This report is part of a broader initiative to analyze online discourse and specifically monitor hate speech in Kenya in the lead up to the general elections in order to help mitigate the threat of hateful language in fueling violence on-the-ground.
Hate speech in this context is defined using section 13 of the National Cohesion and Integration Act, where the threshold is underlined by proof that the speech or expression threatens, abuses or insults others based on their ethnicity, and must be intended to stir up ethnic hatred, OR, under section. 62, inciting ethnic hatred, hostility or violence.
The report also acknowledges the role of “dangerous speech,” which is a heightened form of hate speech that can catalyze mass violence. The report provides an analysis of digital and social media discourse related to the electoral process over a 30 day period, including search trends on digital platforms, social media conversations and popular digital news content. Content is tracked using different newsgathering tools. Twitter is monitored through its API, Facebook and Instagram are monitored through the Crowdtangle platform, WhatsApp insights are gathered through anecdotes from community networks and from third party fact checking reports. Telegram is monitored via TGStat API, while TikTok content is scraped directly from the platform. The resulting monitoring loop (keywords and selected tools) allows to track relevant online discourse data including social media posts, comments or articles.
Data collected is analyzed to identify harmful content and developing themes or trends related to the elections and is analyzed for volume, engagements generated, networked patterns, sentiment of conversation user comments and reactions. Where sentiment and tone analysis is included, it is gathered by humans.