The Bantu languages of Africa constitute a large cluster of languages sharing so many common features that their probable relationship was recognized very early. Doke and Cole summarize the history of this recognition, and of the comparative studies that grew out of it, up to 1943. Given an early start, a closely-knit group of languages, and workers of the quality of Meinhof, Meeussen, and Guthrie, Bantu linguistics has progressed to a stage equalled by few comparative fields outside Indo-European. Detailed reconstructions of Bantu phonology (including tonology and intonation), morphology, syntax, and lexicon have been possible.
The problems that remain are those of internal and external relationships. What are the subgroups of Bantu and what is the history of their development from the hypothetized ancestral language? What languages are related to Bantu, and what is the status of Bantu in any larger language grouping? Both questions together may be put as--does ‘Bantu’ exist? That is, do the languages recongized as Bantu--all and only those languages--constitute a well-defined linguistic group?
The simplest, and most common, assumptions are 1) that Bantu (possibly including some of the languages recognized as Bantu by the followers of Greenberg, but rejected by Guthrie) is a linguistic unit, with a common ancestor, Proto-Bantu, not shared with any other languages; 2) that it constitutes one division of a sub-group of Niger-Congo; and 3) that it in turn consists of two principal dialect areas, showing either an east-west division or a break between the forest languages (Guthrie's Zones A, B, and C) and the remainder. The principal problem for this view is the lack of support in shared innovation for any of the groupings it postulates. There is plentiful evidence in lexicostatistics both for the existence of Bantu as a distinct unit and for either of the internal borders postulated. There is even evidence for these in the form of numerous lexical and grammatical Isoglosses roughly coinciding at the boundaries of Bantu, and primarily lexical isoglosses internally. But there do not seem to be any cases where one of these boundaries is parallelled by an isogloss such that a clearly innovative feature occurs only on one side of the line.