Chapter 1
Introduction
Koen Bostoen and Mark Van de Velde
This second edition of The Bantu Languages consists of two parts. Part 1 contains general chapters that provide an overview of the state of the art in the study of the sound systems and morphosyntactic structures of the Bantu languages and of their classification, reconstruction and different contact situations. Part 2 contains short grammatical analyses of individual Bantu languages for which no book-size grammar is available. Together, these chapters provide a thorough introduction to the grammatical structures of the Bantu languages and to the historical evolutions that have shaped them. The focus on language structure and history means that this volume does not aim at giving an exhaustive overview of contemporary Bantu studies. Such an overview would have to include work on documentation, orthography creation, lexicography, youth languages and so on. This edition is very similar in structure and approach to the highly influential first edition (Nurse & Philippson 2003a), but its contents are almost entirely new. Some chapters from Part 1 have been thoroughly revised. Bonny Sands has renewed the chapter on the Sounds of the Bantu Languages (Chapter 3) in consultation with Ian Maddieson, the author of this chapter in the first edition. Larry Hyman has updated his chapter on Segmental Phonology (Chapter 4). David Odden and Michael Marlo have revised the chapter on Tone (Chapter 5), originally written by Charles Kisseberth and David Odden. The chapter on Derivation was renamed Word Formation (Chapter 6) after an update by Koen Bostoen in close collaboration with its primary author, Thilo C. Schadeberg. The chapter on Aspect and Tense has been revised by its first author Derek Nurse and complemented with a section on mood/modality by Maud Devos to become a new chapter on Aspect, Tense and Mood (Chapter 7). The six other chapters of Part 1 are entirely new. For Part 2, we chose to invite chapters on a new set of languages, because sketch grammars are both extremely useful and relatively hard to get published. For some of the languages included in Part 2, such as Chimpoto N14 (Chapter 23) and Pagibete C401 (Chapter 15), hardly any other published information is available, while others, such as Ngazidja G44a (Chapter 20), have a rich literature but no reference publication that provides a coherent overview of the basic grammatical features of the language. The sketch grammars of the first edition are available on the companian website of this second edition. They have not been revised for this second edition.
The remainder of this chapter is a brief portrait of the Bantu family, starting with its delimitation, number of speakers and geographical distribution (Section 1), and some of its main typological characteristics (Section 2). It finishes with a concise history of its scholarly study, including some early attempts at external classification (Section 3).
1 The Bantu languages: delimitation, speakers and geographical distribution
The most recent inventory of the Bantu languages (Hammarström, Chapter 2, this volume) lists 555 distinct Bantu languages. It uses the same language versus dialect divisions as the 18th edition of the Ethnologue (Lewis et al. 2015). Hammarström observes that a stricter adherence to the criterion of mutual intelligibility would decrease the number of Bantu languages by about 15%, to more or less 472. In order to keep track of so many languages, Bantuists make use of a referential classification, devised by Malcolm Guthrie, in which every language is identified by means of a so-called Guthrie code, which gives an indication of the language’s geographical location (see Chapter 11, Section 3.1.1). The emergence and discovery of new languages and the extinction of others are minor factors to account for, but the difficulty in drawing a discrete line between a language and a dialect is the main reason why numbers diverge considerably in earlier inventories of Bantu languages or language varieties, e.g., 440 in Guthrie (1971), approximately 680 in Mann and Dalby (1987), 542 in Bastin et al. (1999) and 660 in Maho (2003). In the latest online version of his New Updated Guthrie List, Maho (2009) inventories 950 different varieties with a unique extended Guthrie code. Many of these are recognised as different varieties of a single language, i.e., those ending in a lower-case or upper-case letter. The number of distinct varieties without such a final letter in Maho (2009) is 631 (Harald Hammarström, pers. comm.). The 555 languages listed in Chapter 2 are represented by the letter corresponding to their Guthrie zone on the map in Figure 1.1.
Figure 1.1 The Bantu languages represented by the letter corresponding to their Guthrie zone
The boundary between the Bantu languages and their closest relatives of the Bantoid family spoken in Cameroon and Nigeria is to a certain extent established by convention, rather than by a set of shared innovations that are attested in all and only the Bantu languages. According to convention, languages are considered to belong to the Bantu family if they have a Guthrie code. An area where this convention is most likely to be at odds with reality is the little-studied group of Jarawan languages of Nigeria and northern Cameroon, which lack a Guthrie code but which have been argued to be Bantu, possibly Bantu A60 (Gerhardt 1982, Blench 2015). Several studies in genealogical classification based on basic vocabulary (cf. Bastin & Piron 1999, Grollemund 2012: 349, Bostoen et al. 2015, Grollemund et al. 2015) do indeed recognise Jarawan Bantu languages (as well as certain other Bantoid languages) as being most closely related to the languages of the Mbam-Bubi group, which do have Guthrie codes (in groups A30, A40, A50 and A60), without a discrete cut-off point between Bantu and Bantoid or so-called “Narrow Bantu” and “Wide Bantu.”
As for the number of speakers of Bantu languages in Africa, Nurse and Philippson (2003b: 1) estimate that about 240 million Africans speak one or more Bantu languages, multilingualism being the rule rather than the exception in Africa. In 2003, this meant that one African out of three to four spoke a Bantu language (given a total African population of about 875 million at that time). This is more than half of all Niger-Congo speakers, which Nurse and Philippson (2003b: 1) estimate at about 400 million. Patin et al. (2017) estimate that there are about 310 million Bantu speakers. This would correspond to about one African in four, the number of Africans in 2018 being around 1.2 billion. Of the 556 Bantu languages in the 20th edition of the Ethnologue (Simons & Fennig 2017), 529 have a population estimate, whose sum is 276,513,509 speakers (Harald Hammarström, personal communication). This number could be extrapolated to about 290 million for the 556 Bantu languages. However, one could also argue for a higher contemporary number, taking into account that according to the 20th edition of the Ethnologue, the total African population including Madagascar is 929,932,101. This estimate is based on 2039 languages out of the 2178 present in Africa that have a population estimate (from well before 2018). The proportion of this estimate with respect to the actual number of 1.2 billion Africans today is 0.775. Applied inversely to the sum of Bantu speakers estimated in the Ethnologue 20, their number would amount to about 350 million today (Harald Hammarström pers. comm.), which is even more than what Patin et al. (2017) propose without explaining on what their estimate is based.
The Bantu languages are mainly spoken between Cameroon’s South-West region (4°8'N and 9°14'E) in the North-West, southern Somalia’s Barawe (Brava) area (1°6'N and 44°1'E) in the North-East and Cape Agulhas (34°48'S and 20°E), the continent’s southernmost tip, in the Western Cape province of South Africa. Their distribution area is contiguous – some very rare languages surrounded by non-Bantu languages notwithstanding – and spans 23 countries on the African mainland. In alphabetical order, these are Angola, Botswana, Burundi, Cameroon, Central African Republic, Congo-Brazzaville, Congo-Kinshasa, Equatorial Guinea, Gabon, Kenya, Lesotho, Malawi, Mozambique, Namibia, Rwanda, Somalia, South Africa, Southern Sudan, Swaziland, Tanzania, Uganda, Zambia and Zimbabwe. In some of these, such as Burundi, Malawi and Rwanda, Bantu languages are the only indigenous African languages. Long-standing Bantu speech communities are also found on the islands of Bioko (part of Equatorial Guinea), Mayotte (an overseas department of France) and the Comoros (see also Nurse & Philippson 2003b, Hammarström et al. 2017). Nurse and Hinnebusch (1993: 14) report a variety of the Bantu language Swahili spoken on the small island of Nosse-Be, off the northwest coast of Madagascar, with another pocket further down the west coast of this island. In other African countries, especially those in the northern and southern borderlands, Bantu languages coexist with languages that belong to other families, such as Central Sudanic, Nilotic, Cushitic and Omotic, which are part of the wider Nilo-Saharan or Afro-Asiatic phyla, or that are considered isolates today, such as Hadza and Sandawe (formerly considered to be Khoisan). Yet in others, such as the Central African Republic, Southern Sudan and Somalia, Bantu languages are sporadic, not to say nearly absent. All in all, Bantu is the predominant language group in Central, Eastern and Southern Africa. Other Niger-Congo languages – apart from Adamawa-Ubangi and Kordofanian – predominate in Western Sub-Saharan Africa, but have a distribution area that is about one-third to one-half of the Bantu area. Thanks to the vastness of the Bantu area, Niger-Congo is by far Africa’s most widespread language phylum.
The massive spread of the Bantu languages is striking, especially in consideration of the group’s estimated age of no more than 4,000–5,000 years (Vansina 1995: 52, Blench 2006: 126). This time depth is quite shallow compared to the 10,000–12,000 years that have been proposed for the Niger-Congo phylum (Blench 2006: 126). The geographic distribution of Niger-Congo minus Bantu is much smaller than the spread zone of its tardive Bantu offshoot. Bantu languages would have gradually split off from their closest South-Bantoid relatives in the borderland of South-Eastern Nigeria and Western Cameroon, an area of high linguistic diversity within the Bantoid subgroup of Benue-Congo, one of the major Niger-Congo branches (cf. Blench 2015). Ever since Greenberg (1972), there is great unanimity to consider this area as the Bantu homeland. It is from this ancestral homeland that the concurrent dispersal of Bantu languages and Bantu-speaking people across Central, Eastern and Southern Africa started. This phenomenon is commonly referred to as the Bantu Expansion (Oliver 1966, Bouquiaux 1980, Vansina 1995, Ehret 2001, Bostoen 2018). During their initial migration between roughly 5,000 and 1,500 years ago, Bantu speech communities not only introduced new languages in the areas where they immigrated, but also new lifestyles, in which technological innovations such as pottery making and the use of large stone tools originally played an important role, as did farming and metallurgy subsequently. Wherever early Bantu speakers settled down, they left an archaeologically visible culture (Phillipson 2005, de Maret 2013, Bostoen et al. 2015). New insights from the field of evolutionary genetics show that the Bantu Expansion was not just a spread of languages and technology through cultural contact, as was once thought (Lwanga-Lunyiigo 1976, Gramly 1978, Schepartz 1988, Vansina 1995, Robertson & Bradley 2000), but involved the actual migration of people (Pakendorf et al. 2011, Li et al. 2014, Patin et al. 2017). Moreover, Bantu-speaking newcomers strongly interacted with resident hunter-gatherers, as can still be observed in the gene pool (Destro-Bisol et al. 2004, Wood et al. 2005, Quintana-Murci et al. 2008, Verdu et al. 2013, Patin et al. 2014), and/or the languages of certain present-day Bantu speech communities (Herbert 2002, Bostoen & Sands 2012, Gunnink et al. 2015, Pakendorf et al. 2017). The driving forces behind what is the principal linguistic, cultural and demographic process in Late Holocene Africa are still a matter of debate, but it is increasingly recognised that a climate-induced crisis of the Central African rainforest around 2,500 years ago boosted the scale and pace of the Bantu Expansion (Schwartz 1992, Brncic et al. 2009, Ngomanda et al. 2009, Maley et al. 2012, Neumann et al. 2012, Oslisly et al. 2013, Bostoen et al. 2015, Grollemund et al. 2015, Hubau et al. 2015).
2 Typological characterisation
As pointed out in the conclusion of Chapter 3 and in the introduction of Chapters 4 and 9, one of the most attractive features of the Bantu family is that it allows for the comparative study of linguistic variation in a huge set of closely related languages. There is a marked typological divide between the North-Western Bantu languages and the others. The North-Western languages are spoken close to the Proto-Bantu homeland and in a spread zone called the “Macro‑Sudan belt” (Güldemann 2008) or “Sudanic belt” (Clements & Rialland 2008). They typically have dense tone systems, with an equipollent opposition between low and high tones, few or no tonally underspecified morphemes and a high number of floating tones. At the other end of the typological spectrum are the few Eastern languages that have lost tone. In languages with intermediate tonal density, many morphemes are tonally underspecified and receive their surface tone through the application of rules. Chapter 5 discusses many more typological differences between the tone systems of the Bantu languages, such as the nature of the tone-bearing unit or the way in which rules like tone spreading and tone shift work. The high amount of floating tones in the Northwest is due to the loss of segmental material, which is itself due to the existence of maximality constraints on the size of stems (see, e.g., Hyman 2004). These same constraints also explain why the verbal derivational suffixes discussed in Chapter 6 can hardly be stacked in many North-Western languages, whereas they typically can in the East of the Bantu domain, sometimes exuberantly.
The Bantu languages are well known for their rich noun class systems. They have on average about 15 noun classes. On top of those, most languages outside of the North-West also have three locative classes. The few Bantu languages that have considerably reduced or lost their noun class system are either contact languages or spoken in the North of the Bantu domain. For some of the latter, loss of noun classes has been argued to be due to contact with languages from other families, notably Central Sudanic (see Chapters 8 and 12). Another well-known characteristics of the Bantu languages is their high number of past and future tense distinctions, discussed in Chapter 7. Probably less well-known is the pragmatically conditioned freedom of constitu...