Our first measure of dehumanization concerns the linguistic form of reference. Each article in the dataset was scanned for whether it contained collective or individualized mentions of Israelis and Palestinians. The contrast was striking. In 1769 articles (or 63% of the total corpus), Israelis were referenced only as collectives. That is: “Israel,” “the Israelis,” “the mil-itary,” or “the government.” Directly individualized references - phrases such as “an Israeli father,” “a family in Sderot,” “a teacher from Tel Aviv,” - occurred in fewer than 0.5% of the corpus (11 times in 2,454 articles).

Palestinians were also primarily collective, but the gap is meaningful: their individualization rate was 91. This may not seem dramatic, but we must take into account the fact that Israe-lis are mentioned in 2,454 articles, while Palestinians are mentioned in 1521. Therefore, out of 1521 articles that mentioned Palestinians, 6% humanized them as individuals (e.g., “a Palestinian father,” “a young girl in Rafah”) while this is true for 3.26% out of the total corpus (Figure 2.0).

Put differently, CBC was more than 13 times more likely to describe a Palestinian as a person with a name, role, or lived experience than to do so for an Israeli, and Israelis were portrayed with this kind of humanity less than 1% of the time (Figure 3.0). From a discourse-analytic standpoint, this asymmetry is not merely stylistic. Collective reference activates cognitive frames of group identity and abstraction; individual reference evokes personal identity and empathy.1415 When readers encounter “Israel bombed Gaza,” they engage with a geopolitical agent, not a community of human beings. When they read “a Palestinian mother searched the rubble,” or “a toddler held hostage,” they meet a person. The overwhelming collectivization of Israelis therefore performs a form of linguistic erasure: Israelis appear as components of a mechanism rather than as moral subjects capable of feeling, grieving, or suffering.

Research Question #3: Lexical Collocation: Who Gets to be Human?

If the form of reference determines whether people appear as individuals or abstractions, the surrounding language determines what kind of meaning attaches to those references. Words do not operate in isolation; they acquire significance through the lexical environments in which they are embedded. In conflict reporting, these environments quietly shape whether readers encounter a population through the grammar of lived experience - bodies, families, fear, and loss, or through the grammar of institutions, strategy, and power. Examining which words cluster around references to Israelis and Palestinians therefore offers insight into whose humanity is made emotionally legible and whose is mediated through political or militarized frames.

This analysis builds directly on the findings of differential personalization described above by asking a deeper question: once Israelis and Palestinians are named, what kinds of narratives surround them? By examining the most frequent meaning-bearing words that appear in close proximity to “Israeli” and “Palestinian,” this stage of the analysis assesses how language constructs distinct semantic fields around each group. These collocation patterns reveal whether populations are situated primarily within registers of policy and action or within registers of human experience and vulnerability, further clarifying how linguistic framing contributes to uneven moral visibility in CBC’s coverage.

At first glance, both groups’ portrayed identities seem to revolve around conflict. Yet their semantic orientation diverges sharply. “Israeli” is embedded in a vocabulary of agency, militarization, and policy. Figure 4.0 identifies the 50 most frequent colocation words surrounding “Israeli,” after stop words are removed. The terms remaining in the top 50 most frequent collocations surrounding references to Israelis are dominated by institutional, military, and geopolitical language. Words such as “government,” “military,” “minister,” “defence,” “forces,” and “state” situate Israel primarily as an organized political and military actor, while adversarial entities like “Hamas,” “Hezbollah,” and “Iran” reinforce a strategic and conflict-oriented framing. Even when human-related terms appear, such as “people,” “civilians,” or “killed”, they are embedded within a broader semantic field oriented toward action, policy, and warfare rather than lived experience. This produces a narrative environment in which Israelis are rendered legible chiefly through the grammar of institutions and armed conflict, limiting the linguistic space available for civilian subjectivity or everyday life.

By contrast, the collocational environment surrounding references to Palestinians (Figure 5.0) is notably richer in terms associated with bodily harm, displacement, and humanitarian vulnerability. Words such as “killed,” “wounded,” “children,” “displaced,” “health,” “humanitarian,” “support,” and “thousands” foreground physical suffering, scale of loss, and civilian precarity. These terms anchor the narrative in the conditions of daily life under violence, making Palestinian experiences more immediately accessible as human stories rather than abstract political developments.

Taken together, these patterns reveal a consistent asymmetry in narrative framing. Israel is linguistically positioned as an agent within the machinery of war, as an entity that acts, decides, launches, and responds, while Palestinians are positioned primarily as passive recipients of violence and objects of humanitarian concern. This distinction has important narrative consequences. A semantic environment dominated by institutional and strategic language encourages procedural and geopolitical interpretation, whereas a humanitarian and corporeal lexicon invites emotional engagement and moral identification. Over time, repeated exposure to these divergent linguistic patterns shapes how readers understand whose lives are complex, vulnerable, and morally salient, and whose are rendered distant, abstract, or secondary. The result is not simply uneven attention, but an uneven distribution of narrative humanity across the conflict.

Ultimately, agency confers moral readability. The subject of a verb is perceived as intentional and agentive; the object as acted upon. When a population appears almost exclusively as an actor of violence or as a depersonalized collective, the emotional cost of its own losses diminishes in the public imagination. Conversely, when another population appears as the recurrent object of suffering, it becomes the moral focus of the story. CBC’s syntactic choices thus create a hierarchy of empathy: Israel - the monolith - does things; Palestinians - the people - have things done to them.

Research Question #4: Naming Power: Whose Lives and Voices Are Individuated in CBC Coverage?

The dehumanizing effect of this linguistic asymmetry extends beyond style. In moral psychology, empathy arises from perceived similarity and individuation: the mind connects more easily to named, detailed, feeling individuals than to abstract groups.1617 Therefore, another important indicator of dehumanization lies in the scarcity of names. Named individuals are the basic currency of humanization in journalism, and anonymity produces distance. In CBC’s reporting, Palestinian names - often accompanied by ages and familial details - appear routinely (identified by the LLM in 1114 articles). Israelis are named 1,159 times, but only 708 of those are civilians. Conversely, when Palestinians are named, nearly all (1,084) are civilians. In other words, CBC is 1.5 times more likely to humanize and contextualize a Palestinian civilian than an Israeli one. Proportionally, named Israelis are much more likely to be political or military officials or representatives of hostage families. Ordinary Israelis, the parents and children most affected by war, remain nameless.

Further, in the period before December 1, 2023 - the phase most directly shaped by the October 7 attacks - 48% of CBC articles during this time frame included at least one named Palestinian civilian, compared to only 29% that included a named Israeli civilian. This means that in the early months of the war, CBC was approximately 63% more likely to feature a named Palestinian civilian than a named Israeli civilian, even at the moment when Israeli civilian stories arguably should have been most visible. This finding demonstrates that the imbalance was not a late development, but was present from the very beginning of coverage, indicating a structural rather than purely temporal disparity in narrative humanization.

This imbalance compounds the grammatical and lexical trends identified above. The absence of names erases relational context; readers are given little opportunity to imagine Israeli lives within family, community, or ordinary routines. Without these human anchors, Israeli suffering is reframed as a feature of state policy rather than of lived experience.

Even neutral-sounding numerical disparities acquire ideological force when combined with this linguistic framing. Palestinians are quoted or paraphrased in more than twice as many CBC articles as Israelis (536 or 43.3 % versus 222 or 17.9 %). When those appearances are embedded in a humanitarian lexicon, they produce a cumulative narrative bias: volume × empathy = salience. Readers encountering CBC’s corpus over time are thus repeatedly exposed to Palestinian stories rendered through emotive, domestic, and humanitarian frames, while Israeli experiences appear rarely and, when they do, are encased in institutional language. The imbalance does not depend on overt editorial stance; it is woven into the grammar and vocabulary of reporting itself. Once established, these linguistic patterns reinforce themselves. Reporters quoting NGOs or international agencies adopt the same humanitarian lexicon (“the Gaza Health Ministry said at least X Palestinians were killed”), which in turn amplifies the imbalance. Editorial conventions, such as leading with casualty figures, further entrench the grammar of v ictimhood on one side and action on the other. Over time, this feedback loop produces what might be called semantic inertia: it becomes linguistically easier to write new stories within the same moral framework than to break it.

Research Question #5: Representation in Domestic Coverage: Who Speaks for the Community?

The Israel-Hamas war did not unfold only overseas; it reshaped social, political, and communal life across Canada. In the months and years following the attacks of October 7, 2023, Canadian Jews experienced an unprecedented spike in antisemitic incidents.

From harassment and vandalism to threats, intimidation, and violence, synagogues, schools, community centres, and even private homes became targets. This surge did not occur in a vacuum. Public perceptions of Jews and Jewish institutions are profoundly shaped by how media outlets narrate global events, assign moral authority, and frame who is seen as a victim, a threat, or a legitimate participant in public discourse. When coverage foregrounds certain voices, minimizes others, or consistently frames communities through institutional or political lenses rather than as human beings with lived experiences, it directly influences how those communities are treated in the domestic sphere.

To understand how the domestic context was constructed and framed within CBC’s coverage of the Israel-Hamas war and related events, we conducted a targeted analysis of how Canadian Jewish voices appeared within the larger dataset. This section again focuses on the question of representation: whose voices are heard, how they are framed, and what narrative roles they are assigned when the CBC situates the conflict within the Canadian public sphere. Similar to the coverage of Israel and Gaza, particular attention was paid to whether Canadian Jewish perspectives were conveyed through lived experience and personal testimony, or filtered primarily through institutional, political, or symbolic roles.

Using the same dataset of 2,789 CBC articles, we asked the LLM to identify all articles that covered domestic incidents/events (protests, encampments, etc.), and from this subset we asked it to isolate and analyze every instance of quoted speech. The extraction script identified over 41,000 distinct quotations, drawing from both direct (“…”) and indirect (‘…’) speech patterns, and attributed them to a likely speaker using nearby linguistic cues (e.g., “said,” “told CBC,” “according to,” or “from the organization”). The extraction process also incorporated contextual information: the article title, publication date, and a surrounding text window of approximately 200 characters before and after the quote. This provided enough surrounding discourse to detect affiliations, role designations, and geographic or institutional markers.

Each quote was tagged for its identity context: whether it referred to Jews, Israelis, Muslims, Arabs, or none of these, based on nearby lexical cues (e.g., “Jewish community,” “synagogue,” “imam,” “mosque,” “Israeli officials”). On this level, we found that 15,708 quotes appeared in a clearly Jewish context and 16,030 in an Israeli context, compared with 1,257 in a Muslim context and 534 in an Arab context. In other words, Jewish and Israeli identity was highly present in the coverage, but often without corresponding Jewish or Israeli voices - an imbalance that becomes critical when assessing who is granted narrative agency versus who is spoken for.