The NER task can be considered solved for English and a few other European languages given the available research outputs, tools, resources and applications involving NER for these languages. The scenario is sharply different for Nigerian and most of African languages and hence the motivation for the research reported in this paper. The paper presents an exploration of the potency of some language independent features in the recognition of the mentions of persons, locations and organizations in Yorùbá text in a supervised machine learning setup. The results are promising but as further investigations revealed, the size of the training corpus is yet an issue that needs to be addressed.