Exploring Transition Tensions in Public Opinion on the COP26 Coal Phase-out Deal for South Africa as Expressed on Facebook

ABSTRACT

The 2021 COP26 meeting presented South Africa with an $8.5 billion deal to reduce its heavy reliance on coal, sparking a renewed public debate about transforming the country's coal-fired energy system to address emissions, energy deficits, and declining services. This paper examines public opinion on this important energy transition initiative as expressed on social media. While the use of social media platforms for public deliberation on policy matters is increasing in Africa, research exploring the African social media landscape in the context of energy transition is limited. This paper addresses this gap by qualitatively analysing 3,980 Facebook comments on 31 news posts related to the COP26 deal using sentiment and thematic approaches in ATLAS.ti 22. The findings reveal a prevalent negative sentiment and delegitimizing opinions that challenge the deal's credibility. Prominent topics within the discourse encompass concerns about corruption, distrust in public institutions, and perceptions of foreign involvement. Although some motivating factors supporting the deal emerged, negative sentiments and viewpoints dominated the discourse. Studying symbolic practices related to energy visions in this underexplored Global South context yields valuable insights into public opinions on energy transitions, highlighting the link between governance institutions and societal attitudes toward energy transition.

KEYWORDS:

• Energy transition
• energy conflicts
• public opinion
• social media
• global south

Introduction

The Global Coal to Clean Power Transition Statement, a significant commitment from the 2021 United Nations (UN) Climate Change Conference of Parties (COP26) in Glasgow, called for an accelerated “transition away from unabated coal power generation” to meet the Paris Agreement’s goals (UN Climate Change Conference UK, 2021). This focus on the energy sector is crucial, as it currently contributes to about three-quarters of global greenhouse gas emissions (IEA, 2021), making energy transition essential for achieving the 1.5°C global temperature goal.

Energy transitions involves shifting from carbon-intensive energy systems to low-carbon alternatives, with renewable energy (RE) technologies like solar, wind, geothermal, hydro, and biofuel playing a pivotal role (IRENA and AfDB, 2022).

The concept of “phase-out” refers to the gradual reduction of production and use of carbon-intensive energy in favor of renewable energy technologies (Wang et al., 2022). Renewal energy technologies offer the potential to reduce emissions while expanding energy options to meet growing demands (Batinge et al., 2019), particularly in developing countries struggling with energy poverty. Sub-Saharan Africa has the highest population without access to electricity and clean cooking energy (Oluoch et al., 2020) and is identified as a major driver of global energy demand growth (IEA, 2019). Selecting an appropriate energy mix is critical for cost-effective and reliable energy services expansion across Africa (Boulle, 2019). However, despite the abundant renewable energy resources in Africa, their deployment and adoption have remained relatively low (IEA, 2021; IRENA and AfDB, 2022).

Within the framework of a just transition, developing countries require significant financial assistance to facilitate transitions with minimal disruption to societal systems (IRENA and AfDB, 2022). During the COP26 meeting, South Africa secured an $8.5 billion deal from the United Kingdom, France, Germany, the United States, and the European Union to reduce coal usage for energy. This coal phase-out deal, lauded as a “watershed moment” by the South African government (Mkhize, 2021), is meant to support the country's ambitious emission reduction targets outlined in its Nationally Determined Contribution (398–510 MtCO2e in 2025 and 350–420 MtCO2e in 2030). Furthermore, it will provide support to workers and communities whose livelihoods depend on coal-related industries while diversifying energy options for South Africans seeking alternatives to Eskom, a government-owned utility heavily reliant on aging coal-fired power stations (DMRE, 2022).

However, energy transition policies often evoke mixed reactions, particularly in high carbon-intensive economies. “The sunsetting of fossil fuel” challenges established ways of living and doing business, generating tension (Sovacool et al., 2022, p. 3). Choices regarding alternative low-carbon energy sources are contested issues (Fergen et al., 2021; Oluoch et al., 2021). Conflicts resulting from perceived health and environmental risks (Mayer et al., 2021) and concerns over potential impacts on local jobs and economies (Egli et al., 2022) are prominently featured in the literature.

Given South Africa's coal-intensive energy system, the current energy crisis, and the livelihood practices dependent on coal-related industries, it is essential to assess public perspectives on the proposed energy mix changes under the COP26 deal. COP meetings are considered trigger events for public debates on sustainability practices (Chen & Rowlands, 2022), and the announcement of global financial support for the South African energy transition makes this period interesting for analysing public deliberations.

While conventional studies of transition discourse and social acceptance predominantly employ interviews, surveys, traditional media, and institutional reports (Guenther & Joubert, 2018; Lakhanpal, 2019; Miniard & Attari, 2021; Neukirch, 2020), the growing influence of social media as platforms for civic deliberation and engagement necessitates attention (Corbett & Savarimuthu, 2022). The platforms offer arenas for interactions among diverse stakeholders, including state entities, businesses, and non-state actors. They serve as spaces for news consumption, opinion-sharing, and emotional expression on a myriad of public issues, including those related to energy transition (Newman et al., 2021). Consequently, they are of interest to environmental (Bloomfield & Tillery, 2019; Gunster, 2022) and energy communication (Cozen et al., 2018; Katz-Kimchi & Manosevitch, 2015; Neubauer et al., 2023; Olausson, 2018) scholarship since neglecting the discourses they enable can create an “informational blind-spot” in policymaking, as they offer real-time access to public concerns (Corbett & Savarimuthu, 2022).

This paper qualitatively assesses the discourse on energy transition by analysing Facebook comments on news articles shared by South African news media regarding the COP26 deal. Drawing on Wüstenhagen et al. (2007, p. 2684) social acceptance framework, it examines the sentiments and opinions expressed by the networked public concerning the deal. Specifically, the study addresses how netizens (in this case, Facebook commenters) reacted to the news about the deal and their associated arguments and opinions. It analysed 3,980 Facebook comments from 31 news shares by eight South African news outlets using sentiment and thematic approaches in ATLAS.ti 22.

In the following sections, we discuss the study’s background, situating it within the social acceptance of transitions literature and the connection to communicative space of which social media forms a part. Next, we introduce the case context and outline the materials and methods used. Subsequently, we present the results of our analysis and discuss their implication for understanding energy transitions in the global south context.

Energy transition, social acceptance, and energy communication

Amidst mounting concerns over climate change and resource depletion, energy transition has gained significant attention, necessitating radical shifts in how energy is produced and used (Martiskainen & Sovacool, 2021; Sovacool et al., 2020). Energy transition involves shifts from carbon-intensive systems to cleaner and renewable energy systems (Karjalainen & Heinonen, 2018). However, progress remains sluggish despite the urgent need for climate action (Climate Action Tracker, 2021). Consequently, a growing body of research has been directed toward comprehending the social dynamics influencing energy transition (Martinez, 2020; Miniard & Attari, 2021; Neukirch, 2020), including within environmental communication (Cozen et al., 2018).

The literature suggest that energy transition is often characterized by conflicts stemming from divergent opinions on preferred energy mixes and transition timelines (Miniard & Attari, 2021), resulting in heated debates (Loureiro & Alló, 2020). To better comprehend this phenomenon, Wüstenhagen et al. (2007, p. 2684) introduced a three-dimensional framework encompassing “socio-political, community, and market acceptance” to grasp public responses to transitions. The first dimension concerns factors influencing the public, key stakeholders, and policymakers’ acceptance of technologies and policies related to energy transition. The second dimension, rooted in community considerations, encompasses procedural and distributional justice and trust. The third involves the acceptance of consumers, investors, and intra-firm dynamics.

Table 1 provides a concise overview of recent literature on the social acceptance of energy transition. These studies draw evidence from single-country cases (e.g. MacNeil & Beauman, 2022) and comparative studies (e.g. Komendantova, 2021), emphasizing that energy conflicts are multifaceted and arise from domestic and local contexts. They include concerns that cut across the three-dimensional framework, where procedural and distributional justice, among other factors, motivate local communities and allied industry stakeholders to express worries about local economy and threat to local jobs.

Table 1. Selected literature on social acceptance of energy transitions.

Location	Focus	Data source	Key finding	Reference
Germany, Austria, Finland, Morocco, Jordan and India	Behavioral drivers of ET	Surveys + interviews	Economic rationality, including energy prices and socio-economic impacts, takes precedence over environmental concerns	Komendantova (2021)
Australia	Understanding just transition among workers and coal community residents	Interviews	Just transition raises concerns and anxieties among coal workers and residents	MacNeil and Beauman (2022)
Germany	Protest networks over ET	Interviews + documents	Protests can have constructive aspects	Neukirch (2020)
UK	Policy stakeholders’ views on public perception of RE and emotion’s role	Interviews	Opposition is emotionally charged	Cass and Walker (2009)
USA	Perception and support for future energy mix and policy	Survey + interviews	Support is driven by environmental protection, public health, pollution, and economic improvement; opposition is driven by economic and employment concerns	Miniard and Attari (2021)
India	Local opposition to wind power project	Interviews	Contestations arise due to competing resource uses and insufficient community engagement	Lakhanpal (2019)
Kenya	Determinants of public willingness-to-pay for RE	Survey	Environmental concerns, RE type, and job creation influence willingness-to-pay, but not ownership, distance, or visibility	Oluoch et al. (2021)
North Africa	Perceptions of risks affecting RE projects	Interviews	Three classes of risks identified: regulatory, political, and force majeure, with regulatory risks being the most significant	Komendantova et al. (2012)

Exploring the field of environmental communication, which is relatively new in the context of energy communication research, there is a growing emphasis on symbolic practices related to energy resources, production, and consumption (Cozen et al., 2018). For example, Trisiah et al. (2022) conducted a decade-long analysis of geothermal energy framing in two Indonesian newspapers, identifying six common frames: energy security, economy, legislation, environment, knowledge, and social issues. Their study concluded that Indonesian media predominantly portray geothermal energy positively, emphasizing energy security and economic potential frames. The role of sources in news framing of renewable energy has also been studied (Gilbert et al., 2019). In their analysis of the Block Island Wind Farm's news framing in the United States, Gilbert et al.( 2019) found that business sources dominate, deviating from the traditional expectation of government source dominance in traditional media.

Ho et al. (2022) analysis of 26 online focus group discussions on energy campaigns in Southeast Asia highlights the effectiveness of digital media platforms in facilitating dialogical public engagement. Similarly, Katz-Kimchi and Manosevitch (2015) examined Greenpeace's Unfriend Coal campaign against Facebook's energy policy, reporting the social media’s effective use for mobilization and public engagement. In contrast, Neubauer et al. (2023) shed light on pro-fossil fuel groups in digital spaces that employ parallel strategies to promote the fossil fuel industry. Their study, examining Canadian pro-oil advocacy groups on Facebook, particularly references the use of “connective leadership” and “extractive populism” to mobilize support, reframe narratives, and position the industry as a matter of public interest. These studies collectively underscore the significance of communicative spaces in shaping understanding of various energy visions and the role of discursive strategies in shaping public opinion.

Advancing this scholarship from an African perspective, this study concentrates on the social-political acceptance dimensions of energy transition (Wüstenhagen, et al., 2007), particularly regarding the COP26 deal, a policy proposal to transform South African energy system. Leveraging the concept of a networked public or “online community” (Fergen et al., 2021), it aims to uncover insights into how Facebook discussions on the deal manifest markers of social acceptability or its absence. News media pages on social media create online communities where news is consumed, fostering active engagement through comments, likes, and shares (Nielsen, 2012). These platforms provide information-rich repositories of public opinion on topical issues, as they encourage unscripted conversations (Nuortimo & Härkönen, 2018; Singh & Shahid Husain, 2014). Focusing on them can bridge the divide between policymakers and citizens who are likely to be affected by energy policy shifts.

This becomes particularly crucial given the limited attention to “co-production” in the energy policy arena (Galende-Sánchez & Sorman, 2021). To address this gap, various concepts have emerged, including “energy democracy” (Allen et al., 2019; Bloem et al., 2021; Pesch, 2019), “energy citizenship” (Ryghaug et al., 2018), “energy publics” (Pesch, 2019), and “energy justice” (Monyei et al., 2018; Walker et al., 2019). These discussions elevate the role of the “public” to an aspect of collective sensemaking, enabling individuals “to consider themselves part of a greater whole and” believe they contribute to collective decisions through participatory processes (Pesch, 2019, p. 2).

Consequently, it is valuable to move beyond the analysis of conventional energy stakeholders (policymakers, developers, and academics) to assess energy deliberation on Facebook. If public deliberation involves individuals engaging in discourse to consider diverse perspectives and foster common understanding (Ho et al., 2019, p. 4), our goal is to identify and document the viewpoints prevalent in Facebook discussions concerning proposed energy transitions in the relatively underexplored South African energy communication context.

Table 2 provide cursory literature examining the connection between symbolic practices related to energy on social media and the public acceptability of transitions. Various analytical tools from quantitative sciences (machine learning, modeling, network analysis) and qualitative traditions (content analysis, discourse analysis, social network analysis) have proven valuable for mining unstructured social media data, yielding insights into energy deliberation in the digital space. Some studies highlight the extension of energy conflicts into the digital space (e.g. Fergen et al., 2021), while others map the various opinions and arguments underlining the conflicts (e.g. Romanach et al., 2015). This paper complements the literature by examining a less-explored South African context, which offers a significant opportunity to evaluate public perspectives on energy transitions through the COP26 deal.

Table 2. Selected literature on media and energy conflicts.

Location	Focus	Data source	Analysis	Key finding	Reference
Global	Citizens’ opinion on energy	Twitter	Big data/sentiment and complaints analysis	SM analysis can reveal real-time opinions and energy needs of people.	Bedi and Toshniwal (2022)
South Korea	Changes in public opinions on nuclear energy	NAVER articles + comments	Word count/ sentiment analysis	Nuclear power plant and concern about safety featured more prominently in the online discussion	Jeong et al. (2021)
Australia	Social acceptance of geothermal energy	News media	Content analysis	Economic feasibility and uncertainty as to the most reported risks. Industry and government actors as the most cited.	Romanach et al. (2015)
South Korea	Public perceptions of nuclear energy	Facebook, Twitter, NAVER, Kakaostory	Big data mining/ sentiment	Events and accidents affect changes in attitude towards nuclear energy.	Park (2019)
UK and Spain	Sentiments about climate change and energy preferences	Twitter	Sentiment analysis	Similar emotions in UK and Spain; positive sentiments associated with RE, negative with coal, and mixed feelings about nuclear.	Loureiro and Alló (2020)
Denmark	Anti-wind power viewpoints on Facebook pages	Facebook pages	Qualitative discourse analysis (ATLAS.ti)	Protest is localized and shared concerns are diverse.	Borch et al. (2020)
USA	Online-community conflicts around wind energy	Facebook pages	Content analysis	Presence of coordinated anti-wind activism online.	Fergen et al. (2021)
Iran	SM’s influence on intention to use RE	SM Instagram + survey	Structural equation modeling	Perceived risks of climate change and self-efficacy significantly affect intention to use RE.	Zobeidi et al. (2021)
Finland	Debate on biogas and heat pumps	SM Facebook	Social network analysis/ quantitative and qualitative content analysis	Discussion networks are conducive to widening engagement with potential for new ideas.	Rantala et al. (2020)

Methodology

Case description

South Africa, being the most emission-intensive and coal-dependent economy in Africa, accounts for 35% of the continent's energy-related CO₂ emissions, the highest among all countries (IEA, 2021). The country's industrialization depth and reliance on fossil fuels in its energy mix explain this situation. In 2018, coal and oil constituted 83% of South Africa's total primary energy supply (Figure 1). Furthermore, coal dominates electricity generation, accounting for over 90% of Eskom's generating capacity, the major government utility company (DMRE, 2022). Although there have been contributions from onshore wind, photovoltaic power, and concentrated solar power through the Renewable Energy Independent Power Producer Procurement Programme since 2011, the share of renewables remains low (see Figure 1).

Figure 1. South African total primary energy supply, 2018.

Source: Computed from the Department of Mineral Resources and Energy (DMRE, 2022).

Moreover, Eskom's inability to maintain a steady electricity supply has resulted in frequent power outages, locally known as “load-shedding” (Monyei et al., 2018). Due to the historical legacy of segregation, there is also a significant population, primarily black South Africans, who cannot afford or access electricity (Essex & de Groot, 2019). The combination of energy demands, and climate obligations has sparked the government's growing interest in renewables.

However, several challenges confront the South African decarbonization program. According to Tyler and Hochstetler (2021), powerful actors in the energy sector, whose interests are at stake, are concentrated and face potential threats due to the need for transition. Additionally, natural factors like the abundance of inexpensive coal in South Africa, shapes how the nation's energy futures are imagined (Scholvin & Ren Scholvin, 2014). Structural inequalities and high unemployment rates pose additional burdens on the country's energy transition project (Barnes, 2022). The extractive industry plays a significant role in the national economy by providing jobs, revenue, and other benefits to dependent communities. These factors, which have influenced energy debate in other contexts (Barnes, 2022; MacNeil & Beauman, 2022), form the focus of our analysis as we examine their discursive construction in Facebook comments on the COP26 deal.

Data collection

Facebook was chosen as the focal social media for this study due to its widespread use, second only to WhatsApp, and its position as the top platform for news consumption in South Africa (Newman et al., 2021). The data collection strategy is presented in Figure 2. Thirteen South African news media pages on Facebook were identified and examined for news posts specifically addressing the COP26 deal. The search terms used were “COP26,” “COP26, energy,” and “COP26, transition.” Only Facebook posts shared by news media based in South Africa were considered. The initial 36 record URLs were input into Exportcomments.com, a tool designed for extracting comments from social media and generating outputs compatible with ATLAS.ti for analysis.

Figure 2. Data sourcing strategy.

Excel files containing all comments on each record, including replies to initial comments, were downloaded for further screening. During this stage, five records were removed as they only discussed climate finance in the context of COP26 without referencing the phase-out deal for South Africa. A total of 5,833 comments were extracted from the 31 records. The search and data collection period occurred from March 19 to March 30, 2022, allowing for sufficient diffusion of news about the COP26 deal and significant citizen engagement. The 31 news articles on the topic were posted between November 2, 2021, and January 1, 2022 (see supplementary file for details).

Coding and analysis

The 31 records were imported into (“ATLAS.Ti 22 Windows,” n.d.), a qualitative data analysis software useful for organizing, annotating, coding, filtering, searching, and visualizing qualitative data (Rowley, 2012). ATLAS.ti has been employed in analysing energy discourse in newspapers (Batel, 2020) and on Facebook community pages (Revilla, 2020).

Iterative cleaning of the data was performed in ATLAS.ti, removing comments in non-English languages (approximately 6% of the total), empty cells (2%), and those containing unrelated content, advertisements, only emojis, or visuals with insufficient information for interpretation (approximately 25% of the initial total). Non-English content was excluded due to the investigators’ unfamiliarity with the local languages used. This data cleaning process occurred during the initial reading phase and the first cycle of analysis, resulting in a final inclusion of only 3,980 comments for analysis.

ATLAS.ti was utilized for both sentiment (Trisiah et al., 2022) and thematic analyses (Braun & Clarke, 2008). Each comment was considered as a unit of analysis, and marked for tone characterization adapting Trisiah et al.'s (2022) approach. Whereas Trisiah et al. (2022) only accounted for positive and negative tones, we identified and coded four tones characterizing comments on the COP26 deal: positive, negative, positive-but, and neutral. Positive comments indicated a supportive reaction to the deal, while negative comments indicated the opposite. The positive-but code captured instances where positive tone coexisted with reservations about the deal. The neutral code was assigned when tone is indeterminate.

Utilizing ATLAS.ti built-in sentiment analysis, we interpreted, the software's suggestions, guided by the “intertextuality” principle in discourse analysis, which emphasizes the interconnectedness of texts and their broader context (Androutsopoulos, 2009). We analysed each comment in conjunction with the news share preview, previous comments, and the South African energy context from the literature. This approach allowed us to gain a deeper understanding of the comment's meaning. Additionally, considering the diverse language functions employed by netizens on social media, including satirical expressions that might elude a generic computer model, this approach proved especially useful (Ross & Rivers, 2019).

Also, we identified and coded the topics raised in the comments. A comment could have multiple codes if it covered several issues related to the COP26 deal. We identified and coded 37 topics after several iterative coding rounds, using both inductive and deductive approaches (Friese, 2020). These topics were grouped into categories and further into themes utilizing the techniques of constant comparisons and affinity identification (Braun & Clarke, 2008). This stage drew upon the existing literature on social acceptance to interpret the emerging patterns, leading to the identification of three overarching themes: delegitimising and legitimising arguments or opinions, and shared concerns about the COP26 deal. Legitimizing statements encompassed arguments and opinions considering the deal for South Africa as right and socially acceptable, while delegitimizing statements questioned the rightfulness and social acceptance of the deal. Shared concerns referred to borderline issues considered relevant to the success of the deal. The resulting codes, categories, themes, and their relationships are depicted in Figure 3, with supplementary material providing additional information.

Figure 3. Network map of arguments and opinions on the COP26 deal for South Africa.

Author 1 independently coded and analysed the data but held several discussions with author 2 to ensure consistency throughout the process. Additionally, preliminary results were presented in two seminars, one of which focused on energy transition. These deliberative auditing processes acknowledged in methodological literature (Elliott, 2018; Kyngäs et al., 2020), were employed to enhance the analytical trustworthiness of our study.

Ethical considerations

All comments and news shares retrieved from Facebook are publicly visible and accessible. Therefore, obtaining informed consent was not required by the institutional review board that approved this study (Stellenbosch University’s Research Ethics Committee: Social, Behavioural and Education Research, Nr 25206). While demographic information was not relevant to the research and was not collected, precautions were taken to remove any personal details in direct quotes or representations of visuals in this paper. It should be noted that conducting a direct internet search of quoted texts may lead to an individual's comment. However, this is not problematic as the intention of commenters were to publicly express views on a matter of public interest.

Description of the news shares on the COP26 deal

Table 3 illustrates the distribution of news frames, which are interpretative models used to structure the presentation of news (Entman, 1993), for the 31 Facebook news posts. Framing involves highlighting a particular aspect of an issue as salient or important in the news. Most of the articles posted on Facebook (11) announced the deal, while another set (7) framed the issue as a conflict. Furthermore, five articles provided clarification on the deal, four articles presented supporting claims, and an additional four articles grouped under “others” discussed transition themes and made references to the deal. Our analysis also explored whether the variation in news frames influenced netizens’ affective comments on the COP26 deal.

Table 3. Share of articles by news frames.

News frame	Description	Count
#Announcement	Emphasise on the announcement of the deal	11
#Conflict	Emphasis on attacks from individuals/institutions on the deal	7
#Clarification	Emphasis on explanations provided about the deal	5
#Support	Emphasis on statements of support from individuals/institutions	4
#Others	Other themes related to the deal	4
		31

Results

This section provides the study’s findings, beginning with a descriptive analysis that emphasizes the distribution of “groundedness,” denoting the number of quotations associated with specific codes (Friese, 2022). Although counting in qualitative research can be debated, we considered this a valuable approach to draw attention to elements with widespread occurrence and highlight patterns in the dataset (Elliott, 2018).

Attitudes toward the news about COP26 deal on Facebook

Figure 4 offers insights into the sentiment of netizens towards the COP26 deal for South Africa in the Facebook comments analysed. A noteworthy observation is the prevalence of negative sentiments, with approximately three-quarters of the comments expressing negativity. In contrast, only 17% of the comments denoted positive tone, and another 3% were positive-but tone. The affective distribution does not inherently imply opposition to energy transition initiatives in South Africa, a point we will elaborate on later. Rather, it is closely tied to interpretations formed through learning about the deal and its association with existing local complexities.

Figure 4. Treepmap of code groundedness by sentiments in ATLAS.ti (N = 3 980).

To assess whether different news frame types influenced these outcomes, we employed the normalization function in ATLAS.ti within the code-document table, considering variations in document length (i.e. comment size under each news frame type) across our dataset (Friese, 2022). The results show that while negative comments remained dominant, posts framing the issue as a conflict elicited nearly equal shares of positive and negative comments (Figure 5). This implies that news media framing the issue as a conflict appeared to evoke polarized reactions among the audience.

Figure 5. Relative frequency comparison of sentiment by news frame-type.

Note: Differences in document size (number of posts and comments per a news-frame) were normalized in ATLAS.ti.

Underlying motivations for netizens’ attitude

The analysis of Facebook comments on the COP26 deal uncovers diverse motivations underlying the identified sentiments. Figure 6 visually represents the observed patterns by depicting the relationships between the topics and sentiments using the Code-Occurrence function in ATLAS.ti. This function visualizes the number of times two codes co-occur (Contreras, 2011), in our case, the sentiment codes against the thematic codes. The width of the Sankey lines represents the frequency of these co-occurrences, providing a visual indication of the strength of the association¹ between the coded concepts. In effect, Figure 6 highlights that statements with positive tone exhibits a stronger association with topics that legitimize the deal while having a weaker link to positive-but statements. Conversely, negative comments co-exist with delegitimizing topics. Shared concerns are linked to all sentiment groups but are particularly prevalent in negative statements, constituting around 80% of instances.

Figure 6. Sankey diagram of patterns of arguments and opinions in relation to sentiments.

Note: Co-Occurrence analysis visualized 3645 comments due to some comments lacking identifiable topics.

In effect, most commenters, irrespective of their sentiment, were mainly concerned about two governance topics that top the chart presented in Figure 7: “corruption/looting” and “distrust in institutions”. Following these concerns are opinions that draw on ideological postulations related to suspicion regarding foreign influence, perceived hypocrisy of the West and conditionalities associated with the deal, and equity considerations related to threats to jobs and the economy. These rationales were used to delegitimise the deal for South Africa. In contrast, the most frequently mentioned narrative supporting the relevance of the deal is the argument that opponents have vested interests in the coal-related industries. Other leading topics supporting the deal include concern for the environment, the potential for alternative energy and improved services, and the energy crisis as a justification for intervention.

Figure 7. Treepmap of code groundedness by topics (N = 3,980).

Key issues from the topic mapping

From the preceding session, it is evident that corruption and distrust in institutions were dominant topics in netizens’ perspectives on the COP26 deal. These topics were extensively used to delegitimize the deal, with commenters suggesting a link between perceived corruption and a lack of trust in the government's ability to deliver on the deal's promises. Comments 1 through 7 in Figure 8 illustrate the perception that the governing party, the African National Congress (ANC), is corrupt and that the funds would benefit private interests rather than the intended purposes.

Figure 8. Network of statements on corruption and distrust in institutions.

There were no significant differences in the prevalence of corruption and distrust in institutions among different news frames. However, these themes were particularly prominent in news posts announcing the deal. In some instances, corruption and distrust in institutions were used explicitly to reject the deal (e.g. comment 2 in Figure 8). These topics also appeared in comments with positive tone, albeit with recommendations to safeguard the funds (e.g. comment 6 in Figure 8).

Another significant factor was the strongly held opinions about foreign nations, especially the countries offering the deal. As illustrated in Figure 9, commenters expressed ideological postulations, such as accusing Western nations of hypocrisy (comment 1), viewing the deal as an imperial posture (comment 2), and perceiving it as increasing foreign influence and interests in South Africa (comments 3–6). Concern about foreign loans ranked next in the data and was influenced by how foreign actors were perceived. Commenters believed the deal has hidden conditions and would lead to a loss of autonomy (comments 3–6).

Figure 9. Network of statements on perceptions of foreign nations offering the deal.

Anxiety about job loss was another significant topic used to delegitimize the deal. Commenters argued that the deal could disrupt employment, particularly within the coal-related industry, and emphasized South Africa's high unemployment rate.

In terms of motivations supporting the deal, although they were fewer in comparison, they shed light on important aspects of South Africa's energy discourse. These topics predominantly appeared in the comment sections of posts that portrayed the deal negatively. The primary topic under this theme involved accusing opponents of having strategic interests in the coal-related industry. The next leading topic is concern for the environment, frequently appearing as a response to comments delegitimising the deal. These comments highlighted global warming and its impact on the environment and social systems as motivations for the COP26 intervention (Figure 10)

Figure 10. Sample exchanges between commenters opposing and supporting the deal.

Moreover, comments in Figure 11 underscored the connection between frequent power outages and the belief that renewables could enhance the country's energy infrastructure. Specifically, comments 1 through 5 referenced load-shedding as a significant issue, highlighting the necessity for alternative and sustainable energy sources. Additionally, comments 4 and 6 expressed the viewpoint that improving energy services through this intervention could yield ancillary advantages such as job creation and economic growth.

Figure 11. Network of statements on energy crisis and improved/alternative energy topics.

Discussion

This study utilized qualitative interpretation to examine public opinions on energy transition as expressed in Facebook comments on news posts covering the COP26 deal for South Africa. Thus, it responds to the call for recognizing African digital spaces as “sites of knowledge production” about citizens’ experience (Schoon et al., 2020). It also extends Wüstenhagen et al's( 2007) social acceptance framework to studying online communities. Our focus on Facebook, despite its informal nature, yields crucial insights into the social acceptance of the COP26 deal, which represents a significant policy shift in the South African energy context.

Firstly, the overwhelming negative sentiment towards the news of the deal on Facebook may seem paradoxical, considering the recognition of funding as a significant barrier to sustainability initiatives in developing countries (Kiremu et al., 2022; Mungai et al., 2022). Mitigation and adaptation efforts require substantial financing, often lacking in developing nations, including those in Africa (Mungai et al., 2022; Winkler & Marquand, 2009). One might expect public enthusiasm for the availability of substantial funds to support South Africa's transition. However, our analysis of a section of the South African public's discussion on the deal reveals a contrasting picture.

Approximately eight in 10 Facebook comments discussing the deal expressed negative sentiments, indicating a low level of social acceptance for the intervention. We argue that this discrepancy reflects the tension between the universal nature of climate change values (Barnes, 2022) and climate actions as diffused and local. While reducing emissions and addressing climate impacts have global implications and appeal, implementing these actions require mobilizing local politics to support domestic efforts. Consequently, the diverse social, political, and economic values within a society can influence the interpretation of universal climate values within a local context (Barnes, 2022).

Our analysis revealed that environmental values, economic benefits, and energy stability were the main supportive rationales for the COP26 deal. These elements have been shown to motivate support for transition initiatives in other contexts (Crawford et al., 2022; Miniard & Attari, 2021). However, in South Africa, these values were not among the top rationales discussed on Facebook. Instead, most concerns were local, focusing on governance issues, ideological postulations, equity considerations, perceptions of conditionality associated with the deal, and negative perceptions of renewable energy technologies.

This leads us to the second key finding of our study: concerns about corruption and distrust in public institutions are significant factors in public acceptance of energy transitions. While some of the topics uncovered in our analysis, bear semblance with issues raised by Neubauer et al. (2023) (e.g. suspicion of foreign influence and nationalism), our finding strikingly emphasizes the prominence of governance issues as local concerns in relation to the deal. Thus, in addition to previous research highlighting economic rationality, energy security, competition for natural resources, political affiliation, and environmental and health risks as leading causes of resistance to transitions (Jeong et al., 2021; Komendantova, 2021; Neukirch, 2020; Pritchard & Mills, 2021), we contribute empirical evidence on how perceptions of governance frameworks influence citizens’ attitudes.

The evidence of citizens’ distrust in government and the acknowledgement of public corruption in our dataset align with the global trend highlighted in the 2021 Democracy Index (EIU, 2022). The index indicates a decline in trust in government institutions worldwide, with “corruption, insufficient transparency, and a lack of accountability” listed as drivers of this mistrust (EIU, 2022, p. 27). Similarly, our findings suggest that Facebook commenters on the COP26 deal lacked confidence that the $8.5 billion would be utilized judiciously for intended purpose.

Our findings highlight the risks of corruption in transitions and its potential to undermine social acceptability. Sovacool (2021) reviewed study highlights the nature of these risks in South Africa's renewable energy development, including instances of diverting public spending, inefficient allocation of contracts based on patronage, tender rigging, and land grabbing. These concerns were extensively discussed in the investigated Facebook case. Considering that similar concerns were identified for transition projects in Mexico, Malaysia, and Kenya in the same review, one could argue that governance issues are central to transitions in the global South.

Another important finding is that Facebook commenters perceived the deal as an agenda driven by external actors, fostering a sense of suspicion. This aligns with the findings of MacNeil and Beauman (2022) in their examination of resistance to transitions in Australian coal communities and Neubauer et al.'s (2023) Canadian case. Similarly, comments on the COP26 deal for South Africa exhibited apprehensions about the deal being a foreign-led attack on the local economy, which heavily relies on coal. Commenters also used ideological postulations, such as conspiracy theories about climate change being a hoax, to support their suspicions and delegitimize the deal. While we consider Neubauer et al.(2023) “connective leadership” explanation for this phenomenon instructive, the nature of our dataset limits exploring the role of elite mobilization across the heterogenous news pages we analysed. Alternatively, we courted the national context of South Africa to explain our results.

South Africa's turbulent apartheid history is a probable factor that explains the observed disposition towards foreign interventions. This proposition builds on the argument that poorly managed decarbonization processes can lead to or exacerbate existing dispossessions (Manson, 2013; Sovacool, 2021). In this context, dispossession refers to how certain climate actions can deprive “others of their land, wealth, political and economic participation or other assets, exacerbating inequalities in the process” (Sovacool, 2021, p. 2). For example, one commenter expressed concerns about corporate capture and linked it to a fear of losing sovereignty (comment 4, Figure 10). A recent Afrobarometer survey confirmed similar views among the wider public, with more than six in ten South Africans preferring development funds to be sourced locally and 51% expressing aversion towards foreign loans (Mpako & Moosa, 2022).

Additionally, a lack of clear communication to members of the public about the details of the deal may have contributed to the suspicions voiced by citizens in the Facebook discourse. Several commenters questioned the donors’ intentions and the conditions associated with the deal. Lastly, we observed a presence of dialectical engagement among commenters in the Facebook discourse, highlighting the facilitating role of the digital sphere for deliberative democracy. As illustrated in Figure 11, comments often responded to previously posted comments or the original shared article. For this reason, we argue that investigating social media discourse should consider the relational networks of actors and their communicative actions.

Conclusion

By analysing Facebook discourse on the COP26 coal phase-out deal for South Africa, the paper provides valuable insights into public perceptions of energy transitions in the global South. Instead of dismissing transition tensions, it highlights the significance of understanding social practices related to decarbonization visions. Our case illustrates contributions that environmental communication scholarship can make to advance this understanding by providing evidence of various motivations and sentiments embedded in discursive social practice surrounding COP26 deal for South Africa.

The analysis of Facebook comments revealed a spectrum of opinions and arguments both in support and against the COP26 deal. Negative sentiments predominated, with concerns raised about corruption, distrust in public institutions, foreign involvement, loan conditions, and negative perceptions of renewables. However, some topics emerged in support of the deal. The key findings and their policy implications are highlighted below.

• Beyond environmental concerns, renewable energy systems offer solutions to societies facing energy crisis. In South Africa, where frequent power outages disrupt business and household activities, diversifying energy sources beyond coal power resonates with a significant portion of the public. Thus, effectively communicating the benefits of energy transitions can help garner public support.

• Governance considerations are pivotal for public support of energy transitions. The analysis indicates that positive attitude toward transition is low when there are perceptions of corruption and distrust in government. Therefore, a socially acceptable transition requires addressing corruption risks and a commitment to good governance principles throughout the transition process.

• Perceptions of capital sources also sway public views on future energy directions. In the South African case, there is apprehension about foreign actors, including concerns about resource ownership and foreign control during transition process. Promoting transparency and advocating domestic financing as a viable alternative can alleviate these concerns, especially as private sector was taunted as better resource managers in the discourse analysed.

• Citizen participation in energy policy processes is essential for effective energy governance. Dissatisfaction with procedural concerns and a lack of consultations was expressed. Encouraging bottom-up approaches and involving the public in all phases of decision-making can foster trust in the transition processes.

This study has certain limitations. First, the analysis did not include non-English language Facebook comments on the deal, potentially excluding further nuanced perspectives. The interpretative approach used for sentiment and thematic categorization, although supported by the ATLAS.ti analytical functionalities and several trustworthiness checks, could be complemented with deep machine learning in future research. Additionally, incorporating data from other social media platforms like Twitter, Instagram, and YouTube, as well as considering African languages, would provide a more comprehensive understanding. Despite these limitations, this study contributes to understanding public perceptions of energy transition as expressed on Facebook in a Global South context and offers valuable insights with implications for transition acceptability and energy governance.

Acknowledgments

The authors express their gratitude to Prof Phaedra Pezzulo and the anonymous reviewers for their valuable input in enhancing this manuscript. They also appreciate the feedback received during the presentation of the initial version at Stellenbosch University's Economic and Management Sciences Research Day and the Centre for Sustainability Transitions’ Students and Early Career Researcher Colloquium in 2022.

Data availability statement

Information regarding the dataset analysed in this study is available in the supplementary material.

Disclosure statement

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

Additional information

Funding

The first author benefited from the Stellenbosch University’s Committee A 2022 postdoctoral funding during the research period.

Notes

1 The term “association” here implies a general relationship, in contrast to “correlation,” which has a more specific statistical meaning. It indicates that one set of concepts (the sentiments) provides information about the other (thematic codes), and vice versa.