Memory can connect or divide. Whether it becomes a bridge or a battlefield depends on how we talk about the past. Institute IRRIS for Research, Development and Strategies of Society, Culture and Environment Čentur 2026 Memory in the Digital Age: Bridge or Battlefield? Main author: Marjan Horvat Co-authors: Jan Babnik, Darko Darovec, Angelika Ergaver, Jure Koražija, n Urška Lampe, Žiga Oman, Tadej Škvorc, Polona Tratnik Responsible Editor: Darko Darovec Editors: Marjan Horvat, Jan Babnik, Urška Lampe Graphic design: Saša Kladnik, Ervin Markošek opho Some of the visual elements in this work were generated using ChatGPT (OpenAI). Published by: © 2026 Institute IRRIS for Research, Development and Strategies of Society, Culture and Environment, Čentur ( www.irris.eu ) col This publication is a result of the Horizon Europe SoMe4Dem project (Social Media for Democracy), coordinated by the Max Planck Institute for Mathematics in the Sciences, Leipzig. SoMe4Dem is fully funded by the European Union under the Horizon Europe framework (HORIZON-CL2-2022-DEMOCRACY-01-07) grant agreement No. 101094752. Views and opinions expressed are, however, those of the author(s) only and do not necessarily reflect those of the European Union or Horizon Europe. Neither the European Union nor the granting authority can be held responsible for them. Electronic edition URL: https://www.irris.eu/en/publication-memory-in-the-digital-age-brige-or-battlefield/ Kataložni zapis o publikaciji (CIP) pripravili v Narodni in univerzitetni knjižnici v Ljubljani COBISS.SI-ID 268015619 ISBN 978-961-95505-6-4 (PDF) Publikacija je odprto dostopna in licencirana pod licenco Creative Commons BY 4.0. Julian exile no. 30001 (Esule Giuliana n. 30001), Pula (Pola), 6 July 1946. Photograph by an unknown author of the girl Egea Haffner, taken during the Istrian–Dalmatian exodus. 5 s Introduction 6 ent 1. What is Cultural Memory and Communicative Memory? 11 2. Why Does Memory Matter for Democracy? 15 ont 3. The Public Sphere and Communicative Memory 19 C 4. Key Challenges: Polarisation, Echo Chambers, and Memory Wars 23 5. Methodological Framework: How We Conducted the Research 27 6. Case Studies: National, Cross-Border, and European Memory 33 6.a. Day of Uprising Against Occupation (Slovenia) 6.b. Giorno del Ricordo (Italy–Slovenia) 6.c. Europe Day and the Fall of the Berlin Wall (Germany, France, Italy, Slovenia) 7. Results of the Analyses: Key Patterns 37 8. Recommendations 46 9. Conclusion 47 on Collective memory is one of the building blocks of every society. The way we understand the past and the events we preserve as ti significant shape our political choices and our interpretations uc of the present. Memory is therefore not only the domain of his- torians, museums, or state commemorations; it is also a key tr discussing the past, a common language that enables debate, cooperation, and trust, can be lost. In od component of democratic culture and of the broader culture of remembrance. Without at least a minimal shared framework for With the rise of social media, memory has also migrated into digital environments. Historical photographs, commemorations, personal reflections, and political narratives now circulate through tweets, memes, and short videos. This has contributed to the pluralisation of memory, as far more people participate in discussions around memorial days and public holidays. At the same time, memory has become emotionally intense, rapidly shifting, and vulnerable: content spreads in fragmented and viral forms, often without broader con- text. Social-media algorithms frequently amplify content that fuels division rather than understanding. Disinformation, hate speech, memory wars, and conspiracy theories further narrow the space for argument-based discussion. For this reason, the study of digital memory is directly connected to the study of democracy and the quality of deliberation. In a dispersed digital space, communities find it increasingly difficult to identify shared points of reference, which limits the capacity for collective reflection. Memory-related issues thus become a sensitive domain that populist and authoritarian actors can exploit to deepen tensions and consolidate exclusionary narratives. 8 In this publication, we understand memory as a lively debate that unfolds in the realm of both traditional and digital media. We ask whether digital memory, beyond generating conflict, can also support deliberative values: openness, plurality, respectful dialogue, and the capacity for different memories to coexist without exclusion. To this end, we connect research on online and legacy (offline/traditional) forms of cultural and communicative memory with deliberative de-mocracy and complement it with advanced AI-based analytical tools that facilitate the systematic analysis of large datasets and recurring patterns in public debate. This booklet summarises findings and research premises developed within the SoMe4Dem (Horizon Europe) project. It builds on schol- arly articles and project reports on cultural memory, digital public spheres, and deliberative democracy, prepared by the IRRIS Institute independently or in collaboration with project partners. The opening chapters introduce cultural and communicative memory and outline key challenges of the digital public sphere. We then present our meth-odology and a set of case studies from Slovenia, Italy, Germany, and France, showing how national, cross-border, and European memories are shaped online and where opportunities for dialogue emerge. The publication concludes with recommendations for action in the digital memory space. We believe these are relevant for citizens, journal- ists, schools, and decision-makers who seek to build a responsible, inclusive, and reflective culture of remembrance in the digital public sphere. Democracy also depends on the ability of different memories to co-exist. The digital space can facilitate this if we understand its impact, actively shape it, and act responsibly within it. 9 A ChatGPT-generated collage featuring a photograph of Italian soldiers shooting Slovenian hostages in Dane, Loška Valley, 10 Slovenia, 31 July 1942. Collection of the National Museum of Contemporary History of Slovenia. Inventory no. pl1818. s pt n antagonism o ce A mode of discussion in which the ey c other is not an adversary but an enemy; it is marked by exclusion, e – k labelling, and delegitimisation. at eb f d es o d agonism o m A mode of discussion characterised by critical disagreement in which opponents’ views are recognised as legitimate, without exclusion. deliberation A mode of discussion grounded in arguments, verifiable information, and a willingness to listen, weigh different viewpoints, and reflect 11 collective memory is one of the foundations of a democratic community: it shapes how we understand the past, what we consider credible, and which shared symbols we use to build trust. Memory can connect or divide. Whether it becomes a bridge or a battlefield depends on how we talk about the past. 12 1. What is Cultural and Communicative Memory? Berlin Wall, East Side Gallery, Berlin, 31 January 2015. Photograph: Ank Kumar (Infosys Limited). CC BY 4.0. 13 1. What is Cultural and Communicative Memory? Memory is one of the key resources of identity: it links the past to the present and the future, and creates symbolic orientations through which communities recognise themselves. When communities and socie-ties commemorate important past events, they do so not only because of historical facts, but also in order to make sense of the present and of their role within it. Jan Assmann defines cultural memory as the part of collective memory that is transmitted through symbolic forms: rituals, monuments, holidays, textbooks, and media. It preserves durable frameworks of meaning through which a com- munity constructs a narrative about itself. Communicative memory is “living” memory: it is produced through everyday conversations, personal stories, and family narratives, and typically reaches back three to four generations. Communicative Cultural Memory Memory History in the frame of Mythical history, events Content autobiographical memory, in absolute past (“in illo recent past tempore”) Informal traditions and High degree of formalisation, Forms genres of everyday ceremonial communication communication Mediated in texts, icons, Living, embodied memory, dances, rituals, and Media communication in performances of various vernacular language kinds; classical or otherwise formalised language(s) 80–100 years, a moving Temporal Absolute past, mythical horizon of 3–4 interacting structure primordial time, “3000 years” generations Participation Specialised carriers of memory, Diffuse Structure hierarchically structured Communicative vs. Cultural Memory: Key Distinctions (Assmann, 2008). 14 Assmann emphasises that communicative memory is not a mere reflection of cultural memory. Rather, it actively reshapes it. In the digital age, this dynamic is even more pronounced. Photographs, viral videos, and comments enter the public sphere faster and with greater visibility, sometimes even more forcefully than institutional narratives. Institutions increasingly seek contact with members of the public precisely in digital environments, and adapt their modes of communi-cation accordingly. The boundary between private and official memory becomes blurred, and memory turns into a dynamic, rapidly changing field in which per-sonal experiences, political interpretations, and emotional reactions intertwine. Cultural memory has a dual nature. It can connect: it helps communities build a sense of belonging and shared reference points. At the same time, it can divide: social groups may remember the same historical events in divergent ways and assign them different meanings, leading to the formation of distinct identity groups, which are often polarised. This can trigger disputes over the correct interpretation of the past. In such cases, memory quickly becomes a political instrument or a site of cultural conflict. At this point, several key questions come to the fore: Who determines what counts as “real” history when the boundaries between official and personal narratives are dissolving, and when any user can produce or reinterpret mne-monic narratives? How can we distinguish between memory that connects and memory that divides? These questions served as the starting point for our research. key points of the chapter 1 • Memory links the past, the present, and the future, and co- shapes the identity of communities. • Cultural memory is grounded in durable, institutionalised carriers of memory: symbols, texts, rituals, and archives that preserve meaning across generations. • Communicative memory comprises short-term, everyday, and interpersonal narratives transmitted through communication and lived experience. • Both types of memory constantly intertwine and mutually reshape one another. 15 1. What is Cultural and Communicative Memory? Graffiti on the Berlin Wall, 1989. Key References Pierre Nora (1984) – Les lieux de mémoire. A classic text introducing the con- cept of sites of memory (lieux de mémoire): monuments, rituals, and symbols as anchoring points of collective memory. Jan Assmann (2008) – Communicative and Cultural Memory. A foundational account of the distinction between cultural and communicative memory. Assmann & Czaplicka (1995) – Collective Memory and Cultural Identity. A ca- nonical reference for understanding how communities build identities through symbolic markers. Benedict Anderson (1983) – Imagined Communities. A key work on the narra- tion of community; it explains how nations come into being through shared stories and memory. Eric Hobsbawm (1983) – The Invention of Tradition. Shows how traditions and rituals are constructed to consolidate political communities. Jay Winter (2008) – Sites of Memory and the Shadow of War. An analysis of col- lective trauma, wars, and how societies manage memory of historical ruptures. Michael Rothberg (2009) – Multidirectional Memory. A pioneering book argu- ing that different memories do not simply exclude one another; they intersect, compete, and mutually reshape each other. Ruth Wodak & John E. Richardson (2013) – Analyzing Fascist Discourse. Mem- ory in political rhetoric; shows how history and memory are mobilised in politi- cal conflict, especially in the context of nationalism. Andrew Hoskins (2011) – 7/7 and Connective Memory: Interactional Trajectories of Remembering in Post-scarcity Culture. A theory of a “connective memory” explains how digital networks blur boundaries between private and public memory and generate rapidly shifting flows of mnemonic practice. Astrid Erll & Ansgar Nünning (eds.) (2008) – Cultural Memory Studies. One of the most important edited volumes in cultural memory studies. Maurice Halbwachs ([1950] 1992) – On Collective Memory. A foundational work in the sociology of memory. 16 2. 2. Why Does Memory Matter for Democracy A ChatGPT-generated photorealistic image of the Fall of the Berlin Wall, including recognised landmarks and symbols. Created in December 2025. 17 2. Why Does Memory Matter for Democracy? D emocracy is not only a form of government, but also a way of living together and communicating in pu-blic, where debate about the past is central to both. Discussions about what and how we remember reveal whether we understand democracy as a space of plural voices, or as a contest of exclusionary interpretations and narratives. Jürgen Habermas (1989, 1996) emphasises that a democratic public depends on deliberation: open discussion in which arguments matter more than group belo- nging. Erik Meyer (2008) adds that memory forms the background against which we shape our views. Collective memory functions as symbolic capital through which actors legitimise their decisions. In this way, memory becomes the silent background of the democratic process, determining which values carry weight in a society and which do not. Memory can connect. Shared reference points – anniversaries, monuments, symbols – enable a community to develop a common language. Yet when differ- ent groups interpret the same events in incompatible ways, memory becomes a source of polarisation. Disputes over who was on the right side of history shift discussion from the past towards a struggle over identity. In recent years, cultural memory has increasingly been used within the frame- work of the politics of history (Geschichtspolitik) as a tool of mobilisation. Rather than serving as a space for reflection, it becomes an instrument of emotional mobilisation, thereby intensifying antagonisms. Memory is the silent At the same time, memory has a democratic backdrop of democracy: function. Misztal (2005) argues that commu- nities grow stronger when they confront past it determines which injustices. Wüstenberg (2023) adds that memory stories count, which policies can build dialogue and mutual recog- values carry weight, and nition. Here, memory is not an obstacle to de- mocracy, but a condition of its vitality: a way in where we can begin a which society learns, changes, and understands dialogue. itself. This is where memory research directly intersects with deliberation. The key question is no longer only what we remember and what we forget, but how we talk about it. Democracy does not require a single, unified memory; it requires the ability for multiple memories to coexist in a spirit of respect and argument-based 18 debate. Memory that excludes fuels populism. Memory that contributes to un-derstanding strengthens democratic culture. For this reason, reflective remembrance is part of 21st-century political literacy. Societies that can speak about the past openly, argumentatively, and without mythologising can sustain a democratic culture and a shared future that is resilient in the face of manipulation and forgetting. Scholars on Memory and Democracy “It appears that what matters for democracy’s health is not social remembering per se but the way in which the past is called up and used.” Barbara A. Misztal (2005) “What does ‘memory’ do, who ‘does’ memory – and what is being done by politically utilizing memory – are the central concerns for the study of memory politics.” Maria Mälksoo (2023) “Not only is remembering the past not always good for democracy, then, but not all forms of remembering are democratic.” Amy Wüstenberg (2023) key messages of the chapter 2 • Memory shapes how we understand the past and political present. • Democracy requires shared reference points; without them, debate quickly slides into identity conflict. • Collective memory functions as symbolic capital through which political actors legitimise their decision-making and/or decisions. • Memory can connect or divide, depending on how it is used in public. • Political instrumentalisation of history intensifies polarisation. • An open, plural, and reflective approach to memory strengthens democratic dialogue and community resilience. 19 2. Why Does Memory Matter for Democracy? EEAS, elections to the European Parliament, 2024. Key References Jürgen Habermas (1989, 1996) – The Structural Transformation of the Public Sphere & Between Facts and Norms. A foundational theory of public deliberation as the basis of democracy; shows that democracy needs shared reference points and argument-based exchange. Erik Meyer (2008) – Memory and Politics. Defines memory as symbolic capital and explains how history and memory shape democratic processes. Maria Mälksoo (2023) – Politics of Memory: A Conceptual Introduction. Shows how states use history for identity construction, geopolitical goals, and emotional mobilisation. Peter J. Verovšek (2020) – Memory and the Future of Europe. An analysis of Europe’s memory community; argues that democracy can be strengthened through open confrontation with past injustices. Jenny Wüstenberg (2023) – Memory and Democracy . Highlights tensions between memory and democracy; shows when memory enables dialogue and when it leads to exclusion. Barbara A. Misztal (2005) – Memory and Democracy. Develops the argument that democratic culture does not depend on remembering per se, but on how memory is used in public. 20 3. The Public Sphere and Communicative Memory Image reconstruction generated using ChatGPT, based on a historical photograph of an audience wearing special glasses while watching a 3D film at the Telekinema in London during the Festival of Britain, May 1951. Collection: The National Archives UK. 21 3. The Public Sphere and Communicative Memory through posts, photographs, short videos, memes, metaphors, and With the rise of social media, memory has also moved into the digital space. What was once shaped pri-marily by schools, museums, archives, traditional media, and state ceremonies is now also formed hashtags. Memory is increasingly less the exclusive domain of insti- tutions and increasingly produced by a dispersed network of millions of users, each interpreting the past in their own way. Personal posts, emotional reactions, and political interpretations merge into a single but unstable digital flow. Digital environments blur the boundary be- tween private and public, and between individual and collective memory. Andrew Hoskins and Steven D. Brown (2010) call this a new ecology of memory , as the past is continuously reconstructed through the everyday use of digital platforms, often while scrolling. What was once a time-bounded state commemoration has now unfolded into an ongoing, informal, and emotionally intense online conversation. In the digital age, memo- Institutions therefore gradually lose their ry is no longer produced monopoly over the interpretation of history. They share attention with masses of users primarily in institutions, who produce their own versions of the past. but in a continuous Researchers note that digital debates about stream of posts, algo- memory take place in a dynamic and often rithms, and emotional chaotic stream, where different actors, con- texts, and interpretations intersect. Classical reactions, making it one theories of collective memory and the public of the key arenas of the sphere are often no longer sufficient on their democratic public sphere. own. This shift compels us to rethink Assmann’s distinction between communicative memory (informal memories, family narratives, personal accounts) and cultural memory (institutionalised forms of remembrance, such as rituals, textbooks, archives). In digital environments, the two types of memory become hybrid and interwoven: • Personal posts can become part of broader cultural narratives within minutes. • Viral hashtags can durably reshape the meaning of historical events. • Platforms generate new forms of digital ritualisation and commemorative practices. 22 • The changes run deep into the structure of the contemporary public sphere: • The internet creates affective publics in which affect is often more influential than argument. • Fragmented semi-publics emerge, without a shared space of debate. • Mnemonic narratives unfold rapidly, non-linearly, and with high emotional intensity. • Algorithms, driven by commercial logics, often undermine reflection and dialogue. • Digital archives emerge as dynamic systems co-produced in real time by user contributions, algorithmic selection, and platform infrastructure. Because algorithms are guided by attention dynamics, and too often by com-mercial profitability rather than democratic aims and a logic of reflection, digital debates are frequently emotionalised and polarised. The digitalisation of memory has therefore become a cultural and democratic turning point: it enables great-er participation and inclusion of histories that were previously marginalised or silenced, while also increasing exposure to manipulation, disinformation, and polarisation. If we want digital memory to function as a space of understanding rather than division, we must recognise these processes, analyse them, and develop a critical understanding of how they work. key messages of the chapter 3 • In the digital age, memory is increasingly shaped on social media. • The boundary between private and public memory becomes blurred. • Personal and collective memory intertwine in digital environments. • The internet produces affective (emotional) and fragmented publics. • Algorithms amplify conflictual and emotionally charged content. • Digital memory opens new possibilities, but also increases vulnerability to manipulation. 23 3. The Public Sphere and Communicative Memory Vandalized monument at Republic Square, Ljubljana, 2019. Key References Steven D. Brown & Andrew Hoskins (2010) – Terrorism in the New Memory Ecology: Mediating and Remembering the 2005 London Bombings. Digital memory as a real-time, unstable reconstruction of the past; an intertwining of the personal and the collective. Ben-David, Oren Meyers & Motti Neiger (2024) – How Social Memory Works on Social Media. Memory as a fluid, non-linear process in which actors, contexts, and interpretations become entangled. Zizi Papacharissi (2009) – The Virtual Sphere 2.0. Digital media as a space of affective, personalised publics. Jürgen Habermas (2023) – A New Structural Transformation of the Public Sphere. The digital public sphere fractures into fragmented semi-publics, without a shared horizon. Janet E. Marles (2015) – Simulating Memory. Digital media generate emotional, hybrid, non-linear mnemonic narratives. Slavko Splichal (2024) – The Rise of Contractual Publics. Platforms technological interests reshape the conditions of public debate: publics become contractual. José van Dijck (2007) – Mediated Memories in the Digital Age. Digital tools transform how personal and collective memories are stored, shared, and interpreted. 24 4. key challenges: Polarisation, Echo Chambers, and Memory Wars Collage generated using ChatGPT, based on a photograph of the Monument to the Revolution by sculptor Drago Tršar on Republic Square in Ljubljana, combined with documentary scenes from the Friday cycling protests during the Janša government (2020–2022).. (Personal archive) 25 4. Key Challenges: Polarisation, Echo Chambers, and Memory Wars R esearch over the past two decades clearly shows that digital environments create conditions in which political conflicts can quickly escalate into antagonism. Three structural features of contemporary networks contribute to this: • Homophily – people tend to form ties with like-minded others. • Filter bubbles – algorithmic personalisation/ranking may narrow exposure and reinforce preferences. • Echo chambers – self-selected communities reinforce beliefs by repetition. When these mechanisms combine, they produce what Petter Törnberg (2022) calls affective polarisation. Political opponents no longer perceive each other as legitimate participants in democratic debate, but as a threat. Conflict shifts from the realm of arguments to the realm of identities, moral differentiation, and emotional distance. In such an environment, the digital public sphere becomes an ideal breeding ground for antagonistic narratives. Moral judgments dominate over arguments; interactions are short and emotionally charged; us-versus-them logic reinforces exclusionary practices. MEMORY WARS IN DIFFERENT CONTEXTS Germany and the digital transformation of the memory of the fall of the Berlin Wall: The far right uses social media to reinterpret the fall of the Berlin Wall. The internet plays a key role in disseminating these alternative meanings (Richardson-Little et al., 2022). Russia and the post-Soviet space: On digital platforms, disputes over the Second World War, totalitarianism, and the Soviet legacy are intensifying. Online debates quickly shift from historical interpretation to political propaganda and identity-based divisions (Rutten et al., 2013). The Western Balkans: Online debates about the wars of the 1990s shape cross-border memory communities that often retain a logic of moral polarisation – ours versus theirs. The digital space reinforces both solidarity and exclusion (Fridman, 2022). Online extremism: Anti-Semitic, racist, and far-right memes often originate on fringe online forums and then migrate to mainstream platforms. There they become part of a broader toxic memory culture and influence public debates about history (Zannettou et al., 2020). 26 As the public space fragments into affective micro-communities, memory be-comes one of the key battlegrounds of contemporary politics. On social media, debates about history increasingly unfold outside institutional oversight, within rapidly moving and often conflictual online streams. History becomes an identity weapon – a phenomenon researchers describe as memory wars. Digital communication is shaped by the infrastructural functions of contem-porary platforms and by their communicative affordances – memes, tags, likes, and follower mechanisms – which generate hybrid and highly affective forms of expression. The combination of short texts, visuals, and slogans enables mean-ings and novel interpretations to spread rapidly and be continuously recontex-tualised, thereby shaping new modes of collective remembering. Embedded in algorithmic logics, these practices encourage intense emotional reactions and contribute to a transformation of communication towards fragmented, identi-ty-laden memory wars. On the other hand, the internet facilitates new forms of memory activism, such as bringing silenced stories to light, solidarity campaigns, and transnational learning. However, as Yifat Gutman and Amy Wüstenberg (2023) emphasise, in-tervening in the memory landscape also means intervening in the symbolic iden-tity of a community, so activism can strengthen dialogue, or intensify tensions. The key question, then, is how – in an affectively fragmented and algorithmi-cally mediated space – we can recognise when debates about memory slide into antagonism, where agonistic dialogue emerges, and when the possibility for deliberation opens. To address this question, we developed a combined analytical approach to studying digital memory, presented in the next chapter. The method enables systematic analysis of digital commemorations and helps identify where the potential for a shift – from conflict to dialogue – becomes visible. key messages of chapter 4 • Digital environments accelerate affective polarisation. • Algorithms create conditions for echo chambers. • History becomes an identity weapon. • Deliberation weakens in such an environment. • Memory can also be a space of resistance and solidarity. 27 4. Key Challenges: Polarisation, Echo Chambers, and Memory Wars When Memory Becomes Activism George Floyd Square as a digital–physical site of remembrance. Global campaigns: #Srebrenica, #HolocaustRemembranceDay, #BlackLivesMatter. Decolonial movements and the removal of contested monuments. Key References Eli Pariser (2011) – The Filter Bubble. The first analysis of personalisation algorithms that push users into narrow information bubbles. Miller McPherson, Lynn Smith-Lovin & James M. Cook (2001) – Birds of a Feather . Establishes the concept of homophily – the tendency to form networks with like-minded others, a key component of echo chambers. Matteo Cinelli et al. (2021) – The Echo Chamber Effect on Social Media. A large comparative study across platforms demonstrating pronounced echo-chamber formation on social media. Petter Törnberg (2022) – Echo Chambers and Viral Misinformation. Shows that misinformation spreads as a form of complex contagion within dense networks of like-minded users. Ellen Rutten, Julie Fedor & Vera Zvereva (2013) – Memory, Conflict and New Media. A pioneering analysis of online memory wars in post-socialist states. Ned Richardson-Little, Samuel Merrill & Leah Arlaud (2022) – Far-Right Anniversary Politics. Shows how the far right uses historical anniversaries to reinterpret collective memory (e.g., AfD and the GDR/DDR). Orli Fridman (2022) – Memory Activism and Digital Practices after Conflict. Analyses how digital media facilitate the reactivation of marginalised or unwanted memories. Savvas Zannettou, Joel Finkelstein, Barry Bradlyn & Jeremy Blackburn (2020) – A Quantitative Approach to Understanding Online Antisemitism. Empirically shows how extremist historical discourses spread through platform ecosystems – from mainstream networks to 4chan. Yivat Gutman & Amy Wüstenberg (eds.) (2023) – The Routledge Handbook of Memory Activism. A comprehensive overview of memory activism and new digital forms of public pressure and participation. Jessica Yarin Robinson (2022) – Cosmopolitan Memory Formation after George Floyd. A case showing how global events can generate cosmopolitan memory: local traumas become global reference frames. 28 5. methodological framework: How We Conducted the Research The commemoration of the twentieth anniversary of the accession of ten countries to the EU, under the motto “Better Together,” concluded with the ceremonial illumination of the Brandenburg Gate in the colours of the EU. Source: Embassy in Berlin, 2024. 29 5. Methodological Framework: How We Conducted the Research past, we sought to understand whether these processes can be steered A t the core of our research was the question of whether digital memory can contribute to democracy – and if so, under what conditions. Because social media radically transform the ways in which communities imagine the away from conflictual interpretations (antagonism) and towards re- spectful engagement with differences (agonism) and argument-based discussion (deliberation). Our aim was therefore to examine how digital debates move between these three modes, and whether, within the mass of online interactions, “islands” of dialogue emerge – spaces where understanding and co-creation of meaning become possible. antagonism – history as a weapon of exclusion agonism – disagreement that recognises the legitimacy of the other deliberation – arguments, context, fact-checking, and ethical judgment AAD Scores Data Colection Qualitative & Greimas’ Groups of Actants (News & X) Quantitative narrative LLM Analysis structures Media Cloud Gpt-40 Sentence Transformers Selenium UMAP Gpt-40 HDBSCAN Discourse Function Scores 30 To do so, we developed a methodology that combines deliberative democratic theory (Habermas, Mansbridge, Bächtiger), agonistic politics (Mouffe), cultural memory studies (Assmann, Meyer), narratological approaches (Greimas), and artificial-intelligence tools. Our goal was to build empirical tools for measuring and comparing memory debates across different countries. This approach enables us to identify when conversations intensify conflict, when they sustain plurality, and when the pos-sibility for dialogue opens up. Methodological workflow: First, we collected data from traditional media and social media platforms (1). Next, using artificial intelligence, we analysed how the texts are constructed and what kinds of narratives they produce (2). We carried out three types of analysis: (a) narrative analysis, identifying the main roles within the stories; (b) an analysis that classifies entries as antagonistic, agonistic, or deliberative; and (c) an analysis assessing whether texts contribute to knowledge, respect, and inclusion, or instead to ideological polarisation. Groups of Actants Connecting Actants Interpretation with AAD scores 3. 4. 5. Sentence Transformers Mapping Narrative Narrative Structures UMAP structures with Organized by AAD & HDBSCAN MFA & HDBSCAN Discourse function scores 31 a) Narrative Analysis (Greimas) We analysed each text – whether a tweet, a newspaper article, or a comment – as a story. Using Greimas’s actantial model, we examined who or what appears in the text as: (1) the hero/protagonist (subject), (2) what the protagonist seeks to achieve (object/goal), (3) the sender of the message, (4) its addressee/receiver, (5) the helpers who support the protagonist, and (6) the opponents who obstruct them. This approach enabled us to group entries according to the narratives they share. It shows when history is used to mark “ours” and “theirs,” where older divisions are reproduced, and where new shared stories begin to emerge. sender object receiver [source of the narrative] [meaning and purpose [community, public] of the memory] helper subject opponent [supporting actors] [bearer of the story [obstacles, and memory] opposing actors] Greimas’s actantial model represents a story (including a memory narrative) as a subject’s trajectory towards a goal (the axis of quest), supported by helpers or obstructed by opponents (the axis of conflict). The story is also structured by motive and audience – who sends the message and to whom it is addressed (the axis of communication) b) The AAD Scale – Antagonism / Agonism / Deliberation Using our AAD model, grounded in theories of antagonism and polarisation, agonistic democracy, and deliberative democracy, we classify each entry into three categories: Mode Key feature What we looked for Delegitimisation, hostile labels/ “Us” vs. “them,” moral exclusion markers, denial of any shared ground Recognition of the other, Adversaries, not enemies argument-based critique Verifiable sources, reasons/ Arguments, context, openness justifications, openness to changing one’s position The AAD scale (antagonism–agonism–deliberation) provides an overview of three modes of memory-related discourse and the indicators we used to identify exclusion, respectful disagreement, or argument-based debate in posts. antagonism agonism deliberation 32 c) Discourse Functions To assess the democratic quality of debates, we measured four functions: Function Purpose What we track Epistemic Verifiable information, clear Use of facts and sources; responses to counterar-sources, evidence, and guments; verification of openness to correction information Ethical Respect, recognition of Tone of the debate; the other, empathy, and recognition of the other and consideration of different their experience groups’ experiences Equal participation, diversity Diversity of voices; reci- Democratic of voices, and fair exchange of procity in exchanging views; viewpoints inclusion of different actors Dominance of a single Ideological Framing and emphasis on frame; displacement/ power structures exclusion of alternative interpretations Discourse functions of debate: four dimensions (epistemic, ethical, democratic, ideological) through which we assessed whether debates contribute to knowledge, respect, inclusion, and reflection on power – or instead to polarisation. With this methodology, we can show precisely • where digital debates slide into antagonism, • where space opens for respectful disagreement (agonism), • and where signs of deliberation emerge. methodological note: .......................................................................................... Our analyses combine qualitative and quantitative approaches: narrative modelling, the AAD scale, discourse-function analysis, topic modelling, embedding analyses, and LLM-assisted clas-sification. All data were collected from publicly available sources (newspapers, online platforms, X [formerly Twitter]). In our analysis, we followed ethical standards, anonymised sensitive data, and applied the principle of minimal intrusion. Summary of the methodological framework • We distinguished antagonism, agonism, and deliberation as three ba- sic modes of memory-related discourse. • We integrated narrative analysis, the AAD scale, and discourse func- tions into a comprehensive analytical framework. • The approach shows how identities and conflicts are reshaped through digital narratives, and whether digital memory enables a shift from conflict to dialogue. 33 To understand how digital memory operates in practice, we conducted three in-depth case studies – at the national, bilateral (cross-border), and transnational levels. Together, they show how history, memory, and digital communication intertwine in Europe. Across all cases, we applied the same analytical framework, adapted to the specific logic of each commemoration. 34 6. case studies: national, cross-border, and european memory The presidents of Slovenia and Italy, Borut Pahor and Sergio Mattarella, in Bazovica, 13 June 2020. Photograph: Francesco Ammendola/Reuters. 35 6. Case Studies: National, Cross-Border, and European Memory a) Day of Resistance Against the Occupier — National Memory (Slovenia) The Day of Resistance (April 27) is one of Slovenia’s most sensitive public holidays. It commemorates the founding of the Liberation Front (OF) and symbolises anti-fascist resistance, while also being marked by disputes over the role of the Communist Party within the resistance movement, the post-war consolidation of power, and the mass post-war killings of alleged collaborators. Since Slovenia’s independence, this date has become a focal point not only for controversies re- lated to the Second World War, but also for debates over the broader period of Socialist Slovenia and the one-party regime. It is a holiday on which contempo- rary political divisions in Slovenian society are particularly visible. To understand how digital media transform this memory, we analysed coverage in traditional media and posts on social media. In this case study, we examined: • How do social media reshape memory narratives around the Day of Resistance compared to traditional media? • Do social media discussions around the Day of Resistance contain a greater share of antagonistic and agonistic modes of debate than traditional media – and where do new spaces for deliberation emerge? • How do shifts in memory on social media around the Day of Resistance affect the epistemic authority of historical interpretations? april 2023–april 2024 753 articles ance en 202 (collected via 164 y o– st f ia 4 slovene 2 ov si posts media cloud, slovene da supplemented by 202 sl other sources) re 36 b) Giorno del ricordo — Cross-Border Memory (Italy–Slovenia) Giorno del ricordo (February 10) has in recent years become one of the most divi-sive days of remembrance in Italy. On this day, Italy commemorates the victims of the so-called foibe (extrajudicial killings carried out by Yugoslav forces after the liberation of Trieste and the Julian March in 1945) and the post-war exodus of Italians from Istria and Dalmatia. In recent years, the commemoration has become a prominent element of public debate about historical memory, identity, and how societies relate to the past – while also frequently being the subject of controversy and competing interpretations. Alongside remembrance of the vic-tims of end-of-war reprisals, debates also raise questions of guilt and the crimes of Fascist Italy, especially in relation to Yugoslavia. These tensions have made the holiday one of the key focal points of polarisation both in Italy and in the Slovene–Italian border region. In this case study, we examined: • To what extent do debates about Giorno del ricor-do contain elements of agonistic debate, and how does the share of such content differ between X and newspaper articles in Slovenia and Italy? • How does the agonistic potential visible in debates on X and in traditional media align with the po- tential for agonism detected through a user survey questionnaire? el 4 o d 23.000 1.390 68 202 articles – italian italian User survey responses ord 2 collected via questionnaire posts 198 Media Cloud, 148 orno ric to compare 202 manually supple-gi slovene mented Slovene slovene agonistic corpus potential 37 6. Case Studies: National, Cross-Border, and European Memory c) Europe Day & the Fall of the Berlin Wall Europe Day (May 9) and the Fall of the Berlin Wall (November 9) are commem- orations that occupy a central place in European identity. The first emphasises peace and cooperation; the second highlights the end of divisions and democ- ratisation. Online, both serve as triggers for debates about European values, identities, geopolitical challenges, and the future of the EU. Across four countries – Slovenia, Italy, Germany, and France – we conducted a comparative analysis of how these discussions unfold and what quality of dialogue they facilitate. In this case study, we examined: • How do levels of antagonism, agonism, and deliberation differ between debates around Europe Day and around the Fall of the Berlin Wall in Slovenia, Italy, Germany, and France? • How do Europe Day and the Fall of the Berlin Wall, in each country, trigger debates on pressing European issues – and how does the deliberative quality of these debates differ? analyzed we ay april 2023–june 2025 e h all e d 10.300 74.444 f t wn posts posts op (drawn from a larger (drawn from a larger ll orli eur dataset) dataset) fabe why these cases? the cases were selected to capture as wide a range as possible of memory prac- tices across the European Union: • from internal debates about national identities, • through sensitive cross-border issues, • to shared European symbols and historical turning points. This combination allows us to observe different levels at which digital memory operates and to understand how memories are contested, reinterpreted, or consolidated in digital environments. At the same time, it helps us more clearly identify the conditions under which digital memory fuels conflict – and those under which it can open space for dialogue, reflection, and learning. 38 7. results of the analyses Key Patterns Construction of the Berlin Wall at the Brandenburg Gate in August 1961. Source: Wikipedia. 39 Cl uste r r t t ive jec ec der ce Sub Obj Sen Re Commemora- 6 National Unity Historical Themes National Unity tion 3 Resistance and National Sovereignty Aspiration for Sociopolitical Independence and Independence Autonomy Stability 11 Commemora- Commemoration/ Historical Themes National Unity tion National Unity 12 Resistance and National Sovereignty Historical Themes National Unity Independence and Independence 15 Resistance and National Sovereignty Historical Themes Slovenian Society Independence and Independence 21 Commemora- Commemoration Historical Themes National Unity tion 22 Commemora- Slovenian Society/ Commemoration Cultural Tradition tion National Unity 10 Historical Revi- Historical Quest for Truth/ Public Insight sionism Understanding Historical Themes 5 Communication Historical Quest for Truth Public Insight Clarity Understanding 20 Historical Revi- Historical Historical Revisionism Public Insight sionism Understanding 13 Resistance and Systemic Historical Injustice Slovenian Society Independence Transformation 1 Human Societal Dynamics/As- Slovenian Society/ Public Awareness Dynamics piration for Autonomy Colective awareness 2 Political Historical Themes/ Sociopolitical Political Dynamics Dynamics Ideological Dynamics Stability 4 Resistance and National Sovereignty Historical Themes National Unity Independence and Independence 7 Critical Analysis Truth & Accountability Public Insight Accountability Truth & 8 Critical Analysis Historical Themes Public Insight Understanding Historical 9 Public Insight Revisionism Historical Historical Historical Understanding Revisionism 16 Historical Themes Slovenian Society Independence Resistance and National Sovereignty and Independence 17 Critical Analysis Historical Injustice Accountability Truth & Collective Awareness 18 Critical Analysis Dissatisfaction Public Insight Accountability Truth & 19 National Identity Historical Under- Historical Themes Public Insight standing 40 Helper nt nism . al c at gic mi cr al ago te olo tion one tion hic tion tion re Ant Epis Et Ide Demo Size Opp sco Func Func Func func (%) Cultural x x 9.1 8.1 8.1 13.4 6.8 Amnesia Ideological ts Ideological Support ul Deliberation 1.7 30.7 36.0 52.7 27.8 Conflict es ct r Cultural Practices/X ro Deliberation 9.9 27.1 24.9 33.7 19.3 Amnesia Cultural je te p Courage and Will/X Sovereignty ed Deliberationia Threats 8.5 30.5 39.7 53.6 23.8 m Ideological Support Deliberation 2.8 34.8 53.2 53.3 30.7 s fi nt Disruptions seed a reat Historical Study Modern nt na s i er l. x Deliberation 2.0 39.1 37.7 53.4 22.0 re et p Public Engagement Distortions Agonism n X e a s ong Communication Strong er 2.3 11.8 9.1 20.5 10.9 stha Errors Agonism luay c ve c Historical Strong tiy m Historical Study 3.1 39.0 24.3 53.2 15.5 ra Distortions Agonismhe ar. T f n Government Mild ed Public Engagement 2.3 14.3 33.0 49.3 24.1at w o Oversight Agonismid ieal rv ve Historical Study Disruptions 5.4 14.8 9.8 13.9 8.1 pr d nul he sho Historical Strong . T 2.0 47.9 25.0 47.9 19.3nd s Ideological Support Modern ds ot b ea Deliberation he e t Ideological Support n v x Antagonism 3.7 10.1 21.4 19.2 5.0 ve oti ee Political ra et b Ideological Support Antagonism 4.8 22.8 41.4 62.2 31.9 pa ot y Opposition om Collective memory/ C av . Sovereignty e n Colective Empowerment Antagonism 1 20.7 14.0 25.5 42.1 15.1g. at h Threats Fith Historical Study Antagonism 3.1 21.2 45.3 58.0 23.4 Distortions Historical Historical Study Antagonism 2.0 30.7 41.3 69.6 20.6 Distortions Historical Historical Study Antagonism 3.7 34.6 37.1 59.1 22.5 Distortions Historical Ideological Support Antagonism 4.0 19.1 28.1 53.8 15.8 Conflict Ideological Public Engagement x Antagonism 1.7 19.0 37.3 51.5 24.0 Historical Study Antagonism 3.1 14.8 26.3 42.8 14.5 Errors Communication Ideological Support Antagonism 4.0 19.4 27.2 59.4 16.4 tortions Historical Dis- 41 Cluster r t t ive jec ec der ce Sub Obj Sen Re Historical Identity Cultural Heritage Historical 4 National Awareness Preservation Preservation Significance Resistance R National Historical 1 National Identity emembrance Commemoration Significance Historical 2 National Unity X National Identity Significance ; Historical Identity Historical Slovenian Social Future/ he Preservation Significance National Awareness ati on 12 Freedom and Rights s (F en Historical Identity re Historical National g le . 8 Preservation/National National Awareness ic e: m Acknowledgment Identity Crisis rt at is Resistance s a eb on f d ag ew National Unity/ Historical nt 11 National Unity Slovenian Social Future s o n n le Resistance Remembrance Significance g a n i ty on ), a National Resistance National Liberation d Interpretation Slovenian Social Future . 2 – ig nd t ib Historical National Identity/ el 3 er ha t s tr Historical Acknowl-) t Historical Identity en . 1 5 edgment/Freedom Ethical Duty National Awareness er ig Preservation iff ed – s and Rights d ; r ts m n X (F is Historical Identity Historical ligh on 7 Ethical Dutyc National Awareness ed o Preservation Acknowledgment igh ag ifi h nt nt de tely ild a Historical Identity National Historical ia 9 National Awareness re i Preservation Commemoration Significance ed s a m w – m er lo st im el Historical Identity Historical lu on 10 X National Awareness ; y Preservation Injustice uti m ve c is ti ib tr ra on is Ideological National Awareness/ ar Historical Identity Historical d 6 Agenda/Historical Slovenian Social e n ur Preservation Acknowledgment or lo ue – ag Interpretation Future M co bl Our three case studies (the Day of Resistance Against the Occupier in Slovenia; Giorno del ricordo in Italy and Slovenia; and Europe Day and the Fall of the Berlin Wall across four countries) show that digital memory does not operate in the same way in every context. The mode of de- bate depends strongly on the type of commemoration, the political sensitivity of the topic, and whether the memory at stake is national, cross-border, or transnational. The cases also differ in dataset size and in analytical emphasis (for example, in the Giorno del ricordo case we complemented the results with a questionnaire survey). 42 Hel per nt nism . al c at mi gic cr al ago te olo one tion hic tion tion tion re Ant Size Epis Et Ide Demo Opp sco (%) Func Func Func func Cultural/Collective Historical Strong 5.8 74.5 60.8 49.8 36.9 Memory Distortion Deliberation Memory Societal Division Strong Agonism/ ed 4.4 70.0 65.2 56.7 43.3 Mild Delibera Cultural/Collective iate rm Collective Solidarity Oppression & p ed a Strong Agonism/at et 2.9 65.0 70.0 72.5 48.8re Threats Mild Delibera hee t ads Foreign Occupa- Strong Agonism/ reot b Collective Solidarity 10.9 78.4 65.7 68.3 44.7 sp tion Mild Delibed n heulho Memory Societal Division 13.1 71.4 69.2 67.8 52.2 nt Mild Deliber s fi se re Cultural/Collective Strong Agonism/ s i na nte l. Memory Societal Division Strong Agonism 5.1 77.1 66.4 72.9 47.1 nd s tie a arng Cultural/Collective escl. T ap ha er Memory ew Cultural/Collective Societal Division Strong Agonism 12.4 76.9 72.8 71.8 56.7 sp ay c n y m Supportive Alliances rs . T Societal Division Strong Agonism in he 3.6 65.0 48.0 44.2 30.0 ste ed lu at Historical ve al Historical ti Strong Agonism c id Documentation 2.2 83.7 71.0 68.3 43.3 n v ra Distortion aree Supportive Alliances Historical Distor- f n et b o w tion/Oppression Mild Agonism 3.6 71.2 48.0 64.2 42.0ot y ie rve n & Threats veav Cultural/Collective e o t h iv Societal Division Mild Agonism 3.6 81.0 76.0 75.0 58.0ha at Memoryts t parul Cultural/Collective omes Historical Ct r Memory/Historical Mild antagonism 32.1 67.4 62.5 76.1 43.7 . Distortion 2ec Documentation g.oj Fipr The comparative analysis of debates around the Day of Resistance showed, among other findings, that discussions on X are considerably more dispersed and conflictual. In newspaper articles, we identified 12 narrative clusters. Most of these clusters display agonistic debate with elements of deliberation. One cluster stands out with the highest deliberative quality, characterised by above-average epistemic and ethical dimensions, while strong antagonism is rare (only one clearly antagonistic cluster in the entire corpus). On X, we identified 21 narrative clusters, most of which were characterised by an antagonistic discourse style. Nonetheless, a smaller share of agonistic debates also appears, where disagreement is sharp but still recognises the legitimacy of the other, as well as a few deliberative islands, where more justification, verifia-bility, and openness are observable. 43 7. Results of the Analyses – Key Patterns Figures 3 and 4 present online debates surrounding the Day of Resistance Against the Occupier in Slovenia, comparing X with traditional online news media. Overall, discussions on X are more confrontational: when posts slide into an- tagonism, they are often driven by ideological positioning, leaving less room for explanation, ethical reflection, and inclusive exchange. In traditional online media, the pattern is different. Especially in agonistic debates, a more balanced combination of information, value-based emphases, and democratic principles is often maintained, even when conflict is pronounced. The most visible differences, however, emerge in deliberative debates. Traditional media generally provide calmer contextualisation and argument-based expla- nation, whereas on X even deliberative islands often remain more polarised and framed through ideological disputes. To present these differences clearly, we compared the distribution of antagonism, agonism, and deliberation on X (Figure 3) and in newspaper articles (Figure 4) through the lens of four discourse functions (epistemic, ethical, democratic, ideological). 70 60 60 50 50 e cor 40 40 S ge ra 30 30 ve A 20 20 10 10 0 0 Antagonism Agonism Deliberation Antagonism Agonism Deliberation Fig. 3. Graph presenting average scores Fig. 4. Graph presenting average scores of dis- of discourse-function ratings across AAD course-function ratings across AAD clusters in clusters in the X dataset on the topic of the the online articles dataset on the topic of the Day of Resistance Against the Occupiers. Day of Resistance Against the Occupiers. Epistemic Function Ethical Function Ideologic Function Democratic Function The graphs presents intermediate project results that have not yet been validated. They may change and should not be treated as final. In our analysis of Giorno del ricordo, we were particularly interested in whether – and under what conditions – agonistic forms of debate appear on X as well as in traditional media in Slovenia and Italy. For this purpose, we developed two separate coding schemes (one for X and one for newspaper articles) and, on that basis, also designed an online questionnaire to assess agonistic potential directly – that is, beyond media-mediated content and debates on social media. A key finding from our analysis of discussions on X is that agonism – a form of conflict that does not deny the legitimacy of the other side, but enables at least a basic exchange of viewpoints – does occur. This is more visible in Italy, where the categories agonistic tone (6.3%) and strong agonism (1.9%) together account 44 for approximately 8.2% of all posts. In Slovenia, such posts account for approxi-mately 4.5% (agonistic tone), and strong agonism does not appear in this sample. This suggests that the Italian X corpus more often contains a constructive form of conflict: disagreement remains sharp, yet democratic dialogue is still possible. In Slovenia, by contrast, X posts are more often situated within an antagonistic or non-agonistic conflictual frame. At the same time, it is important to note that the Slovene X sample is very small (N ≈ 198) compared to the Italian one (N ≈ 23,000). These figures do not allow for conclusions about the overall public debate; they indicate patterns within the collected corpus. Cluster Italy (N %) Slovene (N %) Antagonism 6977 (29.8%) 74 (37.4%) Agonistic tone 1484 (6.3%) 9 (4.5%) Strong agonism 436 (1.9%) 0 (0%) Deliberation 1513 (6.5%) 3 (1.5%) Non-AAD 5577 (23.8%) 42 (21.2%) Non-agonistic conflict 7382 (31.6%) 69 (34.8%) Unknown cluster 20 (0.1%) 1 (0.5%) Share of agonistic forms of debate on X around Giorno del ricordo: Italy (agonistic tone + strong agonism ≈ 8.2%) shows more agonistic space than Slovenia (≈ 4.5%). The comparison is based on the 2022–2024 corpus; due to the small Slovene sample (N ≈ 198), the results do not represent the full picture, but provide an indicative snapshot of the analysed posts. In newspaper articles on Giorno del ricordo, agonism appears more clearly and in a more structured way than on X. If we compare the results, we see that strong agonism is substantially more common in Slovenia (44.9%) than in Italy (11.7%). This means that Slovene articles in this sample more often combine several ag-onistic signals at once: plurality of perspectives (G1), legitimising the opposing position (G2), bridges (G3), and a restrained tone (G4). In the Italian corpus, by contrast, a larger share of texts remains at the level of procedural neutrality (G4) without clear agonistic bridges (31.2%; Slovenia 9.5%) or falls into the category of no agonism (32.6%; Slovenia 10.9%). At the same time, in both countries we observe the category core without bridges (G1+G2+G4) (Italy 10.9%; Slovenia 20.4%); here, conflict is presented as legitimate, but without connecting opposing memory positions. This is precisely why agonistic bridges matter so much – yet they are not automatic. Where they do appear, debate shifts from the logic of two separate truths towards at least minimal recognition of the other side, thereby 45 7. Results of the Analyses – Key Patterns opening possibilities for dialogue. As with the X analysis, these results should be read with caution: they depend on the collected corpus and the selected time windows. They therefore provide a comparative insight into the analysed articles, rather than a comprehensive picture of the overall public debate. Cluster Italy (N %) Slovene (N %) Strong agonism (G1+G2+G3+G4) 162 (11.7%) 66 (44.9%) Strong core+bridges+plural (G1+G2+G3) 51 (3.7%) 8 (5.4%) Core+No bridges (G1+G2+G4) 152 (10.9%) 30 (20.4%) Core only (G2) 103 (7.4%) 9 (6.1%) Procedural-only neutrality (G4) 433 (31.2%) 14 (9.5%) No agonism 453 (32.6%) 16 (10.9%) Non-core (X-G2+X+X) 35 (2.5%) 4 (2.7%) Distribution of agonistic configurations in newspaper articles on Giorno del ricordo (Italy vs. Slovenia), following the G1–G4 scheme and including memory bridges. In the analysed sample, the Slovene corpus more frequently shows strong agonism (a combination of several markers), whereas the Italian corpus more often remains at procedural neutrality or shows no identifiable agonistic markers. The results apply to the collected articles and the selected time windows. For Europe Day, discourse is predominantly informative in all four languages. Posts in French have the highest share of informative content and are also the least antagonistic. Posts in Italian are likewise strongly information-oriented, but with a somewhat higher level of antagonism. Posts in Slovene stand out for the lowest share of informative content and the highest levels of antagonism and non-agonistic conflict, while posts in German fall between these two extremes and also include the highest share of deliberative posts. Non-AAD Non-ag- Agonism + Ago- Agonism Language / informa- onistic Constructive- nism + Delib-nism Antago- Deliber-ation tive conflict ness (Bridge) (weak) eration 2559 345 246 169 French 30 (0.89%) 17 (0.5%) 3 (0.09%) (75.96%) (10.24%) (7.3%) (5.02%) 1850 409 279 200 18 German 33 (1.18%) 1 (0.04%) (66.31%) (14.66%) (10.0%) (7.17%) (0.65%) 651 86 116 29 7 Italian 6 (0.67%) 0 (0.0%) (72.74%) (9.61%) (12.96%) (3.24%) (0.78%) 169 44 52 12 0 Slovene 4 (1.42%) 0 (0.0%) (60.14%) (15.66%) (18.51%) (4.27%) (0.0%) Europe Day on X: distribution of debate types (non-AAD/informative, non-agonistic conflict, antagonism, deliberation, and agonistic forms) across languages. 46 For the Berlin Wall, the pattern reverses: antagonism and non-agonistic conflict dominate across all languages, and the informative share drops substantially. Posts in German remain the least antagonistic and retain the largest inform-ative component, whereas Italian and Slovene show the highest concentration of antagonistic content. Deliberation is rare in every language. Non- Agonism + Ago- Non-ag- Ago- Language onistic nism nism Antago- AAD / Deliber- Construc- nism (+ informa- ation tiveness deliber- conflict (weak) tive (Bridge) ation) French 82 (0.51%) (39.87%) 6434 4970 4469 113 58 12 (30.8%) (27.69%) (0.7%) (0.36%) (0.07%) German 5160 5265 4482 281 43 152 (0.99%) 31 (0.2%) (33.48%) (34.16%) (29.08%) (1.82%) (0.28%) Italian 2976 1906 2122 65 21 10 (41.65%) 45 (0.63%) (26.68%) (29.7%) (0.91%) (0.29%) (0.14%) Slovene 69 40 57 (41.07%) 1 (0.6%) 1 (0.6%) 0 (0.0%) 0 (0.0%) (23.81%) (33.93%) Fall of the Berlin Wall on X: distribution of debate types (AAD) across languages; compared with Europe Day, the share of conflictual and antagonistic posts increases markedly. A comparison of the commemorations reveals clear patterns: on Europe Day, information-oriented posts predominate, whereas in discussions about the fall of the Berlin Wall the debate is far more often framed in conflictual and polarised terms, with a very small share of deliberation. WHAT WE FOUND: CORE PATTERNS Clearly named actors improve dialogue. Debates in which narrative roles are clearer and more unambiguously defined are less likely to slide into hatred. Unclear narrative structures increase emotional escalation. Epistemic quality and ethical orientation are the foundations of deliberation. Posts that include facts, historical sources, context, and a respectful tone achieve the highest deliberative scores. Agonism is not an obstacle but a driver of debate. Strong disagreement that recognises the legitimacy of the other enables lively dialogue and often serves as an entry point into deliberation. Shared goals reduce polarisation. When debates connect to broader values (democratic culture, Europe’s future, the common good), hostility decreases and willingness for dialogue increases. 47 8. recommendations our analysis – combining narrative analysis, the AAD scale, and discourse functions – shows that users and stakeholders (citizens, teachers, journalists, municipalities, museums, NGOs) can actively strengthen dialogue and reduce antagonism in the digital memory space. how to recognise a manipulative memory-related post? • Presence of an us-versus-them frame. • The text is highly emotional, without context or verifiable facts. • The event is detached from its historical background. how to recognise antagonism in comments? • Personal delegitimisation (e.g., you’re not Slovene). • Moral condemnation without arguments. • Absolutising truth (e.g., only one version is true). • Calls for exclusion or disparagement. how to encourage agonism (respectful disagreement)? • Ask a question instead of issuing a judgement. • Show understanding for the other person’s concern or experience. • Provide additional context or a source. • Avoid labelling. • Greater openness to dialogue. how to avoid emotional escalation? • Do not respond impulsively; time often softens the tone. • Separate the person from the opinion. • Identify trigger points (post-war killings, symbols) and approach them with more context. how to strengthen deliberation? • Use credible sources and historical context. • Acknowledge the limits of your own knowledge. • Create space for reflection, not only for immediate reactions. 48 conclusion 9. A democratic community is not built on a shared memory, but on a shared commitment to discuss memories openly. our research has shown that the digital space is simultaneously a space of fragmentation and a space of new possibilities. It enables the rapid spread of antagonisms, while also creating opportunities for the emergence of agonistic, and under certain conditions, deliberative forms of debate. It is precisely the tension between conflict and dialogue that determines whether digital memory functions as a source of polarisation or as a platform for reflection. The analytical approach we developed, combining narrative modelling, the AAD scale, and discourse functions, and further enhanced with LLM-assisted tools, provides a precise insight into where conflicts emerge and where opportunities for dialogue open. The methodology has proven suitable for studying different types of collective memory (national, cross-border, and transnational) and for identifying areas in which digital memory is particularly sensitive to manipu-lation, affective patterns, and algorithmic pressures. If digital memory is to contribute to democratic culture, we must understand how it is shaped by technologies, emotional responses, political interpretations, and users’ everyday interactions. Researchers, educational institutions, the me-dia, and civil-society actors play a crucial role here: data-informed tools enable a better understanding of the quality of public debates, while critical digital literacy is becoming a foundational civic competence. The methodological framework tested in our three case studies is transferable and applicable to other contexts, countries, and historical disputes where the past intersects with contemporary identities and political tensions. A democratic Democracy does not rest on a single memory, but on the capacity for multiple memories to community is not built coexist, without exclusion even in digital en- on a shared memory, vironments. If we use the digital space respon- but on a shared sibly, it can become a place where differences do not divide a community, but strengthen commitment to discuss it through reflection, dialogue, and a shared memories openly. search for meaning. 49 key takeaway What We Learned about Digital Memory memory does not only reflect the past; it also reflects who we are as a com- munity today, what values we uphold, how we understand difference, and where we recognise common ground. In digital environments, collective identities can be reshaped in real time: through viral narratives, memetics, political interpretations, and users’ every- day reactions. This is why understanding technologies, platform logics, and affective (emotional) patterns is emerging as a key component of 21st-century democratic culture. For digital memory to function as a space of dialogue rather than a battleground of conflict, we need: • Data-informed tools that detect antagonistic patterns and open space for dialogue. • Digital literacy that teaches how to distinguish between information, inter- pretation, and manipulation. • Responsible communication by all actors in the public sphere: media, teachers, memory institutions, and civil society. The analytical approach presented in this publication – combining narrative analysis, the AAD scale, discourse functions, and LLM-assisted analytics – is transferable and applicable across different contexts in which the past is mobi- lised in political conflict. Democracy does not rest on a single memory, but on the capacity for multiple memories to coexist. The digital space offers the possibility to see such coexist- ence not as a threat, but as an opportunity for learning, dialogue, and reflection. about the project SoMe4Dem (Social Media for Democracy) Horizon Europe | Grant Agreement: 101094752 Duration: 1 March 2023–28 February 2026 Call: HORIZON-CL2-2022-DEMOCRACY-01-07 Funding: European Union – Horizon Europe The SoMe4Dem project investigates how social media influence democratic processes, the public sphere, and political participation in Europe – and how digital environments can be steered towards a more open, resilient, and inclusive democratic culture. Project Coordinator: Eckehard Olbrich (Max Planck Institute for Mathematics in the Sciences) Consortium: Max Planck Institute for Mathematics in the Sciences; Max Planck Institute for Human Development; Università Ca’ Foscari Venezia; Leipzig University; Universiteit van Amsterdam; Vrije Universiteit Brussel; Karlsruhe Institute for Technology; Sciences Po; University of Potsdam; Institute IRRIS for Research, Development and Strategies of Society, Culture and Environment (Hop-on facility partner); University of Bristol (associated partner). Details about the project and partners: some4dem.eu Acknowledgements: We would like to thank Eckehard Olbrich, Armin Pournaki, Tom Willaert, Marc Tuters, Stephan Lewandowsky, and Richard Rogers, as well as all other colleagues from the www. some4dem.eu project, for their support and invaluable discussions and feedback. 51 This publication offers a short, accessible, and research-based insight into how memory is shaped in the digital public sphere and how it can be understood as an opportunity to strengthen democracy. Digital memory interweaves personal stories, media interpretations, platform logics, and political narratives. The approach presented here shows how data-informed analytics can help identify antagonistic patterns, build agonistic bridges, and open space for a more argument-based and respectful dialogue about the past.