This opinion piece is written by Yasmin Osama, PhD, a journalism lecturer and researcher focusing on media narratives, dehumanization, and the dynamics of online hate. Her work analyzes how news coverage shapes emotional perception and public discourse in conflict contexts. She is a certified trainer in countering hate speech through the RNTC Training of Trainers programme.
A child is killed.
One headline tells you who she was: her favorite color, her laugh, her unfinished homework. Another tells you only how many died alongside her. When a child dies in a conflict zone, how the story is told determines the world’s response: one is mourned as a stolen future, while the other is archived as a statistical necessity.
By the time the story reaches social media, the emotional ground has already been set. One life feels vivid and close; the other feels distant and anonymous. One demands outrage; the other becomes data.
Online hate is often framed as a product of algorithms, anonymity, and polarization. But this framing overlook something more fundamental: The conditions that make online hostility possible are produced long before content reaches digital media platforms.
By the time it appears in a comment section, the most important shift has already happened.
The narrative has done its work.
Narratives as the Architecture of Dehumanization
In my recent research, “Deconstructing Narratives: An Analysis of Dehumanization Techniques in U.S. Media Coverage of Palestinians at the Onset of the War on Gaza,” I analyzed over 500 news articles from The New York Times and The Washington Post [SS1] during the first weeks of the war, examining how media helped normalize the violence and the acts of genocide that unfolded afterward.
What emerges is not explicit hate speech, but a structured asymmetry in how humanity itself is represented. Palestinians overwhelmingly appear as aggregated figures: “dozens killed,” “hundreds injured”, “thousands displaced.” Their lives are quantified, but rarely narrated. In contrast, Israeli victims are frequently individualized through names, faces, professions, and personal histories disrupted by the war.
This is not simply a difference in style. It is a difference in recognition.
The asymmetry extends to language. The October 7 attacks were described using emotionally charged terms such as “massacre” and “slaughter.” Meanwhile, the killing of Palestinian civilians, including in hospitals, schools, and refugee camps, was often described through technical phrasing such as “airstrikes,” “operations,” or “explosions.”
When one side’s suffering is narrated as violence, and the other as consequence, the moral meaning of both is fundamentally altered.
Even when Palestinian suffering is acknowledged, it is often framed through conditions: displacement, hunger, lack of fuel, rather than lived emotional experience. Pain is visible, but not fully humanized.
What emerges is not random bias, but a patterned way of producing distance, fear, and diminished worth.
These findings are consistent with broader analyses of Western media coverage during the war. A recent study by The Intercept found that Israeli victims received significantly more individualized and emotionally resonant coverage than Palestinians, reinforcing asymmetries in visibility, recognition, and public empathy.
These dynamics align with what Robert Sternberg describes in his Duplex Theory of Hate, particularly the processes of negation of intimacy, passion, and commitment to disparage. In media narratives, these do not appear as explicit hate, but operate subtly through framing, language, and omission, shaping how audiences come to feel, judge, and respond.
These patterns often take more explicit forms of dehumanization. Mechanistic dehumanization occurs when civilians are labeled as “human shields,” a term that strips them of agency and transforms their deaths into tactical outcomes. Simultaneously, animalistic dehumanization occurs when media outlets reproduce metaphors like “human animals” without sufficient critique.
One of the most devastating manifestations of this is the “unchilding” of children, where Palestinian children are not consistently framed as children, but as “youth”, “minors” or potential threats. This weakens empathy and contributes to their treatment as security risks rather than as children entitled to protection.
While this analysis focuses on Palestinians, these patterns are not confined to this context. A broader review of the literature shows that similar narratives have preceded violence across multiple contexts, including the Holocaust, the Rwandan genocide, and the mass violence in Iraq. In Rwanda, for instance, Tutsis were repeatedly portrayed as threats and “cockroaches,” helping prepare the social and emotional ground for mass violence against them. against them. Likewise, post-9/11 representations of Arabs and Muslims often relied on narratives of suspicion and collective threat, while refugees and migrants have frequently been framed as “waves” or “invasions” rather than as individuals with lived experiences. Across contexts, such framing practices shape whose suffering is mourned, and whose humanity becomes conditional.
Emotion is the Engine, Not the Byproduct
If narratives shape perception, emotion determines whether those perceptions go viral. Across both research and practice, one insight remains consistent: people engage with narratives less through logic than through what they feel.
People do not engage with stories as neutral observers. They respond through what feels familiar, threatening, or morally urgent. A narrative that creates distance does not just inform. It numbs. A narrative that creates fear does not just warn. It mobilizes.
This is where hate takes root.
In my study, “Countering Online Hate Speech: A Qualitative Study of Digital Literacy Training Programs”, content creators across different contexts described a shared reality: facts alone rarely shift attitudes. Information is processed, but emotion is what is acted upon. Practitioners in the study noted that relying on logic alone is rarely sufficient to pierce the hardened attitudes of a polarized audience. This is because hate is rarely a product of missing information; it is a product of emotional mobilization.
From Narrative to Normalization: The Amplification Loop
Digital media platforms do not create these dynamics; they intensify them. Once a narrative enters the digital stream, repetition transforms it: a human life is first abstracted into a number, and technical language—”collateral damage,” “operations,” “targets”—replaces human loss. Gradually, shared pain is reframed as strategic necessity, and narratives are no longer questioned; they are internalized.
This normalization is fueled by the systemic logic of social media, where platforms prioritize a business model of engagement in which ensures content triggering fear, anger, or tribal threat is most likely to be amplified. Within business of outrage,” a single post or comment can rapidly escalate into a wave of hatred, a Ripple Effect that is especially perilous in societies already fractured by political, religious, or tribal tensions.
Furthermore, the accuracy of automated detection systems varies significantly across languages, contexts, and cultural cues, leading to deeply uneven enforcement. In many cases, moderation comes only after circulation; even when harmful content is eventually removed, it has often already reached thousands-sometimes millions-of users. In practice, this means that not all groups are equally protected; while some are shielded from harm, others are left more exposed, facing a greater tolerance for abusive or dehumanizing speech before any intervention occurs.
Breaking the Cycle: From Reaction to Construction
For years, the global conversation around online hate has been obsessed with platform-level solutions: better AI detection, faster moderation, and more aggressive content removal. While these are necessary “firewalls,” they are fundamentally reactive, intervening at the end of the cycle only after the harm has already manifested.
If the roots of hostility are embedded in the way we report the news, then deleting a comment on social media is like trying to put out a forest fire with a water pistol while someone else continues to pour gasoline on the trees. Detection mechanisms will always struggle with context and nuance; they can catch a slur, but they cannot catch the “emotional coldness” of a report that treats a thousand deaths as a footnote.
Addressing online hate, therefore, requires moving the point of intervention upstream into the ‘factory settings’ of our journalism and storytelling. This means asking uncomfortable questions:
- Is humanity being distributed equally? If we give five paragraphs to the life of one victim, are we giving the same to the other?
- Who is allowed to be a victim? Are we using ‘active’ language for the suffering of some and ‘passive’ language for others?
- Which emotions are activated, and for whom? Are we framing a conflict to evoke fear of a group or an understanding of a tragedy?
Countering hate is not just about deplatforming ‘bad actors’; it requires an empathy-based pivot toward narratives that sustain human recognition, complexity, and shared dignity before hatred becomes normalized.