Umati Final Report Released

A project by iHub Research and Ushahidi, Umati was born out of the influence social media was seen to have had on the 2007/8 post election violence. The Umati project sought to identify and understand the use of dangerous speech in the Kenyan online space. The goals of the project were:

To propose both a workable definition of hate speech and a contextualized methodology for online hate speech tracking, that can be replicated locally and in other countries.

To collect and monitor the occurrence of hate speech in the Kenyan online space.

To forward any distress calls the Umati team came across online, e.g. on Twitter and Facebook, to Uchaguzi (www.uchaguzi.com). Uchaguzi is a technology-based system that enables citizens to report and keep an eye on election-related events on the ground.

To further education on the possible outcomes of hate speech, so as to promote civil communication and interaction in both online and offline spaces.

Umati Phase 1 ran for nine months from September 2012 to end of May 2013. The project monitored particular blogs, forums, online newspapers and Facebook and Twitter content generated by Kenyans. Online content that was monitored included tweets, status updates, comments, posts, blog entries, videos and pictures. Apart from monitoring online content in English, a unique aspect of the Umati project was its focus on locally spoken vernacular language; online blogs, groups, pages and forums in Kikuyu, Luhya, Kalenjin, Luo, Kiswahili, Sheng/Slang and Somali were monitored.

DOWNLOAD THE FULL REPORT HERE

Key findings from the report include: • The occurrence of online hate speech cannot solely be relied on as a precursor to violence on the ground. Other factors might play a stronger role in determining violent or peaceful outcomes. Instead, it appears that online hate speech could be a window into the conversations Kenyans engage in offline, and thus offer a way to understand recurring issues that need to be addressed. • Most Kenyans online prefer to converse in English, Kiswahili, Sheng or Kenyan English slang. There were very few hate speech statements purely in vernacular. • ‘KOT cuffing’ contributed to the fact that only 3% of total hate speech comments collected by Umati originated on Twitter, while 90% were found on Facebook. We use this new term, ‘KOT cuffing’, to refer to a phenomenon observed on Twitter where tweets not acceptable by the status quo are openly shunned, and the author of the tweets publicly ridiculed. The end result is that the offender is forced to retract statements or even close his/her Twitter account altogether. • There is a huge disparity between what the public perceives as hate speech and what the Umati project defines it as. From an exploratory survey, we found that the public’s understanding of hate speech includes personal insults, and negative commentary about a favoured politician/political party. The public’s understanding of hate speech is also broader than the current constitutional definition, which only takes into consideration discrimination on tribal lines. • Umati defines dangerous speech as a subset of hate speech that contains three out of the five possible calls to action, as defined by the Benesch Framework. Narrowing the definition of dangerous speech further was done in order to fit the Kenyan context. For example, stereotypical insults across tribes can amount to Benesch’s definition of dangerous speech, however, applied to the Kenyan context, such stereotyping across tribes is usually largely perceived as harmless banter. Benesch’s definition of dangerous speech was used by Umati to create a workable methodology that could be implemented in Kenya. Looking more closely at the elements of the Benesch Framework deemed relevant to the Kenyan context by the Umati project, could enable governing bodies, the general public, peace-building organisations, politicians, researchers, to accurately define and identify hate speech in Kenya, and take stronger measures against dangerous speech while protecting freedom of expression.