Harmful speech, including hate speech, incitement and misinformation, is surging in both online and offline spaces. As of 2022, 54% of hate speech cases originated on social media platforms according to the Kenya National Action Plan against harmful speech as presented by the National Cohesion and Integration Commission (NCIC). This calls for an urgent need for robust monitoring across all contexts.
However, a lot has to be put up for an effective monitoring strategy. It must blend advanced tools, cross-sector coordination, and community engagement so that dangerous narratives are detected early.
This article will highlight some of the effective strategies we can employ:
Multi-stakeholder coalitions
This includes building coordinated networks across different sectors to track and report online hatespeech in realtime. For example, Kenya’s MAPEMA consortium united the National Cohesion and Integration Commission, media regulators, AI startups (Code for Africa) and youth platforms (Shujaaz Inc.) into an early-warning unit un-dco.org. that helped track online hate in real time across English, Kiswahili and Sheng languages in Kenya.
Advanced data analytics and AI
Organizations can leverage technology to continuously scan public data to monitor hateful narratives online. This includes using text-mining, network analysis and keyword filters on social media to flag spikes in toxic language. Monitoring tools like CrowdTangle, Meltwater and custom “hate-lexicon” databases can be used to automatically detect and map out harmful speech and generate real-time reports of emerging hotspots or trending hate terms, allowing analysts to zoom in on dangerous content early.
Proactive media monitoring and partnerships
Another alternative strategy is to engage traditional and digital media in the monitoring process. For example, news editors and journalists should be trained to identify and flag inflammatory reports or social media posts for review. Media councils can establish hotlines or reporting desks where audiences and journalists submit suspected hate content. By liaising closely with media houses, authorities can trace how hate speech spreads from social channels into mainstream discourse – and vice versa – ensuring a 360° view of the information environment.
Local-language and context-sensitive tools
Harmful speech often uses local dialects, idioms or coded references and in most cases, generic English-language filters will miss these vocabularies. Developing lexicons and training data in local languages is essential. KICTANet recently helped create hate-speech lexicons in six Kenyan languages including Kikuyu, Luo, Somali, Iteso, Luhya, and Swahili to identify online abuse, including gender-based harassment. Additionally, contextual awareness such as knowing the significance of local festivals, elections or land disputes helps distinguish harmless chatter from content that can inflame tensions.
Community reporting and “peace monitoring
Another strategy is empowering grassroots network groups to support activities such as training local peace committees, civil society groups and even community radio stations to report incidents that can help uncover offline hate and viral rumors quickly. The Kenya’s National Action Plan is a perfect example of community and peace monitoring tool. It explicitly advocates against Hate Speech and calls for “peace and cohesion monitors”, which includes distributing monitoring gadgets to investigators and deploying mobile teams during electioneering periods. Such monitors might log speeches at political rallies, coordinate with neighborhood watch groups, or crowdsource alerts from village elders to raise more awareness on hate speech.
Legal and policy frameworks.
Establish clear definitions and guidelines so monitors know what counts as “harmful speech” without stifling free expression. Many countries including Kenya are drafting/updating laws, codes of conduct and national guidelines on hate speech. Such policies should require both platforms and broadcasters to report on hate incidents and allow space for positive counter-narratives. Also, policy makers should create accountability structures from specialized hate-speech courts to platform transparency requirements that would help reinforce monitoring efforts.
Regular research and feedback loops.
Surveys, polls and public feedback can be used to refine monitoring priorities. For instance, in Kenya the Youth Barometer project surveyed thousands of young people in “7 hotspot” counties about their exposure to hate speech and disinformation under the MAPEMA consortium. These data helped tailor the monitoring focus and response messaging.
Actionable reporting and referrals.
Critically, monitoring must link to action. As harmful posts or programs are detected, they should be flagged to the right responders. Tech platforms and regulators should create clear channels to remove or rebut content. In Kenya’s 2022 election, analysts under the the Mapema consortium identified and reported 890 hate/disinformation incidents to Kenyan authorities during the campaigns.
Ultimately, online platforms lead in accelerating the spread of harmful speech, enhanced monitoring is essential to protect human dignity. We all share a stake in defending our shared values. Now is the time to intensify efforts to identify and counter harmful narratives.