Kenya's migrant crackdowns rarely attract that label, and the record shows why

Anti-illegal-immigration hostility in South Africa and anti-illegal-immigration hostility in Kenya are not the same thing, and the way the world writes about them is not the same either. That is the central finding of a comparative study of media coverage, and it carries a consequence that matters well beyond the newsroom: when state-directed exclusion is filed under security or trade, it stops being given that label at all.

The study, which compares how international coverage frames nativist hostility in the two countries, finds a consistent split. South African hostility is framed as horizontal. Kenyan hostility is framed as vertical. The distinction is partly a description of two genuinely different countries, and partly a distortion with real effects.

Two directions of danger

In ordinary language, the difference is this. In South Africa, the danger to a migrant is most often a civilian crowd in an impoverished urban area, mobilised against neighbours, with the state frequently on the back foot. The research describes recurrent civilian-led crowd mobilisation in poor urban areas, with documented lethal casualties, mass displacements and what it calls recurring civilian pogroms.

In Kenya, the danger to a migrant is most often a state official with a document or a checkpoint. Vulnerability there is produced primarily by the coercive apparatus of the state, through legal encampment requirements, selective immigration vetting, arbitrary police harassment and the systemic denial of identification documentation.

Both are hostility towards foreign nationals. One arrives as a crowd. The other arrives as a form, a queue and a stamp that is not given.

The South African record

Data compiled by the Xenowatch project at the University of the Witwatersrand documents 1 295 verified anti-foreign-national incidents in South Africa between 1994 and 2025. That figure should be treated with care. The Enquirer has not yet confirmed it directly with Xenowatch, and our own workspace files record different Xenowatch totals for overlapping periods. Until that is resolved, the safest description of the South African record is one of sustained, recurring, civilian-led violence against migrants over three decades, in waves, concentrated in poor urban areas, with deaths and repeated mass displacement.

The horizontal framing has an uncomfortable corollary. It places the violence in communities and often against the state’s own attempts to contain it, which means the story is told as a failure of policing, of local leadership and of social cohesion. That is a governance problem, and it is one that badly needs administrative solutions rather than slogans.

The Kenyan record

The vertical framing produces a different story. International coverage disaggregates Kenyan nativist hostility into localised counter-terrorism overreach, regional security buffering, refugee camp management or domestic trade protectionism. Each of those categories is real, and each is a legitimate subject. None of them is given that label.

The clearest example is Operation Usalama Watch, in which Kenyan security forces conducted cordon and search sweeps in Nairobi’s Eastleigh district, detaining thousands of individuals at Kasarani Stadium and subjecting foreign nationals to bribery, extortion and summary deportations. The scale of the detentions is described in the research as running into the thousands. That figure has not been independently verified for this piece, so we describe the scale rather than assert a number.

Economic nativism follows the same pattern. President William Ruto announced a directive barring foreign nationals from informal retail and street hawking, and a proposed Local Content Bill sits alongside it. After reports of intimidation and arbitrary evictions of Burundian traders, Principal Secretary for Foreign Affairs Korir Sing’Oei visited the Burundian embassy to apologise, offer reassurances of police protection and establish administrative amnesties.

That last sequence is the most constructive material in the record. A named official, a named embassy, an apology on the record and a mechanism, however modest, for traders to regularise. It shows what administrative repair looks like when a state chooses to attempt it.

One headline that does not fit

The study’s own evidence complicates its claim that coverage systematically sidesteps the term. It records a DW headline reading: Kenya: Crackdown on foreign traders prompts anti-illegal-immigration sentiment fears.

That headline uses the word. It sits awkwardly beside the argument that the word is avoided in headlines about Kenya. The honest reading is that the pattern is strong but not absolute, that some outlets do name it, and that a single counter-example does not dismantle the broader finding so much as mark its boundary. The tension is worth stating plainly rather than smoothing over.

Why this matters here

South Africans have a direct stake in how this split is drawn. When the country’s problem is framed only as horizontal, the implication is that the state is a bystander to violence it cannot control. When Kenya’s problem is framed only as vertical, the implication is that migrants there face bureaucracy rather than hostility.

Neither framing is complete. Anti-illegal-immigration sentiment can be horizontal, vertical, or both at once, and governance is the common thread running through all of it. The countries differ in degree and mechanism, not in kind. A crowd in a township and a checkpoint in Eastleigh are two expressions of the same exclusion, and both are made worse or better by the quality of the administration that faces them.

What happens next

The Xenowatch figure needs to be confirmed directly with the project before it is published as a hard number. The Kenyan detention figures from Operation Usalama Watch need the same treatment. Until then the trend is clear even where the totals are not.

The wider question is whether coverage will start applying one word consistently to two mechanisms. The evidence suggests the answer depends less on what happens on the ground than on which desk a story lands on when it arrives.