Abstract
This study looks at how mourning is expressed using the hashtag #grief on the social media app TikTok using qualitative content analysis. In a dataset of 100 TikTok videos, this article explores how the TikTok ranking algorithms, which orders content based on previous user engagements, may connect people in mourning across the platform and how these platform-enabled interactions may shape grief expressions. The study shows how grief was narrated on TikTok, which sociotechnical templates (such as duets, stitches, and audios) were incorporated into such expressions, and how these expressions of grief challenged societal mourning norms. This article ends with a discussion about how different subcultural norms on TikTok are linked to the way in which ranking algorithms create social connections across the platform. This study proposes that the “algorithmic closeness” of TikTok users in grief allows them to challenge societal mourning norms in imagined safe spaces, shaped by the algorithmic ranking systems on the platform.
Introduction
All aspects of human existence partake in digital spaces, including our mourning practices. Society has become what Couldry and Hepp (2017) have called “deeply mediated” (p. 7), in that the way we live and make sense of the world is highly entangled with the infrastructures present in the media we use for communication. Even the way we deal with, share, and understand grief is deeply mediated by the online spaces we inhabit in our daily lives, and socially mediated spaces condition digital grieving practices in different ways. Social media can constitute “safe spaces” for individuals in grief, where they challenge societal norms of mourning. For example, when sharing photos of dead children in closed Facebook groups, which is one such practice, grieving parents of deceased children are judged by the majority culture outside of such digital grieving spaces (Refslund Christensen et al., 2017). However, images and videos of the deceased also become incorporated into the everyday content posted by individuals in mourning, where the deceased has a continuing co-presence on social media platforms (Leaver, 2019). Individuals in grief may even continue to interact with the digital remains of the deceased as a continuation of their emotional bond (Hård af Segerstad et al., 2020).
But not only do algorithmic ranking systems on social media platforms create new ways of ordering content, and change the ways individuals relate to these platforms, the inner workings of algorithms also affect subcommunities (Abidin, 2021b). Even how digital mourning is practiced on social media platforms is contingent on algorithmic ranking systems, for example, quantifying measures on social media shapes the ways individuals in mourning feel supported (Lagerkvist, 2019). But how algorithms specifically structure the sociality of digital grief communities is still unclear.
This article attempts to continue this research on digital mourning by explicitly focusing on how the sociotechnical features of social media platforms shape grief expressions and, possibly, work to create spaces for mourning in new ways. Through this aim, this article attempts to fill the gap in the research concerning the role of algorithmic ranking systems for digital grief expressions in what I call “algorithmic closeness,” meaning the ways algorithms shape how individuals express grief in direct and emotive ways on social media platforms, which are inherently public. To do so, this study focuses specifically on the hashtag #grief on the social media app TikTok, sometimes referred to as GriefTok. TikTok has been studied, for example, in relation to its infrastructure (Zhang, 2020), attention economies (Abidin, 2021a), digital dance cultures (Boffone, 2021), political communication and far right-extremism (Medina Serrano et al., 2020; Weimann & Masri, 2020), and the spread of public messages (Basch et al., 2020), and female celebrity culture (Kennedy, 2020) during the COVID-19 pandemic. However, this platform is still relatively new and, as such, underexplored. This research attempts to fill the current gap in understanding how existential issues are expressed on TikTok and specifically how its communities are shaped by its infrastructural properties.
Through a content analysis (Lutz & Collins, 1993) of 100 videos on TikTok using the hashtag #grief, along with a digital ethnography (Pink et al., 2016) of the GriefTok space, I attempt to understand the platform vernaculars of grief. The term platform vernaculars is defined as the “shared (but not static) conventions and grammars of communication, which emerge from the ongoing interactions between platforms and users” (Gibbs et al., 2015, p. 257). I analyze how grief is expressed on TikTok, the sociotechnical features present in these expressions (i.e., the uses of platform affordances of TikTok), and how this expression may extend and/or diverge from societal norms of mourning.
This article sets out to answer the following research questions:
Q1. How do TikTok users narrate their expressions of grief?
Q2. Which sociotechnical templates are incorporated into grief expressions?
Q3. How do grief expressions on TikTok challenge societal mourning norms?
In the sections that follow, I will present previous studies on grief communities and grief expression in digital spaces. This will be followed by a section detailing TikTok as a platform, including its algorithmic ranking systems, content moderation, community guidelines, and sociotechnical features. The data collection, research design, and ethical considerations will then be presented. The analysis of the platform vernaculars present on GriefTok is split into three sections, first, how grief is narrated and by whom, second, how sociotechnical features are incorporated into these expressions, and third, how these expressions may extend and/or diverge from societal norms of mourning. In the discussion, I propose an analytical framework for understanding the role of algorithmic ranking in the creation of grief communities online, defined as “algorithmic closeness,” and provide directions for future research.
Grief on Social Media
Individuals come together on social media platforms, such as Facebook, to collectively mourn their deceased loved ones. It can create meaningful interactions and become an essential part of mourning. In these existential communities, individuals may find that by communicating their bereavement and showing care and affection for one another, they can create spaces for commemoration and bereavement rituals and experience togetherness in their loss (Haverinen, 2015). According to Breen and O’Connor (2007), there is a fundamental paradox inherent in how grief is societally understood, primarily based on classic grief theories, where grief is expected to “pass” through linear stages, culminating in an experience of detachment from the deceased. Instead of aiding the experiences of grief, these ideas have furthered a mounting trend of medicalizing grief. Societal myths of grief have entailed that individuals should not be vocal about those they have lost, or about their grief in general. For example, these societal myths specify who, when, how, and where individuals should grieve, and especially for how long (Doka, 1989). When individuals express grief, these expressions are continuously complicated by these overarching grief myths. The ways individuals in mourning receive social support in times of grief are also hindered by such societal norms of grief, where previous research has focused on the importance of compassionate communities in bereavement care (Logan et al., 2017).
While social media may provide new spaces for finding social support in mourning, both the expressions of grief and how others react to these expressions are still merely reconfigurations of traditional societal norms on mourning, such as perceived privacy, modest expressions, and appropriate ways of remembrance (Wagner, 2018). However, the ways these impressions can spread, and potentially become viral as these are shared on social media, can have ethical implications, become subject of criticism. For example, Abidin (2019) has explored what she calls the “digital grief etiquette,” which consists of the hidden rules behind how and for whom to grieve. Indeed, social media users are sometimes subjected to a form of “grief policing” when expressing grief online, especially when grieving the death of celebrities (Gach et al., 2017).
Indeed, social media users adapt to the different styles, grammars, and logics of specific platforms in what Gibbs et al. (2015) have called the “platform vernacular.” Their study of the posting of funeral images on Instagram found that these photos are used to communicate presence for their loved ones and that these contemporary funerals can become social experiences. In these ways, the deceased’s co-presence with their loved ones is contextualized and visualized in different ways. The individual posters may find themselves shaping both the deceased’s identity and their own on social media platforms. How the deceased become co-creators of content in online spaces, even after their death, can turn them into liminal figures, unable to actively shape their online presence (Leaver, 2019). In this way, their presence can shape the public identity of social media users themselves. This form of co-creating social identity via the memory of the deceased relates back to how Ahmed (2014) theorized about the continuous connections between survivors and the deceased in their impressions: One can let go of another as an outsider, but maintain one’s attachments, by keeping alive one’s impressions of the lost other. This does not mean that the “impressions” stand in for the other, as a false and deadly substitute. And nor do such “impressions” have to stay the same. Although the other may not be alive to create new impressions, the impressions move as I move: the new slant provided by a conversation, when I hear something I did not know; the flickering of an image through the passage of time, as an image that is both your image, and my image of you. To grieve for others is to keep their impressions alive in the midst of their death.
These impressions can be used for shaping the presence of deceased persons in digital spaces. In this way, social media users can use these platforms to keep the deceased alive through their continuous impressions. This might respond to the psychological model of mourning proposed by Klass et al. (1996), who critiqued the idea of grief as linear, where the grieving individuals are expected to move on and “accept” the death of a loved one. Rather, relationships with the deceased should be redefined as continuous, according to this model, where individuals in mourning should expect the emotional bonds continue throughout their lives. As social media has become ever more central in individuals’ lives, so has their postmortem relationships. These forms of non-linear theories of grief have concluded that “one-size-fits-all” approaches to grief are unjustifiable for individual grief experiences (Hall, 2014). Instead, how grief is continuous has become emphasized in more modern bereavement theories, especially concerning the experience of continuing bonds. This can even mean that individuals in mourning continuously interact with the digital remains of the deceased, such as their previously used digital devices (Hård af Segerstad et al., 2020). In this practice of continuing relationships with, and impressions of, the deceased other, the deceased are very much present in digital spaces.
However, a model-based understanding of grief is still very much present as the norm in society, where grief is ultimately overcome to return to everyday life. Because of this, grieving communities online have functioned as spaces for individuals to actively create new and more inclusive boundaries of grief, which function as a form of de-tabooization of grief. These digital spaces can become important for expressing grief in unconventional ways, especially in closed groups that users may perceive as “safe spaces.” Specifically, bereaved parents have used such spaces to challenge the societal taboos that encircle grieving parents’ mourning norms (Refslund Christensen et al., 2017). Allowing others to partake in personal expressions of grief, and releasing some of the taboos surrounding these expressions, may not only create spaces, but brave ones, where speaking honestly about grief, can become a “collective aspiration to repair the world as well as ourselves” (Milstein, 2017, p. 9).
However, grief is culturally specific, and expressions of grief differ worldwide. Grieving rituals are culturally bound and can have deep roots in specific regional cultural practices, such as the Mexican Día de los Muertos (Wojcik & Dobler, 2017). In this current study, the use of an English language hashtag (#grief) infers a general emphasis on Anglo-Saxon communities. Nonetheless, TikTok users from culturally and linguistically diverse communities may use English hashtags to gain visibility. The different traditions present in grief expressions could, for this reason, shape the study.
While finding a community and sharing experiences in grief may positively affect individuals’ mourning practices, the quantifying measures present on social media platforms can have an essential role in the experienced sense of support for bereaved individuals. Each like, comment or share of the bereaved person’s story of the deceased can be understood as a direct measurement of the person lost. Because of this, decreased interaction with such accounts may signify indifference toward the deceased themselves (Lagerkvist, 2019). In this article, I suggest that TikTok users become algorithmically linked together within their subcommunity as a form of closeness to other grieving individuals. This closeness to others on TikTok, despite it being an open and public platform, makes an interesting case for understanding how sociotechnical features of social media platforms may affect mourning practices. But first, we need to take a closer look into the inner workings of the TikTok algorithms, and how users of the platform understand them.
TikTok and the Algorithmic Imaginary
TikTok is one of the most downloaded apps worldwide, according to Forbes (Bellan, 2020), and while it has withheld its status as a “childrens’ app,” this has changed over time, with an increasing number of adult users (Mohsin, 2021). TikTok is one of the most widely used social media platforms, with 689 million daily users and, as such, TikTok is surpassing the number of active users of Twitter, Pinterest, and Snapchat (DataReportal, 2021). TikTok is owned by the company ByteDance, which launched a Chinese lip-syncing video app called Douyin before launching TikTok to non-Chinese markets in 2017 (Elson Anderson, 2020). However, the company’s Chinese origin has entailed some political turmoil in relation to the highly controversial proposal of a ban of the platform in the United States (BBC, 2020).
The ranking and recommendation algorithms structure the content on the app’s home page called the “For You”-page. These suggestions are based on user engagements rather than social networks and are built into the platform’s structure. When setting up an account users are encouraged to choose the content genres that most appeal to them rather than follow their friends from other network sites, the latter of which is a common feature of many other social media apps (Zulli & Zulli, 2020). Instead, users become connected with different communities on the platform through their engagements, for example, by liking specific content or following different accounts. However, the algorithms are also sensitive to user movements on the platform, for example, whether or not a user watches a video from a specific genre from beginning to end (TikTok, 2020).
However, in attempting to specify what “the algorithm” present on digital platforms does, we are usually speaking about the process performed by several algorithms, of which most are “distributed, probabilistic, secret, continuously upgraded, and corporately produced” (Seaver, 2017, p. 3). Since the purpose of ranking algorithms on online platforms is to “hook” its users to stay and engage with the platform, Seaver (2019) has suggested that they should be understood as “traps,” in an anthropological sense. Such a conceptualization encapsulates some of the ethical, cultural, and technological concerns with algorithms.
It is difficult to gauge exactly how the machine learning of TikTok algorithms is constituted. The human and cultural elements present in the decision-making at the creation of these algorithms are essential for understanding their functionality but are often obscured. Instead of seeing these entities as abstract technical achievements, we “must unpack the warm human and institutional choices that lie behind these cold mechanisms” (Gillespie, 2014, p. 16). However, it is not always clear which decisions have gone into their creation, and companies are often unwilling to disclose these choices. By not disclosing such factors present in the algorithmic ranking on the platform, it is difficult to ascertain how algorithms relate to cultural expressions and norms. However, in an internal moderation document, leaked in the online magazine The Intercept, TikTok proposed that users face “algorithmic punishments for unattractive and impoverished users” (Biddle et al., 2020), showing that cultural norms of both attraction and wealth had gone into its ranking systems. As such, we need to understand how “algorithms are multiple, like culture, because they are culture” (Seaver, 2017, p. 5, cursive in original). Which users, then, gain visibility on the platform is dependent on such cultural dispositions. In this study, the videos chosen for the analysis were the most viewed videos using the hashtag #grief, which means that these videos have gained visibility. They may, as such, adhere to these existing cultural norms.
Content Moderation and Shadowbans
While TikTok’s (2021) community guidelines prohibit users from uploading dangerous and exploitative content, 1 it has had ongoing problems with harmful content on the platform, such as sexually exploitative images of children (Cox, 2018). Studies have shown that the content it attempts to prohibit is in fact present on the platform (Weimann & Masri, 2020). Their attempt to suppress specific content actively bans both hashtags and words in video descriptions, as well as text reels in the videos. For this reason, TikTok users who post information about suicide prevention, for example, are given the option of using the hashtag #suicideawareness instead of #suicide. Users may mention harmful words, such as “suicide,” both orally and in their written captions to help hearing-impaired users, either manually or through automatic transcriptions of speech to text on the app, which is still a relatively new function (Perez, 2021). However, to avoid becoming “shadowbanned,” a form of content moderation on the platform, users will instead add, for example, numbers or symbols to avoid moderation. Because of this, when users talk about phenomena deemed to be harmful on the platform, such as suicide, they may instead write “su1cide,” to avoid automatically being banned when posting their content. This tactic has also been shown in content posted by extremists on TikTok to avoid moderation on the platform (Weimann & Masri, 2020). Both the hashtags #death and #depression have been shadowbanned on the platform due to their potential association with harmful content. However, as this article will show, death is highly present on the platform under the #grief hashtag, and as such, this form of content moderation may be ineffective in suppressing specific content while still compromising its potential visibility on the platform.
Sociotechnical Features of Digital Community Engagements
The norms and practices present on social media platforms are dependent on the technological and social structures created on the platforms themselves. These sociotechnical engagements of social media communication have been understood as “affordances,” a term first coined by ecological psychologist Gibson (2015) to describe the relationship between animals and their environmental conditions. However, he also focused on how the relationships offered specific conditions. Media and communications scholars have been interested in affordances at either of these levels, in what Bucher and Helmond (2018) have called high- and low-level affordances. High-level affordances relate to the dynamics provided by technical devices, while low-level affordances have been understood as, simply, the technical features present on platforms.
The technological conditions of social media platforms play an essential role in how individuals use them. However, the way in which these technological features are used and interpreted by different users on the platform is dependent on specific social norms and community practices, which, in turn, becomes rewarded with increased engagements. Because of this, the algorithmic ranking provides more than a personalized feed, it provides feedback and community entanglements on the platform. Gillespie (2014) has emphasized how ranking algorithms enable individuals to connect with calculated publics. Therefore, algorithms can “shape a public’s sense of itself” (Gillespie, 2014, p. 3). On TikTok, these publics are formed based on genre-specific content and through the use of memes, as a shared ritual of both imitation and replication, which Zulli and Zulli (2020) have called imitation publics. These specific genres are sometimes referred to by users as the different “sides of TikTok.” For example, users may ask their viewers to specify on “which side of TikTok” they are currently featured or, in the case of grief, welcome viewers to the grief “side of TikTok.” As such, grief is one of a myriad of niche content available on the platform. Wall Street Journal (2021) recently tested the TikTok algorithms by creating 100 bots pre-designed to pause on content using specific niche hashtags to test how it learns user preferences. The study found that the bot that was programmed to pause longer on videos concerning sadness and depression suggested 93% of similar content on the bot’s For You-page. This indicates, both, how quickly the algorithm corresponds with your platform engagements, and how much other content is filtered out in the process. Specifically, when depressed individuals turn to TikTok, these types of rabbit holes can center the user experience on these specific negative feelings.
TikTok is a platform where visibility is key, and the ranking algorithms present on the platform set the stage for possible virality. The TikTok algorithm is a fickle creature and will boost some posts over others, making TikTok visibility an inherent possibility on the platform while still being challenging to attain. However, some users create content following an “algorithmic imaginary,” which refers to the way in which people “imagine, perceive and experience algorithms and what these imaginations make possible” (Bucher, 2017, p. 31). Depending on how the technological features on TikTok are used, individuals may provide themselves with increased visibility, and this is a repetitive practice, based on “observed patterns, and gut feelings to figure out how the algorithm works, how to please the platform to facilitate their visibility, and how to have their popularity grow” (Abidin, 2021a, p. 85). This repetition can be shared as a form of knowledge exchange that Sophie Bishop (2019, p. 2590) has called “algorithmic gossip,” which refers to the “communally and socially informed knowledge about algorithms and algorithmic visibility.” In this sense, gossip fills an essential function for social media creators in spreading knowledge about the algorithmic ranking systems and how users should relate to these. This shows specifically how the algorithms shape social connections and how users themselves understand and relate to the algorithms. As described above, when users engage with the niche content on TikTok, which centers on sadness and depression, these engagements shape their app experience.
Data Collection and Research Design
This research takes on a digital methods approach (Rogers, 2013), taking into account the advantages of digital data and relevant tools for researching collective phenomena (Venturini et al., 2018). A data collection of 100 videos using the hashtag #grief on TikTok was accessed between 22 and 23 February 2021. This hashtag had 276.8 million views on the platform, which, compared with other hashtags containing the word “grief,” was the most visible hashtag in this category. 2 The videos included in this study were algorithmically ordered as the top 100 videos that had sparked the most engagements (i.e., views, likes, shares, and comments) during this specific time window using the hashtag #grief on TikTok. These videos had a range between 2.4 million and 45.5 K likes per video at the time of data collection. The oldest video in the data was posted in January 2020, and the most recent video was uploaded just days before the time of analysis. These videos were saved in a .mp4 format on a local data server and numbered according to their position on the platform at the time of data collection.
This study focuses on the grieving publics on the social media platform TikTok by studying its search results on the app through the hashtag #grief. In a previous study by Rieder et al. (2018) of the ranking “morphologies” of YouTube, the dataset consisted of search results. However, the 100 videos analyzed in this study were search results from the app version of the TikTok platform, specifically on the researcher’s app version, which was installed several months ahead of time for another study. Instead of doing manual searches on an app that had not yet learned my previous engagements on the platform, as a kind of “blank slate” to understand these search results in a vacuum, I did these searches after using the app. Since my personal and professional interests lie in mental health issues and trauma, I may have affected these results to show a skewed sample relating to these issues. However, it was clear when entering this numerical data that the video which had received the most views was ranked as number one, and subsequent videos were placed in descending order throughout the dataset. While both comments and shares might have affected the algorithmic ranking on TikTok, this proved not to be as important as video views (see Table 1).
The Top and Bottom 10 (1–10 and 90–100) of the 100 Top Algorithmically Ranked Videos in the Dataset.
As the total amount of search results under the hashtag #grief was too vast for qualitative content analysis, I decided to sample the 100 top videos, which made up the dataset for this study. Of course, 100 videos only make up a fraction of the videos available under the hashtag #grief on TikTok. However, these videos made up a realistic sample of the total number to qualitatively analyze the data. The videos were manually coded using content analysis, where each film was assigned codes in a number of categories. Content analysis allows for discovering patterns by assigning them to specific codes (Lutz & Collins, 1993). These codes were inductively created to form a set of codes that reflected the material itself and not the presupposed ideas of the researcher (Rose, 2016, p. 87). Codes must be exhaustive, meaning that each aspect of the videos must be covered by a single exclusive category that does not overlap with other categories. In addition, codes must be enlightening, to create an analytically interesting breakdown of the material (Rose, 2016, p. 92).
First, each video’s basic technical features were added to the analysis. This includes any users mentioned and hashtags used in its description, as well as the number of likes, shares, and comments each video had received by the time of data collection. Then, more complex technical features (such as whether the “duet” or “stitch,” voiceover, or the “text-to-speech” function) were noted, as well as the date on which the video was uploaded to the platform. Second, the videos’ visual elements were analyzed, such as the place of recording, the individuals visible in the film, and the incorporation of visual elements (such as images or videos of the deceased). Third, the grief-specific features were coded, including the forms of grief being addressed (such as death, separation, or injury), and the users’ relationship with the person for whom they were grieving (such as a child, parent, or partner). In addition, the meme incorporated into the video was added to the analysis (such as “getting ready with me,” a make-up session, or lip-syncing) and the incorporated song assigned to the video. This coding was a solo operation, meaning that no other coding partners were hired for this project. Instead, I presented my findings at a conference and asked colleagues to review my manuscript and my coding chart to find patterns I might have overlooked myself.
While this analysis is built upon a content analysis of these search results, the more profound traces of audio memes and mimetic contexts of these videos have been analyzed in an ethnographic sense, following the work of Pink et al. (2016). For example, audio memes have been retraced through their linked content on the platform to contextualize their uses in the GriefTok context.
In addition to this, specific user accounts were explored further to understand how the sampled videos relate to their overall content, such as by examining whether follow-up videos or other information might provide additional contexts within which their original content could be understood. For example, one user was continuously “going live” on the platform each morning as a motivation to get out of bed, providing additional purpose to the technological functions of the platform for her ongoing grief practice. As part of this ethnographic work, I joined such live events or simply scrolled through users’ content to find additional information. In one such instance, I was interested in understanding the context of a “last dance” video of a mother and her daughter. Since the TikTok user had turned off the comments section, I attempted to find additional information in videos uploaded to see if it was ill-received by other users, which I found it had been. Because of these follow-up videos, I was able to understand how the grieving communities on TikTok receive, and relate to, different forms of content on the platform.
Ethical Considerations
While these videos had gained a viral form of visibility on the platform at the time of data collection, this does not entail that they should be understood as public. This study had undergone an ethics review as part of a larger project analyzing digital audiovisual content concerning sensitive issues around victimization in the months before data collection commenced. Even though the content on the platform itself is publicly available, these videos were created by people in grief and, as such, represent a vulnerable time in their life. Because of this, they should not be scrutinized without ethical precaution (boyd & Marwick, 2011; Zimmer, 2018).
Only three of the videos featured in this article belong to users with verified accounts, that is, accounts that have been verified with a tick next to their screen names. As such, few of the videos are what Williams et al. (2017) have called “public figure accounts.” Instead of contacting the TikTok users in my sample, I have attempted to anonymize their stories by, for example, rearranging identifiable words and not giving too much information to the reader, as suggested by Markham (2012). However, as emphasized by franzke et al. (2020, p. 8), this ethical practice can never be definitive. Instead, researchers need to be aware that these anonymizing practices can only help to “de-identify,” not perfectly anonymize, data. With this in mind, I have attempted to anonymize these grieving individuals as much as possible to respect their privacy in trying times. However, certain information is kept in the descriptions of videos in this article, such as the relationship between the grieving and the deceased individuals, for example, when explicitly referencing grieving mothers or siblings. However, specific information, such as their appearance, locations, or otherwise, were anonymized.
The Vernaculars of GriefTok
Of the top 100 videos using the hashtag #grief on TikTok, collected between 22 and 23 February 2021, 77 videos were about death and dying, 6 videos related to other forms of loss (for example, separation, injuries, or sickness), 10 videos talked about grief without specifying the cause of grieving, and 7 videos were about video gaming and proved unrelated to grief in the traditional sense. 3 As previously stated, some hashtags have been banned on the platform for being sensitive subjects, such as #death or #depression, but in this material of the neighboring hashtag #grief, it is clear that these subjects are not discouraged in and of themselves.
In the sampled GriefTok videos, there were 385 different hashtags included in the video descriptions, in addition to the hashtag #grief. While most of these hashtags are only found once in the sample, the most commonly used hashtag in combination with #grief in these videos was the hashtag #fyp (39 instances), and variations of this For You-page type hashtags (#foryoupage, #foryou), which are widely used on TikTok in an attempt to get featured on other users’ pages. These hashtagging practices can allow users to increase their visibility on the platform and maximize their searchability (Abidin, 2021a).
Most of the videos about death and dying concerned the death of a child (27 videos), a parent (20 videos), or a spouse or partner (13 videos). Less common, though still frequent, were videos about a deceased sibling, friend, pet, grandparent, celebrity, or where the relationship to the deceased was not disclosed (five or more videos each). Some accounts were fully dedicated to sharing stories of the deceased individuals, or specifically about their experiences of grief. Some accounts were fully dedicated to users sharing stories of the deceased individuals, or specifically about their experiences of grief. For example, some users referred to themselves as “angel mommies,” having “angel babies” or being “widows with children.” In these ways, the practice of mourning becomes the full context for their presence on the platform. In setting up an account, creating content, and connecting with others in their comments sections, they present their grieving selves.
This article focuses specifically on the socialities of grief as expressed using the hashtag #grief. The way individuals in grief find a connection on the platform, how these connections are created through sociotechnical features, and how users break societal grieving norms will be further explored below. Ultimately, this article asks how these social ties between individuals in grief develop a form of “closeness” to other users, allowing for more open conversations and expressions of grief, which I have chosen to call an “algorithmic closeness.” This closeness is enabled by the ranking algorithms where users find a social connection on the platform, and unconventional forms of grief expressions are valued through increased user engagements.
Expressions of Grief on TikTok
I was interested in the way the stories of grief, or the stories they told about their deceased loved ones, were narrated, specifically focusing on how it was filmed and the images incorporated into the videos, as well as how the narration of the mourning was done, that is, in text or in speech. Most videos were shot in selfie-mode (42 videos), meaning that the TikTok user was filming themselves in front of the camera. In other videos, the TikTok user was talking to, or filming, other people without interacting with the viewers (28 videos). Less common, but still frequent, the TikTok user incorporated photos or videos from before the death of their loved ones (23 videos), and in four of those occasions, they were showing such images on a green screen behind them.
Next, I was interested in whether TikTok users were speaking themselves, using text as their main form of communicating, or using a voiceover function. In most cases (57 videos), the user narrated the video in text layered on top of the video. Sometimes, the actual video was not seemingly related to the story told in the text. For example, in a short video of merely 11 s, a woman filmed herself taking a bite of her food, while the text described how she would have been marrying her fiancé at this point if he had not passed away. While the video is casually shot, with the camera close to the bowl of food, everyday lighting in the room, and simple clothes on the TikTok user, the musical soundtrack was a ballad, showing quite a different emotive state than the one on the TikTok user’s face. In this way, the music on TikTok can function as an expressive component in the storytelling of grief.
In other videos, the narration was done by the TikTok user themselves, by speaking directly to the camera (22 videos). Less frequent were the voiceover effect (eight videos), where a mechanical, Siri-like voice, narrates written text aloud. Even less frequent were the use of handwritten text on paper (one video), held in front of the camera by the TikTok user for their audience to read. However, in five instances, no context was given, in either written or oral form. However, some of these videos provided hashtags to describe its context (for example, #miscarriageawareness to indicate a miscarriage having occurred).
Sociotechnological Features in Digital Mourning Practices
Different forms of audio memes were used in these videos, which should be understood as “the driving template and organising principle” of TikTok (Abidin, 2021a, p. 80). While these do provide an outline for users, different communities will shape these structures to incorporate their specific type of content and, as such, the interpretation of audio memes will differ across the platform. Most commonly, the GriefTok videos did contain some form of musical soundtrack. In fact, on merely 14 occasions, no musical soundtrack had been added to the video. However, most songs were used only once each in these videos. The music was generally melodic, soft, and the lyrics often referred to different forms of grief, loss or sadness. In this way, the music was homogeneous with the overall feeling of the video content, and told the story of grief just as much as the TikTok users themselves in text or speech.
Nevertheless, most audio memes used in these videos were not related to grief specifically. For example, the “Wasted potential”-meme on TikTok is a version of the track “GONE, GONE/THANK YOU” by Tyler, The Creator, where the pitch has been slightly altered. The meme incorporates the section of the song where the singer says, “I hate wasted potential, that shit crushes your spirit, it really does, it crushes your soul,” which breaks into a beat, accompanied by the lyrics, “thank you for the love, thank you for the joy.” This audio meme has been well-used on the platform, where users show before and after versions of the same individual, often themselves, before they got the confidence and style they now have. In the sampled material, this audio meme was used by one TikTok user to show how she and her brother grew up into their “queer selves.” Several pictures were featured of the two siblings before ending on, as stated by the accompanying text, their “last pic together.” Templates, such as the ones described, thus represent technical features and social elements that become meaningful only in how they are used on the platform; in this case, deceased individuals are displayed and incorporated into the GriefTok community. In this way, these templates are technical features and social elements that become meaningful only in how they are used on the platform.
Another function of TikTok, which works as a template for videos, allows users to interact with other users publicly by repurposing previously posted content, called the “duet” and the “stitch” functions. The duet function on TikTok allows users to record a new video side-by-side of already posted TikTok videos. This was, for example, done by one TikTok user who dueted their own previously posted video of themselves. The user shot the original video while driving to hospital to say her last goodbye to her father, who was being treated for COVID-19 symptoms. In the dueted video, however, she filmed herself in front of the camera while crying and shaking her head, presumably in disbelief about her father’s passing.
In contrast, the stitch function allows users to “stitch” together (using sewing terminology) their own videos with videos of other users. These original TikTok videos are almost used as conversation starters, where a user will post a video asking a question or providing a framework to repurpose in other users’ stitched videos. One of these questions, as posed by another TikTok user, was “If you’ve had a really close loved one die, what’s the funniest thing that happened at their funeral?,” preceded in the original video by the line “I’ll go first.” In the videos sampled, one TikTok user answered this question by telling a funny story about a mix-up of the music playing at her friend’s funeral. In this way, stitches and duets function as interactive spaces for talking about, reacting to and sharing grief within the GriefTok community.
Live events are another form of interaction on TikTok, albeit much more direct. In these, viewers ask questions in the chat while the TikTok user responds and talks to the viewers. One TikTok user in the sample had been hosting such live events to cope with the mental issues derived from their ongoing grief, as a way of holding herself accountable to her goal of getting out of bed each morning. In this way, the platform’s technological features also create specific spaces not just for interacting and finding connection in grief, but also for coping and relieving it.
Making (and Breaking) Mourning Norms in Digital Spaces
The presence of both graves and funerals in these videos shows the connection of users to these spaces in sharing their grief. Often, these visuals of places of mourning were combined to create a narrative of grief, showing images of individuals before their passing, followed by images from their funerals or gravestones. For example, in one woman’s TikTok video, a widow filmed herself sitting in remembrance on a sofa, a scene which is interrupted by flashes of images from their life together. These images are, in turn, interrupted by the image of her husband’s gravestone. In other videos, these mourning spaces were shown throughout the film. One such video featured the funeral convoy, where an avenue was lined with mourners of the TikTok user’s deceased father. Instances like these were common in this material, where one moment was captured, often combined with text to provide context to the video. These videos focused much more on the act of mourning and a specific moment in time, rather than as part of the narration of a story, as in the case above.
Several videos were in fact heavily focused on active grieving practices. In one such instance, a short clip showed a woman lying in bed crying, holding a teddy bear in her arms, with the following text layered on top, reading “52 hours and 22 minutes after a drunk driver killed my only child.” This form of instantaneity of an ongoing grief practice was common, even though it challenged conventional norms of expressing grief in public, since societal expectations of grief are centered on how grief should be a primarily private and intrinsic process (Neimeyer et al., 2014). How this level of vulnerability was expressed did, nonetheless, become criticized. While the deceased children displayed in many of these videos exist in a liminal space, as previously described by Leaver (2019), children’s digital footprints seem to be under scrutiny when portrayed in what other users deem the “wrong” way. The prevalence of images and videos of deceased children was, quite frankly, considerable. In one instance, a woman was filming herself dancing with her daughter while wearing matching outfits and flower bouquets, with the text layered on top of the video saying this was “for the last time.” This was one of the most liked videos using the hashtag #grief in the videos sampled. This supposes a relative appreciation of this type of content, while, at the same time, the comments section of the video had been purposefully turned off by the user to protect herself from critique. 4 Generally, there seems to be an invisible drawn line regarding what content gets praise and not in these grieving TikTok circles, as a form of digital grief etiquette (Abidin, 2019), which specifies the norms and collective rules for how individuals grieve. However, this type of boundary-work has also been found in previous studies of closed Facebook groups dedicated to grieving (Refslund Christensen et al., 2017). In this way, unconventional grief expressions can be challenged in different digital settings, even when they generally offer a form of safe haven from the societal conventions of grief, where grief is generally considered to be linear, conclusive, and personal.
Discussion: Algorithmic Closeness
Given ranking algorithms on TikTok connect individuals based on their previous engagements on the platform, digital grieving communities are able to find each other in their mourning. Because these mourning practices are so outspoken and emotionally charged, users’ openness and willingness to share their grief with others are afforded under these conditions. For outsiders looking in, this “side of TikTok” is intense. However, the intensity in grief expressions may fit into the already established codes within the community. In fact, social norms are created within these communities, and the social affordance of these interactions structure sociality, because “how people behave, move, or simply exist in an environment afford important cues as to how others should behave, move or co-exist” (Bucher & Helmond, 2018, p. 9). This might relate to what Abidin (2021b, p. 4) has called silosociality, that is, the intensely communal and localized sociality of subcommunities on TikTok, which “may not be accessible or legible to outsiders.” The silos create different experiences of publicness and privacy and are created and enabled by the overall platform features—such as geo-location access, internet service providers (ISP) settings, and the application programming interface (API) setting—and the unpredictability of TikTok’s ranking algorithms, which structure content according to different communities (Abidin, 2021b). Because of this, strong emotive expressions may be understood as a form of community expression within the GriefTok community as a subcultural norm of connecting on the platform.
As previously stated, algorithms are culture (Seaver, 2017, p. 5). But even more than that, I would like to propose that algorithms also create culture, and specifically, digital community practices which can, and sometimes does, challenge societal norms. However, considering the continuous (and valid) critiques of the so-called online “filter bubbles” phenomenon (see, for example, Bruns, 2019 and Dahlgren, 2021), I do not propose that these communities experience a closed off bubble where algorithmic filtering disallows users to view other forms of content. Instead, these algorithms allow individuals to find similar content to what they have previously engaged with, and find community within these spaces. In the GriefTok community, these members are able to express emotions with their peers in what resembles “safe spaces” of mourning (cf. Refslund Christensen et al., 2017), where unconventional displays of grief are not only allowed, but encouraged as community practice. In this way, even platforms such as TikTok, where content is publicly open, and thus available to everyone and anyone, can provide the sense of safety needed in order for individuals to express grief in direct and emotive ways. As such, not only are grieving practices shared in public in closed groups online, but they are also algorithmically conditioned in these socially mediated spaces.
Given these considerations, I would like to propose that grieving communities on TikTok express a form of algorithmic closeness in their grief, by which I refer to how digital community norms of grief are enabled by the ranking algorithms on the platform. Through the workings of these algorithms, TikTok users are able to find connection within a community of grievers and a sense of safety, where generally, unconventional forms of grief expressions are valued through increased user engagements. While TikTok is used by a much wider public, algorithmic ranking can help individuals find their specific “side of TikTok,” thereby allowing its users a sense of closeness in times of grief. By expressing grief in this way on the TikTok app, users are able to share their deep expressions of grief, which ultimately find their way through the algorithmic ranking to other users who share a similar emotive state. In this way, algorithmic closeness may shape users’ ability to connect in their grief expressions, which in turn may shape the ways in which individuals share grief in unconventional ways in public, on social media platforms.
Given all aspects of our lives have become deeply mediated (Couldry & Hepp, 2017), such platform interactions in times of grief are by no means new. Still, the way these grieving practices may work with, and through, algorithms on social media platforms suggests that there are sociotechnical aspects of grief that need to be explored further. Future research should focus on the expressions of grief in socially mediated spaces and how its algorithmic ranking systems may shape the interactions of grieving communities online.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
