#GreenRecovery for Europe: A Content Analysis of tweets about the Green Recovery from COVID-19 on Twitter

(1)

#GreenRecovery for Europe: A

Content Analysis of tweets about

the Green Recovery from COVID-19

on Twitter

Sheila Schulze and Yvonne Mrukwa

Word Count:_13,624

MA Media and Communication Studies:

Culture, Collaborative Media, and Creative Industries

One-year Master Thesis_{| 15 credits} Submitted:_{VT 2020 | 2020-10-20} Supervisor:_{Asko Kauppinen} Examiner: _{Alessandro Nani}

(2)

Abstract

The aim of this thesis is to investigate how digital activism is conducted on Twitter, particularly in relation to the dialogues and demands for Europe’s green economic recovery plan from COVID-19. It seeks to analyse the communication made using #GreenRecovery on Twitter by various actors over the period of May to June 2020, guided by the theory of public sphere and social movement and literature on digital activism, hashtag activism, Corporate Social Responsibility (CSR) and Corporate Political Activity (CPA) using a qualitative and quantitative content analysis.By analysing the frequency patterns of tweets and by uncovering the different types of communication, this paper sheds light on the users involved as well as the issue frames and mobilisation strategies that were visible in the #GreenRecovery discourse . Results of this study demonstrate that #GreenRecovery is used by varying actors on Twitter such as individuals, social movements, businesses and others. Furthermore, the hashtag has been used to raise awareness, communicate particular information, mobilize action and also employ assertion as dominant digital spectator activity. Tweets with #GreenRecovery was primarily framed towards the need for a redesign of the economy, indicating demands for changes in policies by targeting accounts of political actors from the EU Commission. It is further implied that during the discourse, #GreenRecovery acted as a structural signifier as a response to the leaked proposal of the Recovery Plan demonstrating that it has the potential to create hashtag communities.

Keywords: _{#GreenRecovery, public sphere, social movements, hashtag activism,}

digital activism, citizen activism, businesses, CSR, CPA, Twitter, Green New Deal, economic recovery plan, COVID-19

(3)

List of Figures and Tables

Figure 1. _{Hierarchy of digital activism by George and Leidner (2019)} 12

Figure 2_{. The general architecture of DMI-TCAT} 29

Figure 3._{Number of tweets gathered in the first subsample} 32

Figure 4._{Inductive, deductive and retroductive codebook development} 33

Figure 5._{Frequency of assertion and metavoicing from 18 May to 21 June 41}

Figure 6. _{Hashtag used in tweets alongside #GreenRecovery} 43

Figure 7._{Word cloud of @mentions in tweets with #GreenRecovery} 44

Figure 8. _{Example 1: tweets using multiple frames} 48

Figure 9. _{Example 2: tweets using multiple frames} 48

Figure 10. _{Tweet with a coalition forming strategy} 50

Figure 11._{Actor tweeting habits during peak days} 76

Figure 12. _{Breakdown of tweet mentions} 77

Figure 13. _{Tweet with a link to an online petition} 78

Table 1._{Final subsample to code} 33

Table 2._{Total count of actors} 45

Table 3._{Types of communication visible through #GreenRecovery} 46

Table 4._{Frames used to communicate #GreenRecovery} 48

Table 5._{Use of resource mobilization tactics in tweets with #GreenRecovery 49}

(6)

1 Introduction

As governments and policy makers all over the world are strategizing their COVID-19 recovery plans to rebuild the economy, businesses, citizens, social movements and other actors have been increasingly communicating, campaigning and demanding for a green recovery plan from COVID-19 that is also considering the wellbeing of society, the environment and the economy.

This thesis will explore how the #GreenRecovery hashtag has been utilized on Twitter to share information, express demands and mobilize action in Europe during the period when the bloc’s Recovery Plan proposed by the European Commision was being reviewed and finalized. Guided by theories of social movement and public sphere as well as literatures on Twitter’s role in facilitating activism, the concept of Corporate Social Responsibility (CSR) and Corporate Political Activity (CPA), it will focus on answering the following research questions:

● What is the frequency pattern of tweets using the #GreenRecovery like over the period of May to June 2020?

● What actors are involved in the #GreenRecovery discourse in Europe on Twitter over the period of May to June 2020 and what are they communicating?

Tweets with the #GreenRecovery that were collected via Twitter using the DMI-TCAT toolset from 18 May to 21 June 2020 will be analysed using qualitative and quantitative content analysis. Each tweet will each act as the main units of analysis. Further, this thesis will also lay out the research process and findings which were followed, understood and analysed using a neo-positivist stand point. Lastly, it will bring forward the findings and discussions while also touching on the ethical and methodological limitations of the method chosen.

(7)

2 Background

2.1 The COVID-19 pandemics’ impacts on economies

The COVID-19 pandemic we are currently faced with has had an impact on our personal lives and our societies’. Over the course of this research, there have been over 34 million confirmed cases of COVID-19 globally, including over 1 million deaths reported to the World Health Organization (WHO, 2020). The pandemic has also resulted in severe economic consequences across the globe. Recent data from a UN agency indicate that the financial impacts unleashed by the outbreak of COVID-19 is hurting the economies of many countries, regardless of their income level. Even if they may not be experiencing a high rate in death tolls due to the coronavirus, numerous countries are currently said to be on the brink of a recession (UNIDO, 2020).

Ozili and Arun (2020) point out how the health crisis transformed into an economic crisis, ultimately bringing the global economy to its knees. Social distancing policies have led to the shutdown of financial markets, corporate offices and events. Consumers, investors and international trade partners reacted to the heightened uncertainty and responded with lowered consumption and investment (Engström et al., 2020). These conditions are threatening the survival of companies who will require immediate state aid to pull through _{(Kraus et al., 2020).} This has led policy makers in government and central banks to face a decision dilemma that Ozili and Arun (2020) describe as a situation where it would be impossible during an outbreak to save both the economy and people who respectively also act as economic agents as it would require them to stay home to curb the spread of the virus. As a result of these stay at home activities, economic activities will decrease significantly which will contribute to an economic downturn (ebd., p.20). This dilemma aside, in response to the unexpected outbreak and by taking the threatening recession into account, policy makers around the world have settled on fast unprecedented social distancing and financial recovery measures

(8)

with a goal to curb the spread of the virus, save lives and also the economy (Blaikie, 2009).

Before the outbreak of COVID-19, the European Union set ambitious targets for reducing carbon emissions which they presented in december 2019 (European Commission, 2019). It “aims to transform the EU into a fair and prosperous society, with a modern, resource-efficient and competitive economy where there are no net emissions of greenhouse gases in 2050 and where economic growth is decoupled from resource use.” (p.2). With the emergence of the COVID-19 crisis, this redesign of the economy seemed to be at stake with many initiatives being delayed as most focus shifted towards ensuring an economic recovery (Abnett, 2020). According to the OECD, although tackling the health crisis and helping affected companies and workers recover are currently the main priorities, it is crucial for post crisis programmes to “align public policies with climate objectives and limit the risk of locking-in carbon-intensive infrastructure” (OECD, 2019, p.1).

2.2 Introducing a Recovery Plan in Europe

As outlined in the previous chapter, the COVID-19 pandemic that is currently ongoing has had a major impact on the global economy and wellbeing of society. Our research shows that since the beginning of the corona crisis, the EU countries have already mobilized a great amount of money to support their economies that were focused on providing emergency aid. As an effort to repair the economic and global impacts caused by the current pandemic and begin a recovery that leads to a more sustainable Europe, the European Commission proposed a recovery plan that aims to make use of the full EU budget for 2021 - 2027 on 27 May 2020, which was later approved by EU leaders on 21 July 2020 (European Commission, 2020c). The plan includes a comprehensive package of €1.85 trillion “which combines the multiannual financial framework (MFF) and an extraordinary recovery effort, Next Generation EU (NGEU)” (European Commission, 2020c).

Alongside their proposal of the major recovery plan on May 27th, the European Commission, as stated on their website (European Commission, 2020a), also

(9)

shared their proposal to create the “Next Generation EU”, a recovery instrument with 750 billion euros embedded in it. This tool is intended to ensure that the recovery is sustainable and that it is one that will benefit both the economy and the climate. The recovery package, with the Green New Deal as the heart of its strategy, will focus on investing in building modernisation, renewable energies and green hydrogen as well as clean mobility and the circular economy. In addition to this, the strategy also aims to strengthen the single market and adapt it to the digital age. Last but not least, it will also strive for a fair and inclusive recovery for all, highlighting unemployment reinsurance schemes, a skills agenda, a digital action plan, fair minimum wages, binding transparency pays and lastly, its efforts in fighting against tax evasions to help increase the revenue of its member states.

On 21 July 2020, after four days of negotiation, EU leaders agreed on a 750 billion euro plan to rebuild the region's economies affected by the pandemic (European Commission, 2020b). With this agreement also came a reconstruction to the fund, which was linked to the EU's upcoming seven-year budget of 1.074 billion euros. The central grant component of the reconstruction has been reduced to 390 billion euros, less than the 500 billion euros recommended by the Commission in May 2020. In addition to this, although it has been agreed that 30% of the total expenditure of the long term budget and recovery fund will be devoted to climate objectives (i.e. to successfully implement the Paris Agreement and also the United Nations Sustainable Development Goals) a significant reduction in spendings for critical climate and environmental incentives also occurred as a result of this reconstruction (IISD, 2020). Many agree that despite being a historic step in budget negotiation, disbursements in the budget for climate focused programs must still be developed.

2.3 Negotiating a Recovery Plan on Twitter

Many scholars believe that social media has been influential in mitigating the COVID-19 crisis. According to _{Fenwick et al. (2020), social media and Twitter} particularly, have played a crucial role in “triggering a more effective policy

(10)

response based around social distancing, lockdown, and containment” (p.1) by destabilizing certain narratives that were not backed up by health professionals. Dixit and Panday (2020) believe that the live and real time engagement facilitated by social media has increased the possibility of crowdsourcing content and also outsourcing fact checking and verification needs to the audience.

Aside from being flooded by posts and debates on the COVID-19 pandemic, there had also been an increase in posts linked to economic recovery proposals and plans. To ensure that the economic recovery plan from COVID-19 is still aligned with the goals outlined in the European Green New Deal, Pascal Canfin, French Member of the EU Parliament (MEP), who is also chair of the EU Parliament’s environment and public health committee, initiated the launch of the “Green Recovery Alliance'' in April 2020. The alliance was formed in response to the appeal signed by 12 EU environmental ministers for a green recovery from COVID-19. It brings together 79 MEPs, civil society groups including numerous CEOs and businesses associations, the European Trade Union Confederation, NGOs, think tanks and executives of large companies such as Ikea, H&M, Lego and Microsoft (Simon, 2020). This alliance has led to the popularization of the hashtag #greenrecovery as the announcement was widely spread on Twitter. The hashtag has since been utilized by Twitter users of various backgrounds, from political leaders, citizens, SMOs and also corporations. According to Splash (2020), economic crises such as the COVID-19 pandemic divulge the existence and organization of dominant political economies and opens up the possibility for its reconstruction.

During this debate, particularly in the months of May throughout July 2020, ahead of the package’s approval, approximately 1.2 million citizens and 100 environmental NGOs have signed a petition calling on a green and just recovery plan from COVID-19 during the times where the budget talks were taking place in Brussels. This petition was also widely spread on Twitter (We Move Europe, 2020). Despite being approved, this discourse is still ongoing, with many activists arguing

(11)

that the approved plan does not seem to consider climate change as an emergency, as only 30% of the package is said to be targeted towards climate policies (Taylor, 2020).

3 Literature Review

3.1 Twitter: Facilitating Activism, Crisis and Political Communication Twitter has become a common tool to facilitate communication and debates on societal and political issues. As seen in the previous chapter, people access and share sources of knowledge regarding COVID-19 with increased speed of information exchange.

Aside from this, Twitter is also often used by social movements. The ease of spreading and sharing information via the platform also helps social movements mobilize and recruit new participants to engage in political communication, by eliminating the need for elite support (Miller & Kendall, 2018; Shirky, 2011; Tsatsou, 2018). Twitter has been coined as a platform that is currently “reshaping politics” (Small, 2011). Twitter, like many other social networking sites, have become an avenue and “catalyst tool in the shape of social movements” (Lopes, 2014, p.4). According to Tufekci & Wilson (2012), since the “Arab Spring” burst forth in uprisings in Tunisia and in Egypt in early 2011, social media’s contribution to political change in authoritarian regimes have gained more attention. A survey among participating demonstrators showed that those who used blogs and Twitter for both general information and for communicating about the protests were more likely to attend the protests (Tufekci & Wilson, 2012).

However, while scholars like Papacharissi (2002) also acknowledges the potential of the data storage and retrieval capabilities of internet-based technologies to “infuse” political discussion with information otherwise unavailable (p.9), it has to be noted that, inequalities in information access and the fragmentation of the

(12)

political discourse leaves it open if “a new public space for politically oriented conversation” is emerging (ebd.).

3.1.1 Digital activism

The shifts in social, political and economic trends and practices combined with the rise of social media has changed the way citizen activism is being facilitated and conducted. According to George & Leidner (2019), “Digital activism provides new opportunities for social movement participants and social movement organizations (SMOs)” to communicate and mobilize action as it is considered to reach more people, have a stronger and more immediate impact in comparison to the more traditional forms of activism. Based on the exploratory literature review conducted by the scholars, a framework for digital activism was developed which they have adapted from Lester Milbach’s 1965 layers of activism, which are spectator, transitional, and gladiatorial activities. The former consists of activities with the largest volumes of action, whereas the latter may require extra resources to employ but is guaranteed to have a greater impact.

Actions that fall under digital spectator activities, according to George and Leidner (2019), are clicktivism, metavoicing and assertion. _{Clicktivism, also commonly} referred to as _{slacktivism, represents the endorsements made towards a particular} post that can be done either by “liking,” upvoting, or “following” an activist social media post or blog. Such actions signal endorsements of an existing post. Metavoicing are actions that support the ideas, information and values by reacting to posts made by others, this can be done via the sharing, retweeting, reposting, and/or commenting function on Twitter. Lastly, _{assertion is an activity that can be} performed by anyone with an internet connection, as it describes the social media content creation itself, using video, audio, image, text or media.

Digital transitional activities on the other hand, as stated by George and Leidner (2019) utilizes actions such as political consumerism, digital petitions, botivism and e-funding. Political consumerism involves buycotting, where political

(13)

consumers support their views by financially supporting businesses agreeing with their views and boycott firms that are against it. Digital petitions adopt the concept of online government petitions, whereas botivism consists of actions made by botivists, also known as virtual activists which combines the actions of automated digital actions or digital robots. Lastly there is e-funding, which are actions that utilize technology to create revenue to support a particular cause.

Figure 1. Hierarchy of digital activism by George and Leidner (2019, 7)

It is important to note that the framework developed by George and Leidner (2019) considers both individuals and organizations. It also acknowledges both the senders (e.g. participants and SMOs) and also the receivers (e.g. individuals and targeted organizations). They further highlight major impacts in each tier of activities. Spectator activism activities, according to their study, creates amplification which arouses the emotions of participants. Whereas transitional activities typically result in financial, emotional and cognitive outcomes.

(14)

3.1.2 Rethinking slacktivism

As described by George and Leidner (2019), slacktivism on Twitter involves the “liking,” upvoting, or “following” of an activist account or cause. Despite the negative connotations that have associated the activity with lazy activism, scholars have argued that it may be time to rethink its capabilities in bringing issues to the eyes of the public (Barberá et al., 2015; Freelon et al.; 2020). Slacktivism, also commonly known as clicktivism, armchair activism or feel good activism, is often explained as a the “derogatory epithet used for these activities is ‘slacktivism’, which refers to political activities that have no impact on real–life political outcomes, but only serve to increase the feel–good factor of the participants” (Mozorov, 2009 as cited in Christensen, 2011). It can also be closely tied to the concept of small acts of engagement (SAOE) developed by Picone et al. (2019), which is considered as productive audience practice that can be made with little investment in comparison to others. On social media, these activities can be linked to the profile picture modifications using cause related images (e.g. pink ribbons for breast cancer awareness or a plain image with #JeSuisCharlie), to the viral ALS Ice Bucket Challenge videos (Noland, 2019). Barberá et al. (2015) in their study of analysing three protests in different languages and political contexts on Twitter, provide evidence that slacktivists “are potentially very important as a collective” (p.11) and how their actions have played a peripheral role in increasing the spread of protest related messages. They further argue that slacktivism holds potential to transform a protest action into a social movement and also extend its life cycle.

3.1.3 Hashtag activism

The act of circulating and sharing information on Twitter can take different forms. The platform enables users to generate content themselves and also interact with posts created by others (Moscato, 2016), by either retweeting, liking and commenting on Twitter posts or through the use of hashtags. As an emerging political platform, Twitter’s hashtag feature allows users to view the cacophony of conversations online to identify personally relevant topics and conversations, enabling access to and participation in discussions on social and political issues

(15)

through the form of "hashtag activism" (Xiong et al., 2019), an activity often associated with slacktivism. When speaking of its core features and functions, a hashtag consists of keywords written with the “#” symbol that is often used to index themes or topics on Twitter, allowing people to easily follow topics they are interested in. A keyword is attached to the # symbol so as “to mark a tweet as being relevant to a specific topic and make it more easily discoverable to other users” (Bruns and Moe 2014, p.17, as cited in Pond, 2016).

Studies show that hashtag movements have proven to be a popular approach to bring about several positive socio-political changes around the world (Goswami, 2018). Pond and Lewis (2019) explain that it acts as “genre defining discourses through which action frames may be contested and negotiated” (p. 220). Clark (2016) argues that a hashtag’s narrative logic—its ability to produce and connect individual stories—fuels its political growth, making it a “a tool for building collective identities that serve as the foundations for action” (791). In the context of activism, hashtags also function as an agenda setting tool that enables users to stir dialogue and also divert the public’s attention on certain issues. This tool has also been described by scholars to be a manifestation of participatory culture (Ciszek, 2013) that also enables the transmission of collectively shared personal action frames through the formation of hashtag networks (Bennett & Segerberg, 2012).

3.1.4 Participatory Culture

The concept of participatory cultures first emerged in Henry Jenkins’ earlier works on fandom. This theory speaks about how the advancements in technology we see today have paved the way for the general public to shift from being only consumers, to prosumers of information and knowledge (Jenkins, 2006). He believes that as prosumers, the public previously acting as consumers, now hold power to take the media in their own hands by having an active role in the production, reproduction and distribution of knowledge and information. According to Jenkins (2006), this form of culture allows members to create and share their creations by shifting “the focus of literacy from individual expression to community

(16)

involvement” (p.6), by representing a strong sense of social connection, promoting civic imagination that contributes to sustaining and developing shared civic norms (Jenkins, 2019).

Carpentier (2016) further adds to the explanation from a sociological perspective, where he describes participation as the act of partaking in social processes that “includes many (if not all) types of human interactions in combination with texts and technologies” (p.71), with power acting as a secondary concept that supports such activities. He describes this by pointing to Melucci’s (1989) twofold definition of participation, which is done to advance the individual interest of the actor but at the same time also promoting the collective needs of a community. In their debate about participation and politics, Jenkins and Carpentier (2013) touched on how online activism considers the challenges of building participatory structures and at the same time holds potential to foster participation in their practices. It was further discussed that participation in such practices takes advantage of civil reservoirs (i.e. the citizens’ knowledge and practices) and in turn activates and validates these citizens and their voices, which consequently may “result in more societal happiness and is seen as a better guarantee of good decision making” (p.281).

3.2 Corporate Social Responsibility

Activists and nongovernmental organizations (NGOs) increasingly target companies when concerned about business practises with corporate campaigns to change their business practises. Their weapons of choice range from consumer boycotts to shareholder activism and divestment campaigns (Abito et al., 2019). Through social media platforms such as Twitter, Facebook and blogs, activists can not only capture a broad audience but also quickly disseminate their messages to mainstream media outlets to attack their targets’ reputation (King, 2016, p.230). Companies have started to take these challenges seriously by spending significant resources on corporate social responsibility (CSR) programs and reports, as well

(17)

as reputation and issue management initiatives, and adopting self-regulatory standards at both the firm and the industry level (Abito et al., 2019).

There are diverse definitions of the term Corporate Social Responsibility (CSR). Kolk (2016) proposes a possible distinction in (a) activities by a company to advance a social cause beyond compliance, and (b) management practises that are economically profitable, law abiding, ethical and socially supportive. The author reviewed CSR research published in the leading journals since 1965 and identified the following themes that companies often engage in, such as the (green) environment, ethics, rights, and responsibilities, poverty and (sustainable) development. Scholars argue that there is an increasing influence of businesses on individuals under the notions of CSR across nations as “many business firms have started to assume social and political responsibilities that go beyond legal requirements and voluntarily contribute to regulation making in global governance” (Scherer & Palazzo, 2011, p.1). This leads to questioning the constitution of a corporation as a facilitator for social responsibility. The term Corporate Citizenship (CC) imagines a corporation as a citizen of the state where it operates, whereas other approaches take the different stakeholder groups as potential citizens of the corporation, held to be an analog of the state (Valor, 2005).

The concept of CSR is closely linked to the concept of Corporate Political Activism (CPA). CPA’s roots lie in the concept of CSR, “moves beyond dialogic theory’s emphasis on achieving consensus, but to focus on an organization’s values and how those values are reflected . . . about often controversial social and political issues” (Wilcox, 2019, p. 3). This could take the form of CEO activism that refers to the actions where corporate leaders speak out on social and environmental policy issues that are not directly related to their company’s core business, which distinguishes it from the traditional corporate social responsibility (Chatterji & Toffel, 2019).

(18)

Aside from activists and social movements, similarly, studies show that many companies also use Twitter as a conversation tool. Based on a study conducted by Rybalko and Seltzer (2010) as well as Edman (2010), companies often engage customers by applying dialogic principles on Twitter to cultivate healthy relationships. Within the corporate Twitter communication research, CSR is a newly emerging theme. Gomez’ (2011) pilot study discovered that consultants and strategists lead the CSR discussion on Twitter, followed by general stakeholders. It was expected that companies had a major presence in Twitter, but only a minority of enterprises were twittering CSR messages. This can be related to findings from Gaither & Austin (2016) suggest that CSR initiatives that are highly congruent with a company’s products or services, but to which corporate products or manufacturing processes contribute negatively, are likely to be criticized more heavily and elicit more public skepticism. Generally, they state a company advocates on behalf of an issue of societal concern—one in which the company has expertise due to its products or manufacturing processes—while also demonstrating shared values with its customers.

4 Analytical Framework

This research project analyses the empirical material using several theoretical lenses. In order to comprehend the roles and demands made by citizens, SMOs and companies in relation to this discourse, it explores the theories of the public sphere, social movements, audiences and participatory culture.

4.1 #GreenRecovery as a public sphere on Twitter

Public sphere was defined by sociologist Jürgen Habermas in his English version of the book titled _{The Structural Transformation of the Public Sphere} _(Habermas, 1989), as a process where society engages in a public debate to address a societal problem and in turn influence political action. He conceptualises an informal public sphere as debates supported by civil rights and citizenship, the formal public sphere is represented by politically driven justifications and decision making.

(19)

Habermas’s theory (1989) was a critique on the mass media that emerged in the 20th century. He argues that instead of creating a space for genuine public debate, the media, which is often confined by institutional control (1987), in reality creates what he described as a “pseudo-public sphere” which is defined as a state where the mass media forms a society where it is difficult to form a critical public opinion that opposes the existing elites (1989). His work emphasized that democratic decision making should consider the opinions and concerns of citizens that are voiced either through public forums, informal associations or social movements and this public sphere should not be overthrown by the power of the elites or mass media. His critics, such as Fraser (1990) and Ryan (1992), have highlighted how the presupposed access, discussions and debates that existed at the time did not consider women or other marginalized communities who began to obtain political power and shared their interests publicly during the 19th and into the 20th century (Fraser, 1990; Ryan, 1992). Kluge et al. (1993) further add to this discussion stating that instead of focusing solely on formal deliberation, the public sphere should also facilitate the sharing of shared experiences.

O’Hallarn et al. (2018) argue that the unique architecture of Twitter can promote public sphere-like discourse in hashtags. Bruns & Highfield (2015) claim an extension of the conventional public sphere concept and note that “everyday social interaction between peers and public participation in issue public overlap and are often inextricable intertwined, as users move seamlessly between interpersonal and public topics and registers of expression from . . . one tweet to another” (p.61). Taking corporate communication into account, although the practice of CSR aims to create consensus, it can also create dissensus between corporations and civil society as it “enables corporations and more radical civil society actors to confront each other over issues of fundamental commercial importance” (Whelan, 2013, p.766). Based on this distinction, Whelan highlights that “public spheres can also be arranged to highlight existing or potential points of disagreement” (2013, p.761).

(20)

We consider Habermas’s theory of the public sphere as an important and useful theoretical lens to apply to our study because we are intrigued to explore if Twitter holds a democratic potential in negotiating the Green Recovery in Europe and if The #GreenRecovery holds a public-sphere like discourse and if the hashtag equally facilitates formal and informal public spheres.

4.2 #GreenRecovery as a Social Movement

In this section, we aim to draw a line between the theory of the public sphere to the theory of social movements. Social movements can be defined as “network structures” with a complex aggregation of different organizations, groups and individuals that vary in size, goals, structure and complexity tied by their strong (environmental) ethos (Johnston, 2014). They can be characterized as movements that are formed as a result of shared beliefs and identities, with informal and dense network structures (Diani & Bison, 2004).

According to Zald (2000) social movements are typically formed because of shared grievances about how society should be including the ideas on the ideal role and practices of businesses. Georgallis & Lee (2019) argue that social movements induce corporations to engage in CSR activities and encourage them to enter moral markets, where they create social value by ensuring that their products and services become more superior to the alternatives (p.51). These “moral markets are typically supported by organized actors (usually social movements) motivated by moral or normative considerations rather than only by the pursuit of economic interest” (p.51). Whereas Walker (2015) argues that if threatened by a crisis, corporations often adopt strategies of “insurgent actors” (p.3, as cited in Mcdonnell, 2015), in other words employ social movement strategies of their own.

4.2.1 Framing

Social movements generalize their grievances and enunciate their demands through the concept of framing. Framing allows “people to perceive, understand, and label events and occurrences, organizing them and giving them meaning”

(21)

(Goffman, 1974). As a process that is closely linked to social constructionism, this technique is often used by the media, citizens, social movements, political players as well as other actors to evoke attention on a particular issue (Klandermans & Stekelenburg, 2009).

Emphasis frames focus on “emphasizing different aspects or attributes of an issue” in order to increase the “attributes’ salience in people’s mind, prompting people to use them as the basis to evaluate the issue” (Luong et al., 2019). Luong, Garrett and Slater (2019) indicate that if such frames relate to the existing beliefs or ideologies of the audiences then it will be more persuasive and effective. Some examples of emphasis frames are the use of environmental risks, public health or security frames when communicating a campaign or message supporting climate change mitigation (ebd.).

4.2.2 The life cycle of social movements

Despite their different types, social movements typically experience the same life cycle and go through similar stages. As stated by Blumer (1969), Mauss (1975), Tilly, (1978) in the first stage, social movements _{emerge because of their interest} to solve or change a particular problem. The second stage is _{coalescence, where} movement leaders recruit and mobilize new members or resources and develop strategies to achieve their goals (e.g. collaborate with the media for good publicity or join forces). Third, they go through a stage of _{bureaucratization, where} bureaucracy (and organizational structures) within the movement starts to rise. Volunteers may be replaced by paid employees, and focus may shift towards fundraising. Lastly, they experience a _{decline stage, which could be as a result of} successful campaigns or due to failure caused by the lack of funds and resources.

4.2.3 Collective action

Collective action explains the efforts that are taken collaboratively by social groups and movements to diverge from the social norms of a particular event, directing action to a particular cause to bring about social change. According to Olson (1965), individuals in large organizations are more likely to act (or can be

(22)

incentivized to act) based on a common interest. Smaller organizations, on the other hand, may encounter the “free-rider problem” which he defines as a case where individual members of the organization participate in little to no action in comparison to others but still receive its benefits.

4.2.4 Resource mobilization strategies

In order to comprehend how social movements may succeed as agents of social and political change it is crucial to understand how they function and which tools and tactics they often deploy. Social movements utilize unique repertoires of contention and action, also known as protest-related tools, in the form of marches, speeches, meetings, petitions, strikes, demonstrations, forming associations and coalitions, and many more to coordinate, organize and mobilize action (Snow et al., 2008; Tarrow, 2011; Johnston, 2014). Repertoires of action can take many other forms and depend greatly on the knowledge of social movement members on their features and capabilities (Tarrow, 2011).

As hashtag activism is closely tied closely to social movements, we want to investigate the lifecycle that #GreenRecovery deploys on Twitter. Furthermore, the theory of social movement inspired us to look at the frames and mobilisation resources that are being used and how collective actions and expressions are being used in tweets with #GreenRecovery.

4.3 #GreenRecovery changing the roles of audiences and facilitating participatory culture

In the previous sections, we have discussed how the theory of the public sphere explains the influence that public opinion has on politics and how it represents the ideal of a participative democracy. Further, we have discussed how social movements may act as agents to bring about political and social change. In this section, we will touch upon the shifting roles of audiences as receptors of communication.

(23)

Stuart Hall’s (1973) encoding and decoding model can be used as a foundation to understand the shifting roles of audiences or receivers of communication. Hall’s encoding and decoding model refused the linear (one-way) model of communication by suggesting that both the sender and receiver play an important role in communication and that the message being communicated is sent, received and understood as a result thereof. Encoding is explained as the process where the sender sends a message to the receiver and incorporates cues in order for the message to be understood by the receiver. The decoding of a message, as explained in Hall’s model, is the process where the receiver tries to understand the message by adding meaning to it. The process of decoding can be influenced by several factors that may distort the meaning of the message itself and thus affect the way it is being circulated (Hall, 1973). Similarly, Hill (2017) and Picone et al. (2019) have also pointed out that many factors influence how communication is experienced and how engagements can take place. These factors consider the mood of recipients as well as the time and space of when and where the communication takes place. According to Hill (2017) “there is the creative labour that helps to co-create our expectations of a live event, including the management of liveness as spaces of interaction, and the affirmation of a shared experience” (p.12).

Hall’s encoding and decoding model, audience and reception studies as well as the concept of participatory culture has sparked interests and change in the narrative to consider the increased role of audiences in media and message production. Within political discourses, participatory culture is seen to amplify the freedom of expression of individuals and allows them to share their voice to democratically influence change (_{Gambarato & Medvedev, 2015). Participatory culture can} manifest in various forms. According to Jenkins (2009), it can be in the form of affiliations_{, which is the interest-driven component or expressions, which is the} production component of participatory cultures. It can also take the form of collective problem solving and _{circulations. The former is the knowledge-building}

(24)

component of participatory cultures whereas the latter explains the networks where information is spread.

As means to situate Hall’s encoding and decoding model in interactive media technologies and the digital landscape, Bødker (2016) and Shaw (2017) stress the importance to understand the affordances of the medium itself and how it may become part of the decoding process by either encouraging or discouraging interaction. Placing their views in the context of Twitter, it can be argued that the affordances of the platform itself can affect the way a certain tweet is being decoded and circulated by users online. It can further be argued that these affordances can facilitate participatory culture and promote democratic communication, by enabling participants to take part in shaping and disseminating information.

5 Research Questions

According to Blaikie (2009) all research projects are constructed based on the foundation of specific research questions. He further states that “getting these questions clear and precise requires considerable thought and also comes with preliminary investigation” (p.57) Based on the literature studied and theories reviewed to guide this research, we have obtained the understanding that Twitter, as a platform, is able to encourage the development and distribution of communication and engagement amongst its users. It has also become an avenue that facilitates activism, social movement activities and resource mobilization tactics and CSR practices, and further, promotes attention towards a particular cause. In light of our findings we have formulated the following research problem to be addressed:

Analysing the Green Recovery discourse in Europe over the period of May to June 2020 on Twitter: what is being communicated the most using the #GreenRecovery hashtag, when and what actors are involved.

(25)

As we are interested in discovering the patterns and characteristics of the #GreenRecovery discourse, therefore to address the above research problem, we have formulated several main ‘What’ research questions followed by a set of subsidiary questions. The primary research questions will opt to address the main problem whereas the subsidiary ones will aim to answer and provide background information that support the primary research question to identify and highlight additional findings that we hope can add more context and depth to our research (Blaikie & Priest, 2019, p.92; Blaikie, 2009, p. 66). We have also focused on formulating topical questions, as suggested by Layder (2013) to narrow down our research and obtain answers to the specific issue and problem we would like to address (Layder, 2013, p.10-12). These research questions were then later used to determine which literature should be reviewed.

RQ 1: _{What is the frequency pattern of tweets using the #GreenRecovery like over} the period of May to June 2020?

Subsidiary Research Questions:

a. What digital spectator activities have been used the most in tweets with #GreenRecovery from May to June 2020?

b. What hashtags are being utilized alongside #GreenRecovery in tweets posted over the period of May to June 2020?

c. What actors are being targeted (or mentioned) the most in tweets using #GreenRecovery over the period of May to June 2020?

RQ 2: What actors are involved in the #GreenRecovery discourse in Europe on Twitter over the period of May to June 2020 and what are they communicating?

(26)

Subsidiary Research Questions:

a. What actors are sending tweets with #GreenRecovery?

b. What are the different types of communication visible in tweets using #GreenRecovery?

c. What issue frames are emphasized in tweets using #GreenRecovery? d. If mobilization resources are being used by actors in tweets with

#GreenRecovery, what are they?

6 Methodology

6.1 Choice of Method: Content Analysis

In order to study the green recovery discourse in an online setting, particularly by following the #GreenRecovery hashtag on Twitter over the chosen period, a quantitative content analysis was deemed to be the most suited research method. Content analysis has been termed as one of the most important research methods in social science research that focuses on analysing large chunks of textual data in order to identify its symbolic qualities, meaning and consequences in a given context (Krippendorff, 1989). It is described as a process following a “step-by-step procedure used to answer research questions” (Frey et al., 2000, p.239) that uses what Stemler defines as “a systematic, replicable technique for compressing many words of text into fewer content categories based on explicit rules of coding” (2000, p, 1).

The reason for choosing this methodology as our research strategy for this thesis is twofold. Firstly, due to its non-obtrusive and non-reactive nature (Riffe et al., 2019). Many researchers have also used this method to “to study the characteristics of communication content, but also to draw inferences about the nature of the communicator” (Berger, 2000, p.126). In order to gain insights about

(27)

how Twitter users communicate about the demands for a Green Recovery, this method allows to “draw conclusions from content evidence without having to gain access to communicators who may be unwilling or unable to be examined directly” (ebd.). Secondly, the method’s deductive nature makes it possible to test and analyse our findings using the theoretical lenses that have been chosen (Berg, 2001).

Twitter has become a popular platform to collect data for academic and industry needs. Prior to deciding on this methodology, we documented best practices from existing studies that have deployed Content Analysis while working with Big Data from Twitter. Kim et al., (2013) stated, when analysing Tweets, it is important to acknowledge that this process may typically be hindered by the use of slang, sarcasm as well as other unconventional forms of text based expressions such as hashtags or emoticons. For this reason, the authors further state that the process of manually coding tweets accompanied by high inter reliability checks would be ideal. To ensure that we have followed the crucial processes of content analysis in our research approach, we have adopted the the following procedure as proposed by Frey et al (2000) in chronological order throughout our research process: 1) collecting a sample of texts, 2) selecting the categories to be coded, 3) creating variables, 4) conducting a coding training, and 5) data analysis.

This research uses a mixed method approach in content analysis. We are working with quantitative data that has enabled us to analyse and measure the frequencies, descriptive and inferential data of how #GreenRecovery has been used via Twitter over the given time period. We also conducted a qualitative method of manual coding using a codebook we developed inductively and deductively to analyse the tweets that were tweeted with #GreenRecovery to derive quantitative results and identify patterns of communication.

(28)

6.2 Research Approach, Paradigm and Ontology

Research paradigms are philosophical ways of thinking that tells us “how meaning will be constructed from the data we shall gather, based on our individual experiences” (Kivunja & Kuyini, 2017, p.26). We believe the neo positive approach’s deductive model of explanation provides the right paradigm for this research method. Neo-positivism stems from positivism, which, according to Collins (2010) is a philosophical standpoint that is “in accordance with the empiricist view that knowledge stems from human experience” (p.38), with an “ontological view of the world as comprising discrete, observable elements and events that interact in an observable, determined and regular manner” (p.38). Different from interpretivism, the goal of positivists is to be able to generalise and unearth the truth behind a given context (ebd.).

From positivism emerged a philosophy called neo-positivism or logical positivism that is commonly used to conduct analytical research. According to Blaikie and Priest (2017), the sole difference between positivism and neo-positivism is that the latter “accepts some of the ontological assumptions of the standard view of positivism but rejects its epistemology” (p. 60). With neo-positivism, the authors argue, knowledge may only be produced through a test of trial and error, where theories are tested against observation and findings. By utilizing this paradigm to guide our research via content analysis we are also supporting the methodology’s unobtrusive nature, by strictly observing and not intruding or affecting the outcomes of the research in any way (Berger, 2000).

The reason for having chosen this philosophical approach to guide our research is because it is known to be centered on understanding human behaviour. It follows a research process that is “usually backed up with facts and evidence which will eventually be confirmed through observation and review of action research.” (Ayeni & Kasimu, 2019, p.22). By using the neo positivist approach to this research, “the first task is to establish patterns or regularities, not as the basis for an explanation but what it is that requires an explanation” (Blaikie & Priest, 2017, p.64).

(29)

6.3 Strategy for Data Collection and Analysis

6.3.1 Data collection with DMI-TCAT

We used the Digital Methods Initiative Twitter Capture and Analysis Toolset (DMI-TCAT), which is “an open-source, freely available data capture, and analysis platform for the Twitter micro blogging service” (Borra & Rieder, 2014, p.263) . The tool was developed by researchers of the University of Amsterdam and resulted from continuous interaction with students of a Masters class on digital methods.

We chose this tool because, as Borra and Rieder (2014) stated, DMI-TCAT emphasizes epistemic plurality by “allowing for a number of different sampling techniques, by enabling a variety of analytical approaches or paradigms, and by facilitating work at the micro, meso, and macro levels” (p.266). The tool provides easy data capture and analysis supporting researchers without a strong background in data mining. Further it allows easy import and export of data, integration with existing analytics softwares, and guarantees methodological transparency by publishing the source code.

In order for DMI-TCAT to access the Twitter API, we had to apply for a developer account (Twitter Developers, 2020). There we stated our academic interest as a primary reason for using Twitter developer tools and filling in data about our educational institution. Afterwards we had to state our intended use of the Twitter API, where we briefly described our research design. After the request has been reviewed, we received our Twitter credentials per e-mail. DMI-TCAT runs on Ubuntu 18.04, a Linux operating system that has to be installed.

(30)

Figure 2_{. The general architecture of DMI-TCAT (Borra & Rieder, 2014, p.266)}

DMI-TCAT relies on Twitter’s API to capture tweets in real-time. “Streaming API relies upon a continuously open network connection between Twitter and the receiving host and is designed to support significant volumes of data transfer” (Black et al., 2012, p.232). This makes three different sampling techniques via DMI-TCAT possible. First, researchers can capture a “1 percent” random sample of all tweets passing through Twitter. Secondly it can track tweets containing specific keywords in real-time. Third, it allows for following tweets from a specified set of up to 5,000 users (Borra & Rieder, 2014). The REST API follows “a typical client-server request and response communication pattern where connections between Twitter and the requesting host are dynamically created on a per-request basis” (Black et al., 2012, p.232). This allows a collection of tweets in a so-called “query bin” that can consist of multiple keywords, hashtags and specific phrases. Furthermore, the REST API allows DMI-TCAT to retrieve the last 3,200 tweets for each user in a set, providing a level of historicity for user samples, and the retrieval of follower/followee networks (Borra & Rieder, 2014). Data enrichment of third party shortening services such as bit.ly is provided by a script that “follows all URLs to their endpoint, adds the location to the URL table, and extracts the domain name” (ebd., p.269), ultimately allowing for URL expansion.

(31)

After selecting a data set, as defined by a query bin, different techniques to filter the data set are available. By sub-selecting from the data set, a user can zoom in on a specific time period or on tweets matching certain criteria. She can choose to include only those tweets matching a particular phrase such as a word, hashtag, or mention; she can exclude tweets matching a specific phrase; finally, she can focus on tweets by particular users or tweets mentioning a specific (part of a) URL. All input fields accept multiple phrases or keywords to specify (AND) or expand (OR) the selection via Boolean queries (ebd., p.270).

To analyse the data, the interface of DMI-TCAT allows the download of tweet statistics and activity metrics, tweet exports, networks and other experimental modules in standard tabulated formats (csv.) or as network files. In our analysis we have used the tweet statistics and activity metrics which indicate the number of tweets, the number of tweets with URLs, hashtags, and mentions, as well as the number of retweets and the number of unique users in the selection. Also, tweet exports provided us with a list of actual tweets to further conduct our data analysis as outlined in the following chapter.

6.3.2 Subsampling and Codebook development

This section will lay out the step by step process we followed to develop the framework that guided our data recording process. In developing our codebook we drew inspiration from the methodology and codebook development of Pond (2016, p.12-14) as well as Wheatley and Vatnoey (2019, p.12-14). To ensure an organized and solid structure to the process, we echoed Whetly and Vatnoey’s process of using inductive, retroductive and deductive approaches in the coding process (2020, p.13), with a sequence that we have updated to follow our dominating deductive logic of inquiry which we primarily conducted to identify the set of coding variables.

(32)

We started the data collection on the 18th May 2020 as the launch of the “Green Recovery Alliance'' in April 2020 sparked our interest for conducting a content analysis on the discourse of the Green Recovery for the European Union on Twitter as outlined in chapter 2. By the time of starting the data collection on 18th May 2020, we have been able to technically set up DMI-TCAT and have run a data collection test over one week with another hashtag. The data collection process of #GreenRecovery was run without breaks and was continuously monitored, pulling tweets with #GreenRecovery from a query bin using DMI-TCAT. On 21st of June 2020, a total of 18.057 tweets showed sufficient data capture to perform quantitative as well as qualitative analysis over a time period when the discourse was politically relevant (step 1). By sub-selecting from the data set, we chose to include only those tweets in a first subsample including the expanded keywords “EU” (OR) “European” (OR) “Europe” to analyse the tweets that address the European Recovery Plan leading to a collection of 7.948 tweets (step 2). These tweets were used to conduct a quantitative analysis to answer RQ1, in which we will discuss further in our findings section in Chapter 8.

By withdrawing the tweet statistics and activity metrics, we compared the total number of tweets from the first subsample of step 2 (n=7.948) to the total data collection of step 1 (n=18.057). It shows a similar fluctuation of frequency in engagement was visible. Figure 3. below shows the constant engagement with #GreenRecovery and the EU-related keywords during the selected time frame. Other noticeable peaks occurred on the 22nd of May, 25th of May, 27th of May, 3rd of June and 18th of June.

(33)

Figure 3._{Number of tweets gathered in the first subsample and total data collection}

In order to develop a codebook, we have pulled a randomised sample of 1,000 tweets from this collection using DMI-TCAT (step 3) and removed all non-English tweets, leaving us with 898 tweets (step 4). Further reductions were made until we reached the final sample selection of 601 tweets to code (step 5) because we were interested in identifying the dominant message frames during the high engagements with the #GreenRecovery.

(34)

Table 1. _{Final subsample to code}

Figure 4._{Inductive, deductive and retroductive codebook development}

Setting variables from theories and and literature (Deductive)

Figure 4. above shows that based on our preliminary high-level analysis of randomized tweets in English (n=898), we adopted the deductive logic of inquiry and reverted back to our theories to determine the variables for our codebook to analyse the final selection (n=601) of tweets (step 6). Deductive logic is a concept driven approach to research that opts to test out the effects that the theories and

(35)

models that have been studied has on the data collected (Schreier, 2012). Through this process we established the set of variables that we wanted to measure, which were _{actors, communication type, issue frames and resource mobilization tactics.}

Actors. _{From the background study and literature review conducted, we identified}

key actors that were typically involved in the discourses and debates demanding for a social, environmental and/or economical discourses as well as those who were detrimental to bring about such changes. We wanted to confirm whether these actors also exist in the #GreenRecovery discourse on Twitter and therefore added _{actors as the first variable for our codebook and have set businesses, social} movement actors and organizations_{, news media, political actors as well as} individuals_{as the preliminary categories for this variable.}

Type of message and communication. _{Based on our preliminary study on}

theories (Gaither & Austin, 2016; Gomez, 2011; Miller and Rendall, 2018; Shirky; 2011; Tsatsou, 2018; Whelan, 2013), we discovered that social, environmental and political discourses that occur online via Twitter typically involve users from different backgrounds who use the platform for various purposes. Specific to the context of actors such as activists and social movements we learned that communication tends to be centered towards expressions of collective interests and expressions (Olson; 1965, Habermas; 1989; Klandermans & Stekelenburg, 2009), demands for policy changes (Olson; 1965; Tarrow, 1998) and changes in irresponsible business practices (Zald, 2006; Georgalis & Lee, 2017) as well as to raise awareness and information on particular issues (Miller and Kendall, 2018; Tsatsou, 2018). For businesses, aside from the above, communication on social media is often also focused on sharing the socially responsible business practices and efforts that they have conducted (i.e. CSR or CPA efforts) (Gaither & Austin, 2016; Gomez, 2011; Whelan, 2013). Therefore, the we’ve established the following categories to measure the _{communication type variable:}

a. awareness building and information sharing,

(36)

c. demands and/or recommendations for change in policies,

d. demands and/or recommendations for change in business practices, and e. corporate regulations and mentioning of (own) corporate responsible

practices.

Issue frames. Adapted from the theory of social movements, framing tactics are considered popular persuasive tools often used in social movement communication practices. To lay out a foundation for the categories of variable twe adopted several frames that were categorized by (Alonso-Muñoz & Casero-Ripollés (2018) in their study focused on the _{Political agenda on Twitter during the 2016 Spanish} elections _{(see Appendix A, section 1.2 for details on their complete codebook and} descriptions)_. _{From their list of categories, we incorporated the economy, social} policy, science and technology, infrastructure_{, and environment categories, to our} preliminary codebook with descriptions updated slightly to support the context of our study.

Resource Mobilization. _{Lastly, based on the understanding that all actors we}

have defined from our research (i.e. businesses, individual activists, social movement actors and organizations) often employ social movement resource mobilization strategies in their communication and campaigning efforts, (Johnston, 2014; George & Leidner, 2019; Skirky, 2011; Snow & Soule, 2004; Tarrow, 2012; Walker, 2015) we wanted to confirm if such tactics were visible in the #GreenRecovery discourse on Twitter and determined _{online petitions, online} protests _{as well as association and coalition building as categories to be coded.}

Developing and Enhancing Codebook Categories (Inductive)

The set of coding variables we established were later confirmed and complemented by the list of new and additional categories we have collected to refine and finalize our codebook through the inductive approach we adopted in step 7 as shown in Figure 4. above. Inductive logic in research is described as a

(37)

process often used to search for patterns which are driven by an analysis of text and data (Krippendorf, 1980; Schreier, 2012).

Through this process we made slight adjustments to our codebook and added new categories for frame types based on the themes we identified by closely analysing the tweets (n=601). The new themes that emerged through this process showed tweets that were framed towards demands of a _{redesign of the economy and} neo-liberal capitalist views towards the economy_{. Through the same process we} also identified that more political actors and accounts were involved in the discourse on Twitter, which were governmental institutions, international organizations, national and and EU governmental representatives and bodies as well as EU funded programmes. Considering this we have updated our pre-existing political actors variable, into a more specific variable we termed _National Government, EU Government, Other Political Institutions and updated the definitions accordingly.

Confirming, adjusting and finalizing categories (Retroductive)

As means to confirm and ensure that we developed a framework that is measurable and well formulated, we then used a retroductive approach (Sayer, 1992) where we coded 1% of tweets from our final selection (n=60) to check for consistency and conduct one last stage of thematic exploration (step 8). Through this process no new themes emerged and no changes were made to our codebook as a result. To obtain a complete overview of our codebook as well as the definitions for each category, see Appendix A, section 1.1.

6.3.4 Coding Procedure

We coded the selection of tweets (n=601) based on the final variables and categories we defined for our codebook, the final codebook (see Appendix A, section 1.1. for detailed descriptions). Each tweet acted as the main unit of analysis (Krippendorff, 2004). To code, each tweet was analysed based on the variables we have defined. First the _{user_description was checked to code the}

(38)

actor type. Then we analysed the tweet itself to identify the _{message and} communication type and _{issue frames that were used. Lastly, we checked the} tweet (if a tweet included a short URL then this would be opened and analysed as well) to identify if a _{resource mobilisation strategy was present and coded it based} on the categories we defined for our codebook. All tweets were coded manually in Excel (see Appendix A, section 1.1 for the link to the final codebook).

6.4 Methodological Reflections

Using content analysis as our research method came with several benefits and challenges. Boyd & Crawford (2012) point out that Big Data has transformed knowledge, by stating that “it is a profound change at the levels of epistemology and ethics. Big Data reframes key questions about the constitution of knowledge, the processes of research, how we should engage with information, and the nature and the categorization of reality” (p.665).

One of the most challenging yet valuable experiences gathered throughout this project due to the chosen methodologies was gaining the opportunity to observe and approach the research from a neo-positivist standpoint. Based on the understanding that paradigms cannot be judged on the production of truth, we were more critical in our literature selection and review process. Second, we faced challenges during our coding process due to the variables we have defined. All our variables were latent variables and implicit in nature, which required extra time, attention and effort to code. Lastly, speaking of the more general challenge of the chosen method was the decision to analyse what is being said in the discourse through the hashtag #GreenRecovery. As Bruns et al. (2012) stated “it is virtually guaranteed that some users tweeting about the topic will be unaware of the existence of the central hashtag, or even unfamiliar with the concept of hashtags altogether” (as cited by Pond, 2016, p.157). Considering this, we have only captured a small fragment of the overall Green Recovery discourse.