Monitoring Online Hate Speech in South Sudan: Preliminary Findings
This post is by Hastings Surur, the Lead Researcher on the South Sudan Online Hate Speech Monitoring Project. It first appeared on the iHub blog.
iHub Research, in partnership with Search for Common Ground (SfCG) have been conducting a pilot project on monitoring online hate speech in South Sudan. The study borrows from the Umati Project that has been monitoring for peech with a potential to catalyse mass violence (dangerous speech) in Kenya and Nigeria. It is also premised on Search for Common Ground’s framework of building stronger national platforms for diverse, constructive and non-violent dialogue in South Sudan and has two-fold aims. First, to identify, monitor and map speech that could catalyze violence in the South Sudan social media space. Second, it is to disseminate the research findings and discuss response strategies for constructive peace messages against hate speech in South Sudan within the “I Love My Country”: Strategic Communications for Peace building in South Sudan” project.
The findings presented in this blog post are primarily descriptive insights into online hate speech in South Sudan for the last 2 years. This preliminary report also provides a baseline for the ongoing research as well as future research on the subject.
The methodology involves identifying and monitoring hate speech conversations/messages on the Internet (blogs, Facebook, online newspapers etc.). Two data coders have been scanning online platforms for incidents of dangerous speech and recording the speech acts they perceived to be hateful in an online database. The Umati Project’s breakdown of factors for identifying dangerous speech online is used to derive quantitative measurements to questions such as the extent to which the conversations/messages/texts collected from the Internet are inflammatory.
Before presenting the preliminary findings, allow me to draw an assumption that the ordinary man’s view in the republic of South Sudan of what constitutes hate speech. From the conversations that have been going online — where every so often some Internet users describe personal insults directed at an individual as hate speech — I can infer that these views are not far off from what the ordinary man thinks hate speech is. For the time being, you will agree with me that this inference makes sense, though not empirically grounded.
Preliminary Findings
These preliminary findings are limited only to the online space. Hate speech disseminated through radio and print media and mobile phones (e.g. SMS messages) have not been included in this study.
The metaphors expressed in the data analysed thus far characterize members of an ethnic group in non-human terms such as cancer, a virulent disease, Ebola, incurable disease etc. Words such as ‘nyagat’ (derogatory term for a traitor or rebel), ‘nyamnyam’ (a pejorative term referring to cannibalistic propensities among certain communities), ‘Aryan jeinge’ appeared several times in the speech instances collected. Words or phrases alone were not sufficient in some instances for coders to categorize the text; thus, in such instances the coders have to analyze sentences and perhaps a paragraph to arrive at the underlying meaning of the communication to code accordingly.
Blogs on the economic, political and social issues in South Sudan are normally initiated by those in the diaspora. Most reactions to posts on these blogs amount to ethnicity-based arguments, mainly from South Sudanese nationals in the diaspora in reaction to what they perceive as social, political and economic injustices in the system. 35% of these hate speech incidents coded and analysed come from anonymous commenters, while 61% come from identifiable commenters (people who use real names or pseudonyms). Anonymous and identifiable “ordinary” commenters do not receive significant observable responses from their audience compared to the bloggers who receive significant observable responses from their audience (anonymous and identifiable commenters). The bloggers have more clout than the commenters.
The issues most common in the inflammatory texts are:
- related to land
- related to federalism
- related to poor governance
- Land-based hate speech
The land issue has been a predominant theme in the hate speech instances collected and analysed so far. Speech tensions arise between commenters who support host communities whose lands have supposedly been grabbed by cattle keepers, and those who support the cattle keepers and feel that it is their constitutional right to settle in any part of South Sudan.
2. Federalism based hate speech
Federalism has been a popular demand in South Sudan, predating the southern Sudan secession calls (from present day Sudan), and even after South Sudan became a sovereign state. Again, reactions to blogs on federalism and what is taking place currently on the ground have evoked heated discussions sometimes with others theorizing the country’s 28 states through ethnic lens.
3. Reactions to poor governance
The conversation on poor governance centers around human rights violations, corruption, political rivalry between the President and designate first Vice President, reckless embezzlement, misappropriation of public funds and the national cake being accessible to a few. As supporters of the political divide engaged in these conversations, hate speech directed at the Dinka, Nuer and people of Greater Equatoria surfaced.
Poor governance has been cited as the cause of the conflict in the Republic of South Sudan and the two ethnic groups, Nuers and Dinka, feel they are victims of ‘each other’s political machination’, as in the case of the massacres that have happened in the country. The various ethnic groups subscribe more to local identities rather than to a national identity.
The social sites are seen as informal spaces for venting out anger against the system, ethnic group etc. Most of the websites monitored are run by South Sudanese nationals in the diaspora, and provide online spaces to debate on political, social and economic issues. Comment sections are largely un-moderated, which could indicate why hate speech continues to be found on the platforms. A key point to note is that some websites have their own regulation mechanisms such as not publishing any content that is vulgar or obscene, as evidenced by the complaints from some commenters to the editorial teams for not publishing their contributions.
The websites that most of our samples are drawn from are both rebel and government- affiliated. We have learned from the yet-to-be concluded research that among the ‘Internet warriors’, they also are voices of peace that continually advocate for peaceful co-existence among the country’s 64 tribes. A key point to note is that some of these ‘Internet warriors’ are teenagers who spew vitriolic convictions simply because of a conditioned hatred for another tribe.
It remains difficult to link dangerous speech found online to what is happening on the ground. However, online dangerous speech is a valuable window of insight to conversations and convictions people may hold offline. There is a lot of to be done in terms of harnessing the energy of commenters to preach unity — an issue that the civil society can deal with head-on, informed by insights from a monitoring project such as this one.
