I've heard the Kenya and Nigeria story, but has anyone backed it up with quantitative evidence that the vocabulary LLMs overuse coincides with the vocabulary that is more common in Kenyan and Nigerian English than in American English?