Show Summary Details

Page of

PRINTED FROM OXFORD HANDBOOKS ONLINE ( © Oxford University Press, 2018. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in Oxford Handbooks Online for personal use (for details see Privacy Policy and Legal Notice).

date: 18 December 2018

Niger-Congo: A brief state of the art

Abstract and Keywords

This contribution provides an overview of the research history and contested classification of Niger-Congo, one of the world’s largest phyla. Despite convincing evidence for the genetic coherence of core families, some groupings traditionally seen as Niger-Congo need to be considered as independent families. Recent comparative research on individual families shows systematic correspondences between many grammatical morphemes and suggests that the phylum’s typological heterogeneity is based largely on regional diversification. Although features such as ATR-based vowel harmony, noun classes, verb extensions, serial verb constructions, and constituent order variation are very characteristic of many Niger-Congo languages, their original shape cannot yet be convincingly reconstructed. To better understand the processes that led to the present picture, Niger-Congo studies will have to rely largely on bottom-up approaches such as reconstruction of lower-level units, research on the sociolinguistic history of single languages or clusters, and interdisciplinary approaches to individual linguistic regions and expansion areas.

Keywords: Niger-Congo, classification, African linguistics, comparative method, typology, noun-class systems, verb extensions, serial verb constructions, numeral classifiers, case

1 Introduction

The Niger-Congo phylum is one of superlatives. It is traditionally seen as consisting of more languages—roughly 1500—than any other phylum in the world and covering more than half of the African continent. Its genetic coherence is still contested, but most Africanists consider it—at least in its core—a coherent unit whose hypothetical proto-language may be reconstructed once more individual languages are appropriately described and the composition of its many subfamilies and its internal relations have been studied in greater detail.

Much of the discussion over the past decades concerning Niger-Congo has centered around the inclusion or exclusion of specific languages or language groups. With his method of mass comparison, Joseph H. Greenberg (1963) arrived at a genetic grouping consisting of six primary branches: Atlantic, Mande, Gur, Kwa, Benue-Congo, and Adamawa-Eastern. These in turn were assumed to form a coordinate branch with a group of languages spoken in Sudan and referred to as Kordofanian. The resulting phylum, termed Niger-Kordofanian by Greenberg (1963), was renamed Niger-Congo by Williamson (1989).

Greenberg’s mass comparison method was intended to probe remote relationships rather than to formally provide proof of genetic relationships. It consequently fell upon comparativists to try to adduce additional proof for his hypotheses. Below, we discuss some of the successes and failures of these scholarly attempts. We take a somewhat conservative view in that we exclude Mande and (most of) Ubangi, referred to as the “Eastern” branch of “Adamawa-Eastern” in Greenberg (1963), from Niger-Congo. Moreover, in view of the paucity of supporting evidence, we also exclude from the Niger-Congo family specific language groups treated by Greenberg as part of the Benue-Congo branch, such as Ijoid, or part of the Gur branch, such as Dogon.

Section 2 discusses the background to this stance as well as issues concerning the subclassification of Niger-Congo languages. Section 3 provides a summary of contested families within Niger-Congo, and section 4 contains a brief overview of more recent data on the typological profile of the phylum. Section 5 offers a brief prospect on the future of Niger-Congo studies.

2 The classification of Niger-Congo in a historical perspective

When European scholars “rediscovered” the comparative study of languages during the nineteenth century (after Arabic and Jewish scholars of the Middle Ages had already identified a Semitic family), African languages became a subject of historical comparison. Among the best known scholarly efforts from this period is the Polyglotta Africana by Koelle (1854), a German missionary working for the Church Missionary Society, who began a multilateral comparison of over 150 African languages. Koelle collected word lists and grammatical constructions among Recaptives in Sierra Leone (see Hair 1963). On the basis of his mass comparison he was able not only to identify groupings of similar-appearing languages but also to correlate linguistic structures (such as prefixed noun classes and consonant mutation) with areal patterns of distribution. Koelle suggested that Atlantic, Mande, and other West African linguistic groups can be distinguished from one another, and that some of these groups, such as Atlantic and Bantu, share significant lexical and grammatical features.

In another major contribution, Diedrich Westermann (1911) proposed a division between Western and Eastern Sudanic; many languages of the latter grouping formed the basis for a genetic unit established by Greenberg (1963), Nilo-Saharan. Westermann (1927) established six subgroups within his Western Sudanic family, which would form the basis of Greenberg’s Niger-Congo—namely Kwa, Benue-Cross (now part of the Benue-Congo branch), Togo-Restsprachen (now part of Kwa), Gur, Atlantic, and Mande.

Benue-Congo languages were among the first of the Niger-Congo group to be studied by linguists, who soon realized the family’s cohesiveness. These languages can now be shown to exhibit morphological markers whose original form can be partly reconstructed by applying the comparative method (as further discussed in section 3). At the beginning of the twentieth century, however, the historical processes underlying the development of complex morphologies, as manifested in noun-class systems and verbal morphologies still prevailing in various branches of Benue-Congo, were also interpreted as instances of language mixing within the framework of the so-called Hamitic theory. According to this model, which mirrored evolutionary and racist concepts prevailing in early African linguistics in Europe, Niger-Congo languages (usually referred to by the label “Nigritic”) were originally isolating languages lacking any type of nominal categorization. Noun classes were assumed to have emerged after contacts with speakers of inflecting gender-marking languages, the “Hamites” (Meinhof 1911a, 1911b). Even though Lepsius was able to provide convincing arguments against this model as early as 1880, the Hamitic hypothesis continued as a leading paradigm well into the twentieth century.1

Interestingly, the main protagonist of the Hamitic model, Carl Meinhof, was also the first linguist to apply the comparative method in African linguistics, focusing on Bantu—in terms of the ideology of the time, the prime example of the mixture between Hamitic and Nigritic. Whereas Bleek (1862), as well as other scholars, had identified a genetic grouping that he referred to as Bantu (after the plural form for ‘human beings’ in many of its extant members), it was Meinhof’s (1899) work that demonstrated that Proto-Bantu could be reconstructed by the comparative method. And it was Meinhof’s student Westermann who identified the core of the phylum that is now called Niger-Congo (Westermann 1911, 1927). Westermann (1927) was also one of the first scholars to observe that the similarities shared by the languages of the Sudanic Belt suggested a considerable time depth for the emergence of their linguistic groupings. However, rather than claiming that the relationships among these languages could be explained on the basis of established methodologies such as the comparative method, as Meinhof had done with Bantu languages, Westermann suggested that they shared a common substrate, in the sense of a common areal history and shared concepts.

This concept of the historical development of these African languages was challenged by Greenberg, who published a series of articles on the genetic classification of African languages between 1949 and 1954 (reprinted in Greenberg 1955). In this collection of comparative studies, Greenberg divided the languages spoken on the African continent into sixteen distinct groups. He did this by debunking the Hamitic hypothesis, thereby removing the misleading “noise” provided by racial and cultural considerations, and by eliminating language typology from his historical comparisons. In principle, this approach to linguistic classification meant developing another, “wide angle” perspective by comparing a large number of languages rather than proceeding pairwise in a step-by-step fashion; Koelle (1854) had applied the same approach. Because Greenberg’s method involved not the development of new techniques but rather serendipity, his new hypothesis of the genetic classification of the languages of Africa awaited final proof.

According to Greenberg (1963), Benue-Congo constituted one of the six primary branches of Niger-Congo, next to West Atlantic (renamed Atlantic at a later point), Mande, Gur, Kwa, and Adamawa-Eastern. This concept of a Niger-Congo phylum basically became the reference model for all subsequent comparative work on these languages, often taken as an established fact rather than the hypothesis that it was.

3 The contested subclassification of Niger-Congo

Much of the debate around Greenberg’s (1963) genetic classification centered around the internal classification of Niger-Congo and whether specific branches should be “cut out.” Modifications of Greenberg’s classification in subsequent decades are due partly to new descriptive contributions and partly to re-evaluation of some of his proposed evidence for the phylum’s internal division. An example of the latter is the reclassification of Greenberg’s “Eastern Kwa” as West Benue-Congo. One of the most influential revisions of Greenberg’s classification so farwas the reclassification of Niger-Congo’s largest family, Benue-Congo, by Bennett and Sterk (1977). These authors intended to demonstrate that Greenberg’s proposal to separate Eastern Kwa from Benue-Congo is not supported by comparative evidence. As an alternative, they proposed four isoglosses that divide Western Kwa (which they now called New Kwa) from Benue-Congo (which now included Eastern Kwa, hence “New Benue-Congo”), namely:

  1. (1) Niger-CongoA brief state of the art

The new classification of Benue-Congo consisted of the former (Eastern) Kwa groups Defoid, Edoid, Nupoid, Idomoid, and Igboid plus “Old” Benue-Congo Platoid (with Kainji and Jukun), Cross River, and Bantoid. This was the last widely accepted reformation of Greenberg’s grouping; subsequent decades saw an “obsession” with subgrouping, unfortunately without many lasting effects (see Bendor-Samuel 1989 for overviews).

The continuing reassessment of the phylum’s internal structure is due largely to the preliminary nature of Greenberg’s classification, explicitly based as it was on a methodology that doesn’t produce proofs for genetic affiliations between languages but rather aims at identifying “likely candidates.” However, as descriptive work on individual Niger-Congo languages has continued, the links between Niger-Congo groups have turned out to be less clear than previously thought, and evidence for regular sound changes that would allow a solid reconstruction is still lacking. As McMahon & McMahon (2008: 280) argue with respect to the comparative method, “[l]inguistic trees have typically been drawn by an individual linguist in a way that fits the classification he or she believes in, often rejecting many trees that are different but equally probable….”

Niger-CongoA brief state of the artClick to view larger

Figure 1: Subclassification of Niger-Congo after Williamson (1989).

An example will illustrate how this problem applies to Niger-Congo subgroupings as proposed by different authors.2 Williamson (1989: 21) presents a reformed subclassification for Niger-Congo, shown in Figure 1.

The only comprehensive overview on Niger-Congo published so far, Bendor-Samuel (1989), was organized following this model, which remains an important working hypothesis. But it is not quite clear what this subgrouping is based on. It treats Mande, for example, as a coordinate branch of Atlantic-Congo together with Kordofanian. Whereas in the case of Mande there is virtually no grammatical evidence supporting this claim, such evidence can in fact be adduced from languages that Greenberg (1963) classified as Kordofanian. In the case of the latter grouping, this evidence applies not only to the noun-class system but also to the verbal morphology, as discussed below in section 4.3. Such morphological evidence is likewise lacking from Ijoid, a group of languages spoken in the Niger-Delta (Nigeria) that are proposed to form a genetic unit with the Defaka language. Other proposed modifications of this Greenbergian family tree include the creation of a “Volta-Niger” subgroup, which in Williamson’s model is still part of (New) Benue-Congo. Without further comprehensive comparative research, however, no convincing alternative family tree can be presented. Typological convergence suggests a common history of Ijoid, Defaka, and adjacent language groups in the sense of shared multilingual repertoires and contact, but this constellation does not provide any basis for the proposed subclassification, as it is (still) impossible to distinguish borrowings from inherited items. Given the paucity of grammatical evidence (consisting essentially only of a nominal prefix a- that has been assumed to be cognate with the plural noun-class prefix a- in Niger-Congo), Dimmendaal (2008) prefers to treat Ijoid plus Defaka as an independent language family whose wider genetic affiliations need to be reconsidered.

Similar questions concerning genetic affiliations may be raised about the Dogon cluster in Mali, which Greenberg (1963) had placed within the Gur family. Bendor-Samuel et al. (1989) also treat this cluster of languages as a member of the Niger-Congo family but contend that it “is better … to treat Dogon as an isolate within Volta-Congo until further evidence clarifies its status” (p. 169). Moreover, some presumed members of the Dogon group, such as Bangerime, have now been demonstrated to be linguistic isolates and not part of Niger-Congo (Blench 2006).

Ubangi is another problematic hypothetical member of Niger-Congo. Greenberg (1963: 9) grouped presumed members of this branch, called the Eastern group in his classification, into one subgroup together with Adamawa. But as argued by Moñino (1988, 1995), Ubangian may not constitute a genetic unit. Schadeberg (1989) points toward formal similarities between the noun-class system of one group of languages classified as Eastern (i.e. Ubangi) by Greenberg (1963), the Mba group, and noun classes in Kordofanian (see Table 1). But for the remaining proclaimed “Ubangian” languages, there is no such morphological evidence.

Table 1: Noun classes in Niger-Congo1



4 pl of 3


6 pl of 5




















tree names

head, name










“egg, head”










egg, head, name










egg, head, name


NP mu-






PP ju-








egg, name


  • Mba (Ubangi)

  • (Carrington 1949)













(1) Table adapted from Schadeberg (1989: 72). Numbers in the table and in interlinearizations of examples further below refer to noun classes, which are consecutively numbered following the system commonly used for Bantu.

Atlantic has been a contested concept since the seminal contribution of Sapir (1971), who referred to it as an entity at least equal to “Kwa plus Benue-Congo and perhaps Gur; nothing less” (p. 52). The internal diversification within Greenberg’s West Atlantic (or Atlantic in current terminology) indeed is so huge that some scholars would argue that it is primarily an areal grouping representing a number of independent, early descendants of Niger-Congo. Moreover, apart from Klingenheben’s (1924/25) and De Wolf’s (1985) contributions, not much historical-comparative work has been done on these languages, and with good reason. Its current classification is based largely on lexicostatistic counts (Sapir 1971, Pozdniakov 2009), and in view of the lack of reliable reconstructions, also illustrates mainly the areal relationships these languages share. Work on grammatical features such as the noun phrase demonstrates their striking diversity, suggesting that the comparative method won’t yield any convincing immediate results. A recent re-evaluation of the possibilities of reconstructing the genetic ties between Atlantic units consequently resulted in the suggestion that Atlantic be treated as an areal term and no longer as a genetic concept (Lüpke, forthcoming). For instance, Bijago, a cluster of underresearched languages spoken on the Bissagos archipelago, may represent an early offshoot from Benue-Congo, and the Bainounk cluster (representing the firstcomers’ languages in the Casamance of southern Senegal) within Niger-Congo may be of equal internal complexity.

Niger-CongoA brief state of the artClick to view larger

Figure 2: Subclassification of Benue-Congo after Blench (2006).

But even the internal classification of relatively “safe” families such as Benue-Congo is at times highly problematic. Figure 2 shows a recent suggestion for the internal classification of Benue-Congo.

Several of its (presumed) subgroups have been investigated by way of the historical-comparative method, for example, with respect to noun-class systems (De Wolf 1971) or the development of person marking (Babaev 2008).3 But reconstructions at deeper levels are still missing in many domains. Moreover, recent, groundbreaking work on Kainji languages (Blench & McGill 2012) suggests that the entire picture of Proto-Benue-Congo will change significantly (making it look less “Bantu”) once those diverse and typologically fascinating languages have been subject to more detailed comparative work.

Niger-CongoA brief state of the artClick to view larger

Figure 3: Distribution of Benue-Congo.

Jukun, on the other hand, is a group within Benue-Congo that appears to be rather coherent, both from a comparative perspective and in terms of its geographic distribution in northeastern Nigeria and adjacent parts of Cameroon (Figure 3). Nevertheless, the group’s internal classification, consisting of “Central Jukunoid” (Northern, Wurbo) and “Southern Jukunoid” (including Yukuben-Kuteb), is based entirely on typological features (noun-class suffixes vs. prefixes; after Welmers 1968 and Shimizu 1980) that do not even rest on a sound analysis of the historical processes of morphological change in the individual languages of both proposed branches (Lüpke & Storch 2013). Moreover, the classification ignores the complex sociolinguistic history of the group, which would help to shed light on fairly recent convergence processes that might be the actual reason for observed similarities within single branches. An adequate internal classification and eventually a reconstruction of Jukun will therefore be feasible only once more is known about those convergence processes.

Other branches of Niger-Congo, by contrast, became more widely accepted and have come to appear more coherent after the descriptive situation for their extant members improved. For instance, Volta-Congo as a genetic subgrouping within Niger-Congo (see Figure 1) goes back to Stewart (1976), who proposed this macro-group as the common ancestor of the Benue-Congo and Kwa languages.4 Both subgroups have been shown to share significant parts of their lexicon as well as grammatical morphemes. But the presence versus absence of nasality in noun-class prefixes in Volta-Congo and beyond presents one historical-comparative problem for this grouping. Greenberg (1963) argued that the occurrence of homorganic nasals in noun classes 1, 3, and 4 (1, 3 *u- > *mu- and 4 *i- > *mi-) was a shared innovation of the Bantu branch within Benue-Congo. But in an exemplary and scholarly precise study based on an extensive survey of the literature on virtually all Benue-Congo and Kwa subgroups, Miehe (1991) demonstrated that the same nasals occur outside of Bantu in Benue-Congo and in Kwa alike. She consequently showed that prefix nasals are old, even though there has been a gradual erosion or dismantling of the nasal affixes.

In a parallel collection of studies on noun classes in Gur languages, Miehe & Winkelmann (2007), different authors showed that several noun-class markers with nasals can in fact be reconstructed for Proto-Gur. The origin of this phenomenon and its possible link with nasality as a suprasegmental feature of vowels in Niger-Congo languages remain interesting puzzles for future research. Newly available comparative morphological evidence suggests that there are indeed ancient links between Benue-Congo, Kwa, and Gur. More recently, following up on Jungraithmayr’s (1980) work pointing toward closer links between Gur and Adamawa within Niger-Congo, Kleinewillinghöfer (forthcoming) demonstrated that the core of the phylum clearly also includes Adamawa by providing substantial evidence linking Adamawa languages with Gur (cognate noun classes between Central Adamawa and Gur) and Benue-Congo; also see Kleinewillinghöfer 1996.

Further evidence for the composition of Niger-Congo comes from Kordofanian. Greenberg (1963) assumed that the Kordofanian branch consists of five subgroups: Koalib, Tegali, Talodi, Tumtum, and Katla. According to Schadeberg (1981), however, Tumtum (referred to by Schadeberg as the Kadu(gli) group) should be excised from Kordofanian and Niger-Congo. Blench (2014) and Dimmendaal (2014) have argued that the four remaining groups do not form a genetic unit. One of the few obvious lexical cognates in a basic word list for the extant Kordofanian members is the word for ‘meat’. Because most of the other lexemes do not match up, its distribution is probably the result of borrowing. The paucity of lexical evidence for Kordofanian as a genetic subgroup is paralleled by a lack of grammatical evidence (Dimmendaal 2014). Table 1 shows the noun-class markers that Schadeberg (1989) lists for Kordofanian; these are based on reconstructed forms for Heiban (referred to as Koalib in Greenberg 1963) and Talodi.

But as the table shows, these markers reflect widespread Niger-Congo forms and therefore probably common retentions. Interestingly, the Katla and Rashad group within Kordofanian (as a primarily areal grouping within Niger-Congo) matches up both grammatically and lexically with Benue-Congo rather than with Heiban and Talodi (Dimmendaal 2014). This correspondence applies to a range of noun classes reconstructed for Proto-Benue-Congo by De Wolf (1971), with clear-cut cognates in Rashad (called Tegali in Greenberg 1963) and Katla that do not have cognates in Talodi and Heiban: *ka, *ku (both of which are singular or singulative markers), *I (plural marker),*bu (for abstract nouns), and *lu. As we will show in section 4.2, additional cognate morphemes between Katla and Rashad, on the one hand, and Benue-Congo, on the other, can be found in verbal morphology, again suggesting a closer genetic link among these languages rather than with the remaining Kordofanian groups Talodi and Heiban.

As the brief survey above shows, only detailed historical-comparative research for lower-level Niger-Congo offshoots will lead to further evidence for the inclusion of specific groups and for their genetic positions within the phylum as a whole.

4 Typological variation

One of the key elements in Greenberg’s historical approach was a clear division between evidence leading to a genetic classification of languages and evidence leading to a classification of languages along typological criteria. While Niger-Congo languages share widespread morphological properties, there is also tremendous variation. Niger-Congo languages are often stereotyped as agglutinating languages, but this typological characterization holds true only for part of the phylum. In fact, the surprising diversity of Niger-Congo is due partly to a trend toward morphological reduction in different subbranches, as well as complex rebuilding processes, and partly to the sociolinguistic profile shared by most languages of the Niger-Congo core areas, namely stable multilingualism and convergence phenomena. The following brief survey of phonological, morphological, and syntactic properties illustrates some of the variation found.

4.1 Phonology

As is true for many languages south of the Sahara, most Niger-Congo languages are tonal; the exceptions are languages spoken in geographically peripheral zones such as the Atlantic languages Wolof and Sereer in Senegal or Eastern Bantu languages like Swahili in Kenya and Tanzania. In those languages, contact with (nontonal) Afro-Asiatic languages may have been one cause of the emergence of stress as an alternative prosodic system.

Niger-CongoA brief state of the artClick to view larger

Figure 4: Distribution of subgroups with complex tone systems.

Tonal contrasts in Niger-Congo languages vary between two (with downdrift and downstep) and five (Wedekind 1985, Clements & Rialland 2008). Highly reduced segmental structures and loss of segmental morphology correspond to or are compensated for by additional tonal registers, particularly in the geographically central areas of Niger-Congo, where Gur, Kru, Kwa, and Benue-Congo languages are spoken (see Figure 4). In the following examples from the Gur language Wobe (Bearth & Link 1980: 151), the digits correspond to the respective tone levels (1 as the lowest, 4 as the highest):

  1. (2) Niger-CongoA brief state of the art

Whereas four points of articulation (labial, alveolar, palatal, and velar) are common, especially with stops, Niger-Congo (“Kordofanian”) languages in the Nuba Mountains in Sudan also have dental stops, a feature also common in Nilo-Saharan languages in the area. One “Kordofanian” language, Katla (Hellwig, to appear) combines this areal feature with another characteristic, labial velar stops, an areal feature extending into Nilo-Saharan and Afro-Asiatic languages across Central and West Africa:

  1. (3) Niger-CongoA brief state of the art

Southern Bantu languages bordering on Khoisan (as an areal grouping) borrowed clicks from the latter. Such phonological developments are, however, not necessarily the result of areal diffusion. For instance, some Igboid languages have bilabial and postalveolar clicks /ʘ, !/ (Ikekeonwu 2007); a study of the phonology of Ninkyob, a Platoid language from Nigeria, illustrates how these phonemes emerge out of stop–nasal sequences (“explosive bilabial nasals”) through language-internal dynamics (Harley 2012).

It is common to distinguish between voiceless and voiced plosives in Niger-Congo; in addition, (voiced) implosives are widespread (Clements & Rialland 2008: 550–560). As this three-way contrast is common in geographically peripheral (and discontinuous) zones of Niger-Congo, such as Atlantic or Niger-Congo languages in the Nuba Mountains in Sudan as well as southeastern Bantu, it is likely to go back to their common ancestor.

Clements & Rialland (2008: 58) further point out that such contrasts are less common in an area ranging from the Ivory Coast to western Nigeria. Instead, a contrast between fortis and lenis (or “nonobstruent”) stops is attested. For example, the Kwa language Ebrié (Cama), which is spoken in and around the major port city of the Ivory Coast, Abidjan, has a four-way contrast between “fortis” /p, b, t, d/ and “lenis” /’p, ’b, ’t, ’d/.5

Vowel systems vary rather dramatically not only within Niger-Congo as a whole but also within subgroups. Within Volta-Niger, for example, the number of vowels varies between five (as in other subgroups, such as Kainji or Platoid languages bordering on Chadic) and ten (as in Edoid, Nupoid, Idomoid, Igboid, and Bantoid). Systems with nine or ten vowels usually involve Advanced Tongue Root (ATR) harmony, a phenomenon also found in Nilo-Saharan and neighboring languages belonging to the Chadic, Cushitic, and Omotic branches of Afro-Asiatic; see Casali (2008) for a detailed discussion. At least one Kordofanian language with an ATR system has twelve vowels, namely Tima, an endangered member of the Katla group spoken in the Nuba Mountains in Sudan:

  1. (4) Niger-CongoA brief state of the art

Similarly large vowel systems with ATR harmony are found in West African Niger-Congo groups, for instance, in several Kru languages (Marchese 1989: 128).

The high degree of variation even within lower-level subgroups suggests that vowel systems may change rapidly and also that ATR harmony may be both acquired and lost rapidly under contact conditions. Whereas Proto-Bantu had a seven-vowel system (Meeussen 1967), a system still common in many Bantu languages (with the innovation of a five-vowel system in particular in Eastern Bantu), languages in the northern “Bantu borderland” have developed a nine-vowel system with ATR harmony. Kutsch Lojenga (2003) gives examples of this in Bila, where verb-final vowels vary between -a and -o depending on the harmony set to which the preceding root vowel belongs:

  1. (5) Niger-CongoA brief state of the art

Again, only reconstructions within lower-level units within Niger-Congo can answer the question of whether the widespread system of ATR-based harmony is old or due to areal diffusion. This same question applies to the status of nasalized vowels, another widespread areal phenomenon. These are common in Niger-Congo language groups in West Africa, such as Kru, Kwa, and Gur; in Jukun, Platoid, and Kainji (Benue-Congo); and in neighboring Mande languages (Bole-Richard 1985). Although it has repeatedly been suggested that Proto-Niger-Congo had symmetrical sets of ten oral and nasal vowels (Williamson 1989: 23 ff.), there is again ample evidence for both the emergence and decay of nasalized vowels.

4.2 Class marking

In a wide range of Niger-Congo languages, nouns are prominently characterized by the occurrence of noun classes and their often complex agreement (concord) systems.6 Variation in noun-class markers among Niger-Congo branches and individual languages involves their status as prefixes, suffixes, proclitics, and enclitics (sometimes with transitional stages resulting in circumfixes) as well as their grammatical status. The following examples illustrate this variation:

  1. (6) Wolof (Atlantic; Fal et al. 1990)Niger-CongoA brief state of the art

  2. (7) Kuwaataay (Atlantic; Coly 2012)Niger-CongoA brief state of the art

  3. (8) Kabye (Gur; Kleinewillinghöfer 2000)Niger-CongoA brief state of the art

  4. (9) Ukaan (Benue-Congo; Salffner 2010)Niger-CongoA brief state of the art

  5. (10) Mbembe (Benue-Congo; Richter genannt Kemmermann, 2014)Niger-CongoA brief state of the art

Greenberg(1977) concludes for a number of reasons that Proto-Niger-Congo is likely to have had a system of noun-class prefixes rather than suffixes. First, this is the most widespread pattern genetically across the Niger-Congo phylum. Second, suffixation systems can be shown to be innovations in Atlantic, Benue-Congo, and Gur: encliticized and suffixed noun classes can be explained as resulting from the reinterpretation of postnominal demonstratives as specifiers or definite articles; these in turn may lose their function as definiteness markers and be attached to nouns in all syntactic environments.

Table 2, adapted from Dimmendaal (2011: 292), illustrates this variation in affixation systems across Niger-Congo.

Table 2: Noun-class affixation in Niger-Congo


Agreement within noun phrase


prefixes reconstructed; also suffixation






prefixes (innovating suffixes and proclitics)



petrified suffixes




petrified prefixes



reduced prefixation



(petrified) prefixation



reduced (petrified, some suffixation)



petrified prefixes



  Cross River

prefixes (partly reduced)






prefixation (partly reduced)



toward suffixation



(remnant) prefixes (innovating suffixes and proclitics)



prefixation (reduction and loss in north and west)





The widespread, extensive noun-class prefixation system with agreement marking on dependent categories within the noun phrase as well as cross-referencing for nominal subjects and objects on the verb is widely held to be a retention. Demuth et al. (1986) showed that the historical drift went in two directions: either the agreement system was strongly reduced, or the system of nominal prefixes was reduced and concord remained more extensive. A number of authors have shown that the marking of noun classes on the noun interacts with the presence of nominal qualifiers within an extended nominal group. Kiessling (2012), for example, reports alternations for Grassfields Bantu languages like Isu, which maintains or drops the prefix under certain (syntactic and pragmatic) conditions when the noun is modified. Good (2012) provides a further elegant survey of reinterpretations of Niger-Congo noun-class systems.

The loss of noun-class prefixes may also result in complex morphophonological interactions between class markers and nominal stems, as attested in various western Niger-Congo groupings. In many Northern Atlantic languages, for instance, there is a tendency toward incorporation of the prefix into the nominal root. Whether this has to do with prosodological changes (loss of tone and development of stress) historically is not yet clear. It has been noted for various Niger-Congo groups (e.g. Kegboid and Igboid; Williamson 1985, 1993) that the tonality of noun-class markers is of particular importance for their preservation. In Atlantic (as a primarily areal grouping within Niger-Congo, parallel to Kordofanian), the gradual incorporation of the noun-class prefix results in stem-initial consonant mutation, whereby stem consonants can occur in three grades (realized as fricatives/rhotics, stops, or prenasalized stops). Noun-class markers, being portmanteau morphemes that also indicate number or quantity, may still be present as prefixes (e.g. in Menik and Oniyan) or are rebuilt as suffixes with stem-initial consonant alternation as a reflex of the original noun-class prefix. The latter possibility is illustrated by the following Fulfulde examples:

  1. (11) Niger-CongoA brief state of the art

In Fulfulde, each class is assigned to a particular grade of the stem, which suggests that grades result from regular assimilation processes. However, the class suffix itself also occurs in different forms (e.g. -ʔo, -jo, -ɗo for the singular of humans; -ru, -du, -ndu for a singular of animals and things). As a result the underlying processes of consonantal change are hard to discern, and reconstructing the original shape of the noun stem often seems impossible.

At least two additional historical tendencies for noun-class development are attested in Niger-Congo subgroups, namely increment (or incorporation of former prefixes into the nominal stem) and language manipulation. Williamson (1989) suggests that constant affix renewal is another process that shaped the diverse picture of Niger-Congo noun classes. Noun-class markers allow for stacking in a geographically rather wide variety of languages, as in Bantu, where this strategy results in complex affixes:

  1. (12) Thimbukushu (Maho 1999: 57)Niger-CongoA brief state of the art

The same tendency toward stacking or increment is attested in the Katla group within Kordofanian (Dimmendaal 2014).

There is also evidence that language manipulation contributed to a reversal of the morpheme order, as in cases where syllable transposition is practiced. This has happened in Hone, a moribund Jukun language of Nigeria, where original class prefixes turned into suffixes in a secret language and later spread into the ordinary register (e.g. ʒá-bá ‘friends’ < *bá-ʒá; Storch 2011). Such forms remain about the only instantiations of retained class markers in Hone. Massive loss of noun-class marking elsewhere in the language resulted not only in the inability of most nouns to inflect for number but also in a process whereby endocentric compounds gradually developed into classifiers (Storch 1999), for instance:

  1. (13) Niger-CongoA brief state of the art

The rebuilding in Hone of a locative class alongside five other classes (denoting humans, animates, inanimate objects, liquids, and diminutives) is fairly characteristic of Niger-Congo.

Initial evidence is emerging that Kordofanian languages in the Nuba Mountains also tend to assign a special morphological and semantic status to location marking, as in the following example from Tima (Alamin et al. 2012):

  1. (14) Niger-CongoA brief state of the art

Bantu languages have generally been assumed to be conservative and to reflect archaic morphological properties, since similar morphological systems are attested in the geographical outliers of Niger-Congo, Atlantic, and so-called Kordofanian languages. This assumption may also apply to the system of locatives, which includes an original Proto-Bantu system of three noun-class prefixes, *pa-, *ku-, and *mu- (Meeussen 1967: 97–98, 103–104), each with its own agreement marking on dependent categories:7

  1. (15) Niger-CongoA brief state of the art

Within Bantu, such phrases may occur as locative complements to verbs, but they also may occupy the position of syntactic subject or object, particularly in combination with motional verbs, postural verbs, and verbs of existence and availability. The following example with a locative subject is from Chichewa (Bresnan & Kanerva 1989):

  1. (16) Niger-CongoA brief state of the art

Different authors in the collection edited by Miehe & Winkelmann (2007) reconstruct two locative noun classes,*pʊ and *kʊ, within a total set of 25 noun classes for Proto-Gur. Kleinewillinghöfer (1991) shows that in Waja (Adamawa) there is also a locative class marker pə, which resembles the Proto-Bantu locative class 16 (*pa) in form and meaning.

Future research on these and other languages should clarify whether the typologically interesting properties of Bantu locatives indeed go back to much earlier stages in the history of Niger-Congo.

In the West African coastal area, where many Volta-Niger languages lost noun-class affixation as a morphological strategy, a number of language groups developed a radically different system. Whereas several languages belonging to the Cross River branch within Benue-Congo have retained the “classical” noun-class system, Kana and other, closely related languages replaced this system with a system of numeral classifiers. Ikoro (1996) identifies sixteen such markers occurring between the numeral and the head noun in Kana. Their recent origin is supported by the fact that most of these numeral classifiers still coexist with the nouns that served as their etymological sources, as illustrated in example (17).

  1. (17) Niger-CongoA brief state of the art

There is no distinction between count and mass nouns in Kana; all concrete nouns primarily express substance or material. Numeral classifiers such as those listed in example (18) primarily provide information on shape in the case of concrete nouns, but they may also signify various other concepts:8

  1. (18) Niger-CongoA brief state of the art

Numeral classifiers seem to emerge from associative constructions (‘x of/with y’), as further suggested by the coexistence of such forms with noun-class systems in various languages of the West-Central African geographical area (Kiessling 2013). The shift from a noun-class language to a language with numeral classifiers in Niger-Congo seems to sometimes occur in connection with lexical aspect (a feature that is rather common in Africa’s Sudanic Belt, where most Niger-Congo families can be found). In nominal-aspect languages, nouns can have transnumeral forms, which denote more general concepts (in the sense of referring to a set), in contrast to number-inflected forms, which denote sorts and singular objects (Rijkhoff 2002: 29ff). For instance, in Pular (a variety of Fulfulde spoken in northern Guinea), it is possible to achieve a generalizing concept by omitting the noun-class suffix, as in the following example (Corbett 2000: 12):

  1. (19) Niger-CongoA brief state of the art

The quantification of the morphologically bare set noun entails associative constructions with shape-denoting nouns, such as ‘bag’, ‘heap’, or ‘bundle’. Urban Pular speakers in Conakry incipiently abandon the rich noun-class system of Pular (as a result of contact with Mande languages, which do not have noun classes), retaining only a general plural class and set plurals; these seem to give way to such associative constructions, which are reminiscent of classifier forms (Diallo 2008 & personal communication 2009). Hence the ability of a noun to acquire two readings in the plural, a more distributive, individualizing one and a generalizing one, provides the basis for the development of additional classifying devices besides noun classes.

4.3 Verb extensions

Another outstanding structural feature for much of the Niger-Congo area is the rich inventory of verb extensions found in various subgroups (e.g. see Hyman 2007, 2014). These are predominantly suffixed derivational morphemes, which can be stacked in some languages, resulting in rather impressive verb forms. Verb extensions create applicative, iterative, directional, and other verb stems and therefore usually—though not always—are related to valency-increasing strategies. But they also may cover additional semantic domains, as the list from Fulfulde (Klingenheben 1963) in (20) and the examples in (21) suggest:

  1. (20) Niger-CongoA brief state of the art

  2. (21) Niger-CongoA brief state of the art

Table 3 shows Hyman’s (2007: 151) survey of reconstructed verb extensions for Niger-Congo:

Table 3: Verbal extensions in Niger-Congo1 (Hyman 2007)

Proto-Niger-Congo Voeltz (1977)

Proto-Bantu Schadeberg (2003)

Proto-Atlantic Doneux (1975)






*-ci, *-ti














Reversive (trans.)




Reversive (intrans.)









(1) Table adapted from Hyman (2007: 151).

Current grammaticalization theories have made the search for a lexical origin of such grammatical morphemes popular. There appears to be a popular view that these grammatical markers go back to independent lexemes (as first proposed by Voeltz 1977). However, a number of verbal extensions in Niger-Congo appear to be extremely old and lack any direct evidence for a lexical source. This applies, for instance, to the causative marker *-i, whose presence in geographically disparate and historically distant subgroups such as Bantoid and Kordofanian languages in the Nuba Mountains suggests that it is old in Niger-Congo. For example, the causative suffix -i is present in Moro, a member of the Heiban group (Rose 2014: 47), and in Dagik, a member of the Talodi group (Vanderelst, to appear). This suffix also is attested as a high transitivity marker in the Katla group (Alamin 2009). The same causative suffix is attested in Bantoid languages (see Kiessling 2004). Meeussen (1967: 92–93) reconstructs a suffix *-í̹as an agentive marker for Proto-Bantu (where the cedilla signifies a high closed front vowel).

Meeussen (1967: 92) also reconstructs a so-called impositive marker *-ik, which appears to have been unproductive at the Proto-Bantu stage. This marker corresponds to a productive causative marker -ik in Kordofanian languages like Tima and Katla. These examples of verb extensions (as well as the noun-class markers discussed in section 4.2) help to make clear that there is indeed solid grammatical evidence that Kordofanianlanguages are part of Niger-Congo.

Since high front vowels (as in the causative marker) are highly common cross-linguistically, this formal identity is not significant in and of itself. But the presence of a set of cognate verbal and nominal morphemes (e.g. noun classes, as pointed out above) in Kordofanian (as an areal grouping within Niger-Congo) and Bantoid makes the causative morpheme another likely candidate for a common origin.

There is also evidence for aspectual suffixes as archaisms in Niger-Congo. In particular, perfective-marking suffixes are quite common in Niger-Congo and may constitute an ancient property of this phylum. Meeussen (1967: 110) reconstructs a perfective marker *-ide for Proto-Bantu. Like some of the derivational extensions, this suffix is very persistent. Cross River languages like Kana have a suffix -r-a (Ikoro 1996: 184), Eastern Kru languages use a verbal suffix -e (Marchese 1989: 122), and such a suffixation pattern is also found in Niger-Congo (“Kordofanian”) languages in Sudan, such as Moro. Even in languages whose verbs have largely lost their ability to take any suffixes, there remain distinct aspect stems as well as a variety of petrified extensions (Storch 1999: 177). For instance, in Wapha (Jukun), aspect stems are constructed on the basis of tonal inflection. Even though this language is predominantly isolating, it exhibits fusional inflection in the verb stem in the formation of aspect:

  1. (22) Niger-CongoA brief state of the art

An additional typological property of Niger-Congo languages, however, is the marking of tense and aspect in combination with pronominal subjects by way of a separate constituent preceding the verb, as shown in section 4.4 below.

4.4 Constituent order and syntactic typology

As observed by Bickel (2007), syntactic typologies these days are tending away from gross generalizations over large numbers of languages; instead, researchers prefer to ask questions like “What is found where, and why?” This tendency also applies to the syntactic typology of Niger-Congo (as a genetic grouping).

Gensler (1997) states that over much of Niger-Congo there is a constituent order S-Aux-OV-X, with an auxiliary element covering the slot (marking tense, aspect, and mood) following the subject. However, there are a number of analytical problems here, having to do with the interpretation of pronominal markers, nominalized verb forms, and the structure of multiverb constructions. In Niger-Congo subgroups such as Kwa, Kru, and Benue-Congo, the type of predicate formation often depends on lexical aspect and other semantic features of the verb, which remain grossly understudied in most of these languages. Verb semantics and the way they can change are of relevance for the composition of multiverb constructions (Kiessling 2011: 30 ff.), where a verb can be used both as the core verb and as a coverb in a serial verb construction. In the following example from Isu (Grassfields Bantu), tsìy ‘pass’ can be extended semantically as a center-bounded path coverb from movement on a path to point of reference (Kiessling 2011: 45):

  1. (23) Niger-CongoA brief state of the art

Semantic shift is also at work in the context of (fairly common) split-predicate constructions, where verbs used in the AUX slot function as light verbs or coverbs. The core verb in such constructions often needs to be nominalized, as with wàà ‘drink’ in the following example from Hone (Storch 1999: 114):

  1. (24) Niger-CongoA brief state of the art

Nominalization of the core verb is necessary because Hone requires verbs to always occur before a complement. (There is a consistent SVO pattern in Hone, which is due to the lack of any other ways of differentiating between the major parts of speech, as both verb and noun phrases tend to be bare in this language.) The avoidance of syntactic intransitivity also results in a variety of other constructions, such as intransitive copy pronouns and cognate object constructions in Jukun, Platoid, and to some extent Bantoid.

In Benue-Congo and other Niger-Congo families, the indexation of pronominal subjects and objects differs from that of nominal S and O participants. With respect to the typology of pronominal subject and object indexation, Creissels et al. (2008) observe that descriptions of African languages sometimes do not identify pronominal markers appropriately, treating them as independent words whereas in fact they are bound morphemes: either clitic elements in complementary distribution with noun phrases or affixes. The split is relevant for an appropriate analysis of argument structure. For example, pronominal objects in multiverb constructions in Ọkọ (Benue-Congo) trigger split-predicate constructions (25a), whereas nominal objects trigger bipartite verb forms (25b; data from Atoyebi 2010: 85 ff.):

  1. (25a) Niger-CongoA brief state of the art

  2. (25b) Niger-CongoA brief state of the art

The preradical position of the pronominal object (as opposed to the postverbal position of nominal objects) in a large number of Bantu languages somewhat parallels such constructions and permits a typological differentiation of agreement-marking strategies (Bearth 2003: 124).

Constituent order variation, finally, exhibits strong pragmatic overtones but is also a widespread strategy for achieving passive- and middle-voice meanings of core verbs (Bearth 2003: 130 ff.). In western groupings of Niger-Congo, this strategy is relevant for participant coding in still another way: motion verbs, for instance, undergo semantic shift in intransitive constructions. In Ọkọ, A-participants (as in 25b) become O-participants in such contexts (Atoyebi 2010: 86):

  1. (26) Niger-CongoA brief state of the art

Whereas head marking on the verb by way of valency-changing suffixes is a common strategy for modifying semantic roles within a clause in a wide range of Niger-Congo languages, alternative strategies such as serial verb constructions are also widespread.9 As argued by Ameka (2006: 141) for the Kwa language Ewe, these “are but one type of M[ulti ]V[erb ]C[onstruction]s which form part of the strategies employed to articulate verbal functions in the language…. We cannot understand the nature of S[erial ]V[erb ]C[onstruction]s unless we link the various features to the linguistic type properties of the languages in which we find them.”

Languages spoken in the transitional zones between areas where (equipollent) serial verbs are common and areas where head marking on the verb (verb framing) is the dominant strategy sometimes combine these two strategies to express alternative ways of information packaging in a clause. Ikoro (1996: 254) gives such contrastive examples for the Cross River language Kana (spoken in Nigeria):

  1. (27) Niger-CongoA brief state of the art

  2. (28) Niger-CongoA brief state of the art

Whereas example (27), involving a thetic statement, uses a ditransitive verb ‘give’ without a second verb, example (28), involving verb focus and thereby a categorical statement, requires a serial verb construction.

The interpretation of syntactic constituents as subjects or objects in Niger-Congo languages usually derives from their position within a sentence as well as from cross-reference marking on the verb. The use of dependent marking for these core syntactic functions is rare for this phylum but does exist. Heiban (“Kordofanian”) languages in the Nuba Mountains use accusative case to express syntactic objects, as shown for Moro by Ackerman & Moore (2013: 89):

  1. (29) Niger-CongoA brief state of the art

(where cl:g refers to noun-class prefixes with an initial g-). Similar strategies with accusative case marking occur in Nilo-Saharan language groups in the Nuba Mountains, such as Nubian or Nyimang, and thus may reflect an ancient convergence phenomenon in the area.

5 Prospects

The history of research on Niger-Congo languages is strongly characterized by the search for order in some of the world’s most diverse linguistic regions. Since their colonial beginnings, the postulation and reconstruction of large language families has been professionally more desirable and prestigious for many Africanists than has painstaking work on small clusters and groups. This bias seems to have resulted in one subclassification’s replacing another, without much proof for either of them. The renewed interest in recent years in the comparative study of lower-level units within Niger-Congo seems to finally be counterbalancing this unfortunate situation.

The ongoing descriptive and documentary work on individual languages and their varieties, greatly expanding our knowledge on formerly little-known linguistic regions, is helping to identify clusters and units that allow for the application of the historical-comparative method. Only the reconstruction of lower-level units, instead of “big picture” contributions based on mass comparison, can help to verify (or disprove) our present concept of Niger-Congo as a genetic grouping consisting of Benue-Congo plus Volta-Niger, Kwa, Adamawa plus Gur, Kru, the so-called Kordofanian languages, and probably the language groups traditionally classified as Atlantic. However, we also need new approaches to these small language groups and to historical-comparative research on them. Established diachronic methodologies need to be combined with sociohistorical research if we are to understand linguistic changes due to sociocultural processes as well as those that can be accounted for by the comparative method. Such processes include the constant substitution of lexical and grammatical material in the context of deeply rooted multilingualism, as well as avoidance and taboo (Kleinewillinghöfer 1995, Lüpke & Storch 2013) and the deliberate manipulation of languages (Thomason 2007, Storch 2011). Because the social history of Niger-Congo languages in the past 500 years or so appears to have been profoundly different from the social histories of Indo-European languages (Kopytoff 1987), it won’t be possible to properly understand processes of language change in this phylum without interdisciplinary research.

Such research also entails linking the expansion of Niger-Congo to archaeological findings (Blench 2006). Niger-Congo as a whole constitutes a spread zone in the sense of Nichols (1992), that is, an area where from time to time one language spreads out widely, absorbing or displacing other languages. Successive spreads may have reduced genetic diversity in this zone. The Sahara toward the north and the Atlantic Ocean toward the south created natural obstacles or barriers, virtually impeding entrants from the north or south (with some exceptions) and resulting in what Nichols (to appear) calls a closed spread zone. The Niger-Congo expansion most likely dates back to 12,000 BP, when interglacial climate change resulted in a rather dramatic increase in precipitation in major parts of West Africa that for thousands of years before were covered by deserts and therefore uninhabited by human beings. During this prior era, human habitat was restricted to isolated areas with higher elevations and major river systems. The presence of various linguistic isolates across West Africa (Dimmendaal 2011: 324–329) presumably is a reflex of this earlier genetic diversity before the expansion of Niger-Congo groups set in. Renewed dramatic climate changes and the subsequent expansion of the Sahara over the past five millennia led not only to migration of Chadic (Afro-Asiatic) groups into the area but probably also to the interruption of the Gur–Adamawa continuum (Kleinewillinghöfer, forthcoming), the migration of Niger-Congo groups into the Nuba Mountains in Sudan, and the southward expansion of Bantu.


We are grateful to an anonymous reviewer for very helpful comments on an earlier version of this chapter. We thank Monika Feinen for drawing the maps.

List of Abbreviations

  • ACC =


  • CAUS =


  • CL =

    class marker

  • DEM =


  • FOC =


  • GEN =


  • INDEF =


  • IPFV =


  • P1 =

    immediate past

  • PERT =


  • PFV =


  • PL =


  • PROG =


  • SG =


  • SM =

    subject marker

  • SU =


  • VN =

    verbal noun


Ackermann, Farrell & John Moore. 2013. Objects in Moro. In Nuba Mountain Language Studies, ed. by Thilo C. Schadeberg & Roger Blench, 83–104. Cologne: Köppe.Find this resource:

Alamin, Suzan. 2009. The Nominal and Verbal Morphology of Tima. Cologne: Köppe.Find this resource:

Alamin, Suzan, Gerrit J. Dimmendaal & Gertrud Schneider-Blum. 2012. Finding your way in Tima. In Directionality in Grammar and Discourse: Case Studies from Africa, ed. by Angelika Mietzner & Ulrike Claudi, 9–33. Cologne: Köppe.Find this resource:

Ameka, Felix K. 2006. Serial verb constructions in their grammatical context. In Alexandra Y. Aikhenvald & R. M. W. Dixon (eds.), Serial Verb Constructions: A Cross-linguistic Typology, 124–143. Oxford: Oxford University Press.Find this resource:

Atoyebi, Joseph Dele. 2010. A Reference Grammar of Oko. Cologne: Köppe.Find this resource:

Babaev, Kirill V. 2008. Reconstructing Proto-Benue-Congo person marking I: Proto-Bantoid. Journal of West African Languages 35.1–2: 131–190.Find this resource:

Bearth, Thomas & Christa Link. 1980. The tone puzzle of Wobe. Studies in African Linguistics 11: 147–207.Find this resource:

Bearth, Thomas. 2003. Syntax. In The Bantu Languages, ed. by Derek Nurse & Gérard Philippson, 121–142. London: Routledge.Find this resource:

Bendor-Samuel, John (ed.). 1989. The Niger-Congo Languages. Lanham, MD: University Press of America.Find this resource:

Bendor-Samuel, John, Elizabeth J. Olsen & Ann R. White. 1989. Dogon. In The Niger-Congo Languages, ed. by John Bendor-Samuel, 168–177. Lanham, MD: University Press of America.Find this resource:

Bennett, Patrick R. & Jan P. Sterk. 1977. South Central Niger-Congo: A reclassification. Studies in African Linguistics 8: 241–273.Find this resource:

Bickel, Balthasar. 2007. Typology in the 21st century: Major current developments. Linguistic Typology 11: 239–251.Find this resource:

Bleek, Wilhelm. 1862. A Comparative Grammar of South African Languages, Vol. 1. London: Trübner.Find this resource:

Blench, Roger M. 2006. Archaeology, Language and the African Past. Lanham, MD: Altamira Press.Find this resource:

Blench, Roger M. & Stuart McGill. 2012. The Kainji languages of northwestern and central Nigeria.

Blench, Roger M. 2014. Splitting up Kordofanian. In Nuba Mountain Language Studies, ed. by Thilo C. Schadeberg & Roger M. Blench, 571–586. Cologne: Köppe.Find this resource:

Bole-Richard, Remy. 1985. Hypothese sur la genèse de la nasalité en Niger-Congo. Journal of West African Languages 15.2: 3–28.Find this resource:

Bresnan, Joan & Jonni Kanerva. 1989. Locative inversion in Chichewa. Linguistic Inquiry 20: 1–50.Find this resource:

Casali, Roderic F. 2008. ATR harmony in African languages. Language and Linguistics Compass 2.3: 496–549.Find this resource:

Childs, G. Tucker. 2003. An Introduction to African Languages. Amsterdam/Philadelphia: Benjamins.Find this resource:

Clements, George N. & Annie Rialland. 2008. Africa as a phonological area. In The Linguistic Geography of Africa, ed. by Bernd Heine & Derek Nurse, 36–85. Cambridge, UK: Cambridge University Press.Find this resource:

Coly, Jules Jacques 2012. Morphosyntaxe du kuwaataay. Munich: Lincom.Find this resource:

Corbett, Greville. 2000. Number. Cambridge: Cambridge University Press.Find this resource:

Creissels, Denis, Gerrit J. Dimmendaal, Zygmunt Frajzyngier & Christa König. 2008. Africa as a morphosyntactic area. In A Linguistic Geography of Africa, ed. by Bernd Heine & Derek Nurse, 86–150. Cambridge, UK: Cambridge University Press.Find this resource:

De Wolf, Paul P. 1971. The Noun Class System of Proto-Benue-Congo. The Hague: Mouton de Gruyter.Find this resource:

De Wolf, Paul P. 1985. Die Menschenklassen in den nordwestatlantischen Sprachen. Vienna: Afro-Pub.Find this resource:

Demuth, Katherine, Nicholas Faraclas & Lynell Marchese. 1986. Niger-Congo noun class and agreement systems in language acquisition and historical change. In Noun Classes and Categorization, ed. by Colette Craig, 453–471. Amsterdam/Philadelphia: Benjamins.Find this resource:

Diallo, Abdourahmane. 2008. Language contact between Mande and Atlantic in Guinea. Mande Languages and Linguistics—2nd International Conference St. Petersburg, ed. by Valentin Vydrin, 61–79. St. Petersburg University.Find this resource:

Dimmendaal, Gerrit J. 2008. Africa’s verb-final languages. In The Linguistic Geography of Africa, ed. by Bernd Heine & Derek Nurse, 272–308. Cambridge, UK: Cambridge University Press.Find this resource:

Dimmendaal, Gerrit J. 2011. Historical Linguistics and the Comparative Study of African Languages. Amsterdam/Philadelphia: Benjamins.Find this resource:

Dimmendaal, Gerrit J. 2014. Where have all the noun classes gone in Tima? In In and Out of Africa. Languages in Question in Honour of Robert Nicolaї, Vol. 2: Contact and Language Change in Africa, ed. by Carole de Feral, Maarten Kossmann & Mauro Tosco, 104–125. Louvain-la-Neuve: Peeters.Find this resource:

Dixon, R. M. W. 1997. The Rise and Fall of Languages. Cambridge, UK: Cambridge University Press.Find this resource:

Doneux, Jean L. 1975. Hypothèses pour la comparative des languages ouest-atlantiques. Africana Linguistica 6: 41–129.Find this resource:

Fal, Arame, Rosine Santos & Jean Léonce Doneux. 1990. Dictionnaire wolof–français. Paris: Karthala.Find this resource:

Gensler, Orin. 1997. Grammaticalization, typology and Niger-Congo word order. Progress on a still-unsolved problem. Journal of African Languages and Linguistics 18.1: 57–93.Find this resource:

Good, Jeff. 2012. How to become a “Kwa” noun. Morphology 22: 293–335.Find this resource:

Greenberg, Joseph H. 1955. Studies in African Linguistic Classification. New Haven: Compass.Find this resource:

Greenberg, Joseph H. 1963. The Languages of Africa. Bloomington, IN/The Hague: Indiana University Research Center in Anthropology, Folklore and Linguistics & Mouton.Find this resource:

Greenberg, Joseph H. 1977. Niger-Congo noun class markers: Prefixes, suffixes, both or neither. Studies in African Linguistics, Supplement 7: 94–104.Find this resource:

Hair, Paul E. H. 1963. Koelle at Freetown. Polyglotta Africana (S. W. Koelle), 7–17. Graz: Akademische Druck- und Verlagsanstalt.Find this resource:

Harley, Matthew. 2012. Unusual sounds in Nigerian languages. In Advances in Minority Language Research in Nigeria, ed. by Roger Blench & Stuart McGill, 39–66. Cologne: Köppe.Find this resource:

Heine, Bernd. 1968. Die Verbreitung und Gliederung der Togorestsprachen. Berlin: Reimer.Find this resource:

Hellwig, Birgit. To appear. A Grammar of Katla.Find this resource:

Hyman, Larry. 2007. Niger-Congo verb extensions: Overview and discussion. In Selected Proceedings of the 37th Annual Conference on African Linguistics, ed. by D. L. Payne & J. Peña, 149–163. Somerville, MA: Cascadilla.Find this resource:

Hyman, Larry. 2014. Reconstructing the Niger-Congo verb extensions paradigm: What’s cognate, copied or renewed? In Paradigm Change, ed. by M. Robbeets & Walter Bisang, 103–125. Amsterdam: Benjamins.Find this resource:

Ikekeonwu, Clara. 2007. The Phonetics of Nigerian Languages. Munich: Lincom.Find this resource:

Ikoro, Suano. 1996. The Kana Language. Leiden: CNWS.Find this resource:

Jungraithmayr, Hermann. 1980. Kontakte zwischen Adamawa-Ubangi- und Tschad-Sprachen: Zur Übertragung grammatischer Systeme. Zeitschrift der Deutschen Morgenländischen Gesellschaft 130.1: 70–85.Find this resource:

Kiessling, Roland. 2004. Kausation, Wille und Wiederholung in der verbalen Derivation der westlichen Ring-Sprachen (Weh, Isu). In Sprache und Wissen in Afrika: Beiträge zum 15. Afrikanistentag, Frankfurt am Main und Mainz, 30. September–2. Oktober 2002, ed. By Raimund Kastenholz & Anne Storch, 159–181. Cologne: Köppe.Find this resource:

Kiessling, Roland. 2011. Verbal Serialisation in Isu (West-Ring). Cologne: Köppe.Find this resource:

Kiessling, Roland. 2012. High vowel reduplication and infix genesis in Isu (West Ring). In Aspects of Reduplication in Languages of Cameroon and Senegal, ed. by Pius N. Tamanji & Gabriel M. Mba, 6–31. Munich: Lincom.Find this resource:

Kiessling, Roland. 2013. On the origin of Niger-Congo nominal classification. In Historical Linguistics 2011, ed. by Ritsuko Kikusawa & Lawrence A. Reid, 43–66. Amsterdam/Philadelphia: Benjamins.Find this resource:

Kleinewillinghöfer, Ulrich. 1991. Die Sprache der Waja (nyan wiyáù). Frankfurt: Lang.Find this resource:

Kleinewillinghöfer, Ulrich. 1995. Don’t use the name of my dead father. A reason for lexical change in some northwestern Adamawa languages (northeastern Nigeria). Afrika und Übersee 78: 125–137.Find this resource:

Kleinewillinghöfer, Ulrich. 1996. Relationship between Adamawa and Gur: The case of Waja and Tula. Gur Papers/Cahiers Voltaïques 1: 25–45.Find this resource:

Kleinewillinghöfer, Ulrich. 2000. The noun classification of Cala (Bogong): A case of contact-induced change. Frankfurter Afrikanistische Blätter 12: 37–68.Find this resource:

Kleinewillinghöfer, Ulrich. Forthcoming. Adamawa. In Handbook of African Languages, ed. by Rainer Vossen & Gerrit J. Dimmendaal. Oxford: Oxford University Press.Find this resource:

Klingenheben, August. 1924/25. Die Permutationen des Biafada und des Ful. Zeitschrift für Eingeborenensprachen 15: 180–213, 266–272.Find this resource:

Klingenheben, August. 1963. Die Sprache der Ful. Hamburg: Augustin.Find this resource:

Koelle, Sigismund Wilhelm. 1854. Polyglotta Africana or A Comparative Vocabulary of Nearly Three Hundred Words and Phrases in More than One Hundred Distinct African Languages. London: Church Missionary Society.Find this resource:

Köhler, Oswin. 1975. Geschichte und Probleme der Gliederung der Sprachen Afrikas. Von den Anfängen bis zur Gegenwart. In Die Völker Afrikas und ihre traditionellen Kulturen, Teil I, ed. by Hermann Baumann, 135–373. Wiesbaden: Harrassowitz.Find this resource:

Kopytoff, Igor (ed.). 1987. The African Frontier: The Reproduction of Traditional African Society. Bloomington: Indiana University Press.Find this resource:

Kutsch Lojenga, Constance. 2003. Bila. In The Bantu Languages, ed. by Derek Nurse & Gérard Philippson, 450–474. London: Routledge.Find this resource:

Lepsius, Karl Richard. 1880. Nubische Grammatik: Mit einer Einleitung über die Völker und Sprachen Afrika’s. Berlin: Hertz.Find this resource:

Lüpke, Friederike & Anne Storch. 2013. Repertoires and Choices in African Languages. Berlin: Mouton de Gruyter.Find this resource:

Lüpke, Friederike. Forthcoming. Oxford’s Guide to the World’s Languages: Atlantic. Oxford: Oxford University Press.Find this resource:

Maho, Jouni. 1999. A Comparative Study of Bantu Noun Classes. Gothenburg: Acta Universitatis Gothoburgensis.Find this resource:

Manessy, Gabriel. 1975. Les langues oti-volta. Classification généalogique d’un groupe de langues voltaïques. Paris: SELAF.Find this resource:

Marchese, Lynell. 1989. Kru. In The Niger-Congo Languages, ed. by John Bendor-Samuel, 119–139. Lanham, MD: University Press of America.Find this resource:

McMahon, April & Robert McMahon. 2008. Genetics, historical linguistics and language variation. Language and Linguistics Compass 2.2: 264–288.Find this resource:

Meeussen, A. E. 1967. Bantu grammatical reconstructions. Africana Linguistica 3: 79–121.Find this resource:

Meinhof, Carl. 1899. Grundriss einer Lautlehre der Bantusprachen nebst einer Anleitung zur Aufnahme der Bantusprachen. Leipzig/Berlin: Brockhaus.Find this resource:

Meinhof, Carl. 1911a. Das Ful in seiner Bedeutung für die Sprachen der Hamiten, Semiten, und Bantu. Zeitschrift der Deutschen Morgenländischen Gesellschaft 65: 177–220.Find this resource:

Meinhof, Carl. 1911b. Sudansprachen und Hamitensprachen. Zeitschrift für Kolonialsprachen 1: 161–166.Find this resource:

Miehe, Gudrun. 1991. Die Präfixnasale im Benue-Congo und im Kwa: Versuch einer Widerlegung der Hypothese von der Nasalinnovation des Bantu. Berlin: Reimer.Find this resource:

Miehe, Gudrun & Kerstin Winkelmann (eds.). 2007. Noun Class Systems in Gur Languages, Vol. 1: Southwestern Gur Languages (without Gurunsi). Cologne: Köppe.Find this resource:

Moñino, Yves. 1988. Lexique comparatif des langues oubanguiennes. Paris: Geuthner.Find this resource:

Moñino, Yves. 1995. Le Proto-Gbaya: Essai de linguistique comparative historique sur vingt-et-une langues d’Afrique centrale. Louvain: Peeters.Find this resource:

Nichols, Johanna. 1992. Linguistic Diversity in Space and Time. Chicago: Chicago University Press.Find this resource:

Nichols, Johanna. To appear. Types of spread zones: Open and closed, horizontal and vertical. In Language Structure and Environment: Social, Cultural, and Natural Factors, ed. by Rik De Busser & Randy J. LaPolla, 261–286. Amsterdam/Philadelphia: Benjamins.Find this resource:

Pozdniakov, Konstantin. 2009. Niveaux linguistiques et problèmes de reconstruction dans les langues atlantiques. Sprache und Geschichte in Afrika 19: 175–200.Find this resource:

Richter genannt Kemmermann, Doris. 2014. The Mbembe Language. Leiden: Brill.Find this resource:

Rijkhoff, Jan. 2002. The Noun Phrase. Oxford: Oxford University Press.Find this resource:

Rose, Sharon. 2014. The morphological structure of the Moro verb. In Nuba Mountain Language Studies, ed. by Thilo C. Schadeberg & Roger M. Blench, 25–55. Cologne: Köppe.Find this resource:

Salffner, S. 2010. Tone in the phonology, lexicon and grammar of Ikaan. PhD thesis, London (SOAS).Find this resource:

Sapir, J. D. 1971. West Atlantic: An inventory of the languages, their noun class systems and consonant alternation. In Current Trends in Linguistics 7, Linguistics in Sub-Saharan Africa, ed. by Thomas Sebeok, 45–112. The Hague: Mouton.Find this resource:

Schadeberg, Thilo C. 1981. The classification of the Kadugli language group. In Nilo-Saharan: Proceedings of the 1st Nilo-Saharan Linguistics Colloquium, ed. by Thilo C. Schadeberg & M. Lionel Bender, 291–305. Dordrecht: Foris.Find this resource:

Schadeberg, Thilo C. 1989. Kordofanian. In The Niger-Congo Languages, ed. by John Bendor-Samuel, 67–80. Lanham, MD: University Press of America.Find this resource:

Schadeberg, Thilo C. 2003. Historical linguistics. In The Bantu Languages, ed. by Derek Nurse & Gérard Philippson, 143–163. London: Routledge.Find this resource:

Shimizu, Kiyoshi. 1980. Comparative Jukunoid. Vienna: Afro-Pub.Find this resource:

Stewart, John Massie. 1976. Towards Volta-Congo Reconstruction. Leiden: Leiden University Press.Find this resource:

Stewart, John Massie. 2002. The potential of Proto-Potou-Akanic-Bantu as a pilot for Niger-Congo, and the reconstructions updated. Journal of African Languages and Linguistics 23.2: 197–224.Find this resource:

Stewart, John Massie. 2005. Three-grade consonant mutation in the Fulanic and Akanic languages and in their latest common ancestor (Proto-Niger-Congo?). In Studies in African Comparative Linguistics with Special Focus on Bantu and Mande, ed. by Koen Bostoen & Jacky Maniacky, 7–25. Tervuren, Belgium: Royal Museum for Central Africa.Find this resource:

Storch, Anne. 1999. Das Hone und seine Stellung im Zentral-Jukunoid. Cologne: Köppe.Find this resource:

Storch, Anne. 2011. Secret Manipulations. New York: Oxford University Press.Find this resource:

Thomason, S. 2007. Language contact and deliberate change. Journal of Language Contact—Thema 1. this resource:

Vanderelst, John. To appear. A grammar of Dagik. PhD dissertation, University of Cologne.Find this resource:

Voeltz, Erhard. 1977. Proto-Niger-Congo extensions. PhD dissertation, University of California at Los Angeles.Find this resource:

Wedekind, Klaus. 1985. Why Bench’ (Ethiopia) has five level tones today. In Studia Linguistica Diachronica et Synchronica, ed. by Ursula Pieper & Gerhard Stickel, 881–901. Berlin: Mouton de Gruyter.Find this resource:

Welmers, William. 1968. Jukun of Wukari and Jukun of Takum. Ibadan, Nigeria: University Press.Find this resource:

Westermann, Diedrich. 1911. Die Sudansprachen: Eine sprachvergleichende Studie. Hamburg: Friedrichsen.Find this resource:

Westermann, Diedrich. 1927. Die westlichen Sudansprachen und ihre Beziehungen zum Bantu. Berlin: Walter de Gruyter.Find this resource:

Williamson, Kay, 1985: How to become a Kwa language. In Linguistics and Philosophy. Essays in Honour of Rulon S. Wells, ed. by A. Makkai & A. K. Melby, 427–443. Amsterdam/Philadelphia: Benjamins.Find this resource:

Williamson, Kay. 1989. Niger-Congo overview. In The Niger-Congo Languages, ed. by John Bendor-Samuel, 3–45. Lanham, MD: University Press of America.Find this resource:

Williamson, Kay. 1993. The noun prefixes of New Benue-Congo. Journal of African Languages and Linguistics 14: 29–45.Find this resource:

Williamson, Kay & Roger M. Blench. 2000. Niger-Congo. In African Languages: An Introduction, ed. by Bernd Heine & Derek Nurse, 11–42. Cambridge, UK: Cambridge University Press.Find this resource:


(1) For more detailed surveys of the history of the field, the reader is referred to Köhler (1975), Williamson & Blench (2000), Childs (2003), Blench (2006), Dimmendaal (2011), and Lüpke & Storch (2013).

(2) These problematic aspects of Niger-Congo as a proclaimed genetic unit led Dixon (1997) to a fundamental critique of Greenberg’s methodology and classification hypothesis, resulting in the suggestion to abandon the idea of “Niger-Congo” altogether.

(3) See Dimmendaal (2011) for a survey of the relevant sources.

(4) Stewart (2002, 2005) extended his historical-comparative studies of Volta-Congo to a comparison with (Northern) Atlantic. The resulting historical stage, according to Stewart, would correspond to the earliest stages of Niger-Congo. It is probably no accident that he did not include in his comparative studies two other families that Greenberg assumed to be part of Niger-Congo, Mande and Ubangian.

(5) Phonetic descriptions of such consonants include oppositions such as long versus short duration of articulation, strong versus weak articulation, and greater versus less muscular tension.

(6) Except in Bantu, Niger-Congo noun classes are not consistently numbered. Therefore the class affixes presented in the examples here are glossed simply as class, while Bantu noun classes are represented by their respective numbers.

(7) Meeussen (1967) points toward an additional locative class prefix i-, used with a restricted set of nouns in Northeastern Bantu languages.

(8) Such classifier systems appear to be immanent in a range of West African Niger-Congo languages with or without functioning noun-class systems but seem to have been largely ignored in descriptions (Dimmendaal 2011: 138).

(9) The area where serial verb constructions are particularly common, a zone, ranging from the Ivory Coast across West Africa toward Nigeria, is also the area where noun-class systems were reduced and where fortis/lenis contrasts with consonants are common.