Timbuktu Manuscripts.

I’ve made several posts about the manuscripts and libraries of Timbuktu (put “Timbuktu” into the search box to find them), but I found some details in this story particularly interesting:

Just about 250,000 old manuscripts from the libraries of Timbuktu still survive in present-day Ethiopia. Also, thousands of documents from the medieval Sudanese empire of Makuria, written in at least eight different languages were dug out at the southern Egyptian site of Qasr Ibrim. Thousands of more old manuscripts have equally survived in the West African cities of Chinguetti, Walata, Oudane, Kano, and Agadez.

Upon the real and present dangers posed by fires, insects, and plundering, some one million manuscripts have since survived from the northern edges of Guinea and Ghana to the shores of the Mediterranean. National Geographic even estimates that 700,000 manuscripts have survived in the city of Timbuktu alone.

Ethiopia — that’s quite a journey! If I ever knew about Makuria (Greek Μακουρια, Arabic al-Muqurra), I’d forgotten. And how was “Chinguetti” derived from the Arabic name شنقيط‎ Šinqīṭ? (Thanks, Trevor!)


  1. Jen in Edinburgh says

    Chinguetti seems to have been originally Berber, so the English (French?) and Arabic names might have both come from some other form.

    I watched the Michael Palin episode where he visits it quite recently – I think he talks about the libraries there.

  2. David Eddyshaw says


    Should you ever want to learn a second word, there’s a fairly good grammar of Old Nubian by Gerald Browne, published by LINCOM.

    My favourite Makurian is Prince George of Makuria


    about whom very little seems really to be known. It’s just that the very name encapsulates the fact that the world is so much more complicated and interesting than you might think. The royal courts of Makuria and Alodia may actually have been Greek-speaking.

  3. David Eddyshaw says

    We’ve talked about Makuria before (of course; we’ve talked about everything before. We’re just that cool.)


  4. From the WP article on George:

    Their description of the events is not always conclusive and they provide at several points contradictory information.

    I like that, actually. It’s a letdown when you have several sources who are just redundant copies of the same thing (or of each other).

  5. The etymology of Senegal is thought to be related to the Zanhaya or Sanhaja tribe, which is spelled Zenhaga on this early map:

    Other early maps show the Regnum Senegae:

    And I read that the Arabic locational trinomial Shinqiti isn’t typically meant as a reference to the town/city of Shinqit, but to the entire region – Bilad Shinqiti, encompassing all of Moorish NW Africa. This is apparently sometimes spelled in a way transliterated Shinaqiti, which seems to further establish that it’s related to the Sanhaja/Zenhaga.

    So I’m putting forward the hypothesis that Chinguetti had a French development of its own via Bilad Shinaqiti. But I knew nothing about the region and this is just what half an hour on google led me to. Another half hour and I’ll have the Songhay and maybe even Seneca in there, so I should stop and see if anyone thinks it’s plausible. Can you page Lameen?

    One of the odd things to me is that this important ancient town doesn’t show up on old maps. Perhaps that’s because caravan stations were outside the perspective of early European explorers, and the Islamic maps either aren’t extant or aren’t accessible online to an English speaker.

  6. And how was “Chinguetti” derived from the Arabic name شنقيط‎ Šinqīṭ?

    Catherine Taine-Cheikh offers an etymology of the name in the final paragraph on p. 310 here, in her chapter “Mauritanian and West Saharan Arabic” of the volume Arabic Historical Dialectology: Linguistic and Sociolinguistic Approaches (2018). The name is explained as being Soninke si n-gede “well of the horse” (alternatively written si-n gede). In this, is “horse”, cognate with Maninka and Bambara so and Susu sona within Mande. It is apparently a West African areal term (perhaps originally diffused from Mande?) also seen in Temne asoe, Kru so, Ewe and Fon so, Yoruba esin, and Edo esi. The n is the Soninke definite marker, which is a nasal bearing a low tone and follows the noun it makes definite, if I understand the construction correctly (although the orthography used in Taine-Cheikh’s article hyphenates to the following noun). And gèdé is “well, spring of water, source”.

    Soninke has a series of initial consonant mutations after certain morphemes ending in nasals. In these alternations, /s/ apparently alternates with /t͡ʃ/ (or /c/?) after a nasal. (There is a description of these mutations here.) Perhaps this alternation is responsible for the borrowing into Arabic with initial ش /ʃ/? The writing with qaf ق in شنقيط‎ would doubtless reflects the fact that inherited etymological Arabic q ق is realized as /g/ in Hassaniya Arabic.

    Looking into this has been very fun and informative for me, which is why I wrote up the results for others. But I am very much out of my bailiwick in western Africa. I hope more knowledgeable people can step in and correct my errors, and perhaps explain the ط for Soninke /d/.

  7. David Eddyshaw says

    I don’t know whether the Ewe and Yoruba words for “horse” are ultimately borrowed from Mande (on first principles I’m a bit leery of monosyllabic correspondences just based on initial consonants and a possible vowel) but it’s not at all impossible a priori. There is no reconstructable Proto-Oti-Volta word for “horse”: the etymon seen in Kusaal wief is purely Western Oti-Volta, that in Moba taanm is confined to Gurma, and so on. Some Eastern Oti-Volta languages have a su form IIRC, but I’d need to look it up when I get home.

    Two points that may be relevant: in the Mossi-Dagomba kingdoms, horses are strongly associated with the royal clans that claim to have originated elsewhere; and horses are not found in the Guinea (forest zone) because of sleeping sickness. So both the animal and its name are candidates for borrowing on first principles.

    On the other hand, the wief Western Oti-Volta form is highly irregular in a way reconstructable to Proto-Western and is clearly in no way a recent loan.

  8. West Africans apparently confused horses with asses.

  9. David Eddyshaw says

    Not in any language I know of, they didn’t. And “ass” (bʋŋ in Kusaal) actually is reconstructable for Proto-Oti-Volta.

  10. https://en.wiktionary.org/wiki/asinus#Latin

    Yoruba “esin” is particularly close.

  11. And for the followers of Scando-Niger-Congo theory:


  12. J.W. Brewer says

    off-topic, but of potential Hattic interest (from an appreciation of the just-deceased quiz-show host Alex Trebek):

    Then there was the joke that was turned on Trebek. Preparing for a show, he said, he saw a category called “when the Aztecs spoke Welsh,” filled with difficult pronunciations.

    “I’m making dialectical [sic] marks to help me pronounce these words correctly. And then we’re about to tape this one and I said, ‘What happened to the “Aztecs speaking Welsh?”’

    “And (a producer) said, ‘Have you looked at your watch or your calendar? It’s April 1.’”

  13. West Africans apparently confused horses with asses.

    There have been worse confusions.


    From Proto-Italic *wiros, from Proto-Indo-European *wiHrós. Cognates include Sanskrit वीर (vīrá), Old Prussian wijrs, Lithuanian vyras, Latvian vīrs, Old Irish fer, Old Norse verr, Ossetian ир (ir, “Ossetians”) and Old English wer (English were-).

    (Classical) IPA(key): /wir/, [wɪr]
    (Ecclesiastical) IPA(key): /vir/

    vir m (genitive virī); second declension

    1. man in the sense of “adult male human”
    2. adult, mature, or grown man
    3. brave or courageous man, hero, warrior
    4. husband
    5. (military) foot soldier


    Compare with Georgian ვირი (viri, “donkey, ass”).


    Chechen: вир (wir)
    Ingush: вир (wir)
    Bats: ვირ (vir)


  14. J.W. Brewer says

    “Timbuctoo, Tumbuto, Tombouctou, Tumbyktu, Tumbuktu or Tembuch? It doesn’t matter how you spell it. The word is a slogan, a ritual formula, once heard never forgotten.” The late Bruce Chatwin, in a piece included in a 1997-published anthology but apparently originally published in 1970. Note that the “Timbuktu” spelling that now seems conventional in English is not even among the options listed.

    Per the google books n-gram viewer (although maybe it’s rare enough that the data is a little unreliable?), “Timbuktu” did not permanently pull ahead of “Timbuctoo” (the spelling used by A.A. Milne and thus the first one I knew as a child, although admittedly in a context where Norway was spelled Norroway for metrical purposes …) until 1996. Which is more recently than I would have supposed.

  15. David Eddyshaw says


    Indeed. I had neglected the Scandi-Congo dimension. Your reasoning is ineluctable.

  16. Re Clement Crutwell’s map of Africa: A 1799 map that shows Lake Tanganyika in the right place and proportions is in advance of maps from decades later. It’s one of several very educated-looking guesses, including not showing the Senegal and Gambia as mouths of the Niger. I can find no discussion of this via some quick googling. I wonder what kind of sources he had?

  17. Two points that may be relevant: in the Mossi-Dagomba kingdoms, horses are strongly associated with the royal clans that claim to have originated elsewhere; and horses are not found in the Guinea (forest zone) because of sleeping sickness. So both the animal and its name are candidates for borrowing on first principles.

    I wish comments on this platform had “like” buttons, because I ❤️ this. Very interesting.

  18. PlasticPaddy says

    @rodger c
    For the nile region a similar map (1790) was made by James Bruce based on his own travels.
    The Crutwell map has other inaccuracies, see the corresponding image files in wikimedia.

  19. David Eddyshaw says

    I see from Wikipedia that nobody even knows the origin of the name “Timbuktu.”

    It’s clear to me that it derives from Mooré tembogdo “holes in the ground”, on account of the city being founded before the invention of tarmaq (as the substance is known in Tamasheq.)

  20. I have a serious but tangential question that I have long harbored about the Dogon horse culture. Why are antique Dogon equestrian carvings so valuable? A small one from the early twentieth century will be valued in the tens of thousands of dollars. Larger and older ones go up in price from there. Are they particularly rare, due to their associations with group leadership? Is there something about them stylistically that makes them more appealing to collectors than other kinds of African carvings? Is it something else? What?

  21. I believe someone in 1960s started a rumor that Dogons are aliens from Sirius.

    Obviously an ancient alien artifact ought to cost serious money…

  22. David Marjanović says

    tembogdo “holes in the ground”

    Don’t tell me it’s the original shithole country…?

  23. The people ride in a hole in the ground.
    New York, New York, it’s a hell of a town!

  24. David Eddyshaw says

    For the sake of both of you who have been counting the minutes until I could get home and look it up: the Eastern Oti-Volta “horse” words go: Byali sangə, Ditammari tasãnta, Nateni sãnda (all cognate), but Waama suka. (Waama diverges from the other three Eastern languages in numerous ways and is better regarded as a separate Oti-Volta branch; I’m the the process of getting a treatise which demonstrates this beyond reasonable doubt into sufficiently coherent form that it’s not too embarrassing to plonk on academia.edu.)

  25. David Eddyshaw says

    Why are antique Dogon equestrian carvings so valuable?

    I don’t know, but obviously that’s no reason not to comment …

    Nigel Barley, in The Innocent Anthropologist (much the funniest book about doing anthropological fieldwork in West Africa, against stiff competition) makes somewhat pointed remarks to the effect that the Dogon have been rather successful in selling their traditional culture to the more hippyish kind of anthropologist; I suspect the Dogon themselves are more sinned against than sinning, but they have managed to project an aura of Ancient Wisdom even to their African neighbours (the [very good] Souleymane Cissé film Yeelen illustrates this rather well.) It all helps with brand awareness, anyhow.

    The Sirius nonsense that SFReader alludes to (horrible pun, SF) is this


    Another aspect, I suspect, is that with traditional Dogon artefacts, like distant cousins, there’s a limited supply. Most Dogon nowadays are orthodox Sunni Muslims.

  26. the [very good] Souleymane Cissé film Yeelen

    It’s a wonderful movie which I used to see in theaters every chance I got (New York, New York, it’s a hell of a town); I’m planning to show the DVD to my wife.

  27. Trond Engen says

    The Sirius nonsense

    A Dogon dog story.

  28. David Eddyshaw says

    It’s a wonderful movie

    Merely an excuse to plug a film I very much like, but I would recommend Idrissa Ouedraogo’s Yaaba even more highly. It provoked waves of nostalgia in me, which may have clouded my judgment a bit, but damn, it’s good. It also features child actors who can act.

  29. like distant cousins, there’s a limited supply

    we’re down to the tens, and this is why.

  30. I would recommend Idrissa Ouedraogo’s Yaaba even more highly.

    I’ll keep an eye out for it. I still proudly wear the T-shirt I got at the first of the New York African Film Festivals (brochure [pdf]).

  31. Here’s a nice interview with the festival founder from a few years ago; that year half of the films were made by women, hurray!

  32. Also, the Davids (E and L) are referring to this magnificent song.

  33. Mister Zoot Horn Rollo, hit that long lunar note, and let it float!

  34. @David Eddyshaw: I think Temple blew it with that cover on The Sirius Mystery. If he was going to put the head of a mysterious Egyptian jackal on the cover, it should have squared-off Set animal ears.

  35. January First-of-May says

    В далёкой бухте Тимбукту
    Есть дом у Сары Барабу,
    Сара Барабу, Сара Барабу,
    Есть у неё корова Му…

    IIRC, the story that the lyrics of this song come from was originally written in Norwegian. I wonder what the original Norwegian version of this song looked like.

  36. Those guys remind me of a skiffle band.

  37. Interesting guess. The guys themselves modeled their band on Beatles.

  38. I don’t know whether the Ewe and Yoruba words for “horse” are ultimately borrowed from Mande (on first principles I’m a bit leery of monosyllabic correspondences just based on initial consonants and a possible vowel) but it’s not at all impossible a priori.

    Perhaps at least some of the West African words for horse with (-)s-, Yoruba ẹṣin, Edo esi, etc., might ultimately be diffused from Berber? There is common Berber word “horse (esp. one for riding), stallion” of approximately the right shape (that is, short and with an s):

    Tahaggart ays, pl. iysân “horse”
    Tumẓabt yis, pl. iysan “horse”
    Tamaziγt iyyis, ayis, pl. iysan, isan “racehorse, riding horse”
    Tacelḥit (Chleuh) ayyis, pl. isan “horse”
    Tarifit (Riffian) yis, pl. iksan “horse”
    Tacawit (Chaouia) yis, pl. iksan “horse”
    Zenaga ic, pl. icen “horse”

    There have been attempts (as here, page 371) to derive this Berber word from Arabic حصان ḥiṣān, plural أحصنة ʾaḥṣina “horse, stallion, stud”, but apparently these are untenable. Or could the Arabic word have reached Yoruba and Edo by another route, filtered through mediating languages of Central Africa?

    In any case, it would be interesting to know if there are other words diffused from Berber in these languages and what typical changes they undergo during diffusion.

  39. David Eddyshaw says

    There are Kanuri words and Kanuri-mediated Wanderwörter in West Africa; my own mental map of Africa counts Chad as an honorary part of West Africa, but I suppose it’s properly speaking the northernmost outpost of Central Africa. I don’t think there’s much evidence for loans originating further south, and in general the cultural diffusion has always tended to go north-south in West Africa, for historically obvious reasons.

    FWIW, the Kanuri word for “horse” is fə̂r, which looks vaguely as if it might be from Arabic faras, but I know too little about Kanuri historical phonology to know if that’s really likely (not impossible in principle: for example Hausa kasuwa “market” is – unlikely though it looks – from Arabic su:q via Kanuri.)

    Lameen is the man to tell us about loanwords from Berber; one I can testify to is hal/hali/har “up to, until”, which is all over the savanna and sahel zones of West Africa. The “camel” words are very often ultimately from Berber (not surprisingly), including e.g. Kusaal yʋgʋm and Hausa raƙumi. A lot of words which are ultimately from Arabic have got Berberised along the way, too.

    It’s striking just how little similarity there actually is among the many and various words for “horse” in West Africa; this also means that the great variety of forms out there does tend to make it a bit too easy to find lookalikes.

  40. This scattering of roots matches the polyphyletic racial origin of the west Africann ponies, says Blench p . 93. He also speaks there of possible feral stage (which can explain the variety of forms)

  41. Interesting idea. Yoruba and Edo seem unlikely to have had direct contact with Berber, so if such an etymology can be sustained it would presumably be via Mande. Mande languages certainly include Berber loans; but for the Mande forms the similarity is down to a single consonant…

  42. David Eddyshaw says


    Good find!

    I’m not sure how a feral stage would explain the variety of words for “horse”, exactly, though. Within Oti-Volta (to retreat to my comfort zone), “snake” is securely reconstructable for the protolanguage, for example, and most snakes seem pretty feral. It would help to explain the variety if horses were not common in the relevant areas until relatively recently, whether feral or domesticated.

    It certainly is very odd, though, that the various Oti-Volta stems for “horse” (at least wed-, de-, wusum-, ta:m-, san- and su-) are not only definitely not cognate with each other, but don’t look very likely to be related by borrowing either. Nor do any of the forms (as far as I can see) have any evident kenning-like meanings, unlike some “bear” words in Indoeuropean (nothing like “fast animal” or the like.) It’s all very mysterious.

    It’s not the only example, though: “lion” is similarly not reconstructable for Proto-Oti-Volta, for example. Admittedly there are few-to-no lions left in the area now, but the animal bulks pretty large in folklore etc, and it seems surprising that there’s no common root for such an iconic beast. (The Dagomba royal clan are not allowed to eat them, which must be a hardship.)

    But then there are no reconstructable forms for “brother” or “sister” either, and I dare say that brothers and sisters are not a recent development in West Africa.

    I wonder if tabu avoidance might explain the absence of a common “lion” word … there certainly are languages in which “lion” is expressed by terms which mean “savanna lord” and so forth. Doesn’t seem to help much with horses, though. Tabu horses …

  43. David Eddyshaw says

    But then: horse, ech, hestur, cheval, Pferd, άλογο, kůň … obviously the Europeans must also have had a tabu on naming the horse … or not. As you may think.

  44. David Eddyshaw says

    Actually, that list indirectly suggested an alternative idea: there might have been a Proto-Oti-Volta word for “horse” which happened to be either very short and/or to contain consonants particularly prone to loss by common historical phonetic developments, prompting the creation of replacements; there’s a parallel there in the Proto-Oti-Volta word for “neck”, which seems to have been something like *ɲu (where even the -u is a singular noun class suffix.) The regular outcome of this in Kusaal would have been zero.

  45. The regular outcome of this in Kusaal would have been zero

    “The horse with no name” ! People are claiming there’s a tabu in order to avoid admitting that the word has gone down a phonological rabbit hole.

  46. Proto-Algonquian word for “whiskey”, Proto-Slavic word for “king”, Proto-Oti-Volta word for “horse”…

  47. The Dogon, as is well known, are Nommonalists.

  48. David, the full quotation: “This scattering of roots matches the polyphyletic racial origin of the west African ponies, suggesting that the horses from which they are descended came across the Sahara along a number of different trade routes.”

    An invasive species takes it to the extreme: imagine you’re already scattered all over the continent and speak different langauges and then, independently (“you” doesn’t work well here:-/) you start seeing ponies. You have to come up with different words for them. “Different trade routes” are less radical, they still originate in the same region.

    But then: horse, ech, hestur, cheval, Pferd, άλογο, kůň …
    And the Dutch/German paard/Pferd is identical to proto-Semitic *pVrd🙂

    Tabu horses … Somewhere on Wikipedia they mention a tribe where marrying a speaker of the same language is a tabu (I don’t know how they distinguish between “same language” and “a member of the same tribe”).

  49. David Marjanović says

    There are cases like the Rio Vaupés region in Brazil where it’s part of the shared culture of several tribes that you only marry outside your (father’s) tribe, and almost all tribes coincide with languages. That’s one of the best ways to get everyone to speak like 6 languages fluently, by the way.

  50. David Eddyshaw says


    German Pferd is not connected with Semitic:


    but I suspect you know that.
    The tabu on marrying speakers of the same language is most famously associated with the Vaupés region of Amazonia:


    I don’t think Blench’s account of ponygenesis really works; it seems like a just-so story to me, but YMMV. In particular, I very much doubt whether “the scattering of roots matches the polyphyletic racial origin of the west African ponies.” I’ve never come across the notion that there are dozens of genetically different local kinds of horse in Burkina Faso, northern Ghana, northern Togo and northern Benin, for example, and I don’t believe it. I suspect that by “matches”, Blench just means that there are lots of different ones in both categories, but unless there are actual correlations between horse names and horse genes I don’t think that signifies very much.

  51. There are cases like the Rio Vaupés region in Brazil where it’s part of the shared culture of several tribes that you only marry outside your (father’s) tribe, and almost all tribes coincide with languages.


    The tabu on marrying speakers of the same language is most famously associated with the Vaupés region of Amazonia

    Two Davids with but a single thought!

  52. David Eddyshaw says

    I am not worthy …

    Apparently the linguistic exogamy of the Vaupés doesn’t lead to quite the polyglot paradise you might think, as children tend to marry speakers of their own mother’s language. The convention is that your “own” language is the one that your father speaks, which is therefore exactly not your “mother tongue.” You address people in their own language and they reply in yours. So two languages is the minimum, but you don’t necessary end up with six languages. Most often, you end up speaking to your spouse in the same language that you use to speak to your mother.

    The languages have in fact mostly converged considerably, but only along dimensions that the speakers aren’t consciously clearly aware of, like phonology, syntax and semantics (and even morphology, to some extent.) Per contra there is a strong tabu against mixing vocabulary, which people are aware of. Like everyone else (except linguists), they conceptualise languages as “bags of words.”

  53. …which in extreme cases leads to “mixed languages” like Callahuaya, which assiduously hold on to their vocabulary, while shifting the grammar entirely.

    E too.

  54. Huh:

    The E language’s unusual pinyin-transliterated name, which is also an autonym, consists of a single letter e. The character, which is written “诶” in Simplified Chinese and “誒” in Traditional Chinese, usually denotes an expression of affirmation. The language’s speakers also refer to their language as Kjang E.

  55. January First-of-May says

    But then: horse, ech, hestur, cheval, Pferd, άλογο, kůň …

    And of course лошадь, though IIRC it’s a Turkic borrowing.

  56. David Eddyshaw says


    The Gurense people and their Gurenne language are usually called “Frafra” in Ghana, from the Gurenne all-purpose greeting farefare. I think the principle should be widely adopted, so that (for example) the indigenous language of London might henceforth be called Wotcha.

  57. Jen in Edinburgh says

    The horse with no name

    They might have had to go through the desert to get there, I suppose.

  58. I just got around to reading a post of Lameen’s from almost a year ago, which quotes a recently digitized unpublished Timbuktu manuscript, on the pious speech of animals.

  59. If you wonder what tone goes with E, Wiktionary helpfully explains that in Mandarin 誒 is either ê̄, ế, ê̌ or , interjections expressing: calling attention; surprise; disapproval; or agreement or promise, respectively.

  60. David Eddyshaw says

    a post of Lameen’s

    Yes, I remember that one. What Lameen’s blog lacks (sadly) in quantity, it makes up for abundantly in quality.

    It immediately made me think of Balaam’s Ass (a story of which I am particularly fond) but turns out to be about something quite different. It seems a bit akin to the need felt for apotropaic expressions after someone sneezes (in case they have inadvertently cursed someone, or the like) that you find in a number of cultures. Fear of non-verbal speech …

    In Kusaal you say Win yɛl sida “God speaks truth” when someone sneezes, which was explained to me as meaning that if you sneeze, it means someone elsewhere is praising you; I did wonder whether this was a latter-day sanitising of an older, darker, belief …

  61. Yes, I know that (paard/Pferd):)

    And yes, i suspected that it is Amazonia and somethign rather famous. I wondered why they say “langauge” ( exogamy is not unusual), but from what you said I conclude that “language” is an important concept for local people.

    I suspect that by “matches”, Blench just means that there are lots of different ones in both categories

    This argument works perfectly one way. IF horse was introduced independently to many places under different names, then you must expect a greater variety of names compared to what you would expect otherwise.

    I suspect that by “matches”, Blench just means that there are lots of different ones in both categories,

    Yes, this.

  62. For me personally “feral stage” was a new idea. Usually I imagine ancient people moving from one static ecosystem to another static ecosystem and exchanging domesticated species.

    But I have to deal myself with something else: local climate and ecology are changing quickly where I live, and I see numerous insects and animals which I need a name for. A major disappointment was some 10 years ago: flying green (and beautyful) things began to fill my room every summer. Finally I googled them and learned the name. And when what used to be just a “flying green thing” was assigned a name, it immediately ceased to be mysterios, as if assigning a name to a phenomenon is a way of disposing of it. I’ll never do that again:/

  63. Per contra there is a strong tabu against mixing vocabulary, which people are aware of.

    Kupwar: three languages, a single syntax, no loanwords.

  64. I wonder if anyone here knows, when did scientists become aware that so-called wild turkeys in North America are long-feral descendants of domesticated birds.

  65. David Eddyshaw says

    TIL that “feral” (as a term of art, anyhow) doesn’t mean the same as “wild.” Thanks, Y!

    (Once again my thinking in Latin has betrayed me into misunderstanding of the Modern Barbarian languages. “Feral” doesn’t mean the same as feralis either. Why can’t you young people learn to talk proper, more maiorum, eh?)

  66. Feral Latin.

  67. David Eddyshaw says

    (i.e. French.)

  68. French is wild Latin. Israeli Hebrew is feral Hebrew.

  69. I wonder if anyone here knows, when did scientists become aware that so-called wild turkeys in North America are long-feral descendants of domesticated birds.


    WikiP:Wild_turkey, at least, states otherwise. Meleagris gallopavo silvestris is native to eastern US/Canada. Other turkey subspecies are also native to various regions in the Americas. M. g. domesticus derives from M. g. gallopavo of south Mexico.

    If that’s actually the point, well, I’m not parsing the subtle points of the conversation. Maybe it all flew over my head, much like a majestic wild turkey.

    Somewhat new to me: the striking iridescent plumage of the ocellated turkey.

  70. What I recall hearing is that actual wild turkeys were domesticated in Central America, and either became feral there and spread north on their own; or were brought north by people, and then became feral and continued to spread on their own.

  71. David Marjanović says

    The distribution of wild turkeys is as large as it could be, and the number of subspecies endemic to the US listed in the article is large enough to suggest that turkeys were there long before any were domesticated.

    And when what used to be just a “flying green thing” was assigned a name, it immediately ceased to be mysterios, as if assigning a name to a phenomenon is a way of disposing of it. I’ll never do that again:/

    A rose by any other name would still…

  72. Open Access, and recent:

    Padilla-Jacobo, G., Cano-Camacho, H., López-Zavala, R. et al. Evolutionary history of Mexican domesticated and wild Meleagris gallopavo. Genet Sel Evol 50, 19 (2018).

    Estimates of divergence times agree with range expansion and diversification events of the relict population of M. gallopavo in northwestern Mexico during the Pliocene–Pleistocene and Pleistocene–Holocene boundaries. Demographic reconstruction showed that an expansion of the population occurred 110,000 to 130,000 years ago (Kya), followed by a stable period 100 Kya and finally a decline ~ 10 Kya (Pleistocene–Holocene boundary).


    During the Pleistocene, a large and stable population of M. gallopavo covered a wide geographic distribution from the north to the center of America (USA and Mexico). The mexicana, merriami, and mexicana/intermedia/silvestris/osceola genetic groups originated after divergence and range expansion from northwestern Mexico during the Pliocene–Pleistocene and Pleistocene–Holocene boundaries. Old and new maternal lines of the mexicana/intermedia/silvestris/osceola genetic group were distributed within the Trans-Mexican Volcanic Belt where individuals were captured for domestication.

    Are we talkin’ turkey over here? Yeah, I think we are.

  73. obviously the Europeans must also have had a tabu on naming the horse

    Trond Engen had a relevant comment in the name-of-the-bear thread:

    There’s more than taboo that could cause frequent replacement — a reason to be careful about invoking taboo. In the case of e.g. the badger or the fox, it might just as well be that they lived close to humans, giving rise to a more nuanced vocabulary of both eu- and dysphemisms, cutesie nicknames and technical terminology. Also words for domestic animals tend to be replaced by specialized terms or by words for offspring.

    That would apply to English dogs and pigs, as well as European horses.

    When I was young I used to wonder why words for “horse” were so different in European languages, while cats were always something close to “cat”; I guessed that horses were named for the different work they did, while cats don’t do any work!

  74. Per the same WP article, wild turkeys were nearly extirpated from Canada and the U.S. except for some protected patches: efforts in the 1940s to spread them over their historic range and even beyond (there had been no Meleagris species in California for 10,000 years) were very successful, and the population has risen from a bottleneck of 30,000 birds to 1.3 million. Apparently they do well at the edge of farmlands. But there can be no guarantee in such a situation that there was no gene flow from domestic birds.

  75. Well, I started to say something further about the derivation of Chinguetti. Two days, three scholarly articles later, after poring over a relevant thread or two here, I’m settling in to finally finish the catalog to Northwestern University’s Caravans of Gold, Fragments in Time exhibit on the trans-Saharan trade.

    I’m starting to realize how much I don’t know. Which is good.

    Still, I do want to ask those who might know more, is anyone skeptical of Taine-Cheik’s proposal? Part of me still thinks Chinguetti is too far north, too arid (or much too late) for a Soninke-derived name.

    Anyway, the big win is reading Caravans of Gold. Should have dug in further when I got it last year. Grappling with the subject reminds me of arriving at college and realizing I had no idea what the French Revolution was.

  76. Lars Mathiesen says

    Just to note that Icelandic still has jór, albeit in poetic language. And Spanish has yegua, so maybe the taboo wasn’t that strong.

  77. With horses, it’s certainly not a taboo. As someone up-thread mentioned, in names for farm animals replacement of general terms by more specialist terms, by terms for young animals, or by more emotionally loaded terms (affectionate or pejorative) is frequent. Often both the old term and the new term continue to be used side by side, with the old term restricted to literary / poetic language, like your Icelandic example or English “hound” vs. “dog”.

  78. “hound” vs. “dog”

    Or fowl vs bird/brid.

  79. David Eddyshaw says

    I must find another word for “obviously.” Ideally, one that doesn’t have the confusing alternative meaning “obviously.” I have only myself to blame … it’s not as if it’s my first warning. “Obviously not” might do the trick …

    ktschwartz and Hans are clearly [NOT “obviously” ] right about the horses. <— LOOK, MA, NO IRONY!

  80. January First-of-May says

    Often both the old term and the new term continue to be used side by side, with the old term restricted to literary / poetic language

    This, as I understand it, is basically the case (though not quite complete) with Russian конь vs. лошадь (and for that matter пёс vs.собака).

  81. David Eddyshaw says

    Talking of horses, I seem to remember reading somewhere a spirited attempt to make out that the Hebrew sus is a loan from Somewhere in Indoeuropean. Has anyone with any linguistic street cred ever maintained such a thing? It doesn’t look very plausible (but stranger things have happened, I guess …)

  82. Lars Mathiesen says

    Because we can obviously never tell when DE is applying irony.

    Q: How can you tell that a Welshman is typing something ironical?

    A: His fingers move.

    (Apologies to M. Headroom).

  83. Hebrew sūs: I happen to have the chapter for this from the Theological Dictionary of the Old Testament, which is currently the most exhaustive giode to the literature on those selected words it happens to cover in its fifteen volumes.
    So: it widely occurs in the Semitic languages, though they are not clear about what’s inheritred and what’s borrowed. It says, “Akkadian sisû or sisā’ū is to be viewed as a loanword of unknown origin,” but also “The etymology of sûs is disputed. Derivation from Sanskrit aśva(s) was widely supported [refs. Albright, Mitannian maryannu, “chariot-warrior”, and the Canaanite and Egyptian Equivalents, Archiv für Orientforschung 6(1930/31, 217-21; Salonen, Hippologica Accadica, p. 21; A. F. Rainey, Tell el-Amarna Tablets, p.77]. G. R. Driver refers to Hittite aššušani, ‘rider’, which is of Hurrian origin. He points out, however, that the loss of the initial a- creates difficulties, and thus suggests understanding sūs from the perspective of a repetitive susu or sisi, either an onomatopoetic term or Lallwort [Really??].”

    Later they say that “Finally, any derivation of rhe Semitic word from ‘horse’ from Indo-Aryan aśva(s) has proven to be extremely doutful.” But don’t say why.

    Albright’s paper goes into some detail into the Sanskrit etymology, particularly on the Semitic side, but a lot has happened since 1930.

  84. I’d like to say I knew “tabu on the horse” was ironic, but I can’t swear to what my state of mind was. If I understand correctly, we’re both saying that if almost all European languages swapped out their ancestral horse word for something else, then so could the Oti-Volta languages—though I have no idea what time frame we’re talking about. If we didn’t have written records of Latin, then it would be mysterious how German got Pferd (and Kopf for that matter), wouldn’t it?

  85. David Eddyshaw says

    Oti-Volta is a good bit less diverse internally than Indoeuropean, though considerably more diverse than any individual branch thereof; at any rate, I think it’s clear that the timescale for loss of any presumed common Proto-Oti-Volta “horse” word would be much less than that for Indoeuropean.

    But the situation is different with regard to the actual animal, too; whereas all Indoeuropean speakers would presumably have known horses as familiar everyday things from the beginning, this is not the case with speakers of Oti-Volta languages (whatever one makes of Blench’s speculations about wild horses); as I say, to this day in the Mossi-Dagomba states, horses are very definitely associated with chieftainship (with the chiefly clans claiming alien origin); and horses (unlike donkeys) are very much not a feature of everyday life. (If you see a horse tethered in a compound, it’s a chief’s residence.) So while the “familiarity breeds diversity” explanation works fine for Indoeuropean horses, it doesn’t fit Oti-Volta well at all.

    In a way, the diversity of words for “horse” in Oti-Volta is not really surprising or problematic in itself; the only words for animals reconstructable for Proto-Oti-Volta are (unsurprisingly) those for the most familiar animals of all: cow, goat, dog, donkey, snake, monkey. Not even “sheep” is universal, though the pattern of distribution there suggests replacement of a common Oti-Volta etymon in the Atakora region of Benin as an areal phenomenon. So it’s pretty much what you’d expect given that horses (pace Blench) aren’t actually common animals there.

  86. David Eddyshaw says

    On animal-name tabus, I’ve just noticed that Wayne Suttles’ paper on Sasquatches (located by Stephen Goranson) speculates (on p9) about whether an original Salish name for the presumed beast has been ousted by euphemisms, as the original words for “wolf” and “grizzly bear” seem to have been:


    Unfortunately I have been unable to discover the Kusaal for “sasquatch.” Further research is needed …

  87. I have wondered whether sasquatch, Halkomelem sásq’ets (morphological breakdown here), literally something like “split-on-the-back”, makes reference to the reportedly enormous trapezius muscles (as mentioned here and in many other descriptions of the creature as well-muscled) and especially the very deep natal cleft (as here and here). I have been dilatory in contacting native speakers of Halkomelem in order to learn their opinion of this analysis—I don’t want to seem presumptuous in seeking the knowledge of the community on just this one point, for a linguistic analysis that might be construed as disrespectful or even obscene.

  88. whereas all Indoeuropean speakers would presumably have known horses as familiar everyday things from the beginning, this is not the case with speakers of Oti-Volta languages — oh, that’s the point I was missing. Thanks.

  89. Trond Engen says

    tabu on the horse

    A curse on both your horses.

  90. David Eddyshaw says

    Horses for curses.

  91. Ye’ve both made me laugh, curse ye!

  92. (No, it was not a horse laugh. I am far too sapient for that sort of thing.)

  93. per incuriam says

    Icelandic still has jór, albeit in poetic language. And Spanish has yegua

    And Gaelic still has each as the generic term.

    A curse on both your horses

    Compare Gaelic an t-each (the horse) v Irish an teach (the house).

  94. David Eddyshaw says

    Welsh has ebol “foal”, though a grown-up horse is ceffyl (or march, which Victor Mair is firmly convinced is related to the Chinese 馬. It would have been borrowed from my cousin Evans the Tocharian.)

  95. David Marjanović says

    Horse Shakespeare & I


  96. January First-of-May says

    or march, which Victor Mair is firmly convinced is related to the Chinese 馬

    I mean, it could be; it’s one of the less ridiculous of Mair’s hypotheses. Very similar roots show up in Mongolic (whence Russian мерин “gelding”) and in a bunch of other Asian families (an extensive list is here).

  97. David Eddyshaw says

    it’s one of the less ridiculous of Mair’s hypotheses


  98. David Eddyshaw says

    AFAIK (admittedly not very far) the march etymon turns up in Celtic and Germanic within Indoeuropean, and that’s it. While it’s perfectly possible that some unattested Central Asian twig of the Tarim-Liffey family passed this form (lost in all other branches) on to Mongol and Chinese, I maintain a certain scepticism.

    Mair has supported the view that there is specifically a Germanic component to PIE influence on Sinitic, so anything is possible … anything at all, in fact. From the proposition “zero equals one” you can deduce …

    On the other hand, the Mongolian connection to 馬 looks quite plausible, though a lot of the forms listed at the Proto-Sino-Tibetan link seem most likely to be borrowed from Chinese itself.

    The very variety of forms of words for “horse” that we’ve been talking about makes it all too easy to find lookalikes. I mean, the Waama suka is obviously derived from Hebrew sus

  99. @David Eddyshaw: Actually, “zero equals one” isn’t enough to derive anything except that zero is equal to everything. (It does not, for example, allow anyone to derive Mrs. Lenhart’s phone number.) That is, proving 0 = 1 just shows that a system is actually trivial (not inconsistent); in fact, it’s a very useful way to prove that there is no consistent nontrivial system with certain properties. If the goal is to prove that there is, say, no ring that satisfies some set of conditions {X}, the easiest way to do that is usually to use the ring axioms and the additional relations {X} to prove 0 = 1, because that means that the only ring with those properties is the zero ring.

  100. Lars Mathiesen says

    You want P and not P innit? Not vastly more useful since it’s now true that any random phone number is that of Mrs. Lenhart — unless of course she answers on all of them, but how can I call my Mum then? (It’s also true that any given phone number is not hers. This is why we can’t have nice things).

    What’s that, I just proved that Mrs. Lenhart is my Mum? NANANANA, I can’t HEAR you!

  101. David Eddyshaw says

    It does not, for example, allow anyone to derive Mrs. Lenhart’s phone number.

    If zero equals one, Mrs Lenhart’s phone number is (of course) zero. And one, of course. And the same as Lars’ mother’s.

  102. Miss Lenhart’s phone number is zero, because she is in Timbuktu. Please call her horse instead…

  103. David Eddyshaw says

    Miss Lenhart’s phone number is zero, because she is in Timbuktu.

    Makes sense.

  104. isn’t enough to derive
    when we have the axioms of ring and nothing else (not even a wristwatch). Most people have natural numbers though.

  105. Lars Mathiesen says

    Well, if 0 = 1 (as sets) then those sets are not the base of an inductive set and its successor, so having the natural numbers in the same universe is just confusing.

    (In the construction of the natural numbers used in Zermelo–Fraenkel arithmetic, 0 = {} and 1 = {{}}, so I guess you could say that 0 = 1 proves that your model does not fulfil those axioms. But I’m not sure 0 = 1 even means anything if you’re trying to do arithmetic, as opposed to rings).

    TIL that Fraenkel’s family name was just that, and he published as Adolf/Adolph Fraenkel at least up through the 20s, later as Abraham F — and who can blame him? (He was a German Jew and had both names from birth). ObHat: how would Fraenkel be pronounced in (early 20c) German, just like *Fränkel?

  106. Probably, but I like the idea of Fra Enkel.

  107. We had an entire thread on various Frenkels


  108. David Marjanović says

    how would Fraenkel be pronounced in (early 20c) German, just like *Fränkel?

    “fraenkel”: 4,900,000 ghits
    “frenkel”: 2,410,000 ghits
    “fränkel”: 699,000 ghits

    All pronounced the same.

  109. Lars Mathiesen says

    So Edouard Fraenkel was considered a namensvetter of Hermann Fränkel, that seems to indicate a yes.

    @DM, thanks. I know how to pronounce Frenkel and Fränkel, but I wasn’t sure if there was some other convention behind the Fraenkel spelling.

  110. David Marjanović says

    No, just elegant variation, like th, pff and gkh.

  111. There is a convention in German to write umlaut letters as ae, oe, ue in cases when the umlaut letters are not available (e.g. when using foreign language keyboards); this convention is also applied in crosswords. And the two dots historically go back to a superscript “e”. So Fraenkel is not unusual, and actually represents the historically oldest way of writing “ä” (distinct from writing it “e”, as they did in Old High German).
    This sometimes clashes with another tradition, from North-Western Germany, to use “e” and “i” after back vowels to express length, leading to place names like Straelen or Soest wrongly being pronounced with “ä” and “ö”, respectively.

  112. As mentioned by Stu (then still Grumbly) a decade ago; I remember being quite surprised.

  113. And again in 2019, with subsequent discussion.

  114. In case anyone didn’t get my joke about Mrs. Lenhart: xkcd “Principle of Explosion.”

  115. Thanks, I in fact missed the reference.

  116. There was an anecdote about a teacher, who was working with Aboriginal children. With a translator: she didn’t know the langauge. She wrote 2 + 1 = 3 and asked the guy to translate. He objected, saying it is going to be nonsense. She insited, he refused, she insisted, he finally translated, kids were perplexed.

    (I guess, no one here needs an explanation)

  117. David Eddyshaw says
  118. David Marjanović says

    distinct from writing it “e”, as they did in Old High German

    There were two distinct ones, one (short only) written e and probably pronounced [e], one (long and short) written a (or â by Notker) and probably pronounced [æ]. They’re still distinct in many Upper German dialects; in mine they’re largely /e/ and /a/, both distinct from /ɛ/ (which is not an umlaut product) and from /ɒ/ (the outcome of a without umlaut).

  119. David Eddyshaw, thanks! The anecdote is from an article referenced on this page. Page 33, right column.

    I read it million years ago. It was in a collection of scans on a server in a university in Papua New Guinea.

    The first paper I opened there was so exciting (it began with a several pages long list of typos on p. 16 which began with: “p 16, line 5”. But it was the list of typos that now populated p.16., the actual article with a typo now three pages down!) that I thought, it is dumb not to read it in its entirety. It didn’t disappoint me and I read more.

    This paper was one of them but the site now is inaccessible, and even figuring out what was the title was a problem. But I just recognized it somehow.

    It is dedicated to debunking “one – two – many” number system story (commonplace in textbooks)

  120. David Marjanović says

    or â by Notker

    …that’s specifically the long one.

  121. My understanding was that, except by writing “e” (i.e. the same as /e/ from Proto-Germanic /*e/ and /*i/, OHG didn’t distinguish umlauted from non-umlauted “a”, and what Notker did was marking length, not umlaut. Or is my recollection wrong here?

  122. David Marjanović says

    Notker indeed marked length. Umlaut of *a gave two different things under different conditions: [e] e by default, [æ] a in the presence of “blocking factors” of which length was one; that’s why I wrote “a or â“.

  123. Just now going back to the article that prompted this post and realizing its odd polemics.

    It says the libraries of Timbuktu are reticent about bringing out their manuscripts today because of supposed book-burning by the French 70 years ago, rather than the burning of manuscripts by Islamic extremists in the last decade, which it doesn’t mention.

    Were there any colonial attacks on these libraries that led to destruction of manuscripts (as opposed to plundering them)? The book Bad-Ass Librarians of Timbuktu, focused on recent troubles, didn’t mention anything like that. I’m well aware of the depravities of colonialism, but that particular depravity doesn’t fit my template.

  124. I’m also curious about what fact or factoid might have been behind this paragraph:

    >Most of the manuscripts themselves are of special importance to the owners for some reason. For instance, many who hitherto claimed royalty have been discovered to be from the servile class on account of evidence from the manuscripts.

    It just seems like a very odd claim.

    And what happened in 1985?:

    >Finally, not until 1985 that much life has been breathed into the intellectual life of this region

    I see that there was a border war between Mali and Burkina Faso, but a quick glance doesn’t suggest it had a major cultural, religious or intellectual impact of it. Was there a culture shift across the Sahel in 1985?

  125. Apparently in 1985, the parliament of Mali adopted Loi n°85-40/AN-RM relative à la protection et à la promotion du patrimoine culturel national

  126. John Cowan says

    Oti-Volta is a good bit less diverse internally than Indoeuropean, though considerably more diverse than any individual branch thereof

    That seems pretty remarkable: is it really more diverse than Indic? In any case, looking at the WP article on O-V languages lead me to the one on Kusaal. Is the stuff on the talk page addressed to you?

  127. Since turkeys came up in this thread last year, I thought I would drop this here:


  128. “the Aztec name for turkey, guajolote”

    That is, of course, huehxōlōtl, saith Wiktionary.

  129. David Eddyshaw says

    I’ve often (well, once or twice) wondered where the Hausa talotalo “turkey” originated. (The WOV languages have borrowed the Hausa word, as has Yoruba, if Wiktionary is to be believed.) It doesn’t seem to be from Portuguese, which is usually a good guess for words of this sort.

    I suppose that if porco “pig” can end up as Kusaal kukur (which it did), it’s not totally inconceivable that something like guajolote might end up transmuted into talotalo. Foreign words are hard to get your tongue around …

  130. David Eddyshaw says


    (Replying to you on Jan 28; better late than never, and this is LH, so)

    Yeah, I think Oti-Volta is more diverse than Indic (it’s much more diverse than Germanic, for example, which is more or less on a level with just Western Oti-Volta in terms of internal diversity, and WOV itself is a much more cohesive grouping than Eastern Oti-Volta or Gurma. The most divergent branches of Oti-Volta show only about 50-60% matches on the Swadesh 100 list FWIW; and although reconstructing noun morphology for the protolanguage is fairly straightforward, nobody so far has made much of the verbal system. (Manessy himself wimped out and decided that the protolanguage didn’t really have conjugations at all, and the branches have all independently innovated some, but I think progress can yet be made on it.)

    The stuff on the Kusaal talk page is actually (mostly) by me, and addressed to Tony Naden. I really ought to clean up the Kusaal WP page itself at some point. (I’ve always though that if anybody is really interested in more information they can follow the links to Urs Niggli’s grammar and to mine*, though I’ve just noticed that they could do with some freshening up too.)

    * I previously purged the link to academia.edu on ideological grounds, in favour of the one to the good and deserving Zenodo, though it’s still on academia.edu for those who don’t mind that sort of thing …

  131. Could talotalo be a locally-made onomatopoeia?

  132. David Eddyshaw says

    That’s what the dictionaries say. I never heard a turkey go “talotalo” myself, but perhaps they do in Africa.

    I suppose that both could be true: a less bizarre form more obviously related to guajolote might have been reworked for the onomatopoeia.

    Hausa talotalo has to be either a loanword or onomatopoeic on phonological grounds alone (obviously the bird itself is an import, but in principle the name for it might have been created by repurposing some inherited word. But it wasn’t.)

  133. @David Eddyshaw, what are your ideological objections to academia.edu?

  134. David Eddyshaw says

    I share Peter LeFanu Lumsdaine’s opinion on the matter, though in a rather more lackadaisical fashion:


    It’s not that there isn’t some very good stuff there (there are some really excellent grammars of the Agolle dialect of Kusaal, for example …) But that is not quite the point, I guess.

    I put my own work on academia.edu in the first instance really just because it was so easy to do, and I tend to use that version as a sort of scratchpad, keeping the Zenodo version more stable. I don’t feel strongly enough about it to pull the academia.edu version altogether, and quite a lot of Ghanaians have found it there over the years. I don’t know how much actual use it is to them (not much, I suspect), but I like to imagine that they appreciate that it exists …

  135. Yes, I read Peter’s opinion. But I see academia.edu as “imperfect”. Ideologically I am friendly to them (Peter is clearly hostile). That’s why I was curious.

  136. Let’s say, if Elsevier, Springer and co get bankrupt tomorrow, the world becomes a better place. If I could make them bankrupt I would have done that:/ I do not feel the same about academia.

  137. David Eddyshaw says

    If I could make them bankrupt I would have done that

    Let’s not abandon hope!

  138. It’s easy to imagine a much better academic social network than academia.edu — better metadata, better transparency, better integration with other services and networks (trivial, since the current level is just about none); but as far as freemium commercial enterprise versions go, it seems to be about as good as can be reasonably expected. And as always there’s a principal agent problem in who should be working on doing better? (Technically, e.g. I am already, but only in a very elementary sense, on some resources that an academic online platform probably should have.)

    Some much better commercial social media / database websites exist (e.g. I’ve been a happy regular on RateYourMusic.com for some 15 years now), but generally these seem to have additional income sources available and better crowdsourcing, unlike the academia.edu model.

  139. Y:

    Could talotalo be a locally-made onomatopoeia?

    David Eddyshaw:

    That’s what the dictionaries say. I never heard a turkey go “talotalo” myself, but perhaps they do in Africa.

    I’ve never heard a turkey go “gobble”, but that is how its call is commonly rendered in English.

    I suspect some onomatopoeia is just scrabbled up from whatever one has in one’s inventory of consonants and vowels.

    Βρεκεκεκέξ κοάξ κοάξ

  140. To Ryan above, belatedly but since this rather overlaps with my interests:

    There are attested cases of French colonial destruction of manuscripts, by way of collective punishment against their owners. In Algeria, many of the manuscripts of Chellata (near Bejaia) were burned by French troops in 1957 as punishment for their owners’ suspected help to the independence fighters, and the library of the University of Algiers itself was burned down in 1962 by the OAS in an effort to deny the nascent Algerian state the resources it contained (destroying books as well as manuscripts). I wouldn’t be surprised if Malian history attests similar cases.

    As for the claim about manuscripts revealing the humble origins of parvenus, I don’t know the details but… People in the Sahara love to use genealogy to build social status and strengthen alliances, a process which not infrequently involves doctoring the earlier stages of a genealogy to establish ties to higher-status groups. In such a context, it would be natural to see the discovery of older genealogical documents as potentially status-threatening.

  141. David Marjanović says

    Βρεκεκεκέξ κοάξ κοάξ

    Oh, there are frogs in Europe that sound like that. The basic treefrog (Hyla hyla) comes to mind.

    Conversely, ribbit is a phenomenon of the US west coast if I’m not confusing things.

  142. In Russian translation of Andersen’s Tommelise it was
    — Коакс, коакс, брекке-ке-кекс!

  143. David Eddyshaw says

    The late* Stefan Elders’ truly superb grammar of Kulango shows its excellence not least in having no less** than 41 pages on ideophones. The section on onomatopoeic words contains some which occasion no great perplexity, like

    ʔiʔã̀ã̀ “cri de l’âne”
    ʔwà “bruit de vomissement”
    ʔwɛ̃ɛ̃ ʔwɛ̃ɛ̃ “les pleurs d’enfant”
    pr̀r “bruit d’un moto”

    and (apropos)

    ʔyããʔyã̀ã̀ʔyãã “bruit de beaucoup de petites grenouilles”

    but apparently Kulango dogs go ʔyɛ̀ɛ.

    * He died, aged only 42, of a short but severe tropical illness, while doing fieldwork among the Dogon in Mali. While this is obviously extremely sad, I can’t help but feel that that is the way for a great descriptive linguist of African languages to go …

    ** Or fewer, either.

  144. I guess it has some 41 pages on ideophones.

  145. David Eddyshaw says

    Indeed. Indeed.

  146. I realized that a Russian prefix pere- is onomatopoeic.

  147. I dislike academia.edu, but not intensely. Separately, one should probably separate consideration of Springer as a journal publisher from Springer as a book publisher; for textbooks in math, there is just no one that can compare to Springer, frankly.

  148. Stu Clayton says

    but apparently Kulango dogs go ʔyɛ̀ɛ

    Basenji Yodel

  149. Lars Mathiesen says

    In H.C.Andersen’s original of Tommelise it is Koaks, koaks! Brekkekekeks!. Very recognizable if you’ve had it read to you as a child, but I didn’t know it was from Aristophanes. (Possibly through intermediate sources, I don’t think Andersen had much Greek).

  150. David Eddyshaw says

    Basenji Yodel

    I did actually seriously wonder about that, for a moment. I don’t think Basenji dogs have caught on in those parts, though. Maybe Kulango dogs just wish they were Basenji.

    “Basenji” is the only Lingala loanword I can currently think of in English.

  151. @DE: the OED only adds likuta/makuta, likembe (though not makembe) and, perhaps via Dutch, gobbe < ngúba

  152. The etymology of Basenji given by the OED is quite interesting:

    Probably < Lingala basɛ́nzi (formerly written basenji ; singular mosɛ́nzi ) wild ones, indigenous people or animals < ba- plural class prefix (for animate beings) + -sɛ́nzi indigenous, wild (probably < Swahili -shenzi uncivilized, probably ultimately related to the words discussed at Zinjanthropus n.)

    Here is Zinjanthropus:

    < scientific Latin Zinjanthropus, former genus name (L. S. B. Leakey 1959: see quot. 1959) < Zinj (variant of Zanj (Arabic Zanj , Persian Zang), a medieval name used in Arabic and Persian sources for an area in East Africa, including the Horn of Africa and also parts of modern-day Kenya and Tanzania (compare the first part of the name of Zanzibar : see Zanzibari n.)) + ancient Greek ἄνθρωπος man (see anthropo- comb. form), on account of the first specimen having been found in Tanzania.
    With the place name Zanj, Zinj compare Arabic zanj, zinj (noun) dark-skinned inhabitants of Africa collectively, Persian zangī (adjective) of or relating to the dark-skinned inhabitants of Africa (Pahlavi zangīg), Hellenistic Greek Ζίγγις, the name of an unidentified landmark on the East African coast (2nd cent. a.d. in Ptolemy), all probably ultimately independently < an otherwise unrecorded name in an East African language.

    As I understand the situation from this, the Swahili -shenzi element is cognate with the ancient source of Greek Ζίγγις, Pahlavi zangīg, etc., then?

    In various sources, as here (p. 106), there is a different account, taking -shenzi from Arabic زنج Zanj. And then diffused from the east coast to central Africa? But how would the phonology work for the Swahili then, getting -sh- out of Arabic z-?

  153. David Eddyshaw says

    Nurse and Hinnebusch’s Swahili and Sabaki lists the word (in the section “Doubtful Proto-Sabaki Lexis”, p664) as from *musenzi “barbarian”, but just says “Omani/Zanzibar msenzi“; “this must be a recent loan from Omani/Zanzibar Arabic”; “s : š correspondence unclear.” It doesn’t relate it to the Zanj “Black Africans” word. I don’t know anything about Omani Arabic, but it’s hard to see how msenzi could be connected with Zanj, unless Omani has undergone some very odd sound changes. On the other hand, each word has its own history …

    Diffusion to central Africa would be unsurprising, however. Swahili has got that far; and there are undoubted Swahili loans in Lingala.

    “Zanzibar” is yet another “Land of the Blacks” etymologically, like “Sudan” and “Guinea.”

  154. David Eddyshaw says



    is quite informative on Omani Arabic. Interestingly, it seems to share the Egyptian [g] for Classical /ɟ/ thing, which presumably would make the cognate of Zanj into Zang or Zeng.
    Some dialects have [d͡ʒ] though, and it seems to be undisputed that the first element of Zanzibar is the Zanj word.

    There’s no suggestion that Omani does anything peculiar with /z/ though, like devoicing it.

  155. And then why -j- in basenji?

  156. David Eddyshaw says

    Lingala doesn’t have a z/j contrast. Guthrie (in 1939) writes j throughout, but says

    Le son qui s’écrit j varie beaucoup. Il est souvent comme “dj” dans “djinn”, mais chez quelques tribus il se prononce plutôt comme dans “jeu”, tandis que parfois il devient un simple z comme dans “zéro.” La prononciation ordinaire est “dj.”

    It’s written “z” in the modern orthography, though.

  157. To uphold the ancient tradition of using opaque orthography to fool the uninitiated.

  158. David Eddyshaw says

    There may have been an actual diachronic change.* Michael Meeuwis’ Lingala, from 1998, makes no mention of any other realisation of “z” than “voiced alveolar fricative.”

    Mind you, Meeuwis’ work is an irritating and incomplete effort, of the sort where the author spends a surprising amount of time dissing his predecessors instead of getting on with the job.

    (I see there’s a much-extended new version. Not sure if my interest extends to stumping up 158 euros, though …)

    * Sixty years is a long time in the history of a language less than 200 years old …

  159. I just wonder what are the contrasts for African speakers, that Arabic j appears as z in transcriptions from Swahili, but then as z/j in possibly the same (or not) word in Lingala, while Arabic z gets devoiced.

  160. David Eddyshaw says

    Assuming that Nurse and Hinnebusch have given the Omani Arabic form correctly, the odd sound correspondences (if they really are correspondences, as opposed to Zanj and msenzi just happening to look similar) are already there within Arabic itself, rather than having arisen in the process of borrowing into Swahili or onward into Lingala.

    Swahili in fact does have both /j/ and /z/ (in native vocabulary, too.)
    I’ve never actually thought about the phonology of Arabic loans in Swahili; I can certainly think of cases where Swahili has j for classical /ɟ/ (e.g. jaribu “try”, jinsi “kind, sort.”) On first principles I would guess that there are at least two strata to them, viz Omani, as spoken by actual live immigrants, and Classical, available to all pious Muslims (like the Swahili.)

    I would think that what the OED is postulating is that the -senz- and -zanɟ- are related not so much in the sense that the former is directly borrowed from the latter but that both ultimately go back to a place name in some unknown African language, whence it got borrowed into Greek and Persian and thence Arabic (and thence returned to its homeland.)

    It seems to me quite possible that the resemblance is just coincidence. On the other hand, *√snz looks pretty iffy for an Arabic root unless it’s of foreign origin (but somebody who knows a lot more Arabic than me would really need to pronounce on that.)

  161. David Eddyshaw says

    N & H conflate Omani and Zanzibari Arabic here; but I wonder if this msenzi word is actually found in Oman, or only in Zanzibar?

  162. Assuming that Nurse and Hinnebusch have given the Omani Arabic form correctly

    I thought it is a typo or something…. “msenzi” does not sound very Arabic 🙁

    The also have on p. 295. mušenzi ‘barbarian’, (Omani Ar/Zanzibar msenzi), °mušendzi (Gi), mšend̪i (Am), mšenzi (Mw,Ung), mšendzi ‘slave’ (Ng), this must be a recent loan from Omani/Zanzibar Arabic, note that Am has /nd̪/ (as also in /kand̪u/ for /kanzu/)

  163. Could it be somehow a Greek or Greek-via-Turkish word, like efendi?

  164. David Eddyshaw says

    I was actually wondering if it could be a loan into Zanzibari Arabic from Swahili (or a recent ancestor thereof), a hypothesis which pretty much founders if the word is found in Oman too.

    Mind you, if it’s Swahili-ish, it doesn’t seem to have a plausible Bantu etymology, so that wouldn’t help much. Unless it was from the indigenous descendant of whatever-it-was that the Greeks and Persians heard as Zing-/Zang-

    I agree msenzi doesn’t look particularly (classical) Arabic, but initial ms- for mus- is perfectly OK for Omani, as the paper I linked to demonstrates.

    What we need now is for Lameen to tell us that msenzi is a perfectly cromulent post-classical Arabic word for “barbarian” (ideally, one derived from Berber, to confuse the issue further …)

    (The Kusaal word for “Hausa person” is Zangbɛog. I just thought I’d throw that in to muddy the waters as much as possible.)

  165. Msunobari, n. (mi-), pine-tree, fir-tree, deal,—timber imported in quantities to Z. chiefly from Norway. It is rapidly destroyed by white ants. (Ar. and Hind.)

    Norwegian Wooood!!!! Sadly, Lennon did not know about the White Ants.

  166. everything above was written by me based on assumption that UNLESS it is African, then it is Swahili rendering of Perso-Arabic zVng/j (with various Swahili prefixes including m-).

  167. David Eddyshaw says

    (And a Muslim trading centre in a West African town is a “Zongo”, from Hausa zango “camping ground, lodging place.” Sadly, I cannot think of any remotely plausible way for this to be the origin of Zanj-/Zang-. Coincidence is a bitch …)

    “Zanj” would be a great name for a jazz quartet.

  168. According to Hunwick (1970:102-8) the zanji, zanjiyya, zanaji, zanajiyya etc.
    were servile castes in the Mali and Songhay empires during the period 1400-1700. They included fishermen, boat keepers and their crews, domestic servants, bodyguards and blacksmiths along the Niger river in the sub-saharan region. See also Wansbrough 1970:97-99.

    (from) Oh, I did not know.

  169. David Marjanović says

    White Ants

    Termites. (Not ants, but cockroaches.)

  170. …Shirazi, a generic term used even today, especially in Zanzibar where a large majority of the people call themselves Washirazi.

  171. David Eddyshaw says

    My favourite word of Persian origin in Swahili is that word beloved of colonialist-era ripping yarns, bwana (N&H p317; Wiktionary’s account of this word looks like a folk etymology, perhaps from the same enthusiast who informed us that maji “water” was also an Arabic loanword. The interesting paper that drasvi linked mentions him.)

  172. Is that true, though? The Encyclopedia Iranica says,

    The paucity of lexical material at this stage as well as the total absence of nonlexical material suggest very light Persian influence. The few items of general reference at this stage include *(m)pula “steel,” *bwana “man, gentleman,” *(m)pamba “cotton,” and *(n)kasa “turtle,” which correlate with Middle Persian pōlāwad, bān, pambag, and kašawag.

    By about 1700, when the first recorded documents in Swahili appeared, Swahili vocabulary was much as it is today, containing several hundred items from Persian (Krumm; Knappert) and many more from Arabic. Thus, most of these entered Swahili between 800 and 1700. […]

    However, the word ‘man’ as a borrowing, among one animal and two trade items, raises an eyebrow. John R. Perry’s review of that article and others (Iranian Studies 31 (3/4), 517–525, 1998, p. 521) writes,

    […] it seems inherently unlikely that the well-known word bwana derives from Middle Persian bān, “lord,master” ([footnote:] There is no separate entry bān in EIr, but cf. its feminine derivative BĀNŪ); and it is a pity that there is not a single example of the “hundreds” of Persian borrowings in Swahili between 800 and 1700, nor any attempt at diagnosis of indirect borrowingof Arabic through Persian.

    Is there any inherent reason the Wiktionary etymology, “From Arabic أَبُونَا‎ (ʾabūnā, ‘our father’)”, should be rejected?

  173. David Eddyshaw says

    I am merely parrotting Nurse and Hinnebusch, whose work inspires confidence by its page layout, nice design and by the amount I paid for it … (sunk costs fallacy?)

    However, the paper drasvi linked to has lots of loans attributed to Persian. (It’s worth a look.) Though it also implies that bwana is actually indigenous Bantu …

    “Total absence of non-lexical material” seems an odd counterargument to me. There are (for example) a lot of quite unmistakable Hausa loanwords in Kusaal (and many more in Mampruli and Dagbani), but nothing much in the way of anything else, which is hardly surprising in view of the fact that Hausa morphology and syntax are pretty different from WOV (as different as Persian from Bantu, say …) About all I think you could say might be due to non-lexical foreign influence in Kusaal is the presence of /h/ as a phoneme (as opposed to an allophone of /s/), which occurs in just one common word, which doesn’t seem to be borrowed from Hausa itself anyway … a couple of prepositions have been borrowed, but that didn’t involve innovating a whole new grammatical category or anything as radical as that.
    Again, there are at least two Songhay loanwords in Kusaal. Other Songhay influence on the language? Zilch. None whatsoever.
    I can even vouch for at least three Portuguese loanwords in Kusaal … but my best efforts have revealed no traces of Portuguese phonology or syntax … total absence …

    In fact, the only case where I think I can trace a morphological peculiarity in Kusaal to foreign influence is the word Mɔr “Muslim”, which has the completely unparalleled plural Mɔɔm; it seems likely this has come about from adopting the Mooré More “Muslim” along with its plural Moeemba, which is very slightly less isolated in Mooré than in Kusaal.

  174. Riffing off David Marjanović’s mention of termites/”white ants” (but I’m not going to make this “@David Marjanović”) because I’m sure he personally already knows all of it:

    It is amazing how conclusions about evolutionary relationships based on morphological (or behavioral) arguments have been subject to so many revisions over the last few decades. When I was a child, we still learned that the termites were the antecedents of the Hymenoptera (wasps, broadly construed, so including, besides the obvious bees, also ants and sawflies; the cladogram for the order is extremely complicated), but now we know that the termites are actually a sister group to the cockroaches. Despite the common eusociality between the Isoptera and Hymenoptera, they are not especially closely related insect groups.

    Not all these changes to our understanding of the Tree of Life actually come from DNA sequencing though, and sometimes, even when major mistakes are found, they turn out not to change the cladograms that much. The relations of chordates and hemichordates illustrate both these points. The stomochord, which stiffens the acorn worms,* was long seen as a basal homologue of the notochord, but it is now known to be unrelated. Microanatomical histology shows that the two structures are composed of completely different cell types and are held together with different compounds. However, because they are, like us, deuterostomes (meaning the original embryonic invagination in their embryonic development became the anus, rather than the mouth, when the mouth and anus became separate openings), the hemichordates have only moved from being the sister group to the chordates to being the sister group to the echinoderms, with the chordates being the out group among the deuterostome phyla.

    * As a teen, I memorized common names for ten different phyla of worms: flatworms, roundworms,** proboscis worms, spiny-headed worms,*** segmented worms,**** tube worms,***** jaw worms, penis worms, acorn worms, and arrow worms.

    ** The roundworms of the time are now known to be paraphyletic, and another phylum—the horsehair worms—has been split off.

    *** Spiny-headed worms are now known, based on genetic evidence, to be oversized but degenerate rotifers.

    **** These have also been split. Now that it is known that the earthworms-type segmented worms are not particularly close to the arthropods among protostome (mouth first) animals, the ecdysozoan velvet worms (which actually are very close to the Arthropoda) have necessarily been split off into a separate phylum.

    ***** The tube worms, on the other hand, have been merged into the annelids based on genetic evidence, even though they show relatively little evidence of their ancestral segmentation.

  175. “Oversized, degenerate rotifers.” I’ll file that for future use.

  176. Lodhi’s dissertation (the one linked to by drasvi) says that bwana comes from mwana (under dubwana, p. 159). But doesn’t mwana mean ‘child’? How do you get from ‘child’ to an honorific?

  177. David Eddyshaw says

    Good question. I can’t see it either. (Though Childe Rowland to the dark tower came, if memory serves.)


  178. containing several hundred items from Persian (Krumm; Knappert) and many more from Arabic.

    Krumm seems to be one of Nurse and Hinnebusch’s sources.

    Doubtful PSA Lexical Reconstructions p. 658: For reference purposes, we give a reconstructed shape for each entry; these are marked with a question-mark followed by an asterisk. This is followed by a gloss. If we have information on a source, this is listed in parentheses following the gloss-thus CB, PSC (Ehret 1980a), Arabic, Persian, Indian (for the three latter we rely on Sacleux 1939/41, F. Johnson 1939a, or Krumm 1940).

    Johnson 1939a. A Standard Swahili-English Dictionary.
    Krumm, B. 1940. Words of Oriental Origin in Swahili. London: Sheldon Press.
    Sacleux, C. 1939/41. Dictionnaire swahili-francais, vols. 1 and 2 (Travaux et memoires de
    l’Institut d’Ethnologie
    36-37.) Paris: Musee de l’Homme.

    Krumm looked promicing, so I tried to read it in Google books (Google snippets…).


      The structure of the name Zanzibar must be regarded as wholly Persian. For “ bar ” is not—as it is often alleged in the literature on East Africa—the Arabian “ barra ”=برّ=Swahili bara. In Arabian “ bara ” signifies “ land ” in the sense of continent. It does not seem likely that an island of the small size of Zanzibar has ever been called a “ bara,” i.e. mainland or continent. “ Bar ” (in Iranian) is a formative syllable, used to construct nomina locis (and temporis) “ where something is found in multitude ” (see Beck, Persian grammar, p. 191), or “ which have something to do with the sea or with water ” (see Hess in Z.E.S. Bd. 10, p. 148). That does not exclude the possibility that Arabians are convinced, “ bar ” in Zanzibar was the برّ of their own language , just the same as they believe the “ dar ” in Dar-es-salaam was the Arabic دار. The original name of the island was: Zang-e-bar, or, more correct: Zang-i-bar. The ‘i'(e) is the Persian ‘i'(ya) of relation
    or connection.
      But how could Zang-i-bar change into Zanzibar? This transmutation is very remarkable indeed, as it is due to a phonetic peculiarity of Bantu. There is in Bantu an i (called by Meinhof “ heavy i ” and written î) by which a preceding ‘ng’ is changed into ‘nz’ ; thus ngî> nzi. The Swahili must have regarded or felt the i in “ Zangi ” to be an î and consequently changed Zangi into Zanzi.[1]
      [1] A native from Zanzibar informed the author and insisted on this explanation that “ Zanzibar ” was derived from the name for “ ginger ” (of which the botanical name is Zingiber officinale) which is called by the Swahili “ zengebil.” Ginger grows in the isle of Zanzibar and is often used to be mixed with coffee given to a person when ill. There can be no doubt a relation exists between zengebil, zingiber, and even ginger (from Old English : gingiber) and Zanzibar. But the fact will be that the name for the plant was derived from the name of the place where it grows and not vice versa.

  179. Second,

      Is a term of contempt employed by the natives of the coast when speaking of the native of the interior . That learned Africanist, Dr. K. Röhl, author of the new translation of the Swahili Bible, informed me that, in his opinion, shenzi was derived from a Bantu verb “ shenga = to cut , to skin , to bark ; from this “ shenga ” a nomen agens was constructed : m-shenzi= cutter or skinner of sugar-cane, and this word for “ sugar-cane skinner ” or “ sugar-cane eater ” became a nickname.
      I do not share this view, for I fail to see why “ sugar-cane eater ” should have become a derisive nickname, especially when it is remembered that to have sugar-cane and to be able to feed on it has always been the ardent desire of every native. Therefore, owners of sugar-cane fields were more likely to be envied than derided. Thus, it would seem highly improbable that the original coastal population used to look down upon their brethren from the Hinterland. The term may have come into use when—after the immigration of Oriental people into East Africa—a mixed population sprang up. This mixed coastal race, lighter in hue, and devoid of negroid habits, were particular about being differentiated from the pure black Bantu. On the conversion to Islam of the coastal population, the Waswahili became conceited, and spoke scornfully of the shenzi, because the latter was uncircumcised, wore no clothes, and could not read the Koran. The Waswahili had learnt the term shenzi from their Arabian masters who employed this word in a derogatory sense. I think that there must exist some relation between shenzi and the above-discussed “ zengi.” Is it possible that “ zenği ” changed to shenzi ??
      When endeavouring to verify this conjecture, we must remember §51(d) where it was stated that according to BROCKELMANN (7, I, p. 271) in Arabian dialects metathesis may take place between z and ğ. And we have proof of such a metathesis in the Swahili jozi (=ğozi)<Arabic زوج=zauğ (zoğ). Therefore a similar metathesis may have taken place in zenği resulting in a ğenzi.
      But now the question arises: Could “ ğenzi ” become shenzi, i.e. could a ğ change into š, or—to express it in phonetical terms—could the voiced alveolar fricative change into an unvoiced sound? There are cases when voiced sounds may become voiceless; for instance, when voiced sounds are assimilated to voiceless sounds. And there is a phonetic worn rule (Müller’s law) that voiced sounds may become unvoiced in constant everyday use, when the voiced sound is “ worn out.” According to GAIRDNER (17, p. 52), “ in Arabic both classical and colloquial, such unvoicing of initial voiced sounds does not occur,” but it may take place in South Arabian dialects, and the supposition remains that in these dialects we shall find phonetical changes which explain the transmutation from ğ to š; or could it be that the transmutation is due to a hybrid Bantu m-shenzi, which K. Roehl mentions ?
    (See above.)
      The thesis, that ğ may change to š gives us an explanation not only for the etymology of shenzi, but also for—

      With regard to shamba I have to thank the above-mentioned Dr. Roehl for the following information : “ The word…

  180. ğ (not ǧ) because I wanted the text to be googlible. “m-shenzi” – in the original “m-shenzi” (others were in italics and I inadvertedly italicized this one too).

  181. David Eddyshaw says

    to have sugar-cane and to be able to feed on it has always been the ardent desire of every native.

    It’s the same in Wales.

    Nothing seems to be harder for some ardent etymologists than to say “Alas, we don’t know.”

  182. David Eddyshaw says

    The thing about *g -> z before tense i is real enough, though: that’s part of Bantu spirantization.* N & H discuss it extensively on pp113ff.

    * Proto-Bantu had a seven-vowel system; the tense i and u, but not the corresponding lax vowels, cause many preceding consonants to become fricatives. This is the reason why Swahili -fa “die” corresponds to Kusaal kpi

  183. David Marjanović says

    When I was a child, we still learned that the termites were the antecedents of the Hymenoptera

    Huh. I didn’t know that had ever been thought.

    I’ve never heard of onychophorans as annelids either – as part of a transition between annelids and arthropods, sure, but always as their own phylum and closer to the arthropods.

    But doesn’t mwana mean ‘child’? How do you get from ‘child’ to an honorific?

    I once read bwana actually means “father of sons”…?

    just the same as they believe the “ dar ” in Dar-es-salaam was the Arabic دار.

    …Huh. What is it, then?

    This is the reason why Swahili -fa “die” corresponds to Kusaal kpi …

    That reminds me of the origin of /f/ in much of Sinitic: Middle Chinese bj- pj- pʰj- > bʝ- pç- pçʰ- > bv- pf- pfʰ- > v- f- and a short-lived fʰ- > Standard Mandarin w- f- f-.

  184. David Eddyshaw says

    The fricativisation before originally-tense i is actually still a synchronically active process in Swahili, eg. jenga “build”*, mjenzi “builder.”

    In fact, I think that really does dispose of the difficulty as far as the final consonants of Zang- and -senzi is concerned.

    Vowels, as we know, count for nothing in etymology … (though, seriously, both *Zang and *Zeng really do seem to be attested for the “Black Africans” word) …

    So the only remaining difficulty with equating the roots of Zanj and mshenzi is the initial z/s/sh correspondences.

    Krumm’s “wearing out” idea has nothing whatsoever to recommend it, but I suppose voicing dissimilation of z-z to s-z is conceivable. And there does seem to be something fishy going on with the intra-Sabaki s/sh variants in any case.

    All this is really predicated on N & H being wrong about msenzi being “Omani” Arabic. If the word really is found in Oman and not just Zanzibar, that would be awkward. A fortiori, of course, if it turns out to be a pukka Arabic word found all over the place ** …

    * Yes, that’s where the game gets its name from.

    ** But then, I just used the word pukka, and I’ve never been to India …

  185. I have always wondered whether the portmanteau name “Tanzania”, created from “Tanganyika” and “Zanzibar” when the states merged in 1964, suggested any particular meaning to the locals.

  186. Spiny-headed worms are now known, based on genetic evidence, to be oversized but degenerate rotifers.

    No wonder Lovecraft hated sea life.

  187. I guess “eldritch” = “oversized but degenerate.”

  188. Goodness, I had no idea the word had such a long and flourishing history (OED entry from 1891):

    Weird, ghostly, unnatural, frightful, hideous.

    1508 W. Dunbar Goldyn Targe (Chepman & Myllar) in Poems (1998) I. 188 There was Pluto, the elrich incubus.
    1513 G. Douglas tr. Virgil Æneid vii. 108 Vgsum to heir was hir wyld elriche screik.
    1535 W. Stewart tr. H. Boethius Bk. Cron. Scotl. (1858) II. 636 Thair cleithing quhilk wes of elritche hew.
    c1540 J. Bellenden tr. H. Boece Hyst. & Cron. Scotl. vi. ix. f. 71/1 Mony haly and relligious men..fled in desertis and elraige placis.
    1584 King James VI & I Ess. Prentise Poesie sig. Miiij The king of Fary..With many elrage Incubus rydand.
    1598 J. Melville Diary 25 Feb. 320 The amazfull, ugly alriche darkness.
    1793 R. Burns Poems (ed. 2) II. 220 Ye’ll find him snug in Some eldritch part.
    1834 T. Pringle Afr. Sketches ii. 144 Loud bursts of wild and eldrich laughter.
    1850 N. Hawthorne Scarlet Let. vii. 127 Pearl..gave an eldritch scream.
    1860 Ld. Lytton Lucile i. iii. § i. 87 Truth is appalling and eltrich, as seen By this world’s artificial lamplights.
    1866 W. D. Howells Venetian Life iii. 40 Joy that had something eldritch and unearthly in it.

  189. David Eddyshaw says

    You can be eldritch and petite. And “degenerate” is in the eye of the beholder, puny humans!

  190. David Eddyshaw says

    Krumm, frankly, seems to be full of it. Immediately after the nonsense about Dar-es-Salaam he says

    The original name of the island was: Zang-e-bar, or, more correct: Zang-i-bar. The ‘i'(e) is the Persian ‘i'(ya) of relation or connection.

    Nope. It can’t be, because the order is back to front. (Otherwise it would be “Africans of the Place.”)


    EDIT: on reflection, I suppose he actually does mean that, parsing it as “Africans of the Sea” or some such. He’s still full of it. And his explanation that the final g has become z before Persian ezafe in “Zanzibar” because Proto-Bantu *g became /z/ before original tense /i/ (long before there even was any Swahili language, in which the tense/lax versions of i and u have merged completely) is a masterpiece of confusion.

  191. David Marjanović says

    No wonder Lovecraft hated sea life.

    Acanthocephalans are parasites in the guts of vertebrates; free-living rotifers are ubiquitous in freshwater.

  192. David Eddyshaw says

    The order of the elements in “Zanzibar” is neither Arabic nor Persian-with-ezafe; the WP article on “Zanzibar” says it’s from a Persian compound Zangba:r, which not only makes perfect sense as “Land of the Black Africans”, but actually is the Persian name for “Zanzibar”:


    Arabic /ɟ/ for Persian /g/ and the insertion of an epenthetic vowel to break up the CCC cluster are exactly what you would expect. The Zanz- would be a European mangling of the Arabic, and nothing to do with African languages at all.

  193. @languagehat: The eldritch quote from The Scarlet Letter is an example of Hawthorne trying to portray the child, Pearl, as some kind of creepy faerie being. Unfortunately, his portrayal is somewhat at odds with the contemporary associations of phrases like “elf-child.” I don’t remember the text ever explicitly suggesting that Pearl was a changeling, that she is not Hester’s natural child, but maybe I just missed that idea, or it was supposed to be so obvious to Hawthorne’s contemporary readers that he never spelled it out. However, it does seem to conflict with another symbolic purpose that Pearl often serves, as a physical manifestation of Hester and the reverend’s sin.

    This reminds me, however, how much the discussion of characterization in Hawthorne’s fiction in my American Literature class was dominated by a paradigm that required us to fit his characters into six archetypal models. These six types were not useless, by any means There was value in pointing out the similarities between Hester Prynne and Reverend Mr. Hooper from “The Minister’s Black Veil,” as exemplars of the “exalter” character type, and how Hawthorne felt that overtly proclaiming one’s sinfulness was not the answer to the problem of human sin, but rather a dangerous dodge that added to the alienation of those protagonists from their communities. However, the character categories are not generally so clear cut as the authors of that scheme seemed to want. For example, Arthur Dimmesdale is clearly both Hawthorne’s canonical “concealer” of sin and yet, at the end of the book, also the “redeemed.” Even worse, we had a long discussion, attempting to shoehorn the character of Doctor Prynne/Roger Chillingworth into the “scientist” archetype, which he does not clearly fit, because the narrative does not reveal enough about his internal thought processes to understand his attitude toward his own failings. Only much later, when I read The House of the Seven Gables, did it become apparent that the “scientist” was modeled on Jaffrey Pyncheon, the villain of that novel, so some of the specifics in the description we got of the “scientist” really only fit him. And ultimately, after groups of students were assigned to pick and analyze additional short stories by Hawthorne, even our unfortunately unprepared teacher admitted that the six purported Hawthorne character types had no relevance to a story like “The Gray Champion.”

    @David Eddyshaw: I generally dislike describing organisms as “degenerate,” but those acanthocephalans have lost so much interesting anatomy, it’s hard to see them as otherwise. About their only derived trait is the eversible proboscis (along with its spines), and an eversible proboscis has evolved many times. Such structures are also found in acorn worms and give penis worms and proboscis worms (also known as “ribbon worms”) their names. The arrow worms, interestingly, achieve a similar morphological effect in a different way, with a retractable hood that uncovers the mouthparts when needed.

  194. Krumm… чудит.
    чудо “wonder”,
    чудак “an excentric person”,
    -т 3sg present. -и- one of verbal stem suffixes.

    (чудачка is a person too. (I am not sure if *мудачка is attested, though))

  195. I wonder if the etymology “Zanzibar<"ginger" can be true. Krumm so confidently dismissed it, that he made me curious.

    The fricativisation before originally-tense i is actually still a synchronically active process in Swahili, eg. jenga “build”*, mjenzi “builder.”

    And his explanation that the final g has become z before Persian ezafe in “Zanzibar” because Proto-Bantu *g became /z/ before original tense /i/ (long before there even was any Swahili language, in which the tense/lax versions of i and u have merged completely) is a masterpiece of confusion.

    @David Eddyshaw, don’t you contradict yourself here? He did not say Proto-..

  196. dar-es-salaam
    I looked up ‘dar’ in the book.
    p.114, Krumm 1940
    ‘…and Arabs will interpret Dar-es-salaam in this manner. But in former times and in old writings (cf. p. 5) the place was always called: Bendar-es-salam. Therefore no doubt can exist that the original meaning of the name was “ harbour of peace or salvation.” ‘

  197. David Eddyshaw says

    don’t you contradict yourself here?

    Good catch; I don’t think I do, but I did express it pretty sloppily.

    The process is still synchronically active in Swahili word formation, but it can’t be expressed as a synchronic morphophonemic rule because there is no longer any distinction in Swahili between the original /i/, which caused “spirantisation”, and /ɪ/*, which didn’t. If you try to set up two distinct phonemes which always coincide in realisation, Paul Kiparsky will come along and eat you.

    Moving along …

    For the fricativisation to work before the ezafe of *Zang-e-bar (which would actually be lax, not tense, in Persian, not that it matters) the Swahili speakers would have to interpret the Zangi as a mistaken form of a deverbal agent noun (which for some reason was appearing without any noun class prefix) and “correct” it to zanzi- by analogy. This is what Krumm’s blether about “the Swahili must have regarded or felt the i in “ Zangi ” to be an î” actually entails. Of course, in reality no Swahili speaker who hasn’t studied Bantu historical linguistics “feels” that the /i/ suffix of deverbal agent nouns, even, is an “î”, let alone random lax ɪ’s in loanwords.

    Another weak point in my own argument (if i may say so) is that I am making an assumption that the falling together of *i and *ɪ in Swahili predated the supposed changing of *Zangibar to “Zanzibar”; if such a change actually had taken place, that might well not be the case. The fact that its effects haven’t been levelled even in modern Swahili suggests to me that the sound change is probably comparatively recent.

    But there is no reason even then to think that the i in this “Zangibar” was tense (in fact, if it had been Persian ezafe, it wouldn’t have been; and Krumm’s hypothesis in any case founders on the fact that the order of the elements pretty much proves that the original form was actually a Persian compound noun of a perfectly straightforward construction, still in use in present-day Persian.

    * Bantuists write them as i̢ i respectively, in order to épater les bourgeois and because they’re basically bad people.

  198. David Eddyshaw says

    Bandar-es-Salaam looks right enough, though. Interesting.

    I imagine that it would be natural enough to change it to Dar-es-Salaam, not only because the first bit is Persian, but because of the analogy with دار الإسلام


  199. David Eddyshaw says

    The -bar element also turns up in “Malabar”


    though it looks like this isn’t as neatly soluble as the etymology of “Zanzibar.” But I notice it’s Ma:la:ba:r in Persian (and Arabic), FWIW …

  200. David Eddyshaw says

    I should have written: “Bantuists write /i ɪ/ as i respectively”, not “i̢ i.” Bantuists may be bad people, but they have not sunk to that level of depravity.

  201. How does “Afghanistan” work?

    I do not mean

    افغان • (afğân)
    Etymology 1
    Afghan, Pushtun
    Etymology 2.

    I mean, how do you tell it from ezafe?:/ Syn-/diachronically, anyhow. cf.


  202. David Eddyshaw says

    Well, obviously you can tell the difference from the meaning because in the ezafe construction the head comes first (so *Afghan-e-Stan would be “Afghan of a Land”); presumably you mean, Is there any phonetic difference between ezafe and an epenthetic vowel? I don’t know, really, but I don’t think so.

    Lambton’s grammar says that compounds are stressed on the last syllable, but her actual examples of compounds actually include several which look like set phrases rather than run-together stems, so I suspect stress won’t distinguish set phrases with ezafe from compounds (but again, I don’t really know.) I don’t know about subtleties like junctures in Persian, either.

    However, in this particular case, at any rate, I don’t think you could segment “Afghanistan” as *Afghan-e-Stan, because Persian doesn’t like initial /st/ clusters, so I presume it would have to have been *Afghan-e-Estan.

    Xerîb is likely to be able to give a proper answer to this, I suspect.

  203. David Eddyshaw says

    I see that although I was right about the Persian short and long vowels differing in quality, I’m wrong to say that the short high front vowel is [ɪ]; in fact it’s [e] (at least in Western Persian):


    (Similarly, [o], not [ʊ].)
    It doesn’t affect my argument: nevertheless, I was Wrong. Wrong!

    It did occur to me that the Persian might conceivably have retained short [i] [u] up until the period of Krumm’s alleged Swahilification of a supposed Persian *Zangibar to Zanzibar (even up until a time when Swahili itself still distinguished /i/ from /ɪ/, in fact) but it appears not. That hasn’t been the case since Persian became Modern.

    The WP article seems to confirm that stress alone doesn’t reliably distinguish ezafe constructions from compounds.

  204. I wonder whether the shift of i to e took place in the southern dialects of Persian to which Zanzibar would have been more exposed, but I don’t think it would matter: the Arabic form (with typical Omani g for jiim) would still presumably have been Zangibaar.

  205. David Eddyshaw says

    What do you think about this Omani/Zanzibari Arabic msenzi, that Nurse and Hinnebusch cite, Lameen?

    On another tack, I wonder if Krumm’s Swahilification idea for the second z in Zanzibar could be rescued by positing Omani Arabic rather than Persian as the immediate target? Short i as a tense /i/ vowel in any sort of Arabic doesn’t seem all that likely, but I know very little about this. The paper I linked to on Omani Arabic does in fact have short /i/, but as far as I can see, only in cases corresponding to Classical /i:/ …

  206. presumably you mean, Is there any phonetic difference between ezafe and an epenthetic vowel?


    The context was Krumm’s “The ‘i'(e) is the Persian ‘i'(ya) of relation or connection.” (where he seems to identify both sounds).

    Then the Syntacticians discuss somehting called “reverse ezafe” in Caspian languages : farha:d-e xa:xer-e hemsa:ye “neighbour of Farhad’s sister” (and Balochi – but their examples come from a grammar of the Balochi of Turkmenistan, and I do not know about the Balochi of Balochistan).

    Then WP has “Zangba:r” (you quote it above) but elsewhere it is zangebâr.

    Farsi Wiktionary splits them as gul-istan, afghan-istan ([اَفغان/اِستان] /گُل/اِستان/) but zangi-bar:
    زنگی + بار، «ساحل سیاه‌پوستان»
    WHich looks like cheating.

    And I realized that I do not understand somethging as simple as “Afghanistan”. I do not know Persian, though:(

  207. the Syntacticians
    I mean, Chomskian Syntacticians:not very interested in philological information.

  208. Is there any phonetic difference between ezafe and an epenthetic vowel?

    Not necessarily phonetic. Maybe morphological. …
    What you seem to be saying: one is “epenthetic”, the other is a morpheme.
    What Krumm seems to be saying: both are the same morpheme.
    I do not understand how I tell one from another:( Say: a connective vowel morpheme (like connective -i- in Latin), identical to the ezafe clitic.

  209. David Eddyshaw says


    You could well be right about the modern Persian form really being Zangeba:r with an epenthetic vowel. (Some Hatter will know …)

    Three-consonant clusters do seem to occur in Persian, but I don’t know what constraints there are on them.

    The -esta:n forms wouldn’t need an epenthetic vowel, which is presumably why Wiktionary writes those first components without a trailing -i. Writing زنگی just looks wrong though: that would presumably imply that the whole word should have been *Zangi:ba:r, which it isn’t.

    If Krumm really was implying that an epenthetic vowel in compounds was the same morpheme as ezafe he was without doubt just plain wrong, though they may well (for all I know) be homophonous.

    If Baluchi does indeed have the opposite order to Persian in so-called ezafe constructions, that is of no significance for Persian.

    According to Lewis, traditional Turkish grammarians use the term izafet to describe the construction of e.g. Orhan’ın ismi “Orhan’s name” (“Orhan’s his-name”), so the term has got dissociated from the Persian type.

  210. David Eddyshaw says

    I dare say it might be possible (if I knew any Persian) to contrive an example that was ambiguous between the noun phrase X + ezafe-e + Y and the compound noun X + epenthetic-e + Y; but the existence of ambiguity would not make the two the same.

    In Persian ezafe constructions the first element is the head. In Zang(e)ba:r the second element is the head. Ergo, it cannot be an ezafe construction in reality (unless a plausible place name meaning as “Black Africans of a Something” can be cooked up by picking a suitable – if, apparently, unparalleled – meaning for the ba:r “Something” component.)

  211. If Baluchi does indeed have the opposite order to Persian in so-called ezafe constructions, that is of no significance for Persian.

    Just for clarity: the example above was “Mazandarani”, from (the paper also has “Modern linguistic research strongly suggests that children come to aquisition armed with …”).
    Baluchi is here.

    I only mentioned it to explain the context of my question. It may have significance : “similar sounds within the same language area are used to connect elements of chains”.

  212. David Eddyshaw says

    I would hazard a guess that Krumm had in mind something like a mangled Persian form from Arabic بحر as the second element, creating a sort of reversed form of بحر زنج


    and hence “Zanj of the Sea.” But you’d need to show that there actually was such a word in Persian … and, really, why bother, when there is a perfectly satisfactory etymology there already, without such gymnastics?

  213. The i is tense in Standard Arabic at least, if not in most dialects.

    Reinhardt’s grammar of Zanzibar Arabic doesn’t seem to give the actual form for “Zanzibar” anywhere (much less this improbable supposed Arabic word msenzi.) He does transcribe “woman from Zanzibar” as zengyje, though. I could see msenzi as a loan from Omani Arabic zengi if at some relevant point Swahili had no phoneme z; is that compatible with the sound change history?

  214. I would hazard a guess that Krumm had in mind something like

    Krumm says: The structure of the name Zanzibar must be regarded as wholly Persian.

    Well: The structure of the name Zanzibar must be regarded as wholly Persian. For “ bar ” is not—as it is often alleged in the literature on East Africa—the Arabian “ barra ”=برّ=Swahili bara. In Arabian “ bara ” signifies “ land ” in the sense of continent. It does not seem likely that an island of the small size of Zanzibar has ever been called a “ bara,” i.e. mainland or continent. “ Bar ” (in Iranian) is a formative syllable, used to construct nomina locis (and temporis) “ where something is found in multitude ” (see Beck, Persian grammar, p. 191), or “ which have something to do with the sea or with water ” (see Hess in Z.E.S. Bd. 10, p. 148). That does not exclude the possibility that Arabians are convinced, “ bar ” in Zanzibar was the برّ of their own language , just the same as they believe the “ dar ” in Dar-es-salaam was the Arabic دار. The original name of the island was: Zang-e-bar, or, more correct: Zang-i-bar. The ‘i'(e) is the Persian ‘i'(ya) of relation or connection.

    If Krumm really was implying that an epenthetic vowel in compounds was the same morpheme as ezafe
    This is how we both understood him….

    Neupersische Konversations-Grammatik : mit besonderer Berücksichtigung der modernen Schriftsprache / von Sebastian Beck. and Zeitschrift für Eingeborenen-Sprachen. mentioned by him can be found on Hathitrust: Beck, ZES. I can only access it through VPN.

  215. And, by the way. Zhao_Rukuo or

    Chau Ju-Kua: his work on the Chinese and Arab trade in the twelfth and thirteenth centuries, entitled Chu-fan-chï, Translated from the Chinese and Annotated by FRIEDRICH HIRTH and W. W. ROCKHILL, (1911) with W. W. Rockhill, ST. PETERSBURG, Printing Office of the Imperial Academy of Sciences, Vass. Ostr., Ninth Line, 12. 1911.

    It deals with Zanzibar on pages 142 and 165 of the pdf.

  216. I went to the library to see what Krumm and Knappert have to say about bwana. Knappert, the more recent source, doesn’t mention it, presumably because he doesn’t think it’s a Persian (or Turkish) loanword. He acknowledges Krumm’s contribution, while adding that Krumm’s etymologies often suffer from lack of rigor. Krumm says:

    In the German edition I have explained bwana as probably being formed by a contraction of Arabic ‘bu’ (<abu) and the Swahili stem for child “ana.” In a note I added that from the use of the word bwana for addressing superiors we may conclude that the Iranian بان = bān “prince, master, chief,” must have influenced the meaning of bwana, but that it would be wrong to derive Swahili bwana from Iranian bān, as we would miss then an explanation of the ‘w’ in bwana. This etymology has been criticised by K. R. in Hochland, Dez. 1933. K. R. opines that bwana is contracted from Bantu “bu” = great, and the Bantu diminutive suffix “-ana.” Thus K. R. and I differ in two points, (1) K. R. says, “-ana” is a diminutive suffix, whilst I stated “-ana is the stem for “child.” I do not see in this so big a difference that it seems worth while to discuss it. Anyhow, we agree that -ana is Bantu. (2) K. R. says ‘bu’ is the Bantu-stem for great, whilst I maintain ‘bu’ is short for Arabic ابو = “abu” = father. It may be possible to derive a Swahili ‘bu’= “great” from a Bantu root, and perhaps the frequent Bantu word mbuya = grandmother is connected with this stem. But mbuya is not met with in Swahili, and bwana would be the only example. K. R. mentions babu = grandfather as another example, but I cannot see proof in this word, as babu is a very common oriental expression (cf. Pers. Hind.: بابو = babu = an old man), and among the Bantu languages confined to Swahili as far as I know.¹ There is a form of the A. abu with the pers, pron. suffix ‘na’ = our; thus “abuna” or “abana” = “our father.” “Abu” can be shortened to “bu” (cf. Reinhardt, 43, p. 79), and this “bu” has in some Arabian dialects become an adjective with the meaning of “proprietor of” (cf. Stumme, Handbook of Shilhish, p. 22), so that a possible explanation might be “bu-ana” = proprietor of children (in the same sense as the biblical name Abraham was given as “father of many children”). My opinion therefore is: bwana has several sources. Formed by contraction of “bu” and “ana,” it is influenced in its meaning by Arabic “abuna” and Iranian “bān.”

    ¹ In Bertram Thomas’ (51, p. 11) we read the following passage which is of some interest for our word in question: “Did not the ruler habitually address his slave as abana = our father!…” I think it worth mentioning that in Hausa (which has nothing to do with Swahili) we have “uba-na” = my father.

    [inconsistent italics retained. I don’t know who K.R. is.]

    I think it’s a mess.

  217. “K. R. says, “-ana” is a diminutive suffix,”
    “K. R. says ‘bu’ is the Bantu-stem for great, ”

    little great!

  218. Stumme, Handbook of Shilhish

    H. Stumme, Handbuch des Schilihschen von Tazerwalt, Leipzig, 1899
    Hathitrust (again, VPN)

  219. THe Ancients knew how not to waste endless evenings italicizing the stupid bibliography consistently…

  220. David Eddyshaw says

    Thanks Lameen!

    I have been pronouncing Classical Arabic /i/ /u/ as lax all this time, and I was wrong … (Again!)

    Glad you confirm the idea that this msenzi word doesn’t look very Arabic. Not just us, then …

    Zengyje seems to be zengīye when unGermanicated, which makes sense: zanjiyya.

    I could see msenzi as a loan from Omani Arabic zengi if at some relevant point Swahili had no phoneme z; is that compatible with the sound change history?

    Proto-Bantu *ɟ has become /z/ (or /j/) in most Sabaki languages, but Swahili itself seems to mostly have /ɟ/, as in jenga “build”, maji “water”, and everyone’s favourite Swahili word jambo. In fact, most cases of Swahili /z/ actually seem to arise from that very fricativisation process we’ve been discussing (including a word I’ve discussed here before, zaa “give birth”, which to my great joy turns out to be related to Kusaal biig “child.”) There’s been a lot of dlalect/language mixture to complicate things, though.

    N&H think that the reduction of the seven-vowel system to five has diffused across the Sabaki languages “relatively recently” (p549), and later in Swahili than most, although they don’t stick their necks out with any actual dates, even rough ones.


    At the time that a supposed *zengi would have undergone “Bantu spirantisation” changing the g to /z/, it is at least conceivable that there actually wasn’t any other /z/ in the language prior to this very spirantisation.

    So, yes, I think it could have gone *zengi -> *musengi -> *musenzi -> msenzi.

    I don’t know about mshenzi, though; as far as I can see, native Swahili sh also pretty much only comes from spirantisation (of *k.) Awkward …

  221. David Eddyshaw says


    Both Guthrie’s and Meeussen’s lists of Proto-Bantu roots are innocent of any “Bantu-stem” *bu “great” (though there is a *bu “bad” …)

    *(j)ana “child” is not a “diminutive suffix” (though it does occur with diminutive meaning as the first stem in compounds, according to Meeussen’s rather nice Bantu Grammatical Reconstructions, which you can find on the intertubes.)

    Yes, it’s a mess. Lamentable stuff.

  222. The only point that makes sense is that Persian /b/ wouldn’t be borrowed as Swahili /bw/.

    Going back to the Encyclopedia Iranica article, I don’t see what their source was for including bwana among Persian loans, since both of their references don’t think it was one.

  223. David Eddyshaw says

    Persian /b/ wouldn’t be borrowed as Swahili /bw/

    I’m not so sure about that. Swahili b, despite the orthography, is actually an implosive, except in mb.
    I can conceive of (pre-)Swahili speakers having some difficulty with the exotic sound [b]*, and consequently (over)labialising it.

    [Quite irrelevantly, Kusaal for “dog” is baa, whereas Swahili has mbwa. I actually suspect that these words really are related, though sadly not by borrowing of the Kusaal word by Swahili. The Proto-Bantu form was *mbúa; WOV as a whole seems to have irregularly lost the vowel rounding which actually does turn up elsewhere in Oti-Volta, e.g. Moba bɔg “dog.”]

    *Before their Arabic improved.

  224. let us remember that
    – “diminutive” and “great” are arguments that he disagrees with, while Omani msenzi is something that he does not propose.

    – Guthrie’s and Meeussen’s lists did not exist. He is using whatever sources are available to him.

  225. WOLD:

    The role of other [than Arabic and Indian languages] languages as direct donors of Swahili loan words is less prominent or even doubtful.
    Shirazi descent has enjoyed a high prestige in traditional Swahili society, and there are indeed quite a number of loanwords originating from Persian. However, since there is no convincing evidence of any actual Persian presence in East Africa, and since Persian has been a prestigious donor language for Hindi and the varieties of Arabic spoken on the eastern and southern coast of the Arabian peninsula, we assume that all or most such words entered Swahili via one of these other languages. [7]

    [7] Note however six words in the database having a probable Persian etymology without any indication of an alternative donor language: pamba ‘cotton’, gurudumu ‘wheel’, bwana ‘Sir’, malaya ‘prostitute’, boma ‘fortress’, balungi ‘grapefruit’.

  226. Lodhi’s dissertation (the one linked to by drasvi) says that bwana comes from mwana (under dubwana, p. 159). But doesn’t mwana mean ‘child’? How do you get from ‘child’ to an honorific?


    But doesn’t mwana mean ‘child’? How do you get from ‘child’ to an honorific?

    I once read bwana actually means “father of sons”…?


    mwana can be a honorific too:
    The ruler was a king (the Mwinyi Mkuu, the Great One) or a queen (Mwana wa Mwana, Daughter of the Daughter), but the administrative head was an appointed male minister, the Diwani (pl. Madiwani), whose office also tended to be hereditary.

    Also Mwana Khadija.

  227. David Marjanović says

    Zengyje seems to be zengīye when unGermanicated

    (entdeutscht, surely)

    I was wondering about the y! Y for /iː/ is official in Lithuanian, but it’s also a strong tradition in Swiss German, including such surnames as Wyss and Schnyder that have diphthongized in flatter lands.

  228. Bantuists may be bad people, but they have not sunk to that level of depravity.

    What, there are levels of depravity below “total”? You surprise me.

  229. David Eddyshaw says

    I was speaking in terms adapted to the imperfect understanding of the secular.
    Naturally, the depravity of Bantuists is total – unless, of course, they repent of their detestable orthographic perversions. There is hope, yea, even unto Bantuists!

  230. Bantuists write /i ɪ/ as i̡ i

    No, Bantuists write *į *i for two vowels for which */i ɪ/ is a proposed reconstruction. Given the disproportionately high degree of spirantization that the “super-high” vowels trigger (e.g. one study that draws its evidence disproportionally from Bantu finds that “fricativization is the prototypical sound change”, surely false otherwise), at least I would side with the alternate proposal that they were originally fricative, *z̩ *v̩ or the like.

    I am not informed enough to claim if this also implies that *ɪ *ʊ should be readjusted as plain *i *u, for complete vindication, or if they do retain enough [-ATR] behavior, but at least I see no reason to gripe about the diacritics for the super-high vowels. (I am not certain what is the one you’ve used there, the most common one I’ve seen *į *ų with an ogonek.)

  231. David Marjanović says

    I was looking for that post for a long time! 🙂 I had once tried to download the paper linked there, but somehow that didn’t work, and I found out too late to just try again. Unfortunately the link doesn’t work anymore: Semantic Scholar gives me an error message.

    Tumblr is being weird, BTW. Following your link gives me a “consent” page. I click on rejecting all cookies and am redirected to a general Tumblr login page every time. The trick is to reduce the URL to https://possessivesuffix.tumblr.com/post/166495229723/ – that works, and redirects to the URL you provided.

  232. David Eddyshaw says


    Analysing *į *ų (or whatever) as (phonetic) fricatives is possible, I suppose, though it seems difficult to square with their reflexes in the actual Bantu languages, cognates in non-Bantu Volta-Congo languages, Bantu syllable structure, and the fact that they are tone-bearing.

    “Bantu spiratisation” is a misnomer, in the sense that it is not found in all Bantu languages:


  233. David Eddyshaw says

    While I take your point that symbols in a reconstructed protolanguage are in principle arbitrary, not directly representing anything phonetic at all, and that there is value in using odd symbols simply to avoid premature judgments about their possible realisation: it still seems a bad idea to pick symbols with outright misleading connotations. Ogonek is usually used on vowel symbols to indicate nasalisation, and nobody has proposed that as the distinctive feature of these vowels. It’s actively misleading as a notation.

    (The symbol I was using was a backwards ogonek. as seen for example here:



  234. Fricative vowels are regardless attested from the transition-towards-Bantu zone of Benue-Congo. It seems to me that these arose relatively late in the pre-Bantu development and were also lost right away in early Common Bantu, but maybe leaving something like *Czi *Cvu (but also depending on the area why not Janson’s *Cji *Cwu) as initial reflexes. Simplification to /z/, /h/, /f/ etc. has been obviously later and massively areal. But it makes no sense to me to posit that the full process happened independently in every involved language.

    Known cases of fricative vowels occur quite often in languages that are tonal and which have otherwise very simple syllable structure, so these seem like non-issues to me.

    Tumblr’s cookie check redirect seems to me to be triggering basically randomly as of late.

  235. David Eddyshaw says

    Good point about it being an odd process to occur many times independently if it actually was just triggered by the vowel tenseness; and the fact that it differs in detail across the languages suggests exactly that, viz that it did happen independently many times.

    And the Bantu area is a bit on the large (and diverse) side to atttribute it all to diffusion; mind you, the current geographical dispersion is also largely late …

    Do you have examples of languages with tone-bearing fricatives (I don’t doubt you, I just cant think of any)?
    (Though this would actually not be a relevant point anyway, if your idea is couched in terms of *Czi *Cvu sequences, at that …)

    I suppose there’s Mandarin, come to think of it, which I believe has one or two speakers …

  236. David Marjanović says

    The tone-bearing oddities of Mandarin are mostly syllabic approximants on average, I’d say. There’s rarely any friction.

  237. A silly question:

    I think there are people here much better familiar with Persian history and language than I am. Does anyone know this “bân”, other than in compounds? It is not in fa.WP. In fa.wiktionary it mentioned as a part of comound words: باغبان، دربان، دبده‌بان.
    در – это door.
    باغ – это bagh.
    what is دبده‌ I do not know:( en.Wiktionary adds tonbân “trousers”, darvâze-bân “goalkeeper”

  238. PlasticPaddy says

    You can find the word (بان‎) on its own in the Shahnameh using text search. The word “keeper” in English would be archaising outside of compounds, except for the back-formation, which is preferred by some to goalie.

  239. @PlasticPaddy, thank you! I will do it. As I understand, phonetically bwana and bân just “sound similar”, and there is no reason to think that local development is impossible.

    Semantically, if it is just “a word from the dictionary”, such an etymology has the status “maybe”.

    Usually, when you borrow a honorific it is a word already used this way in another language.
    Of course, sometimes we borrow words in a different meaning (cf. Slavic “parking”: it is a парковка “parking lot, car park” that wants to sound English). Also I knew rais (from pirate names and from the Central Asian version of “the chairman of a kolkhoz”) before I began learning Arabic, the word is very productive – but if with time it ceases to be so, I can imagine that future scholars won’t even know that it once was.

    Yet if the word is known to have once been used in the region, it would improve the status of this etymology. It bothers me that I never heard the word and

    Note however six words in the database having a probable Persian etymology without any indication of an alternative donor language: pamba ‘cotton’, gurudumu ‘wheel’, bwana ‘Sir’, malaya ‘prostitute’, boma ‘fortress’, balungi ‘grapefruit’.

    implies: not attested in India.

  240. trousers
    The idea to call them “body guard” is funny though.

  241. David Eddyshaw says

    We should do the opposite.

    “The President will be accompanied on his visit by his elite military trousers.”

  242. But what about presidents who appear in public without body guards?

  243. David Eddyshaw says


  244. That’s when you need a child.

  245. Then an aquitance of mine once found the president of Lithuania standing right behind him in the crowd in Vilnius. Very shamelessly (or else the trousers were invisible…)

  246. I was surprised to find out that trousers comes from a Celtic source.

  247. David Marjanović says

    And not just any Celtic source, according to Wiktionary:

    Attested since the 1610s, from the earlier form trouzes (attested since the 1580s), extended from trouse (1570s), with plural ending typical of things in pairs, from Middle Irish triubhas (“close-fitting shorts”), of uncertain origin. The unexplained intrusive second -r- is perhaps due to the influence of drawers.[1]

    Breeches is a loan between much earlier Celtic and late Pre-Germanic, probably from the latter into the former (and then into Latin: brācae) according to what’s in Wiktionary.

  248. Must be a Celtic word for cowards (Russian trusý “underpants”, trúsy “cowards”)…

  249. OED, as follows, in what must be a record for a mealy-mouthed exposition of a tangled etymology


    Alteration of trouses, plural of trouse, apparently after other words for paired items ending in -ers, which are the plurals of formations with -er suffix (probably especially drawers, plural of drawer), but also, from other semantic fields, pliers, a pair of pincers at pincer, pair of twitchers at twitcher, etc.


    Apparently partly a variant of trews, and perhaps partly independently < Irish triús trews.

    Perhaps also influenced by Middle French, French trousses, denoting a kind of breeches or hose (1570; specific use of the plural of trousse truss; compare truss).
    Early evidence for trouse appears to refer more commonly to Ireland than Scotland (unlike trews which is found in both Scottish and Irish contexts in early use); this may suggest a partially separate path of borrowing for trouse, perhaps in connection with the English conquest of Ireland in the early 16th cent., which may also have had an influence on the word’s pronunciation.


    < Irish triús and Scottish Gaelic triubhas (Early Irish triubus), of uncertain origin, perhaps a borrowing < a Romance word related to Anglo-Norman trebuz, tribuz and Old French trebus (late 12th cent., apparently rare), and its etymon Old Occitan trabucs, trebucs (late 12th cent. as trebuz) < post-classical Latin trabrugi (8th cent.; also tribuces (12th cent.); although this is probably already influenced by Romance forms), variant or alteration of tubruci kind of leg coverings (a636 in Isidore), probably < an early compound in a Germanic language (reflected by, cognate with, or formed similarly to Old High German diohbruoh loincloth, breeches) < the Germanic base of thigh + the Germanic base of breech.

    (“probably” x 3, “perhaps” x 3, “partly” x 2, “apparently” x 2.)

  250. Does anyone know this “bân”, other than in compounds?

    The Dehkhoda dictionary has this note under بان :

    چنانکه از متنها و لغت نامه ها معلوم شد کلمه بان پساوند است و بتنهایی بکار نمی رود

    As is clear from texts and dictionaries, the word bān is a suffix and is not used by itself.

    I was looking through the Shahnameh, and I could only find one بان bān not used in a compound (such as دیده‌بان dīdabān “sentinel, lookout” and ساربان sārbān “leader of a caravan, camel keeper” and جهانبان (jahānbān) “powerful king, lord”, literally “world-guardian”, etc.), but it was not an instance of this bān. Rather, it was another word bān, meaning “benzoin, styrax”. It occurs in the description of the reign of Jamshid and the many beneficial things he introduced to mankind:

    دگر بویهای خوش آورد باز که دارند مردم به بویش نیاز
    چو بان و چو کافور و چون مشک ناب چو عود و چو عنر چو روشن گلاب

    digar bōyhā-yi xwaš āvurd bāz / ki dārand mardum ba bōyiš niyāz
    ču bān u ču kāfūr u čōn mušk-i nāb / ču ’ūd u ču ’ambar ču rōšan-gulāb

    Furthermore, he brought back perfumes that men hold dear for smelling,
    like benzoin, camphor, pure musk, agarwood, ambergris, and splendid rosewater.

    what is دبده‌ I do not know

    I think this  دبده‌بان is a typo for the very common word دیده‌بان dīdabān “watchman, lookout, guard, sentinel”.

    Lodhi’s dissertation (the one linked to by drasvi) says that bwana comes from mwana (under dubwana, p. 159). But doesn’t mwana mean ‘child’? How do you get from ‘child’ to an honorific?

    As for the etymology of bwana given by Lodhi (“Dubwana is a derogative/amplicative formed from the original Swahili word mwana > bwana > dubwana, and it is not a loanword”), I came across this this note (no. 139) in Jan Vansina (2004), p. 253, saying that in the greater Bantu context, the honorific mwana and the mwana meaning “child” are two different, tonologically distinct words, as far as I understand it from this note.

  251. Re trousers: I’d rather see a heavy use of “perhaps” when appropriate than an unqualified statement, wouldn’t you?

    And what great timing: these are all hot off the press, updated December 2021! Many of the “probably”s come from pushing back the origins farther than before:

    trews: in old edition “ad. Irish trius, Gael. triubhas n. sing., ad. Eng. trouse” — that last derivation from English seems circular. The revision abandons that and instead tries to trace the Irish/Scottish words further: “perhaps” from Romance “probably” from Germanic. That’s far enough back that almost everything has to be “perhaps”.

    † trouse: in old edition it was already “apparently” from Irish. The influence of French trousses was dismissed in the old edition, but now, with earlier French evidence available, it is allowed into discussion with a “perhaps”.

    trousers: in old edition “perh. directly after drawers”; now “apparently after other words for paired items… probably especially drawers”, which is a slight upgrade in certainty.

  252. David Eddyshaw says


    That low-tone mwàna does indeed look like a contender. Very interesting!

    What actually is the modern Persian name of Zanzibar, btw?

  253. Mwàna looks really good, indeed.

    Farsi Zanzibar is زنگبار, zangibār I guess.

  254. ktschwarz: yes, I was snarky, but at some point the etymology seems unanchored and lost, and depends on too many uncertainties. If the OED wanted to present the full current state of knowledge, sure, but that seems unlike their usual practice.

    all hot off the press

    Yes, trousers should be regularly pressed.

  255. @David Eddyshaw, about diminutives: possibly he meant -ana forms like msichana/mvulana .
    N&H put the former into proto-Sabaki (figure… 12? I think, in the book)

    (But again, “he” is not Krumm, “he” is the K.R. who Kriticized Krumm’s etymology.)
    (likely I should write here “she”, for consistency)
    (and by the way:

    the same enthusiast who informed us that maji “water” was also an Arabic loanword. The interesting paper that drasvi linked mentions him.)

    Did you meant Zawawi? Then you are wrong about “him”.)

  256. I think this دبده‌بان is a typo for the very common word دیده‌بان dīdabān “watchman, lookout, guard, sentinel”.

    Aha!!! I googled it, and saw a few links like Human Righs Watch (in Iranian news*) or a site “Science Watch”. But they were few. A typo explains it.

    *cluster bombs used by Georgia in Ossetia

  257. Re trousers: The anchor is the Irish and Scottish Gaelic words. Those are known, and known to be the source of English trews; and it’s known that the path goes from there through trouse to modern trousers. All of that is certain; it’s just some details that aren’t, e.g. to what extent was trouse a development of trews within English vs. a re-borrowing of the Irish word.

    Getting beyond Irish is tentative, but would you rather have them leave it out? I think “to present the full current state of knowledge” *is* their usual practice — consider boy (revised 2008), what a nightmare! (Now that you mention it, the etymology of trews would probably read more clearly if they first gave a few notes on the dating of the Irish and Scottish words, to indicate that they are secure, then after a paragraph break proceeded with the tentative earlier sources.)

    What they didn’t do, and I think they should, is provide references to the specialist literature. Some entries do, e.g. marzipan, blue, boy.

  258. David Eddyshaw says


    Fair enough. I should not have assumed that crackpots are necessarily male …
    (Though statistically …)

  259. I would rather have the pros and cons on either side of the perhapses spelled out. For example, why would the Irish word not be a borrowing from the proposed Romance source? Are the Scottish and Irish sources impossible to discern apart as the source of the English from phonological reasons?

  260. David Eddyshaw says


    Figure 12 (p290) “Proto-Sabaki lexical innovations” has *mušįkyana “girl”, presumably a compound with “child” as the second element. I can’t see any mvulana, but I may not be looking hard enough.

    I concede that if there had been a stem *bu “big”, it would have been possible in principle to create a compound noun with “child” as the second element, presumably meaning “small big person.” Peter Dinklage, perhaps …

    Bantu languages typically make diminutives by changing the noun class membership of a stem, rather than via derivational suffixes.

    Many Oti-Volta languages do the same, using the diminutive-class suffix -la; though not Kusaal, which uses compounds with bil “little” instead – the sole word which still takes the aforementioned diminutive-class suffix in Kusaal. Thus e.g. nu’ug “hand”, nu’ubil “finger.” The root is the same as in the ubiquitous (except in Bantu) Volta-Congo word for “child” (Kusaal biig), so this is pretty much parallel to compounding with -ana in Bantu.

    [N&H do use the ordinary ogonek (unlike Meeussen.) They should be ashamed. Bitterly ashamed. Depravity! Depravity!]



  261. David Eddyshaw says

    Meeussen has this to say about compounds in his Bantu grammatical reconstructions (p96):

    It is not very frequent for the first stem in the compound to be a nominal, except stem -(j)ána “child”, which is often found with diminutive meaning, in such a way that we can tentatively ascribe the formation to Proto-Bantu, including perhaps the feature that the second member can be just a stem (not preceded by prefix) : **mu-ána(m)búdi̡ “young of goat”.

    In other compounds with nominal as first member, the frequence of a peculiar second member (-ntu “person”, -kádi̡ “woman”, -dúme “man”, -ána “child”, -kúdú “grown up”) has led to a present-day situation in which these stems, without a prefix, are used more or less as suffixes. The lack of particular features in these correspondences prevents us from setting up this type as reconstruction, since it can be easily understood as independent parallel development. Some languages point to bu-dí̡a “vassalage” and bu-kó-o “in-law relationship” as second members.

    Rather than being a nominal, the first member is a verbal stem with final -a (low); this type is so widespread and so particular that it can be said to constitute a firm reconstruction: **(ki)-búmba-kaju̡mbá (“house-moulder”), “mason wasp”; **(mu)-dámba-bi̡día “cook”, **(n)-kíta-midimo “work-doer”, etc. In these examples, the second member is an analog of an object vis à vis the verbal stem (kukítá midimo “to do works”) but in accordance with 9.4 it may also be an analog of a subject : **(mu)-píta-nkíma “sp. tree” [“(where) the monkeys pass”].

    Other types, e.g. connective or connective-stem (5.3), or adjective (4.9) preceded by noun-stem (S-pp-a-S, S-a-S, S-A) are not, for the time being, within reach of reconstruction.

  262. Yes, I saw mvulana elsewhere (e.g. WOLD).

    (mušįkyana < mušįki- "girl" – Venus in Furs by the Velvet Underground. "Shiny shiny shiny boots of leather / whiplash girlchild in the dark…".)

  263. David Eddyshaw says

    In Meeussen’s examples, -búdi̡ “goat” is cognate with Kusaal bʋʋg, which can be shown by comparison with the rest of Oti-Volta to go back to a stem *buʎ- (cf Nawdm burgu “goat”); -kúdú “grown up” looks remarkably like Kusaal kʋdʋg “old”, but I’m pretty sure that one is a false friend: the WOV root basically means “dry out, wither”, not “mature”, and the root-final consonants don’t correspond properly between Bantu and Oti-Volta (though this is still a work in progress, to put it mildly.)

    The bi̡día element of (mu)-dámba-bi̡día “cook”is presumably “food”; the root “eat” is pan-Volta-Congo (Kusaal di.)

  264. Yan makarantar Chomsky
    Ga karatu ba salla!!! (from)

    (music. I have no idea if the original tune is similar to this. It is playing in my head today (I managed to remember it somehow) so I release it)

  265. David Eddyshaw says

    It’s interesting that Bantu compounds are basically back-to-front from the point of view of Oti-Volta, in which noun stems are the usual first component of compounds, and the stem representing an object precedes a deverbal noun, e.g. ninkʋʋd “murderer” (“person-killer”: O kʋʋd nidib “He kills people.”)

    No doubt this is connected typologically with the fact that noun class affixes precede their stems in Bantu, but follow their stems in Oti-Volta.

    They seem to share the phenomenon that two (non-deverbal) human-reference stems in a compound can occur either way round, i.e. they’re basically appositional (Kusaal bina’ab = na’abiig “prince/princess” (child-chief/chief-child.)

  266. GB says the book Xerîb links to is “By Gan Fansina, Jan Vansina” (it’s by Jan Vansina.) That’s really weird.

  267. Yan makarantar Chomsky

    I just saw Chomsky interviewed last night and was startled — he’s grown out his facial hair and looks like the Old Man of the Mountain. This is from July; he looks even bushier now.

  268. My association was:

    1. in Swahili and Hausa words like “scientist” are formed by wana- and yan respectively.
    astronaut – child-sky
    So actually they can be translated as -ist.

    2. I was thinking about -ist, -er, etc., and here people sometimes discuss “Chomskyite”, “Chomskianite” etc.

  269. David Eddyshaw says

    ‘yan is actually the plural of ɗa “son” along with the “genitive linker” -n (cf Usman ɗan Fodio/Usman ɓii Fudiye/ عثمان بن فودي.) It’s used very widely indeed to make descriptive epithets of people (e.g ɗan iska “irresponsible person (m)” (“son of wind”), ‘yar ƙwaya “drug addict (f)” (“daughter of drugs”), and elsewhere as a diminutive form ‘yar takarda “booklet” (“daughter of book.”)

  270. David Eddyshaw says

    Interesting to see all these backwards (from an Oti-Volta-centric standpoint) Swahili compounds (mwanasheria “jurist” = Kusaal sariakat, literally “shari’a-driver”; you “drive” law in Kusaal, and the “child” stem is not used to make this kind of compound, but only diminutives.)

  271. is “driver” a verbal or nominal stem?

  272. David Eddyshaw says

    Nominal. It’s the agent noun of kad “drive (away)”, formed with the derivational suffix -d-, and showing a regular change of *dd -> t. Most (but not all) agent nouns are segmentally (but not tonally) homophonous with the corresponding finite imperfective form (kàt here, whereas the agent noun is kāt.) It’s not altogether clear what the significance of this is; it’s very tempting to suppose that the finite imperfective is derived from a deverbal adjective historically, but the comparative evidence throws up quite a few problems with this neat idea. If it is valid, there must have been a great deal of levelling in the verbal system subsequently (but there is quite a lot of evidence for that in any case.)

    Oti-Volta doesn’t have compounds with verbs. In a word like danuur “beer-drinking”, the second element is a gerund form (just as in English: you can’t say “I beer-drink” in either language.) However, regular gerunds are in fact formed by stem conversion: you basically stick a noun class suffix on the verb stem.

  273. Yes, sorry! wana is “children” too.

    But is “child”-“a person who belongs to” is common African thing?
    Arabic and Hebrew have it, but…

    Is it an African expression that spilled to the Middle East, or is it an areal smaller than the whole continent, or is it a Semitic idea?

  274. Even McDonalds (the chain of public toilets, they have McDelivery®) has it.

  275. In USSR there was a serious issue with public toilets.

    I simply knew all of them in the part of Moscow city center where I studied. And it was like “two kilometers to the east, near metro station such and such”. In the garden near the Kremlin. Etc.
    And they were at least relatively clean in Moscow. And there never ever was toilet paper in any of them.
    On the periphery there are … Trees:)

    Thus the problem was solved only by Capitalism: there are cafés everywhere, all have toilets, all have toilet paper, all are clean. But for a group of my freinds the prototypical toilet café is McDonalds. They just would say: “do you know a McDonalds around?” or “I need to find a McDonalds or something. Urgently”.

  276. @PlasticPaddy

    The word “keeper” in English would be archaising outside of compounds

    Not since the ’80s. (Admittedly a different sense, but you said ‘word’, not ‘meaning’.)

  277. David Eddyshaw says

    I did wonder if the use of “child” in this way is calqued on Arabic. Both Hausa and Swahili show huge Arabic influence. (A Ghanaian Hausa-speaking colleague of mine was very struck by the great number of Swahili words he discovered that he already understood when he went for a training attachment in Kenya.)

    It’s certainly not pan-African: as I said above, it’s not found in Kusaal or its relatives at all, for example. The nearest similar construction uses not “child” but daan “owner of …”, which is similarly applied very broadly, e.g. tieŋ daan “bearded man” (“beard-owner”), pʋpielim daan “righteous person” (“righteousness-owner”) and even bʋgʋsiga daan “cautious/gentle person” (“softly-owner.”) Hausa and Arabic have their own commonly used equivalent “owner” words used in a very similar way, though of course in the opposite order (Hausa mai gemu “bearded man.”)

  278. Y: I would rather have the pros and cons on either side of the perhapses spelled out.

    Now I finally get what you’re saying, thanks. Yes, that would be better.

    Are the Scottish and Irish sources impossible to discern apart as the source of the English from phonological reasons?

    I interpret them as saying not that they’re unsure which was the source, but that both Scottish and Irish sources contributed and eventually coalesced into a single word in English. The citations are predominantly from Scotland, with the first one spelled “trevis”.

  279. David Marjanović says

    And there never ever was toilet paper in any of them.

    In the 1960s, my dad visited Kyiv once. There was toilet paper in the university… it was Українська правда.

  280. And they were at least relatively clean in Moscow.

    Boy, things had changed since 1971. (Except for the lack of toilet paper.)

  281. My freinds fantacized in Kiev about a дворянский туалет (aristocratic toilet) hidden somewhere deep inside Kiev train staition. t was supposed toinclude a library…

  282. David Eddyshaw says

    There is a magical-realist novel just waiting to be written there.
    The Secret Lavatory of Kiev.
    (Possibly the title needs work.)

  283. Boy, things had changed since 1971. (Except for the lack of toilet paper.)

    Relatively. “at least” was mindlessly calqued from “ещё”
    “и в Москве они ешё относительно чистые!” with an implication: “think about other cities where they are not”.

    But there is an important distinction: вокзальный/привокзальный туалет, that is one at a train station, and a normal one. The former type were terrible. In a smaller town it is the only type and they are even more terrible… Or I do not know. What is worse: when no one ever cleans it, or when someone is trying to wash floor in your stall when you are using it? (it was 1997, not USSR, though.).

  284. @ktschwarz: thank you for leading me to clarify my argument.

    both Scottish and Irish sources contributed and eventually coalesced into a single word in English

    That bothers me. When does such coalescence ever occur? Is there, say, phonological evidence for one source, then the other?

  285. The Secret Lavatory of Kiev
    I beg to differ. It’s a Dan Brown mystery novel (or is it??)

  286. David Eddyshaw says

    “Middle-aged, heavily built lavatory cleaner and internationally renowned symbologist Konstantin Chernenko staggered across the linoleum floor, his golden lavatory brush falling from his failing grip …”

    The toilet paper proves to be the crucial link in the mystery. Why was it embossed with the Vatican crest …?

  287. He seriously surpried me. I read it and it turned out to be a simplified version of a well known genre. So can I write a simplified version of a Brazilian soap opera and be even more popular with housewives than the prototype?

    (a snobbish comment, but it is quite an achievement, to write a book where I find nothing new:()

  288. John Cowan says

    If zero equals one, Mrs Lenhart’s phone number is (of course) zero. And one, of course.

    TIL that a (the?) mathematical ring in which the additive and multiplicative identities are the same is called a (the?) zero ring. Which accounts for why calling Mrs. Lenhart never works: there’s nothing but silence at the other end.


    Surely in good time the President of the Republic of Dalriada will appear from time to time in the kilt! (If not in the robes.)

  289. There was toilet paper in the university… it was Українська правда.

    Serendipitously, I just hit the following passage in Pelevin’s Желтая стрела (1993):

    — Да, Григорий, — сказал Иван, выпив и выдохнув, — совсем забыл. Слушай. Предлагают большую партию туалетной бумаги с Саддамом Хусейном. Она после войны осталась, а спрос упал. Очень дешево. Сколько она у вас может стоить?

    — Стоить-то она может много, — сказал Гриша. — Но я тебе, Иван, могу сразу сказать, что заниматься этим нет мазы. Реальный рынок для туалетной бумаги очень маленький — только СВ. Из-за этого даже браться не стоит.

    — А общие и плацкарта? — спросил Иван.

    — В сидячих она вообще никогда не шла, а сейчас из-за инфляции плацкарта тоже на газеты переходит.

    I presume СВ = спальные вагоны.

  290. David Marjanović says

    The former type were terrible.

    Yup, can confirm from experience in the train station of Lubliniec in 2007. There was toilet paper there, but of the worst quality I had yet seen in my sheltered life.

  291. David Eddyshaw says

    Surely in good time the President of the Republic of Dalriada will appear from time to time in the kilt!

    You’ve got to watch out for the Kilt. They are much more ruthless than the Trousers. They don’t care if people think that what they do is not nearly so fitting, and they disdain pusillanimous concealment. (You can’t handle the Truth …)

  292. Congratulate me.

    Only NOW I realized why some ladies in short skirts cross legs when sitting.

    In my defence I can say that they are out of fashion here since 90s.

  293. Bat ladies handle skirts rather masterfully and kilts are not ‘short’.

    Are not like what I meant by a “short skirt” anyway.

  294. @John Cowan: The zero ring is indeed unique (and I referred to it by name above, back in November 2020).

    @drasvi: Under traditional English protocol, it was considered risque for a lady to sit with her legs crossed anywhere above the ankles. Supposedly, by having her legs crossed, she was suggesting (or acknowledging) the crude possibility that someone might be able to peek up her skirt. (Of course, as with many elements of British class-based morality, there is the question of whether this is a post hoc rationalized explanation for something—in this case, the way a lady should sit—that was settled upon as being proper merely because it was a commonplace practice among the aristocracy at a particular point in time.)

  295. David Marjanović says

    In the 1960s, my dad visited Kyiv once.

    But in 1970 he visited Moscow (possibly on the same trip, so “1960s” may be inaccurate), and he confirms everything including the lack of toilet paper…!

    Реальный рынок для туалетной бумаги очень маленький — только СВ.


  296. I think I mentioned that before; one ofthe worst toilets I ever saw was in a student dorm in Zelyonograd, where I spent a couple of days in 92. Shit was everywhere. And that was actually a relatively luxurious dorm, with 3 beds per room (I was told that the standard was 8 beds per room). Toilet paper didn’t figure at all.

  297. Reminded me about correspondence between a freind of my friend and Lacan’s daughter. He stidied Lacan’s psychoanalisis (two freinds of my friends did) rented a flat or a room from her. Someone left his (I won’t add “/her”) poo* in the toilet and did not flush it and others complained and she thought – or maybe others thought and told her – it must be him and wrote him a letter and he responded her with a letter too, explaining that is surely must be somone else, but someone kept doing that…

    какашки. Not sure if it was his or her formulation…

    P.S. well, maybe it is not very good idea to write here such stories about someone identifiable who is not a celebrity and not my friend either. But I will leave it.

  298. David Eddyshaw says

    This thread has uncannily converged with


    All threads are One …

  299. the fact that it differs in detail across the languages suggests exactly that, viz that it did happen independently many times

    The devil’s in the details. What exactly might or might not have happened several times independently? Palatalization plus assibilation of *k, *g and/or *t, *d before front vowels, especially /i/, is a bread-and-butter innovation, but the fact that it is in Bantu basically always both, and also basically always just *į but not the lower front vowels, looks already more suspicious. Palatalization of *p, *b to anything more advanced than /pʲ bʲ/ seems almost vanishingly rare elsewhere and I don’t buy that this would have happened there more than a handful of times independently. Same goes for any “palatalization” before *u. “Palatalization” of *pj *bj, as in Mandarin or the “Rule of Change” in French (Lt. cambiāre > Fr. changer) seems a bit more common but this is usually really glide fortition + stop loss.

    On the other hand, once some sort of initially fricated reflexes like “*kų” > *kfu or “*pį” > *pçi had arisen, later development of various simplified reflexes (maybe even some back-developments to stops that end up looking archaic) could be easily independent. Also doesn’t mean that I’m proposing grouping all varieties with a similar pattern in a strictly a common subgroup necessarily, just as coming from common areals within early Common Bantu. I suppose you’re still right that this also weakens the initial motivation to set up fricative vowels, but then I’ve really not tested how far this might stretch exactly, when there’s differences in the details of spirantization even within e.g. Southern Bantu. (You linked Jansen and another paper of his notes the reflexes of “*dų” as Nguni /vu/, Venda /bvu/, but Tsonga /dzu/ and Sotho /du/ ≠ /dʊ/, so necessitating something like *dvu as the common proto-form even just for these.)

    Tone-bearing fricatives: a classic example is Nuosu, most other reports I’ve seen are roughly from the same region, e.g. Ersu.

  300. @drasvi: That sounds awfully bourgeois for a so-called Maoist.

  301. David Eddyshaw says


    I was agreeing with you, not disagreeing (though I see I didn’t make that very clear …)

    The “Bantu spirantisation” change is, as a whole, indeed cross-linguistically very unusual, even if individual parts of it may be less remarkable; consequently, it seems unlikely that it could have taken place several times independently if it were purely a matter of simple consonants being affected merely by the height of the following vowels, without some preceding, shared, groundwork like affrication (or something.) Unless one attributes it to Sapir-style “drift” … or (more fruitfully) could adduce evidence that it really was a matter of diffusion.

    One difficulty that occurs to me is that whatever these “preliminary” changes may have been, they presumably had to be capable of disappearing without subsequent trace in languages which don’t have the spirantisation, without, for example, any effect on the original, preaffricativisation (or whatever) vowel qualities etc. This, too, would presumably have had to have happened several times independently.

    If one posits that the languages that don’t have the spirantisation never actually had the preliminary changes in the first place, what remains seems to be really a proposal for what the intermediate steps of the spirantisation sound changes were in just those languages where it occurred. For that to be of further significance, you’d probably want evidence that the spirantising languages formed an actual Bantu subgroup, recognisable by other shared innovations.

  302. David Eddyshaw says

    It would probably help if we had a better handle on the vowel system of Proto-Volta-Congo. Proto-Oti-Volta seems to have had a (West-Africa-standard) seven-vowel system, and the qualities seem (very broadly) to match in POV/PB cognates, but there are a lot of unresolved problems with the Oti-Volta material, and not a huge number of secure POV/PB cognates to work with. Even when forms look pretty certainly cognate, there are often awkward rough corners: Kusaal kɔnbir /kɔ̃bɪɾ/ “bone” must surely have something to do with Swahili mfupa “bone” (from *-ku̡pa) for example, but while Kusaal /ɔ/ often does go back to earlier /ʊ/, it doesn’t go back to /u/, and the Oti-Volta cognates disobligingly have /a/, e.g. Moba kpabl. And what’s with the nasalised vowel?

    There’s no reason to think that Proto-OV had syllable-initial clusters like Cw- or Cy, but on the other hand it certainly had many more initial consonants than PB seems to have done, with palatals contrasting with alveolars, and labiovelars contrasting with velars. Oti-Volta has all that as well as a seven-vowel system, but there are some places where these consonants are in complementary distributions (there’s no evidence for a velar/labiovelar contrast before rounded vowels, for example.)

  303. @drasvi: That sounds awfully bourgeois for a so-called Maoist.

    “…This occurred after she handed out course credit to someone she met on a bus, and subsequently publicly declared in a radio interview that the university is a capitalist institution,[2] and that she would do everything she could to make it run as badly as possible. After this, she was demoted by the French education department to a lycée teacher.”

    Wow (about demotion in rank – is it an army?)

  304. Brett, I did not know that! What I have in mind is miniskirts: in this case, I think can be hard not to notice whatever they want (not) to notice, if the wearer is not careful. I am not sure. It never happened to me (as a viewer). But it is unsurprising indeed if excessive precaution is immediately proclaimed inappropriate by people who love the concept of “inappropriate”.

    Also “women’s underwear” is not what a yougn male Russian really wants to see (same for British ladies and kilts, and the famous absence of underwear, I assume?) so the actual motivation (again, Russia) must be more about looking better.
    Honestly, I remember exactly one such episode.

    I of course had seen all sorts of ladies in public transportation, but it is is different. My “mathematical” friends usually dressed the same way as they would do when hiking:) They could wear something beautiful or not, modest or reveling, under certain (rare, if we do not mean romantic situations, but sometimes it is acceptable) circumstances they could even wear nothing, but never something “fashionable”. Philological girls are a very different story, but would not put on a miniskirt for university. It was when a guy with famously Brittonic name Iwan was treating us with Breton crêpes. The girl later would become a rock star (J1M once mentioned her) and my first thought was “wow, just like in movies!”. I had seen this gesture and ladies sitting this way in films. I just took it holistically as a gesture and dress from films.

  305. Зензубель!!!!!!!!!!!!!!!!!!!

    Another ginger word! < Simshobel

  306. PlasticPaddy says

    In France a university prof (prof d’uni, enseignant-chercheur) is a civil servant (fonctionnaire).
    Normally the competent authority for disciplinary proceedings is the secton disciplinaire at the uni where the offences occurred or the principal place of work of the prof. But
    S’il existe une raison objective de mettre en doute l’impartialité de la section disciplinaire initialement saisie dans son ensemble, l’examen des poursuites peut être attribué à la section disciplinaire d’un autre établissement.
    so another authority can “prorogue” the investigation.

    The possible penalties are:
    1 Le blâme ;
    2 Le retard à l’avancement d’échelon pour une durée de deux ans au maximum ;
    3 L’abaissement d’échelon ;
    4 L’interdiction d’accéder à une classe, grade ou corps supérieurs pendant une période de deux ans au maximum ;
    5 L’interdiction d’exercer toutes fonctions d’enseignement ou de recherche ou certaines d’entre elles dans l’établissement ou dans tout établissement public d’enseignement supérieur pendant cinq ans au maximum, avec privation de la moitié ou de la totalité du traitement ;
    6 La mise à la retraite d’office ;
    7 La révocation.

    1. verbal sanction
    2. minor promotion bar (change in echelon) for up to 2 years
    3. demotion in echelon (I think this was the case for your friend but see 5)
    4. major promotion bar (change in class, grade or “corps de service”, as well as echelon) for up to two years
    5. Partial or full bar from teaching or research in the uni where the offence occurred or in all third-level institutes
    6 . Enforced retirement
    7. Revocation (I don’t know if this applies to the academic title or to the related status allowing the holder to teach/do research at a uni).


  307. for your friend

    The lady who corresponded with my friend’s friend about poo in the toilet:)

    Thank you. I knew the French system is rather peculiar, but I did not know that it is so up to demotion.
    Now it clear why they did not like the professor and why the professor did not like them.

  308. And the craziest of ginger words:

    On the island of Corfu, Greece, a traditional drink called τσιτσιμπύρα (tsitsibira), a type of ginger beer, is made.”

    ginger beer. Somehow it managed to obtain that J-G-B-R sound by adding ‘beer”!!!!

  309. “J-G-B-R” “J-N-J-B-R” I mean:(

    By ginger words I mean words with this shape: 𑀲𑀺𑀁𑀕𑀺𑀯𑁂𑀭#Sauraseni_Prakrit.

    There is Russian zinziver 1. “mallow” 2. “tit” I do not know why mallow. Why tits borrowed the word is clear: it is several Russian small-bird (it is those tits, not the tits that women have) onomatopoeias. It consists of them wholly.

    -i- is Russian (and not only) for high-pitched sounds. Russian mosquitos say -eeeeee-, and our ee can be very tense and front.
    -z- : dzʲinʲ is the sound of a wine glass (dictionary gives variant dzʲinʲdzʲiver and Ukrainian дзiндзiвер for malva).
    – for three syllables compare “chickadee”.
    -ver- : vorkovatʲ “to coo”, vorchatʲ “to grumble”, murmur, Barbarian etc., -e- again gives high-pitched sound.

  310. That meaning of tit is not really part of my normal American English—although titmouse is, since it refers to a common North American bird. Of course, tit is quite familiar from British sources, most notably The Mikado. (My father called this the best-plot-integrated song in all of musical theater, but that was not a genre he was really very knowledgeable about. Another skillful rendition, by a dog and bird, is also available; it also features an extra “dicky bird” gag that I had forgotten about.)

  311. That meaning of tit is not really part of my normal American English

    Really? I would have thought it was Basic American Slang. I guess it’s just another sign of my being behind the times.

  312. @languagehat: Slang, as in, “What a little tit”? That sounds affectedly British to me. The mammary meaning seems to have squeezed out any other meanings of tit in American slang.

    Which all reminds me, in turn, of this.

  313. Oh, I completely misunderstood you — I thought you meant the mammary meaning.

  314. It would probably help if we had a better handle on the vowel system of Proto-Volta-Congo

    For this in turn it would probably help if there was some kind of a stock of even 100ish basic etymologies well-distributed across Volta–(Atlantic–Niger–…)Congo. A goal that seems far away, but I don’t known, maybe it could be not that far; perhaps someone just needs to start collating and comparing all the subgroup lexicons to put together something of a first-pass etymological dictionary of the family (and hopefully not go full Ehret in the process); since this does not strictly require subgroup reconstructions.

  315. Brett, thanks for that Rowlf and Sam clip, it is really funny. Only I hear somehow “tick” for “tit”. John Reed is usually very good, but in this song he is too stilted for my taste.

  316. jack morava says

    cf pair de Tits


    mostly verbal, hard to find any written examples…

  317. @jack morava: Yeah, I’ve heard “Tits pair” for that before. It’s not the easiest thing to search for though.

  318. Y: When does such coalescence ever occur? Is there, say, phonological evidence for one source, then the other?

    There are zillions of words that entered English in medieval/early modern times labeled as “Partly a borrowing from French. Partly a borrowing from Latin.” That may not be what you’re asking, though, since in many cases the French and Latin are indistinguishable (e.g. abjuration < French abjuration; Latin abiuration-, abiuratio) and in many others the form is only French but the meaning is also influenced by Latin (e.g. ability < French abilitie; Latin habilitāt-, habilitās).

    An example of borrowings through separate paths that eventually converged: Eskimo is “Partly a borrowing from Spanish. Partly a borrowing from French. Partly a borrowing from Cree.” The word was originally picked up from Algonquian speakers on the Labrador coast by Basque whalers, who brought it back to France and Spain; Richard Hakluyt must have learned “Esquimawes” in France before he published his book in 1584. A century or two later, the Hudson’s Bay Company was trading with the Cree, picking up their cognate word, and using it for the people to the north; the OED attributes the “β. forms” (Ehuskemay, Eusquemay, Uskemaw, etc.) to this Cree influence. But eventually, the spelling standardized on the French-influenced Esquimau and finally Eskimo.

    Thousands of examples can be found by searching the OED for “Of multiple origins”, mostly phonologically very close: aardvark < Dutch and Afrikaans; abatagati < Xhosa and Zulu; Abba < Latin, Coptic, Geez, Hebrew, and Aramaic (whoa, is that a record?).

    But there are probably some others that are as complicated as Eskimo. One possibility: dowf (Scottish, Irish English, and English regional (northern)), ‘Dull, flat; lacking in spirit (etc.)’, origin: “Probably partly a borrowing from early Scandinavian. Probably partly a borrowing from Dutch.” (Again with the Dutch-Scottish connection!) More detail: “The variety of modern Scots forms and pronunciations reflects the word’s multiple origins. Forms indicating a diphthong… reflect the early Scandinavian word…, while forms indicating a front rounded vowel or its reflex … reflect Middle Dutch …” One of many etymologies that are *way* the hell more thorough in the Third Edition than the old ones!

  319. PlasticPaddy says

    Re Dutch origin for dowf, the word dof is now found in dialect and seems to have meant dull as applied to sounds made e.g., by striking something (cf. German dumpf).

  320. That the OED uses the “multiple origins” term indicates to me not an actual phenomenon, but an older practice of linguistic imprecision. There are separate phenomena here. In words like abjuration, the phonological form alone cannot be used to choose between two different sources. In words like Eskimo, different attestations from different times and places have different sources, but are under the same headword in the dictionary (in the case of Eskimo, the etymologies are, again, sloppy: there is no explanation as to why the intermediate language is Spanish and not French or Basque.)
    In the case of trews which we discussed above, it sounds like the etymologists suggest that both the Irish and the Gaelic words had some part in shaping the English. Does that mean some English speakers were exposed to both Celtic sources? Did the English take certain phonological elements from each language and mixed them up? That is what it sounds like they are saying, and I am skeptical that that happens often if at all. If that is what they mean they should justify it rigorously.

  321. There are zillions of words that entered English in medieval/early modern times labeled as “Partly a borrowing from French. Partly a borrowing from Latin.”

    My go-to example is zero, simultaneously from Italian, French and Spanish.

  322. Updated in 2018:

    Etymology: Partly (i) < Spanish †zero (1512; now cero), partly (ii) < French zéro (1485 in Middle French; 1690 as adjective), and partly (iii) < their etymon Italian zero (late 14th cent.), all in sense ‘number 0’ < post-classical Latin cifra, cifre, cyfre (early 12th cent.), zephirum (1202 in Leonardo da Pisa) < Arabic ṣifr cipher n.

    That seems like a reasonable formulation to me.

  323. But what does it mean? Does “simultaneously” mean people heard equal amounts of the three languages? What did people hear and repeat? Was the z pronounced distinctly in the three languages as it is now? Or was the form first transmitted by writing?

  324. John Emerson says
  325. Lesson 15. Ordering of adjectives.

    1. a pair of great Russian tits [pic]
    2. a pair of Russian great tits [pic]

    a) Once a pair of [ ] great [ ] tits considered my room* for nesting. I relocated to another room.
    b) Once a pair of [ ] great [ ] tits considered the same and I did not relocate to another room.
    Put the adjectives ‘great’ and ‘Russian’ in the correct order.

    * a closet, more precisely

  326. Y: in the case of Eskimo, the etymologies are, again, sloppy: there is no explanation as to why the intermediate language is Spanish and not French or Basque.

    (I think you mean “Spanish and French but not Basque”?) The evidence for French is obvious, Hakluyt spent time in Paris. I would guess the evidence for Spanish and against Basque has to do with the amount of known contact, commercial and diplomatic, of English with Spanish and not Basque. That’s where they should have given a reference to some source with full detail.

    trews … both the Irish and the Gaelic words had some part in shaping the English. Does that mean some English speakers were exposed to both Celtic sources? Did the English take certain phonological elements from each language and mixed them up?

    I take it to mean that some English speakers were exposed to Irish, some to Scottish Gaelic, and later others heard both versions and *considered them to be the same English word*, perhaps spoken in different accents. That’s also what I think happened with the French and Cree transmissions of Eskimo. But that’s just my amateur level of understanding. I need to read Philip Durkin’s book.

    Alon: zero, simultaneously from Italian, French and Spanish
    Y: Does “simultaneously” mean people heard equal amounts of the three languages?

    I read Alon as saying that the statements “zero is from Italian”, “zero is from French”, and “zero is from Spanish” are all true, not necessarily that they were all borrowed at the same time. Come on, nobody is saying silly things like “people heard equal amounts”. I think it means that some English speakers learned it from Italian, some from French, and some from Spanish, and others then understood them all to be talking about the same thing; note the spelling was the same in all four languages except for an accent in French. The OED’s first cite is a translation from Spanish and the second says it’s a French word; I assume the attribution to Italian is based on some evidence of English-Italian contact, but here as well it would help to have a reference.

    Are you satisfied with the OED’s etymology for zebra as partly from Italian and partly from Portuguese? Their citations include works translated from both Italian and Portuguese.

  327. I think it means that some English speakers learned it from Italian, some from French, and some from Spanish, and others then understood them all to be talking about the same thing

    Exactly — modulo the fact that a fair bit of this dissemination probably took place directly in writing rather than speech, through works like Filippo Calandri’s De arithmetica opusculum

  328. David Eddyshaw says

    It’s often difficult to determine the immediate source of the (ultimately) Arabic loanwords in Kusaal. It’s likely to be Hausa for the most part, but the evidence for this is generally only that Hausa is the source of most identifiable loanwords that are not of Arabic origin. To be able to tag a different language as the proximate origin for certain, you need evidence of some sort of intermediate stage only attested in that language (as with Agolle Kusaal maliak, Toende malɛk, Mooré malɛka “angel”*), which is not usually possible. Even the fact that word doesn’t occur (or doesn’t occur in that particular form) in “standard” Hausa doesn’t prove that it couldn’t have come from a western regional dialect (most of which are not very thoroughly documented.)

    But if I knew more about the relevant languages, I’d probably be able to identify a lot more of these Arabic words as having come via Songhay or even Mande (especially Dyula.)

    * This particular case gains in plausibility from the fact that the first Christian missionaries to the Kusaasi are known to have used Mooré, before they learnt (Toende) Kusaal. (Though angels were already known of in West Africa through Islam, of course.)

  329. OED on festa ‘a religious or other festival’ (via kelmakelma.com):

    Origin: Of multiple origins. Partly a borrowing from Italian. Partly a borrowing from Portuguese. Partly a borrowing from Catalan. Partly a borrowing from Maltese. Etymons: Italian festa; Portuguese festa; Catalan festa; Maltese festa.

    Etymology: Originally < Italian festa religious feast day (late 12th cent.), secular festival (mid 13th cent.) < classical Latin fēsta (see feast n.).

    In later use also < other Romance (or Romance-influenced) languages, as Portuguese festa (1154), Catalan festa (14th cent.), and Maltese festa (itself < Italian), all in the sense ‘religious feast day, festival’, all < classical Latin fēsta (see feast n.).

    A citation exemplifying borrowing from Maltese:

    1961 Malta Year Bk. 260 A curious example of the mixture of religious ceremony and rustic revels, the festa plays a very important part in their social and economic life.

  330. Abba < Latin, Coptic, Geez, Hebrew, and Aramaic (whoa, is that a record?)

    Nope! It’s exceeded by yogurt: “A borrowing from Turkish. Partly also a borrowing from Italian … French … Dutch … German … Greek.” (Where the Second Edition had only “[a. Turkish yōghurt.]”) Etymology continues:

    Etymons: Turkish yoġurt, yoğurt; Italian iogurt; French yogourt, yaourt; Dutch yoghurt, yogourt; German Joghurt; Greek γιαούρτι.

    In early use via forms in other European languages …

    Ottoman Turkish -ġ- (corresponding to Turkish -ğ- ) originally stood for a voiced velar fricative consonant and was variously transliterated in European languages by -g- , -gh- , and -h- : compare α. and β. forms. [ α: spelled with -gh- or -h-, β: with -g- ]

    The γ. forms (and ultimately also the δ. forms) [spelled with no g or h] reflect a variant pronunciation with the fricative elided, which is now standard in Turkish.

    I think they’re saying that by borrowing the word while the fricative /ɣ/ was still pronounced (in the 1400s-1700s), and adapting it to their own phonology, European languages preserved a trace of it while it weakened and disappeared in Turkish. Is that about right?

  331. a voiced velar fricative consonant

    Was it not a uvular one? In Tatar, it’s a uvular fricative before back vowels and a velar plosive before front vowels.

  332. David Marjanović says

    Is that about right?

    Yes, except it was uvular (and still is in unspecified eastern Anatolian dialects).

    The loss of this is part of the same phenomenon as the loss of the [q]-[k] allophony (preserved in Tatar apart from loans) and of [q] altogether. I’m not sure when that happened. Some accents have a new front/back vowel allophony for /k/, namely [k]-[kʲ], shared with most Standard Greek accents.

  333. Thanks! Wikipedia says the letter ğ could represent either a velar or uvular fricative depending on the language: velar in Turkish (Ottoman, or dialects where it is still pronounced) and Azerbaijani, uvular in Tatar and Kazakh. In Ottoman Turkish it was written with the letter ghayn غ , which in the various languages that use it could be velar or uvular.

    Anyway, the point is: were the Western Europeans who first wrote down words for yogurt actually hearing a fricative in it and trying to write it in the Latin alphabet, rather than (or in addition to) transliterating the Ottoman written form and then giving it a spelling-pronunciation? Does that seem right?

    Incidentally, the OED’s only mention of a uvular fricative (outside of quotations) is in the etymology of razzia ‘Esp. in North Africa: a hostile or aggressive incursion, foray, or raid’, from Maghribi Arabic both directly and via French. Under etymology, they write:

    The initial ghr- in the α. forms (which are perhaps directly < Arabic) and the initial r- in the β. forms (which are < French) are attempts to represent the pronunciation of the Arabic voiced velar fricative ġ (= /ɣ/), which in French came to be spelt as r because it was perceived as closest to the standard French voiced uvular fricative /#/.

    Unfortunately they’ve screwed up the IPA symbol: that /#/ should be /ʁ/.

  334. David Marjanović says

    Wikipedia says the letter ğ could represent either a velar or uvular fricative depending on the language:

    Or this is simply copied from sources that failed to distinguish velar from uvular fricatives – which is amazingly common. Apparently, if you aren’t used to either, it’s hard to hear the difference…?

    What little Arabic I’ve heard all had uvular fricatives, but I’ve also been told the most prestigious accents use velar ones instead.

  335. David Eddyshaw says

    failed to distinguish velar from uvular fricatives

    IIRC there’s an actual word for this in Inuit languages, with the sort of associations that baby speech or lisping has for Anglophones. Kinda cute …

  336. DM: OK, what’s your reference for Ottoman Turkish ghayn representing a uvular fricative, and not velar? (It wouldn’t be terribly surprising to find it was both, or variable.)

  337. ktschwarz: The change was post-latinisation. That’s the reason orthograohic ğ exists in Turkish in the first place. Happy new year!

  338. David Marjanović says

    …as usual: I once read that somewhere and can’t find it this year. 🙁

    However, the en:WP article on Turkish phonology contains this sentence: “[q] is an allophone of /k/ before back vowels /a, ɯ, o, u/ in many dialects in eastern and southeastern Turkey, including Hatay dialect.” While no such thing is mentioned for ğ, the continued existence of [q] makes [ʁ] for ğ in something like the same area more plausible.

  339. @V: The change was post-latinisation. That’s the reason orthograohic ğ exists in Turkish in the first place.

    That indicates that it became standard post-latinization, but there may have been at least some dialects that dropped it earlier, since the spelling yaourt is recorded in English, French, and Dutch as far back as the late 1700s.

    @David M, I would think the continued existence of [q] as an allophone of /k/ in some dialects makes [ʁ] for ğ more likely *as an allophone or dialect variant*, it doesn’t mean it was only ever [ʁ] and not [ɣ] in all Turkish at all times. From a previous discussion of Turkish phonology here:

    Annette Pickles says
    … ğ, representing a segment that in the Wikipedia article linked to above was conventionally notated /ɰ/ (corresponding in Turkish dialects to a sound realized as a voiced velar/uvular fricative in back vowel environments).

    That suggests the OED’s “velar fricative” was not wrong, just simplified; they may well have meant “velar or alternately uvular at some times in some dialects”.

    Turkish hasn’t ever distinguished velar from uvular fricatives as different phonemes, has it? *Are* there many languages that distinguish those as different phonemes? There’s Halkomelem, which according to Wikipedia distinguishes not only /x/ and /χ/, but also labialized versions of both of them. Probably other Pacific Northwest languages; marie-lucie mentioned working with some.

    The Turkish Language Reform: A Catastrophic Success mentions one writer in 1939 who “employed ğ for ghayn to show the pronunciation of French words: programme as pğogğam, and carte postale as kağt postâl.” That’s presumably /ʁ/ in French, but the writer may not necessarily have meant ğ was exactly /ʁ/, he may have meant it was the closest thing in Turkish to /ʁ/.

  340. *Are* there many languages that distinguish those as different phonemes?

    Chechen, for one (not to mention that they come in aspirate, ejective, and geminate varieties):


  341. According to those pages, Chechen and Uzbek distinguish velar from uvular *stops*, but not *fricatives*, which is what I was asking about. Distinguishing the fricatives seems to be less common than distinguishing the stops.

  342. David Marjanović says

    Turkish hasn’t ever distinguished velar from uvular fricatives as different phonemes, has it?


    I wonder, however, when ğ started occurring in front-vowel words.

    *Are* there many languages that distinguish those as different phonemes?

    No, this seems to be limited to languages with very large consonant systems.

  343. David Eddyshaw says

    Eskimo languages distinguish velar from uvular fricatives (and have relatively small consonant systems.)

  344. David Marjanović says

    Oh, yes. Aleut even does that for both the voiceless and the voiced ones.

  345. Smallest consonant inventory I know of with a [ɣ] : [ʁ] contrast is Northern Selkup, where both are allophones of the stops /k/, /q/ respectively. There are also some classic examples of languages with /k x χ/ but no **/q/, e.g. Seri; and by some descriptions northern Pashto. (The front pair of fricatives in the latter are traditionally called “palatal” x̆ ɣ̆ and then sometimes transcribed as /ç ʝ/, but they come from retroflex /ʂ ʐ/ and the latter dialectally goes to /g/, so I think this is confusion from old phonetic terminology where all velars were called “palatals” vs. uvulars “velars”.)

    Chukotkan is interesting in having a robust /k/ : /q/ contrast but having as their fricative counterparts mostly instead /ɣ/ : /ʕ/ (the latter often further > /ʔ/).

    Anyway insofar as Turkish -ǧ- is due to medial lenition (quite natively so also in front-vocalic words; siğil ‘eye’, yeğen ‘nephew’ etc.) I would agree with ktschwarz: I don’t think there is reason to think it has ever been generally uvular, in contrast to those Turkic languages that get their “ɣ” from general lenition of back-vocalic *g i.e. the highly unstable [ɢ].

  346. David Marjanović says

    Oh, the native lenition process had escaped me entirely. So, yes, definitely never uvular in front-vowel words.

  347. Adyghe: that makes sense, Caucasian, lots of consonants. The -gh- in the language name is pronounced [ɣ] in the language itself, but [g] in Russian and English.

    And Adyghe distinguishes velar from uvular fricatives in spelling: г vs. гъ and х vs. хъ in Cyrillic, g vs. ğ and x vs. xh in Latin. But what about plain /g/? Doesn’t exist, except in loanwords. Instead they have labialized [gʷ], spelled гу in Cyrillic, gu in Latin.

    Eskimo languages distinguish velar from uvular fricatives: though their [ɣ] is an allophone of /g/, not distinguished from it in Latin spelling or syllabics, and may not be found everywhere. According to Wikipedia: “/g/ is realized as voiced velar fricative [ɣ] in Siglitun. In other dialects, this realization may also be found between vowels or vowels and approximants.” That seems to imply that some speakers *don’t* realize it as [ɣ], but only [g].

    Aleut: yes, and the Latin spelling also distinguishes all four: g, x are velar voiced and unvoiced fricatives; the same letters with circumflex are uvular. (ĝ is a character in Unicode, but x̂ isn’t, and has to be composed, so it looks bad in some fonts.) Like Adyghe, Aleut doesn’t seem to have a plain /g/, at least not as a separate phoneme.

    Aleut was also previously written in a version of the pre-1918 Cyrillic alphabet, with inverted breve added to г and х for the uvular sounds; the modern Cyrillic equivalents ӷ and ӽ are also used in Nivkh to distinguish uvular from velar consonants.

    Northern Selkup, where both are allophones of the stops /k/, /q/ : for all I know there could be lots of languages where the Wikipedia chart just shows /k/, /q/ but they’re actually realized as fricatives in some environments.

  348. John Cowan says

    ĝ is a character in Unicode, but x̂ isn’t

    Because Esperanto had a pre-Unicode encoding (ISO 8859-3[*]) and therefore its characters fell under the “must have a 1:1 mapping to Unicode” rule, whereas Aleut did not.

    [*] Also intended for Maltese and Turkish, but Turkish eventually got its own 8859-9 encoding that was less incompatible with 8859-1.

  349. David Eddyshaw says

    A kind relative got me Michael Meeuwis’ Grammatical Overview of Lingala for Christmas. It’s actually pretty good; he treats the Kinshasa dialect exclusively, and makes no mention of a [ɟ] realisation of /z/ at all. It also seems that the seven-vowel system has definitively collapsed into five in Cityspeak: apparently distinguishing e/ɛ and o/ɔ is an eminently mockable feature of up-country hick Lingala.

    Meeuwis has also got a more lucid account of the history of the language, and indeed of its name, which (though now sanctified by norma loquendi) seems to have originated as a mistake several times over:

    Lingala derives from a creole spoken in a place called Bangala. So far, so good. However, “Bangala”, though of unknown origin, seems at least definitely not to be an ethnonym, so there is no warrant for segmenting it Ba-Ngala “Ngala people.”

    The name Lingala was invented (on the back of this misanalysis) by missionaries, who used the term exclusively for their invented “rebantuized” version of the Bangala creole which they created for their own purposes. The Li- is also a mistake; the language itself uses Ki- just as [Ki]Swahili does, for all other language names.

    The missionaries had some limited success in imposing their “Lingala” as a spoken language in areas where schooling was under their direct control. Everywhere else (including Léopoldville/Kinshasa) people just happily carried on speaking the Bangala creole. However, “Lingala” was imposed as a church and literary language, though apparently it is largely incomprehensible to speakers of the creole who have not learnt it separately.

    Eventually, because “Lingala” was supposed to be the “purified” version of this selfsame creole, and hence “the same language”, people eventually just called the Bangala creole “Lingala” too. And there we are …

  350. Annales Aequatoria, 22 (2001), 327-421, Grammaire et vocabulaire du Lingala ou langue du Haut-Congo by De Boeck (1904, with a preface by Meeuwis’ and an introduction of De Boeck, both in Freinch. Basenji is also mentioned in the introduction)

  351. David Eddyshaw says

    Meeuwis, in his Grammatical Overview, talks about De Boeck quite a bit, essentially characterising him as the inventor, not only of the name “Lingala”, but of the “rebantuized” artificial language “Lingala” itself.

    In a way, you have to admire this (quite apart from the sheer amount of work and commitment it all must have entailed, which is no small thing at all.) Regrettable though it is that De Boeck could not appreciate the unique and considerable beauty of the Bangala creole in its own right, that is hardly surprising given the prejudices of his time about the superiority of flexion and fusion (or at least agglutination) over isolating structures*; and at least he valued the complexity of mainstream Bantu languages, and did not regard them as unworthy of higher purposes.

    * Sapir, in Language, twenty years later, has the superb footnote:

    One celebrated American writer on culture and language delivered himself of the dictum that, estimable as the speakers of, agglutinative languages might be, it was nevertheless a crime for an inflecting woman to marry an agglutinating man. Tremendous spiritual values were evidently at stake. Champions of the “inflective” languages are wont to glory in the very irrationalities of Latin and Greek, except when it suits them to emphasize their profoundly “logical” character. Yet the sober logic of Turkish or Chinese leaves them cold. The glorious irrationalities and formal complexities of many “savage” languages they have no stomach for. Sentimentalists are difficult people.

    Preach it, Ed!
    This bias was persistent; It’s behind Guthrie’s determined refusal to accept that his beloved Bantu could possibly be related to those horrid West African languages like Yoruba (he might have been a bit less hostile if he’d known any Fulfulde.) As is (again) sadly of its time, the bias tended to express itself in bogus “racial” terms, too. Whatever you say about Greenberg (and I’ve often said a lot) he rescued us definitively from all that crap.

  352. D’autre part je m’aperçus qu’on n’avait pas appelé sans motif cette langue universelle le ‘Bangala” ou, comme disent les noirs, le “Lingala”., ibidem, 3. And:
    Son emploi du glossonyme ‘lingala’ se situe également au niveau de ce recours aux langues locales de Nouvelle-Anvers, dans lesquelles les glossonymes prennent le préfix li- et dont les locuteurs ont dû utiliser cette forme pour renvoyer au bangala, ne fût-ce que de façon non-systématique. (p. 333)

  353. The linguistic features of Bangala before Lingala: The pidginization of Bobangi in the 1880s and 1890s
    an open access paper by Michael Meeuwis: https://biblio.ugent.be/publication/8644523/file/8645617

    I am still confused: he calls using Bangala as an ethnonym “erroneous” but does not explain why:

    The European officers had chosen the name “Bangala-Station” for their new post on the basis of what during his first passage in 1877 the explorer H. M. Stanley had thought was a local ethnic population called the “Bangala” (Stanley 1878:287; Mumbanza 1973; Hulstaert 1974; Burssens 1954; 1958; Samarin 1989a, Samarin 1989b; Mbulamoko 1991).

    Because of its tied link with the station, the variety soon, i.e. from the late 1880s onwards, came to be called after it, i.e. “Bangala”. Thus, there was first, i.e. from 1877, “Bangala” as an imagined ethnonym, i.e. the name Stanley and Europeans after him erroneously but determinedly continued using to refer to the populations living around the northwestern bend of the river; secondly there was “Bangala-Station”, the name of the station the colonial occupiers founded in 1884 and named after the alleged ethnic group living there; and now, in this second half of the 1880s, “Bangala” also became a language name, referring to the Bobangi pidgin imported to the station and imposed there.

  354. Vocabulaire pratique Français, Anglais, Zanzibarite (Swahili), Fiote, Kibangi-Irébou, Mongo, Bangalas
    par le Leutenant Charles Lemaire du 2e Régiment d’artillerie, 1894
    Fr. urine En. water …

  355. David Eddyshaw says

    Lingala má(y)i is actually just “water”, but Guthrie has nɛna “déféquer”, so I suppose nɛna mái could be a way of expressing the concept. Guthrie has an actual verb súba for it, though, and masápo (or masúba) for the product of the process.

    (I expect the missionaries introduced this hifalutin vocabulary as part of making LIngala a fit language for evangelisation and education through rebantuization.)

  356. Yes, I thought the leutenant’s Victorian informants are responsible for this, but then I noticed mayi.

  357. For two days I have been trying (in my free time) to check the references in Meeuwis and Samarin in hope to come across a paper or better two that discusses the etymology of “Bangala” (and hopefully “lingala”).

    They both use words “erroneously” (M), “arose out of some onomastic misunderstanding” (S) etc.
    But elsewhere (Lingala, 1998, a 50-page predecessor of the 202-page predecessor of 333-page David’s Christmas gift) M. has:

    Hulstaert (1974: 177), however, maintains that the term was already known and used by the Africans in the region before colonization as the name of an extant ethnic group, while Mumbanza (1973: 473), like Tanghe (1930: 343-344) and Samarin (1989: 80-81), argues that the terrn was not authentiс but inadvertently ‘designed’ by Stanley when he misunderstood local populations referring to “the people from the town of Mankanza”.

    And S has a footnote: 7. The origin and history of the name has come in for a considerable amount of discussion, reviewed by Mumbanza mwa Bawele 1973, whose account is subjected to a critique by Hulstaert 1974. See also Mumbanza mwa Bawele 1974. The issue is too complex for simple summary. Suffice it to say that the name owes its existence to some kind of deformation on the part of whites—of what is uncertain—and that it came to be used for many different ethnic groups. Whether or not there ever was a ‘Bangala’ ethnicity is still an open question. (The Black Man’s Burden, 1989 p 81).

    Thus the discussion of “Bangala” is in:

    Gustaaf Hulstaert,
    A propos des Bangala. Zaïre-Afrique, 1974p n° 83, 173-185. (Au sujet de l’article de Mumbanza mwa Bawele “Y a-t-il des bangala? Origine et extension du terme”, l’auteur apporte quelques informations complémentaires et attire l’attention sur quelques points qu’il faudrait élucider.)
    Jérome-Émilien Mumbanza mwa Bawele na Nyabakombi Ensobato,
    Y’a-t-il des Bangala ? Origine et extension du terme, dans: Zaïre-Afrique, XIII (1973) 78, pp. 471-483
    Les Bangala du fleuve sont-ils apparentés aux Mongo ? dans: Zaïre-Afrique,-XIV (1974) 90, pp. 623-632

    🙁 Can’t call it “success”: JMwB 1973 is not even in Google Books.

  358. And I do not understand “black” and “white” as linguistical cathegories. It is not that blacks and whites do not exist: there are both cultural and linguistical differences between Europeans and Africans (what about tones?). But they have been interacting… Treating a term to whose trajectory whites contributed as “some kind of deformation on the part of whites” is mad.


    comme disent les noirs, le “Lingala”.
    de Boeck, 1904.

    que les Blancs ont appelée le lingala,
    Mumbanza mwa Bawele, 1980.


  359. David Eddyshaw says

    The “deformation” referred to is surely just European mishearing or misunderstanding of whatever the original word or phrase at the back of “Bangala” was; it seems to be clear that that must have happened somewhere along the line. (I like to blame Stanley, just on account of his being such a nasty piece of work in general.)

    Moreover, I suspect the word has got carried over into English from the French déformation, which doesn’t have quite the same connotations: it’s “bending out of shape, distortion, warping, misrepresentation.” “Deformity” in French is difformité.

  360. David Eddyshaw says

    (Driving in France, one is frequently warned of chaussées déformées, which appear not to be deformed highways. But the ones you really need to slow down for are the denivellated ones.)

  361. David, but what does it mean?

    Words (and people/place names) change all the time. “Deformation” is a deformation of déformation.
    “French” is a “distortion, misrepresentation” of a word too.

  362. I love how LH threads migrate to tangential but still interesting topics.

    But I also have to mention that for more than a year now, each time I’ve seen Timbuktu Manuscipts pop up in the recent comments, I’ve been disappointed not to find new comments concerning Timbuktu manuscripts. Sigh.

  363. A friend of mine grew up in Kinshasa, so am quite curious about this particular word.

    But Timbuktu manuscripts are definitely a more interesting topic.

  364. It wasn’t meant as criticism of your comment in any way. It’s amusing to me that after being teased this way for a year, my brain still thinks “maybe this time?”

  365. I understand and I feel the same. “Old books” is not something people here expect to find in Africa. Africa is 1. wild life 2. dances (and ophtalmologists). It would be an understatement to say that I was excited when I first learned about books: I am a bookworm, after all.

  366. and ophtalmologists
    Dr. Eddyshaw has gained a further morsel of immortality 😉

  367. OphtHalmo-, sorry.

    See the definition of Africa here: https://www.culture.ru/poems/33118/barmalei

  368. David Eddyshaw says

    No problem.
    I was an ophtalmologue whenever I visited Burkina Faso or Togo.

    I object only to “opthalmologist” …

  369. @drasvi: That brings back memories. I remember watching the cartoon film based on that poem with my daughter when she was little.

  370. I rememebred it because of “1. wild life”.
    “In Africa gorillas, in Africa big angry/evil crocodiles.”, a definition:)
    One of the most famous poems for little children here. It does have a doctor and it is clearly inspired by the dr. Dolittle (The Russian doctor’s name is “Oh it hurts”, though).

  371. Dr. Dolittle
    …poorly known here. I am much better acquinted with the Doolittles of Pygmalion:(
    (I am often temtped to use “Not bloody likely!”)

  372. David Eddyshaw says

    Sadly, I am unable to talk to the animals. Or even to trees.

    There are, curiously, few stories about heroic ophthalmologists. A gap in the market …

    The excellent anime Monster has a leading character who is a saintly neurosurgeon, but apart from that it’s pretty realistic.

    Stanley Holloway was born to play Alfred Doolittle.
    (I’ve just this minute discovered that he was named after that aforementioned nasty piece of work Henry Morton Stanley. But then, I don’t expect he was consulted about it.)

  373. A heroic ophthalmologist was widely discussed 2 years ago. But it is a sad story.

  374. David Eddyshaw says

    True. Alas.

  375. One of the most memorable threads about the English langauge for me was when a Chinese learner asked native speakers’ opinion about her accent (she added a recordign, a fragment from some novel) and also asked if anyone can teach specifically Oxford accent.

    It provoked dozens comments where (male) English speakers both analyzed her phonetics and posted their own recordings of the very same fragment, first in their accents, then in accents that they can imitate. (cf. also http://lughat.blogspot.com/2013/06/how-different-are-egyptian-and-algerian.html and links therein). Sadly, they posted their recordings on vocaroo.com that only keeps them for a few months.

    Mostly they thought that her pronunciation is affected by Chinese tones, but her intonations reminded me how ladies speek in films, I suspected* she is trying to imitate them. And I watched several different versions of Pygmalion:) Apart of that Pygmalion (mostly My Fair Lady) was quite famous among learners in USSR: there are not many lays and films about English…

    *There were issues with her rhythm as others said to her, but I suspected that it was not only “substrate” influence but also attempts to imitate a particular intonation.

  376. Since I have no friends who read the NYRB, I have to share this here instead, one of the more ridiculous corrections I’ve ever read, from the latest issue:

    >We regret to inform you that the essay we published as new last month had been previously published in its entirety, 19 years ago, in another magazine. We were unable to consult the writer on its publication history because he has been deceased since 1993. And while we believe the 2002 edition was the fist airing of these particular words, the writer had previously published an essay about the same novelist in 1955, under the same title, suggesting that even in 2002, the ideas may not exactly have been fresh.

  377. Great heavens above!

  378. Is this the Jan. 13 issue? What page is it on?

  379. The correction letter is in the Feb. 10 issue I received yesterday. I did see the original but don’t remember which issue.

    To be fair, which usually means less funny, it’s a letter (which I paraphrased though pretending to quote) from the daughter of Irving Howe, who wrote the relevant piece(s?) on Isaac Babel. She took over as his literary executor after his son’s death. She was responsible for forwarding it to the NYRB for its latest publication and hadn’t realized her brother had done the same with the New Republic in 2002.

    It’s not clear whether the son realized in 2002 that the previous piece had been published in 1955. The letter doesn’t explain how the two pieces may be related. Publishing under the same title seems odd, and to me at least, suggests that the 2002 publication may have suffered from a similar lack of familiarity with the publishing history.

    The most likely story for the 2002 publication, to me, is that Howe’s son found the manuscript for the original article, with its title, didn’t know it had been published, and sent it to the New Republic. By then, the original was nearly 50 years old, so the number of people who had read the original and remembered it must have been vanishingly small, and internet archiving wasn’t nearly as advanced, so no one noticed, particularly given that by 2002, TNR’s focus was much narrower and its readership smaller.

    Also, when published a few weeks ago, it was made clear that it was written long before and was being published very posthumously. They wrote that it had been “unearthed.” I think it was fair to elide that in making fun of a ridiculous mistake, but it’s also right to acknowledge it after the joke.

  380. Ah, here it is — the December 16, 2021 issue. That’s hilarious; I can’t wait to get the new issue and see the correction for myself. Here’s the prefatory note:

    My father, Irving Howe, wrote this essay on Isaac Babel in the late 1980s, so far as I can tell, as the introduction to a book of Babel’s stories that Patricia Blake had planned to edit. The book was never published, and my father apparently tucked the introduction away in a drawer in his apartment. At the time of our father’s death, in 1993, my brother, Nick, began to collect the unpublished essays that subsequently appeared in the posthumous A Critic’s Notebook. Nick was aware of the Babel introduction and wanted to include it but could not find it; only recently did our father’s widow, Ilana Weiner Howe, come across the manuscript. Given my father’s long history with The New York Review, starting from its very first issue, he would have been amused to know that his essay on Babel now appears in these pages.

    —Nina Howe

  381. 40,000 Timbuktu mss. just went online!

    Djenné, too.

  382. 40,000 pages. Well, it’s a start, and the website is super-fancy.

  383. There’s a good deal more of it online at the HMML and EAP.

  384. Left out of my earlier lists of the common names of worms was the clade of spoon worms. I think that by the time I was memorizing all those picturesque worm names, it was sufficiently clear that the spoon worms are not actually a separate phylum echiura, but are highly derived annelids that have lost all segmentation in their adult forms. So I never picked up that name along with the others.

  385. The people ride in a hole in the ground.

    We do indeed. What is more, hole is apparently the technical term used by maintenance-of-way workers for the passages between stations that contain rails. They are only tunnels if they are under water.

  386. Trond Engen says

    Here’s another link from Dmitry, to recent paper on the genetic history of the Swahili ports:

    Brielle et al: Entwined African and Asian genetic roots of medieval peoples of the Swahili coast Nature 615 (2023).

    The urban peoples of the Swahili coast traded across eastern Africa and the Indian Ocean and were among the first practitioners of Islam among sub-Saharan people. The extent to which these early interactions between Africans and non-Africans were accompanied by genetic exchange remains unknown. Here we report ancient DNA data for 80 individuals from 6 medieval and early modern (AD 1250–1800) coastal towns and an inland town after AD 1650. More than half of the DNA of many of the individuals from coastal towns originates from primarily female ancestors from Africa, with a large proportion—and occasionally more than half—of the DNA coming from Asian ancestors. The Asian ancestry includes components associated with Persia and India, with 80–90% of the Asian DNA originating from Persian men. Peoples of African and Asian origins began to mix by about AD 1000, coinciding with the large-scale adoption of Islam. Before about AD 1500, the Southwest Asian ancestry was mainly Persian-related, consistent with the narrative of the Kilwa Chronicle, the oldest history told by people of the Swahili coast. After this time, the sources of DNA became increasingly Arabian, consistent with evidence of growing interactions with southern Arabia. Subsequent interactions with Asian and African people further changed the ancestry of present-day people of the Swahili coast in relation to the medieval individuals whose DNA we sequenced.


    A key finding of this study is evidence of mixture at roughly AD 1000 between peoples of African and Persian ancestries (Fig. 3 and Table 1). This is consistent with the Kilwa Chronicle, which describes the arrival of Persians on the Swahili coast and interactions between them and coastal residents. Whether or not this history has a basis in an actual voyage, the ancient DNA provides direct evidence for Persian-associated ancestry being derived overwhelmingly from males and arriving on the eastern African coast by about AD 1000. This timing coincides with archaeological evidence for a substantial cultural transformation on the coast, including the widespread adoption of Islam. At Kilwa, coin evidence has dated a ruler linked to a Shirazi (Persian) dynasty, Ali bin al-Hasan, to the mid-eleventh century. The genetic evidence suggests that this arrival was accompanied by mixture, which began by AD 1000, and continued later. People of both African and Asian ancestry made major contributions, with African proportions of approximately 57% on average at Mtwapa and Faza, 32% at Manda, 67% at Songo Mnara and 74% at Kilwa (Table 1, Extended Data Table 2 and Supplementary Table 8).

    Archaeological evidence provides important context for our genetic findings. The individuals that we analysed lived in the thirteenth to eighteenth centuries, and were excavated mostly from elite contexts. However, coastal sites from around AD 1000, when the mixture occurred, showed little evidence for distinct societal elites. Three of the sites sampled here (Mtwapa, Songo Mnara and Faza) did not exist as towns in AD 1000, and so these admixed populations moved to those towns later. Thus, the elite inhabitants of Mtwapa and other sites developed from admixed populations and were not foreign migrants or colonists.

    Linguistic evidence provides further context. Kiswahili is a Bantu language, and since most ancestry in medieval Swahili people derives from African people, our results suggest that the children of immigrant men of Asian origin adopted the languages of their mothers, a common pattern in matrilocal cultures. However, Kiswahili also has non-African influences, reflecting a millennium and a half of intense interaction with societies around the Indian Ocean rim. Persian loanwords contribute up to 3% of Kiswahili, but it is unclear whether they are derived directly from Persian or through adoption into other Indian Ocean languages. Arabic loanwords are the single largest non-Bantu element in Kiswahili (16–20% of words), and may be primarily due to incorporations after AD 1500.


    [TL;DR]The first population of the Swahili towns was a mixed population of local women and Persian men. This started no later than 1000 CE and went on until 1500 CE. Admixture after that is mainly Arabic.

    The obvious gamechanger in 1500 CE was the arrival of the Portuguese, which eventually made the local sultanates dependent on Oman.

  387. Thanks! (As a Yank, I still find it unintuitive that “Asian” = Persia and India, no matter how many times I’ve seen that usage.)

  388. John Cowan says


    That really is etymologically just what it looks like: ‘unspeaking/unreasoning’. It’s also used for ‘horsepower’, though the standard Contemporary Modern Standard Greek word for ‘horsepower is the otherwise-lost ίππος!

    apotropaic expressions after someone sneezes

    My understanding is that the point of God bless you and the like is that when you sneeze, your soul flies out your nose, and you need the blessing to keep the devil out while your soul creeps back in.

    Hebrew sus is a loan from Somewhere in Indoeuropean

    Now if it meant ‘pig’, the matter would be beyond all doubt. But alas, one who would milk the sow is doubly at risk, for backache and for disembowelment.

    traditional Turkish grammarians use the term izafet to describe the construction of e.g. Orhan’ın ismi “Orhan’s name” (“Orhan’s his-name”),

    Ottoman was very big on izafet, not just the construction (which is the original meaning of Arabic iḍāfah إضافة ‘construct case’, lit. ‘addition’) but the particle as well, first in Persian phrases and then in native ones. The particle was dropped as part of the Turkification of Turkish. It was used in Hind(ustan)i too, as in Kaisar i-Hind ‘Empress of India’.

    Peter Dinklage, perhaps

    The 19C American politician Stephen A. Douglas also comes to mind: he was known as “The Little Giant”, little because he was 163 cm and 40 kg, Giant because of his forceful, not to say outsize, personality: he defeated Lincoln in an 1858 Senate race and would probably have beat him for president in 1860 if the Democratic Party had not fractionated.

    the worst quality I had yet seen in my sheltered life

    The worst toilet paper I ever struck in a public toilet consisted of a set of overlapping sheets in a cardboard box. That was bad enough, but the paper was waxed.

    zingiber […] ginger beer.

    Coincidence. Ginger beer is what Americans call ginger ale (sweetened carbonated, usually non-alcoholic), only stronger.

    Some accents have a new front/back vowel allophony for /k/, namely [k]-[kʲ]

    As in the name Kemal [kʲemɑl]. This is a front-vowel allophone in native words, but can appear before /a/ or /u/ in loanwords and foreignisms.

    education through rebantuization

    Sounds like a slogan from 1990s South Africa.

    I object only to “opthalmologist”

    That took some figuring out. English phth can be /pþ/, /fþ/, or just /þ/ (as in phenolphthalein).

  389. January First-of-May says

    and would probably have beat him for president in 1860 if the Democratic Party had not fractionated

    Hard to say; the electoral vote math works out such that Lincoln would still have won in 1860 even if all non-Lincoln votes were combined into a single candidate*, but I can’t rule out that if Douglas was indeed the sole D candidate he’d actually have gotten a little more votes than OTL’s three combined (or indeed Lincoln a little less than OTL), even if naively I’d expect a slight difference in the other direction.

    I’ve seen a neat alt-hist short story where some gaffes by Lincoln in the run-up to the election cost him California, Oregon, Illinois, and Indiana (though he still got all his OTL electors in New York and New Jersey) – which meant that Lincoln was just short of an electoral vote majority, and Douglas just barely passed Bell to third place. In the resulting House of Representatives mess Douglas emerged as the obvious compromise candidate.

    *) In fact compared to OTL he would only have lost California and Oregon. The only other state where he received electors but did not win a majority was New Jersey, where the Democrats used a compromise setup that did effectively combine their votes into a single candidate, but some technical issues in implementation meant that Lincoln ended up with several electors anyway.
    That said, even without the New Jersey electors, he’d have gotten 169 – easily enough for a majority.

  390. David Marjanović says

    English phth can be /pþ/, /fþ/, or just /þ/ (as in phenolphthalein).

    Never /ft/?

  391. And, per Wiktionary, even /t/ initially, in phthisis.

  392. David Marjanović says

    Looks like I have to find some entomologists and make them say Phthiraptera.

  393. Looks like I have to find some entomologists and make them say Phthiraptera.

    Ah, to dream…

    (OED: BrE /(f)θʌɪˈrapt(ə)rə/, AmE /θaɪˈræptərə/. They probably sent a boy out to ask an entomologist or two.)

Speak Your Mind