveDonbassPeople hashtag. However, our study demonstrated that the hashtag was used not only by advocates of a particular side, but also bystanders, including media organizations, think tanks, and individual users. In contrast to activists, bystanders produced less emotional and political messages, and mainly used the hashtag for passing on relevant information and discussing the conflict in Eastern Ukraine. Similar to earlier studies (Aday et al., 2010; Lysenko & Desouza, 2012; Morozov, 2011) that question Twitter’s organizational potential during protest campaigns, we found that neither Novorossiya nor ATO activists used Twitter primarily to plan their actions. Instead, both sides used the hashtag mainly to mobilize public support and/or discredit their opponents. While doing so, both camps employed a number of strategies to convince the wider Twittersphere of the righteousness of their cause, including attempts to infiltrate information streams, organize online flashmobs, manipulate data sources, and spam influential bloggers. Our analysis of two of these strategies, which involved the use of ]auxiliary hashtags and the addition of external content, suggests that even small-scale activity by the opposing camp can have a serious impact on an online protest campaign. Although Novorossiya advocates were rather active in sending messages and engaged a significant number of sympathizers, ATO supporters were able to penetrate the #SaveDonbassPeople information stream and use the campaign’s medium to challenge claims of anti-government supporters. For this purpose, pro-government advocates resorted both to covert infiltration (e.g., the inclusion of links to pro-government resources in #SaveDonbassPeople tweets or the use of infiltrating hashtags) and direct interference (e.g., the creation of parallel information streams or the dissemination of refutations of anti-government accusations). The main struggle between pro- and anti-government supporters in the course of the campaign evolved around the framing of the conflict in Eastern Ukraine. A number of frames were used for this purpose, yet almost all of them were based on either real or imagined cultural, religious, and ethnic differences between Eastern Ukraine and the rest of the country. Both sides tended to use these differences for constructing negative image of their opponents and framing the conflict as a clash of identities. The #SaveDonbassPeople campaign was focused on the international audience, but neither ATO nor Novorossiya advocates attempted to facilitate an understanding of these frames for those users, who were unfamiliar with local context. Instead, the majority of frames used in the campaign relied heavily on familiarity with Ukrainian history as well as the ethnic and religious characteristics of Ukrainian society, making their use more efficient for the Ukraino- and Russophone audiences than for the Anglophone one. Together these findings suggest that, instead of tweeting protest, the #SaveDonbassPeople hashtag was mainly used to tweet propaganda. In contrast to the widespread belief in the pluralizing power of social media, our study showed that online platforms can be easily used to propagate a certain point of view. Moreover, digital media might be particularly vulnerable to such usage because of the ease with which fake or provocative content can be produced, uploaded, and distributed across social networking sites. The extensive use of links to external materials in #SaveDonbassPeople tweets is illustrative in this respect, as ATO and Novorossiya supporters used Twitter to disseminate references to offensive videos, propagandistic images, and hate messages, which had to dehumanize their opponents both in the eyes of the local population and the international community. These conclusions, however, should be tempered by recognition of the limits of our research. Although our study captured the heyday of the #SaveDonbassPeople campaign, the hashtag has also been used after the end of our study period. Consequentially, our findings are not necessarily representative for the later stages of the #SaveDonbassPeople activity, which should be investigated separately. Similarly, although we examined the major #SaveDonbassPeople information stream on Twitter, our analysis suggests that the campaign involved a number of parallel streams and external resources, which were only partially considered in our study. A thorough investigation of these digital media mechanisms is beyond the scope of our study, yet it can certainly contribute to a better understanding of the #SaveDonbassPeople activity. Our study also highlights a number of possibilities for further research on the use of social media, in particular Twitter, during the Ukrainian crisis. Complex interactions between different actors involved in the crisis can be thoroughly examined in future studies, which will focus on competing strategies used by activist and non-activist groups for mobilizing public support and discrediting their opponents. Further research is also required in order to explore the impact of social media on internationalization of the crisis, including differences between information/propaganda campaigns focused on domestic and on international audiences. Finally, future studies can examine the evolution of frames used for representing the conflict in Eastern Ukraine by comparing our findings with observations related to the later stages of the crisis. Our findings echo those observations that question the role of social media in organization and mobilization efforts of protest movements, yet this does not mean that Twitter had no impact on the conflict in Eastern Ukraine. The study of the use of the #SaveDonbassPeople hashtag allowed us to explore the goals and methods of anti- and pro-government advocates, and revealed some of the ideological contradictions between them. However, Twitter can be used not only to examine differences in self-identification, but it can be also employed by individual actors to accentuate these differences in the course of the online campaign. Although it is difficult to assess the exact impact of such manipulations, we suggest that the use of Twitter as a propaganda outlet might not only distort coverage of the conflict, but also escalate hostilities by portraying events in a one-sided manner and dehumanizing the opposing side. As the use of digital technologies becomes a common practice for contemporary protest movements, we need further research in order to comprehend how digital media can be used for propaganda and what potential dangers may arise. REFERENCES Aday, S., Farrell, H., Lynch, M., Sides, J., Kelly, J., & Zuckerman, E. (2010). Blogs and bullets: New media in contentious politics. Washington, DC: U.S. Institute of Peace. Aguado, E. (2014, February 20). Redes sociales rompen la barrera informativa de las protestas en Venezuela [Social networks breach the information blockade on protests in Venezuela]. Marti News. Retrieved from http://www.martinoticias.com/content/redes-sociales-barrera-informativavenezuela-protestas /32225.html Antiterroristicheskaja operacija v Donbasse: Hronika za 1-31 Maja [Anti-terrorist operation in Donbass: May 2014]. (2014, May 31). Retrieved from http://news.liga.net/articles/politics/1596426-vtorzhenie_v_ukrainu_poslednie_sobytiya_v_D onbasse_i_krymu.htm Attia, A. M., Aziz, N., Friedman, B., & Elhusseiny, M. F. (2011). Commentary: The impact of social networking tools on political change in Egypt’s “Revolution 2.0.” Electronic Commerce Research and Applications, 10, 369–374. doi:10.1016/j.elerap.2011.05.003 Bikhchandani, S., Hirshleifer, D., & Welch, I. (1992). A theory of fads, fashion, custom, and cultural change as informational cascades. Journal of Political Economy, 100, 992–1026. doi:10.1086/jpe.1992.100.issue-5 Bociurkiw, B. (1995). Politics and religion in Ukraine: The Orthodox and the Greek Catholics. In M. Bourdeaux (Ed.), The politics of religion in Russia and the new states of Eurasia (pp. 131–163). Armonk, NY: M. E. Sharpe. Bohdanova, T. (2013, December 9). How Internet tools turned Ukraine’s #Euromaidan protests into a movement. Global Voices. Retrieved from http://globalvoicesonline.org/2013/12/09/how-internet-tools-turned-euromaidanprotestsinto-a-movement/ Boyd, D., Golder, S., & Lotan, G. (2010, January). Tweet, tweet, retweet: Conversational aspects of retweeting on Twitter. Paper presented at the 43rd Hawaii International Conference on System Sciences, Koloa, HI. Retrieved from http://www.danah.org/papers/TweetTweetRetweet.pdf Breuer, A., Landman, T., & Farquhar, D. 2014. Social media and protest mobilization: Evidence from the Tunisian revolution. Democratization, 22, 764–792. doi:10.1080/13510347.2014.885505 Bruns, A., & Burgess, J. (2011). #Ausvotes: How Twitter covered the 2010 Australian federal election. Communication, Politics & Culture, 44, 37–56. Bruns, A., Burgess, J., Crawford, K., & Shaw, F. (2012). #qldfloods and @QPSMedia: Crisis communication on Twitter in the 2011 South East Queensland floods. Brisbane, Australia: ARC Centre of Excellence for Creative Industries and Innovation. Retrieved from http://www.danah.org/papers/TweetTweetRetweet.pdf Bruns, A., & Stieglitz, S. (2013). Towards more systematic Twitter analysis: Metrics for tweeting activities. International Journal of Social Research Methodology, 16, 91–108. doi:10.1080/13645579.2012.756095 Burns, A., & Eltham, B. (2009, November). Twitter free Iran: An evaluation of Twitter’s role in public diplomacy and information operations in Iran’s 2009 election crisis. Paper presented at the Communications Policy & Research Forum, Sydney, Australia. Retrieved from http://vuir.vu.edu.au/15230/1/CPRF09BurnsEltham.pdf CEC. (2012a). Pozacherghovi vybory narodnykh deputativ Ukrajiny 30 veresnja 2007 roku: Vidomosti pro pidrakhunok gholosiv vyborciv v mezhakh reghioniv Ukrajiny [Early parliamentary elections in Ukraine (30 September 2007): Voting results by region]. Retrieved from h ​ ttp://www.cvk.gov.ua/pls/vnd2007/w6p001 CEC. (2012b). Vybory narodnykh deputativ Ukrajiny 28 zhovtnja 2012 roku: Vidomosti pro pidrakhunok gholosiv vyborciv v zaghaljnoderzhavnomu vyborchomu okruzi v mezhakh OVO [Parliamentary elections in Ukraine (28 October 2012): Voting results by region]. Retrieved from ​http://www.cvk.gov.ua/pls/vnd2012/wp315?PT001F01=900 Couldry, N., & Curran, J. (2003). Contesting media power: Alternative media power in a networked world. Lanham, MD: Rowman & Littlefield. Eltantawy, N., & Wiest, J. B. (2011). Social media in the Egyptian revolution: Reconsidering resource mobilization theory. International Journal of Communication, 18, 1207–1224. Fisher, A. (2010). Bullets with butterfly wings: Tweets, protest networks, and the Iranian election. In Y. Kamalipour (Ed.), Media, power, and politics in the digital age: The 2009 presidential election uprising in Iran (pp. 105–118). Lanham, MD: Rowman and Littlefield. Freedman, L. (2014). Ukraine and the art of limited war. Survival, 56, 7–38. Friedman, U. (2014, February 19). Why Venezuela’s revolution will be tweeted. The Atlantic. Retrieved from http://www.theatlantic.com/international/archive/2014/02/why-venezuelas-revolution-will-be-t weeted/283904/ García, C., Chauveau, P., Ledezma, J., & Pinto, M. (2013, August). What can social media teach us about protests? Analyzing the Chilean 2011–12 student movement’s network evolution through Twitter data. Paper presented at the 4th International Conference on Collaborative Innovation Networks COINs13, Santiago de Chile, Chile. Retrieved from http://arxiv.org/ftp/arxiv/papers/1308/1308.2451.pdf Genç, K. (2014). Turkey’s Twitter doi:10.1177/0306422014522385 army. Index on Censorship, 43, 106–110. Gladwell, M. (2010, October 4). Small change: Why the revolution will not be tweeted. The New Yorker. Retrieved from http://www.newyorker.com/reporting/2010/10/04/101004fa_fact_gladwell Gruzd, A., & Tsyganova, K. (2014, September). Politically polarized online groups and their social structures formed around the 2013–2014 crisis in Ukraine. Paper presented at the 2014 Internet, Politics, Policy Conference: Crowdsourcing for Politics and Policy, Oxford, UK. Retrieved from http://ipp.oii.ox.ac.uk/2014/programme-2014/track-d-networked-politics/ political-and-personal-networks/anatoliy-gruzd-ksenia-tsyganova Hermida, A. (2010). Twittering the news: The emergence of ambient journalism. Journalism Practice, 4, 297–308. doi:10.1080/17512781003640703 Information Resistance. (2014a). About us. Retrieved from h ​ ttp://sprotyv.info/en/about-us Information Resistance. (2014b). Vojna za umy. Kak vzryvali odesskuju OGA cherez e-mejl genprokurora i 37% Jarosha [War for minds. Blowing up the Odessa regional administration through the prosecutor-general’s e-mails and 37 per cent of Yarosh]. Retrieved from http://sprotyv.info/ru/news/679-voyna-za-umy-kak-vzryvali-odesskuyu-oga-cherez-e-meyl-ge nprokurora-i-37-yarosha?_utl_t=fb Iyengar, S. (1991). Is anyone responsible? How television frames political issues. Chicago, IL: University of Chicago Press. Kappeler, A. (2009). From an ethnonational to a multiethnic to a transnational Ukrainian history. In G. Kasianov, & P. Ther (Eds.), A laboratory of transnational history: Ukraine and recent Ukrainian historiography (pp. 51–81). Budapest, Hungary: CEU Press. Khmelko, V. (2004). Linhvo-etnichna struktura Ukrainy: Rehionalni osoblyvosti ta tendentsii zmin [Ethnolinguistic structure of Ukraine: Regional features and tendencies of change during the independence period]. Naukovi Zapysky Naukma, 32, 3–15. Lievrouw, L. (2011). Alternative and activist new media. Cambridge: Polity, UK. Lin, Y.-R., Keegan, B., Margolin, D., & Lazer, D. (2014). Rising tides or rising stars?: Dynamics of shared attention on Twitter during media events. PLoS ONE, 9, 1–12. Lotan, G., Graeff, E., Ananny, M., Gaffney, D., Pearce, I., & Boyd, D. (2011). The revolutions were tweeted: Information flows during the 2011 Tunisian and Egyptian revolutions. International Journal of Communication, 5, 1375–1405. Lynch, M., Freelon, D., & Aday, S. (2014). Syria’s socially mediated civil war. Washington, DC: U.S. Institute of Peace. Lysenko, V., & Desouza, K. (2012). Moldova’s Internet revolution: Analyzing the role of technologies in various phases of the confrontation. Technological Forecasting and Social Change, 79, 341–361. doi:10.1016/j.techfore.2011.05.009 Marples, D. (2007). Heroes and villains: Creating national history in contemporary Ukraine. Budapest, Hungary: CEU Press. Meraz, S., & Papacharissi, Z. (2013). Networked gatekeeping and networked framing on #Egypt. The International Journal of Press/Politics, 20, 1–29. Morozov, E. (2009a). Iran: Downside to the “Twitter revolution.” Dissent, 56, 10–14. doi:10.1353/dss.0.0092 Morozov, E. (2009b, December 29). Iran’s propaganda hits the “Spinternet.” CNN. Retrieved from ​http://edition.cnn.com/2009/OPINION/12/29/morozov.dicatorships.internet/ Morozov, E. (2011, March 7). Facebook and Twitter are just places revolutionaries go. The Guardian. Retrieved from http://www.theguardian.com/commentisfree/2011/mar/07/facebook-twitter-revolutionaries-cy ber-utopians Morstatter, F., Pfeffer, J., Liu, Y., & Carley, K. M. (2013, July). Is the sample good enough? Comparing data from Twitter’s Streaming API with Twitter’s Firehose. Paper presented at the 7th International Conference on Weblogs and Social Media, Cambridge, MA. Retrieved from ​http://arxiv.org/pdf/1306.5204v1 Mungiu-Pippidi, A., & Munteanu, I. (2009). Moldova’s “Twitter revolution.” Journal of Democracy, 20, 136–142. doi:10.1353/jod.0.0102 Nikiporets-Takigawa, G. (2013). Tweeting the Russian protests. Digital Icons: Studies in Russian, Eurasian and Central European New Media, 9, 1–25. Noakes, J. A., & Johnston, H. (2006). Frames of protest: A road map to a perspective. In J. A. Noakes, & H. Johnston (Eds.), Frames of protest: Social movements and the framing perspective (pp. 1–33). Lanham, MD: Rowman & Littlefield. Onuch, O. (2015). EuroMaidan Protests in Ukraine: Social media versus social networks. Problems of Post-Communism, 62, 217–235. doi:10.1080/10758216.2015.1037676 Operacija po antiterroru v Donbasse: Hronika za 1-30 ijunja [Anti-terrorist operation in Donbass: June 2014]. Available at http://news.liga.net/articles/politics/1658614-vtorzhenie_v_ukrainu_poslednie_sobytiya_v_do nbasse_i_krymu.htm​. Accessed July 4, 2014. Plokhy, S., & Sysyn, E. F. (2003). Religion and nation in modern Ukraine. Toronto, Canada: CIUS Press. Poell, T., & Borra, E. (2012). Twitter, YouTube, and Flickr as platforms of alternative journalism: The social media account of the 2010 Toronto G20 Protests. Journalism, 13, 695–713. doi:10.1177/1464884911431533 Portnov, A. (2013). Memory wars in post-Soviet Ukraine (1991–2010). In J. Fedor, A. Etkind, & U. Blacker (Eds.), Memory and theory in Eastern Europe (pp. 233–255). New York, NY: Palgrave Macmillan. Radchenko, D., Pisarevskaya, D., & Ksenofontova, I. (2012). Logika virtual’nogo protesta: Nedelja posle vyborov—2011 [The logic of virtual protest: A week after elections—2011]. Antropologicheskij Forum, 16, 108–126. Razumkov Centre. (2011). Relihiia i vlada v Ukraini: Problemy vzaiemovidnosyn [Religion and government in Ukraine: A question of interrelations]. Retrieved from http://www.uceps.org/upload/prz_2011_Rlg_smll.pdf #SaveDonbassPeople! Iniciativa PDRS—v dejstvii! [#SaveDonbassPeople! The initiative of DDWU in action!]. (2014, May 31). Retrieved from ​http://pdrs.dp.ua/type/type78.html Scherbak, A. N. (2013). Nationalism in the USSR: A historical and comparative perspective. Moscow, Russia: NRU Higher School of Economics. Segerberg, A., & Bennett, W. L. (2011). Social media and the organization of collective action: Using Twitter to explore the ecologies of two climate change protests. The Communication Review, 14, 197–215. doi:10.1080/10714421.2011.597250 Sloviansk. (2014, May 28). MAKSIMAL’NYJ REPOST! Novyj vitok informacionnoj agressii!!! #SAVEDONBASSPEOPLE pic.twitter.com/gPfQS6qlG8 [MAXIMUM REPOST! A new round of information aggression!!! #SAVEDONBASSPEOPLE pic.twitter.com/gPfQS6qlG8] [Twitter post]. Retrieved from ​https://twitter.com​/Sloviansk/status/471618826132791296/photo/1 Sullivan, A. (2009, June 13). The revolution will be Twittered. The Atlantic. Retrieved from http://www.theatlantic.com/daily-dish/archive/2009/06/the-revolution-will-be-twittered/200478 / Szostek, J. (2014). The media battles of Ukraine’s EuroMaidan. Digital Icons: Studies in Russian, Eurasian and Central European New Media, 11, 1–19. Theocharis, Y., Lowe, W., Van Deth, J. W., & Albacete, G. M. G. (2013, March). Using Twitter to mobilise protest action: Transnational online mobilisation patterns and action repertoires in the Occupy Wall Street, Indignados and Aganaktismenoi movements. Paper presented at the 41st ECPR Joint Sessions of Workshops, Mainz, Germany. Retrieved from http://dl.conjugateprior.org/preprints/ecprjs-theocharis-et-al.pdf Tremayne, M. (2014). Anatomy of protest in the digital era: A network analysis of Twitter and Occupy Wall Street. Social Movement Studies, 13, 110–126. doi:10.1080/14742837.2013.830969 Tufekci, Z., & Wilson, C. (2012). Social media and the decision to participate in political protest: Observations from Tahrir square. Journal of Communication, 62, 363–379. doi:10.1111/j.1460-2466.2012.01629.x Turchinov ob’javil o nachale antiterroristicheskoj operacii na jugo-vostoke Ukrainy [Turchynov declared the beginning of anti-terrorist operation in South-Eastern Ukraine]. (2014, April 7). Retrieved from http://vesti-ukr.com/strana/46250-turchinov-objavil-o-nachale-antiterroristicheskoj-operacii-n a-jugo-vostokeukrainy Vicari, S. (2013). Public reasoning around social contention: A case study of Twitter use in the Italian mobilization for global change. Current Sociology, 61, 474–490. doi:10.1177/0011392113479747 vita_linka. (2014, June 2). Marazm krepchaet [Dementia grows stronger] [Blog post]. Retrieved from h ​ ttp://www.liveinternet.ru/users/4956405/post326552113/ Wasyliw, V. Z. (2014). Orthodox churches in Ukraine. In L. N. Leustean (Ed.), Eastern Christianity and politics in the twenty-first century (pp. 312–334). London, England: Routledge. Weller, C. A. (2002). Mass attitudes and ethnic conflict in Ukraine. In T. Kuzio, & P. D’Anieri (Eds.), Dilemmas of state-led nation building in Ukraine (pp. 71–103). Westport, CT: Praeger. Yandex. (2014). Ohliad sotsialnykh merezh i Tvittera v Ukraini [Review of social networks and Twitter in Ukraine]. Retrieved from http://download.yandex.ru/company/Yandex_on_UkrainianSMM_Summer_2014.pdf