Joseph Greenberg and the Current State of Niger-Congo

Full text

MOTHER TONGUE
Journal of the Association for the Study of Language in Prehistory • Issue XVIII • 2013
50th Anniversary of J.H. Greenberg’s The Languages of Africa (1963)
17
Joseph Greenberg and the Current State of Niger-Congo
Kirill Babaev
Russian State University for the Humanities, Moscow
The year 2013 was the 50th anniversary of Greenberg’s fundamental book [1963]
which marked the beginning of a new era in African linguistics. Since then, any linguist
working in this field should have inevitably defined himself as ‘pro-Greenberg’ or ‘anti-
Greenberg’, but the overall opinion of the linguistic community about Greenberg’s four
macrofamilies of Africa has been quite positive. The elegance of the theory has captured
many hearts in the community which otherwise did not have a clue of how to deal with
hundreds and hundreds of African languages seeming so dissimilar to each other. The four
macrofamilies of Greenberg gave an easy and comfortable basis for classification.
Today, after 50 years, Greenberg’s theory is more often criticized than supported.
Monographs and articles dealing with genetic grouping of the languages of Africa tend
either to revise or completely deny the four macrofamilies of Greenberg as an obsolete,
methodologically incorrect and factually inaccurate version. Moreover, the farther we go
in time from Greenberg’s book, the more legends start to appear about his activities, views
and research. This paper aims to identify and possibly clarify some of these legends of
contemporary African linguistics, as well as to propose a proper way to avoid these in the
future.
1. Greenberg’s Revolution
The 1963 work (and its early versions starting to appear since 1949) is sometimes
regarded as a revolutionary view on the genetic classification of African languages, which
was elaborated by Greenberg from scratch and drastically changed the landscape of African
linguistics.
This idea is quite incorrect. In reality, in many aspects of his theory, Greenberg
followed earlier hypotheses. The kinship between Afro-Asiatic (Hamito-Semitic)
languages was widely discussed ever since the middle of the 19th century, and Adolf
Erman in the beginning of the 20th century did not doubt much about the relatedness of
Egyptian and Semitic. The Afro-Asiatic hypothesis was formulated in much detail by
Marcel Cohen, Greenberg’s contemporary, in 1947 [Cohen 1947].
The idea of kinship between the languages of West, Central and South Africa,
known as Niger-Congo since Greenberg, was actually proposed a century before. In an
introduction to his collection of glossaries of languages of Mozambique, Wilhelm Bleek
[1856] wrote: “The languages of these vocabularies all belong to that great family which,
with the exception of the Hottentot dialects, includes the whole of South Africa, and most

MOTHER TONGUE
Journal of the Association for the Study of Language in Prehistory • Issue XVIII • 2013
50th Anniversary of J.H. Greenberg’s The Languages of Africa (1963)
18
of the tongues of Western Africa; certainly the Otsi, or Ashantee, the Bullom, and the
Timneh of Sierra Leone. The Gōr family, which includes the wide-spread Fulah, the Accra,
and the Wolof, may be considered as related to these; as may also the Ukuafi, spoken near
the source of the White Nile, and the Tumale in Darfur.”
It should be noted here that Bleek mentions Fula among those related languages of
Sub-Saharan Africa, another discovery attributed to Greenberg. The later “Hamitic Myth”
actively promoted by the giant authority of Karl Meinhof and Alice Werner in the early
20th century overshadowed Bleek’s hypothesis a little, but Diedrich Westermann actually
said pretty much the same as Bleek in his [1927] paper on Sudanic languages.
The same must be said about the Khoisan languages which, because of their
distinctive phonological features (click consonants) and even more distinctive
anthropological traits of their speakers have been identified and treated as a whole long
before Greenberg.
But what Greenberg actually did was to gather all these early scholarship attempts
at classification and to construct a logical and uncontroversial system out of them. Before
him, linguists were mostly guessing; Joseph Greenberg postulated these guesses into an
elegant theory which summarized early achievements and could encompass all languages
of the huge continent. Moreover, he was the first who tried to prove these early guesses
with a mass of lexical data, even though this method of his has always been regarded as his
weakest point.
2. Greenberg’s Method
The method of mass or multilateral comparison used by Greenberg in his 1963 work
(as well as many papers afterwards) has been increasingly criticized by the community of
comparative linguists. The main thesis usually proposed to disqualify this method states
that Greenberg projected too scant data to conclusions that were too far-reaching.
Indeed, in his 1963 research on Niger-Congo we can find 49 lexical items taken
from 186 languages (out of the 1,500 which are currently attributed to Niger-Congo). The
Niger-Kordofanian addendum to the paper adds to this 52 correspondences between Niger-
Congo and Kordofanian. The lexemes used mostly belong to the basic lexicon but not
limited to the Swadesh list or any other selection. No explanation is provided on the
principles of selecting either languages or lexemes, so they might look random to those
accustomed to a strict methodology. Taking into account the huge number of languages in
the macrofamily, one may imagine that it would be possible to find a cognate for almost
any lexical item which will be a pure coincidence in fact. No attempt was made to work
out a system of regular phonetic correspondences between the families within Niger-Congo
or to reconstruct the proto-language phonological system. Apart from a brief analysis of
noun class markers across Niger-Congo, no morphology was analyzed. So, one may ask,
how could Greenberg prove anything with such an imperfect and incomplete method?
However, the author of Niger-Congo actually did not intend to prove anything. He
always regarded his idea as a “first proposal” [1977], not a final solution to the genetic
classification of the languages of Africa. It would be completely wrong to view
Greenberg’s paper as the proof of the classification. He did not aim at creating an ultimate
systematic reconstruction of Proto-Niger-Congo, nor did he plan to establish a system of

MOTHER TONGUE
Journal of the Association for the Study of Language in Prehistory • Issue XVIII • 2013
50th Anniversary of J.H. Greenberg’s The Languages of Africa (1963)
19
phonetic correspondences or a morphological database. This goal could probably not be
achieved at that time with the volume of data accessible to date, when most African
languages were known by just their name or at most a small glossary of inaccurately
recorded lexemes.
Later Greenberg always emphasized that his method of multilateral comparison can
only be treated as a proposal, a dotted line which was still to be verified or modified by
means of the strict comparative method, of which Greenberg remained a passionate
supporter through all his life.
Consequently, it would be absolutely inappropriate to try to prove Greenberg’s
hypothesis without the utilization of the comparative method. Both mass comparison and
any other types of lexicostatistics widely used in African linguistic literature, will
inevitably suffer from a subjective approach. At the same time, the typological approach,
also quite popular in African linguistics in the West in the 20th century, may be
characterized by exaggerating the role of typological characteristics of languages in
defining their genesis. Speaking about Niger-Congo, anyone would now agree that the
presence or absence of the system of noun classes cannot be regarded as a valid proof of
kinship. The opinion that languages using noun classes should be included in Niger-Congo
and those lacking them should be excluded from it, is still seen sometimes in linguistic
papers, but is surely wrong. Nominal classification is a structural phenomenon and can
appear both as inherited and as areally spread.
No other method may be used for confirming Greenberg’s African hypothesis (or
suggesting another one in this field) but the comparative method. But was it ever done so
in practice?
3. Greenberg’s Legacy
In recent decades, “splitters” seem to overwhelm “lumpers” in African linguistics.
Their claim is that there is not enough proof that the mid-level families of African
languages are related, and that at least some of the larger groupings should be reconsidered
as areal or areo-typological units rather than genetic ones. Splitters continue arguing that
Greenberg’s language map of Africa was not confirmed by any strict scientific method
despite all the efforts made in the past 50 years.
The truth here is that almost no attempts in fact have been made to verify
Greenberg’s Niger-Congo hypothesis. This might seem strange but the path laid by Joseph
Greenberg to Proto-Niger-Congo was not followed by much research. Most scholars have
focused on individual families or groups, and classifications as well as reconstructions were
made on lower levels. Compared with the volume of literature on Atlantic or Mande
languages, the list of papers considering the aspects of Niger-Congo reconstruction per se
is quite scarce. Apart from efforts of Hans Mukarovsky [1976-1977] and John Stewart
[2002 etc.] who proposed their pilot versions of reconstructing the proto-language, not
quite in line with Greenberg’s ideas, not much has been done in this regard.
This is true both for lexicon and grammar, let alone phonology. As for today, most
objective problems that Greenberg must have faced regarding the amount and quality of
comparanda for comparative analysis have been overcome. Hundreds of new language
descriptions have appeared in free access, and a dozen new ones appear annually as theses,

MOTHER TONGUE
Journal of the Association for the Study of Language in Prehistory • Issue XVIII • 2013
50th Anniversary of J.H. Greenberg’s The Languages of Africa (1963)
20
monographs or articles. Numerous dictionaries follow grammars. The level of their
accuracy has increased greatly, since they all now mark tones (a rare advantage in the 60s)
and in general reflect more or less adequately the complicated phonetics of African
tongues. For a great number of languages, we can compare between two or more
descriptions. Sure, the number of languages left undescribed is still big, and new
discoveries are still made (the Russian Linguistic Expedition to West Africa alone has
proudly discovered two new Niger-Congo languages in the past four years), but the amount
of data is quite sufficient to make a decent lexical database including at least Swadesh
100/200-item lists. This is still to be accomplished.
Until the lexicon is treated systematically, no advance can be made in
reconstructing phonetics. It is fully understood that the reconstruction of an uncontroversial
phonetic system for a proto-language of over 10,000 years is a challenge. But other
postulated macrofamilies of the world, including Afro-Asiatic and Nostratic, can already
boast huge progress in this field. For Niger-Congo, no decent proposal has been made for
the reconstruction of its phonology which would include lexical data with correspondences.
Grammatical reconstruction is another big task, which may seem the most
important from the point of view of splitters of the 21st century. As Campbell & Poser
[2008] rightly note, morphological reconstruction is the most convincing evidence for any
deep language reconstruction, given the wide spectrum of various speculations a lexical
reconstruction may provoke. The Afro-Asiatic hypothesis was mainly based on
grammatical comparisons, and the Indo-European one started with them. In Niger-Congo
linguistics, the only aspect which was considered by linguists in more or less detail, is the
system of noun class marking. More research in morphology should generate a plausible
reconstruction of Proto-Niger-Congo which will turn the largest macrofamily of languages
in the world from a phantom into a reality.
This is how we can save Greenberg’s legacy.
4. Reconstructing Niger-Congo person marking
The following briefly summarizes the results of the comparative analysis of Niger-
Congo person marking systems published in [Babaev 2013] in Russian. The objective of
the research was to perform a consistent step-by-step analysis of person marking in as many
languages of the macrofamily as possible, and to suggest a systematic and uncontroversial
reconstruction of what the proto-language system of person marking could look like both
in terms of forms and meanings. Such a task was dealt with for the first time.
Data from over 650 Niger-Congo languages were used in the research, with all the
major families and groups of languages included. This allowed the creation of the biggest
database on personal markers in Africa and the provision of a necessary level of reliability
for the conclusions made. All the data was presented according to a unified structure of
four syntactic series of person markers: subject markers, direct / indirect object markers,
possessive markers, and independent pronouns widespread in Niger-Congo languages to
mark focus or topic or to serve in nominal clauses. Person marking was analyzed by family,
in accordance with the currently recognized genetic tree of the macrofamily, starting from
Benue-Congo and then up to Atlantic, Mande, Dogon and Kordofanian. For each family, a
reconstruction was made on the basis of lower-level reconstructions of person marking

MOTHER TONGUE
Journal of the Association for the Study of Language in Prehistory • Issue XVIII • 2013
50th Anniversary of J.H. Greenberg’s The Languages of Africa (1963)
21
systems for groups and subgroups where necessary. This means, e.g., that the Proto-Benue-
Congo reconstruction was not suggested before Proto-Bantu, Proto-Cross River, Proto-
Central Nigerian, Proto-Edoid and other reconstructions were elaborated. This cautious,
detailed and scrupulous approach allowed one to solidify the use of the comparative
method and increase the level of reliability of the results received.
As the main conclusion of the research, Greenberg’s idea of Niger-Congo should
be confirmed. The core families of Niger-Congo (the so-called Volta-Congo languages
including Benue-Congo and Kwa, Kru, Gur, Adamawa, Ubangi, and Atlantic) show
distinct genetic relationship in person marking, not only in the shape of individual
morphemes but also in paradigms which is a much more stronger argument for kinship.
The Dogon and Kordofanian data also confirm their status as Niger-Congo, even though
more distant from the core.
It appeared however that the same cannot be stated for sure for Ijoid and Mande,
the two families whose Niger-Congo affinity may not be reliably established through the
system of person marking. Both families must have either been more distant relatives of
Niger-Congo, departing from the proto-language community very long ago, or (in the case
of Ijoid) may not be non-Niger-Congo at all, having adopted some of the Niger-Congo
language features due to lengthy and intensive language contact. This conclusion, agreeing
with some earlier research (see [Dimmendaal 2011]), should however be reconfirmed by
further analysis.
At the same time, person marking systems of other languages of Central and West
Africa were also included in the analysis in order to demonstrate a proof by contradiction.
Person markers from Chadic, East and Central Sudanic, Songhai, and Kadugli-Krongo
languages, all adjacent to the Niger-Congo-speaking area, show drastic differences in form
and meaning from those of any Niger-Congo branch. Contrary to the statements of some
Niger-Congo skeptics, this macrofamily has its distinct borders and may not be projected
to the other families of African languages.
Some interesting conclusions on the internal classification of Niger-Congo can be
made from the research. The boundary between Benue-Congo and Kwa appeared to be
almost non-existent, and Kwa should probably be treated as a collection of branches of a
single node (Benue-Kwa) rather than a single node. Several groups, formerly regarded as
peripheral Gur languages, including Senufo, seem not to belong to Gur but rather should
form separate branches of Volta-Congo. The same may be true for Ubangi and especially
Adamawa, where several subgroups (e.g., Yungur) do not show any Adamawa affiliation
whatsoever.
All these conclusions, however, must be taken into account only as hypothetical,
since the system of person marking is not the only and ultimate marker of language kinship,
even though it can provide a strong argument on the issue. Further research in
morphological systems of Niger-Congo are now essential to provide additional evidence
for the validity of the macrofamily, including the research in noun class marking, in verb
extensions, verbal auxiliaries, numerals and other paradigmatic systems so important for
the proto-language reconstruction. The more detailed comparative research we present to
African linguistics, the clearer will be the picture that Joseph Greenberg sketched for us
fifty years ago.

MOTHER TONGUE
Journal of the Association for the Study of Language in Prehistory • Issue XVIII • 2013
50th Anniversary of J.H. Greenberg’s The Languages of Africa (1963)
22
Literature
Babaev 2013 - Бабаев, К.В. 2013. Нигеро-конголезский праязык: личные
местоимения. М., ЯСК. [Niger-Congo protolanguage: personal pronouns.
Moscow: “Languages of Slavic Culture.”]
Bleek, W. 1856. The languages of Mosambique: vocabularies of the dialects of Lourenzo
Marques, Inhambane, Sofala, Tette, Sena, Quellimane, Mosambique, Cape
Delgado, Anjoane, the Maravi, Mudasu etc. London: Harrison & Sons.
Campbell, L. & W. Poser. 2008. Language Classification: History and Method.
Cambridge: Cambridge University Press.
Dimmendaal, G. 2011. Historical linguistics and the comparative study of African
languages. Amsterdam – Philadelphia: John Benjamins.
Greenberg, J. 1963. The Languages of Africa. The Hague: Mouton; IJAL 29.1,
Bloomington: Indiana University Press.
Greenberg, J. 1977. “Niger-Congo noun class markers: prefixes, suffixes, both or
neither.” // Mould, M. & T.J. Hinnebusch (eds). Proceedings of the 8th
conference on African linguistics, Suppl. 7 to Studies in African linguistics. Los
Angeles: African Studies Center & Dept. of Linguistics, University of California.
Pp. 94-104.
Mukarovsky, H. 1976-1977. A Study of Western Nigritic. Vols 1-2. Vienna: Institut für
Aegiptologie und Afrikanistik.
Stewart, J. 2002. “The potential of Proto-Potou-Akanic-Bantu as a pilot Proto-Niger-
Congo, and the reconstructions updated.” Journal of African Languages and
Linguistics 23: 197-224.
Westermann, D. 1927. Die westlichen Sudansprachen und ihre Beziehungen zum Bantu.
Berlin: de Gruyter.