A potential critique of this approach is that it focuses specifically on Canadian Jewish speakers and does not apply the same institutional classification to Muslim, Arab, or Palestinian speakers. This focus is intentional. During the period under study, Canada - and the world more broadly - experienced a shocking and well-documented spike in antisemitic incidents, threats, and attacks, unlike anything seen since the Holocaust. Violent episodes such as the recent attack at Bondi Beach did not emerge in isolation, but within an international climate in which antisemitic rhetoric, normalization, and justification have become increasingly visible. The central concern of this analysis is therefore whether Jewish voices are being accurately portrayed in media coverage, and whether patterns of visibility, framing, or marginalization in that coverage may be contributing to the broader social atmosphere in which antisemitism has intensified.
This inquiry does not require a comparative evaluation of how other communities are represented. The representation of Muslim, Arab, or Palestinian voices - while important in other analytical contexts - is not a relevant metric for assessing how Jewish communities are perceived, understood, or mischaracterized in public discourse. Media framing of Jews and Israelis operates independently as a driver of social meaning and moral interpretation. The question under examination is whether CBC’s treatment of the conflict, both in its international reporting and its domestic coverage, participates in shaping an environment in which Jewish humanity is flattened, politicalized, or rendered morally suspect. The representational dynamics at issue are therefore not symmetrical, and apply-ing identical analytical treatment across groups would obscure rather than clarify the mech-anisms through which media narratives may influence attitudes toward Jews in Canada and beyond.
Benefits of This Methodology:
This approach relies on a clear division of labour between computation and interpretation. The large language model was used to apply the same analytic criteria to the entire dataset, identifying patterns in framing, tone, sourcing, and narrative emphasis based on fixed prompts. The human researcher’s role was not to supervise or correct the model’s outputs, but to interpret those outputs, probe them with follow-up questions, and situate the observed patterns within a broader scholarly and journalistic context.
One advantage of this approach is consistency. Human labelers, even highly trained ones, are subject to fatigue, shifting attention, and contextual drift. The same article may be labeled differently depending on whether it is read early or late in the day, before or after exposure to particularly emotive material, or in light of evolving assumptions about the conflict. For example, a human labeler might classify a passage as “neutral” on one day and “sympathetic” on another based on subtle changes in mood, memory of prior articles, or growing familiarity with recurring narratives. Over the course of thousands of articles, these small variations can accumulate into significant inconsistency. Indeed, this is the best case scenario, assuming that the same human can label all of the data. This doesn’t account for variance between labelers, which may introduce an even greater source of variability.
By contrast, the language model applies the same criteria in the same way every time. It does not tire, recalibrate intuitively, or unconsciously adjust thresholds for what counts as “emotive,” “humanizing,” or “institutional.” Each article is assessed against the same definitions and prompts, regardless of order, volume, or emotional content. This does not eliminate interpretation, but it standardizes it. The result is a coding process that reduces variability introduced by human subjectivity and produces a stable, replicable baseline from which meaningful patterns can be identified and analyzed.
Crucially, interpretation enters at the analytical stage, not the labeling stage. The model identifies recurring narrative structures and representational patterns across the corpus; the human researcher then evaluates what those patterns mean, how they relate to journalistic practice and public impact, and how they align with the CBC’s stated standards and obligations. This separation strengthens, rather than weakens, the rigor of the analysis by ensuring that judgment is applied to results, not inconsistently during their generation.
In addition, our methodology incorporates temporal awareness. By using an LLM to evaluate the corpus as a whole, one article at a time in some cases, we were able to identify shifts in tone, emphasis, and representation that a human reader would struggle to track manually across thousands of articles. This temporal dimension allows for a richer understanding of how CBC’s narrative framing evolved throughout the course of the Israel–Hamas war and its aftermath.
Importantly, the findings of this report are not meant to be read as definitive counts of representation but as indicators of recurring narrative patterns, and the analytical significance lies in the consistency and direction of these patterns. Whether a particular voice appears 10 times or 100 times does not alter the core conclusion: across a corpus of nearly 3,000 articles, any sustained absence or marginal presence remains analytically meaningful and should be explored. When entire categories of experience appear only sporadically rather than routinely, (or, are over-amplified beyond their real-world frequency) this signals a structural pattern of representation, not a numerical anomaly.
Results:
Research Question 1: Whose Voices Were Heard?
One of the clearest indicators of bias in conflict reporting is the question of whose voices are amplified, humanized, and granted narrative authority. In the context of war coverage, the inclusion (or, indeed, exclusion) of civilian testimony is central to shaping audience empathy and moral perception. Civilians provide a moral and emotional anchor to geopolitical events; their voices convey the lived realities behind policy decisions and military actions. The ways in which media outlets select, quote, and frame these voices determine not only what is reported, but how readers come to understand the human cost of war. The CBC Journalistic Standards and Practices12 seemingly affirm this principle, committing the network to “tell[ing] the stories of… the people affected by the fighting” in their conflict reporting and to represent the perspectives of ordinary people caught in violence. Our analysis therefore takes this standard as both a benchmark and a point of comparison, evaluating whether CBC’s coverage of the Israel-Hamas war fulfils its own stated commitment to seemingly human-focused storytelling.
Of the total 2,789 articles reviewed, the LLM identified 1,238 (44.8%) that included some form of direct civilian perspective. The remainder of the coverage was dominated by institutional, military, or governmental sources, as well as third-party analysts, diplomats, and international aid organizations. When civilians are absent from more than half of the reporting, the conflict is presented primarily through policy statements and strategic analysis rather than lived human experience. This emphasis risks dulling the emotional and moral complexity of the war, making it harder for readers to grasp how events on the ground affect ordinary people’s lives.
Within the subset of articles that did feature civilians, the disparity is clear. Palestinian civilians were directly quoted in 43.3% (536) of these articles, while Israeli civilians were quoted in 17.9% (222). The remaining 38.8% (480) were either unable to be definitively classified as Israeli or Palestinian by the LLM, or were some other category of civilian (i.e., a Canadian civilian speaking about their family in Gaza, etc). In this case, Palestinian voices appeared in more than twice as many articles as Israeli ones. The gap is further widened when considering paraphrased accounts: 29.2% (814) of articles further para-phrased or summarized Palestinian civilian experiences, while only 5.7% (159) did so for Israelis. These findings indicate a strong editorial emphasis on Palestinian subjectivity, grief, and suffering, while Israeli civilian perspectives were either minimized or omitted entirely.
Moreover, the decision to quote versus paraphrase is not a neutral one. Direct quotations convey authenticity, immediacy, and emotional resonance; they invite the reader into the speaker's world and establish an authoritative connection between audience and subject. Paraphrasing, meanwhile, creates distance and often strips away the affective force of original speech, opening the subject's words up to interpretation, creative license, and potential narrative bias. That Palestinian civilians were not only quoted more frequently but also paraphrased more often than Israelis suggests that CBC coverage was disproportionately constructed around the Palestinian lived experience, with Israeli civilian realities occupying a more marginal narrative space.
This asymmetry becomes even more pronounced when viewed through a temporal lens. After December 2023, as it became clear that the war would continue and international scrutiny intensified, the LLM identified only 12 out of the 1090 articles published in this time frame - a mere 1% - as featuring any direct quotations from Israeli civilians (figure 1.0), compared to 207 quotations of Palestinian civilians.
This decline is striking given the material conditions inside Israel during this same period and the abundance of civilian experiences that could have been reported. Israeli hostages remained in captivity, border communities continued to face persistent security threats, and Israeli society was undergoing widespread trauma, disruption, and displacement. Daily rocket alerts routinely sent civilians across the country to bomb shelters, while more than 320,00013 army reservists were called up - equivalent, proportionally, to mobilizing approximately 1.3 million Canadians - forcing large numbers of families into prolonged separation. At the same time, nearly 100,000 Israeli civilians in the north and south were displaced from their homes; schools closed, towns were evacuated, and entire communities lived for months under constant threat. Israeli military data indicate that more than 30,000 rockets were launched at Israel between October 2023 and June 2025. Yet CBC’s coverage made little to no effort to document how these conditions affected ordinary Israelis or to humanize their experience of living in an active war zone. The near-total absence of Israeli civilian testimony during this period suggests not a momentary editorial gap, but a sustained narrowing of whose suffering was treated as narratively relevant, particularly when contrasted with the regular, detailed accounts of displacement, fear, and psychological toll provided for Palestinian civilians.The absence of these stories is not neutral. By omitting Israeli civilian experiences during an ongoing period of threat and upheaval, CBC implicitly framed Israeli suffering as either resolved, insignificant, or, at worst, deserved. Palestinian suffering, by contrast, remained dynamic, personalized, and newsworthy. This editorial selectivity did not simply reflect the evolving facts on the ground; it actively shaped the moral landscape of the coverage, signaling to audiences whose pain mattered, whose voices merited space, and whose lived reality could be safely ignored.
This pattern of civilian absence is mirrored by a contrasting pattern in institutional sourcing. While Israeli civilian voices were largely excluded from coverage, Israeli officials continued to appear frequently as spokespersons for events on the ground - Israeli officials were quoted in 45.8% of the articles, while Palestinian officials were only quoted in 20.3% (Figure 1.0). This reinforces the broader framing dynamic identified throughout the analysis: Israeli presence is maintained primarily through institutional and state actors rather than through lived, civilian experience.
Research Question #2: The De/Personalization of Actors: How does CBC Frame Human Identities?
Beyond whose voices are included in coverage lies a more subtle but equally consequential question: how are the actors in the conflict linguistically rendered as human subjects, or abstracted into collectives and institutions? In conflict reporting, the difference between referring to people as individuals with names, relationships, and lived experiences versus as impersonal groups or state actors has profound implications for empathy and moral interpretation. Individualized language invites readers to recognize vulnerability and shared humanity; collective language, by contrast, situates events within the logic of policy, strategy, and power, often at the expense of personal experience.
The way media outlets navigate this distinction shapes whose suffering is emotionally accessible and whose is rendered distant or procedural. When one population is consistently described through institutions and abstractions, while another is more frequently individualized, audiences are guided - often unconsciously - toward differential identification and concern. This analysis therefore asks whether CBC’s coverage of the Israel-Hamas war frames Israelis and Palestinians symmetrically as human subjects, or whether patterns of de-personalization and abstraction systematically/asymmetrically structure how each group is understood. By examining the form of reference used to describe Israelis and Palestinians, this stage of the analysis seeks to assess how language itself contributes to the uneven humanization of actors within the conflict.
