#fukushima Five Years On : A Multimethod Analysis of Twitter on the Anniversary of the Nuclear Disaster

International Journal of Communication 12(2018), 928–949 1932–8036/20180005 Copyright © 2018 (Anna Rantasila, Anu Sirola, Arto Kekkonen, Katja Valaskivi, and Risto Kunelius). Licensed under the Creative Commons Attribution Non-commercial No Derivatives (by-nc-nd). Available at http://ijoc.org.       ANNA RANTASILA ANU SIROLA ARTO KEKKONEN KATJA VALASKIVI RISTO KUNELIUS University of Tampere, Finland1 This article examines how the fifth anniversary of the Fukushima Daiichi nuclear disaster was commemorated on English-speaking Twitter in March 2016. By combining social network analysis and critical discourse analysis, a research design is developed that can be applied to study the structure of actors and interpretative resources invoked in the crafting of communal remembrance of a disruptive, global media event. In the study, we explore the most visible actors and the most dominant meanings in the #fukushima stream. According to our analysis, the most significant players were the mainstream media and other established organizations. While most of the retweeted messages contained a ritual element of collective memory work, grief, and observance, another prominent feature was the strongly politicized discourse surrounding the aftermath of the disaster. Keywords: multimethod, Twitter, hybrid media, social network analysis, critical discourse analysis, Fukushima Daiichi Anna Rantasila: [email protected] Anu Sirola: [email protected] Arto Kekkonen: [email protected] Katja Valaskivi: [email protected] Risto Kunelius: risto.kune[email protected]i Date submitted: 2016–10–12 1 This research has been supported by the Academy of Finland, grant number 277465. In addition, we received assistance from the CSC IT Center for Science, administered by the Ministry of Education and Culture of Finland, to collect and store the data for this study. International Journal of Communication 12(2018) #fukushima Five Years On 929 March 11, 2016 marked five years since the 9.0 magnitude earthquake and subsequent tsunami struck the northeastern coast of Japan’s main island, severely damaging the Fukushima Daiichi nuclear power plant 240 kilometers north of Tokyo. From the first tremors of the earthquake to the ongoing remembrance, the media coverage of the threefold disaster has played out in transnational communication networks where legacy media coverage overlaps with the new platforms and practices of social media. In this new kind of evolving media environment, the inherited platforms, brands, forms, and genres of mass media and the emerging modes and affordances of social media platforms interact (Chadwick, 2013; Harder, Paulussen, & Aelst, 2016). This rapidly changing landscape now constitutes a new kind of environment where the meanings of controversial issues and dramatic events are negotiated across national and institutional boundaries. Media research is only beginning to make sense of the consequences of the new dynamics that are in play. The Fukushima disaster involved an exceptionally dramatic, disruptive, and traumatic series of events. In addition to the human costs, they focused on nuclear power―a particularly loaded energypolicy domain where difficult questions related to science, expertise, economic interests, and political power intersect. Hence, the events of March 2011 caused complex systemic disruptions, ranging from lost lives and contaminated environments to the major policy decisions of nations and energy futures. This potential of meanings and consequences makes the interpretation of the event an exceptionally interesting object of study for understanding how the memory of such a traumatic event is communicated in the networked media landscape. In this article, we take the microblogging service Twitter as an entry point to the transnational communication networks activated when such commemoration work takes place. In the 2010s, Twitter has emerged as one of the key platforms through which the new conditions of the contemporary media environment are played out. Indeed, in Japan, the aftermath of the 2011 Fukushima Daiichi disaster represented a particular breakthrough moment for Twitter, and the platform has been a focal point for discussion both globally and in Japan since the events began to unfold (Cho & Park, 2013; Friedman, 2011; Li, Vishwanath, & Rao, 2014). In this paper, we focus on the transnational flows of information relating to the Fukushima Daiichi nuclear disaster on English-speaking Twitter during the fifth anniversary of the event, and specifically on two of Twitter’s key functions: hashtags (#) and sharing or “retweeting.” We treated hashtags such as #fukushima as “hybrid forums” (Burgess, Galloway, & Sauter, 2015) that create a context for discussion and enable different kinds of subforums to form under a particular topic. Our data consists of 17,619 English-language tweets containing the hashtag #fukushima. This article has two main aims. Methodologically, we wish to develop and test a research design that helps us better understand the use of Twitter as one key locus of the current global media landscape. As an effective tool for both quick commentary and the filtering and redistribution of content, Twitter user data has been used to study various phenomena, from fan cultures to political participation. Much of Twitter research has understandably been quantitative, relying on massive amounts of tweets. However, in addition to revealing the networks of issues and actors, there is an evident need to flesh out a more synthetic approach that lies between the quantitative analysis of network relations and the qualitative analysis of the discourses that articulate these relationships (Marwick, 2014; Sumiala, Tikka, Huhtamäki, & Valaskivi, 2016). More substantially, our aim is to reflect upon the role of Twitter networks in the 930 A. Rantasila et al. International Journal of Communication 12(2018) context of a traumatic event such as the Fukushima disaster. In particular, we want to explore questions related to the interplay between a moment of commemoration and the political potential opened up by collective, emotionally loaded, attention. By applying two complementary perspectives―social network analysis (hereafter SNA) and critical discourse analysis (hereafter CDA)―we hope to uncover deeper insights into what remains in discourses relating to a complex disruptive event such as the Fukushima Daiichi disaster five years on, and how these discourses are produced, reproduced, and circulated in the contemporary, global, and transnational hybrid media environment that encompasses various cultures and national media systems from Japan to the EU and the Americas.2 To elaborate our research object, we identify three analytically distinct “logics” that intersect in commemorating the Fukushima disaster. First, the logic of hybridity refers to the interplay of new and emerging institutions and modes of communication in traditional mass media and on social media platforms. Second, the logic of ritualizing trauma refers to how collective traumatic experiences are negotiated towards a shared, cultural interpretation of the disaster. Third, the logic of politicizing memory refers to the space of opportunity, and the attempts of various social actors to take advantage of it. Dynamics of the Global Hybrid Media Environment In recent years, the notion of “hybrid media system” has often been used to capture the blurring institutional boundaries, shifting actor roles, and multimodal representational opportunities of the digitalized media environment. In his influential book in which he coins the term, Andrew Chadwick (2013) uses the concept to zoom in on the changing interface between politics, journalism, and social media. He empirically and convincingly shows one form of hybridity: how old institutions (political parties, mainstream journalism) have been able to incorporate and exploit the new logic(s) of social media (van Dijck & Poell, 2013), partly allowing social media logics to shape these legacy institutions in return. While our study constructs its object somewhat differently, we have drawn inspiration from the theoretical sensibility developed by Chadwick (2013). In crafting the “ontology of hybridity,” he recognizes multiple boundaries where the notion of hybridity has been made use of. This inventory stretches from analyses of democratic vs. authoritarian political systems to analyzing new modes of governance through public-private partnerships, and from the reflective and innovative remix of media genres to the blurring of human and nonhuman actors in Actor-Network-Theory (Chadwick, 2013, pp. 9–15). We take seriously this general definition of “hybridity,” and note that its ontological anchoring in the existing institutionalized practices of social subsystems―through their sometimes nonlinear and often “contrapuntal” interaction―keeps the systems in a constant state of becoming. However, instead of focusing on specific institutionalized, interacting logics as such, we link the notion of hybridity to a question concerning a social process. We use the networked commemoration work on the Fukushima anniversary as an opportunity to consider how the process of collectively handling the trauma of a dramatically disruptive event plays out in 2 The article is part of a wider international research initiative on the mediation of the Fukushima nuclear power plant disaster that followed up on five anniversaries of the disaster. International Journal of Communication 12(2018) #fukushima Five Years On 931 the contemporary networked space. Here, the social theory of trauma outlined by Jeffrey Alexander (2012) offers a useful background. Starting from a culturalist-functionalist perspective, Alexander (2012) models an institutionally differentiated social process where a society slowly crafts a master narrative of a traumatic event. This involves identifying the loss or pain related to the traumatic events, identifying victims or “carrier groups” (people affected by the loss), and situating the wider audience. Such narrative reconstructions of traumatic events, then, are handled and filtered through a specialized, differentiated institutional order, where legal, aesthetic, religious, scientific, media, and state bureaucratic actors all process parts of the narrative, finally helping “the society” to come (more or less) culturally “to terms” with the event. Finally, in this model, as events become ritualized and normalized, the process allows “members of wider publics to participate in the pain of others” and thus “broaden the realm of social understanding and sympathy” (Alexander, 2012, pp. 15–30). Such a neo-functionalist approach to social systems has been criticized for theoretical reasons (see Joas & Knöbl, 2010, p. 336). In terms our research object, however, it helps in identifying an intersection between the intensified hybrid condition of the new media environment on the one hand and the more generalized, differentiated elements of the process of cultural trauma on the other. Reading Alexander’s model, we can see that it lies on a theoretical foundation that sees modern society as a set of differentiating social subsystems (or institutions) that usually play a role (through their own logics) in rationalizing and ritualizing disruptive events. Reading Chadwick’s build-up of the notion of hybridity, we can begin to see how such a process is situated in a radically new communication context. Analyzing the commemoration of the Fukushima disaster anniversary on Twitter provides an opportunity to reflect on how the old, institutionalized pattern processing of social trauma takes place in the new hybrid media environment. Social media serves as a communication resource for a wide variety of actors and institutions―from science to religion to NGOs and legacy media outlets―and makes their mutual relationships more complex. At the same time, however, these complex mutual relationships can still be seen as a part of an ideal shared collective process. As elements of a hybrid media environment, Twitter and other social media provide new channels to challenge and negotiate discourses produced by the mass media (Eriksson, 2016). During major media events, such as presidential election debates, interaction between Twitter users appears to be based on retweeting rather than the expression of personal opinions (Lin, Keegan, Margolin, & Lazer, 2014). However, widely retweeted content often seems to be produced by “elite users,” such as established media organizations, high-profile individual journalists, political parties or individual politicians, widely known organizations, and celebrities whose Twitter accounts have large numbers of followers (Freelon & Karpf, 2015; Lin et al., 2014). Previous studies also indicate that while Twitter is preoccupied with mainstream media, the relationship is rarely reciprocated (Rogstad, 2016). While the overall production of content by all users of Twitter increases during a media event, the dynamics of attention in the networked media environment clearly reward the aforementioned elite users who have pre-existing large audiences (Lin et al., 2014). Such an uneven distribution of attention is not unique to Twitter; online audiences of mainstream media websites and political blogs are strongly concentrated and follow a “winner-takes-all,” power-law distribution (Benkler, 2006; Hindman, 2009). The user’s number of followers and followees, 932 A. Rantasila et al. International Journal of Communication 12(2018) and the length of time they have owned their account, also increases the “retweetability” of their content, as does the presence of URLs and hashtags in a tweet (whereas the user’s number of previous tweets does not) (Suh, Hong, Pirolli, & Chi, 2010). Similar dynamics also appear to apply during crises (Bruns, Burgess, Crawford, & Frances, 2012). Despite the dominance of elite users, all users play a significant role in deciding what is shared, particularly as we look into the differentiated clusters of actors-relations in a Twitter network. Unlike in mass-media contexts, the users’ individual decisions (however predictable) ultimately constitute the network. Singer (2014) calls this “a two-step gatekeeping process” in which users have the power to enhance the visibility of content produced by media and other elite users by sharing it with secondary audiences. Indeed, in terms of the most retweeted messages, it is largely the nonelite users that usually upgrade the visibility of content produced by the elite (Lin et al., 2014; Singer, 2014). Data and Methods We collected data using Twitter’s Streaming API and the Digital Methods Initiative Twitter Capture and Analysis Toolset program (DMI-TCAT; Borra & Rieder, 2014). The former offers almost realtime access to Twitter’s global stream, with tweets retrieved using keywords or by focusing on specific users. While the partial matching of keywords is not possible, hashtag versions of keywords are matched (i.e. “fukushima” matches “#fukushima” but not “#fukushimadisaster”). We collected 17,619 Englishlanguage tweets that included the hashtag #fukushima between March 10 at 10:00:00 and March 12 at 11:59:59 UTC, 2016.3,4 Of these, 5,012 were original tweets and 12,607 were retweets.5 In total, 10,788 users participated or were retweeted in the discussion; 2,377 users produced their own tweets, while the rest only retweeted. As we are interested in the dynamics of power in the social process of commemorations, this material can be approached from two perspectives and using two complementary methods. While SNA constructs the Twitter feeds as networks, and thus articulates the actor-relationships that are a key aspect of social and political power, CDA looks at another aspect of political power, namely representations of the event and its affective dimension, victims, and political implications. Social network analysis is a strategy for researching various social structures based on graph theory, a branch of mathematics. In this approach, the research subject is conceptualized as consisting of actors (or nodes) and the connections (or ties) between them. This form of analysis has been used in the 3 The period in Coordinated Universal Time (UTC) when it was March 11 in at least one time zone on Earth. 4 To test the reliability of Twitter’s automatic language detection, we had two humans code a sample of 1000 tweets as either primarily English or non-English. The human coders then identified tweets where they disagreed on language. We compared the results against Twitter’s automatic classification and found a high agreement rate (Cohen’s κ = 0.95). Disagreement mostly concerned extremely short tweets or those containing multiple languages. 5 Not including “manual retweets” performed using expressions such as “RT @,” “MT @,” or “via @.” International Journal of Communication 12(2018) #fukushima Five Years On 933 social sciences to provide a deeper understanding of diverse phenomena, including belief systems, alliance and treaty systems, and international and transnational organizations (Cioffi-Revilla, 2010). Data obtained from social networking services such as Twitter can be conceptualized as a network, making SNA a powerful method and a sound starting point for analyzing these services. We constructed networks from our tweet sample and then examined them visually. Network visualizations are both representations of network structures and a means of communicating them to others6 (Freeman, 2000). Our approach resembled previously presented models of visual network analysis that focused on iteratively filtering, visualizing, and computing metrics in making sense of network data (Hansen, Rotman, Bonsignore, Milic-Frayling, Mendes Rodrigues, Smith, & Shneiderman, 2009; Huhtamäki, Russell, Rubens, & Still, 2015). An open source network analysis and visualization software Gephi (Bastian, Heymann, & Jacomy, 2009), and its Force Atlas 2 layout algorithm, were applied to visualize the data.7 As a result, actors that share a large number of connections are situated close to each other in the visualization. Finally, we used modularity to locate communities within the network, and nodes were colored according to the community to which they belonged.8 In addition to visualizations, we used degree centrality within a retweet network to identify key users. The weighted degree of node V is the number of times a user’s tweets were retweeted by others (for simplicity, we refer to this simply as “degree”). We also differentiated between in-degree (the number of times V’s tweets were retweeted) and out-degree (the number of times V retweeted other users’ tweets). We then identified the top users from each category. To determine dominant users, we looked at the number of retweets received, the number of retweets made, the number of original tweets posted, or the total number of tweets, and produced Lorenz curves and computed Gini coefficients for these distributions. A Lorenz curve shows the cumulative share of all retweets made or received, or tweets posted, at % y by the bottom % x of users. Thus, if this share is distributed evenly among all users, the Lorenz curve becomes linear, whereas a convex shape indicates that a small number of top users account for a large share of activity. The Gini coefficient can be defined as the ratio of the area that lies between a 45-degree line (representing a perfectly equal distribution) and the Lorenz curve, to the area beneath the 45-degree line. A high Gini coefficient indicates an unequal distribution of attention or activity. We also examined whether the users whose tweets had been retweeted were the same as those who had retweeted other users’ tweets or who tweeted more actively overall. For this, we used the Kendall rank correlation coefficient to measure whether a user’s ranking in one category (e.g., retweets received) correlated with their ranking in another category (e.g., retweets made). Therefore, it does not 6 We understand that visualizations are themselves discursive, and therefore are not objective representations of the data. 7 Force Atlas 2 is a force-directed layout algorithm, meaning that nodes in the visualization repulse each other while ties draw them together in an attempt to turn structural proximities into visual ones (Jacomy, Venturini, Heymann, & Bastian, 2014). 8 Modularity (Newman, 2006) is a property of networks that can be used to divide a network into clusters; members of a cluster have a large number of ties between them compared to actors outside the cluster. 934 A. Rantasila et al. International Journal of Communication 12(2018) depend on the absolute values of these variables. Finally, lists were created of the most popular tweets, news articles, and various other attributes. CDA extends from linguistically focused microanalysis to broader formations and constellations that shape larger universes of meaning, always focusing on how the use of language is tied to politics and power in society (e.g., Gee, 2014; Gee & Handford, 2014; Maeseele, 2015). The focus of CDA is on the specific meanings that utterances construct and the power relations these meanings shape and reinforce (Blommaert & Bulcaen, 2000; Fairclough, 1995; Wodak, 2013). Through an analysis of how actors and realities are constructed, it provides an opportunity to capture the ritual and political aspects of the meanings that played out in the #fukushima stream. A commemorative moment for a major traumatic disaster offers a strategic moment to study such discursive strategies that naturalize and legitimate social order (van Dijk, 1993; see also Alexander, 2012). Building on the results of SNA, we focused on tweets that were retweeted at least 10 times (n = 208), and paid attention to the accounts from which they were sent and the affiliations thereof. We looked particularly at the utterances and expressions used in depicting the initial event, and at whether or not the tweet included references to other events or actors. We also focused only on the content of the tweet itself, bracketing out URLs from the analysis. By combining these findings with the results yielded by the SNA, we were able to shed light on the combinations of actors and discourses that came to dominate the #fukushima feed on March 11, 2016. In the analysis below, we first describe the network structure of the #fukushima stream, with an emphasis on the connections between users and other hashtags used in the discussion. To define the relationships between mainstream mass media organizations and other actors, we also look at the identities of the most influential actors in the network. After depicting this “structure of commemoration,” we move on to consider the discursive action in this networked space by taking a more detailed look at the tweets retweeted at least 10 times. Focusing on the most retweeted content allows us to pinpoint the type of discourses that dominated the circulation of meanings, and enables us to explore the relationship between the most circulated discourses and the most influential users highlighted by the SNA. As Sumiala, Tikka, Huhtamäki, and Valaskivi (2016) argue, this kind of multimethod approach is essential in understanding how media events unfold in the contemporary hybrid media environment. The Network: Structure of Commemoration We formed networks from our data by interpreting Twitter users and hashtags as actors, and interpreting retweets and mentions of both users and hashtags as connections. In the analysis, connections were treated as directed, meaning that they did not apply the other way around. An adjacency list of connections in the form of {actor, actor} pairs was generated using Python scripts. We visualized and inspected three networks: 1) a user-hashtag mention network, formed from direct and indirect mentions of hashtags by users; 2) a user-user retweet network; and 3) a network that International Journal of Communication 12(2018) #fukushima Five Years On 935 showed all direct and indirect connections between users and hashtags.9 A visualization of this third network (Figure 1) shows some key user and hashtag actors, colored based on their modularity in order to make different communities (or subforums) stand out. Figure 1. Visualization of user-user and user-hashtag connections. (Several prolific private users within the #nuclear subforum are not shown.)10 9 In this case, if user A posted a tweet that included mentions of user B and hashtag H, the connections shown would be A -> B and A -> H. If user C were to retweet this tweet, the resulting connections would be C -> A, C -> B, and C -> H. 10 The large concentration of grey nodes at the edges are mostly users who tweeted using only the #fukushima hashtag either directly or by retweeting, and who did not explicitly connect to any of the 936 A. Rantasila et al. International Journal of Communication 12(2018) Many of the most popular hashtags and key users appeared as the central actors of clusters, which we interpreted as subforums of the larger, hybrid #fukushima forum in the network visualization. As the visualization shows, many of these subforums relate to powerful organizations that either have a stake in the nuclear energy debate or can be characterized as established media institutions. A Greenpeace subforum, shown in the top-right section of Figure 1, formed around Greenpeace International’s user account @Greenpeace and several other Greenpeace-affiliated accounts—such as that of the crew of the Greenpeace ship Rainbow Warrior III (@gp_warrior)—and around the hashtags #5yearsago, #nonukes, and #renewables. The activity within this subforum consisted mainly of a large number of retweets of several popular tweets created by Greenpeace and, to some extent, of replies to those tweets. Most tweets that included #5yearsago were authored by Greenpeace or were retweets of such tweets. A smaller subforum appeared on the opposite side of the visualization, centered on the International Atomic Energy Agency (IAEA) (@iaeaorg) and its hashtag #iaea. Another notable subforum appeared around the hashtag #japan, in which CNN International correspondent Will Ripley (@willripleyCNN) featured prominently. Other media-centric forums could be observed around the Russian state-funded RT11 (@RT_com, previously Russia Today), Ripley’s affiliate channel CNN International (@cnni), and National Geographic (@NatGeo). What these forums had in common is that they included a large number of retweets from users who did not otherwise engage in discussion relating to #fukushima. This is indicated in the visualization by the large number of separate smaller clusters around these accounts. In addition, users who retweeted these influential accounts most likely did not do so for other accounts. For instance, few users retweeted both @RT_com and @Greenpeace. The center of the network visualization is dominated by a large and sparse forum around the hashtags #nuclear, #radiation, #chernobyl, and #fukushimaanniversary, and by several highly active users that we could not identify as belonging to any established organization. Whereas the aforementioned, more tightly knit, forums formed due to the large number of users retweeting or mentioning content posted by a small number of users, the #nuclear-#radiation forum featured many connections between many users, although it is marked by an absence of users who dominate the forum as a whole. In a formal network analysis, this “forum” looks like a level field of discussion or interaction between diffuse groups of readers. Actors within this forum are also interconnected to the less central parts of the network. Table 1 lists users who received the most retweets (see Appendix) and who were therefore the most successful in spreading their message during the anniversary discussion. These users notably include international media outlets Agence France-Presse (@afp) and the German channel Deutsche Welle (@dwnews) (cf. Bruns et al., 2012; Lin et al., 2014). subforums (at least not by using the appropriate hashtags). In some 8,230 tweets, the only hashtag used was #fukushima. 11 See, for example: https://www.rt.com/usa/rt-government-broadcasting-radio/ (Accessed October 4, 2016 at 11:52 GMT+2). International Journal of Communication 12(2018) #fukushima Five Years On 943 possible variety of discussions that took place on Twitter during March 2016. Thus, we cannot make any claims about how dominant the tweets about Fukushima Daiichi were in the overall commemoration of the triple disaster, or about which actors and discourses dominated other Twitter networks outside the #fukushima one. Second, a major shortcoming is that the number of tweets retrieved using the Streaming API cannot exceed one percent of Twitter’s global traffic (Morstatter, Pfeffer, Liu, & Carley, 2013). If the number of tweets matching the search terms used is larger than this, some of them are discarded. During our data collection period we hit this limit several times, meaning that some data was lost. However, the incompleteness of any data obtained from Twitter is a well-known problem (boyd & Crawford, 2012; Driscoll & Walker, 2014). Another significant limitation is that keyword-based matching collects only those tweets that include the keyword. In our case, since we were looking for tweets that included the hashtag #fukushima, replies to those tweets were included only if they, too, included the hashtag (Bruns & Burgess, 2012; Lorentzen & Nolin, 2015). Studying a transnational media event in the hybrid environment demands that we simultaneously map the formation of networks and the circulation of meanings and discourses therein. This requires both a systematic multimethod approach and a healthy dose of humility regarding the conclusiveness of the evidence. Comparing different anniversaries over time would provide a deeper understanding of how discourses around remembrance are developed and shaped, and of the identity of those actors who gain the visibility and power required to reconstruct the event. Although Twitter is well suited to research focusing on dominant actors and discourses in large datasets, it will be important to study other media platforms to better understand the dimensions of the hybrid media environment and its workings during major events such as disaster anniversaries. One important step for future studies would be to examine how content circulates through multiple platforms in the hybrid environment. References Alexander, J. C. (2012). Trauma. A social theory. Cambridge, UK: Polity Press. Bastian, M., Heymann, S., & Jacomy, M. (2009). Gephi: An open source software for exploring and manipulating networks. In Third International AAAI Conference on Weblogs and Social Media, pp. 361–362. Benkler, Y. (2006). The wealth of networks: How social production transforms markets and freedom. New Haven, CT: Yale University Press. Bird, D. K., Haynes, K., van den Honert, R., McAneney, J., & Poortinga, W. (2014). Nuclear power in Australia: A comparative analysis of public opinion regarding climate change and the Fukushima disaster. Energy Policy, 65, 644–653. Blommaert, J., & Bulcaen, C. (2000). Critical discourse analysis. Annual Review of Anthropology, 29, 447–466. 944 A. Rantasila et al. International Journal of Communication 12(2018) Borra, E., & Rieder, B. (2014). Programmed method: Developing a toolset for capturing and analyzing tweets. Aslib Journal of Information Management, 66(3), 262–278. boyd, d., & Crawford, K. (2012). Critical questions for big data: Provocations for a cultural, technological, and scholarly phenomenon. Information, Communication & Society, 15(5), 662–679. boyd, d., Golder, S., & Lotan, G. (2010). Tweet, tweet, retweet: Conversational aspects of retweeting on Twitter. In 2010 43rd Hawaii International Conference on System Sciences, pp. 1–10. Bruns, A., & Burgess, J. (2012). Researching news discussion on Twitter. Journalism Studies, 13(5/6), 801–814. Bruns, A., Burgess, J., Crawford, K., & Frances, S. (2012). #qldfloods and @QPSMedia: Crisis communication on Twitter in the 2011 south east Queensland floods. Brisbane: ARC Centre of Excellence for Creative Industries and Innovation. Burgess, J., Galloway, A., & Sauter, T. (2015). Hashtag as hybrid forum: The case of #agchatoz. In N. Rambukkana (Ed.), Hashtag publics: The power and politics of networked discourse (pp. 61– 76). New York, NY: Peter Lang. Chadwick, A. (2013). The hybrid media system: Politics and power. Oxford, UK: Oxford University Press. Cho, S. E., & Park, H. W. (2013). Social media use during Japan’s 2011 earthquake: How Twitter transforms the locus of crisis communication. Media International Australia, 14(149), 28–40. Cioffi-Revilla, C. (2010). Computational social science. Wiley Interdisciplinary Reviews: Computational Statistics, 2(3), 259–271. Driscoll, K., & Walker, S. (2014). Big data, big questions. Working within a black box: Transparency in the collection and production of big Twitter data. International Journal of Communication, 8, 1745–1764. Eriksson, M. (2016). Managing collective trauma on social media: The role of Twitter after the 2011 Norway attacks. Media, Culture and Society, 38(3), 365–380. Eyre, A. (2007). Remembering: Community commemoration after disaster. In H. Rodriquez, E. L. Quarantelli, & R. R. Dynes (Eds.), Handbook of disaster research (pp. 441–455). New York, NY: Springer. Fairclough, N. (1995). Critical discourse analysis. Boston, MA: Addison Wesley. Freelon, D., & Karpf, D. (2015). Of big birds and bayonets: Hybrid twitter interactivity in the 2012 presidential debates. Information, Communication & Society, 18(4), 390–406. International Journal of Communication 12(2018) #fukushima Five Years On 945 Freeman, L. C. (2000). Visualizing social networks. Journal of Social Structure, 1(1), 1–10. Friedman, S. M. (2011). Three Mile Island, Chernobyl, and Fukushima: An analysis of traditional and new media coverage of nuclear accidents and radiation. Bulletin of the Atomic Scientists, 67(5), 55– 65. Gee, J. P. (2014). An introduction to discourse analysis: Theory and method (4th ed.). New York, NY: Routledge. Gee, J. P., & Handford, M. (2014). Introduction. In J. P. Gee & M. Handford (Eds.), The Routledge handbook of discourse analysis (pp. 1–6). New York, NY: Routledge. Hansen, D. L., Rotman, D., Bonsignore, E., Milic-Frayling, N., Mendes Rodrigues, E., Smith, M., & Shneiderman, B. (2009). Do you know the way to SNA? A process model for analyzing and visualizing social media network data. In Proceedings of the 2012 ASE International Conference on Social Informatics, pp. 304–313. Harder, R. A., Paulussen, S., & Aelst, van P. (2016). Making sense of Twitter buzz. Digital Journalism, 4(7), 933–943. Hindman, M. (2009). The myth of digital democracy. Princeton, NJ: Princeton University Press. Hommerich, C. (2012). Trust and subjective wellbeing after the great east Japan earthquake, tsunami and nuclear meltdown: Preliminary results. International Journal of Japanese Sociology, 21(1), 46–64. Huhtamäki, J., Russell, M. G., Rubens, N., & Still, K. (2015). Ostinato: The exploration-automation cycle of user-centric, process-automated data-driven visual network analytics. In A. S. Matei, G. M. Russell, & E. Bertino (Eds.), Transparency in social media: Tools, methods and algorithms for mediating online interactions (pp. 197–222). Cham, Switzerland: Springer International Publishing. Jacomy, M., Venturini, T., Heymann, S., & Bastian, M. (2014). ForceAtlas2, a continuous graph layout algorithm for handy network visualization designed for the Gephi software. PLoS ONE, 9(6), e98679. Joas, H., & Knöbl, W. (2010). Social theory: Twenty introductory lectures. Cambridge, UK: Cambridge University Press. Kalmbach, K. (2013). Radiation and borders: Chernobyl as a national and transnational site of memory. Global Environment, 11(1), 130–159. 946 A. Rantasila et al. International Journal of Communication 12(2018) Kasperski, T. (2012). Chernobyl’s aftermath in political symbols, monuments and rituals: Remembering the disaster in Belarus. Anthropology of East Europe Review, 30(1), 82–99. Katz, E., & Lazarsfeld, P. (1955). Personal influence: The part played by people in the flow of mass communications. Glencoe, IL: Free Press. Katz, E., & Liebes, T. (2007). ‘No more peace!’ How disaster, terror and war have upstaged media events. International Journal of Communication, 1, 157–166. Kyriakidou, M. (2014). Distant suffering in audience memory: The moral hierarchy of remembering. International Journal of Communication, 8, 1474–1494. Lagerkvist, A. (2014). 9.11 in Sweden: Commemoration at electronic sites of memory. Television & New Media, 15(4), 350–370. Li, J., Vishwanath, A., & Rao, H. (2014). Retweeting the Fukushima nuclear radiation disaster. Communications of the ACM, 57(1), 78–85. 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(5), e94093. Lorentzen, D. G., & Nolin, J. (2015). Approaching completeness: Capturing a hashtagged Twitter conversation and its follow-on conversation. Social Science Computer Review, 35(2), 277–286. Maeseele, P. (2015). Risk conflicts, critical discourse analysis and media discourses on GM crops and food. Journalism, 16(2), 278–297. Marwick, A. E. (2014). Ethnographic and qualitative research on Twitter. In K. Weller, A. Bruns, C. Puschmann, J. Burgess, & M. Mahrt (Eds.), Twitter and society (pp. 109–122). New York, NY: Peter Lang. Morris-Suzuki, T. (2015). Re-animating a radioactive landscape: Informal life politics in the wake of the Fukushima nuclear disaster. Japan Forum, 27(2), 167–188. Morstatter, F., Pfeffer, J., Liu, H., & Carley, K. M. (2013). Is the sample good enough? Comparing data from Twitter’s streaming API with Twitter’s Firehose. In Proceedings of the 7th International Conference on Weblogs and Social Media, ICWSM 2013, pp. 400–408. Newman, M. E. J. (2006). Modularity and community structure in networks. Proceedings of the National Academy of Sciences, 103(23), 8577–8582. International Journal of Communication 12(2018) #fukushima Five Years On 947 Rogstad, I. (2016). Is Twitter just rehashing? Intermedia agenda setting between Twitter and mainstream media. Journal of Information Technology & Politics, 13(2), 142–158. doi: 10.1080/19331681.2016.1160263 Siegrist, M., & Visschers, V. H. (2013). Acceptance of nuclear power: The Fukushima effect. Energy Policy, 59, 112–119. Singer, J. B. (2014). User-generated visibility: Secondary gatekeeping in a shared media space. New Media & Society, 16(1), 55–73. Starbird, K. (2017). Examining the alternative media ecosystem through the production of alternative narratives of mass shooting events on Twitter. Retrieved from http://faculty.washington.edu/kstarbi/Alt_Narratives_ICWSM17-CameraReady.pdf Suh, B., Hong, L., Pirolli, P., & Chi, E. H. (2010). Want to be retweeted? Large-scale analytics on factors impacting retweet in Twitter network. In Social Computing (socialcom), 2010 IEEE Second International Conference on Social Computing, 177–184. Sumiala, J. (2013). Media and ritual: Death, community and everyday life. London and New York, NY: Routledge. Sumiala, J., Tikka, M., Huhtamäki, J., & Valaskivi, K. (2016). #JeSuisCharlie: Towards a multi-method study of hybrid media events. Media and Communication, 4(4), 97–108. Sznaider, N. (1998). The sociology of compassion: A study in the sociology of morals. Journal for Cultural Research, 2(1), 117–139. van Dijck, J., & Poell, T. (2013). Understanding social media logic. Media and Communication, 1(1), 2–14. van Dijk, T. (1993). Principles of critical discourse analysis. Discourse and Society, 4(2), 249–283. Wodak, R. (2013). Editor’s introduction: Critical discourse analysis—challenges and perspectives. In R. Wodak (Ed.), Critical discourse analysis. Volume 1: Concepts, history, theory (pp. xxi–xlv). London, UK: SAGE Publications. World Health Organization. (2013). Health risk assessment from the nuclear accident after the 2011 Great East Japan earthquake and tsunami, based on a preliminary dose estimation. Retrieved from http://apps.who.int/iris/bitstream/10665/78218/1/9789241505130_eng.pdf?ua=1 948 A. Rantasila et al. International Journal of Communication 12(2018) Appendix: Tables Table 1. Users With Highest Degrees in the Retweet Network (Ranked by Their In-Degrees). User In-Degree OutDegree Degree Description @Greenpeace 1472 4 1476 Greenpeace (environmental organization) @RT_com 837 1 838 RT (TV/news network) @cnni 587 5 592 CNN (TV/news network) @willripleyCNN 585 2 587 Will Ripley (CNN International correspondent) @NatGeo 407 0 407 National Geographic (magazine) @iaeaorg 321 0 321 International Atomic Energy Agency @gp_warrior 282 3 285 Rainbow Warrior (Greenpeace ship) @newsbreakslive 278 0 278 BreakingNewsFeed.com (news aggregator) @environews 216 4 220 Enviro News (alternative environmental news website) @greenpeaceCA 194 3 197 Greenpeace Canada @gpj_english 162 11 173 Greenpeace Japan @NatGeoPhotos 158 0 158 Photographs by National Geographic @TEDtalks 149 0 149 TED (nonprofit organization that hosts conferences with speakers from various disciplines) @greenpeaceusa 144 0 144 Greenpeace USA @naturenews 143 0 143 Nature (academic journal) @efmania 133 0 133 Japanese account posting Formula 1-related content @doomsdayscw 121 177 298 Noninstitutional user @dwnews 112 3 115 Deutsche Welle (international public broadcaster) @afp 103 0 103 Agence France-Press (news agency) @frediteres 98 23 121 Noninstitutional user International Journal of Communication 12(2018) #fukushima Five Years On 949 Table 2. Gini Coefficients and Correlations. Type N / Gini / correlation A: Users who tweeted at least once 10,788 Gini retweet in-degree 0.99 Gini retweet out-degree 0.42 Gini all tweets posted 0.36 Gini original tweets posted 0.89 Correlation rt in-degree and rt out-degree -0.32 Correlation rt in-degree and all tweets posted 0.20 B: Users who tweeted and were retweeted 884 Gini retweet in-degree 0.82 Gini retweet out-degree 0.67 Correlation rt in-degree and rt out-degree 0.14 Correlation rt in-degree and all tweets posted 0.26 Correlation rt in-degree and original tweets posted 0.27 C: Users who posted original tweets 2377 Gini original tweets posted 0.48 Correlation rt in-degree and original tweets posted 0.22 D: Users who posted original tweets and were retweeted 873 Gini retweet in-degree 0.82 Gini normal tweets posted 0.61 Correlation rt in-degree and original tweets posted 0.27