With so many conflicts around the world, violence and war discourses have permeated various larger discourses, such as mass media, institutional, and others, in both verbal and visual forms. In mass media, war journalism, “a mode of practice” and “a litmus test for journalism” (Allan and Zelizer, 2004, p. 4), usually acts as ‘a fourth power’, the role of which is to expose and condemn the abuses of power, inequalities, (counter)propaganda, war crimes and other abuses that arise during any armed conflict. After all, “good journalism can play a mediatory role in society and can contribute to the creation of a ‘mediative space’ that makes it possible for communities to explore conflict in a more collaborative manner” (du Toit, 2019, p. 24).

At the same time, society tends to establish certain rules to limit the display of and exposure to violence and death among its group members, which results in publishing policies in mass media or community policies on social media, for instance. For example, in conflict reporting, publishing images glorifying violence, such as photos of combatants, is typically considered unethical (Windelspecht, 2022). However, in the Editorial Guideline of the BBC, it is written “7.3.42 In the immediate aftermath of an event involving death, suffering or distress, the use of more graphic material is normally justified to provide a reasonable account of the full horror” (BBC, 2024).

The full-scale Russian invasion of Ukraine, which started on 24 February 2022, has already been nicknamed “the most intensively documented in history” (Fidler and Grove, 2023). However, when it comes to media reporting, a “conflict sensitive approach” (du Toit, 2019) is advocated.

The article aims to trace how one particular event involving violent death has been communicated in the mass media reports. I examine how the verbal text interacts with the visual text, thereby guiding or supplementing the interpretation of the photos from the alleged war crime scenes in the town of Izium (Kharkiv region, Ukraine). These photos were posted after the town’s liberation by the Ukrainian army in selected mass media. I strive to explore the questions of how violent death is represented in mass media and to what extent national media “protect” people’s (local) realities from the consequences of a violent conflict.

Humanization/dehumanization are discourse strategies that influence how agency is ascribed to various social actors, which especially in the case of mass media in turn influence audiences’ perceptions and judgements. Dehumanization is the term used to refer to the presentation of the outgroups or their members as less than human to justify harm or violence against them (Bandura et al., 1975; Bandura, 1999; De Ruiter, 2023; Enock and Over, 2023, p. 1). Humanization, on the other hand, refers to the representational process through which individuals or groups are depicted in ways that evoke empathy and justify moral or social support, primarily by conferring recognizably human traits, emotions, and capacities on them (Kirkwood, 2017). In humanizing or dehumanizing individuals or groups, the linguistic and visual representation of social actors is important.

When it comes to violent deaths, mass media often attributes agency to the deceased, and thus they tend to “retain some sort of posthumous personhood” (Meese et al., 2015, p. 410). The coverage of violent mass civic death has been addressed in several publications (Hanusch, 2010; Morse, 2014, 2017). Hanusch (2008, p. 304) observes that the Western media’s approach to the representation of death is not uniform. At the same time, there is empirical data suggesting that while death itself is a popular subject in the news reporting, its visual representation is often “sanitized” (Hanusch, 2008, pp. 304, 314; Morse, 2014, pp. 109–110). When it comes to reporting of violent deaths, unlike the living, “victims [of violent death. – S. Sh.] never chose to become the subject of a public spectacle” (Morse, 2014, p. 100), and thus the agency ascribed via multimodal presentation of the victims of violent death could be of potential scholarly interest.

To analyze how social actors are represented linguistically and visually in various contexts, van Leeuwen (1993, 2005, 2007, 2008) offers a sociosemantic and sociosemiotic framework, focusing on the meanings and functions behind representational choices. In mass media discourse, social actors are typically “entextualized” through practices that may be habitual or intentional (Gruber, 2024, p. 6) rather than neutral, as is the case of the representation of the deceased (see examples in Hanusch, 2008; Morse, 2014). Below, I offer an overview of van Leeuwen’s model.

Van Leeuwen (2007) points out that in discourse the actors may be excluded or included linguistically or visually. “Representations include or exclude social actors to suit [the addresser’s] interests and purposes in relation to the readers for whom they are intended” (van Leeuwen, 2008, p. 274). Exclusion, “a symbolic form of social exclusion” (van Leeuwen, 2008, p. 142), can occur when certain groups are entirely omitted from depictions, suppressed, or backgrounded. When included, social actors can be personalized, or represented as humans. Bateman and Tseng (2023, p. 264) mention that personalization in multimodal news discourse could happen “by means of expressions of attitudes and emotions from characters appearing in the news, or emotionalisation of people and events either verbally or by any other means, such as displays of emotional responses, facial expressions, visual framing, and so on.” On the other hand, social actors could be impersonalized, through abstraction or objectivation, often minimizing agency. Visual impersonalization in reporting is understood here as an “image act” (Van Leeuwen, 1993, p. 215) and refers to visual representation of social actors metonymically, through the relation of contiguity, for example, to parts of human bodies, masses of bodies, or objects or spaces, representing or personifying acts of violence, verbally identified in a text accompanying the image. “Objectivation occurs when social actors are represented utilizing reference to a place or thing closely associated either with their person or with the action in which they are represented as being engaged” (van Leeuwen, 2008, p. 46).

Role allocation determines whether actors are portrayed as active agents (activation) or passive recipients (passivation), influencing how responsibility and power are distributed. Actors may also be represented generically (genericization), as types, stereotypes or categories, or specifically (specification), as identifiable individuals, with generic portrayals often leading to dehumanization. They can be individualized or assimilated into groups, through collectivization or aggregation, often affecting how personal or distant they appear. Association and dissociation show how individuals are linked with or separated from others, shaping perceptions of unity or division. Further, actors may be anonymized through indetermination or set apart via determination. “Overdetermination occurs when social actors are represented as participating, at the same time, in more than one social practice” (van Leeuwen, 2008, p. 47). They can be referred to by name (nomination) or categorized by roles or identities (categorization). They can be functionalized, or depicted by what they do; identified (identification) by what they are, such as by age, gender, or ethnicity; or described in explicitly evaluative terms (appraisement) when portrayed as admirable or contemptible, sympathetic or blameworthy. Finally, visual categorization can occur through cultural markers (e.g., dress or behavior) or biological features, which are often exaggerated as the ingroup constructs descriptions of themselves to distinguish them from the others.

Violent death is usually associated with extreme suffering. Whitehead (2004, p. 68) views violent death as “a cultural expression of the most fundamental and complex kind.” Traditionally, it is anthropology that is concerned with “how the destruction of bodies has been integrated into the social and symbolic space of societies affected by violence” along with the “meaning and role of corpses, their parts, and their bodily fluids in those spaces of violence” (Parra and Ubelaker, 2023, p. 5). The CDA approach offers linguistic and multimodal perspectives on the discourses of and about violent deaths in mass media (e.g., García Marrugo, 2017; Easteal et al., 2019; Bunnag and Chaemsaithong, 2023).

In news reporting, the reporting of violent deaths falls under the umbrella term “spectatorship of suffering” (Chouliaraki, 2006). The term refers to the media-constructed ways in which audiences are positioned to emotionally and morally engage/disengage with the pain of distant others (Chouliaraki, 2006, p. 8). Building upon Boltanski's (2005) notion of the “politics of pity,” Chouliaraki (2006, pp. 8–11, 70) differentiates in news discourse three “regimes of pity,” or types of “regimes of meanings” which discursively construct suffering in relation to the desired spectator’s engagement with the sufferer. These are adventure news, or news of suffering without pity; emergency news, or news of suffering with pity; and ecstatic news, or news of suffering and identification. In news reporting, geographical hierarchies play an important role in inducing the politics of pity, when the “prevailing power relations and geographical and cultural proximity” influence the way the media presents suffering (Xu and Zhang, 2024, pp. 638–639, 649). Chouliaraki (2006) suggests the framework of multimodal news analysis by looking at: the properties of language and image in co-appearance and combination (multimodality), representation of symbolic proximity/distance to the sufferers (chronotope) and agency (Chouliaraki, 2006, pp. 8, 74). Focusing on the relationships between the verbal and visual, she zeroes in on the construal of suffering through the “claim to the facticity of suffering in perceptual realism, claim to the emotion of suffering in categorical realism, claim to justice concerning the cause of suffering in ideological realism” (Chouliaraki, 2006, p. 79). While Chouliaraki’s focus is not specifically on the representation of death—although she does touch upon some issues related to the normalization of images of death and war, imminence of death, verbalizations of death (e.g., Chouliaraki, 2006, pp.56, 85, 139)—below I will apply the scholar’s approach to focus on the multimodal representation of the victims of violent death in the news reports of the selected world media.

Digitization of production systems, media and organizational convergences (Erdal, 2007, p. 76) and affordances of Web 2.0 have led to increasingly multimodal news production (Powell, 2017). Following Bateman et al. (2017, p. 132), news reports are understood herein as multimodal/multisemiotic texts, or “artefacts,” embedded in a certain communicative situation and context. Bateman and Tseng (2023, p. 266) point that “combinations of reoccurring elements create multimodal ‘texture’ by which certain elements are constructed as being relevant for interpretation.” In news reports, the nature of text-image relations, I argue, also has an effect on the representation of social actors, since as it is known, visual and verbal framing allows mass media to “structure and interpret news information and relate to the audience’s knowledge of the world” (Geise and Baden, 2015, p. 46). The interaction between image and text is thus central to the construction of meaning in multimodal news, forming a multimodal frame, which has greater potential to influence audiences than either mode alone.

Text-image relations have been a point of scholarly attention, and several attempts have been made to classify and describe the relationship (see, e.g., Martinec and Salway, 2005; van Leeuwen, 2005; Baumgarten, 2008; Liu and O’Halloran, 2009; Alikhani et al., 2023; Bateman and Tseng, 2023, to name a few). As news reports are types of “multisemiotic artefacts” (Parodi, 2012; Bateman, 2021), or, more broadly, “communicative occurrences” (de Beaugran and Dressler, 1981, p. 3), coherence and cohesion construct and stabilize their intersemiotic texture. Intersemiotic coherence refers to inferred relations that are established cross-modally/intersemiotically and form an integrated discourse, where each mode fulfills a communicative function and supports shared inferential goals (Alikhani et al., 2023, pp. 2–6). Alikhani et al. (2023, p. 5) identified five relations of coherence between text and image: (1) visible relation, when the content of image and text overlap, i.e., when the text conveys information directly perceptible in the image, such as objects, attributes, states, etc., without extending beyond what is visibly accessible to the perceiver; (2) action relation, when the text describes an event or activity captured by the image, thus temporally anchoring the visual content within an unfolding process; (3) subjective relation, which is established when the text supplements the image with an evaluative stance, emotional reaction, or personal judgement; (4) story relation, which occurs when the text provides a broader narrative content for the image, framing it as part of larger event or scenario; (5) meta relation, which applies to the cases when the text comments on the image, by explaining how, when, why the image was produced.

Intersemiotic cohesion is defined as an attribute of multisemiotic texts (Liu and O’Halloran, 2009, p. 369) based on recurrence of intersemiotic features that are significant for the discourse (Bateman and Tseng, 2023, pp. 265–266). Martinec and Salway (2005) offer a model of possible image-(verbal) text relations based on the status (equal vs. unequal) and logico-semantic (expansion vs. projection) aspect of image-text relations and identify the units and realizations of such relations. While the model is indeed useful for understanding cohesive ties between the elements, the scholars’ understanding of “whole text” presents a methodological difficulty for the following article.¹ The authors state: “we will consider the ‘whole text’ to mean a paragraph or smaller units, such as clause complexes, clause, and even words, whenever these are the units that images relate to” (Martinec and Salway, 2005, p. 347). In the context of the news reports, we often see images supplied with captions. Captions serve as “cut lines” (Saberi and Chan, 2015, pp. 183–184) that function as inseparable parts of visual “denotation” (Barthes, 1977, p. 26) and help to create various forms of understanding. As Mendelson and Darling-Wolf (2009) report, the readers might look at the pictures and read the caption and ignore the text of the article altogether. I suggest regarding image and caption as a “multisemiotic text” (Liu and O’Halloran, 2009, p. 369) capable of existing on its own as a single meaning-making unit, which is further interwoven into the texture of a newspaper report. To analyze image-caption cohesion and further ‘image-caption’-article coherence, I adopt Liu and O’Halloran's (2009) discourse-based approach, which treats verbal text and image as covariate structures. The authors acknowledge that “[i]n a multisemiotic news report, […] the picture is likely to be connected to its caption, the headline and the report proper in terms of logical meta-function” (Liu and O’Halloran, 2009, p. 379). In addition, this approach provides a flexible and context-sensitive account of how multimodal texts achieve cohesion, coherence, and intersemiotic texture. The scholars acknowledge that based on the established logical relations between the image and verbal elements, various intersemiotic cohesive devices could be used, such as correspondence or parallelism (Liu and O’Halloran, 2009, p. 379). The analysis focuses on the following intersemiotic cohesive links (Liu and O’Halloran, 2009): (1) intersemiotic correspondence, or cross-modal reference to the same experiential content in two or more modes; (2) hypo/hypernymy, or hierarchical relations of generality and specificity between modes; (3) mero/holonymy, or part/whole relations across semiotic resources; (4) homospatiality, or parallel alignment of verbal and visual elements in the same spatial location; (5) intersemiotic ellipsis, or omission of meaning in one mode but inferential supplying by another; (6) intersemiotic reference, or explicit anchoring or pointing from one mode to elements in another; (7) intersemiotic addition, or contribution of information not present in the other mode; (8) intersemiotic temporal sequence, or organization of meaning through chronological ordering across modes; and (9) intersemiotic consequence, or linking of modes via causal or purposive relations.

I argue that understanding text-image coherence and cohesion can help understand how the discursive strategy of (im)personalization used in media discourse to represent the sufferers, perished as the result of violent death, construes “the same situation in alternate ways” (Alonso Belmonte and Porto, 2020, p. 56).

The town of Izium² is located in Kharkiv region in eastern Ukraine. During the Russian invasion of Ukraine in February 2022, Izium became a point of intense fighting. Russian forces occupied the town in March 2022, making it a key military base for their operations in eastern Ukraine. In September 2022, Ukrainian forces successfully recaptured Izium as part of a large counteroffensive in the Kharkiv region. The Ukrainian forces entered Izium on 10 September while the Russian forces hastily abandoned the positions, leaving even their equipment behind. The town suffered extensive destruction, and after the liberation came the reports of severe infrastructure damage and civilian casualties.

As the Ukrainian officials and military started clearing the area, on 15 September 2022 they discovered mass graves in the forest near Izium, with the bodies of both civilians and soldiers. This discovery drew international attention as President Zelenskyj invited international mass media and organizations to visit the site and record the evidence of the alleged war crimes. On 16 September 2022, international media and human rights organizations started arriving in Izium as Ukrainian forensic teams began the exhumation of bodies. On 23 September, Ukrainian officials reported the completion of the exhumation at a mass burial site near Izium, where 411 civilians and 21 military personnel were found, many showing signs of violent deaths and torture (Coynash, 2022). Bodies displayed evidence of horrific treatment, including nooses, bound hands, broken limbs, and some had been mutilated (Cernuzio, 2022). Investigators also uncovered multiple other mass graves in the region, as well as torture chambers used by Russian forces (Centre for Information Resilience, 2023).

Among the first outlets to report about the mass burial were Voice of America, the BBC, Reuters, the Guardian, and over the next few days, most major world media covered the tragedy. Meanwhile, Russian and pro-Russian media accused Ukraine of staging the tragedy with the aim of provocation and further escalation of the conflict. For example, TASS (2022), a Russian state-owned news agency wrote: “The Kiev regime needs the provocation in Izyum involving the alleged discovery of a mass grave just for the sake of photo and video content for the Western mass media.”

The material for analysis (see Table 1) was collected from the selected mass media from Ukraine, independent Russian media, pro-Ukrainian European (UK, France, Germany), Middle East (UAE and Qatar) and American (US, Canada) media as well as India.

A total of 12 articles were collected (slightly more than 10,300 words in total). Only the first reports of the Izium tragedy were taken into consideration. Of these, 2 articles are in Ukrainian, 1 article is in Russian and 9 articles are in English. While the material is geographically diverse, it represents mainly Anglophone data and is somewhat ideologically homogeneous. Russian media reports were excluded from the corpus due to excessive propagandist focus of state-controlled media and therefore different institutional logic, since the reports not only reversed the blame, but also “fictionalized” the victims (see Shurma, forthcoming)³. All articles were accessed during 18–20 September 2024. The choice of the media was based on the popularity of websites, from which the articles were collected (see FeedSpot, 2024; Press Gazette, 2024a, 2024b), and/or availability of texts with accompanying photographs in English without subscription. Reporting for the Russian audience is represented by the independent news web resources Meduza (in Russian) and The Moscow Times (in English), both headquartered outside of and banned in Russia. For the purposes of the current study, only still images were analyzed, while video messages were not taken into consideration. A total of 40 photos supplemented the articles in question, of which 3 are maps and 1 is a satellite image.

Suggested analysis falls under the category of a case study focusing on the multimodal representation of violent death. While such types of research have been criticized for focusing on the “individual occurrences of items” (Bateman and Tseng, 2023, p. 282), I believe it could serve as the first step prior to a larger corpus-based analysis reliant on statistical techniques (as suggested, for example, by Bateman and Tseng, 2023) with the outlined patterns tested on larger geographically and ideologically heterogeneous datasets. The paper contributes to the existing body of research into violent death representation in the media (esp., Hanusch, 2010; Morse, 2017). Herein, the multimodal framing of the event in question is analyzed from the perspective of (Multimodal) Critical Discourse Analysis (MCDA) (Chouliaraki, 2006, 2010; Machin and Mayr, 2012; Ledin and Machin, 2018) and Social Semiotics (Hodge and Kress, 1988; van Leeuwen, 2005; Kress and van Leeuwen, 2006; Kress, 2010). CDA helps to uncover ideological workings underlying the representation of social actors and events by critically analyzing how mass media discourse contributes to the construction of the discourses of death from the perspective of power (Van Dijk, 1993; Reisigl and Wodak, 2001; Fairclough, 2013). Multimodal critical discourse analysis (MCDA) is used as a tool for examining how violent death is framed across text and image by revealing the interplay between these modes and the underlying power dynamics, ideologies, and social identities they construct. By integrating the critical perspective with the Social Semiotic approach, I analyze how the linguistic choices, along with the visual elements, shape the framing of the discovery of Izium mass burials by selected news outlets and subsequently public agency in such framing. This analysis uncovers how the news reports around the world highlight or obscure sufferers’ agency, death and public pity (Chouliaraki, 2006), and influence social attitudes towards the Russo-Ukrainian War. Ultimately, the analysis aims to contribute to an understanding of how multimodal framing both reflects and shapes broader social contexts, especially with regard to attitudes and interpretations of death and death-related practices. SketchEngine was used for collecting statistical information from the texts. Additionally, content analysis was performed to examine intersemiotic cohesion. Drawing on the previous studies about intersemiotic cohesion (Martinec and Salway, 2005; Liu and O’Halloran, 2009), the following intersemiotic cohesive links were identified: correspondence, hypo/hypernymy, mero/holonymy, homospatiality, ellipsis, reference, addition, temporal sequence, consequence. Since captions are viewed herein as subordinate to the image, the relation of verbal text to image was explored. The coding was done by the author and a trained coder with an MA in linguistics who speaks English, Russian, and Ukrainian. The second coder was trained on the process of coding, and the differences were discussed by both coders. In the discussions with the coder, ellipsis and homospatiality were excluded from the list of cohesive links. Homospatiality is the link between text and image on the “expression plane” (Liu and O’Halloran, 2009, p. 372), such as readable names on the crosses in the dataset. Yet, when applied to image-captions, seen here as a multisemiotic text, the captions could be seen as a property of the expression plane, albeit optional; hence, it can be argued that homospatiality is a property of captions. Similarly, ellipsis, defined as a cohesive device in which a meaning omitted in one mode is recovered from another (Hendriks, 1995), proved difficult to identify or justify: every caption in the dataset had a narrower focus than the inferences possible from the images, on the one hand, and the cohesive device of addition in captions, on the other hand, could be argued to create visual ellipses. The coding was done on 34 visuals (individual and groups of images in case of Zelenskyj’s Telegram post) supplied with captions, source credits (in case of Suspilne and TSN) or posts (in case of Yermak’s X and Zelenskyj’s Telegram). Kippendorff’s Alpha (Marzi et al., 2024) was equal to 0.874, which indicates excellent reliability.

While it is generally acknowledged that most of the experience with death comes through media, it is also true that the news select as noteworthy mostly the more unusual deaths (Hanusch, 2010, p. 39). Izium tragedy does fall into this category, since, as the media themselves label it, the mass burial found at the site is one of the largest in Europe. The article addresses the gap, identified by Hanusch (2010, p. 60), by looking at how the textual representation of the violent death is related to the visual coverage of the tragic event. The results collected show several strategies in reporting the aftermath of violent events involving the death of a large number of people. News narratives discursively construct the representation of the diseased, including ascribing the roles of protagonists to sufferers, as is the case with Meduza, for example, thus leading to desired/intended narrative interpretations on the part of the readers, which become familiar and habitual (sensu Bateman and Tseng, 2023, pp. 261–262). While the news in question mostly do fall in the category of the “spectatorship of suffering” with pity, they are influenced to a degree by the discourse of death and social practices underlying it.

The results show that across the corpus intersemiotic correspondence dominates in image-caption relations, and thus the captions simply anchor what is already visible rather than reinterpret it. Such alignment contributes to perceptual realism in the representation of violent death in the selected media. Interestingly, although wooden crosses are frequently included visually, they are underrepresented lexically in captions. It might be argued that such asymmetry happens due to the positioning of the cross as a cultural sign that does not require linguistic interpretation, and thus contributes to visual symbolism that is aimed to carry most of the representational weight. In fact, the burial site plays a crucial role in the construction of the chronotope, where present actions, mostly visible, implicitly index past atrocities and enable moral inferences (without graphic depiction, though).

It should be noted that verbal impersonalization results in the strategies of spatialization (association through place) and somatization (association through body parts) (van Leeuwen, 2008, pp. 46–47). This results in two effects: on the one hand, we are faced with the requirements of the genre—unless we are talking about obituaries or eulogies, there is a tendency in the media to impersonalize the diseased; on the other hand, we notice the elements of sensationalism and the media rush to be the first to report an issue of such a scale. Aggregation leads to the variation in reporting mentioned above, which possibly comes from two factors. The time of the news publication influences the data available at that particular time. Translation could be another factor, possibly explaining the variation in references to the numbers of bodies vs. graves. Additionally, the source(s) used for the details provided seem to have/report different information. Originally, President Zelenskyj announced at 16:16 EET in a Telegram post of more than 400 bodies (Zelenskiy / Official, 2022), while Syniehubov posted at 16:26 EET about 450 bodies found at the site (Syniehubov, 2022a); by the end of exhumation on 23 September 2022, the latter reported 436 bodies (of which 21 were soldiers) (Syniehubov, 2022b). Verbal strategies of aggregation, abstraction, and objectivation cumulatively produce symbolic distance and assimilate sufferers into numbers and “bodies.” On the other hand, the selective verbal personalization introduces moments of emotional proximity, which is also consistent with the geographical or cultural proximity.

Chouliaraki’s (2006, p. 88) definition of humanization presupposes “endowing sufferers with the power to say or do something about their condition”; however, when we speak about the deceased, they are devoid of such power. To paraphrase Chouliaraki, in the discourse involving violent death, the humane (diseased) sufferer is the sufferer who symbolically “acts” in death. Via the framing through the narrative details about the deceased, such as the names visible on the crosses or mentioned in the articles, etc., as well as the choice of verbal euphemisms in the texts, such as zakatovani ‘tortured to death’, the sufferers—in the case of the analyzed material, collectively—mainly evoke the “beneficiary action of others” (Chouliaraki, 2006, p. 88). The latter could include, for instance, sending aid to Ukraine either as personal or state contributions in forms of donations, monetary, humanitarian, and military support, condemnation of the aggressor, etc. While the sufferers can no longer be protected or rescued, the idea of the denial of a “good death” (sensu Coret and Martimianakis, 2023, p. 4)—“the innocence of victims” (Fishman, 2017, p. 18)—discursively creates the “regime of pity” ranging from emergency to ecstatic news types based on the symbolic proximity of the event to the reader. At the same time, we also notice that “a reluctance to display death remains steadfast” (Fishman, 2017, p. 11) in the mass media, and the sufferers remain less or more distant (cf. reporting in the Times of India and The Globe and Mail, for example).

Despite the openness of, especially Western, societies to the discussion of various issues, which have been earlier tabooed, death remains a sensitive topic. “Modern societies have been traumatized by confrontation with death in magnitudes not experienced in previous eras” (Moore and Williamson, 2003, p. 10). Scholars explain this tentative approach to death by the fact that humans associate it with fears, including the fear of the destruction of the body after death (Hoelter and Hoelter, 1978; Allan and Burridge, 1991). As members of the community, we thus create symbolic systems to preserve self-esteem in the face of terror, and institutions legitimize those systems. We could speak about the beautification of death or even “the age of avoided death” when modern neoliberal hedonistic culture allows only brief reminders of death (MacLean and Williams, 2003, p. 753). I argue that “avoided death” results in the strategy of visual euphemisation (Pratama et al., 2024; Van Winkle, 2024). Visual euphemisation becomes a discursive practice of replacing the original signifier perceived as unpleasant by another one metonymically connected to the original one to conceal the former. Visual impersonalization thus becomes a suitable strategy to soften the unpleasant “fear of death.” Much like in language, such an “agentless passive style” (Bandura, 1999, p. 195) of representing suffering and sufferers works as a tool in the visual representation of removing the unpleasant from the spectators’ eyes, even if the narrative contains more gruesome details.

In the articles under consideration, objectivation is used to avoid showing sensitive content. The crosses pervasively appearing in the photos acquire symbolic meaning as they stand as euphemisms for violent death and sufferers buried under them. “Symbolic meaning is related to its referent […] by discursive associations based on conventional knowledge and value” (Chouliaraki, 2006, p. 80). While sufferers are visually excluded—we do not see their remains—we associate the cross with the deceased and thus buried people underneath it. Nevertheless, the reliance on the cultural knowledge of the ritual of placing the cross at the head of the grave is addressed differently. The captions are used to help to complement the context of the image. Yet, for the audiences familiar with the burial practices of Ukraine and Russia this determination is not necessary. Thus, Ukrainian media omit captions altogether, on the assumption that the schematic knowledge is sufficient for the interpretation of the visual stimulus.

Avoidance of images of human remains, which can be upsetting for the viewers, also resulted in the instrumentalization in visual representation. Closed body bags or coffins in certain images metonymically stand for the sufferers but also protect the viewers from violent/graphic content. The finding confirms the observation made by Hanusch (2010, p. 61) about newspapers showing “very little actual death.” At the same time, while the scholar mentions that graphic images might still appear if the dead are from abroad, we do not see this happening in the case of the Izium mass burial. It might be argued that one of the reasons for this is the attitude to Ukraine and its efforts of resistance to the undeclared war, which made reporters treat these deaths as not so distant. Additionally, several photojournalists are of Ukrainian origin, and therefore their selection of visual framing reflects their cultural values and stance of cultural producers in the historically important event. On the other hand, the press representation of the mass burial in Izium is a part of what Fairclough termed “a hegemonic practice” (Fairclough, 2013, p. 129) of largely uniform reproduction of journalistic practices across ideologically similar, but geographically distant, media. Therefore, we notice how conventions and rules of specific agencies, such as AFP, AP or Reuters, are “naturalized” (Fairclough, 2013, p. 129), recontextualized and operationalized (Fairclough, 2013, pp. 76, 233) by specific media outlets (see the discussion of the changes in the same AFP article reprinted by The Moscow Times and Le Monde). Through repeated visual motifs as well as verbal and visual framing strategies a common-sense understanding of responsibility and suffering is constructed, and the media discursively reproduce consent around dominant geopolitical narratives related to the Russo-Ukrainian War. At the same time, the role of (photo) journalists, who mediate death through professional, ethical, and cultural norms (see Hanusch, 2010), in socially constructed discourses of violent death cannot be overlooked. Through multimodal and discursive coherence, they have the power to shape how the audiences understand mortality, responsibility, and suffering, and render death as meaningful and morally interpretative.

However, why would certain media then choose to be different, as is the example of Zelenskyj’s Telegram message published by Suspilne? Prior research suggests that “domestic deaths often evoke a heightened level of sensibility, so as not to upset readers or viewers who may have personally known the victims” (Hanusch, 2010, pp. 42–43). The answer to why Ukrainian media would not want to protect its readers in a similar way as the other media do, lies, in my opinion, in the nature of legitimation in the Ukrainian context. Authority and moral evaluation legitimation (van Leeuwen, 2008, pp. 105–111) underlie the message. The Ukrainian President calls for moral value assertion by appealing to the audience’s judgement of good and evil. The caption offers evaluation of what the viewer sees—zhorstokist ‘cruelty’ and terorysm ‘terrorism’—and explicitly points to the “evil” side: I imia tsiomu – Rosiia ‘And the name to it is Russia’ (Riazantseva, 2022). The breaching of the taboo in the photos—a kind of “conflict with respect towards dead” (Morse, 2014, p. 106)—increases the effect of the message and shifts it to Chouliaraki’s category of ecstatic news. In the country torn by the undeclared war, both visual and verbal messages have a very high degree of agency, as this breach of socially established lines in how death should be treated holds great power for action. This humanization of the sufferer as a condition (Chouliaraki, 2006, p. 124) alludes to the humanity of Ukrainians: death should be “good.” Messages like these motivate the public to action, yet also paint the enemy as evil. At the same time, “[w]hile news coverage is essential to convey information to a populace during any crisis (2), the amount, content, and biases of such coverage must be balanced against the likely harmful effects of this media exposure on mental and physical health” (Holman et al., 2024, p. 2).

This study has contributed to the understanding of the multimodal framing of violent death in mass media. At the same time, it has several limitations. First, the analysis of impersonalization in the reporting of violent death was based on a relatively small corpus, which limits the representativeness of the findings, yet the results could be used in further larger-scale mixed method research, which could potentially include reports in other languages. The latter could also add a much-needed cross-cultural dimension to the study. Moreover, with the exception of India, the sources analyzed are predominantly Western, pro-Western, and pro-Ukrainian; this introduces a potential bias and narrows the geopolitical (and ideological) diversity of the dataset, narrowing the analytical lens through which media representations of violent death are understood.