The lexical frequency of labial-velar stops as a window on the linguistic prehistory of northern sub-Saharan Africa

Abstract

Cross-linguistically, labial-velar stops are rather rare, but they are known to be common in the phonological inventories of the languages of northern sub-Saharan Africa (NSSA) (Cahill 2008, Maddieson 2011). For this reason, labial-velar stops are usually considered to be a distinctive areal feature of NSSA (Clements & Rialland 2008, Güldemann 2008). At the same time, a cursory examination of the descriptions of the languages that have labial-velar stops quickly reveals that they can vary significantly with respect to the status of these consonants in their phonologies and lexicons. This paper presents the results of a survey of the lexical frequency of labial-velar stops in 336 languages of NSSA and discusses their implications for reconstruction of the history of the area and its languages. To the best of our knowledge, our survey is the first example of such a large-scale survey of the lexical frequency of an areal sound pattern. It was made possible thanks to the recently created lexical database source RefLex (www.reflex.cnrs.fr).We statistically analyzed the spatial patterns in the distribution of lexical frequencies of labial-velar stops within NSSA and the patterns of distribution of labial-velar stops within the lexicon. The spatial analysis of the data using Generalized Additive Modeling shows that there are three major areas with high lexical frequency of labial-velar stops within NSSA. The first two areas roughly correspond to coastal West Africa with the divide between them centered around the Dahomey forest gap. The third area corresponds to CAR & northern DRC and is separated from the closest West African area by a major discontinuity in Cameroon and northeastern Nigeria. When considered against the geography of NSSA, this spatial distribution suggests that the two areas are hotbeds not so much for spread but for retention of labial-velar stops, with the hotbeds arguably correlating with higher incidence of language shif