Conversational Gatekeeping—Social Interactional Practices of Post-Publication Gatekeeping on Newspapers’ Facebook Pages

ABSTRACT

Digital platforms, such as social media networks, have become intertwined in the news ecosystem, leading news media to lose their role as the sole gatekeeper in the public space. This development has given an active voice to audiences and turned journalism more into conversations between journalists and their audiences. The starting observation for this article was that alongside journalists, platforms and audiences play a part in the gatekeeping process that takes place post-publication, and therefore we need to gain a better understanding of this triadic relationship. Furthermore, as conversations are one of the main functions of social media platforms, more understanding of the role of social interaction in post-publication gatekeeping is needed. After analysing posts (N = 180) and their comments on Finnish newspapers’ Facebook pages utilising content and digital conversation analysis, we extend the traditional gatekeeping theory to post-publication practices of gatekeeping and finally suggest the concept of conversational gatekeeping. The concept explains how through social interaction journalists and social media audiences are able to build mutual understanding and create norms as well as decide on the content and action that is appropriate or wanted in the public news space formed on the particular online platform.

KEYWORDS:

• Conversational gatekeeping
• digital conversation analysis
• Facebook
• gatekeeping
• newspapers
• post-publication gatekeeping
• social interaction
• social media

Introduction

Journalists’ possibilities to interact with their audiences have multiplied in recent years. Since the journalism reform that took place along with the rise of Web 2.0., audiences have gained greater importance and journalism is now increasingly seen as a dialogue—journalism-as-a-conversation (Marchionni 2013). Consequently, the premises of gatekeeping theory have been challenged. In the current social media era, news media and their social media audiences can engage in conversations online to build mutual understanding and decide interactively what kind of content is accurate and acceptable for news media’s social media accounts. In this article, we introduce the concept of conversational gatekeeping to account for this interactional process. Drawing on qualitative content analysis of 180 Facebook posts (by four Finnish newspapers) and digital conversation analysis of three exemplary cases (posts), we demonstrate how journalists and their audiences together shape and negotiate news items after they have been published.

In the digital era, news organisations are no longer gatekeepers in the traditional sense as they cannot fully decide which items are in or out of the public sphere (Welbers and Opgenhaffen 2018; Vos 2020). The news ecosystem is constituted through various platforms of which social media are one (van Dijck, Poell, and de Waal 2018), and in this platformed news ecosystem the news organisations have lost control over their products (Nielsen and Ganter 2018). Due to platformisation and digitalisation, audiences now play a role in gatekeeping processes by affecting the visibility of a news item (Singer 2014), for example, and algorithmic platforms and technology companies influence the process through their settings and decisions (Wallace 2018; Vos and Russell 2019). Consequently, gatekeeping research has turned to exploring post-publication gatekeeping (Hermida 2020): gatekeeping as the processes and practices taking place post-publication, especially on social media.

In the pre-Internet era journalism has been described by some (e.g., Marchionni 2013) more as a top-down process where professional journalists, officials, and elites merely informed citizens. However, it can be argued that this has not entirely been the case. When media was still analogue, “letters to the editor” served as a conversational forum (and it still does) in most of the printed media. At the turn of the twenty-first century, the great potential of ´citizen journalism´ was emphasised by many journalism scholars to trigger conversations especially in the local and regional media (Franklin 2006). This was also the case in Finland where several research projects facilitated on-the-spot town hall meetings. However, the results remained modest (Heikkilä 2001; Kunelius 2001).

Consequently, the relationship of journalists with their audiences has been important to them even before the rise of the Internet. Yet, the ways to connect with audiences have multiplied and the relationship has become more complex as audiences are more present in the digital news ecosystem. That is why for example Marchionni (2013) calls for a theory of conversational news, i.e., audience participation in commenting on the news after its release. She further claims that clear conceptual and operational definitions are missing even though there is a plethora of audience-centred literature that is concerned with conversation. Thus, there is a need for research that focuses on conversational aspects of the news—in participatory journalism and in the post-publication environment.

One main function of social media is the conversations they afford and make publicly visible (Kavada 2015). As journalism is increasingly turning into conversations (Marchionni 2013), understanding social interactional processes between journalists and their audiences on social media can help to develop a better understanding of the conversational side of news and news engagement. While (digital) journalism has engaged with several other subfields (Ahva and Steensen 2020; Steensen and Westlund 2021) there is a lack of studies on social interactional relationships. Thus, we argue that the field benefits from a method such as conversation analysis (CA) that explores how mutual understanding is achieved through interaction between conversationalists (in this case journalists and their audiences) and specifically looks at the logic and function of conversation (Sacks, Schegloff, and Jefferson 1974).

The focus on social interaction is particularly warranted in studies of platformed journalism, where interaction and conversations between the news media and audiences increasingly take place through the social media platforms and news media have become dependent on platforms’ policies and affordances (van Dijck, Poell, and de Waal 2018, 61). When mediated by a platform, none of the participants can fully decide the premises of the conversation, as the affordances shaping it are dictated by the platform company. Furthermore, some aspects of conversations could be seen as problematic to the news media. After news media have opened their gates and posted on social media, audiences could be offering unwanted conversational practices such as hate speech or altering news stories and therefore perhaps spreading mis- or disinformation. This is an issue of gatekeeping, as news media are to some extent responsible also for the actions and content that take place post-publication on their social media accounts (Finnish Council for Mass Media 2017).

In this paper we introduce the novel concept of conversational gatekeeping to theorise the conversations between news media and their audiences on social media platforms. By building on the premises of digital conversation analysis, we see this as a process of social interaction, and thus a mechanism through which journalists and audiences build mutual understanding. Conversational gatekeeping is intertwined with interactants’ orientations to technology and describes the phenomenon of social interactional relations between journalists and their social media audiences as practices of post-publication gatekeeping. Our conceptualisation is based on a combination of previous theory and current empirical findings. More precisely, we present conversational gatekeeping as a conceptual starting point and through empirical analysis, we show how this concept is played out through two complex and overlapping dimensions—gatekeeping of conversations and gatekeeping through conversation, that jointly explain how norms and mutual understanding are being built and how conversations are being looked after on newspapers’ Facebook pages.

Empirically, we examine four Finnish newspapers’ Facebook pages and the social interactional relationship between the newspapers and their social media audiences on the third-party platform. Facebook is still the most important social media platform for news, across all markets (Newman et al. 2020). It is also the most used social media platform among Finnish newspapers (Simola 2019) and the social media platform where Finns consume news the most (Newman et al. 2020). Our data consists of 180 Facebook posts and their comments published by the newspapers between November 2018 and February 2019.

Against this backdrop, the paper asks:

RQ1: How is post-publication gatekeeping intertwined with journalist-audience conversations on Finnish newspapers’ Facebook pages?

RQ2: How do journalists and their audiences build mutual understanding on Finnish newspapers’ Facebook pages?

Over the following pages, we explore gatekeeping theory and look at how audiences, algorithmic platforms, and journalism ethics have shaped the digital era of news. Second, we discuss digital conversation analysis as the approach utilised in the current research. Third, we introduce our data and analysis. Fourth, in the findings section, we propose conversational gatekeeping and its two dimensions and provide empirical findings for it. We conclude by discussing possible implications as well as practical and ethical issues that newspapers should consider in their social media activities in light of the proposed concept of conversational gatekeeping.

Gatekeeping in the Digital Era

In the traditional sense, gatekeeping is the journalistic process of selecting, writing, editing, scheduling, and making other decisions concerning news production that have an impact on how reality is presented to the public (Shoemaker, Vos, and Reese 2009, 73). As gatekeeping is one of the foundational theories in journalism studies, its modernisation is pivotal for the field to remain up to date. As Shoemaker and Vos (2009, 130) state: “The challenge is for scholars to think creatively about applying the theory to a changing world and to adapt research methodology that keeps pace.”

The theory has transformed notably since Lewin’s (1947) notions of “gates” in the field of social psychology and White’s (1950) introduction of “Mr. Gates”; a study about a wire editor of a morning newspaper and his editorial gatekeeping decisions. The transformation of the theory has been underscored e.g., by Shoemaker, Vos, and Reese (2009), who provide an overview of the developments from the subjective perspective (White 1950) to the organisational level (Gans 1979), and on the impact of gender (Bleske 1991) and race (Heider 2000) on gatekeeping. Further, technological changes led to a new line of gatekeeping research, as is shown by Berkowitz (1990), who examined the gatekeeping processes of local television news, or Abbott and Brassfield (1989) who compared print and electronic media. In more recent years, gatekeeping has been applied to various kinds of online research, for example into user-generated visibility via media websites (Singer 2014), the rise of news events on social media (Meraz and Papacharissi 2013), social media editors’ impact on news diffusion (Welbers and Opgenhaffen 2018), and visual gatekeeping practices focusing on the roles of non-professionals and professionals (Pantti 2015).

Gatekeeping can also be viewed from the perspective of pre- and post-publication practices, the latter in particular emphasising the audiences’ role in the gatekeeping process. The more traditional gatekeeping research (e.g., White 1950; Abbott and Brassfield 1989) has looked at the processes of gatekeeping before news items enter circulation, i.e., pre-publication practices of gatekeeping. More recently, research (e.g., Bruns 2005, 2018; Singer 2014) has turned to the processes and factors that take place after a news item has been released—in this case, we can talk of post-publication practices of gatekeeping. Hermida (2020) introduced his idea of post-publication gatekeeping, where the interplay of publics, platforms, paraphernalia, and practices in the circulation of news are all part of the factors that shape the news. Our current study makes the same notion as his (2020, 1), namely that when talking about “how issues and topics rise to prominence and gain attention following publication in a digital hybrid media ecosystem,” we address post-publication gatekeeping. In addition, Bro and Wallberg (2015, 95–98) introduce three models of journalistic gatekeeping: first, the process of information (a linear process of information transmission, e.g., White 1950), second, the process of communication (a non-linear communication process where sources, journalists, and audiences interact, e.g., Bruns 2005, 2018), and third, the process of elimination where journalists have lost their role as gatekeepers (news media is no longer needed as the prime intermediary between sources and audiences). The current study adds to the list of research that sees gatekeeping as the process of communication. However, this study focuses on post-publication gatekeeping.

Users’ Rising Role as Gatekeepers

Digitalisation has created new ways for audiences to participate in the hybrid news ecosystem where old and new media meet. Social media has created opportunities for audiences to interact with each other and with journalists. The gatekeeping process in the current open media environment involves more participants than in the old media environment, as platforms give audiences the ability to comment and share what journalists have regarded as newsworthy (e.g., Singer 2014). Or, as Shoemaker and Vos (2009, 124) present it: “ … we must conceptualize readers having their own gate, and they send news items to others in the audience when the interaction between newsworthiness and personal relevance is strong enough.” The users’ role in the gatekeeping process is also emphasised by Shaw (2012, 367) who describes, in the context of political news blogs, the centralized and decentralized mechanisms of gatekeeping as decision processes that are formed together among users online: in decision processes users jointly “establish, negotiate, enforce, and adapt boundaries, norms, and standards that constitute the site.”

We extend the understanding of post-publication practices of gatekeeping conceptually by proposing the concept of conversational gatekeeping: a concept that sees gatekeeping as the process of communication—social interaction—between journalists and their audiences on a respective platform where affordances of the platform interact with conversations that take place post-publication. Namely gatekeeping materialising as and in interaction between audiences and news media. Conversational gatekeeping adds to some previous studies that we categorise into post-publication gatekeeping studies: Bruns’s (2005, 2018) gatewatching, where users (and more recently journalists) keep watch on important and interesting content and publicise it online; Singer’s (2014) secondary gatekeeping, where users can upgrade or downgrade the visibility of a news item online; and to Vos’s (2020) gatebouncing, where already circulating news items can be marked as illegitimate and bounced back through the online gates by journalists and users. And lastly, to Hermida’s (2020) framework where the 4P’s (publics, platforms, paraphernalia and practices) interplay in the circulation of news, in the post-publication environment.

The aforementioned concepts highlight the audiences’ growing role in gatekeeping processes and emphasise the significance of post-publication practices of gatekeeping. However, none of these works takes an explicit focus on conversation and social interaction, which are our main premise. As audience participation in contemporary gatekeeping practices is increasing, more research on communication practices between journalists and audiences from the current time and platforms is needed. Or as Bro and Wallberg (2015, 102–103) formulate when talking about their three models of journalistic gatekeeping (of which the model of communication is one): “ … they can help highlight, inspire and suggest future research into the ways in which gatekeeping has evolved over space and time.” Further, in this paper, our novel concept addresses the call of Reese and Shoemaker (2016, 407): “new media configurations must be identified and their emergence accounted for, even as they may prove elusive and transitory.”

Platformed Gatekeeping and Journalism Ethics

Conversational gatekeeping is a social and communicative process, but still deeply embedded in the technological and professional context in which it takes place. Wallace (2018) models digital gatekeeping as a combination of different kinds of gatekeepers and their mechanisms: journalists, individual amateurs, strategic professionals, and algorithms. Similarly, Hermida (2020) emphasised the technological context by placing platforms as one central element of post-publication gatekeeping. The mediating role of platforms introduces “invisible gatekeepers” to the process of post-publication gatekeeping: third-party platforms’ algorithms have a central yet invisible part in the gatekeeping process. They affect news items’ visibility on the platform (e.g., Zamith 2019; Hermida 2020). Yet, their code is proprietary and undisclosed, and frequently changes, bringing volatility to the platformed news ecosystem. For example, in 2018 the Facebook algorithm was updated to prioritise “meaningful interaction” over organic reach, which essentially lowered the visibility of news media content (Boyd 2019) and urged news organisations to seek ways to foster conversations with their Facebook audiences. What is notable is that not only news, but also the conversations that take place between media and their audiences are technologically mediated by the social media platforms.

Another form of algorithmic gatekeeping is the removal of inappropriate content from the platform, which typically works as a combination of automated moderation and human flagging (Caplan 2018; Gillespie 2018). The competencies of both algorithmic and human moderation have raised public concerns particularly in smaller language contexts (e.g., Mansikka 2019). For example, hateful, discriminatory language is considered unacceptable but it still flows somewhat freely on platforms, especially in small languages, as algorithms struggle to read the cultural and linguistic contexts (Caplan 2018). As a consequence, moderation responsibilities on the textual level often rely on users and their reports (Roberts 2016). This further accentuates the role of audiences on news organisations’ social media sites. Thus, the actions of humans/audiences are increasingly important in defining what content is prominent and favoured—or allowed to stay on platforms in the first place.

Finally, the emergence of the conversational audiences on news media's social media sites also poses new questions of media ethics and responsibility. The public’s expectation towards journalists has not changed: trust and transparency are still expected (e.g., Karlsson, Clerwall, and Nord 2017). Despite the novel digital context, news media are still required to follow the principles of good journalistic practice and ethical guidelines. In Finland, those are enforced by the media’s self-regulation council, the Finnish Council for Mass Media. The council (Finnish Council for Mass Media 2021) has instructed news media that, if needed, essential factual errors need to be corrected without delay on the Internet. Further, the council has made a ruling that when operating on social media sites, news media must follow the same guidelines as on other discussion forums maintained by them, and further, even though comment threads are not editorial material, content created by the audience needs to be monitored and the media has the obligation to remove content that insults human dignity and privacy (Finnish Council for Mass Media 2017). Thus, keeping social media sites of news media clean and accurate is a complex question that brings together gatekeeping, media self-regulation, and media ethics.

While audiences do not have any specific ethical guidelines, they also end up having responsibilities as citizens and audience members on news media’s sites and platforms; through their communication style, they shape the language and content shared on the platforms. The concept of conversational gatekeeping looks at journalists’ and audiences’ social interaction and pinpoints the importance of humans in constructing joint norms and common rules of the news forum. If shared norms are missing, individuals may orient to their own norms. As conversations are one of the main functions of social media (Kavada 2015), the norms and platform ethics are formulated through conversations, often on the respective platform, in interaction shaped by the affordances of the platform. This makes the triadic relationship between platforms, media organisations and audiences intriguing also from the viewpoint of journalism ethics.

Digital Conversation Analysis

Recent work calls for a clearer conceptualisation of “journalism-as-a-conversation,” reflecting current developments in the way news is covered and increasingly becoming democratised (Marchionni 2013). This gives support to more research that investigates actual practices of interactive journalism and news-related conversations online in order to be able to advance theory building. Accordingly, the approach this paper draws on is conversation analysis (CA). Classic CA starts out from the assumption that each turn provides the grounds for the next one, and that each next turn exhibits an understanding of the previous one. This enables participants to monitor and adjust the understanding of their contributions on a turn-by-turn basis and is therefore seen as the central building block for the accomplishment of interaction and of intersubjectivity in particular. Although it has been largely utilised in the study of direct, face-to-face interaction, the malleability of CA allows for its application to all sorts of set-ups (including technology-mediated and online). Giles et al. (2015, 48) even advocate CA as an approach that is “perhaps most equipped to deal with” the specifics of online interaction. Indeed, a growing number of conversation analytic studies look into human interaction in the context of digital technologies (e.g., Moore 2015; Arminen, Licoppe, and Spagnolli 2016), and some have specifically studied (chat-)interaction on Facebook (Meredith and Stokoe 2014; Farina 2018).

While contemporary CA-research is increasingly interested in video- and text-based interaction, the focus on technology-mediation is not all that new. Schegloff’s (1968) classic study on openings of telephone calls is a good example of how technological affordances may surface in interaction. Schegloff (1968) noted that an appropriate understanding of opening sequences in telephone calls must include the ringing of the phone as a first-pair part of a so-called summons-answer adjacency pair. Adjacency pairs constitute sequences of two turns, which are bound to each other by conditional relevance, i.e., the first turn (such as a question or a summons) builds a strong expectation for a certain next (such as an answer). If the second does not occur, it is treated as “officially absent” (Schegloff 1968, 1083) leading to corrective actions, such as repeats of the first item. Similar to face-to-face interaction, where a summons might include some form of an address term, the summons/ringing of the phone fulfils the function of an availability check in phone calls. In other words, in phone calls the technical device itself plays a pivotal role for participants in the organisation of opening sequences. In this paper, we approach newspapers’ opening posts on Facebook as a form of digital summons, designed to invite others to contribute to the topic at hand.

Excessively, talk today takes place in online environments, and some of the traditional concepts and findings of CA cannot be readily applied to what people do when interacting asynchronously on different social platforms. The aforementioned adjacency, for example, can be missing entirely from online talk, which poses questions with regard to interactivity. According to Giles et al. (2015, 48), “discussion thread turns are not, technically, conversational, only conversation-like” (emphasis in the original), because they do not rely on immediacy. In line with such and similar challenges and a genuine interest in the affordances of technology in interaction a new strand of CA, “digital CA,” is beginning to emerge and is increasingly applied to the rigorous study of online talk (see, for example, Giles et al. 2015; Meredith 2017). While digital CA appears similar to CA in a number of details, digital CA explicitly takes the technological embeddedness of talk in various online environments into consideration, i.e., analysis approaches online talk as being tied to the respective social platform in use (see, e.g., Giles et al. 2015). Similarly, in this study we engage in an analysis that follows the principles of CA and at the same time takes into account the data’s specific digital context (such as reliance on written interaction and orientations to public visibility of individual contributions/turns), whenever it becomes observably relevant to the participants.

Data and Analysis

The dataset consists of 180 Facebook posts and their comments made by four Finnish newspapers during November 2018 and February 2019. The posts were produced by journalists from the newspapers as part of a research project: in the second phase of the project (Lauk, Salonen, and Koski 2019) all four newsrooms tested three types of posting strategies and two interactional strategies that were based on the results of the first phase of the project (Lauk, Salonen, and Sormanen 2018), and on previous practices (Mayer et al. 2016). Conversation analysis was used as a basis to create interactional strategies for the newsrooms in the second phase. During the experiments in the second phase, journalists were asked to make notes on whether they moderated (hid or deleted audience comments) or increased interaction by some means on the posts they made. Therefore, if a newspaper’s Facebook post was moderated during the test period, there is additional evidence of it.

The dataset for the project was gathered in the form of screenshots during February 2019. Any excerpts of the material presented in this paper have been translated from Finnish to English by the authors. By way of translation, we are able to better protect the anonymity of the audience members taking part in discussions. For the same reason, the names of the conversationalists have been changed. The newspapers have given their informed consent to the research.

In the first step, we reviewed the 180 Facebook opening posts (“summons”) and their comments, drawing on (theory and data-driven) qualitative content analysis (Silverman 2011; Tracy 2013). The purpose of this initial inspection was to get a sense of the dataset at hand, which enabled us to organise the data into smaller sections and to locate interactional instances that contained gatekeeping or elements of its secondary practices in a post-publication social media environment. As such instances, we treated audience members’ or newspapers’ visible orientations to opening posts and comments that indicated the news content, its presentation, or public reception were treated as either problematic or acceptable. In addition, these post-publication gatekeeping practices needed to have an observable connection with the news item shared on the platform, without exiting Facebook. To grasp a better understanding of what occurs in the light of gatekeeping after the news item is posted, we formed a tabulation (frequency of instances, Silverman 2011, 66–67) of the 180 Facebook posts: 31 posts clearly indicate practices of post-publication gatekeeping. Of these 31 posts, 19 had instances of conversations that handled what is appropriate or wanted content for the forum, and 17 covered misunderstandings, reassurances, and factual errors that were visible through the forum. In addition, we noted three posts that were not addressed in terms of normative conventions of the forum, even though one of the three posts had been moderated by the newspaper. All three posts were news covering immigration. Later we will discuss further the deviant case that is seen to be an essential ingredient in CA: A case that disconfirms the developing analysis, which already recognises the normative status of the rule or pattern under-identification, but at the same demonstrates participants’ normative goals (Edwards 2006).

In a second step, in order to provide a detailed account of these instances in the data set, we narrowed down three exemplary cases for closer inspection that clearly exhibited post-publication gatekeeping practices or a lack of them, and analysed the posts using digital conversation analysis (Giles et al. 2015). This choice allowed us to trace the threads (and their interactional constituents) as they unfold moment-by-moment, and to conduct a fine-grained, genuinely qualitative analysis of participants’ visible orientations to the content and representation of opening posts and comments.

In line with a conversation analytic understanding of the role of single instances for an adequate, empirically-based understanding of social interaction, we argue that the three cases provide comprehensive insight into what is actually possible in terms of (conversational) gatekeeping practices on social media. In other words, we follow Psathas’ (1995, 50) assertion that further cases essentially would not add to the credibility of our analysis and findings, and ultimately of our conceptualisation of conversational gatekeeping.

As we will demonstrate further below, among the pivotal practices that emerged during the analysis were activities that mark previous contributions as problematic in terms of social norms and journalistic facts. We treated these instances as relevant for the study’s purposes because they provide a window into the interactional relationship (including post-publication gatekeeping practices) between audience members as well as between audience members and journalists.

Findings: Conversational Gatekeeping

Content analysis revealed that of 180 posts 31 had signs of post-publication gatekeeping practices that occurred in conversations between audience members and journalists. Further, the empirical findings revealed two kinds of dimensions of gatekeeping that occurred through their social interaction. Of the 31, 19 posts demonstrated characteristics of what we call gatekeeping of conversations, and 17 posts gatekeeping through conversation (see Table 1.). The former dimension means that journalists and audiences can gatekeep their conversations; build their own norms by determining what is appropriate or wanted content or action for the online forum. The latter dimension means journalists and audiences can act as gatekeepers in online settings through conversations; give prominence to misunderstandings, reassurances, and factual errors that are in need of repair, assurance, or correction. These two dimensions are also dynamically interconnected, as gatekeeping of conversations and gatekeeping through conversations can occasionally occur at the same time—an example of this is provided in Post 2 and Extract 3 later in this section. Together the dimensions form the novel concept of conversational gatekeeping (see Table 2).

Table 1. Occurrences of conversational gatekeeping on newspapers’ Facebook posts.

Newspaper	Total number of Facebook posts	Posts with conversational gatekeeping	Posts with gatekeeping of conversations	Posts with gatekeeping through conversation
NP1	60	13	7	8
NP2	68	7	5	3
NP2	25	7	5	3
NP4	27	4	2	3
Total	180	31	19	17

Table 2. Dimensions of conversational gatekeeping.

Conversational gatekeeping
Gatekeeping of conversations	Gatekeeping through conversation
Journalists and audiences gatekeep their conversations and build their own norms by determining what is appropriate or wanted content or action for the online forum. > conversations are the target of gatekeeping.	Journalists and audiences act as gatekeepers in online settings through conversations. By participating, they give prominence to issues (for example misunderstandings, reassurances, and factual errors) that are in need of repair, assurance, or correction. > conversations are the means for gatekeeping

From a conversation analytic perspective, the concept of conversational gatekeeping is intertwined with the phenomena of repair and correction: repair is seen as a mechanism by which interlocutors confront trouble or problems in speaking, hearing, or understanding (Schegloff 2007, 100), and correction is a class of repair that includes an actual error (Meredith and Stokoe 2014, 186). Jefferson’s (1974) study considers small errors in natural talk, occurring both in production of speech and in meaningful interaction. She also introduces two domains of error: “production errors—that is, errors in the production of a coherent utterance, or interactional errors—that is, errors in speaking “appropriately” for the recipients of the talk and for the occasion of the exchange” (Meredith and Stokoe 2014, 186). This notion of error and repair is important in the context of our study, where expectations of adherence to certain social online norms may surface in posts that treat previous contributions as inappropriate or otherwise problematic. This is also encouraged or enforced by the self-regulation practices, as the Finnish Council for Mass Media (2021) suggests that factual errors should be remedied by the news media on social media. In addition, our findings show instances of repairs and errors (corrections) by viewing self- and other-initiated self- and other-repairs (see Schegloff, Jefferson, and Sacks 1977, 364–365; Schegloff 2007, 101–106). Schegloff, Jefferson, and Sacks (1977) talk about self and other repair/correction, depending on who does the repair: the speaker themselves or another participant. They conceptualise “that which the repair addresses as the ‘repairable’ or the ‘trouble source’” (1977, 363). Repair/correction can be either self-initiated or other-initiated; the one who repairs is not automatically the one who initiated the repair operation. As Schegloff, Jefferson, and Sacks (1977, 364) explain, “ … both self-repair and other-repair (and failure as well) can be, and sometimes are, arrived at from either of the (for conversation) exclusive types of repair initiation: SELF-initiation of repair (i.e., by speaker of the trouble source) and OTHER-initiation of repair (i.e., by any party other than speaker of the trouble source).”

Exemplary Cases

In the current study, we consider the first turn—the opening of the conversation and news media’s Facebook gates—as a summons. Furthermore, the interlocutor of the newspaper is unknown because multiple journalists of a given newsroom are able to post and comment via the company's Facebook account. Nevertheless, the newspaper’s actions in a post are considered a unit; no individual journalists are singled out.

Post 1: Our first example looks at a post covering a news article concerning the local church council elections. The post has two main comments made by audience members and one sub-comment by the newspaper. This post exemplifies the dimension of gatekeeping through conversations where errors and misunderstandings are treated as problematic and in need of repair.

Appendix 1 - Posts

Post 1

Summons 1

Extract 1.

Main comment:

Tom (Nov 19, 2018. 9:56):

I wonder where these numbers 25 and 13 come from. There are only 27 seats in the council. The numbers are more likely 17 and 10.

Sub-comment:

Newspaper (Nov 19, 2019. 10:22):

Thank you, Tom! We apologize for the mistake, we will correct these numbers to the story!

In Extract one, an audience member claims that the referenced news article has a factual error. The article is located on the newspaper’s website and, therefore, the trouble source cannot be seen on the post: the problem is made visible through the platform. In this case, by making other-initiated other-repair, the audience member, Tom, is now acting as a gatekeeper through conversation, giving priority to the accuracy of news. Indeed, it is interesting here that Tom does not react to the news itself, but his response to the summons displays an orientation to the correctness of the content before it may become a topic for further conversation. Tom’s actions thus also resemble what Vos (2020, 92) calls gatebouncing: news items in circulation can be bounced back through the gate due to fact-checking outcomes. The newspaper responds to Tom’s comment by thanking him and by promising to make a correction. By doing so, the journalist’s actions resemble acknowledgement of an actual error leading to other-initiated self-repair. However, there is no evidence of the promised correction as the trouble source was located on the newspaper’s website, and, therefore, it is not visible to social media audiences whether the correction was ever made. The newspaper has therefore fulfilled its duty to some extent: admitted the error and promised to correct it, as per the ethical guidelines of the Finnish Council for Mass Media (2017, 2021).

As the factual error is not visible on the post, there is nothing to correct on Facebook. However, by engaging in the conversation the news media observably takes responsibility, and in a way contributes to building trust with its audience via social interaction. The audience member draws on the affordances of the digital environment by using the public page to make visible to others that the numbers in the news article are possibly faulty: through his comment, he is acting as a conversational gatekeeper. By the journalist’s promise to make a correction, it is made visible that the audience member and the journalist are agreeing there has been an error. With the admission and by building mutual understanding the journalist observably acknowledges the audience member as a gatekeeper. This post demonstrates that an audience member can initiate and participate in the gatekeeping process through conversation. Also, by other-initiation of repair, the news media was given the opportunity to correct its error.

In cases where no actual error is made, news media can clarify or repair misunderstandings, as happens in the following post that continues to exemplify the dimension of gatekeeping through conversation (Post 2, thread Y). The latter part of the following post also introduces the dimension of gatekeeping of conversations (Post 2, thread X).

Post 2: Our second example looks at a post concerning the newspaper's news about a plan to install landscape lighting on a local bridge. The post had a total of 40 comments of which most formed into two sub-comment threads (Y and X) that had 25 sub-comments combined.

Post 2

Summons 2

Extract 2 (Comment Thread Y)

Main comment:

Mary (Nov 22, 2018. 17:02):

I believe this is the new bridge on Highway 4, that is passed over when bypassing the centre of Äänekoski. The lighting is beautiful.

Sub-comment 1:

Newspaper (Nov 22, 2018. 17:25):

Äänekoski Bridge is in the centre of Äänekoski, on the road leading to Suolahti. There are also plans for impressive lighting for the new bridge on Highway 4.

(Sub-comments 2, 3, and 4 are not included, they are though explained later)

Sub-comment 5:

Ryan (Nov 22, 2018. 21:05):

So this is the Hiski Bridge?

Sub-comment 6:

Newspaper (Nov 22, 2018. 21:36):

Yes. Its official name is Äänekoski Bridge. The new bridge on Highway 4 on the other hand is called Tärttämäki Bridge.

Sub-comment 7:

Ryan (Nov 22, 2018. 21:47):

So I did still recognise right

(Seven more sub-comments follow)

In the main comment (thread Y), Mary states that the bridge in the post (summons) is the bridge nearby Highway 4. She also responds to the question (“How does it look?”) in the summons: “The lighting is beautiful,” thereby explicitly engaging in conversation with the newspaper. Mary’s comment treats the summons as a trouble source: by stating where she believes the bridge is situated, she implies that the post does not clearly reveal the location, which is therefore in need of a clarification. The first sub-comment is a reply made by the newspaper treating Mary’s comment in turn as a trouble source: “Äänekoski Bridge is in the centre of Äänekoski, on the road leading to Suolahti. There are also plans for impressive lighting for the new bridge on Highway 4” (Äänekoski and Suolahti are small neighbouring towns). In the comment the journalist is making other-initiated self-repair: they are attending to the summons and Mary’s comment with a clarification. The next three sub-comments of the comment thread constitute another topic that we will return to later. The fifth sub-comment is made by Ryan, asking: “So this is the Hiski Bridge?” With this question format, he is also treating the newspaper’s summons (and perhaps the journalist’s comment, see first sub-comment) as a trouble source. The journalist then produces the sixth sub-comment by replying: “Yes. Its official name is Äänekoski Bridge. The new bridge on Highway 4 on the other hand is called Tärttämäki Bridge.” Again, with this comment, the journalist makes other-initiated self-repair, providing additional information that treats the summons as possibly incomplete (and therefore acknowledging the source of the audience’s problems to locate the bridge). With the seventh comment, Ryan accepts the journalist’s explanation by confirming that he indeed recognised the bridge in the picture (in the summons).

Other-initiated self-repair not only gives the news media the opportunity to correct errors (as in Post 1), but it also makes visible if the news media’s actions are treated as problematic. These aforementioned examples point out occasional needs for repairs—clarifications of summons or problematic comments. They also show that through conversation the audience may become a conversational gatekeeper. It could be even argued that the audience members and journalists are “educating” one another as they are working towards achieving mutual understanding.

The next example is another comment thread (X) in Post 2. This thread introduces and demonstrates how journalists and audience members may perform gatekeeping of conversations by creating norms for the forum. By engaging in meta-communicative practices, conversationalists may not only display expectations of certain standards for their interaction, but also maintain them.

Extract 3 (Comment Thread X)

Main comment:

John (Nov 22, 2018. 18.25):

It is good to have landscape lighting in a county where the unemployment rate is 21%!!!

Should something else than “landscape lighting” of millions be done?

I mean really?

Just asking?

Sub-comment 1:

David (Nov 22, 2018. 23.49):

How is unemployment connected to the lighting of the bridge? It looks the same whether you are unemployed or employed. The attitude ‘progress is shit’ is shit.

Sub-comment 2:

Newspaper (Nov 23, 2018. 11:12):

John, let’s keep the commenting at the appropriate level! The opinion will surely come clear without powerful wordings and going to a personal level.

Sub-comment 3:

John (Nov 23, 2018. 11:13):

So, if a troll goes around here giving idiotic comments while living under parents’ roof and says something like ‘shit is shit’ stuff.

Then we are no longer at the “appropriate” level.

Sub-comment 4:

Newspaper (Nov 23, 2018. 11:20):

The use of vulgar words is not appropriate. This goes to all conversationalists on our pages.

Sub-comment 5:

John (Nov 23, 2018. 11:22):

So, why do you only say this to me?!?!?

Say it to all who use them!!

FOR CRYING OUT LOUD!!

Is there a student working there this week?? A trainee number 2?

Sub-comment 6:

David (Nov 23, 2018. 11:43):

John millions of euros? Are you in the business, or how such an estimation? Is the contentedness of the town an extra investment? Can’t wait to hear more?

Sub-comment 7:

John (Nov 23, 2018. 11:45):

Oh sorry.

I didn’t know you live under the bridge. So sorry about that. So just enjoy!

Sub-comment 8:

David (Nov 23, 2018. 11:57):

An illustration with a caption: The unemployment rate of the town of Äänekoski was 12,8 at the end of August.

(Three more sub-comments follow)

Comment thread X is opened by John, an audience member who takes part in conversations in both threads (Y and X) of Post 2. In fact, the three missing sub-comments in Extract 2 were made by John. The content of the comments in thread Y is similar to the ones from thread X presented next. In this comment thread’s opening post John states that the county has an unemployment rate of 21%, and he continues by questioning the priority given to landscape lighting. This prompts a response (in the form of a sub-comment) by another audience member, David, who counters that unemployment rate and lighting are unconnected: “How is unemployment connected to the lighting of the bridge? It looks the same whether you are unemployed or employed. The attitude ‘progress is shit’ is shit.” With this comment David’s actions resemble self-initiated other-repair, treating John’s contribution as unrelated to the summons, and he even dismisses it as “shit.” He thereby establishes what kind of conversation topic or even opinion is appropriate in response to the post; he is initiating gatekeeping of the conversation.

Later on, the journalist submits a comment (sub-comment two) that directly addresses John: “John, let´s keep the commenting on an appropriate level! The opinion will surely become clear without powerful wordings and going to a personal level.” This comment indicates that the journalist is treating something John has said as problematic, as a “social” trouble source. According to the newspaper’s notes which they provided to us, this discussion was moderated, and one comment was removed. Neither we nor other social media users any longer have access to this deletion, which is an interesting feature of this kind of text-based mediated interaction, leaving gaps in an ongoing public conversation that may become visible in seemingly unrelated contributions or—like in this case—disconnected scolding. Considering that some comments are kept, and others deleted, the newspaper has powerful rights (or opportunities) to gatekeep the content, to moderate or even delete it. In this sense, journalists also act as traditional gatekeepers in the light of post-publication gatekeeping.

In the third sub-comment, John expands the newspaper’s comment, not only providing an account for appropriateness in this context, but also treating David as the troll. This prompts another sub-comment (4) by the journalist, which can be seen as self-initiated self-repair: they stress that this etiquette applies to all participants, thereby repairing a possible understanding of singling out John. By initiating other-repair (sub-comment 2) and self-repair (sub-comment 4), the newspaper is creating norms for its forum in terms of how interactants are expected to behave. The comment thread initiated the need for some ground rules on behalf of the newspaper, and by commenting the journalist is engaging in gatekeeping of conversations.

The fifth sub-comment by John shows a clear understanding of being a subject of discrimination. He is drawing on several resources of written online talk to indicate emotions, e.g., capital letters and exaggerated use of punctuation marks (Farina 2018). By this way of expression, John shows that he is not accepting the attempted repair by the journalist. Indeed, the journalist’s actions lead to more problematic talk. This seems to indicate that when news media performs conversational gatekeeping, directly blaming individuals might be detrimental to the course of the interaction since this could give the conversationalist more reason to counter. In the subsequent six comments, David and John continue the discussion over the unemployment rate. In sub-comment 8, David provides an illustration that shows that the unemployment rate according to officials was 12.8 per cent. By doing so, David returns to John’s main comment performing self-initiated other-repair by correcting a factual error, i.e., he is gatekeeping through conversation.

In this comment thread, the newspaper and its audience were jointly gatekeeping the news content in conversations and repairing the ways the interaction was unfolding between the conversationalists, i.e., in terms of appropriateness. These joint gatekeeping practices made visible (social) trouble sources, and with other-repair, the newspaper and the audience were creating and reinforcing social norms for their forum. Extract 3 also demonstrates that the two dimensions of gatekeeping can overlap. In the next example, we illustrate how the audience is starting to develop norms for themselves as journalists and the so-called invisible gatekeepers are “missing.”

Post 3: Our third example looks at a newspaper’s post covering an armed robbery. The post has a total of 23 comments, 21 main comments, and 2 sub-comments. We consider this instance as a deviant case, i.e., a case that seems to disconfirm the developing analysis that has already formed an initial understanding of a normative pattern, but at the same time it exhibits features that may yet support previous analysis (Edwards 2006). We have shown earlier, in Posts 1 and 2, how journalists and audiences interactively do gatekeeping through self- and other-repair—we regard the phenomenon that has emerged in our analysis as conversational gatekeeping. However, we have also observed in our data that the audience’s norm building may unfold in a manner that is against this normative pattern. Closer inspection of such instances reveals that conversational gatekeeping is not occurring, nor does the platform, the invisible gatekeeper, notably moderate the flow of the conversation. At the same time, the non-normative behaviour forms its own pattern that unveils the conversationalists’ joint performance and reinforcement of hate speech.

In Post 3 all the responses are affirmative by nature and they can be seen as clear replies to the summons. The conversation that is unfolding in the main comments is concerned with the crime suspect who was presented textually and visually in the summons. However, the focus is on the background of the suspect and all the comments contain slur words. We provide two extracts from the main thread to demonstrate how the use of slur words forms social interactional behaviour that is non-normative and harmful to societal discussion.

Post 3

Summons 3

The face of the person in the image has been blurred for ethical reasons.

Extract 4

Main comment 1:

Bruce (Dec 26, 2018. 10:13)

You’ll probably find him in the nearest refugee center. [Comment had 28 likes.]

Main comment 2:

Steve (Dec 26, 2018. 10:16)

Look at that, that’s Niilo from Utsjoki. [Niilo is a traditional Finnish name for a man and Utsjoki a county in northern parts of Finland, in Lapland. The comment had two likes and two laughing emoticons].

Main comment 3:

Dan (Dec 26, 2018. 10:16)

Were you surprised … ? [The comment had three likes].

Main comment 4:

Miles (Dec 26, 2018. 10:18)

A man?

Main comment 5:

George (Dec 26, 2018. 10:23)

Did he speak Savonian or Arabic?

[Comment has been written in Savonian dialect that is used mostly in Eastern parts of Finland, and the comment had 5 likes.]

[Next to the comment is an image of a book cover that says a black-bearded man causes stir. The book cover has also an illustration of a black-bearded man talking on phone and stroking his beard.]

(Five main comments follow)

The first main comment states “You’ll probably find him in the nearest refugee centre.” This answer to the summons (eliciting 28 likes) seems to set the tone for the conversation as all the following comments display the non-normativity that reinforces hate speech—aggressive or hostile talk towards different groups of people e.g., based on ethnic background, gender, or age (Pöyhtäri, Haara, and Raittila 2013, 26). The main comments 2–4 make a joke of the man’s ethnic background, question if the person is a man, and contemplate whether people are really surprised about the case. The main Comment, 5, is a combination of text and image. The comment states “Did he speak Savonian or Arabic?” which refers to the language that the person might speak (Savonian is a Finnish dialect). In the image attached to the comment, there is a book cover that says a black-bearded man causes a stir. Thus, the comment thread displays a hate speech competition of sorts, where elements of satire are picked up and reiterated by the users, which creates a new norm for interaction in this context.

In Extract 5, the first two comments affirm the flow of hate speech: in Comment 11 James uses the demeaning word “blackamoor,” and in Comment 12 Andrew states the man should be sent back to where he is from as soon as he is caught.

Extract 5

Main comment 11:

James (Dec 26, 2018. 11:53)

Christmas peace doesn’t really matter for these blackamoors … well, they don’t respect anything else either. [Comment had two likes.]

Main comment 12:

Andrew (Dec 26, 2018. 12:42)

Back where he came from as soon as he gets caught. [Comment had one like.]

Main comment 13:

Brian (Dec 26, 2018. 12:55)

Points for publishing a clear image for once. It could have been easy to add something else besides the bearded man into the description, but that would probably be considered as discrimination.

(Six main comments and two sub-comments follow)

Even though there is no clear evidence of other- or self-repairs that indicate conversational gatekeeping in Post 3, Brian’s actions resemble prospective repair. In Extract 5, in the main Comment 13, he replies to the summons by stating “Points for publishing a clear image for once. It could have been easy to add something else besides the bearded man into the description, but that would probably be considered as discrimination.” His statement is controversial as he claims that more description could have been given, but that this would have been possibly perceived as discriminatory. In a way, his comment resembles self-initiated self-repair as he projects that his contribution might be recognised as hate speech; he thus does a pre-emptive move, prospective repair, anticipating trouble. However, the audience does not treat his comment as a trouble source, but as socially acceptable. The use of slur words and hate speech continues in the following comments. Brian’s actions can perhaps be seen to opening the floor for hate speech as a “norm” for conversationalists.

Post 3 has not been moderated by the newspaper according to the notes they provided to us. Nor has the platform, the invisible gatekeeper, deleted comments that clearly are derogatory by nature. In addition, the conversationalists have been fuelled by each other’s comments and allowed hate speech to flow freely—no one attempted to repair the talk. Thus, this post indicates why conversational gatekeeping can be an important element for constructing and reinforcing interactional norms in Facebook commenting. If nothing (neither human actors nor algorithms) interrupts, the non-normative behaviour might flourish freely in discussions related to the news. Algorithms struggle to read the cultural and linguistic contexts and there is also an evident lack of human moderators who could fill this gap on behalf of the platform (Roberts 2016; Caplan 2018; Mansikka 2019). It is therefore often left to journalists and audiences to negotiate what kind of behaviour and conversations are socially acceptable related to news in the light of societal discussion—in other words to employ conversational gatekeeping.

Conclusion

180 posts and their comments on four Finnish newspapers’ Facebook sites were analysed using content analysis and digital conversations analysis to understand how gatekeeping is intertwined with audience participation in post-publication practices and how journalists and their Facebook audiences build mutual understanding. Through our empirical findings we demonstrated that gatekeeping is negotiated in the post-publication environment in social interaction, in conversations among journalists and social media audiences. Together—in conversations—they build mutual understanding and create norms as well as decide on the content and action that is appropriate or wanted in the public news space formed on the respective online platform. We call this concept and interactional phenomenon conversational gatekeeping. As our empirical analysis showed, the concept is formulated through two dimensions—gatekeeping of conversations and gatekeeping through conversation—that are dynamically interconnected but occasionally occur at the same time. The former dimension explains how norms over behaviour and shared content are constructed among journalists and their audiences. The latter dimension explains how journalists and their audiences in conversations seek to repair misunderstandings and errors that occur on the forum. This means that in conversations they together frame the journalistic and ethical “gates” that define what kind of news and discussion over news is considered appropriate on that particular forum.

On a theoretical level, conversational gatekeeping belongs with discussion of post-publication gatekeeping, that is, what occurs when the news item is already in circulation (Hermida 2020). Bro and Wallberg (2015) presented a three-way model of gatekeeping where the second model sees gatekeeping as a non-linear process of communication that is aimed to ensure communication between private citizens, authoritative decision-makers and journalists. Our study contributes to their theory-building by introducing a communicative view of gatekeeping that looks at the communication practices of journalists and their audiences. By adopting this view, we demonstrated the active role and potential power of communicating audiences in gatekeeping processes. Furthermore, we would like to highlight that it is important to recognise these online communication practices as new media configurations that need to be identified and examined (Reese and Shoemaker 2016).

Our empirical findings demonstrate the importance of journalists’ presence on their Facebook pages’ conversations. Algorithms struggle to remove inappropriate content in small languages (Caplan 2018) for which our findings on Post 3 provide support: Extracts 4 and 5 have signs of hate speech that were left unnoticed by both algorithmic and human moderators. Further, the absence of journalists’ actions inadvertently supported the non-normative behaviour of the audiences and, in this case, resulted in a long thread of discriminatory comments. This demonstrates the importance of journalists’ presence in conversations so that they can guide the conversations and take part in the norm building. Further, without journalists’ presence, the traditional gatekeepers are missing from the post-publication environment. Thus, we need human parties—journalists and audience members—to understand the given contexts, to create the norms and boundaries for their forum, and to ensure that discussions on public online spaces are democratic. As Bruns (2018, 355) highlights: algorithmification is not always the solution for everything in the context of news on social media. However, we acknowledge that the platform context is affecting the conversational gatekeeping and its dimension through the logics of the newsfeed algorithm: posts favoured by the algorithm will be shown to more users and they may gain more comments. Therefore, in the future, it would be beneficial to find ways to research the influence of algorithms on the mechanisms of conversational gatekeeping.

Our empirical findings also highlight that norms are created and reinforced—built collectively—in public news spaces as journalists and audiences jointly in conversations negotiate the rules for the respective platform (see Post 2, Extract 3). A previous study by Shaw (2012) also points out how users form norms in decision processes. The study’s applicability to other contexts has been suggested (Wallace 2018, 278) even though Shaw studied only (homogeneous) data from a political blog. Our current study is based on heterogeneous public pages that are administered by news media, which provide several kinds of news topics for audiences to discuss. By introducing the concept of conversational gatekeeping, we also hope to provide a basis for an understanding that is empirically more approachable, and that highlights the interactive aspect of a new kind of gatekeeping also in the light of norm-building.

With respect to methodology, our study contributes to the new wave of digital CA studies (e.g., Meredith and Stokoe 2014; Farina 2018) that are reforming the ways conversation analysis can be applied in the context of social media. In the current study, we drew on CA’s concept of repair to inspect the post-publication gatekeeping practices of news media and their audiences for the very first time. By analysing three posts in detail, we were able to show how conversational gatekeeping is enacted on Facebook. For future studies, we suggest that more cases, involving different social media platforms and international comparison, should be investigated in order to better understand the phenomenon and concept of conversational gatekeeping.

Apart from the theoretical and methodological contributions, our study has practical implications for newsrooms. The way posts, summons, are designed makes a difference to the audience: the topic of the post should be carefully considered by journalists to a given situation so that the gates are not left open to hate speech, while ensuring too that freedom of speech and of the press are not jeopardised. In addition, if audiences request clarifications concerning a particular piece of news, it might be beneficial for newsrooms to edit the posts and through conversational practices to build mutual understanding with their online audiences (see Post 2, Extract 2). When journalists engage in conversational gatekeeping, it might be advisable not to directly blame individuals for their “bad manners” as this could provoke more negative talk (see Post 2, Extract 3). Furthermore, essential factual errors should be corrected (see Post 1, Extract 1) as the Finnish Council for Mass Media (2021) advises so that the news remains accurate.

Finally, we raise ethical questions concerning news media’s social media activities and look at the future of the concept. When considering content moderation on third-party platforms, such as Facebook, where should the line be drawn by moderation—how much gatekeeping is necessary? Conversational gatekeeping could bring us more answers to the question if more research and larger data sets were applied. The concept could help researchers and media organisations to better understand audiences’ viewpoints and which kinds of settings and topics need more guidance on behalf of media and platform companies, especially in smaller languages and different cultural contexts. Indeed, future research is needed to further develop the concept and it would be interesting to discern different styles of conversational gatekeeping and how they unfold. Further, as platforms continue to triumph in contemporary society, we need to better evaluate arising new ethical issues brought by digitalisation: all human members of the platform society—audiences and journalists in the frontline—have a central position to construct ethical gates for the digital spaces.

Acknowledgements

The first version of this paper was presented at the Future of Journalism conference 2019, Cardiff. People have been generous and donated their time to formulate this work ever since. We would like to thank the research group of Jyväskylä University’s LUOTSIVA project led by Epp Lauk. In addition, we are indebted to Tim Vos, Avery Holton, Minna Koivula, Lauri Haapanen, Anna Rantasila, Mikko Villi, Pasi Ikonen, and Ville Manninen for their insightful comments and help. Lastly, we would like to express our gratitude to the anonymous reviewers who helped us to make this paper better.

Disclosure Statement

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

Additional information

Funding

This article is a part of a dissertation project that has been funded by the C.V. Akerlund Media Foundation, the Ellen and Artturi Nyyssönen Foundation, the Media Industry Research Foundation of Finland, the Finnish Cultural Foundation, and the Department of Language and Communication Studies (Univeristy of Jyväskylä).