3.1 Empirical analysis of the spiral of silence

Extensive research on the spiral of silence theory predominantly centers on its application within political communication [16, 21–25]. These studies typically employ survey designs to empirically validate the theory, using regression analyses to ascertain relationships between individuals’ willingness to express support for political candidates, their perception of the prevailing opinion climate, and their level of fear regarding social isolation. For instance, Kushin et al. [16] found the significant role of fear of isolation in the prediction of individuals’ expression of support for Hillary Clinton or Donald Trump during the 2016 US presidential election. Similarly, Chan [21] observed, within the context of Hong Kong’s 2015 elections, a negative correlation between fear of isolation and self-censorship, noting that individuals were more likely to express their opinions within groups of like-minded political ideologies. However, certain research indicates the theory’s constraints, suggesting its diminished applicability among individuals with pronounced attitude certainty [23, 26].

In addition to survey methods, qualitative approaches such as interviews have also been instrumental in exploring the spiral of silence theory. For instance, Fox and Warber [27] conducted interviews with 52 participants before and after the 2012 U.S. election, uncovering a spiral of silence among participants whose LGBTQ+ identities were concealed within their social networks, thereby subjecting them to the perceived heteronormative majority. Furthermore, experimental designs have yielded insightful data; Masullo et al. [28] orchestrated an online experiment that revealed individuals’ emotional responses, particularly negative ones elicited by rude comments, as pivotal in determining their propensity to engage in discourse, overshadowing the prevailing opinion climate. These findings suggest that emotional reactions may significantly influence the dynamics of the spiral of silence, indicating a nuanced interplay of factors that dictate public expression. The relevance and implications of these findings, particularly in non-political contexts such as the entertainment industry scandal discussed herein, warrant further exploration.

A limited corpus of research has ventured into testing the spiral of silence theory using observational methods on large-scale data sets. For example, Morales [17] capitalized on an event during which an automated bot retweeting Venezuelan President Maduro’s tweets was deactivated, leading to a notable surge in critical tweets against the president. This study, using the framework of the spiral of silence theory, posited that public opinion can shift when the promotional efforts of an autocratic regime are interrupted. Despite these advancements, an important gap remains in understanding how silence is disrupted and public opinion subsequently escalates, particularly in non-political contexts like the entertainment industry. By meticulously documenting this transition through the analysis of extensive social media data and news archives, the present study seeks to capture the dynamism inherent in social issues, breaking the spiral of silence. Furthermore, this research aims to significantly enrich the field of public opinion studies by elucidating the specific actions undertaken by social media users and news media personnel throughout this transformative process.

3.2 The impact of social media on the spiral of silence theory

There exists a small body of literature discussing the impact of social media on the Spiral of Silence Theory. For example, Chaudhry [29] analyzed comments attached to articles about race, racism, and ethnicity on Facebook pages. The results indicated that minority users with discriminatory views did not hesitate to express their opinions. This suggests that the effect of the Spiral of Silence Theory may be limited on Facebook. Furthermore, Laor [30] demonstrated through interview research that Facebook groups serve as a receptacle for diverse opinions. What the interviewees highlighted was the value of closed Facebook groups where they could share opinions different from those in the general society. This study suggests the potential for discourse on social media to be picked up by news media. Sohn and Sohn [31] clarified the behavior of the Spiral of Silence through simulation. They found that while the Spiral of Silence phenomenon is less likely to occur when an individual’s social network is small, it becomes more probable as the network expands. The authors suggest from these results that the Spiral of Silence phenomenon might occur in the context of mass media with universal access, whereas the formation of opinions might differ in online environments where opinions are locally concentrated.

The theoretical background for such occurrences can be explained by considering the structure of the Spiral of Silence Theory. The background for minorities to remain silent includes the fear of isolation. That is, the premise is recognizing that one is in the minority and fearing it. However, on social media, due to the existence of echo chamber phenomena or communities that share values like Facebook groups, individuals are more exposed to opinions similar to their own. This means that even if one is in the minority in real society, it is possible to avoid the fear of isolation on social media. In this way, opinions aggregated on social media can be picked up by news media and have the potential to shape public opinion.

3.3 #MeToo movement

The issue of Johnny Kitagawa can be contextualized within the broader #MeToo movement, which spotlighted sexual abuse allegations globally. This movement bears notable similarities to the case in Japan, echoing the scandal surrounding film producer Harvey Weinstein, who faced allegations of sexual abuse from over 80 individuals across three decades. Additionally, the #MeToo’s impact is further underscored by the recognition of the women who spoke up against sexual abuse, called the “Silence Breakers” [32], which infers a more analogous relationship can be found with the current issue.

In Western countries, extensive studies have been presented on the theme of the #MeToo movement and the media. Benedictis et al. [33] examined how #MeToo was handled in British newspapers, revealing that the newspapers played a significant role in enhancing the campaign’s visibility. Benedictis et al. [33] pointed out that, although there were some differences among newspapers, many treated #MeToo positively. However, they criticized the focus on the experiences of wealthy white female celebrities and the lack of discussion on solutions to sexual victimization. Noetzel [34] conducted a quantitative content analysis over two years, from one year before to one year after the #MeToo tweets, of news articles published in four U.S. newspapers. As a result, it was shown that not only physical sexual violence but also verbal sexual violence began to be reported, and there was a shift from a posture of pursuing third-party responsibilities to one of pursuing individual perpetrators’ responsibilities. However, the authors have warned against individualizing the issue, emphasizing that sexual violence is a societal problem.

In the literature on the spiral of silence theory and #MeToo and the media, there are few studies that dynamically and detailedly describe the process in which the silence state is broken, and public opinion subsequently gains momentum. This study, by using a large amount of social media data and news records to describe that process, records the dynamism in breaking the social silences, making a significant contribution to public opinion research by clarifying how social media users and news media editors reacted in that process. This study particularly excels in comprehensively analyzing the discourse across diverse mass media (TV, newspapers, tabloids, magazines, online news) and social media in a chronological manner. This effort, as examined in the previous section, provides profound insights into whether social media weakens the effect of the Spiral of Silence Theory. Ultimately, it adds insights into an essential discussion in social sciences regarding the visualization of social issues.

Furthermore, it is worth noting that the #MeToo movement has not gained as much traction in Japan compared to international arenas [35, 36]. Cultural factors allegedly create an environment where public discourse is challenging [33]. In such a constrained atmosphere, the instances of the present paper where societal silence is effectively shattered are infrequent. Therefore, our study’s exploration of of spiral of silence e in the context of Japan’s unique cultural landscape offers critical insights into the dynamics of silence breaking within movements like #MeToo.

3.4 Research question

Previous research has highlighted the importance of “fear of isolation,” the relationship between social media and the spiral of silence theory, and the spread of the #MeToo movement in the media. However, these studies faced methodological limitations such as the use of sample data, interview, and simulations. In contrast, this paper utilizes a comprehensive and chronological large-scale dataset to elucidate the impact of social media on the spiral of silence theory. Specifically, it provides a detailed description of the dynamics involved when silence is broken and offers a contemporary interpretation of the spiral of silence theory. The research question is as follows:

1. RQ: In the context of sexual abuse issues that were silenced by mass media, what role has social media played in shaping public opinion?

This study responds to the above research question by analyzing a large-scale dataset and seeks new developments in the application of the spiral of silence theory.

4.1 News media data

The news media data for this study is sourced from Ceek.jp News (https://news.ceek.jp/), a comprehensive aggregator of virtually all Japanese news media articles. The dataset encompasses various details, including the publisher name, headline, and the initial segment of the news articles that are accessible without a paywall. For the purposes of this study, we extract news articles that contain “ジャニーズ” (Johnny’s: the most common naming of Johnny & Associates in Japan) either in the headline or the introductory section of the news content. The timeframe for the articles for this study spans six months, from March 1, 2023, to August 31, 2023, encompassing a total of 14,808 articles. These articles were derived from 219 distinct news domains, including tabloids, newspapers, and televisions, providing a broad spectrum of perspectives and reporting styles on the issue. Here, to concentrate solely on Japanese media outlets, we excluded BBC news from the dataset. The rationales for this process are that BBC is exceptional as a silence breaker in the issue; we could not identify any other foreign news (e.g., CNN) written in Japanese reporting on Janney Kitagawa among the top 150 news domains; it is the domestic media that are considered to have substantial commercial ties to the Johnny & Associates [37].

4.2 Social media data

For the social media component of this research, we employ data from X (formally Twitter), a platform that boasts significant popularity in Japan, accommodating approximately 67 million active users, nearly half of the country’s population [38]. This figure is roughly on par with the number of daily active users in the United States [39], emphasizing X’s substantial reach and influence in the Japanese social sphere.

The data collection process is conducted via X’s search API. Our search queries are Japanese keywords (translation in parentheses): “ジャニーズ” (Johnny’s) OR “ジャニー” (Johnny) OR “ジャニ” (Johnny) OR “#ジャニーズ” (#Johnny’s). This approach is designed to encompass a wide range of public discourse pertaining to the Johnny’s scandal. The timeframe for data aggregation is from March 1, 2023, to August 31, 2023, aligning with the period of news media collection. During this period, we amass a substantial corpus of approximately 14.5 million posts. Of these, 11.3 million are reposts (constituting 77.5%), while the remaining posts were a mix of quotes, replies, and original posts. The number of users is 1.5 million.

8.1 Methods

8.2 Results

Fig 3(a) distinctly illustrates the fragmentation of the repost network, mirroring the well-documented echo chamber phenomenon [70], which is represented by groups with pronounced polarities. Such groups typically have opposing viewpoints and tend to communicate predominantly within their own clusters, rarely engaging in cross-group interactions. It is interesting to see the emergence of this phenomenon in the presented case.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Network of user groups and the transition of their activity.

(a): Network visualization indicating nodes as users and edges as reposts. The color of nodes indicates the cluster that nodes belong to. (b): Table depicting the labels of clusters and representative words frequently associated with profiles of groups’ users, determined the scores by Tf-Idf. (c): Temporal trend of the weekly volume of posts by each group, with the upper chart showing post counts and the lower one presenting the relative ratios. (d): Temporal network snapshots showing the repost dynamics on specific dates (Apr. 12, May. 14, Jun. 12, and Aug. 31), indicating increasing polarization over time.

https://doi.org/10.1371/journal.pone.0306104.g003

Fig 3(b) shows the user profilings derived from the top 10 Tf-Idf words (translated from Japanese; see S5 Table for original terms), revealing a nuanced landscape of fan groups and their affiliations. The groups are composed of there are three types of fans (Fan1 (14.8% of the network nodes), Fan2 (11.6%), and Fan3 (11.3%)) and political users (Politics1 (32.2%) and Politics2 (14.5%)). On social media, political users often join emerging discussions (e.g., [56]), which was true in this case. We see divergent attitudes between fans of Johnny’s idols. Fan 1, typified by idol groups like Kinki (https://en.wikipedia.org/wiki/KinKi_Kids) and individuals like Takuya Kimura (https://en.wikipedia.org/wiki/Takuya_Kimura) and Jun Matsumoto (https://en.wikipedia.org/wiki/Jun_Matsumoto), showcases allegiance to the current Johnny & Associates. Conversely, Fan 2 and Fan 3 contain the names of groups such as SMAP (https://en.wikipedia.org/wiki/SMAP) and King and Prince (https://en.wikipedia.org/wiki/King_%26_Prince) that have distanced themselves from Johnny & Associates, either in the past or imminently. This division of the fanbase reflects the situation of Johnny & Associates, where there has been a recent trend of several idol groups leaving the company to pursue independent careers even before the reveal of scandal [71, 72].

By looking at the most shared posts by each group (see S6 & S7 Tables), Fan1 primarily shared the posts claiming “no evidence of sexual abuse” or posts with the hashtag “#ジャニーズ事務所を応援します” (Support Johnny’s Office). Notably, even in the face of allegations related to sexual abuse, these posts appear to defend Johnny & Associates and criticize the accusers. On the other hand, the most shared posts by Fan2 are somewhat neutral, many of which simply indicated news that showed the progress of the case. Fan3, contrarily, mainly shared posts criticizing Johnny & Associates and the company’s president. Thus, we can tell the stances of Fan1, Fan2, and Fan3 as pro-Johnny’s, neutral, and anti-Johnny’s, respectively. In the Politics groups, the most shared posts composed of criticisms of varied factors of this issue, including the injustice that only Johnny’s company has been allowed for years compared to other scandals, the media and entertainment industry for their silence, the political party’s handling of the situation, and feminists. Note that Politics groups are basically anti-Johnny’s.

Fig 3(c) shows the temporal volume of posts for each group. It shows that Politics group and Fan2 (Neutral) are the first to respond, followed by Fan3 (Anti-Johnny’s). Then, Fan1 (Pro-Johnny’s) finally showed up in May, which suggests there is an association between the strength of the relationship with Johnny & Associates and the delay in response.

The chronological breakdown of networks post-BBC documentary paints a compelling narrative (Fig 3(d)). Initially, one side of the polarity dominated the discourse. However, after the apology video of the company’s president (May 14) and with the onset of the external investigation (Jun. 12), the other side of the network emerged and eventually shaped the echo chamber phenomenon. This transitionary phase, marked by the looming crisis at Johnny & Associates, potentially urged the current fans into action to protect the company.

In the analysis above, we found a slow response from Fan1 (Pro-Johnny’s). Then, how the level of their interest transcend? Fig 4 shows the ratio of sexual abuse posts in each user group in their Johnny’s related posts. Fig 4(a) indicates the aggregation over the entire time period and shows that overall Fan1’s engagement with sexual abuse topic is low. Also, Fig 4(b) shows the transition of users’ interest in the scandal, which shows that Fan1 (Pro-Johnny’s) continued to have low engagement and finally caught up with Fan2 (Neutral) in August, when their engagement was highest. In other words, for a while after the scandal came to light in March, Fan1’s interest was not directed toward the scandal, indicating a strong atmosphere of silence regarding this issue.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Ratio of sexual abuse posts in each user group in their posts related to Johnny’s.

(a): The aggregation of the whole period (six months) and (b): transition.

https://doi.org/10.1371/journal.pone.0306104.g004

To gain a deeper understanding of their opinion, we applied topic modeling and sentiment analysis to the posts related to sexual abuse (see Methods for details). For topic modeling, we obtained six topics (parenthesis are the labels of topics that we put):

1. Contents of the press conference of survivors, the investigative team, and the UN group (Press conference)
2. Reports on scandals and investigative findings (News and fact-finding)
3. Company’s responsibility and compensation (Responsibility)
4. Johnny Kitagawa and the details of the scandal (Johnny and crime)
5. How the media’s surmise exacerbated the scandal (Mass media’s surmise)
6. Concerns of fans about the idols (Fans’ feelings)

We also perform sentiment analysis to categorize each post into eight sentiments (see Methods). Fig 5 is a heatmap showing how many topics and sentiments each group of posts had. The color of the heatmap indicates the degree of deviation from expectations (see Methods), with pink indicating more posts than expected.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Heatmap representations of topics-users (a) and sentiments-users (b).

Color of cells indicates the degree of deviation of post counts from expectations in each cell about topics and users (see Residuals). A shift towards pink indicates more posts than expected. Note that we omitted the “trust” from the sentiments due to its exceptionally low counts.

https://doi.org/10.1371/journal.pone.0306104.g005

Fig 5(a) reveals that Fan1 demonstrates a marked interest in news and fact-finding, while exhibiting diminished concern for other topics. Contrarily, Fan2 and Fan3 show limited interest in news and fact-finding, with a pronounced focus on corporate responsibility. The Politics category, along with Others, predominantly pertains to the media’s surmise and the contents of the crime. The topical interests align with the posts that they most shared, which would lead to Fan1’s maintained critical perspective on media coverage of the scandals, while Fan2 is interested in the contents of news, and Fan3 frequently critiques Johnny & Associates.

Fig 5(b) indicates the prevalent sentiments among groups. Fan1 predominantly exhibits feelings of anger and disgust, signifying dissatisfaction with news coverage. Fan2 shows heightened surprise and anger, which makes sense considering their interests are directed to the contents of the news. Fan3 expresses disgust, which is seemingly towards Johnny & Associates. Politics groups display heightened anticipation, talking about improving the situation and future.

9.1 Main results

Compared to traditional news media, users on platform X responded more promptly to Johnny’s sexual abuse issue (Fig 1). This suggests that social media users contributed more to breaking the spiral of silence around the matter. This observation aligns with the recent study that social media can undermine the effects of the spiral of silence [29]. Within the social media realm, users often form reference groups, as demonstrated by several studies [15–20]. Specifically, when individuals recognize that their peers on platform X are engaging in discussions of a particular issue, they become more likely to express their viewpoints, regardless of whether mainstream media has acknowledged the issue [18, 73].

Contrarily, mainstream Japanese news media outlets–television–lagged behind the BBC and X users in reacting to the scandal (Fig 2). This study thus offers quantitative evidence of the “silence of news media” highlighted in various news articles and reports [5, 8, 74]. Also, the pace of Japanese media coverage stands in contrast to the rapidity with which Western media outlets reported on the MeToo movement [33, 53]. Several potential factors may explain this discrepancy. For example, in addition to the suspected collusion between the mainstream media and the entertainment industry, Japan may have had few proponents of the debate, such as the feminists who supported the #MeToo discussion in Western countries. One way to speed up news media coverage, for example, would be to deregulate the industry and increase the number of television stations [75].

On the other hand, the news media’s reaction was most evident following the apology video by the president of Johnny’s company (Fig 2(a) and 2(b)). Subsequent to this press conference, the prevailing public opinion began to perceive Johnny’s issue as a genuine concern, elevating it to the status of a “social issue.” Actually, TBS, a major Japanese media outlet, admitted in a retrospective analysis that its staff initially regarded Johnny’s issue as mere gossipp [37]. The progression of this issue—becoming a public concern primarily after the apology video from the company’s president, who was aligned with the perpetrator rather than centering on the survivors’ accusation—may suggest potential commercial interests intertwining the news media and entertainment industries. Such dynamics require further research and scrutiny.

As for the users on platform X, we found that whether or not their favorite idol groups belong to the company might have influenced their perception of the sexual abuse scandal. Particularly for Fan1 (Pro-Johnny’s), escalating reports of the abuse would likely damage the reputation of their favored idol group, which could be a reason to remain silent on the matter. Such inherent biases of idol enthusiasts may play a role in shaping the opinion climate and fostering the spiral of silence. Yet, since idol fans do not have any incentives to break their silence, disrupting this spiral might necessitate external intervention, such as media exposure. Additionally, the departure of the idol group supported by Fan2 and Fan3 might be substantially attributed to the demise of Johnny Kitagawa. An in-depth evaluation of these factors in dismantling the spiral of silence would be necessary.

As depicted in Fig 3(d), echo chambers emerged among user groups. It’s critical to highlight that, amidst the controversy, Fan1 (Pro-Johnny’s) maintained a minority viewpoint (14.8% of the network nodes), defending Johnny’s company. In an environment where the overwhelming majority criticized Johnny’s company, these fans stood out as the sole defenders. In this context, they can be categorized as the “hard core” group, as described by Noelle-Neumann [14, 26] in relation to Johnny’s controversy. Actually, Fan1, who is pro-Johnny’s, had been silent for some time since the scandal came to light (Fig 4). If they were indeed a hardcore group, even if they were aware of the fear of isolation, a strong attachment to the Johnny company could have fostered a high level of conviction in their stance, making them less resistant to expressing opinions divergent from public sentiment [26]. Alternatively, they might have been insulated from the fear of isolation by the echo chamber effect, being surrounded by peers sharing similar views [16].

9.2 Social media and the spiral of silence

Originally, the theory of the spiral of silence was posited in an era predating social media. However, the advent and ubiquity of social media have reshaped the communication landscape, necessitating a reevaluation of the theory’s applicability. Three salient points from the Johnny’s case illustrate this shift:

• Rapid Social Media Reaction: The immediacy with which social media users responded to the BBC report, juxtaposed against the lag in news media response, highlights the agility of digital platforms in disseminating information.
• Echo Chambers: Even when faced with an overwhelming societal narrative to the contrary, the visibility of like-minded opinions within their echo chamber enabled them to voice their support without apprehension.
• Amplification of Opposition: The public opinion condemning Johnny & Associates, which was in the minority before the issue was made visible, became the majority after being reported by the major media. Subsequently, an opposing public opinion emerged in defense of Johnny & Associates as a counter to the majority, which then expanded its network through echo chambers.

Especially at the time the Spiral of Silence theory was proposed, it highlighted the phenomenon where, fearing isolation, people conformed to the majority opinion, leading to a situation where the majority grew increasingly dominant, the minority increasingly marginalized, and the free market of opinions distorted. However, what this study has implicated is a story in which the minority (initially those criticizing Johnny & Associates, and later, those defending it) gains strength potentially through the power of echo chambers. Thus, in the social media epoch, the spiral of silence theory’s core tenets requires contextualization. This study is novel in that it clarifies the role of social media under circumstances where silence prevails, through a comprehensive and chronological analysis of the interaction between multiple mass media and social media regarding this issue.

9.3 Implications for the media industry

The media fundamentally serves to illuminate societal injustices and elevate them to the status of public discourse. In this context, the BBC, an international broadcaster, assumed this responsibility. Yet, it was the discussion on the social media platform X that catalyzed widespread awareness in Japan. This digital fervor prompted an apology video by the president of Johnny’s, marking a turning point wherein Japanese mainstream media amplified its coverage, solidifying the matter as a nationally recognized issue.

This study suggests that in the current media environment, where social media is widespread, social media can break the silence even if the mainstream media remains silent, provided there is a trigger. On the other hand, it also revealed that reversing the majority and minority can lead to the emergence of an opposition to the new majority, and this opposition can be amplified.

Japan’s television industry is notably oligopolistic, with a starkly fewer number of TV stations than countries like the United States [76]. This is attributed to the permissive nature of broadcasting rights, which have become significant vested interests. Given this structure, Johnny’s company is a pivotal player, influencing viewer ratings across nearly all TV stations. This entrenched relationship potentially fosters undue influence. Introducing more TV stations, ensuring their independence from specific entertainment entities [75], and promoting inter-station competition could mitigate such concerns.

9.4 Limitation and futurework

In this study, we explored the disruption of silence surrounding the most significant scandal in the annals of Japanese entertainment. We acknowledge several limitations of our current approach. Firstly, while our keyword-centric method for identifying posts about sexual assault prioritizes the recall rate, it may overlook posts that don’t explicitly contain our chosen keywords but are still relevant to the topic. Employing machine learning models could address this oversight. Additionally, our categorization of users and posts relies on unsupervised learning rather than theory-driven coding, which presents an area for future refinement.

Regarding future research directions, as delineated in our discussion, several pertinent issues emerge. These include the relationship between commercial ties of media and entertainment agencies and the spiral of silence, the influence of Janie Kitagawa’s passing on this spiral, and the external pressures imposed on the Japanese media by foreign entities such as the BBC. While these are crucial subjects, they could not be comprehensively addressed given the constraints of our current dataset. A more in-depth exploration of these topics would provide valuable insights into strategies for breaking societal silences and fostering an environment more conducive to supporting the socially vulnerable.