The report concludes with a set of national and regional recommendations aimed at strengthening
responses to online hate speech and gender-based violence, including actions that can be taken by
governments, political parties, technology platforms, and civil society organizations.
Main Insights and Key Findings
Quantitative Insights
● Neutral discourse remains dominant but is gradually declining. Neutral content still
represents the largest share of online discussions; however, its share declined over the
monitoring period from 57% to 50% in the Arabic dataset, indicating a shift toward more
confrontational and harmful discourse as the elections progressed.
● Patterns of harmful speech differ across language contexts. The Arabic-language dataset
shows higher levels of harassment (18.5%), primarily driven by political harassment and
sexual insults. In contrast, the Kurdish-language dataset shows lower harassment (4.3%) but
higher levels of discriminatory narratives (17%), particularly accusations of treason (10.4%)
and dehumanizing language (5.1%).
● Gendered and sexualized narratives appear across multiple forms of harmful speech. Sexual
insults constitute a large share of harassment in both language contexts, accounting for 83%
of harassment cases in Kurdish content and 48% in Arabic content, indicating the persistent
gendered nature of electoral hostility toward women candidates.
● Violent discourse remains limited but highly severe. Although violence accounts for only
1.2% of Arabic content and 0.7% of Kurdish content, it largely involves the most extreme
forms of online harm, including incitement to kill (84% in Arabic) and sexual violence
discourse (66% in Kurdish).
● Hostile narratives intensified as the elections progressed. Across both language datasets,
disagreement, insults, and discriminatory narratives increased over time while neutral
discourse declined, suggesting that the electoral context contributed to escalating hostile
online behaviour.
● Electoral discourse is largely driven by official political actors. Political figures and media
accounts dominate the conversation across platforms, while influencers and content
creators account for 3% or less of the activity, indicating limited organic participation in
election-related discussions online.
10