South Arabian Languages.

Edward Fox writes for Al-Fanar Media:

I used to think that Arabic was the only language native to the Arabian Peninsula. As a student of Arabic, I learned how the ancient grammarians built the rules of standard Arabic from the speech of the desert Arabs. I thought the language of the Qur’an and of classical Arabic literature was an only child: I knew it had ancestors, but I didn’t know it had any living relatives.

I learned that the story was not that simple many years later in Oman, when I travelled from Muscat to the sultanate’s southern province of Dhofar, a region separated from the capital by 1,000 kilometers of mostly featureless flat land. There I met a young man who told me he spoke a local language that was not Arabic. When I asked to know more, he obligingly, and to my astonishment, spoke a sample of a language called Shahri, a linguistic rarity spoken by a few thousand people in this part of Arabia. It sounded nothing like Arabic. He seemed delighted by my surprise, and proud of his ability. […]

Shahri, also known as Jibbali, is language that is spoken mainly in a remote mountainous region of Dhofar province. (The name Jibbali is derived from the Arabic word for mountain.) It is one of a handful of related languages spoken in the southern part of the Arabian Peninsula that are distinct from Arabic, and are mostly unintelligible to speakers of the majority language.

The Unesco Interactive Atlas of the World’s Languages in Danger identifies Shahri as one of six languages in this group. The others are Bathari, Harsusi, Hobyot, Mehri and Socotri (spoken on the island of Socotra). It’s unclear how many people speak these languages, since Oman, like many countries in the region, has never conducted a census, but experts estimate the total numbers of speakers of these languages in the tens of thousands. One of these languages—Bathari—is believed to be spoken by only 11 people.

Collectively, these languages are called the Modern South Arabian languages. Aaron Rubin, a professor of linguistics at Penn State University in the United States and the author of a grammar of Shahri/Jibbali, explains that the Modern South Arabian languages are a branch of the Semitic languages group that includes Arabic and Hebrew.

“If you look at the Semitic languages as a family tree, you will see that the Modern South Arabian languages diverged from the rest of the Semitic languages a long time ago,” he said in an interview. “Structurally, Jibbali has straightforward similarities to other Semitic languages. But the vocabulary is very different.” […] “There are sounds in Shahri that we think existed in an ancestor language called proto-Semitic, but which have been lost in Arabic, Hebrew and Amharic (a language of Ethiopia),” says Aaron Rubin, of Penn state.

There’s more on the South Arabian languages at the link, as well as two clips of Shahri being spoken. And I’m glad to be introduced to the Atlas of the World’s Languages in Danger. Thanks, Trevor!

Comments

  1. David Eddyshaw says

    Very much a cut above the usual stuff on linguistics (it’s nice to be able to give praise where it’s due.)
    Not least because the author has consulted not one but two actual experts on the subject.

    I was giving a lecture lately in which I had cause to mention Augustin-Jean Fresnel; I was happy to be able to tell the troops that he had a much more interesting younger brother, Fulgence

    https://en.wikipedia.org/wiki/Fulgence_Fresnel

    (Actually they seem to have been one of those families. Like the Ramsays (as in numbers and Archbishops.)

  2. David Eddyshaw says

    (I should perhaps explain, having just noticed that Wikipedia doesn’t, that FF was the first linguist to discover the Modern South Arabian languages.)

  3. David Eddyshaw says

    After a bit of fossicking, I’ve remembered that the sickeningly polymathic family that was at the back of my mind were the children of Herbert William Fisher:

    https://en.wikipedia.org/wiki/Herbert_William_Fisher

    who included HAL Fisher the historian and MP, an admiral, a chairman of Barclays Bank, the most celebrated editor of Tacitus (unfortunately killed in the Battle of Jutland), Ralph Vaughan Williams’ first wife and Charles Darwin’s daughter-in-law. No archbishops, though.

  4. John Cowan says

    I used to have a plastic Fresnel lens stuck to the glass of the window in my living room. Very beautiful when the sun shone.

    the children of Herbert William Fisher

    Roland Fisher, statistician and co-founder of population genetics, belonged to another kettle of fishers altogether, however.

    I once worked with one of those Fodors: like me, almost the only non-Ph.D. in the family.

  5. Savalonôs says

    These languages deserve better terminology, since the “Modern South Arabian” languages are not descended from “Old South Arabian”, plus Old South Arabian is not completely extinct in the modern era and Modern South Arabian languages presumably had ancestors that existed in the old days. Also “Arabian” might tend to confuse people into thinking that they are a subset of Arabic. I’ve heard Old South Arabian also referred to as Ṣayhadic (no idea where that name comes from) and Wikipedia suggests Yemenite as another option. For Modern South Arabian, the only alternative Wikipedia suggests is Eastern South Semitic, which I suppose is no more syllables than Modern South Arabian is.

  6. David Eddyshaw says

    Old South Arabian is not completely extinct in the modern era

    Oh!

    https://en.wikipedia.org/wiki/Razihi_language

    I did not know that!

  7. Stu Clayton says

    (Actually they seem to have been one of those families. Like the Ramsays (as in numbers and Archbishops.)

    The Ramsey number “solves the party problem”.

  8. David Eddyshaw says

    like me, almost the only non-Ph.D. in the family

    I am the first in my paternal line even to have been to university; in my wife’s family, on the other hand, to have only one higher degree marks you as an underachiever.

    As I have remarked on several occasions to my daughter, no man worth bothering with for a moment wants a stupid woman as a wife (and, I dare say, vice versa.) Her significant other is pursuing a PhD which involves Python and Latin; I feel he is Worthy.

  9. David Eddyshaw says

    The Ramsey number “solves the party problem”

    Small parties are always the best.
    (And otherwise we have to try to destroy the alien.)

  10. no man worth bothering with for a moment wants a stupid woman as a wife

    I was told this was a reason for South Korean female higher education.

    The entire point of it is to get a degree, work for a few years in a prestigious company, meet a promising young man, get married and become a housewife.

    Because, you know, any men worth having won’t marry stupid women.

  11. David Eddyshaw says

    I am fairly confident that my daughter does not read this blog; if I’m wrong, we may be in for a flame war … (she lacks my invariable concern for courtesy above all. And she seems to have a remarkable number of Korean friends …)

    Actually, even that somewhat curdled take on the phenomenon would at least reflect an appreciation on the part of Korean manhood that clever girls are worth catching. These things take time …

  12. I was told this was a reason for South Korean female higher education.

    The inside joke in these Mormon parts of the world is that the girls go to college to get their R.M.

    Strictly speaking it isn’t one of the degrees of the alphabetic soup, but stands for Returned Missionary

  13. Stu Clayton says

    I imagine some men appreciate the extra power kick in the idea of subduing a woman smarter than they are. It’s beset with conceptual risk, though, since if they succeed it suggests she wasn’t so smart after all.

    As Gertrude Stein used to say: plus ça change, plus c’est la même rose.

  14. John Cowan says

    Furthermore, it’s doubtful that the six MSA languages are actually a clade. It’s clear that Mehri, Harsusi, and Bathari are closely related, but the common elements in all six may well be shared retentions: there is no clear evidence of shared innovations, nor of any shared innovations between MSA and Ethiopic Semitic. It’s the Central Semitic languages (Arabic, Aramaic, Hebrew, and their extinct relatives) that are clearly innovative, and the rest are just various leftover groups that don’t really fit together.

  15. OT/BDT (Off Topic/But Down Thread, so maybe it’s okay …)

    An article on the difficulties of translation, from that old favorite of linguists and intellectuals everywhere – Entertainment Weekly:
    https://ew.com/books/2018/10/19/harry-potter-translators/

    If it starts with a fart joke, things can only get better, right?

    I actually think it’s pretty good. Here’s a sample:

    >“There is an unhappy wizard in the Goblet of Fire who sadly is unable to attend the Triwizard Tournament, suffering from a condition named “lumbago.” For this medical term we have the common expression ‘Hexenschuss’ – witches’ shot. Of course I used this funny coincidence with gusto. There was no pun intended in the original, but sometimes it is legitimate to use a pun in our own language to make up for puns that we possibly lost in other parts of the translation.”

    I believe that was Newton’s 3rd Law of Translation – the conservation of puns.

    One amusing mistake that got past the C.E. – “Isreal.” The author would seem to be Jewish. Did she never read the final copy and ask for a correction?

  16. Athel Cornish-Bowden says

    No archbishops, though.

    Not Geoffrey Francis Fisher, Archbishop of Canterbury from 1945 to 1961, therefore?

  17. SFReader: “The entire point of it is to get a degree, work for a few years in a prestigious company, meet a promising young man, get married and become a housewife.”
    David Eddyshaw: “even that somewhat curdled take on the phenomenon would at least reflect an appreciation on the part of Korean manhood that clever girls are worth catching.”

    David, not necessarily. The young men at prestigious companies might marry their female coworkers, not because they’re clever, but just because they’re the ones around. And with the working hours they have to put up with, they might not have time to look for potential wives elsewhere. I’ve heard this about Japan too, by the way.

  18. Stu Clayton says

    In a way those could be called mariages de convenance, if it is permitted this once to mistranslate “convenance” as “convenience”.

  19. David Marjanović says

    I’ve heard Old South Arabian also referred to as Ṣayhadic

    IIRC, that’s a script, not a language.

  20. Lars (the original one) says

    Also MSA so much means Modern Standard Arabic that any other name for the group would be better.

  21. David Eddyshaw says

    Not Geoffrey Francis Fisher, Archbishop of Canterbury

    Apparently not, though I had originally misremembered him as one of the clan.

    Aristophanes’ Frogs starts with fart jokes. It’s got something for everybody.

    The young men at prestigious companies might marry their female coworkers, not because they’re clever, but just because they’re the ones around. And with the working hours they have to put up with, they might not have time to look for potential wives elsewhere.

    From Richard Gordon’s Doctor in Love:

    It is a fact well known to the medical profession that doctors marry either nurses, other doctors, or barmaids. During the most marriageable years these are the only women they meet.

  22. John Cowan says

    get their R.M.

    A very dated mainstream-American version of this is “going to college to get your M.R.S.”

  23. @X, when I did my graduate work at BYU, I heard more people use the sexist line that L.D.S. women went for their “M.R.S.” degree, read as three letters. That they specifically went to find an “R.M.” was a given, because young men who failed to “serve a mission” were less “worthy”. (Need to wash my mouth out after all that Mormonspeak.)

    What I found fascinating were the EE/EE dances, specifically held for Elecrrical Engineering and Elementary Education majors – large programs with severe gender imbalances. Sadly this just 20-25 years ago.

    As a closeted gay graduate student who was nevertheless an R.M., I found the sociology fascinating.

    Okay, enough derailment.

  24. An anecdote of how the Soviet central planning worked at its best.

    There was a large center of textile industry in central Russian city of Ivanovo with such overwhelmingly female (and quite young) labor force that it got a nickname – “bride town”.

    The Soviet planners decided to fix the dangerous gender imbalance by building in Ivanovo one of the largest airbases of the Soviet Air Force.

    It worked very well – lots of young factory girls married young air force lieutenants to everyone’s satisfaction.

  25. John Cowan says

    Aristophanes’ Frogs starts with fart jokes.

    The Arrowsmith translation has the hero of The Acharnians say about his wife: “The dawn comes up like thunder. So does she.”

  26. David Eddyshaw says

    that L.D.S. women […] specifically went to find an “R.M.” was a given, because young men who failed to “serve a mission” were less “worthy”

    In William Foley’s superb (if somewhat intimidating) grammar of the Yimas language of New Guinea, I read:

    only boys who had killed in a headhunting raid were permitted to wear a pubic covering of a flying fox skin and hence be eligible to take a wife.

    Mormons, Yimas … truly, all men are brothers.

  27. John Cowan: I don’t want to toot my own horn too much here, but your comment to the effect that Modern South Arabian (=MSA) languages (and/or Ethiopian Semitic) may not form a genetic subgroup within Semitic is very interesting: If they indeed do not, this would seem to be *very* compatible with the scenario I suggested here (July 12 4:05 comment) involving the spread of Semitic out of Africa (and I meant “West Coast of Arabia” when I wrote “East Coast of Arabia”, nota bene):

    http://languagehat.com/natufian-origin-for-afroasiatic/

  28. Here‘s a direct link to Etienne’s comment.

  29. David Marjanović says

    I’ve come across the idea that Proto-Semitic was spoken in the Rub al-Khali. When that place, which at one point was a river delta, became one of the driest on Earth, some of the people moved north and founded East Semitic, while others moved south and founded West/Rest Semitic.

    Elementary Education

    I’ve recently learned that many women of various fundamentalist stripes study that to better educate their own children, whom they expect to have promptly after they’ve married and become housewives.

  30. Savalonôs says

    Well, if “Modern South Arabian” isn’t really a clade, then maybe “Eastern South Semitic” communicates the appropriate level of vagueness.

    @David Marjanović See https://www.academia.edu/2603460/The_subgrouping_of_the_Semitic_languages , part 7 on Old South Arabian a.k.a. Ṣayhadic.

  31. David Marjanović says

    I must have confused it with something, maybe Safaitic.

    BTW, as useful as this introduction is, its discussion of the evidence in terms of “what is reliable, and what must be ignored” reminds me of the 1950s in biology. Right at the beginning, the most important innovation of West Semitic – and the only one we’re shown – is said to be the replacement of a past tense by a perfect tense (which comes from an older stative, preserved as such in East Semitic). But such things are capable of areal spread just like phonological and phonetic innovations. The English present-perfect tense has analogs all over Northwest Germanic today, but that is a borrowing from Romance, which in turn got it from Greek; and while I suppose it could be a borrowing into Proto-West-Germanic, I’d be quite surprised if Proto-NW-Gmc. already had it, given how rare it is in the oldest texts and how old it could be in Romance. Next, its practically complete replacement of an inherited past tense in French and Upper German has identifiable internal motivations on both sides, but the geography makes that suspicious – some or all of the internal motivations could be motivations for borrowing instead of for an internal development.

    Even further by the way, the author has a number of interesting publications on his Academia.edu page one click away. How about the first-ever study of Judeo-Urdu?

  32. Someone should publish on the Odessa dialect and call it Judeo-Russian.

  33. David Marjanović says

    Sure, but Judeo-Urdu is written in Hebrew letters, so it’s 273% more Jewish.

  34. Didn’t anybody ever write Russian in Hebrew letters?

  35. John Cowan says

    Ṣayhadic, says the WP talk page, was a name proposed by A. F. L. Beeston “as a convenient term after the medieval name [Ṣayhad] for the desert Ramlat al-Sab´atayn where many inscriptions have been found”. The desert is on the undefined border between Yemen and Saudi Arabia; the WP article on it says it “extends from al-Khawr to the edge of the Rub’ al-Khālī or Empty Quarter.”

    The fact that we now know about a modern Ṣayhadic language makes “Old South Arabian” an even more unapt (not to say inept) term.

  36. Owlmirror says

    Didn’t anybody ever write Russian in Hebrew letters?

    The link posted by Savalonôs is to a work by Aaron Rubin (one of the linguists who contributed to the OP). Other works on by the same author include “The Jibbali (Shahri) Language of Oman: Grammar and Texts” (specifically relevant to the OP), and a work he co-edited and co-introduced, “Handbook of Jewish Languages”.

    While the latter is unfortunately just a ToC, it includes a chapter titled “Jewish Russian”, by Anna Verschik, followed immediately by “Judeo-Slavic”, by Brad Sabin Hill.

    Rubin contributed a chapter on Judeo-Italian, and the epilogue, “Other Jewish Languages, Past and Present”.

  37. David: I think you are a little unfair, the author refers several times to the difficulty involved in separating inherited and diffused changes. The cautious, consensus classification he offers, with all non-Central Western Semitic varieties spoken today (or, in the case of Ethiopic, spoken in the past) in the far South of the Arabian peninsula, does seem to indicate that West Semitic first broke up there, which in turn fits the theory I presented here rather nicely (whether, in addition, the rub’ al-Khālī played a role in the diversification of Semitic or not I will leave to people more qualified than I to answer). In fact, if his classification is accepted, it seems well-nigh impossible to believe that the migrants who introduced Proto-Semitic into the Middle East could have arrived from Africa via a route near the Mediterranean.

  38. Owlmirror says

    Anna Verschik has an Academia page as well, which includes works titled: “Yiddish, Jewish Russian, and Jewish Lithuanian in the Former Soviet Union” and “Jewish Russian and the field of ethnolect study”.

    [ Not linking because WordPress doesn’t seem to like me doing so. Or at least, a comment with a link disappeared for a while before resurfacing.]

  39. January First-of-May says

    Or at least, a comment with a link disappeared for a while before resurfacing.

    Anything with more than one link, or anything with one link that gets edited, is sent into moderation, and has to be rescued manually by the blog owner. This usually takes several hours.

  40. just a ToC, it includes a chapter titled “Jewish Russian”, by Anna Verschik, followed immediately by “Judeo-Slavic”, by Brad Sabin Hill.

    A bit more than just a TOC. Opening pages are available too.
    https://brill.com/view/book/edcoll/9789004297357/B9789004297357_021.xml?lang=en

    There, the discussion starts from Old Czech “Cnaanic” glosses but later on Hill promises to discuss “East Judeo-Slavic” spanning a later era up to XVII c., of which to my knowledge nothing exists but a lone XVII c. note in one of 9 documents appended to Wilno rabbi Shabbetai haKohen’s treatise on the halacha. Still it’s fascinating to think that somewhere in far-flung Lithuania, some – quite likely a Jew – needed a comment on the Ashkenazi traditional law worded in Slavic.

    The note is credited to Shabbetai’s father Meyer “Katz” (Kohen Tzedek) Ashkinazi, who spent his late years in Mogilev in today’s Eastern Belarus, then on the Eastern-most fringes of the Ashkenazi Jewish settlement area , having been invited from Frankfurt area to spread wisdom in the East.

  41. This usually takes several hours.

    Sometimes it only takes a few minutes; it all depends on when I notice there’s something in moderation.

  42. Dmitry Pruss says

    In active discussions, sometimes a delay means that a comment appears pages above the apparent end of thread. And it may be a lot harder to notice, then. Is it feasible to update publication time when approving?

    I have one in this thread btw

  43. I’ve rescued it!

  44. later on Hill promises to discuss “East Judeo-Slavic” spanning a later era up to XVII c., of which to my knowledge nothing exists but a lone XVII c. note in one of 9 documents appended to Wilno rabbi Shabbetai haKohen’s treatise on the halacha. Still it’s fascinating to think that somewhere in far-flung Lithuania, some – quite likely a Jew – needed a comment on the Ashkenazi traditional law worded in Slavic.

    The note is credited to Shabbetai’s father Meyer “Katz” (Kohen Tzedek) Ashkinazi, who spent his late years in Mogilev in today’s Eastern Belarus, then on the Eastern-most fringes of the Ashkenazi Jewish settlement area , having been invited from Frankfurt area to spread wisdom in the East.
    Sorry for an extensive self quote. I was duped by a popular misinterpretation of R. Katz’s passage which travels from one study to another without assessing the original. Alexander Beider carefully debunks it in “Origins of Yiddish Dialects” (Oxford University Press, 2015). There is no “responsa in Hebrew-glossed Slavic”. Instead, a lone word is glossed, and it’s a Slavic toponym. Basically R. Katz, a Westerner, gripes that the Jews of Mogilev can’t stick to proper Yiddish and instead use Russian, as in using a toponym Brest for a town the Jews should properly call Brisk. To an uninformed observer, it may sound like a prescriptivist observation of Russian or rather Belarussian borrowing. Or maybe even, gasp, of a complete switch to Judeo-Slavic.

    Beider shows, however, that it was the form Brisk which was borrowed, from Poland’s German settlers, and by mid XVII c it has become a norm in Polish Yiddish but not yet further East. He also quotes from several Slavic documents with Jewish names of mid-XVII c Mogilev, like contracts and list of pogrom victims, to support the conclusion that Yiddish was the primary language of all of them.

    So the old wise man was just dissing the local usage for a misplaced reason, and then the modern researchers made (as both Ashkenazi and Slavs would say) an elephant out of a fly (which would be a mountain out of a molehill in English)

  45. Fascinating, I love molehills like that!

  46. Hill promises to discuss “East Judeo-Slavic” spanning a later era up to XVII c.

    An addendum from Alexander Beider: Brad Sabin Hill discusses another Hebrew-glossed Slavic line also linked to the legacy of R. Katz of Mogilev, and first discussed by Dubnow in “Jewish Antiquities” 1909 (1), 15. In this second case, the text is 100% Slavic, w/o a “Judeo” part. There, a Jewish witness retells what a Christian witness of a murder said, in Belorussian recorded with Hebrew characters.

    Here is how Dubnow re-glossed it in Cyrillics: ваш школьник, что на школьном (дворе?) живал: зарезали яго як куры резав… там богаты мужики не бардзо везьмут ~~ Your synagogue assistant, who used to live at the synagogue compound, they slaughtered him like they slaughter hens … there are rich peasants (or many peasants) who wouldn’t take much [to do it?]

  47. Owlmirror says

    школьник

    Hm; I know I’ve seen/heard “Skolnik” (or some variant) as a family name. It means . . . beadle (Wikipedia says the Hebrew terms are gabbai or shamash)? Is that right?

    Is “shkol” the Russian/Slavic word for “shul”?

  48. Owlmirror says

    Anything with more than one link, or anything with one link that gets edited, is sent into moderation,

    I think there’s more to it than that. The comment I posted had one link, and the whole thing disappeared as soon as I posted it (so no editing took place).

    I think the other factors may include the length of the comment, the age of the original post, the number of previous comments to the original post, the number of previous comments made by the commentator, the character sets used in the comment, certain words used in the comment, whether approval has been given previously, and for all I know, the phase of the moon and whether a carp has been properly sacrificed to the Highest God.

  49. My great-grandfather (the one who was president of the Workers’ Circle in Chicago) said he spoke two languages back in Bobruisk*: Yiddish and Russian. I think he would have scoffed at the notion of a Judeo-Russian dialect; the use of the specifically Jewish terminology in running Russian speech was just code switching into Yiddish.

    * He and his brothers talked about Bobruisk as being part of the Ukraine, not White Russia. Although the city is quite centrally located in modern Belarus, they had all gotten out of the Russian empire by 1909.

    @Owlmirror: There are other things, besides having two HTML links that can push a comment to moderation. Cyrillic letters were causing a lot of trouble a while back, although they might be safer now. However, links are by far the most common issue. And if your comment ever does not appear, you can always e-mailed Hat to let him know he has something in the moderation queue.

  50. whether a carp has been properly sacrificed to the Highest God.

    The carp is very important.

  51. I’ve seen/heard “Skolnik” (or some variant) as a family name. It means . . . beadle (Wikipedia says the Hebrew terms are gabbai or shamash)? Is that right?

    Is “shkol” the Russian/Slavic word for “shul”?
    A Skolnick was my genetics mentor. Yes it is a fairly common “professional” surname; Yad Vashem has over 70 pages of Skolnicks/Shkolniks among the genocide victims…

    I couldn’t find a good English translation to shkolnik as an occupation, and regular Russian dictionaries offer a false-friend shkolnik = grade school student, a pupil. The two words are full cognates of course, since Yiddish shul => Russian calque “shkola” (school). Only smaller congregations were officially designated as “prayer schools”, while the larger ones were officially “synagogues”. The law prescribed how many Jewish houses of prayer can be in a shtetle, depending on its population, and 2 “schools” were equivalent to 1 synagogue. But a Shulhoff was always calqued as “a school yard”, школьный двор.

    Russian translation of shkolnik, служка, is a full equivalent of shamash, but gabbai is used in Russian primarily as a Karaite title.

  52. spoke two languages back in Bobruisk*: Yiddish and Russian. I think he would have scoffed at the notion of a Judeo-Russian dialect; the use of the specifically Jewish terminology in running Russian speech was just code switching into Yiddish.
    Just like Yinglish, there was a variety of Yiddish-infused Russian dialects, with peculiar grammar often showing Ukrainian influences as well, peculiar phonetics (part-reflected in spelling since Russian spellings hew closer to phonetics) and borrowings (some of which permeated slang and over time became mainstream if low-register Russian words, like хохма or шмон). But they were universally glossed in Cyrillics, and universally derided as illiterate. Few people would consider Yinglish a “Judeo-English”, and the same ought to apply to the transitional Russian dialects which similarly didn’t last and weren’t meant to last. There is also an extant Russian idiom of Israel, grammatically and phonetically quite correct but heavy on borrowings (мазган!). It’s in all likelihood a transitional immigrant dialect, too, but we will know better in 30 or 40 years.

  53. David Marjanović says

    David: I think you are a little unfair, the author refers several times to the difficulty involved in separating inherited and diffused changes.

    And then he uses this difficulty as a reason to exclude all phonological innovations from consideration, because they are potentially areal, while the fact that morphological innovations can be areal, too, is not mentioned.

    This is how phylogenetics worked in biology, too, back when it was an art: authors who understood the difficulties perfectly well drastically simplified these difficulties into reasons for considering a very small part of the evidence and discarding all the rest.

    In fact, if his classification is accepted, it seems well-nigh impossible to believe that the migrants who introduced Proto-Semitic into the Middle East could have arrived from Africa via a route near the Mediterranean.

    Yes.

  54. Sounds good, though one remaining option might be to treat southern Arabia as a residual zone, populated by speakers of various groups coming in from the north in several waves over the centuries. The dialectology of Arabic itself sure looks close enough to something like this… but maybe that’s actually rather due to substrate effects.

    I like the counterclockwise-eastern-Semitic theory for imaginativeness, but I’d expect to then find some evidence for it from the usual stuff like placenames along the Gulf. I would consider also a pilot expansion model: East Semitic heads north along the same route as Central Semitic, just a millennium or so earlier, so as to create (1) a trail of related varieties that can be easily assimilated entirely under a CSem. expansion, (2) a relatively distinctive language boundary once only Akkadian in the far northeast is left. (Compare Chuvash vs. Common Turkic, or Brythonic vs. Romance.)

    Some time ago I also used to change views with a guy who thinks Semitic broke up in Ethiopia and crossed north from there in multiple waves, one perhaps clockwise around the Red Sea; but this never got developed in good enough detail and I kind of suspect Teeter’s Law (he’s Tigrinya and talks up a lot the diversity of “Afro-Semitic”) and/or overapplied decolonial theory (he has tended to spend a lot of time on how racist Africanists of yore were vs. not that much on why exactly their views were wrong).

  55. Teeter’s law: The language of the family you know best always turns out to be the most archaic.

  56. David Eddyshaw says

    Teeter’s law
    Heh!

    I can see an obvious mechanism for this, especially with relatively understudied families: your pet language is the only one well enough described for the archaic feature you are obsessing about to have actually been documented.

    For example: a closed but fairly large subgroup of Kusaal verbs have an imperfective form in *-ya rather than the default *-da, and I have found no description of anything similar in any other Western Oti-Volta language. It is undoubtedly ancient, with exactly parallel formations in the much less closely related language Nawdm, where the cognate suffix is quite productive; Nawdm jeɦra “is standing”, for example, matches its synonymous Kusaal cognate zi’e(ya) exactly, phoneme by phoneme.

    I strongly suspect that this appearance of uniqueness in antiqueness is simply an artefact of the relevant formations having been missed in other Western Oti-Volta languages. There are tantalising glimpses in the example sentences in some of the dictionaries …

  57. John Cowan says

    But, but, but, there are no shared innovations (that I know of) between Modern South Arabian and Ethiopic. Surely the obvious hypothesis is that Semitic broke up in Central Semitic-land, with the homeland later innovating and the outliers (Mehri/Harsusi/Bathari, Jibbali, Soqotri, Hobyot, Proto-Ethiopic) not. I mean, like English.

    (And not like Niger-Congo, where all the outliers are concentrated on the northern edge. I discovered this truly delightful map of Niger-Congo the other day, which rather than assigning a different color to each group, assigns colors to non-Atlantic-Congo, non-Volta-Congo, non-Benue-Congo, non-Bantoid, non-Bantu, and Bantu, showing the Russian-doll structure of Niger-Congo, where at each level Bantu there are a few outlier groups and languages and then the great bulk of the languages and speech areas are in the next level, or doll. You can go further and say that Bantu A and much of Bantu B are the outliers for a True-Bantu doll.)

  58. David Eddyshaw says

    There is the spread of k from the first person of the suffix conjugation to the second. I believe that this is generally discounted as not meaning much, on the grounds that it’s not difficult to imagine it happening independently (Central Semitic also levelled, but the other way, after all.) It depends on what you mean by “shared”, I suppose.

  59. John Cowan says

    On moderation: if you do not edit again after adding your one permitted link, you do not go into moderation no matter how many times you have already edited, a fact I used to advantage above. Add the link last, when you are sure everything else is right!

    UPDATE: Wrong, wrong, wrong! Doing that just postpones moderation until the timer expires.

  60. David Eddyshaw says

    Indeed. Your comment that I replied to has now disappeared. I must practice my parallel programming skills. Or my Kriegspiel.

    (I am manfully resisting the temptation to wait ten minutes and then add a link for the evulz.)

  61. David Marjanović says

    I kind of suspect Teeter’s Law (he’s Tigrinya and talks up a lot the diversity of “Afro-Semitic”)

    I know who you mean. On the one hand, he absolutely has a point that “Afro-Semitic” is dramatically underresearched (one does not simply walk into Ethiopia); on the other, he has resorted to ideas like [f], widespread in Ethiopia, being a retention from Proto-Semitic and [p] being derived from [f] – the only way to get that to work would be a substrate strong enough to completely destroy the famous Semitic grammar.

  62. AJP Crown says

    a closed but fairly large subgroup of Kusaal verbs have an imperfective form in *-ya rather than the default *-da, and I have found no description of anything similar in any other Western Oti-Volta language.

    You could probably clear this up if you had some examples with goats.

  63. David Eddyshaw says

    Very true. I should have thought of that.

    Bʋkaŋa la pʋ zi’eya.
    Bʋpielkaŋa la daa pʋ zi’eya.
    Bʋsabilbamma la pʋ zi’e kpɛɛ.

    I hope that clarifies the matter.

  64. David Marjanović says

    kpɛɛ

    Is that the sound Kusaal-speaking goats make?

  65. David Eddyshaw says

    Yup. Same as modern Greek sheep go vi, vi.

  66. AJP Crown says

    Bʋkaŋa – this goat, but unfortunately
    Bʋpielkaŋa la daa pʋ zi’eya – did not match any documents. I’m going to have to read A Grammar of Kusaal.

    kpɛɛ

    seems crazy but it makes sense when I think about it. Goats have voices several octaves higher than sheep, at least the ones around here do.

  67. I just saw Tarkovsky’s last movie, Offret/The Sacrifice, and at one point a completely unmotivated flock of sheep wandered across the field of view, bleating. Also, a dog barked in the distance.

  68. David Eddyshaw says

    seems crazy

    Indeed it does.
    I believe that the caprine vocal apparatus has been shown to be incapable of producing doubly articulated consonants.

    I owe the observations on the cry of the sheep to Erasmus, that fine zoologist.

  69. David Eddyshaw says

    not like Niger-Congo, where all the outliers are concentrated on the northern edge

    Nah.

    This is an artefact of the tail-wagging-the-dog notion still all too unthinkingly propagated that gives Bantu (a subgroup of a subgroup) a primacy in reconstruction of Niger-Congo for which there is no evidence. Bear in mind that Bantu has expanded over central and southern Africa in something less than three thousand years (much less, for most of its current range.)

    You could create exactly the same sort of illusion for the Tarim-Liffey languages by drawing a map detailing all the Germanic languages with all their dialects in detail along with some big labels for “Slavonic”, “Italic”, “Indic” etc. Germanic outliers …
    The map shows three and a half languages in Ghana (which has about forty) and about a dozen in Nigeria, which has something over two hundred.

    The Bantu languages (or at least the better known of them) have a wonderful agglutinative exuberance combined with relatively extensive documentation which all too readily led to the idea that in reconstructing proto-Bantu (a comparatively straightforward task for a subgroup of a subgroup) you were pretty much reconstructing proto-Niger-Congo. But there’s good reason (or at least an argument not to be cursorily dismissed) to suppose that in the verbal system in particular this is the result of innovation, welding previously independent pronouns and particles to verb words, not an ancient morphological survival. As Tom Güldemann has pointed out, the very regularity and transparency of the agglutinative Bantu verbal system is a strong argument against it being a survival from proto-Niger-Congo. It must be either a relatively recent creation, or have been extensively and repeatedly remodelled by analogy.

    My own impression is that Oti-Volta as a group is pretty conservative (possibly illustrating Teeters’s Law, I guess, but …) The crucial comparative work on proto-Oti-Volta is all there still waiting to be done: the standard references are still the many-decades-old work on “Gur” of Gabriel Manessy, who achieved much and had the right ideas but was working with vastly less extensive data than are now available and made some stonking methodological errors.

    Mande is probably not related to Niger-Congo at all, and is therefore not an example of an outlier; and the connexion of (bits of) Kordofanian and Atlantic to core Niger-Congo is a long-range hypothesis right up there with Altaic, not a demonstrable fact like Indoeuropean. Calling them outliers is putting the cart before the horse.

    [Addendum: on reading your post more carefully, I see that I am in fact agreeing with you. For “Nah” at the top, read “Yup” and make minor adjustments in polarity as required …]

  70. David Eddyshaw says

    To put it fractionally more succintly: there is indeed a significant decrease in diversity of structure within Niger-Congo as you go southward, but it’s not because the northern languages are outliers: it’s because the southern languages all belong to a single twig of the tree which has expanded over a huge geographical area comparatively recently.

  71. AJP Crown says

    bleating
    The sheep around here make deep, stentorian roars.

    incapable of producing doubly articulated consonants
    Our goats didn’t communicate very much vocally, except always as a greeting or to otherwise attract your attention. They expressed details to humans by eye movement, looking in the direction of different foods or doorways and then back at you, but between themselves – and besides the social subtleties of butting – there always seemed to be subliminal interactions that were harder for me to identify. As with dogs, smells tell them an awful lot about what’s been going on.

  72. Trond Engen says

    David E.: Bear in mind that Bantu has expanded over central and southern Africa in something less than three thousand years (much less, for most of its current range.)

    On the waves of expansion and near extinction of Bantu languages in the Congo rainforest:

    Dirk Seidensticker et al 2021: Population collapse in Congo rainforest from 400 CE urges reassessment of the Bantu Expansion, Science Advances, Vol. 7, no. 7, eabd8352

    From Dmitry as usual, It’s a really cool paper using pottery classification and 14C data to identify periods of expansion and retraction of human settlement, broadly confirming a pattern already arrived at by linguistic means*. Then they use population genetics to test the hypothesis and find that all the populations tested carry evidence of reduced population size in the identified period of depopulation.

    They discuss several reasons for depopulation. The rough coincidence with the Justinian plague is mentioned, and it’s speculated that the Justinian strain of Yersinia pestis might have orginated in e.g. Ethiopia instead of the Eurasian Steppe. That’s interesting but this being the rainforest, there should be enough of local diseases capable of producing a prolonged pandemic.

    *) Or so they say. I haven’t checked the data.

  73. The map shows three and a half languages in Ghana (which has about forty) and about a dozen in Nigeria, which has something over two hundred.

    Ethnologue’s figures are 80 and 500 respectively, though they are of course splitters. I have never asked a Ghanaian how many languages are spoken in their country, but I have asked Nigerians living in the U.S. the analogous, and the general response is “About a thousand”, making them supersplitters.

  74. It’s a really cool paper

    It sure is. Here’s the abstract:

    The present-day distribution of Bantu languages is commonly thought to reflect the early stages of the Bantu Expansion, the greatest migration event in African prehistory. Using 1149 radiocarbon dates linked to 115 pottery styles recovered from 726 sites throughout the Congo rainforest and adjacent areas, we show that this is not the case. Two periods of more intense human activity, each consisting of an expansion phase with widespread pottery styles and a regionalization phase with many more local pottery styles, are separated by a widespread population collapse between 400 and 600 CE followed by major resettlement centuries later. Coinciding with wetter climatic conditions, the collapse was possibly promoted by a prolonged epidemic. Comparison of our data with genetic and linguistic evidence further supports a spread-over-spread model for the dispersal of Bantu speakers and their languages.

  75. Trond Engen says

    Thanks. I usually quote the abstract, but somehow I forgot.

  76. David Eddyshaw says

    Ethnologue’s figures are 80 and 500

    I think the Ghanaian figures may reflect subdivision of mutually comprehensible dialects into distinct languages in order to conform with ethnicity: does Ethnologue count Twi and Fante as two distinct languages, for example? (They aren’t by purely linguistic criteria.)

    [The reverse also happens, though: Toende and Agolle Kusaal are distinctly more different from each other than Twi and Fante are, but the speakers regard themselves as all belonging to the same ethnic group, and all speaking the same language. Still, Burkina Faso Toende speakers, who haven’t been exposed to Agolle Kusaal to the same extent as their Ghanaian Toende cousins, self-report quite low levels of comprehension of the Agolle dialect; though it may be significant that they also regard it as less “pure” than their own, so they may just be too snooty to understand it.]

    Don’t know about Nigeria, and strongly suspect nobody else does either, when you come right down to it. I’ve seen estimates from 250 to 400, and wouldn’t be surprised in retrospect if there really were more than 400. There are a lot of tiny Chadic and Adamawa languages and the like which still haven’t been quite gobbled up by Hausa, and coastal languages unrelated to anything, like Ijaw and Defaka …

    It’s a really cool paper

    Interesting stuff. It would help to account for the fact that the Bantu languages (once you get beyond the weirdo northwest-corner languages) do not seem diverse enough in the light of some of the archaeological evidence.

  77. Oddly, Ethnologue-ers (Eth-nologers? Ethno-loggers?) are splitters at the language level and lumpers at the phylum level. Why that should be I cannot tell.

  78. Could it be that a lot of people left the forest to become farmers or pastoralists elsewhere, as in latter-day migrations from village to city? I doubt it myself, but it’s worth asking.

  79. Oddly, Ethnologue-ers (Eth-nologers? Ethno-loggers?) are splitters at the language level and lumpers at the phylum level. Why that should be I cannot tell.

    Linguistic syncretism.

  80. David Eddyshaw says

    Subsidiarity. It’s all an EU plot.

  81. Trond Engen says

    Y: Could it be that a lot of people left the forest to become farmers or pastoralists elsewhere, as in latter-day migrations from village to city? I doubt it myself, but it’s worth asking.

    I’ve been thinking the same thing. They do draw a merged quasi-population curve of Bioko Island and the southern woodlands, which follows the opposite pattern, with hardly any pots=people until about 500 CE, but the numbers are small. It would be interesting to see a similar study for the whole of Bantu-speaking Africa.

  82. David Eddyshaw says

    All Hatters will surely be glad to know that the Yoruba ikú “death” (which features prominently in the above article) is cognate with the Kusaal kūm “death” (kùm- as the first component of a compound); the stems are segmentally identical and the tonal correspondence is also regular, but Kusaal belongs to the Northwestern, noun-class suffixing end of Volta-Congo, and Yoruba to the Southeastern, noun-class prefixing end (though in Yoruba, the prefixes have become fossilised and invariant.)

    The corresponding Proto-Bantu (apud Meeussen) is -kú̧-o, where the root again matches exactly ( is Bantuist-speak for tense [u], as opposed to [ʊ].)

  83. I am indeed very glad to know that!

  84. David Eddyshaw says

    Just to demonstrate the wonders of the comparative method, the Swahili verb fa “die” is also cognate, from the selfsame root as the “death” words above; likewise kifo “death” (where ki- is, of course, a noun class prefix, not the root, and is probably of the same origin as the Yoruba i- in ikú.)

  85. Still, Burkina Faso Toende speakers, who haven’t been exposed to Agolle Kusaal to the same extent as their Ghanaian Toende cousins, self-report quite low levels of comprehension of the Agolle dialect

    self-understandings and levels of exposure:

    when i visited brno, in 2002, i remember being told by czech-speaking folks i met there that they (in their 20s) found slovak entirely comprehensible, but that their younger siblings who’d grown up mainly after the velvet divorce could barely understand it at all. my polish-cradle-tongue traveling buddy had no serious problems with understanding either, but found slovak a little easier.

  86. David Eddyshaw says

    My Agolle-speaking informants seemed to have little difficulty understanding Toende Kusaal, so the situation appears to be asymmetrical (although a confounding factor may be that my informants were all linguistically sophisticated polyglots, even compared to the West African norm.)

    I think the asymmetry is probably real, though; factors may be a negative attitude to the supposedly “mixed” Agolle dialect on the part of Toende speakers*, and the fact that Toende Kusaal objectively is phonologically more conservative in many respects, so the situation is rather like Castilian Spanish versus Portuguese.

    *Agolle Kusaal has many more speakers, is spoken around Bawku, the only sizeable local town, and is the basis of the written language; still, traditional disdainful attitudes die hard.

  87. does Ethnologue count Twi and Fante as two distinct languages, for example?

    They do, and code them ‘tw(i)’ and ‘fan’ respectively. But there is a third code ‘ak(a)’ for situations in which one does not wish to distinguish. Similarly, there are codes for each of 16 Sinitic languages (which are certainly linguist’s languages) and a code for Chinese-as-a-single-thing (which, however, excludes Dungan because of its separate sociolinguistic development, even though it is linguistically part of the Mandarin dialect continuum).

  88. David Eddyshaw says

    Depending on what one is using the codes for, that seems actually quite reasonable; however, it means that Ghana (for example) has 80 Ethnologue codes, rather than eighty languages. Eighty seems quite plausible for the total given such a fine-grained approach.
    (Does Ethnologue have separate codes for the two Kusaals?)

    Akan (in the macrolanguage sense, rather than the much broader ethnic sense) has other dialects besides Twi and Fante which ought to have their own codes on this basis, such as Akuapem (at one stage a candidate for the single Akan literary language) and Bron, which Ashanti find hard to understand, though Bron speakers understand Twi well enough (another of those asymmetrical situations.)

    Oop North, both Farefare and Dagaare are dialect continua, and you could make two or three languages out of either of them without getting too silly. Even Mampruli has dialects with and without grammatical gender, which might merit ethnologoschisis.

  89. David Marjanović says

    but that their younger siblings who’d grown up mainly after the velvet divorce could barely understand it at all

    Well, not after five seconds of exposure. Ten seconds should do the trick.

    An asymmetry that doesn’t seem to be related to social factors is that Czechs understand Polish “right away”, while the other way around requires two days of exposure. Palatalization is easier to ignore than extreme vowel length, I suppose.

  90. David Eddyshaw says

    A maximalist approach to labelling would account for both Ethnologue’s splittery when it comes to individual languages and its lumpery when it comes to families and phyla; presumably the idea is to have a standardised label available, without prejudice as to the precise linguistic status of the labelled.

    Given that both language-versus-dialect questions and the higher-level-groupings questions can be anything from works-in-progress to questions-of-definition to highly political hot potatoes, that’s probably not unreasonable, so long as people aren’t misled into thinking that the labels represent settled-consensus linguistic facts.

    (In days of yore, when Ethnologue still was available to mortals, I was deeply unimpressed by its entries for Western Oti-Volta languages, which, though based on legitimate published sources, gave no indication to the unwary that the legitimate published sources in question were often both old and unreliable.)

  91. Palatalization is easier to ignore than extreme vowel length, I suppose.
    Don’t forget the umlaut. At least, that’s what trips me up most often when I’m trying to understand Czech based on my knowledge of other Slavic languages.

  92. David Eddyshaw says

    (On Ethnologue)

    “And even I can remember
    A day when the historians left blanks in their writings.
    I mean for things they didn’t know.”

    https://www.poemhunter.com/poem/canto-13/

  93. I quoted Canto XXV three minutes before you quoted XIII! Synchronicity strikes again…

  94. David Marjanović says

    Speaking of things I didn’t know… umlaut? I’ve never noticed, and I’ve (slowly) read most of a MSc thesis in Czech. What do you mean?

  95. Stuff like Slavic /ju/ > Czech /ji/ etc.

  96. David Marjanović says

    Ah, that. That doesn’t quite register to me; maybe I’m used to vowel randomness from German(ic).

  97. John Emerson says

    The quote from Pound on leaving blanks traces back to lament from Confucius, though XIII is not one of the Confucian cantos.

    Rashid al-din, though not a Confucian, did leave blanks in his history of the Mongols where he was uncertain about names.

    The transmitters of the Daodejing,by contrast, allowed themselves enormous freedom, which is consistent with that book’s unenthusiastic attitude toward verbal expression in general

  98. PlasticPaddy says

    So Jiri = Yurii. There is a Czech/Bohemian saint Jitka with name day 5 Dec., but the name Judita is a competing version for Judith. Julie=Julia/Yulia (no Jilie).

  99. David Eddyshaw says

    I’m used to vowel randomness from German(ic)

    All languages worth bothering with have umlaut: Welsh, Czech, Kusaal … what more could anyone want?

  100. …Takelma…

  101. … Chechen (mil(a) ‘who (sg)’ -> mülsh ‘who (pl)’.

  102. it means that Ghana (for example) has 80 Ethnologue codes, rather than eighty languages.

    I’m reasonably sure that the number excludes macrolanguage codes.

    (Does Ethnologue have separate codes for the two Kusaals?)

    No, it doesn’t, nor are there any BCP 47 language variant codes for Agolle and Toende, because variant codes are created only on request (if you would like to request them, let me know).

    However, there are variant codes for Akuapem and Asante. The underlying story is complex. Before Ethnologue was harmonized with ISO 639, the former had codes for Twi and Fanti, but the latter had codes for Akan and Twi: probably no one remembers why this was so. In order to align the two systems, Akan was promoted to a macrolanguage encompassing Twi and Fanti. Then someone requested the creation of the variant tags, so one can tag spoken or written materials as “ak(a), “fan”, “tw(i)”, “tw-akuapem”, or “tw-asanti”.

    Don’t forget the umlaut.

    In the Romance languages, a similar process is called metaphony: following /i/, /j/, or in some languages /e/ causes the stressed vowel to be raised rather than rounded. Tuscan, and therefore Standard Italian, does not have it, perhaps because it stands on the boundary between full metaphony and metaphony from /i/ only. See Wikipedia for details.

  103. Savalonôs says

    Roger Blench had a couple articles up on Academia recently about Modern South Arabian. One (“Semitic upside-down: a new proposal for the origins of Modern South Arabian languages”) argues that a) the Modern South Arabian group is genetically related and has a fairly recent common ancestor, rather than being just a regional description; b) Semitic urheimat in East Africa, spreading to Asia via south Arabia; and c) a subphylum of Afro-Asiatic (“Southern Afro-Asiatic”) which includes Semitic and Chadic.

    Blench uses the term “Modern South Arabian” and does not propose an alternative. “Eastern South Semitic” seems like a good enough alternative if it’s really just a regional descriptor. Otherwise, I propose “Rawremic”, from the Mehri rawrem “sea”, a common ESS etymon.

    https://www.academia.edu/43019644/Semitic_upside_down_a_new_proposal_for_the_origins_of_Modern_South_Arabian_languages

  104. Thanks! Here’s the abstract:

    The Modern South Arabian (MSA) languages constitute a significant outlier in Semitic, maintaining a phonological inventory which has all but disappeared elsewhere in the family. Although they have been thought to represent an archaic migration to the south of the Arabian Peninsula, their high level of lexical similarity argues that the current MSA languages dispersed quite recently. Despite being in neighbouring regions, they seem to have no special relationship with Epigraphic South Arabian (ESA), the cluster of inscriptional languages found in Southwest Arabia. However, they do have a distinctive relationship with Ethiosemitic. Ethiosemitic is highly internally diverse, suggesting that it must have been speakers of MSA languages who migrated across the Red Sea from Ethiopia relatively recently. Archaeological evidence points to the ancient settlement of the Arabian Peninsula by pastoralists, and thus MSA speakers must have replaced an earlier pastoral population.

    If this is so, then something is wrong with our understanding of the genesis of Semitic and its relationship with the other branches of Afroasiatic. The paper proposes there was a ‘South Afroasiatic’ node consisting of Semitic, Cushitic and Chadic, co-ordinate with Egyptian, which developed after the split with Omotic. This node was characterised by a complex phonology which included both lateral fricatives and an ejective series. Although these features survive in peripheral branches of Chadic, Cushitic and Semitic, they have disappeared in Ethiopia proper, as a result of a levelling process which took place after the dispersal of MSA. The paper proposes that MSA languages resulted from a seaborne migration of pastoralists from the Ethiopian mainland around 500 AD.

    I find this sort of thing fascinating.

  105. David Eddyshaw says

    I think we already talked about “Semitic Upside Down”:

    http://languagehat.com/trasks-historical-linguistics-online/#comment-3929908

  106. I never know what to think of Blench. I’m pretty sure he’s some mixture of genius and of quite wrong, but at what ratio I can’t tell. I don’t know much about most of the many fields he works on.

  107. David Eddyshaw says

    He does a great deal of fieldwork, especially in the way of collecting word-lists and the like from languages that nobody has ever studied in detail, if at all. I find the resulting papers frustrating because of their lack of detail (my own temperament being much more in favour of narrower focus and greater depth), but it has to be said that he is the sole decent source for quite a lot of languages; and I get the strong impression that he’s pretty accurate in what he does cover.

    He’s published a lot of very interesting and detailed ethnobiology stuff, which I reckon is probably his greatest contribution. His heart is definitely in the right place when it comes to language study belonging in its proper ethnological setting.

    He’s published a lot on language relationships, almost all of which I think is poorly thought through and/or just wrong; part of this impression of mine may just be an artefact of the fact that he is something of an uberlumper, whereas I’m – not.
    Moreover, it has to be said that there isn’t (outside of well-studied low-level domains like Comparative Bantu) anything like a firm consensus on higher-level language relationships in Africa, so he can hardly be accused of being a maverick in his opinions (though do I get the impression that the consensus is slowly shifting away from Greenbergian ultralumpery as better data become available and more people get involved in real comparative work.)

    I’ve never come across a paper of his that I didn’t at least find interesting.

  108. January First-of-May says

    anything like a firm consensus on higher-level language relationships in Africa

    For what it’s worth, my own amateur impression is that 1) higher-level language relationships in Africa probably exist [even outside Afro-Asiatic and Bantu], and 2) good luck picking them out with that kind of data. For that matter, exactly the same, except even worse in terms of data, applies to the Americas and/or New Guinea.

    I was never much of a fan of the current situation in Amerindian linguistics; logically one should have expected a single branch from whatever people originally crossed the Bering Strait back around 16k BP or thereabouts (I forgot what the current consensus date is), with possible later influxes from later migrations that could be accounted for (Dene-Yeniseian seems to be one of those cases), but even if the 16-kiloyear branch is unreconstructable, it feels unlikely that we have that many families going on, particularly in South America.

    OTOH, with the twin attacks of Greenberg (lumping everything too closely together without regard to what the data actually looked like) on one side and Chomsky (pushing everyone away from that sort of “irrelevant” thing at all) on the other, I understand how the opinion had never moved away from the utter-loads-of-families consensus, and by the time (if ever) it gets unfixed from that place it would probably in most cases be too late to collect more data (and the exceptions tend to belong in well-established subfamilies anyway).

  109. David Eddyshaw says

    higher-level language relationships in Africa probably exist [even outside Afro-Asiatic and Bantu]

    Oh, yes. No question about it. Most of “Niger-Congo” is undeniably valid, for example. Kusaal is, beyond reasonable doubt, genetically related to isiZulu.

    The data are often not so much the problem nowadays; it’s more the legacy of premature classifications made without proper evidence, which have been sanctified by long acceptance without ever being properly scrutinised.(But this is changing now.)

    But even someone as splittery as I am would readily concede that there will almost certainly turn out to be not nearly as many independent unrelated families in Africa as in North America (which is itself odd, when you think about it.)

  110. Dimmendaal’s book allows for Niger-Congo, minus Mande. A 2019 handout by Güldemann (“African language classification beyond Greenberg”) says:

    + against the currently available diagnostic evidence, not all units conventionally subsumed under Niger-Kordofanian are robust members – subsets according to likelihood
    a) robust: Benue-Kwa (>20), Dakoid, Atlantic (7), Gur (8), Adamawa (14), Ubangi (7)
    b) promising: Kru (except Siamou), Pere, Dogon, Bangime
    c) less likely/weak: Kordofanian (4), Katlaic/ Ijoid, Siamou, Mande

    Maybe Blench used to be an enthusiastic lumper, but now he’s an enthusiastic splitter. He has called for considering a number of the languages of NE India isolates, not Tibeto-Burman as they had been onsidered before. I can’t judge the data, but in general I think that, in the Old World especially, people have been reticent to declare new phyla (=isolates), and have been biased toward shoehorning languages to where they don’t belong.

    Back to Africa, I like the map in Dimmendaal’s book (map 10, p. 327) illustrating his point that most African isolates are right on the edges of language spreads. That explains the Americas: fewer big spreads=more small families.

  111. David Marjanović says

    He has called for considering a number of the languages of NE India isolates, not Tibeto-Burman as they had been onsidered before.

    Here’s the “Declassifying Arunachal” paper. It’s a good read.

  112. David Eddyshaw says

    Maybe Blench used to be an enthusiastic lumper, but now he’s an enthusiastic splitter.

    Oddly (to me, at any rate, who am catholic in my splittism) it seems quite common for people to be lumpist about one phylum/family and splittist about another, without any particular basis in the actual degree of evidence as far as I can see.* Dimmendaal, who in general seems properly sceptical to me, is mysteriously gullible/enlightened about Nilo-Saharan. Of course, he knows a great deal about at least some parts of Nilo-Saharan; but it seems to me that if anything, the usual tendency is for in-depth knowledge of a group of languages to correlate with increased splittism rather than the reverse. Of course, he may simply be right.

    * The disquieting conclusion that should probably follow is that such things are much less evidence-driven in reality than it is comfortable to believe (a conclusion valid for all sorts of belief, after all.)

  113. Güldemann’s graded approach seems better to me. It at least suggests someone looked at the data critically and without a bias.
    Have you seen Lyle Campbell’s 1997 American Indian Languages? He has a section with many different proposals for larger-scale genetic groupings, and he gives them each two marks: “probability”, meaning how likely the hypothesis is to be true, and “confidence”, how sure he is of the evaluation. So Tlingit-Eyak-Athabascan gets +75%/40%, Aztec-Tanoan gets 0%/50% (= “IDK, I guess”), Zuni-Penutian (likely a hoax proposal) gets -80%/50%, Nostratic-Amerind (!) gets -90%/75%, and Takelma-Kalapuyan gets +80%/60% (wherein he gets caught as an enthusiastic lumper; our Penutianist thinks the proposal is unlikely).
    I have some gripes about Campbell, but I like this idea.

  114. David Eddyshaw says

    In fact, I’m not sure that Blench is actually inconsistent across phyla; it’s more a question of taxonomic level. He is (to my mind) much too prone to accepting grand overarching schemes without anything like sufficient evidence (he’s toyed with Niger-Congo-Nilo-Saharan, even), but he’s also rightly identified individual African languages as probable isolates; IIRC he was the first to point out that Bangime was certainly not not Dogon and probably not even “Niger-Congo.”

    @Y:

    Yes, I have seen (indeed, own) American Indian Languages; while I’m a paid-up Campbellite in many respects, I must say I have my doubts about where he gets his numbers from there.

    I also get the impression sometimes that Lyle Campbell has sort of painted himself into a corner, where Old Believer Orthodox Splittery has become so much his particular thing that he feels obliged to put on a show of maintaining the True Path in a naughty world in all circumstances, as if we’ll all slide back into a Proto-Sapiens morass if he doesn’t keep the faith.

  115. John Cowan says

    Exactly. As I pointed out back in 2014, Algic is the most recent phylum proposal that Campbell accepts as correct. But he recognizes a Yurok-Wiyat node called “Ritwan” which nobody accepts today. So he isn’t an anti-lumper, he’s just a get-off-my-lawn conservative: if it was proposed less than a century ago, it’s Not Proven. (I said in my earlier comment that it had to be established a century ago, which is wrong, which is why I’m not linking it.)

  116. marie-lucie says

    From Africa back to North America: Swadesh’s hat

    Perhaps those of you reading the word “hat” saw in their mind the pre-WW1 Western bourgeois’s obligatory headgear worn when out of his house. But among Salishan and other people of the rainy NW coast, a waterproof hat has always been a necessity. These hats were made of very tightly woven cedarbark strips, and in several communities some people – men and women – have taught themselves the techniques of harvesting and treating the bark before weaving it. Some hat-weavers also finish the hat by painting a traditional design on it. Making one of these hats is very time-consuming as well as highly skilled, so they are very expensive. Many people save in order to afford such hats for themselves or close relatives in order to wear them during ceremonies.

    Among the Tsimshianic languages the name for a hat is “qayt” or “qa:yt”. This word cannot be from Proto-Tsimshianic, otherwise the central vocalic element would not be identical. I don’t know from which (probably Salishan) language the word must have been borrowed.

  117. marie-lucie says

    In Lyle Campbell’s 1997 American Indian Languages….. Takelma-Kalapuyan gets +80%/60% (wherein he gets caught as an enthusiastic lumper; our Penutianist thinks the proposal is unlikely.

    Your Penutianist here to the rescue (not of LC’s proposal, which is originally based on a misleading article by Frachtenberg 1918.).

    The other encyclopedic book to mention the topic (Mithun 1999) quotes approvingly from the handout on “T-K” (Tarpent & Kendall 1998, still unpublished but confirmed by further research (T 2015). More precisions will be included in my book in progress on Penutian.

  118. @marie-lucie, thank you! I did imagine bourgeous’s headgear, and also a Russian hat (to see if it is indeed something cross-culturally basic), but I realized that its presence in the list is motivated by Salishan reality and was curious.

  119. marie-lucie says

    Thanks drasvi!

    I just realized that the people who order the hats in question are mostly women. The ceremonial attire for high-ranking men includes a headdress with a carved wooden frontlet above the forehead, made for the wearer with a design appropriate to the man’s rank and clan symbol, but there is no traditional equivalent for women, so nowadays women wear the woven hat for that purpose. Sometimes grandparents order woven hats for their grandchildren for the same reason.

    The social use of these hats that I am describing reflects my experience among the Nisqa’a ( one of the Tsimshianic-speaking groups), so details may vary with Salishan and other groups.

  120. John Cowan says

    I have always held with Asimov that there are only two kinds of hats worth wearing: the Russian, against cold, and the Vietnamese (and apparently PNWC), against rain. And even then I’d rather wear hoods than hats: harder to leave behind.

  121. I wanted to see what these Salishan hats looked like. The link below shows a Haida version, along with a cedar strip raincoat!
    https://cifunderground.wordpress.com/2015/11/23/16-11-15/comment-page-1/

    Ignore the kitten at the top and scroll. Or enjoy the kitten at the top as suits you.

    JC, you may not be out as often as I am, or have darker skin. Protection from the sun is a necessity for some of us.

  122. Thank you for finding those images, Ryan!

  123. marie-lucie says

    Thanks! The Haida hats shown are not quite the same shape as the Nisga’a ones (which are less flat) but otherwise the same general model and techniques. I have never seen a real cape like the cedarbark one shown (with fur edging), but there are many photographs of coastal people wearing cedarbark clothing (for instance in Curtis’s work). Most of this clothing consists of blankets wrapped around the body with a belt. The coat shown here looks rather ragged, no longer waterproof, probably from long use.

  124. John Cowan says

    JC, you may not be out as often as I am, or have darker skin.

    I am quite pale, and in fact don’t go out in the sun much (neither did Asimov).

  125. (Mithun 1999)

    This and queens here reminded me about my old question. Mithun has: “Thomas Jefferson, an early president of the Society, encouraged the wide collection of vocabularies, sending out questionnaires and interviewing speakers, partly at the suggestion of Catherine the Great of Russia.” (p. 5)

    I know that they corresponded, but I was surprised that they discussed this and I was not able to find anything about it (it was a decade ago though). Does anyone know any details or where to look for them? I know that in Paris he tried to get a permission from her for an expedition that wanted to reach the west through Russia (Catherine called the plan ‘chimeric’, they tried anyway, disguiced as merchants, but were intercepted in Siberia). But it is a different story.

  126. One of Catherine the Great’s pet projects was COMPARATIVE DICTIONARIES OF ALL LANGUAGES ​​AND DIALECTS COLLECTED BY THE RIGHT HAND OF THE ALL-IMPORTANT PERSON (I know the title translation sounds stupid, but it is barely better in Russian).

    https://www.prlib.ru/en/node/371073

    In 1795, Catherine wrote to Thomas Jefferson asking to contribute translations of several hundred words from Shawnee and Delaware languages for this dictionary. George Washington also contributed a translation of Cherokee and Chocktaw words.

  127. David Marjanović says

    Why “important”? Всевысочайшая strikes me as a calque of allerhöchste, “the very highest”, “the highest of all”.

  128. In 1795, Catherine wrote to Thomas Jefferson asking to contribute translations of several hundred words from Shawnee and Delaware languages for this dictionary. George Washington also contributed a translation of Cherokee and Chocktaw words.

    That’s absolutely amazing. Russo-American relations would be entirely different if this historical tidbit were taught in schools!

  129. Three founders of North American Indian linguistics – Catherine the Great, Thomas Jefferson and George Washington…

  130. What her left hand was doing that moment?

    P.S. ah, it was десница, as in יְמִינְךָ . SFReader thank you!

  131. “…As for Suzdal dialect, it is mixed, in part from arbitrary words, in part from Greek words turned into Russian, the same way as German language used by the Jews, distorted by Jewish words. The trade that is led from Suzdal as far as to Greece can be a reason for this change.

    Among the Celtic dialects one should understand [sic] Basconian proper, so called in France and not the one that is known in Spain by the name vascuenza, which is not similar at all to Celtic and which will be discussed specifically in the second part.”

    Yheir Indian informants are Indians living in Astrakhan.
    I knew that India was… hm, the main trading partner of cities like Bukhara but it just did not occur to me that there was an Indian diaspora on Volga:/
    Astrakhan is an extension of Central Asia/Persia, but it just did not occur to me.

  132. I think the translation is wrong. Catherine the Great’s Ned Stark was Prince Potemkin who doesn’t seem to have been involved in the project at all.

    The Latin text says simply “Augustissimae cura collecta”.

    By the way, it turns out, Catherine the Great was an early proponent of the Proto-World theory and her comparative dictionary was supposed to prove it.

    Plusieurs raisons expliquent le soutien de Catherine II au projet. L’impératrice s’intéresse aux langues, en particulier à la question d’une origine commune de toutes les langues, abordée par exemple par Antoine Court de Gébelin, dont elle a lu les travaux. Il s’agit également d’établir que c’est la langue russe qui est à l’origine de nombreuses autres langues.

    The last sentence is intriguing.

  133. https://en.wikipedia.org/wiki/Razihi_language

    Funnily, Razihit (but of course English nisba adjectives are not gendered…).

  134. David Eddyshaw says

    Toende and Agolle Kusaal

    I’ve myself sometimes cited the survey reported in

    https://www.sil.org/system/files/reapdata/96/97/71/96977114718357064414587880573428680498/SILESR2001_009.pdf

    which concluded that there were “84% apparent cognates between the agole (Bawku) and the tonde villages (Groff 1983:5)”, which has always struck me impressionistically as surprisingly low.

    It just struck me today that Berthelette (excellent man!) actually gives the data that this is based on, so I decided to go through the list myself (leveraging the much better lexicographic data now available.)

    I make it something over 96% …

    (It’s also interesting, in the light of actually understanding Kusaal, to see how the helpful informants have inventively coped with being asked to give single-word equivalents of single, often polysemous, French words presented to them in isolation …)

Speak Your Mind

*