Tweets That Matter: Reconsidering Journalistic Sourcing and Framing Processes in the Context of the #Grexit Debate

: This study explores the news media Twitter messaging on the issue of Grexit, as an exemplary case of transmediatisation of problems in highly polarized contexts. Our analysis focuses on media tweets (in English, French, Italian, and Greek) using the Grexit hashtag between March and July 2015. There are three main questions on the potential reshaping of journalistic sourcing and framing on Twitter. The ﬁrst focuses on the milieu of actors used by media outlets as sources in the #Grexit debate, the second on the types of news frames that dominated #Grexit media tweets, and the third on how sourcing and news frames interact to construct a space of power positions. The above processes took shape within a close information system, which included politicians, media elites, and economic experts that marginalized alternative voices and critical perspectives. These ﬁndings indicate that mainstream news media normalized Twitter to ﬁt their traditional sourcing and framing norms and practices. More speciﬁcally, our ﬁndings indicate the following: ﬁrst, traditional sources and powerful economic actors get easier access to online media reporting on Twitter; second, the negative and episodic media-driven frames take the lead in the frame-building process; and third, the non-elite political and socially-driven frames are marginalized in the framing building process. The Twitter a ﬀ ordances were essentially normalized by media to ﬁt into their understandings of the negotiation process as a high-stakes international politics and economic game with predetermined winners and losers. It is also likely that this normalization reﬂects the normalization of Twitter by powerful political and economic elites aiming to o ﬀ er journalists on Twitter easy and instant access to their narratives. et study journalists’ tweets U.S.


Introduction
Twitter has been widely exploited by media outlets and journalists alike. Firstly as a free service to drive traffic to their news' pieces, but also as a handy tool for sourcing leads, newsgathering, and news reporting (Malik and Pfeffer 2016;Heravi and Harrower 2016;Newman 2009Newman , 2011Lasorsa et al. 2012;Hermida 2013;Vis 2013;Artwick 2013;Hernández-Fuentes and Monnier 2020;Russell 2019). Twitter, however, means much more to journalism. The blending of information, of discursive possibility, and social networking (Bruns and Burgess 2012) seems to be a determinant factor to the role of Twitter in professional mainstream journalism today. The first-hand reporting of events as they occur and instant assessment of the newsworthiness of events combined with Twitter's massive use for ongoing discussions construct an ambient news environment (Papacharissi 2015;Van Aelst et al. 2015;Hermida 2010;Burns 2010). In this environment, anyone, not just professional journalists and news media, can publish, aggregate, reframe, and define "news".
The possibilities offered by social media for audience engagement in news production have provided the ground for extensive academic discourse and research on the ways mainstream journalistic norms and practices on sourcing and framing are re-shaped in this networked, ambient news environment. Two contrasting and in parallel complementary narratives emerged from relevant academic discussions and research. The first narrative is structured around the wider theme of "networked" or "participatory" journalism and its implications on mainstream gatekeeping norms and framing practices that create new paradigms of journalism. The second narrative is based on the "normalization" theme, which suggests that traditional news media use social media in ways that fit them into established journalistic norms and practices. These narratives essentially function as working hypotheses aimed to advance academic inquiry into how mainstream professional journalism is (re)shaped in social media platforms.
Using these narratives as a backdrop, this study aims to explore and discuss how international and national (Greek) mainstream news media covered the Grexit issue on Twitter and in particular their sourcing and framing practices. The purpose of this study is to explore how mainstream news media use Twitter to cover polarized political debates in the context of negotiations between international actors and particularly between international creditors and debtor states and to discuss whether and to what extent their tweets reflect mainstream journalistic norms and practices in sourcing and framing or less conventional ones. As will be discussed later, academic discourse and research on the ways in which mainstream journalistic norms and practices in sourcing and framing are re-shaped on Twitter are focused predominantly on such issues and events as demonstrations and political campaigns, about which mainstream news media narratives have been greatly challenged by counter publics and by populist and new 'anti-establishment' media and populist political actors. However, much less is known about how media use Twitter to cover polarized high-stakes political debates that involve international economic and political actors. This paper covers a largely transmediated problem (mediatized through nationally different media arenas) as a case study of analysis of sourcing and framing practices in highly polarized negotiation processes.

Sourcing and Framing Journalistic Norms and Practices in the Age of Social Media
During the last decade, a body of research suggests that Twitter gives voice to marginalized issues and publics, particularly under situations where access to media is limited or restricted by repressive regimes (Hamdy 2010;Tufekci 2017;Lee 2018). From a wider perspective, this body of research has focused on the role of social media during popular protests, where networked counterpublics, struggling to amplify their messages, mobilize support, and coordinate collective action, turn to social media to articulate their narrative of the events and create news frames that oppose those created by mainstream media (Hermida 2013;Ince et al. 2017;Jost et al. 2018;Agur and Frisch 2019). Within this increasingly contested boundary space between news creation, publishing, and use (Lewis 2012), particularly in the context of popular protests and civil unrest, the traditional journalistic gatekeeping norm that is reflected in the heavy reliance on elite sources to get newsworthy and credible information (Sigal 1973;Hall et al. 1978;Gans 1979) has been greatly challenged.
In the cases of Occupy Wall Street (Penney and Dadas 2014), the Egyptian uprising of 2011 (Meraz and Papacharissi 2013), and the Gezi Park resistance in Turkey (Çoban and Ataman 2017), it has been argued that Twitter enabled activists, bloggers, and citizens to document events 'as they happened'. Social media enabled marginalized voices to sustain alternative narratives alongside mainstream media's framing of the protests in Egypt (Hamdy and Gomaa 2012). Twitter was used to break news, but also to monitor rumors reported as facts during the 2008 terrorist attacks in Mumbai (Jewitt 2009). It also enabled the outside world to get insider information on the 2009 Iranian presidential election protests when the authorities had enforced a media blackout (Meraz and Papacharissi 2013). Other research further suggests that Twitter enabled a form of alternative journalism aimed to challenge the mainstream media's point of view when covering protest activities and criticize it as biased and manipulative (Hermida and Hernández-Santaolalla 2018). According to Hermida et al. (2014), their study on the use of sources by National Public Radio's Andy Carvin on Twitter during key periods of the 2011 Tunisian and Egyptian uprisings pointed to " . . . the emergence of a new style of near real-time gatekeeping, where journalists cite a potentially broader set of sources through social media." (p. 495). As Hermida (2013) has argued, the use of Twitter in popular protests illustrates "that core journalistic norms of gatekeeping and framing are being reshaped from outside the field" (p. 304). Such reshaping took place also by the extensive exploitation of social media by populist movements and leaders, who traditionally gained little attention from mainstream media, to disseminate their "unmediated" messages. They used social media to circumvent mainstream media to reach wider audiences and to attack "media elites" (Engesser et al. 2017;Stier et al. 2017). Research findings indicate that such exploitation of social media has also led to a rise in mainstream media attention on populist candidates and their messages (Wells et al. 2020) and even to the unintentional amplification of disinformation by foreign-sponsored actors (Lukito et al. 2020).
Although pressures from outside the field do create huge tensions, professional journalism has struggled to reconcile them by engaging its online audience in processes of news production and sharing. This presupposes that established media and professional journalists are willing to move away from their traditional gatekeeping role towards gatewatching, news moderating, and facilitating (Jarvis 2006;Beckett and Mansell 2008;Bruns 2005Bruns , 2018Reese and Shoemaker 2016). Scholars in the field have coined several terms to describe joint forms of online news content production and sharing and discuss how journalists, news sources, and audiences interact online (Jarvis 2006). The term "networked 4th estate" (Benkler 2011) points to the combined action of professional journalists, citizens, and social movements to form a decentralized democratic discourse, a "networked journalism" (Jarvis 2006), that blends collaborative and collective action into "participatory journalism" (Allan 2006;Singer et al. 2011;Hermida 2012;Borger et al. 2016;Lawrence et al. 2018). Other terms like news "produsage" (Bruns 2005(Bruns , 2018 and "audience mutualisation" also aim to describe the processes through which news creators and audiences/users interact to construct news communities " . . . where ideas and news are shared rather than delivered" (Editorial, The Mutualisation of News 2009, Guardian, 27 July). The above indicate that online networking offers the potential to blur the boundaries between professional journalism and citizen engagement in news production and sharing in ways that are not antagonistic, but synergistic.
However, other scholars in the field argue that professional journalism tends to absorb and exploit rather than genuinely embrace participative practices. For example, the normalization hypothesis proposed by Singer (2005) and further explored in a considerable body of research (e.g., Lasorsa et al. 2012;Parmelee 2013;Molyneux and Mourão 2019;Duffy and Knight 2019) suggests that traditional news media tend to use social media in ways that, to a large extent, fit them into established journalistic norms and practices, while in parallel, making some adjustments in response to their affordances for audience engagement and round-the-clock newsgathering and reporting. Boundaries are bound to be defended by those who have a strong interest in keeping them. It is to be expected that, in times of an increasingly contested professional boundary space, journalists and media will defend their interests; but this offers only a partial explanation for why journalism may resist embracing genuinely participative practices in the age of social media. As Reese and Shoemaker (2016, p. 390) argue, the new public spaces may not fit easily into traditional journalistic norms and values. Thus, journalists are bound to defend their profession against contesters also for ideological reasons based on their professional logic that " . . . serves as a general conceptual frame through which to organize the discourse on journalism's norms, routines, and values" (Lewis 2012, p. 845) and privileges them as gatekeepers over news content on behalf of society. Boundaries have to do with power and it is, therefore, imperative that the new media configurations supporting these spaces should also be understood within a context of a larger framework of power (Reese and Shoemaker 2016, p. 390).
On the other hand, perhaps more than anything else, the huge growth of fake news in social media (Allcott and Gentzkow 2017;Guess et al. 2018;Fletcher et al. 2018;Mourão and Robertson 2019;Bradshaw et al. 2020) and the " . . . 'weaponisation' of social media to target, threaten, and harass independent news media, individual journalists, and their audiences" (Posetti et al. 2019, p. 10; see also Thornton 2020;Lewis et al. 2020) offer a fresh ground to defend and reassert the critical importance of professional journalism in open and democratic societies. As Beckett (2017) has characteristically argued, "in my sector of journalism, fake news is the best thing that has happened for decades. It gives mainstream quality journalism the opportunity to show that it has value based on expertise, ethics, engagement and experience". Gatekeeping, as a core journalistic role in an era of abundant and increasingly "disordered" online information (Wardle and Derakhshan 2017), is reaffirmed based on fundamental journalistic norms. If professional journalism aims to defend the public interest in this environment, it should be able to act as independently as possible, try to avoid bias and be impartial, stick to verified facts and credible and transparent verification methods, be accurate, and refrain from taking sides on issues of public controversy including politics. As Reuters proudly states, "our reputation for accuracy and freedom from bias rests on the credibility of our sourcing" (Reuters. n.d.).
Early research on the blogging activity of journalists in the USA showed that, while most of the journalists did normalize their blogs by maintaining control over the information provided, they also tended to express more openly their personal opinion, thus moving away from non-partisanship (Singer 2005, p. 192). Overall, however, j-bloggers appeared to be appropriating the blogging format into their commentary formats in traditional media. Lasorsa's (2012) study on journalists' use of Twitter concluded that " . . . j-tweeters appear to be normalizing microblogs to fit into their existing norms and practices but, at the same time, they appear to be adjusting these professional norms and practices to the evolving norms and practices of Twitter." (p. 31).
Numerous empirical studies have shown that legacy media traditionally turn to powerful actors when writing their stories (e.g. Bonfadelli 2000;Gans 1979;Sigal 1973;Tresch 2009). According to Davis (2010), new media use has encouraged patterns of "closed information systems", where policy elites (politicians, officials, and journalists) form networks hidden from the wider public. Recent research shows that this norm is also predominantly applied by established news organizations regarding online news sourcing on Twitter and other social media. Research conducted by Knight (2012) indicated that journalists covering the 2009 Iranian presidential election protests turned to more traditional sources of information-political statements and expert opinion, rather than information coming from citizen journalists on Twitter. Lawrence et al.'s (2014) study of journalists' tweets during the 2012 U.S. presidential campaign concluded that, despite the inclusion of opinion and expression, the traditional scheme of gatekeeping did not change. Parmelee's (2013) qualitative study on how political journalists at U.S. newspapers used Twitter in their daily reporting of the 2012 presidential campaign indicated that, although Twitter changed the daily work routines, their practices on Twitter were conforming with the " . . . generations-long tradition of journalists serving as objective gatekeepers" (p. 302). Other research further supports these conclusions. Lecheler and Kruikemeier (2016), in their review of 22 research studies related to online news sourcing, concluded that "journalists still gravitate toward elite sources, and do not show a decrease in traditional news sourcing techniques. In a sense, online news sources seem to have simply shifted a part of the information search routine into a new space (i.e., normalization)." (p. 167). Thorsen and Jackson (2018), exploring the sourcing practices in online news and live blogs of three U.K. news organizations, found that elite-centered sourcing practices prevailed in both formats. Particularly regarding political news, the news organizations were found to rely heavily on politicians and media as news sources. Mourão and Molyneux (2020), in their study of how political reporters in mainstream and non-mainstream outlets in the United States covered on Twitter the first presidential debate in 2016, found out that mainstream journalists essentially reinforced their role as gatekeepers by predominantly using as sources or by interacting with other journalists instead of other actors outside the mainstream media field. Several other studies have also shown that journalists mainly interact with their colleagues on their Twitter accounts and tend to predominantly retweet content published by other journalists and media outlets (Nuernbergk 2016;Folker and Nölleke 2019;Bentivegna and Marchetti 2018).
The academic discourse and research around how mainstream media sourcing practices are re-shaped by and in social media are particularly important because news sources provide their interpretations of events that have the power to frame the news content according to their agendas and ideology. According to research conducted by Hänggli (2012) on the relationship between political actors and news media in the context of the 2008 political campaign on immigration in Switzerland, the political messages of powerful political actors were more likely to be reported by media as compared with those of weaker ones. Research on the framing of the Euro crisis between 2010 and 2014 in German and Spanish online quality newspapers revealed that their preferred frame of "conditional assistance" (i.e., financial support to troubled countries on the condition of austerity measures) aligned to the preferred policy proposals of the political system in power (Kaiser and Kleinen-von Königslöw 2017). These findings confirm the "power bias hypothesis", which states that media attention is biased towards more powerful actors (Hänggli 2012). Such actors as government officials or heads of financial institutions are likely to get more media attention not only because their decisions and actions may potentially influence the life of a large number of people, but also because they often devote resources and develop tactics to manipulate reporting routines in their favor (Shoemaker and Reese 2014, p. 190;Garland 2017;Marland 2017). Therefore the choice of news sources can potentially have a huge impact on the impartiality and objectivity of media's coverage of events and issues. Sourcing and frame-building, as a process that takes place in a continuous interaction between journalists, their sources, and audiences (De Vreese 2005), are two closely linked journalistic practices, and the study of their relationship in the context of online news creation can generate interesting insights into journalism. According to Entman (1993), "to frame is to select some aspects of a perceived reality and make them more salient in a communicating text, in such a way as to promote a particular problem definition, causal interpretation, moral evaluation and/or treatment recommendation for the item described" (p. 52). Generic news frames, e.g., news frames that can be applied in the coverage of any issue or event, are often distinguished between episodic and thematic frames. As has been pointed out by Iyengar (1990Iyengar ( , 1991Iyengar ( , 1996, political news coverage in the United States is predominantly relying on episodic frames. Reporting is limited to the day-to-day events and circumstances and often fails to also employ thematic frames that provide the historic, social, and economic context of an issue. According to Iyengar (1991) the prevalence of episodic framing elicits individualistic attributions of responsibility for social problems and diverts attention from the wider political and economic causes and remedies of social problems. Another example of dominant generic framing in political news is the "strategy frame", which emphasizes the game aspects of politics rather than the social problems that political action aims to address. As Cappella and Jamieson (1996) have argued, strategic coverage, which was found to be prominent in U.S. media, activates cynicism about campaigns, policy, and governance because, as an underlying interpretive frame, it reduces the motivation to political action " . . . to a single, simple human motivation: the desire to win and to take the power . . . " (p. 81). Research on generic frames in political news coverage indicates that the national political culture and context, national government policy, and the structure of the media field influence the framing choices of legacy media. In other words, as Shoemaker and Reese (2014) propose in their hierarchical model of influence, frame-building is a process influenced not just by individual journalist choices and standardized professional practices or by individual media policies and norms (micro to meso levels of influence), but also by a wider system of political, economic, and cultural factors (macro levels of influence). For example, research by Semetko and Valkenburg (2000) on four newspapers and three TV outlets in Holland revealed that, although television news was " . . . just as episodic in character as in the U.S . . . ." (p. 106), the attribution of responsibility on issues and problems was predominantly directed to the government instead of to the individuals. On the other side, the second most common generic frame in Dutch news was found to be the "conflict frame", which, just like the "strategy frame", focuses upon the political conflict between individuals, groups, or institutions. A comparative framing analysis of media coverage of terrorism in prominent U.S. and U.K. newspapers by Papacharissi and Oliveira (2008), revealed that the U.S. newspapers were more episodic in their coverage, while the U.K. ones more thematic. A comparative analysis of election news coverage in Sweden and the United States provided similar findings (Strömbäck and Dimitrova 2006). On the other side, Camaj (2010), in her study on Kosovo's status negotiations, underlined the congruency in international news agencies, which mainly utilized the 'episodic frame' and emphasized the 'conflict' nature of the issue, particularly by the end of the polarized negotiations over Kosovo's final political status when the negotiation process reached a heated point. The above indicate that, within a globalized news environment, there are observed, on the one hand, tendencies of homogenization of framing practices, and on the other hand, persistence of discrete national/regional media norms guiding framing practices.
The technological affordances of the different social media platforms, as well as their actual use by millions of people around the world, have exerted their own direct and indirect influence on frame building by mainstream media and journalists in their social media accounts. But how much and in what ways have established news frame-building practices changed in these platforms? The questions that emerge are indeed numerous. In the case of Twitter, one could ask how, in what kinds of events, and to what extent frame-building practices by established news media are reproduced on Twitter or new frame-building patterns emerge depending on the news source(s)? How, to what extent, and why the Twitter format may privilege specific types of frames over others in news reporting? On the other side, do different types of news media (e.g., print, broadcast, digital-only) use their Twitter accounts in differentiated ways in terms of both sourcing and frame-building? When, in what kinds of events, and why? In a field as fluid as this, such questions are likely to resurface and be addressed by new research only temporarily as the communication formats; the affordances of online services; as well the collective behaviors, tastes, and preferences change over time.

The #Grexit Media Tweets as a Case Study
In 2010, Greece attracted international interest because of its severe sovereign debt crisis. Greece agreed with the so-called "Troika" (comprised of the European Commission (EC), the International Monetary Fund (IMF), and the European Central Bank (ECB)) on a bailout program that led to harsh austerity measures and reforms that caused significant political ruptures and social upheaval. According to Moreira Ramalho (2020), the role of the Troika was one of the 'master-framer' of the euro crisis, managing to successfully frame the euro-crisis as a result of Greece's (as well as Spain's and Portugal's) domestic failure and moral hazard, which provided a strong legitimation for austerity.
In the 2012 national elections, the two main political parties, the conservative center-right ND and particularly the center-left PASOK, suffered huge losses. SYRIZA, a radical left party, became the official opposition. In January 2015, Greek national elections SYRIZA formed a coalition government with ANEL, a minor right-wing party. The new government under PM Alexis Tsipras followed an aggressive strategy in negotiations with Greece's official lenders. During the six months of negotiations (February to July 2015), several scenarios were circulated based on the possibilities of an agreement between Greece and its creditors, a temporary compromise, or a complete break between Greece and the Eurozone. In the above context, the term "Grexit" has been used to describe the possibility of Greece's exit from the Eurozone. The stakes fueling the Grexit issue were high enough to potentially shake the stability of the Eurozone.
How Greece's crisis was covered by legacy media has attracted the interest of researchers. Tracy (2012), for example, analyzed the main frames of the early phase of the Greek crisis (2009)(2010) in the U.S. mainstream media, arguing that, as a problem, the case of the Greek crisis was presented as the 'Greek Contagion', and that the recommended solution was 'austerity'. A considerable body of research has pointed out that, during the economic crisis, mainstream media were aligned in their support to Troika's framing of the crisis, while alternative voices emerged (mainly online) because of pressures in journalistic industry and culture (Pleios 2013;Mylonas 2014;Iosifidis and Boucas 2015;Nikolaidis 2015;Touri et al. 2016;Doudaki et al. 2016). Furthermore, Stavropoulos (2017) discussed how news concerning the Grexit discussion between January 2016 and February 2017 cultivated an uncertainty frame, resulting in moral and economic panic. Similarly to Greece, research on how mainstream media framed bailout programmes in other southern European countries, such as Cyprus (Doudaki et al. 2019), Spain (Rios Rodríguez 2020), and Portugal (Sousa and Santos 2014), has indicated that coverage was dominated by frames that legitimized austerity.
In contrast to the mainstream media narratives of the Greek crisis, research has revealed the use of configuration and diagnosis generic frames of the debt crisis by countervailing voices (Kountouri and Nikolaidou 2019). The first of those frames defined the crisis as the economic and moral bankruptcy of the model of development adopted by the country since its transition to democracy (1974) and considered the deepening of the crisis a result of a long period of austerity. The second one defined the crisis as humanitarian, while the third considered it to be a crisis of democracy. All three frames interpreted the Greek crisis through an economic and moral, but also a political lens. Crucially, they all linked to a powerful diagnostic frame that unveiled the crisis as a systemic problem of capitalism. Those thematic rather than episodic frames viewed the Greek crisis as an international systemic problem.
Within this wider context of media coverage of the Greek crisis, this paper focuses on the #Grexit tweets of mainstream news media in English, French, Italian, and Greek language during around three and a half months (March-July 2015) when Greece was engaged in heated negotiations with its international creditors. We focus on the #Grexit discussion based on five interrelated research questions: RQ1: Which were the sources used by news media on their contributions to the #Grexit discussion on Twitter?
RQ2: To what extent those sources reflect traditional journalistic sourcing norms or emergent ones?
The above questions are framed within the wider discussion on whether the use of Twitter has somehow changed and in what ways the established norms and values of journalism on gate-keeping in general and sourcing in particular.
RQ3: Which were the dominant frames of the news media's #Grexit posts on Twitter?
RQ4: To what extent those frames reflect traditional norms on news media frame-building or emergent ones?
To address the above questions, we built on previous work by applying framing theory (Scheufele 1999;Scheufele and Nisbet 2008;Druckman and Nelson 2003;Borah 2011) to understand the processes through which prominent sources negotiate the flow of information and advance frames to prominence. It is assumed that the main frame-builders are the prominent sources producing dominant structures of narratives (Entman 1993). Furthermore, by building on news values theory (Staab 1990), we argue that, while negativism and conflict lead to more publicity in mainstream media, it could be expected that news on Twitter focuses more on consequences than on positions taken by negotiating sides and that journalists try to keep control on their field of action through ignoring frame constructions of other actors. Based on this assumption, we go a step further to ask the following: RQ5: How sourcing and framing interact in the news media's #Grexit posts on Twitter? To what extent and in what ways such interactions produce established hierarchies in news reporting or challenge them?
In our view, the meaning of sources and frames is also relational and cannot be grasped by only looking at each component separately. Reconsidering the practices of sourcing and frame-building raises new questions of power relations and control that need to be revisited in the context of the global audience interaction, the abundance of news online, and social media platforms.
Before turning to the methodology, let us summarize the hypotheses. First, we expect that the power-bias hypothesis, which states that traditional sources, but also powerful (political and media) actors get easier access to the media than weak actors, is also valid in the social media environment. Second, we expect that the news value theory, stating that negativism and episodic coverage of media news, matters for the social media too. In that case, the salience of the media-driven frames takes the lead in the frame-building process. Third, the political-driven frames are multiplied by the news media accounts as a specific framing process leading to the negative impact of the Grexit discussion.
Our analysis of the news sources in media's #Grexit tweets shows that, in all four languages, elite media and journalists, powerful institutions, and experts were the more visible actors in the #Grexit debate. Activists, non-elite media, and journalists did not emerge as dynamic agents in reshaping the sourcing and frame-building choices made by mainstream media outlets on Twitter. In other words, even in social media, the traditional hierarchy of sourcing that privileges some voices over others appears to persist in high-stakes media stories. We can safely assume that this was a calculated mainstream media's choice to maintain their privileged role in controlling the flow and form of news even in the hybrid news space of social media. Within the #Grexit discussion, media outlets predominantly zoomed in on the evolving story of the talks, the factors that affected this process, and the accompanying scenarios of a Grexit. However, they somehow failed to widen their lens to place into the picture background factors (e.g., Greece's failing economy or the Eurozone's structural problems) and other emerging threats to the stability of the European Union as a whole, which, less than two years later, did materialize with the Brexit. Overall, our findings suggest that media outlets adopted almost exclusively episodic frames, reporting and commenting upon day-to-day events and circumstances and constructing a 'conflict' narrative of a negotiation game involving blame and scare tactics, and effectively ignoring thematic frames that would provide the historic, social, and economic context of the Grexit issue. Comparatively, Greece's positions received much less attention from media outlets in all languages. This finding reveals a fundamental imbalance in covering Greece's perspective on the issues and the events of the period. Even the Greek media, by relying heavily on international media, journalists, and politicians as news sources, effectively managed to frame much of their news coverage on #Grexit from the perspective of the positions of the creditors. On the other hand, a more balanced framing approach was applied to the creditors' side in the negotiation table, focusing not only on their positions, but also on the challenges they face in the case of an exit of Greece from the Eurozone. This 'one-sided balance' news framing reveals an underlying generic framing strategy to apply a rounded coverage (both reporting and in-depth analysis) on the most powerful side in an evolving negotiation, while effectively ignoring the weaker side's positions. Interestingly, the media studied did not tweet anything on the creditors' negation assumptions. Perhaps, in a negotiation process where there is such a clear imbalance in the negotiating power between the two sides, media outlets do not find it necessary to tweet on the "obvious", i.e., that at the end of the day, the more powerful side will make use of its superior power to coerce, if necessary, the weaker side. Perhaps they also do not want to simply succumb to cynicism. Overall, the 'one-sided balance' news framing on Twitter during the negotiations between Greece and its creditors may be interpreted as a result of factors specific to dominant journalistic norms and practices regarding sourcing and framing (episodic framing emphasizing conflict and game aspects of negotiations), as well as political and economic factors originating from outside the field reflected in the actual imbalance of negotiation power between the negotiating parties, and finally from ideological factors given the dominance of the 'austerity-for-bailout' narrative among international and national political and economic elites. The Twitter affordances were essentially normalized by media to fit into their understandings of the negotiation process as a high-stakes international politics and economic game with predetermined winners and losers. It is also likely that this normalization reflects the normalization of Twitter by powerful political and economic elites aiming to offer journalists on Twitter easy and instant access to their narratives.

Sourcing in a Networked Public Sphere. Business As Usual?
Our analysis of the news sources in media's #Grexit tweets shows that, in all four languages, elite media and journalists, powerful institutional actors (political leaders and ECB/IMF), and experts (market economists and academics) were the more visible actors in the #Grexit debate (see Table 1). In contrast, the voices of "other" sources that could provide countervailing views, such as union leaders, activists, protesters, or citizens, were not given much space, i.e., they did not appear to be valued as important or relevant in this debate. This finding further supports the normalization hypothesis, that traditional news media use social media in ways that fit them into established journalistic norms and values. It should also be noted that almost 61% of the media outlets' (except for the Greek outlets) promoted their own, internal news sources, mainly via mentions to posts from their staff and links to pieces published on their platforms. Host media organizations and their top journalists were, by far, the most frequent sources of #Grexit news tweets by media outlets' accounts. Both those sourcing practices, either through promoting on Twitter #Grexit news stories published in the media outlet's platform or through mentions and links to pieces of their staff, indicate that the boundaries of the media-journalists elites were strongly defended on Twitter. It is also interesting that English, French, and Italian news organizations did not tweet news from news outlets outside their national/linguistic boundaries. These findings complement those of earlier studies. For example, Armstrong and Gao (2010) as well as Holcomb et al. (2011) found that news organizations use Twitter not so much as a reporting tool, but as a means to drive traffic to their news sites. In Holcomb et al.'s (2011) study, 93% of the postings offered a link to a news story on the organization's website. Thurman and Walters' (2013) study offers similar evidence. Their study on Guardian.co.uk's use of Live Blogs showed that, although reporters made use of information from Twitter, the information flow was one-way as Live Blogs were housed in Guardian and did not feedback the Twitter.
In contrast, news outlets in Greece re-tweeted #Grexit headlines and shared stories from global/international elite-oriented media, which have a privileged role among globalized elites. It appears that foreign media (especially Bloomberg, Financial Times, Reuters, Guardian, Spiegel, and Wall Street Journal) had a prominent role in the sourcing process of Greek media. This finding indicates what Corcoran and Fahy (2009) have pointed out, that power follows within and across national contexts through elite-oriented media and global journalism. Based on a "closed information systems" perspective, the predominance of foreign elite media as sources in the Greek media #Grexit tweets indicate that, in Greece, a cross-media agenda was established. This finding also points to what Lasorsa et al. (2012) identify as the distinction between elite and non-elite media outlets, arguing that journalists affiliated to more prominent outlets tend to adhere to existing norms and values far more than their colleagues in non-elite outlets. This practice was possibly adopted by Greek media on Twitter as a means to promote their credibility as news outlets, also given the highly polarized political climate in Greece at that time.
Furthermore, Table 1 shows that, in all languages, politicians are ranked in the second place of the most popular sources in the #Grexit debate. On the other hand, political parties are almost absent. This indicates that media outlets considered political parties-in Greece or abroad-as non-significant sources on the Grexit issue. The personalization of the debate seems to remain a value that discriminates the newsworthiness of the news. Furthermore, actors that seem to have played a significant role in sourcing on Grexit are economists and academics. This can be justified on the basis of their status as experts. Their expertise ensured source credibility, which is essential to the journalistic norm of objectivity. On the other side, the absence of activists, citizens, and even politicians that had a differentiated view on the Grexit debate is striking (e.g., Yianis Varoufakis, the vocal and, to many, unconventional Greek minister of Finance at that time, got only 10 references). These findings agree with those of Tracy (2012) in their study of the U.S. media outlets' coverage of the Greek crisis back in 2009, where they observed that "the majority of sources for the stories are financial industry, government, or academic or think-tank economists, at least some of whom share fairly similar beliefs about economic dynamics as they relate to Greece" (p. 519). Interestingly, a study by Borchardt et al. (2018) on how mainstream media covered the Brexit negotiations between September 2017 and March 2018 in eight European countries also provided similar findings. As they point out, "there is relatively little reporting on how ordinary citizens will be affected by Brexit and how it will impact upon their lives. Brexit is an abstract spectacle taking place in Brussels and London, far removed from ordinary citizens' lives." (p. 43).

#Grexit Frame-Building
The approach and steps to identify all the different frames adopted in the #Grexit debate are discussed in the methodology section. The result of this process was the development of the #Grexit coding scheme. Its logic, the main categories, their conceptualization, and characteristic tweet examples are presented and discussed below. We discuss the emphasis on three master frames during the Grexit debate. These include the consequences frame (Semetko and Valkenburg 2000), the strategic frame (Xenos and Foot 2005;Entman 2007), and the episodic frame (Iyengar 1991). The consequence frame is identified as a common frame in the news. It reports an issue in terms of the consequences (especially economic, political, and social) it will have on an institution, region, or country. The strategic frame refers to the rationale and strategy underlying the different actors' rhetoric and positions. The episodic frame shows the restricted coverage of the negotiation process, while a thematic coverage would insist on a more integrated coverage of the debt crisis (see the methodology section for a more analytical presentation of frame building).
First, the media outlets studied only rarely posted #Grexit tweets on the causes of Greece's crisis. Moreover, only a couple of tweets were found to directly refer to the importance of negotiation between Greece and its creditors. Following media outlets' contributions to the #Grexit discussion during the critical months between March and July 2015, two "whys" are largely missing: (i) why Greece is in crisis and (ii) why the negotiation is important? Grexit was somehow a phenomenon with no history apart from this told in the context of the talks between Greece and its creditors and the scenarios on the possible consequences of a Grexit.
Second, only a few tweets referred to arguments that a potential exit of Greece from the Euro was not as important for Europe as other looming crises, particularly a Brexit. Within the #Grexit discussion, media outlets predominantly zoomed in on the evolving story of the talks, the factors that affected this process, and the accompanying scenarios of a Grexit. However, they somehow failed to widen their lens to place into the picture background factors (e.g., Greece's failing economy or eurozone's structural problems) and other emerging threats to the stability of the European Union as a whole, which, less than two years later, did materialize with the Brexit. Overall, our findings suggest that media outlets adopted almost exclusively episodic frames; reporting and commenting upon day-to-day events and circumstances; and effectively ignoring thematic frames that would provide the historic, social, and economic context of the Grexit issue.
From the qualitative analysis of the range of media frames in the #Grexit discussion, we now turn to quantitative analysis to identify which were the prominent ones. As shown in Table 2, across all languages, one of the most prominent or the most prominent frame was this of "Greece's challenges", which is included in the consequences frame. Research on the Brexit negotiations has similarly indicated that media predominantly focused on how Brexit would affect the United Kingdom, and only 15% of the overall coverage was focused on the impact of a potential Brexit on the EU and expressed worries about the future of the EU (Borchardt et al. 2018, p. 27). This finding suggests that mainstream media assumed that the consequences of both Grexit and Brexit would be much more critical for Greece and the United Kingdom, respectively, as compared with the EU.
Comparatively, Greece's positions (strategic frame) received much less attention from media outlets in all languages (even by Greek media). This finding reveals a fundamental imbalance in covering Greece's perspective on the issues and the events of the period. Similarly, one of the key findings of Borchardt et al.'s (2018) study of mainstream media coverage of Brexit was that "there was very little to no support for the UK's view and approach to negotiations" (p. 43). The Greek media, by relying heavily on international media, journalists, and politicians as news sources, effectively managed to frame much of their news coverage on #Grexit from the perspective of the positions of the creditors. The challenges faced by the creditors' side are comparatively downplayed in all languages apart from French, where they receive equal attention to Greece's challenges. In the French-speaking media, it also appears that the negotiation process (episodic frame) has some importance to the Grexit issue. Perhaps they were the only ones to value the talks between Greece and its creditors as a critical negotiation process and not just an irrelevant or minor factor in the Grexit issue.

How Sourcing and Frame-Building Interact
To further explore the relationships between sourcing and framing, a multiple correspondence analysis (MCA) was performed (see Benzécri 1992;Greenacre 1993;Le Roux and Rouanet 2010) with the statistical software R using the FactoMineR package (Lê et al. 2008). This statistical method supports an inductive approach to data analysis to explore the empirical relations between categorical or nominal variables within a low dimensional space.
The MCA was performed on two active variables: the synthetic #Grexit coding scheme comprised of six issue-specific news frames: (a) the negotiation process, (b) Greece's positions, (c) Greece's challenges, (d) creditors' positions, (e) creditors' challenges, and (f) challenges faced by all negotiating parties. The synthetic media sources variable also comprised six categories: (a) media/journalists, (b) politicians, (c) economists, (d) ECB/IMF, (e) academics, and (f) other.
The above active variables contributed directly to the formation of the factorial axes in the MCA analysis. As a supplementary variable, the name of the media outlet was used. Supplementary variables do not contribute to the creation of the factorial axes, but provide useful information for their interpretation.
The factorial model that was adopted as more interpretable had two main axes, which explained 39.5% of the total inertia (the first axis explains 15.5% of the inertia, λ1 = 0.77; the second explains 12.3%, λ2 = 0.61). The variable categories with their v test value on each axis are presented in Table 3. The negative sign of the v test indicates that the category is loading on the negative pole of the axis, and vice versa. The higher a category's absolute value of the v test, the higher its importance in the construction of the factorial axis. The first axis reflects the emphasis put on frames by media and political actors and the second axis for the distribution of power positions between actors taking sides. As shown in Table 3, on the positive pole of Axis 1 are grouped media outlet's #Grexit tweets that predominantly focused on the consequences frame that all sides (individually and collectively) were faced with within the context of their ongoing negotiations to avert an exit of Greece from the Eurozone. The news sources are mainly media/journalists and economists, but also academics and others. This side of the first axis has then to do with one of the main framing processes of news media. On the other side of the axis, we have news on the positions of the negotiating parties, what we had described as strategic frames obtained from sources on the negotiation table, i.e., politicians, ECB, and IMF. Overall, Axis 1 has to do with a bipolar frame-building process: the media-driven and political driven frames. This function, as our analysis showed, was also performed by media outlets' use of Twitter in the context of the #Grexit discussion. In that case, the consequence frame is media-driven, and the strategic frame is politically-driven. This finding coincides with earlier research on the emphasis put by media on the disagreements, and the conflict nature of negotiations (Camaj 2010). On the other side, politicians and experts act as agents of their positions and tactics. Regarding specific media outlets, ABC and Bloomberg are the most characteristic cases of media that used Twitter to post #Grexit tweets that provided inputs for more in-depth analyses of the events and conditions related to a possible Grexit. On the reporting function, the most characteristic examples are three Greek media that mainly used Twitter to report what politicians and other officials of all negotiating parties were saying about their positions, their tactics, and moves on the negotiation table, including assessments (and sometimes blames) of the other side's positions and moves.
On the second axis (see Table 4), we have indices of the imbalanced distribution of power between the different sides of the negotiating parties. The data in Table 4 show that, on the positive pole, we have mainly #Grexit tweets focusing on the creditors' positions and the challenges they face. On the other side, we have Greece's positions and the challenges with which it is faced. Therefore, the second axis has to do with media outlets' effort to offer balanced news about a negotiation side: its positions, but also the complexities of the challenges it is faced with in trying to deal with a possible Grexit. What is noteworthy here is that this balanced approach was preferably applied to just one, not both negotiating sides. Those media outlets focusing on the creditors' side relied mainly on officials from ECB and IMF, on market economists, on academics, and on other sources. The most characteristic cases of media posting news predominantly about the creditors were two Greek media: ANT1, a mainstream national TV channel; and Athens Voice, a free magazine with wide circulation in Athens. Among the media that predominantly posted #Grexit tweets on Greece's positions and challenges to avert Grexit and bargain better terms on the negotiation table were two news agencies (Agence France-Presse and Reuters); a Greek tabloid (To Proto Thema); Guardian, an elite newspaper from the United Kingdom; TF1, the most popular TV channel in France; and China.org.cn, a government-controlled Chinese portal. These media outlets mostly relied on journalists and politicians as news sources. Overall, this axis could be seen as a power position field and a confirmation of the power-bias hypothesis, which suggests that powerful actors get easier access to the media than weak actors. In the negotiation process, the Greek government was seen as a weak actor. This one-sided balance news framing in the context of high-stakes international negotiations reveals an underlying generic framing approach that provides a rounded coverage of the stronger side, while effectively ignoring the weaker side.

Discussion
This exploratory study into media outlets' tweets on the Grexit issue has highlighted the need to better understand online journalistic practices of sourcing and frame-building in the context of the coverage of international negotiations.
The #Grexit debate has revealed a top-down approach in news coverage by media outlets. Grexit has been covered as mainly a macro-economic and top-level policy issue negotiated by a small number of elite actors. This approach seems to reproduce the schema of the unequal distribution of resources between resource-poor and resourceful competing political and social groups and competing frames. As in the case of Brexit, negotiations took shape within a close information system, which included international political and media elites and economy experts that marginalized countervailing voices and critical perspectives (Borchardt et al. 2018). The above fit well to the "closed information systems" or "elite discourse networks" perspectives (Lewis et al. 2005;Davis 2007), which suggest that policy elites (politicians, officials, and journalists) form networks that are relatively shielded from the wider public (Davis 2010, p.110).
In the case of the #Grexit debate, the multiplication of actors and the amplification of voices never materialized in the media narrative on Twitter. The reproduction of some core journalistic norms seems to be the case. Actors such as political leaders; mainstream media; and representatives of the Troika institutions, economists, and academics had been the major sources and key frame-shapers on the #Grexit debate. Activists, non-elite media, and journalists didnot emerge as dynamic agents in reshaping the sourcing and frame-building choices made by mainstream media outlets. In other words, the traditional hierarchy of sourcing that privileges some voices over others appears to persist in high-stakes media stories even in social media. The findings of this study on media's sourcing practices in the #Grexit debate complement those of earlier studies (e.g., Lecheler and Kruikemeier 2016;Nuernbergk 2016;Folker and Nölleke 2019;Bentivegna and Marchetti 2018;Borchardt et al. 2018), which show that news media even on their Twitter news feeds tend to rely heavily on politicians and media as news sources.
As long as news media fail to use the collaborative format of Twitter (and of other social media) to engage citizens and traditionally marginalized voices in the co-creation of news in high-stakes international negotiations, we can safely assume that this is a calculated choice to maintain their privileged role in controlling the flow and form of news. In todays fragmented, fluid, and hybrid networked public sphere, such news media practices practically undermine rather than strengthen their place and value in the public life of democratic societies.
The finding of this study, that thematic and socially driven frames and not only media and political driven frames, that would provide the historic, social, and economic context of the Grexit issue, as well as the multilevel consequences of the austerity bailout program, were largely absent in media's tweets, complement those of earlier research on media's coverage of the economic crisis in southern Europe (e.g., Tracy 2012; Doudaki et al. 2019). Furthermore, the prevalence of the episodic and negative framing, which characterized the crisis as a national and ethical problem rather than a systemic and European one, was also indentified by earlier research (Tracy 2012;Mylonas 2014).
What we can detect in the above sourcing and framing processes is that a power position field on a high stakes negotiation process is formed by media, political, and economic actors with important resources who manage to imply a legitimacy to their interests frame that marginalized the voices of powerless political, economic, and media counter publics, as well as socially driven and thematic frames.

Methodology
Tweets with the hashtag Grexit between 25 March and 13 July 2015 were collected using Twitter's streaming API to execute a standard query using as search operator the term "#Grexit". As pointed out earlier, this was a period during which the negotiations between the newly elected Greek government and its creditors were not making any real progress, and the imposed capital controls as well as the victory of "No" in the July 2015 referendum only heightened the risk of a Grexit. Through Twitter's API, we were able to collect 192,627 tweets containing the hashtag #Grexit, of which 91,806 (47.7%) were original and the remaining were re-tweets. From this corpus of the #Grexit tweets, we singled-out for further analysis only those written in English, French, Italian, and Greek language that were posted by official accounts ofnews media outlets in these languages (N = 1142).
Often, news media have different Twitter accounts for their different news services (e.g., Bloomberg, a global news network, has separate Twitter accounts for its major news divisions, such as "Business", "Markets", "TV", "View", and so on). For example, in our corpus of the English language #Grexit tweets, we were able to identify 88 unique accounts of 45 parent news organizations. We chose to analyze the #Grexit tweets from all the accounts belonging to the same parent media organization, thus avoiding the challenge of having to choose which accounts to exclude. In Table 5, information on the parent media organizations that posted on #Grexit by language during the period under study is summarized. The method followed for the analysis of the tweets was initially qualitative content analysis (Elo and Kyngäs 2008). The aim was to identify the range of information and opinions expressed in #Grexit tweets by the media outlets during the period of negotiations between Greece and its creditors.
The approach to qualitative content analysis adopted was based on grounded theory (Strauss and Corbin 1998). Firstly, all #Grexit tweets were open-coded, based on the most readily understood meaning of the message(s). This process was focused on the identification of the characteristic, value-laden or emotional words used in the tweet such as "tragedy", "fears", "failure", "risk", turmoil", "respect", or "conflict", among others. In parallel, attention was given to the identification of different topics on the Grexit issue and the direction of opinions expressed. On several occasions, coding a tweet in a category required the inspection of other hashtags in the message as well as images, drawings, charts, and even linked texts or videos (De Vreese 2005, p. 54). Gradually, a coding scheme emerged that appeared to be reasonably well suited to help us code all #Grexit tweets in a rather straightforward manner. This was the point where, from a methodological perspective, the content analysis approach became purely quantitative, aiming to count the number of tweets coded under each category of the scheme.
Very often, the tweet as a whole could be coded in a single category. In a few cases, a tweet could be coded in more than one category. For example, the tweet "#Greece: Close with EU. But #Merkel thinks of #Grexit. And stocks Go Down . . . " actually contains three units of analysis. The first is about the position of Greece, the second is about the position of Chancellor Merkel, and the third is about the reactions of the markets. This was the case for only 14 of the coded tweets.
The result of this process was the development of the #Grexit coding scheme. Its logic, the main categories, their conceptualization, and characteristic tweet examples are presented and discussed below. The internal logic of the #Grexit discussion by news media outlets that emerged from the pilot analysis was further elaborated upon in the main analysis phase. In the analytic phase, a total of 38 coding categories were identified and all #Grexit tweets were coded accordingly. At the synthetic level of analysis, these 38 coding categories were grouped into six main categories: (a) challenges faced by Greece, (b) challenges faced by creditors, (c) challenges faced by all negotiating parties, (d) Greece's positions, (e) creditors' positions, and (f) negotiation process.
The above six framing categories were utilized to identify three master frames that have been discussed in the literature on media framing. Based on the above, we were able to identify the following frames.
The consequences frame category included "challenges faced by Greece", "challenges faced by creditors", and "challenges faced by all parties". Tweets offering information and opinions about potential consequences and/or enablers of each party's strategy, capacity, and bargaining power during negotiations were of particular interest. In the above frame category, we discerned multiple frame sub-categories that are presented in detail in Tables 6-8. Table 6. Categories and subcategories of media tweets pointing to challenges faced by Greece.

Categories and Subcategories Tweet Examples
Blame on Greece's tactics Merkel's Bavarian allies say Greeks act like 'clowns' in debt talks . . .
A Grexit will hurt Greece #Greece's GDP to shrink 20% in case of '#Grexit'-SandP Other countries are preparing for a Grexit Exclusive: The secret plan of Italy in the case of Greek drama Critical factors affecting Greece's negotiation power Financial constraints Greece is officially nearly out of cash . . .

Impact of economic crisis on Greek people Greeks run at ATMs #easyJet Founder Delivers #Athens Meals to the Poor
Political antagonism within Greece 2 at night in Greece. #Tsipras still in #Parlamento to listen to opposition #Grexit Protests Greeks protest against #Grexit, as #EuroZone leaders hold crucial summit on #DebtDeal Table 7. Categories and subcategories of media tweets pointing to challenges faced by Greece's creditors.

Categories and Subcategories Tweet Examples
Blame on creditors' tactics IMF: the debt dictator that just needs to cut Greece some slack?

A Grexit will benefit Greece
Greece would be better off outside the Eurozone Table 8. Subcategories of media tweets pointing to challenges faced by all negotiation parties.

Categories and Subcategories Tweet Examples
Blame on all negotiating parties' tactics Never underestimate the ability of politicians to do the wrong thing Blame on past bailout solutions adopted The Greek bailouts are incredibly stupid . . .

A Grexit is undesirable by all
Our 40% #Grexit risk implies that a last-ditch deal will be reached before the end of June Market nervousness "#Grexit scares the markets. European stocks in red . . .
Grexit is not as important an issue as compared to other looming crises "#Brexit Bigger Shock for Germany Than #Grexit Why, in political terms, #Brexit may be a bigger risk than #Grexit The strategic frame included Greece's positions and creditors' positions, and refers to the rationale and strategy underlying the different actors' rhetoric, positions, and tactics. In detail, the "Greece's positions" framing device included #Grexit tweets on Greece's and creditors' positions and, more specifically, on the stated and end goals of negotiation, their intentions, assumptions and interests, their negotiation moves, and their understandings of others' intentions (see Table 9). This frame had four subcategories: stated goals, preferred settlement processes, key negotiation assumptions, and negotiation moves. Table 9. Categories and subcategories of media tweets mentioning Greece's positions.

Categories and Subcategories Tweet Examples
Stated goals #Greece plans to stay in #Eurozone, reignite economic growth-[Greek] minister Preferred negotiation settlement processes @atsipras said the issue had to be settled by leaders . . .

Key negotiation assumptions
"#Greece @yanisvaroufakis #Grexit would be the beginning of the end for the euro. Getting out of the single currency would be catastrophic Negotiation moves #Tsipras: 'Ready to compromise, but not if EU asks for submission' . . .
Regarding the "creditors' position" (see Table 10), the sub-category "key negotiation assumptions" is missing. The reason is that no media outlets' #Grexit tweet was about what creditors took for granted to formulate their negotiation strategy. This is an interesting finding on its own. Perhaps in a negotiation process where there is such a clear imbalance in the negotiating power between the two sides, media outlets do not find it necessary to tweet on the "obvious", i.e., that at the end of the day, the more powerful side will make use of its superior power to coerce, if necessary, the weaker side. Perhaps they also do not want to simply succumb to cynicism. Table 10. Categories and subcategories of media tweets mentioning creditors' positions.

Categories Tweet Examples
Stated goals #Europe must do its best to avert #Grexit-German FM Steinmeier