Bantu

Contents

• Introduction

• Bantu

• Swahili

• Noun classes and noun prefixes

• Munukutuba

• Semantic content

• Odds and ends

• Gemini’s effort on genders

• Conclusions

• References

Introduction

On page 80 (of my OUP paperback) of reference 1, in a section entitled ‘Using the language instrument’, Torey talks of the 14 categories of Bantu, a good example of the language-based schemas which brains seem to be keen on. This prompted the investigation of this matter which is described in what follows. This in the context of an extended argument that human consciousness and language are very much bound up together, an argument that excludes other primates, bats, cows, cats and cephalopods (amongst others) from consciousness as we know it.

Confused, I thought to ask Gemini about these categories, who appeared to do rather well. What follows is an attempt to check him.

In the course of preparing this post, there was a substantial digression around the Oneida and their pronominal pronouns. This is to be found at reference 12.

The opening snap was one of the first images offered by Bing on the search key ‘Bantu Gemini’, taken from reference 24.

Bantu

The large Bantu family of languages, maybe 500 of them with 350mn speakers, is spoken across a wide swathe of sub-Saharan Africa and is to be found at reference 7. While the probably older Khoisan family is spoken in the south-west of the continent and is to be found at reference 10. Part of the interest of these languages appears to arise from the fact that there are a lot of them and that you get all kinds of interesting shifts in both grammar and language as you move from one to another, with such a move perhaps amounting to a move from one area to a neighbouring area.

According to Wikipedia at reference 7, ‘…The most prominent grammatical characteristic of Bantu languages is the extensive use of affixes … Each noun belongs to a class, and each language may have [as many as twenty] numbered classes, [otherwise] somewhat like grammatical gender in European languages. The class is indicated by a prefix that is part of the noun, as well as agreement markers on verb and qualificative roots connected with the noun. Plurality is indicated by a change of class, with a resulting change of prefix…’.

According to Ruhlen at reference 6, the world’s languages can be organised into 21 phlya (a term more widely used in taxonomy) of which Niger-Kordofanian is one and of which last Bantu is a subordinate group, sometimes divided into forest in the northwest and savannah or central for the rest. Indo-Hittite, which includes most European languages, is another such phylum.

A Bantu language story which is amplified and largely confirmed by the DNA story of Reich at reference 11 – although Khoisan does not appear in his index. As it happens, a story which is summarised, after a fashion, in the opening snap.

Swahili

By number of speakers, Swahili is the most popular of these languages, widely used in east Africa, originally in the coastal ports, and accounting for around a quarter of all Bantu speakers. Swahili, which used sometimes to be called a Pidgin or a Creole rather than a ‘proper language’, includes lots of loan words from Omani Arabic – there is learned dispute about exactly how many – and was first written down in Arabic, although it moved to something like our alphabet under European, initially Portuguese, influence. 

From reference 14, I learn that, after French and German and a few of the nearby languages, Swahili is considered (by the US State Department) to be one of the easiest for English speakers to learn. Perhaps because, like English, Swahili has been something of a melting pot, during which some of its original complexity was tidied away.

While at reference 17, I see that it is getting a foothold in the US.

Noun classes and class prefixes

Wikipedia includes a useful summary of the Bantu noun (aka nominal) classes and their associated prefixes which is snapped above. Bearing in mind that there are a lot of Bantu languages and although the macro structure and organisation is conserved, there is also a lot of micro variation. Furthermore, the prefixes shown in this summary are morphemes and are subject to a lot of variation driven by context and phonetics rather than by geography. The best description of all this that I have found is at reference 15.

Noun classes are not about sex, are not about gender. But they often do distinguish animals from vegetables, singular from plural, they include locatives, which we would do in English with prepositions, they do diminutives and they can convert verbs to infinitives. So a rather eclectic mixtures of jobs. Noun classes drive adjectival agreement, more of it, but something in the way of Latin or French – but not English, where such agreement has dropped out, along with most gender marking.

An accessible description of locatives is to be found at reference 13. The more serious student is referred to reference 18, the 373 page book snapped above.

Munukutuba

Guided by Bing, my first stop had been reference 2, which was more about how gender changed across 179 of the Bantu languages, rather than providing tutorial material on the gender system more generally. But it did offer an impressive looking consensus tree, which took me to reference 19 which contained a larger version and which is snapped above.

A larger version which is itself exploded in the Supplementary Material, from which the top right-hand corner is snapped above. Perhaps of interest to a specialist.

Reference 2 also pointed to reference 3, about gender with the Munukutuba, which looked more promising, a Creole language spoken in and around the Congo and which is introduced at reference 4. Creole languages are introduced at reference 20: a term which is widely used and which is convenient – but which strikes me as rather dated. A pity no-one has come up with something a bit more neutral.

In the course of her study of the Munukutuba noun class system, Buchanan looked at around 1,500 words, of which we are given the translation and noun class of something over 200. The class of a noun is marked by its prefix and those for Munukutuba are listed in the left-hand column of the table below. Other Bantu languages have rather more and it seems likely that it is the process of Creolisation which has thinned them out here.

Class 1a is words which never had a prefix, words which have lost their prefix and imports which have not been given one. We suppose that a fluent speaker knows all the words concerned. These have the null ‘Ø-’ prefix. While the ‘N-’ prefix is null in a different way, in that the first letter of the noun stem is nasalised (a term in phonetics).

If we allow the nasal ‘N-’ and the null ‘Ø-’ prefixes, then the noun class of every noun is marked by a prefix. A noun is audibly attached to its class, an attachment reinforced by the agreement of any adjectives there may be. Pronouns may get into the story too. Munukutuba uses 12 prefixes, rather less than most other Bantu languages, of which eight cover four pairs, a pair being a singular class and its corresponding plural class. By which we mean if ‘rock’ is a noun stem, then something like ‘AB-rock’ is the singular form ‘rock’ and something like ‘BC-rock’ is the plural form ‘rocks’, where ‘AB’ and ‘BC’ are the prefixes for the two classes. While in English we don’t bother with classes and just mark the plural with one of various suffixes, mostly involving an ‘s’. These four pairs are classes [1, 2], [3, 4], [5, 6] and [7, 8].

With one complication being that the plural class 2 is shared with some other singular classes – 1a, 9 and 11. I have not yet found out why class 1a was not given its own class number: perhaps it was nothing more than keeping the diagrams tidy. A bit like popping 29b into a row of houses, renumbering from scratch not really being an option. With Lowden Road in Chippenham, which I came across recently, being a particularly difficult example of same.

And another complication being that lots of words which take a singular prefix do not have plurals and lots of words which take a plural prefix do not have singulars. Class 14 only contains abstract nouns, none of which have plurals. This accounts for the one class column in the table above. 

Plus, there are some irregularities, accounting for the half a dozen or so small, off-diagonal entries.

Buchanan uses the term ‘gender’ to refer to a class plus its associated plural class, if present. Remembering that it does not seem to have anything to do with ordinary gender, that is to say sex: it just occupies a gender like slot in the grammar.

The rather simpler story from Wikipedia at reference 4.

Semantic content

Bearing in mind that this study used just over 1,500 words out of some much larger total, the small class 1 contains just words for humans, including the word for human. Other such words are scattered across most of the other classes.

Liquids and paired body parts mostly go to classes 5 and 6, with class 5 for the few liquids which come discretely, for example the drops from a dripping tap or tears from the eyes.

Animals are all over the place, outside of class 1, although mainly in 5, 7 and 9. I don’t suppose speakers of Munukutuba needed to trouble themselves with the colonial animals like coral and some small jelly fish - for which see reference 21. Furthermore, in the course of checking up on colonial animals, I stumbled on reference 22: a useful reminder that even the definition of an individual gets a bit tricky at the margins.

Class 14 is quite tidy, containing only abstract nouns – although such nouns are to be found in other classes too.

So it would seem that there is some correlation between the classes of words and their meanings. But not all that much. Perhaps very much the same as gender in French, except that there is rather more of it.

An analysis which does not seem unreasonable in the circumstances: things of interest to a speaker of Munukutuba. But perhaps more interesting in that it reminds me how tricky these sorts of seemingly simple classifications are – which always seem to involve words which should belong in more than one group.

Odds and ends

There is a small community – the Siddi community of India and Pakistan – say around 50,000 souls – of African descent in the sub-continent, thought to have arrived at the time of the Muslim invasion of the 7th century. They speak – or spoke – one of the Bantu languages. See reference 8.

 was reminded of the nonsense classification, retold by Borges, possibly derived from the work of a 17th English divine, the top of which is snapped above. See reference 9.

In the course of taking a look at Munukutuba, I was puzzled by the noun class prefix ‘N-’, where all the other prefixes were lower case and this one, this upper case letter, did not appear in the alphabet. More by luck than judgement, I decided that perhaps it was something to do with nasal phonetics, but got no further. Neither Bing nor Google was much help, so I tried Gemini who did well, his story only being marred by two elementary and visible blunders in one of the examples he used to back up his story – in fact, two instances of almost the same blunder, the second after I had complained about the first. ‘N-’ is not a letter, rather it stands for the nasalisation of the first letter of the noun stem concerned. An operation which one supposes does not destroy its function as a marker.

My mother used to talk of a language without written literature, without a proper history barely being a language at all. Which might be a rather old-fashioned view, but she did have a point. A point which I have not looked into here at all. While more recently I have been reading Ong at reference 25, where he talks of there being step changes in consciousness when we moved up from oral to written and then from written to printed. A point which has not been taken on here.

Incidentally and unusually, my copy of the book at reference 25 comes from the old-style second-hand bookshop at reference 26. A place with a shop you can visit, rather than some warehouse on some provincial industrial estate. Although, to be fair, it is a shop fashioned out of a house in an otherwise residential street in Chalk Farm.

I used to own a bookcase on wheels, like those shown in the snap above, a bookcase which I wrongly knew as a carrel, solidly made but long retired. An error which I knew about back in 2012 and noticed at reference 27, but an error which subsequently floated back to the top again. I feel sure that it came up again quite recently, but have so far failed to remember why or where.

Gemini’s effort on genders

The start of the exchange is snapped above, starting with my question.

Gemini starts quite well, but then we get: ‘Torey highlights the complexity of this system in Bantu languages, which can have around 14 or more noun classes. This contrasts with European languages like English, which have a simpler gender system (masculine, feminine, neuter) or have lost grammatical gender altogether’. Which is the sort of thing that Torey might have done – but didn’t. Torey actually goes on to make different points. Gemini is joining up the dots when he shouldn’t.

There is also a drafting awkwardness. The paragraph would have been better ‘… 14 or more noun classes. This contrasts with European languages, which have a simpler gender system (at most masculine, feminine, neuter) or English which has lost grammatical gender altogether’. 

Gemini does not tell us that these classes are audibly and visibly marked by the use of prefixes: a noun does not just belong to a class, that among Bantu speakers, we are continually being reminded of the fact. The sort of omission for which one might mark down a pupil’s or student’s assignment.

Gemini goes on to say that ‘Torey uses the example of Bantu noun classes to illustrate the diversity of human thought and language, and to challenge the notion that European languages are somehow superior or more logical. He argues that the complexity of Bantu noun classes reflects a different way of categorizing and understanding the world’. Which, again, is the sort of thing that Torey might have done – but didn’t. Again, Gemini is joining up the dots when he shouldn’t.

It is rather as if Gemini is guessing at what Torey might have said on the basis of general knowledge, rather than having access to or making used of this text in particular – although a copy is available on the Internet, turned up by a routine search.

The closing paragraph of this first answer is fair enough: ‘It's important to note that the exact number of noun classes can vary across different Bantu languages. Also, linguists prefer the term "noun class" to "gender" when discussing Bantu languages to avoid confusion with grammatical gender in European languages’.

I followed up with a question about Monukutuba.

Gemini does quite well here, and I did not spot any howlers, although ‘7-9’ is a bit on the low side. On the low side, but not really disturbing the line of thought in a serious way. But the drafting is, once again, awkward. ‘It typically has around 7-9 noun classes, depending on the specific analysis’ would be better ‘It has around 12 noun classes, although different analyses have come up with a range of numbers here’.

Gemini has chosen to focus on Monukutuba being a Creole language, a simplified amalgam of its parent languages. Which is fair enough given the ‘anything in particular’ bit in the question. Gemini is being invited to choose. Nevertheless, we are not told much about this language, as opposed to its genesis. And on this last, Wikipedia includes the useful reminder that a lot of the ground work here, as in so many other places, was done by or for missionary societies, a by-product of trying to bring these people to the Lord. Perhaps a point a teacher would bring out in discussion of a pupil’s or student’s assignment.

Perhaps an easier question as it is more directly about Monukutuba, rather than about what Torey might or might not have said about Bantu genders.

So, all in all, he did quite well. Almost as good as Wikipedia, from where he may well have drawn a lot of his information. I suppose we start to worry when it all starts to get a bit circular. To use a bit of jargon from a neighbouring discipline, re-entrant rather than strictly feed-forward. With one consequence of the former being that the result can be very hard to analyse: one might quite like the answer, but it is hard to see exactly how one got there.

Conclusions

I suppose that I have done here rather slowly what Gemini managed rather quickly: a trawl through likely looking material on the Internet, followed by an attempt to sum it all up.

He is still making elementary mistakes, of varying significance.

I wonder about the boundaries of his knowledge. We are told that he has hoovered up huge tracts of knowledge from the internet. Which, understandably, does not include the sort of knowledge which needs to be extracted from databases. But it also does not appear to include the copy of the Torey book in question. I get the impression that his knowledge is closed: an edifice has been built which is all very impressive, but which cannot be modified on the fly, which limits his ability to respond to errors and omissions. For that he has to go back to the factory. In IT-old-speak, it is a compiled system which has to be recompiled if you want to make a change.

On the classes more generally, I have been impressed by these prefixes, which forcefully remind one of the classes to which the nouns one is using belong, with a lot of these classes being about classifying nouns, not with conversions like mapping a verb onto its infinitive, a noun onto its plural or a noun onto a diminutive. Reminders which must result in clusters of information in the brain being activated, adding some flavour to the cluster of information which makes up what is experienced as consciousness. These noun clusters do drive consciousness to that extent.

PS: and while we are on bookshops, a puzzling email, part of which is snapped above, has just turned up. So far I do not have a clue who these people are. Maybe it will float up into to the light...

References

Reference 1: The crucible of consciousness: A personal exploration of the conscious mind – Zoltan Torey – 1999. 

Reference 2: A typology of northwestern Bantu gender systems – Francesca Di Garbo, Annemarie Verkerk – 2022. 

Reference 3: The Munukutuba noun class system – Deborah L Buchanan – 1997. 

Reference 4: https://en.wikipedia.org/wiki/Kituba_language. For Munukutuba.

Reference 5: Bantu grammatical reconstruction - Meeussen A E – 1967. 

Reference 6: A guide to the world's languages. Vol. 1: Classification – Merritt Ruhlen – 1987. Book_133.

Reference 7: https://en.wikipedia.org/wiki/Bantu_languages. 

Reference 8: https://aaregistry.org/story/the-siddi-people-of-india-and-pakistan/. 

Reference 9: https://en.wikipedia.org/wiki/Celestial_Emporium_of_Benevolent_Knowledge. 

Reference 10: https://en.wikipedia.org/wiki/Khoisan. 

Reference 11: Who We Are and How We Got Here: Ancient DNA and the new science of the human past - David Reich – 2018. 

Reference 12: https://psmv5.blogspot.com/2024/11/oneida.html. A digression.

Reference 13: Chapter 25: Locatives – Jochen Zeller – 2021. 

Reference 14: The story of Swahili – John M. Mugane – 2015. More the story of Swahili speaking people than of the language itself. 

Reference 15: Swahili Language Handbook – Polomé, Edgar C – 1967. 250 pages of typescript.

Reference 16: https://www.cal.org/. The original home for reference 15.

Reference 17: https://swahili-institute.org/. At Chicago.

Reference 18: Les locatifs en bantou – Grégoire, Claire – 1975. Tervuren: Musée royal de l’Afrique Centrale.

Reference 19: Bantu expansion shows that habitat alters the route and pace of human dispersals – Rebecca Grollemund, Simon Branford, Koen Bostoen, Andrew Meade, Chris Venditti, Mark Pagel – 2015. 

Reference 20: https://en.wikipedia.org/wiki/Creole_language. 

Reference 21: Life in the Colonies: Learning the Alien Ways of Colonial Organisms – Judith E. Winston – 2020. A digression into natural history.

Reference 22: Genetic variation and the natural history of Quaking Aspen: The ways in which aspen reproduces underlie its great geographic range, high levels of genetic variability, and persistence – Mitton JB, Grant MC – 1996. 

Reference 23: https://psmv5.blogspot.com/2024/10/denham-two.html. The last outing for quaking aspens, presumably the European variety.

Reference 24: https://support.ancestry.co.uk/s/. 

Reference 25: Orality and Literacy: The Technologizing of the Word – Walter J. Ong, S.J. – 1982.

Reference 26: https://www.waldenbooks.co.uk/. 

Reference 27: https://pumpkinstrokemarrow.blogspot.com/2012/09/ambling.html.