TikTok's For You Page (FYP), an algorithmically curated starter page with an endless stream of videos determined interesting for the user based on an opaque set of variables, is of extraordinary importance for its users (Bhandari and Bimo, 2022, p. 5). Bhandari and Bimo (2022) aptly note that the foregrounded position of the algorithm on the platform takes away some of the classical “social” aspects of social media platforms. As TikTok users do not have to follow each other to engage with content, the algorithm constitutes the platform itself. View counts and viewership are less dominated by the classic social media metrics of reach by followership but are driven by virality or the content's ability to be spreadable and to spark sustained interest (Jenkins et al., 2013; Guinaudeau et al., 2022). This focus on immersive and prolonged interaction can lead to problematic use (Montag and Hegelich, 2020; Montag et al., 2021). In research on uses and gratifications of TikTok use entertainment, affective needs, escapism and self-expression emerge as the most relevant drivers of TikTok usage (Bucknell Bossen and Kottasz, 2020; Omar and Dequan, 2020; Shao and Lee, 2020). However, research has shown that the platform is also used for political communication (Medina Serrano et al., 2020) and hate speech (Weimann and Masri, 2021).

While users seem to mostly enjoy the content offered to them by the algorithm (Bhandari and Bimo, 2022), it also makes the platform prone to censorship and other types of content-regulative intervention. The use of specific words in comments can trigger ambiguous keyword lists and may lead to blocking by TikTok. As a recent German investigative report (https://www.tagesschau.de/investigativ/ndr/tik-tok-begriffe-101.html) confirms, the aforementioned keywords include “pornography” or “crack” but also “gay”, “LGBTQ”, “sex work”, even “gas” or “slavery”, and some phrases related to the Russia–Ukraine War, such as “international law” or “special operation”. Of course, the relatively rigorous censorship practice and uneven distribution of visibility for different issues and communities at the margins are not exclusive to TikTok. Other mainstream video- and image-sharing apps, such as Instagram, employ similar techniques (Fitzsimmons, 2021; Jaramillo-Dent et al., 2022).

Furthermore, social media platforms always foster different forms of digital self-performance but inhibit others (Szulc, 2018) and users form complex ideas of algorithmic interaction, i.e., in the form of “algorithmic gossip” (Bucher, 2017; Bishop, 2019). Thereby, any platform logic is, first and foremost, an economic one, built for the revenue interest of the concerned platform company and its stakeholders, constructed in a deeply capitalist logic (Gillespie, 2013, 2014; Hearn, 2017). User identities operate within the platform affordances, testing and pushing boundaries (van Dijck and Poell, 2013; Hearn, 2017). Over time, users” practices within a platform's affordances shape the platform reflexively. On TikTok's predecessor Musical.ly, for instance, users developed complex sets of performative gestures as their voices took a backseat to lip-syncing playback (Rettberg, 2017). Creative self-performance and economic motivations of reach monetarization and connectivity merge into a “like economy” (Gerlitz and Helmond, 2013), a “remix culture” (Lacasa, 2020) or “imitation publics” (Zulli and Zulli, 2022), in which publication does not mark the end of content but the beginning of an evernew reinterpretation of content fostered by a participatory paradigm with its very own resistances and demands (Neuberger et al., 2019).

Due to TikTok's FYP-centered content curation, the videos are largely detached from the contexts of their creator profiles. Bhandari and Bimo (2022) refer to this feature of the platform as “content without context” (p. 7). However, instead of the creator-centered curation prominent on Instagram or YouTube, TikTok content is contextualized by topic and, most importantly, socially contextualized by the imitation practices and viral trends that are prominent characteristics of the platform. This re-enactment of video snippets and stories and the recontextualization and personalization of proven concepts, such as specific sounds and camera action combinations, speak more for what we might call recontextualized content.

This paper aims to contribute to the growing body of literature exploring the particularities of TikTok. In particular, we offer an in-depth qualitative analysis of the platform-specific representation of the current Russian–Ukraine War. In the following, we discuss the literature on war and conflict reporting and representation. We present TikTok as a timely consequence of the technological progress in war coverage and within the contemporary participatory paradigm of hybrid media systems. Then, we provide a detailed description of the sampling and analysis methods that we used in our research before presenting our analysis and discussion of our study results and findings.

The Russia–Ukraine War is not the first to be extensively covered on social media. The Syrian uprising in 2011 was widely shared and commented on by citizens on the scene (Al-Ghazzi, 2014). The Iraq War of 2003 was considered the first war that was accompanied by an online struggle for sovereignty of interpretation. Traditional news media were considered biased, and war blogs attracted substantial interest (Cammaerts and Carpentier, 2009). This was a turning point because previously, such as in the Gulf Wars or the Vietnam War, journalists had to negotiate complex dependency relationships with the military leadership on the ground to be able to report on the war onsite. These dependencies were increasingly loosened by technological developments such as smaller cameras and storage media, but above all, by the spread of the internet, whose prominence in the midst of a conflict situation was seen for the first time during the Kosovo Conflict of 1999 (e.g., Matheson and Allan, 2009, p. 28).

The development of smartphones and social media platforms further lowered the hurdles of demanded specialist knowledge and skill for the documentation and reporting of conflicts, which soon became multimodal, connective efforts (Bruns and Hanusch, 2017). This technological advancement marks the “era of becoming a witness” (Givoni, 2011, p. 165). The smartphone camera allows citizens to record any event, often from a dangerous setting, and widely spread their recordings to make them available to a broad public and, at times, to make sense of the event and repurpose and adapt its context (Andén-Papadopoulos, 2014; Schankweiler et al., 2018). In today's digitally networked world, witnessing and providing testimony to events such as armed conflicts is no longer exclusively tied to editorial decisions and newsroom resources. While the scholarly work on war and conflict on social media still largely focuses on news media and institutions and their use of platforms (Wall, 2010; Cowart et al., 2016; Parry, 2018; Crilley and Chatterje-Doody, 2020; Dhanesh and Rahman, 2021; McCrow-Young and Mortensen, 2021; Hedling et al., 2022), analyses of user-generated content have become increasingly prevalent (Andén-Papadopoulos, 2009, 2014; Al-Ghazzi, 2014; Makhortykh and Sydorova, 2017).

A shortcoming of digital communication research is that it primarily focuses on Twitter, a development that is often attributed to the data availability and access restriction issues shaping the computational turn of the discipline (Berry, 2011; Burgess and Bruns, 2015; Freelon, 2018; Lazer et al., 2020). However, public organizations engaged in conflict, such as the police, also primarily use Twitter and Facebook in their crisis communication (Jungblut et al., 2022). The general preference for Twitter is also prevalent in activism and war research (Seo, 2014; Bruns and Hanusch, 2017; Manor and Crilley, 2018; Özkula et al., 2022), but studies on, for instance, Instagram, YouTube and thematic forums also exist (Andén-Papadopoulos, 2014; Dinnen, 2016; Crilley and Chatterje-Doody, 2020; Al-Rawi, 2021). However, TikTok research is still scarce (Medina Serrano et al., 2020; Subramanian, 2021; Guinaudeau et al., 2022; Kaye et al., 2022).

The research on the self-representation of US military personnel on Facebook shows the prevalence of “moto photos”, posed, personal pictures in military gear (Silvestri, 2014). While these photos can also serve to boost one's morale, they are often targeted outward, to yet-undeployed comrades (p. 115; see also Andén-Papadopoulos, 2009). Aesthetic shots of artillery hardware alone exist but are less common than personnel informally posing with their guns or “casually engaging with them in some way” (p. 115). Notably, the personnel posing on such photos show emotionless faces and position their military gear centrally, suggesting that “the weapons, rather than the people, are the central foc[i]” (p. 115). This aestheticizing focus is also common in employer-branding campaigns of armed forces and is used in the best interests of the arms industry. In her paper on arms producers' YouTube content, Jackson (2019) notes that arms producers construct a discourse of militarized national security that promotes certain normative citizen ideals and aims at prohibiting citizens' questioning of militarized national security (p. 271).

Recent studies have shown an increased focus on humanitarian and personalized perspectives in news reporting on war (Hellmueller and Zhang, 2019). This points to the important role of affective intent in conflict-related audiovisual content (Papacharissi and Fatima Oliveira, 2012; Pantti, 2013; Bruns and Hanusch, 2017; Papailias, 2019). Displays of feelings and solidarity represent important audiovisual news values in crisis contexts. Chouliaraki (2008) notes that “through their systematic choices of word and image, the media not only expose audiences to the spectacles of distant suffering but also, in so doing, simultaneously expose them to specific dispositions to feel, think, and act toward each instance of suffering” (p. 372, italics in original). Chouliaraki (2008) outlines a hierarchy of misfortune coverage from adventure to ecstatic news and claims that the former offers “maximal distance from the sufferer and no options for action toward the misfortune [one] watches”, whereas the latter allows for “reflexive identification” (p. 378) with the suffering subjects. It is important to note, however, that the particular local contexts of conflicts and the witnessing of these must be taken into consideration to fully grasp “the intentionality behind the use of digital media” (Al-Ghazzi, 2014, p. 449).

Bruns and Hanusch (2017) similarly argue that the aforementioned highly engaged form of coverage “extends the concept of connective witnessing beyond the circulation of purely factual footage” and can culminate “to a point where that affective response becomes newsworthy in its own right” (p. 25). On social media, memes are a popular affective response format. Silvestri (2018) argues that remixing is an essential tool in the rhetorical function of memes that can evoke memories and emotions (p. 4001). TikTok content gains a new dimension of spreadability through networked imitation and memetic reinterpretation of contexts in the form of trends that make particular forms of storytelling more salient than others at a given point in time. We thus view TikTok as a consequential progression in the digitalized user coverage of war and argue that in our analyses of TikTok-mediated conflict, we must place particular emphasis on a particular affordance of the platform: its fostering of creative self-performance.

Because we followed an inductive research approach that one might call “grounded theory lite” (Braun and Clarke, 2006, p. 81), it was important to us to have frequent discussions on the study, which is also the recommended procedure for inducing reflexivity in qualitative research (Mauthner and Doucet, 2003). After we decided to use the final research approach explained above, we drafted a first analytical framework and collected data from students. We then met again to go through all the responses and videos together and conducted a test coding of five videos each. With this test coding, we reworked our analytical framework together. The final version of the analytical framework resulting from our deliberations can be found in Appendix 1. Finally, we split the sample and met again after the coding phase to go through the notes and content together several times for the final analysis.

Our analytical framework was oriented toward qualitative video analysis and emphasized the affordances of TikTok. An obvious focus of video analysis is the action on the screen. However, any comprehensive video analysis must also include an analysis of the camera action: the acts carried out by the camera, such as zooms, framing, what is seen and what is unseen, and the post-production acts, such as effect and filter application and music and sound editing. The importance of these dimensions is recognized by different approaches to video analysis (Moritz, 2014; Moritz and Corsten, 2018). Reichertz (2014) aptly points out the following:

On TikTok, the aforementioned acts of engagement with the viewer, and hence also the intentionality of the content, become salient in the various forms of remixing contexts and the application of filters and sound.

Our analysis focused on two main dimensions: the creator and the content. Video analysis often applies a sort of partiture or sequence protocol, sometimes even on the level of individual frames. In the present study, we used the video as a unit of analysis because TikTok videos tend to be short already. We coded in three subdimensions within the content and user dimensions. We traced the contextuality of the videos, with a specific focus on the uploading user's profile: gender, user interactions in content, user and content metrics and change of profile over time. The narrative construction dimension focused on the action in front of the camera: where, what and who is depicted in the content. We also addressed the intentionality of the content, which we coded last. Finally, the production dimension focused on the camera action (angle, cuts, and sound) and the platform-specific practices of remixing sound and footage (trends).

It is important to note that the aforementioned coding framework, like any partiture of a piece of audiovisual content, carries out only an orientation function as it is inseparable from the videos and is already an interpretative step (Reichertz, 2014, p. 68). We therefore provide many examples in our analysis chapter and supply a list of all the video creator handles included in our sample (Appendix 2). We can provide the videos as MP4 files upon request.

The theme of suffering is conveyed in the TikTok videos in the sample through the use of visuals of the war-torn country that are frequently employed in videos. The most prominent of these visuals are images of shelled buildings or cities and vehicles. The war and its consequences are made visible through images and footage of material objects, but the suffering of civilians or soldiers and human casualties are rarely depicted. Nonetheless, one student (Student 20) wrote that their video choice, which does not demonstrate human affectedness, “shows the consequences properly”, and another student (Student 22) noted their empathetic response to the depiction of the war-torn country. TikTok videos such as 2201 or 2901 are photomontages that were edited to match emotive music, showing before-and-after photographs of cities and buildings in Ukraine. The consequences of the war are thus shown as purely material; in the cases in which people are also shown, they happen to be soldiers, usually presented in full tactical gear, hence remaining uniform and impersonal. Importantly, however, there are moments of humanization, when soldiers are depicted doing something unrelated to fighting, such as joking around while on the highway or dancing. These are the instances that allow for ecstatic responses from the viewers to military personnel. However, ripped clothes or bloody bodies are not shown, seemingly suggesting that what is war-torn is the country, not its people. This is interesting because shocking pictures are usually important for the depiction of conflict and catastrophe. A reason for this could be the creation of a unifying sense of the war-torn nation and its resilient people. It can also be both an intended and unintended consequence of algorithmic curation and censorship by the platform. The creators counter such censorship by using hashtags such as “fake body” or “airsoft”. In this way, creators disguise real war action as fictional, hence circumventing the platform's censorship regime, which flags and removes potentially violent content.

The aforementioned depiction of the material consequences of the Russia–Ukraine War rather than its human casualties transforms the image of the war on TikTok, showing that the war is not really life-threatening but only affects people's livelihoods and material environment. Thus, the actual meaning of the war for the Ukrainian people remains opaque. A type of TikTok content that exemplifies this very well is the military anesthetization video. Here, we often see machinery and weapons in action: shooting missiles or artillery projectiles from somewhere at something, demonstrating a kind of unaffected excitement over “blowing something up”, as in video 2401, which is described as “strange” by Student 24. In one video (2002), two soldiers in a trench are shown being shot at with artillery or mortars and shooting back at their opponents. Yet, together with the engaging music, no urgent sense of danger is made salient, just the excitement of the action conveyed. This is further emphasized by the fact that the soldiers even adjust the camera after a particular close hit. No one seems to be in any immediate danger of dying.

While the viewer is mostly distanced from the deadly realities of the war, content creators often practice a sort of emotional response cuing to elicit an appropriate emotional response from their audience. They achieve this by placing themselves in front of the content they adopted, either in the sequence or the frame, and showing the audience their emotional state, perhaps with the intention to create emotionally contagious effects (e.g., Coplan, 2006). These types of videos can be said to belong to the popular genre of reaction videos (e.g., McDaniel, 2021). TikTok fosters the reaction video style by means of the “stich” function and the overall lowered hurdles to content editing. In this sense, smartphones, more specifically social media apps such as TikTok, allow users not only to cover armed conflict and share their related experiences (or allow other users to witness this firsthand testimony) but also to share their embodied response to the suffering. We find that the consistent advancing of collapsing contexts eventually amounts to a recontextualization of embodied experiences on TikTok, where the “embodied performance of eyewitnessing” (Andén-Papadopoulos, 2014, p. 760) mediated through the smartphone camera action is furthered by the simplicity and affordance of remixing images, sound and movement. At this instance, creators function as mediators, bridging the emotional distance between their audience and the war and allowing ecstatic coverage (Chouliaraki, 2008) within the platform affordances and censorship regime. In other words, the lack of graphic or drastic images and explicit stories is compensated by emotional response cuing.

The practice of remixing (Lacasa, 2020) shapes the reimagination of the Russia–Ukraine War on two levels. Firstly, creators employ visual editing practices to adopt television and other social media footage related to the war, which they recontextualize to their profiles. It is this practice that follows and facilitates the collapse of contexts (Marwick and Boyd, 2011) on the platform. Users imagine their audiences and adopt and adapt content of other creators to fit their personal branding. On the one hand this recontextualization is facilitated and constrained by the platform affordances. It seems trivial yet is important to realize that stitching or remixing as a platform feature has become industry standard (see YouTube shorts or Instagram remix). It is a defining feature for short-video content production and users perform their remixing practices within this affordance. In other words: Remixing has been written into the social media logic (van Dijck and Poell, 2013).

On the other hand, users perform these reimagination practices within the boundaries of their profiles' commodified self-performance. As Szulc (2018) notes, users' profiles are to a large extent defined by the abundance created through constant updating. The core self of a user, that is the profile created upon signing up (name, gender etc.), retreats behind the datafication logic of platforms (van Dijck and Poell, 2013); on TikTok with its strong focus on the algorithmically curated For You Page in particular. The profile core self on TikTok is so minimized that the self is performed almost exclusively in content. The remixing of content thus embeds it not within a static context of a profile but within the continued performance of the self.

Secondly, sound remixing facilitated by the platform-defining feature of trending sounds is widely used. Zulli and Zulli (2022) write that “[TikTok] networks form through processes of imitation and replication, not interpersonal connections, expressions of sentiment, or lived experiences” (1873). While we agree with the idea that interpersonal connection, such as through followership or community-building practices, is not as relevant on the platform as on others, we found that it is precisely the embodiment of expressions of sentiment that, facilitated by the use of sounds and sound editing practices, creates what we call affective audio networks. As such, the sounds used match not only a single creator's visual content through rhythm and motif but also that of others. This has two functions: to elicit appropriate emotional responses from the viewers and to connect the content to other trending creations with the same motif and similar contexts. While the first function thus serves as an orientation for the audience within the creator's self-performance, the second function orients the creator's self within the platform. Papacharissi (2011) aptly wrote that in the “networked and remixed sociabilities” of SNSs “a sense of place is formed in response to the particular sense of self, or in response to the identity performance constructed upon that place” (p. 317). In that sense, trending sounds can be seen as a signpost to aid that sense of place. As Ramati and Abeliovich (2022) point out, voices can thus, as an original sound, become and remain an integral part of a network even when the bearers of the voices themselves have long since gone.