This paper aims at examining the function of prosodic prominence in Naija, an extended pidgin spoken in Nigeria (West Africa) by the vast majority of its inhabitants. Prosodic prominence remains a complex notion that has received great attention among scholars this last decade, particularly in Romance and Germanic languages (for a recent review, see Wagner et al. 2015) ; in contrast, it has yet been little studied in African languages. Our work is based on the Rhapsodie model, developed for the intonosyntactic modelling of French, and its functional interpretation in terms of communication devices and shaping of the informational flow of speech (Lacheret-Dujour et al 2018). Our study presents at least two major interests. First, it deals with a language that remains underexplored at the prosodic level. At the lexical level for example, Naija’s different varieties have been variously described as tonal, as stress, or as pitch accent languages (Mafemi 1971, Obilade 1976, Oyebade 1983, Faraclas 1996, Elugbe & Omamor 1991). More than eighty-five percent (85%) of lexical items are of English origin (Faraclas 1984). These English source words are used by Naija speakers who also speak any of about five hundred and fifty (550) different local tonal languages, which all have different tonal patterns (Blench 2012). The interaction of these two major sources results in a prosodic structure that is neither that of English nor of any of the local tonal languages. Despite and thanks to this complexity, Naija presents a fantastic opportunity to explore cross-language prosodic invariant features and crosslanguage interferences. Naija’s prosody beyond word level has yet never been described : the supra-lexical characteristics of prosody and its interplay with syntax in the marking of informational structure is unknown, and its place in prosodic typology remains unclear. We address the following questions: despite the diversity of its sources, is there a unified strategy to encode and perceive prominences in specific distributions, and how are these distributions correlated to informational structure ? The second major interest lies in the methodology of this study. We adapt a methodology, in terms of annotation, instrumentation, and automatic processing, developed for French, a typologically different language, to analyse the prosody of Naija, which makes it possible to highlight what is invariant and what is specific to prosodic measurement andmodelling. This study is based on twelve five minutes monologues recorded from twelve speakers representing all the regions of Nigeria, presenting a balanced sampling between the Northern and Southern parts of the country. The data were orthographically and phonetically transcribed, then syllabified semi-automatically using the SPPAS tool. The resulting Praat textgrids were then annotated for prominence by three native speakers, all specially trained linguistics students at the University of Ibadan. Their points of convergence were noted by the researchers and automatically checked, making focused words within utterances visible. It then became possible to observe the most frequent syntactic units marked by prominences: their nature, their function and their distribution in the utterance.