Y wrote me as follows:
Semiticist Benjamin Suchard retired his Twitter account in favor of a blog. His specialty is biblical Hebrew, but he’s interested in everything else Semitic and Afro-Asiatic. Check it out.
(The site design is awful but I’m sure it’ll get fixed sooner or later.)
The design really is awful, and there’s not even an About page [There is, just click on WHO? — thanks, Jen! –LH] where one might hope to learn why it’s called Bnuyaminim (if indeed it is — that’s in the URL but not on the blog itself), but never mind, the posts look interesting: August 10, Is Arabic Cushitic?; August 9, Is Beja Berber?; August 8, The genitive in Semitic and Cushitic (and Beja), etc. His now abandoned Twitter page said:
Comparative Semiticist trying to leave his dark Indo-Europeanist past behind him. /ˈsuʃɑɹd/, /ˈsyʃɑɹt/, or /syˈʃaʁ/ preferred; /ˈsʌkhɑɹd/ dispreferred.
Which is fun, but I kind of wish he’d said which of the three preferred versions is his own usage. Thanks, Y!
You need to look under ‘WHO?’ 🙂
Ah, thanks! “*bnu yamīnim is how I’d reconstruct the name ‘Benjamin’ back to Proto-Semitic.”
To me, /ˈsuʃɑɹd/ looks like an English pronunciation, syˈʃaʁ/ French, and I would guess /ˈsyʃɑɹt/ is Dutch (although I’m curious about the /ɹ/ there). Perhaps his own usage differs depending on the language being spoken?
Very likely!
Bunched [ɹ] in coda is pretty common in Netherlands Dutch, called (TIL) “Gooise R” and “associated with the upper middle class and social climbers”, which presumably includes former Indo-Europeanists.
Heh: “Synonym: bekakte r.”
Is Beja Berber?
Well, I’m going to go out on a limb here and say: No. No it isn’t.
The “evidence” is proclitic Beja articles which look similar to some Berber noun prefixes, but the only really striking similarity is the feminine t, which is undoubtedly proto-AA and thus of no use at all for subclassification. The Beja articles mark case, too, which Berber prefixes don’t.
The development of flexional prefixes from agreeing “articles” in fundamentally suffixing languages is familiiar in Niger-Congo. Needless to say, it’s happened in Oti-Volta, e.g. with Ditammari and Konkomba, which are much more closely related to prefixless Nateni and Moba, respectively, than to each other.
I’m psyched to see someone leaving Twitter to start a blog.
RSS readers would be more widely adopted if they had a more inviting name than RSS reader.
In Dutch, we wouldn’t use the ʃ there, rather the x, unless you would know it is a French name.
https://www.universiteitleiden.nl/en/staffmembers/benjamin-suchard#tab-1
has an email (and telephone)
The wikipedia article for Lior Suchard (of an age to be his brother) gives pronunciation as ” [ ˈsuʃaʁd]”, another variant.
Hi everyone, thanks for the shout-out! Good work on the last name (I’m not related to Lior though).
Obviously, Beja is not Berber (nor is Arabic Cushitic; you heard it here first). But there are a few more similarities than just the t-, enough to really make me wonder whether this isn’t a shared Beja-Berber development. There’s also some things in the verb that are different from what we find in Semitic, but hard to say at this point whether that’s innovative or archaic.
Hope to see some of you in the comments!
For some reason he’s put the heading of his blog in a column of its own to the left taking up half the page. If you resize your browser to make it narrow like a mobile phone, it goes on top like it should. It looks to me like it would be very easy to fix (although I don’t know WordPress…)
The wikipedia article on the other (unrelated) Suchard says that: “In 2010 he placed 28th on People Magazine’s Sexiest Men Alive list.” Which naturally raises the question of who the highest-ranked comparative Semiticist and Indo-Europeanist, respectively, on that list were.
Not in People Magazine, but in Diachronica.
(Comparing linguists is the domain of comparative linguisticians.)
I suppose the title of the subsequent post, “Is Arabic Cushitic?” is as tongue-in-cheek as the earlier one, only more obviously so. It all comes to how some innovations, thought to neatly follow a family tree, appear in unexpected branches and ruin everything.
Nice blog. Exactly that tone of interrogative speculation that makes me both wiser and hungry for more.
To the renaissance of blogs! And time to set up an RSS feed again.
To the renaissance of blogs!
Amen!
@David Eddyshaw Berber prefixes most definitely do express case!
iwət a-rgaz “he hit the man”
iwət t u-rgaz “the man hit him”
The annexed state u- fem. tə- is used for marking the post-verbal subject and post-prepositional nouns.
The free state a- fem. ta- is used for marking the oblique (and thus marks direct object)
Of course, as PhDnix knows, Kabyle Berber prefixes have been argued to express something typologically unique that is not case due to its interactions with information structure. But, even if one accepts Mettouchi and Frajzyngier’s arguments to this effect, they don’t apply to most other Berber languages, where information structure has less of an effect and it seems much more like a prototypical marked-nominative two-case system.
Either way, I was evidently wrong.
The latest is an interesting summary of recent papers by Jungraithmayer. Chadic has long been considered iffy as a member in Afroasiatic status, and he argues that that comes from over-reliance on Western Chadic languages such as Hausa, whose morphology bears only a weak resemblance to that of other AA languages. J shows that Eastern Chadic has much clearer connections to other AA families. Suchard says,
Chadic has long been considered iffy as a member in Afroasiatic status
No competent scholar now doubts it.
It was only comparatively recently that the AA affilation of Chadic became the consensus, admittedly. And it is true that some Chadic languages are much more obviously AA than Hausa: Lameen had a post on this on his blog, in fact.
Apparently, Chadic is the least-obviously-Afroasiatic family that ended up being included in the phylum
No, that’s Omotic.
One easy way to illustrate this is by looking at whether a family preserves the Afroasiatic prefix and/or suffix conjugations of the verb
From which we conclude that Egyptian (no trace of the prefix conjugation) is not Afroasiatic?
May I suggest looking a a Hausa grammar? There are some rather good ones. There are actually reasons why Chadicists don’t simply equate the “subject markers” with Semitic verb prefixes.
There is a lot of literature on this. The Chadicists are well into comparative-historical stuff.
Isn’t Omotic excluded (after Theil)? Or is there more recent lumping to be read?
It ought to be excluded, but inertia is powerful.
I imagine that Jungraithmayer’s proposition leans heavily on the evidence of Mubi, which is indeed fond of Semitic-style ablaut
http://lughat.blogspot.com/2020/01/mubi-plurals-from-arabic.html?m=1
(and other posts from Lameen.)
There are, however, plenty of East Chadic languages much less obviously AA than Hausa (I’m looking at a grammar of Lele.) There just isn’t some East-West cline where tone progressively ousts ablaut. And Mubi has tone (like all Chadic) and Hausa has ablaut.
Probably the least AA-like of Chadic languages, the near-isolating Goemai (Western Chadic) retains just one bit of inherited verbal morphology: making pluractionals by vowel change.
All this is also making the same sort of error for Semitic and AA as has been made in Niger-Congo with Bantu: the best-documented branch is just assumed to reflect the protolanguage closely and all non-Semitic/non-Bantu features are routinely attributed to degeneration or substrates or whatever.
Most AA languages are Chadic.
I find it more useful to think in terms of a North-South cline: to a first approximation, the further south you go in Chadic, the less archetypally AA the languages look (less ablaut, no gender, weirder pronouns…)
The final write-up of that series on Mubi is here: Pattern borrowing and hybridization in
Mubi (East Chadic): The importance of congruence.
Yes, it certainly is the case that Chadic languages have been influenced by languages from other language families, often pretty profoundly. (Hausa seems to have borrowed the word for “two” …)
Goemai is a kind of ne plus ultra of this, in fact. Birgit Hellwig’s exemplary grammar goes into some detail about the areal phenomena involved.
Perhaps one needs some familiarity with Niger-Congo languages to be aware of how very distinctive and different the vast majority of the Chadic languages remain. The entire morphological mechanism of Hausa (allegedly one of the only-dubiously AA ones) is radically different from anything in Niger-Congo. One of the most striking differences is in actual fact the pervasive ablaut: there’s nothing at all comparable in Niger-Congo. This remains highly characteristic of Chadic in general: in fact the Central Chadic languages seem to regard pretty much all vowels are mere playthings of the morphology, particularly in the verbal system.
I think someone starting from the Semitic languages and assuming that they are normative for AA in all respects is almost bound to end up interpreting perfectly everyday sub-Saharan language features like tone as evidence of deep substrate influence, simply because it’s not familiar to them in languages they actually know anything about. Hausa, in fact, has a strikingly simple tone system compared with typical Niger-Congo languages: that seems to be common in Chadic too.
[Very nice paper, Lameen. This sort of extensive borrowing of morphological patterns is fascinating. I suspect it’s happened a good bit in Niger-Congo, too: Greenberg’s idea that the NC noun-class system is so unusual that its mere presence guarantees that a language is genetically Niger-Congo begs so many questions …]
I wanted to know how to pronounce Goemai, so I went to Wikipedia, where I discovered this horrifying information: “The spelling Goemai originates from the 1930s. Orthographic oe stands for the mid-central vowel ə, a practice that had been adopted by missionaries working among the Goemai in Shendam during the 1930s, such as Father E. Sirlinger.” What the hell is wrong with people? Why would you do that??
And of course it leads to absurdities like Russian Гоэмаи.
Hellwig says that the actual speakers like the convention; apparently Sirlinger’s catechism played a major role in promoting literacy in Goemai. Anyway, she keeps it.
And of course it leads to absurdities like Russian Гоэмаи
This (rather nice, though not completely accurate) map of Gur languages has some similar mechanical transliterations into Russian letters that don’t really work:
https://ru.wikipedia.org/wiki/%D0%93%D1%83%D1%80_(%D1%8F%D0%B7%D1%8B%D0%BA%D0%B8)#/media/%D0%A4%D0%B0%D0%B9%D0%BB:Gur-ru.png
If we assume that Fr. Sirlinger was a native speaker of German, it may be significant that “oe” is how you spell ö when you don’t have umlauts ready to hand in your typeface, and the sound so denoted (IPA /ø/) is probably, in German phonology, the closest in vowelspace to ə, with German being one of the many languages that has no specific orthographic representation of ə because it just happens “naturally” when syllables are unstressed. If he was instead a native speaker of French, French orthography also sometimes uses the ligature œ for /ø/.
On the non-phonological front, here’s a photo of Fr. Sirlinger in the Shendam region in 1923. On a motorcycle. https://collections.library.yale.edu/catalog/16713864
UPDATE: Fr. Sirlinger is referred to in various internet sources as being a “French” missionary, but he was reportedly born in Alsace while it was under German rule (in 1887). So he almost certainly was familiar with the orthographic conventions of both French and German.
Ah, that makes sense. Thanks, it makes me less crazed. Though I still think it’s dumb.
And of course the actual speakers like the convention — they’re used to it. Just as English speakers like English spellings, except for a few cranks.
Fr. Sirlinger
I thought he might be a White Father (like the astonishing André Prost) but apparently his organisation “ne doit pas être confondue avec la Société des missionnaires d’Afrique (Pères blancs)”:
https://fr.wikipedia.org/wiki/Soci%C3%A9t%C3%A9_des_missions_africaines
German ö, ASCIIfied oe, is both /øː/ and /œ/. The trick is that:
1) The most widespread value of the fully reduced vowel in German is not in fact [ə], or at the very least not an unrounded one; I think it’s [ɵ].
2) The reduced French vowel isn’t [ə] either as far as I can tell. The IPA insists that the symbol [ə] does not specify an unrounded vowel, and I think that’s specifically so it can be used for French; but I don’t think it’s a central vowel in the first place. I think it’s a front vowel halfway between /ø/ and /œ/.
3) It’s interesting how different perception can be from the vowel chart. The German vowel most similar to a perfectly unrounded BrEng [ɜː] is the diphthong /œɐ̯/ which very much starts out rounded, and the second most similar one is /œ/ alone, rounded throughout. The German vowel most similar to a very specifically unrounded Chinese [ɤ] is… /œ/ again; there are even transcriptions from the beginning of the 20th century, produced in China, that use ö for it. So if the Goemai vowel is anything other than an unrounded [ə] or [ɐ], ö is intuitively obvious.
Hellwig just gives [ə] without going into any fine phonetic detail as far as I can see.
However, /ə/ behaves as a full-status vowel, can be the nucleus of a stressed syllable, and also occurs long (though it’s not clear how robust length distinctions really are in Goemai, it seems.)
Omotic.
(Missing is the next level: “Perhaps Dizoid is AA after all”…)
There is now a scary “Log in to reply” field. Just click on the e-mail (uh… letter) symbol, and everything returns to normal.
A few thoughts here…
1-Chadic languages are Afro-asiatic. Okay. But before talking about their spread in prehistoric times, might it perhaps be wise to establish that ALL Chadic languages have a common ancestor which is distinct from Proto-Afro-asiatic? Within Afroasiatic, nobody doubts that Beja is a member, but there is real doubt as to whether it is Cushitic or not, or indeed whether Beja and Cushitic (assuming for a moment that they are distinct genetic entities) form a subgroup within Afro-asiatic.
2-Jungraithmayer seems to be conflating two utterly distinct matters: the geography of the typological structure of Chadic languages today on the one hand, and the history of the spread of Proto-Chadic (assuming Chadic is a real subfamily: see 1) on the other.
Quoting Suchard: “For instance, the further west you go (roughly speaking), the less important vowels become for aspect marking and the more important tone becomes”.
It is a HUGE leap to conclude therefrom that this geographical distribution of tone- versus vowel-marked verbal aspect must mirror the spread and diversification of Proto-Chadic. For all we know perhaps Proto-Chadic spread from West to East, with no major typological differences arising in the Chadic continuum during this initial spread, and the spread of tone at the expense of vowels as aspect markers (which may or may not have involved language contact) took place in the West at a much later date.
3-Just because a feature in Chadic is similar to a non-Chadic Afroasiatic feature needn’t mean that said feature must be older in all Chadic languages than more areally typical (=West African) features.
There is a good example of this: Greenberg thought he had identified an inherited Afroasiatic feature in Chadic, internal /a/ as a nominal pluralizer: later work by Paul Newman showed that this Afroasiatic-Chadic similarity was in fact an illusion: the Chadic instances of internal /a/ in nominal plurals all appear to be instances of umlaut triggered by a plural-marking /a/ suffix.
4-Indeed, perhaps Proto-Chadic (assuming it existed: see 1) used both vowels and tone to mark aspect: in which case perhaps Chadic languages which make no use of tones for aspect-marking today might now look like non-Chadic Afroasiatic as a result of changes rather than of preservation of an inherited structure.
Actually, unless the vowel-marked system of aspect marking in those Chadic languages which have it can be proven beyond a doubt to be an inherited feature, I do not see how we could exclude the possibility that Proto-Chadic (if it existed, see 1) used *nothing but tone* to indicate aspect, with the use of vowels an innovation: in which case the Chadic languages without tones and with vowels as aspect markers would be the most innovative languages.
5-The above scenario is not as improbable as some Hatters might think: what if Proto-Chadic (Assuming it…okay, by now you know the drill) was originally a very West African-like language, typologically, which spread from South to North, expanding at the expense of various less West African-like languages (some of which indeed might have been Afroasiatic as well, perhaps close genetic relatives of Proto-Chadic and Proto-Berber: I believe I once, here at Casa Hat, pointed out that the more recent spread of Tuareg and Hausa may have eliminated whatever non-Berber, non-Chadic varieties may once have existed between West Africa and North Africa) spoken further North, whose influence caused the more northern part of the Chadic dialect continuum to become less West African-like?
6-Indeed, returning to point 2, the typological profile of a language family (or subgroup of a family) is typically unrelated to the history of its spread. For example, let us take the Romance varieties spoken in what used to be Roman Gaul: the basic divide in medieval/early modern times involves four varieties: Proto-Gascon, Langue d’oc, Franco-provençal and Langue d’oïl. Now, the last two definitely form a subgroup (they share innovations alien to the first two varieties).
Now, a linguist knowing nothing of the history of the spread of Latin/Romance in Gaul might conclude that the Franco-provençal- and Langue d’oïl-speaking parts of Gaul must have been romanized at the same time, before or after the spread of Latin/Romance which gave birth to Proto-Gascon and to Langue d’oïl.
The conclusion would be wrong: the Southern parts of both the Franco-provençal area and of the Langue d’OC area were romanized first (They were part of of the Gallia Narbonensis, conquered by the Romans starting in 125 BC), with the remainder of the Franco-provençal and Langue d’oc areas, plus all of the Langue d’oïl- and Gascon-speaking areas, romanized much later, as a result of Caesar’s conquest of the rest of Gaul from 58 to 52 BC.
Please note, incidentally, that while Gascony and the Langue d’oïl-speaking areas were conquered (=romanized) at the same time, they have nothing in common linguistically (beyond being Romance, of course): indeed Proto-Gascon may well have arisen as a distinct variety before any of the three others did -DESPITE Latin/Romance having been introduced there long after its introduction in (part of) the Franco-provençal and Langue d’oc areas.
7-Oh, and as a postscript: in present-day France the standard has almost wholly supplanted the older dialects/languages. It is worthy of note that the regional diversity found in spoken French today is wholly unrelated to substrate influence: in contradistinction to the now extinct North/South divide (Langue d’oïl + Langue d’oc + Gascon, with Franco-provençal straddling this North/South line), Modern spoken French in France exhibits a basic East/West divide in its regional variants. Which, incidentally, has nothing to do with the chronology of the spread of French from Paris to Eastern and to Western France.
Depending on who you ask, by now it might be also instead Hadza 🙂 proposed recently by Militarevs Sr. & Jr. to be a Dahalo-type case with substratal clicks and not an ultraconserved “Khoisan” hangout after all. Their draft article does not inspire a giant amount of of confidence when e.g. the second piece of evidence cited in favor is /tʰitʰi/ ‘little bird’, but sure enough it’s now proposed at least, and some other lexical comparisons cited do look like they could be onto something.
Indeed!
Incidentally if anyone’s wondering, my blogging is not being slow for too-busy-on-Twxttxr reasons as much as unrelated personal life issues; plans-of-blogging at least continue to be made unabated.
edit:
Looks like Militarev’s already on throwing spanners at that too, siding with a view that Kujarge “is not East Chadic or Chadic at all”, but a third branch of “Berber-Chadic” (for some kind of not-highly-scrutable lexicostatistic reasons).
Incidentally if anyone’s wondering, my blogging is not being slow for too-busy-on-Twxttxr reasons as much as unrelated personal life issues; plans-of-blogging at least continue to be made unabated.
Glad to hear it (apart from the life issues, of course)!
Previously proposed by Starostin Jr. in 2007 – but only in Russian. Apparently the basic vocabulary is unsettlingly AA-like.
Edit: oh, they knew that of course:
Also, apparently the book only came out in 2008, but the chapter was ready a year earlier as usual.
I would suspect that “AA-like basic vocabulary” remains currently very underdefined though and that just comparision with whatever chunk of the family would turn up way too many false positives to be a useful metric. E.g. the Militarevs compare /iŋtʰawe/ ‘nose’ with equivalents in just a diverse handful of Chadic languages; /unu/ ‘person’ with two Chadic languages, a chunk of Omotic, and a single Cushitic one. Several classic actually-widespread-in-AA cases like #lib- ‘heart’, #lis- ‘tongue’, #mwt- ‘die’ not really in evidence (‘to die’ is still /misi/, but rejected on account of no explanation for t ≠ s).
Yeah, the Hadza paper looks like a good illustration of the major problems with mass comparison.
Though not with lexicostatistics in general; that should be improvable with conceptually minimal effort. One would just want to build first, say, the Western Chadic / Central Chadic / Eastern Chadic basic vocabulary lists, then see what parts of those can be combined into the common Chadic one, only then start comparing any of it with the rest of Afrasian (including probably some “upwards reconstruction”), and only after that start checking anything with some not-established-member like Hadza. It would mean the data gets whittled down from a full 200- or 100-list as one goes, but that’s a feature and not a bug, there is no a priori guarantee that all of that can be reconstructed for every dang protolanguage.
Might still be good enough already to book the issue for later investigation, but I don’t have high hopes that investigating “what if this one modern isolate is a primary branch of a 12,000-year old family” is very tractable before you’ve actually reconstructed its proto-language in decent detail.
(Militarev would claim to have already provided himself a detailed PAA though, that’s kind of the proximate issue in any of this.)
A fully satisfactory reconstruction of proto-Central Chadic alone would already be a massive accomplishment, and one I don’t see happening any time soon (a recent review I did expands on the issues a bit.) Western and Eastern Chadic are even harder, and most Chadic languages have histories of rather intense contact to complicate the issue (Hausa’s word for “two” is apparently from Niger-Congo.) In that context, I’m sceptical about ever being able to get very far with Kujarge, for which all we have or are likely ever to have is 100 words recorded from a drunk semi-speaker.
Their draft article does not inspire a giant amount of of confidence
“compared according to regular consonant correspondences and fairly strict semantic criteria”
Ignoring vowels always helps when it comes to finding cognates. Luckily for the Militarevs, you can do this at will when it comes to comparing anything with (or within) AA.
As an excellent example of “fairly strict semantic criteria” we have “mouth”, equated with “lip”, “cheek”, “tongue” and “muscle longitudinal de chaque cóté du cou.”
I notice that Hausa biyu “two”, regarded by actual Chadicists as not AA but a loan from NIger-Congo (as Lameen noted above), is cited as evidence for the AA cognacy of piʰe “two.”
On the other hand
“occupying the same area over 50,000 years”
How clever of the Militarevs to know this!
Gravina’s actual thesis is more impressive than Wolff’s version of proto-Central-Chadic, though open to very much the same criticisms. It is well done, I reckon, but could easily be subtitled “Extremely common palatalising prosodies of unknown function are likely to make proto-Central Chadic forever effectively unreconstructable.”
Ignoring vowels always helps when it comes to finding cognates. Luckily for the Militarevs, you can do this at will when it comes to comparing anything with (or within) AA.
Well, as I have pointed out before, ignoring vowels is a great help in understanding someone speaking to me a kind of German I’m not familiar with.
The consonant remainder can be compared with standard German consonant remainders. Sometimes I have to twiddle one or the other consonant (k/g, t/d etc) as well. Of course if I didn’t know standard German this would not be effective. I suppose a far-reaching knowledge of Germanic sound changes over the millenia might help, but that’s overkill for the purpose at hand.
@stu
I suppose if you are not talking about bread rolls, this is a good strategy. With bread rolls you might be faced with:
Brötchen
Weckle
Schrippen
Semmeln
Bollen
https://youtu.be/TnG3A95Q1lk
Their draft article does not inspire a giant amount of of confidence
No kidding. Apparently the Militarevs have come to believe, like all too many Africanist historical linguists, that if you pile enough bad cognates sets they’ll add up to something convincing. If we ignore all the silly ones, there are a couple that actually look good on the face of it (like “I” or “foot”), but do I really have time to sift them out?
The Berber comparisons include some really annoying howlers:
“Berb. *yaw-n/t (<*yawʕ-) ‘one’": more like *yiwăʔ-ăn/ăt (see Kossmann 2022), and there’s obviously no possible Berber-internal justification for a pharyngeal here.
“Audjila iša ‘sleep’” – is from *ănsəʔ, and the n is part of the root
“Berb. *ʔawk ‘to give’” – no, it’s *ăβḱəʔ; the first radical is definitely not a semivowel.
“AA *(ʔa-)kʷay- ‘not’: =Berb.: Ghadames ak, Fodjaha ənk” – no. Ghadames ak cannot be regularly related to El-Fogaha ənk to begin with (which comes from ənki, as seen in Sokna), and neither has any chance of being reconstructed as a proto-Berber negator.
‘thou’… “Berb.: =Ghat ti-unti (f.)” – yer what? The obvious sources for Ghat record 2f.sg. as kemmounan and cam (i.e. kămmunan, kămm), just as would be expected in any Tuareg variety; ti-unti would appear to be a mistranscription of Ghat tounti “female”, a noun with no particular connection to the 2nd person.
“Berb.: Siwa ta-məǯa (<*mVga) ‘neck’, Timimun ta-məg-n-a ‘head’" – there is no Berber-internal justification for treating the n or the a as a suffix, or for relating taməğğa to taməgna in the first place.
“Berb.: Mzab beddu ‘perdre la raison, e^. fou’, Wargla a-beddiw ‘faible d’esprit, idiot, fou’, Nefusa beddu ‘e^. fou” – all of these are transparently related to the most widespread Berber word for “monkey”, itself frequently compared to Chadic terms for “monkey”.
And seeing all that does not increase my confidence in the forms from branches I know less about…
itself frequently compared to Chadic terms for “monkey”
Hey, I know this one: Hausa biri! (Also, “elderly man whose behaviour is not suitable for his age”, says Newman’s dictionary. Sounds like a plan …)
There is actually a problem with the Hausa word, if Newman’s History of the Hausa Language is to be believed, non-initial *r is supposed to become y/i. He does say that there are lots of exceptions with intervocalic *r, but doesn’t proffer an explanation (though many individual cases are presumably loans.)
I sometimes get the impression (as an outsider) that in Chadic comparative work, Nobody Knows Anything. (This would bring it nicely into line with comparative Niger-Congo.)
“Two” seems to be widely regarded as the least likely numeral to be borrowed. But if it’s just Hausa and Hadza…
Yep. But part of the reason is that in German generally there’s been very little interaction between consonants and vowels in the last 1200 years. There’s been more in Slavic and Romance, for example.
That’s why I feel quite uneasy about Wolff’s decision* to just not reconstruct the tones of Central Chadic. Tones interact with consonants pretty often, and sometimes also with vowels or entire syllables, so if you’re out of luck, you need to reconstruct the tones if you want to reconstruct the consonants.
* Well, I don’t know if it was a decision, really; it’s easily possible that there’s just not enough documented information on the tones of the extant languages. But that would cast all the same doubt on the results as a mere pragmatic decision to reconstruct the tones in the next volume or the one after that.
And in this case, two of them are widespread enough to be considered standard and widely understood (1. Brötchen, 2. Semmel). But there are things for which every dia- and mesolect seems to have a word but Standard German does not, and the words are generally wholly unknown outside the regions they’re used in, e.g. the ends of a loaf of bread…
The heels !
the knubl!
Skalken!
“Two” seems to be widely regarded as the least likely numeral to be borrowed.
To be honest, although this seems to be accepted by Chadicists, it looks pretty iffy to me.
The justification seems to be the negative one that the form doesn’t match proto-Chadic *sər-; however, it is not just found in Hausa, but also elsewhere in West Chadic A, e.g. Bole bolou, Angas vəl, Ron ful (all plagiarised from Newman.)
Newman does not, in fact, identify a source in Niger-Congo in the History, and it looks rather hand-wave-y to me. I presume the idea is that it would be something like proto-Bantu *-badɪ́/-bidɪ́.
However, Bantu itself is of course hardly likely as a source, and the proto-Oti-Volta *-ɹɪ looks a whole lot more likely to represent the continuation of the actual proto-Volta-Congo form. The *ba- in proto-Bantu seems to be a bit of mystery: it would do for “human plural”, but it’s a bit difficult to see why “two” in particular would end up stuck as a human-agreement form when other numbers didn’t.
Pozdniakov’s The Numeral System of Proto-Niger-Congo makes a big deal of “one” and “two” both seeming to reflect something like *di, and actually does cite Oti-Volta, but (not having the benefit of my own groundbreaking contribution) he does not really recognise that the initials definitely differed in POV; cf Moba yènd̀ “one”, (ń)lé “two.”
Anyway, it looks to me suspiciously as if the Chadicists have here committed the sin of assuming that proto-Bantu can be used straightforwardly as a proxy for “Proto-Niger-Congo.”
Be that as it may, the Militarevs are still cherry-picking to get pretend cognates: the actual proto-Chadic form is (according to Newman) both well established and nothing like piʰe.
Endeskiver, how hard can it be. (I can warmly recommend the end slice of a newly baked black rye loaf, ideally with nothing but goose fat and salt. You need to let it cool just enough to be possible to slice, but not all the way. [With no functional gluten it gets its consistency from caramelized sugars, and until it’s cooled it’s basically goo inside]).
WOLD gives “Benue-Congo belu” as the source for Hausa biyu; but I don’t have the chapter handy, so I hope they’re not just quoting Skinner or something
Apparently this is the article I should be looking at (I blush to confess that I don’t recall ever having read it):
Hoffmann, Carl. (1970). Ancient Benue-Congo loans in Chadic. Africana Marburgensia 3(2)
Pozdniakov (4.1.1.2) says
Most of Benue-Congo seems to have initial labials in “two” in fact, though not Kainji.
More significantly, Plateau languages seem often to have *pa or *fa.
(Not Kwa, though.)
Perhaps there was a word for “two” something like *pa, which got concatenated with *ɹɪ in Bantu to bolster the very short and doubtless potentially confuseable inherited form (cf Kusaal (a)yɔpɔi “seven”, where the yɔ element is the expected reflex of POV *-ɹoɹɪ “seven” and the pɔi part appears alone for “seven” in counting: this seems to be an unconnected Wanderwort also meaning “seven”, cf Kassem (n)pɛ “seven.”)
There doesn’t seem to be much evidence for this presumed concatenated form of “two” outside Bantu, though, nothing really like *belu, which would be required for the loan hypothesis. (Also the vowels are wrong.) I suspect some circular reasoning behind all this.
It occurs to me that I encountered an obscure language spoken In Plateau State in which the words for “two” have the variants “pair” and “brace” (the latter with an “extension” /s/ of unknown origin.)
The awesome power of Sheer Coincidence.
And if Volta-Congo could manage to come up under its own steam with two (or more) quite unrelated roots meaning “two”, why shouldn’t Chadic?
(In which case, although I still think the Militarevs’ methodology is terrible, it’s not impossible in principle that only Western Chadic A and Hadza happened to preserve a particular proto-Chadic form. Just much less likely than pure change resemblance, that’s all.)
It occurs to me that I encountered an obscure language spoken In Plateau State in which the words for “two” have the variants “pair” and “brace” (the latter with an “extension” /s/ of unknown origin.)
Is this the same obscure language that is spoken in the Mountain State, perhaps?
Hausa and Hadza
Which are themselves cognate names, obvs….
the ends of a loaf of bread
Eng. heel(s) (non-regional. The first OED quotation in this sense is from1611 in Cotgrave’s French-English dictionary s.v. esquignonner ‘to cut, or breake off a lumpe, cantle, crustie heele, or peece from a loafe of bread’. However, Piers Plowman wrote I nolde ȝeue for þi pardoun one pye hele!, glossed by the OED as ‘(perhaps) the crust or end of a pie’.
As it happens, Algerian Arabic has replaced the word for “two” with a borrowing from Greek – which wasn’t even a numeral in Greek, and wasn’t borrowed as a numeral. Zuj “two” is ultimately from zeugos “yoke” (and hence “pair”).
The Chadic-internal evidence seems sufficient to argue that biyu is a West Chadic A innovation (and surely Militarev doesn’t want to discard one of the best AA etymologies we have?), but that doesn’t entail that the scenario of direct borrowing from Benue-Congo must be correct.
Fair enough. Indeed, how about a PIE *s-stem: *pér-, *pr-és-… indeed, find a prefix and a suffix, and we can even keep abreast of Verner’s law. 🙂
On the mysterious *ba- component of proto-Bantu *bàdɪ́ “two”, and Pozdniakov’s idea that it could be a fossilised class 2 (“human plural”) agreement prefix, I’m reminded of the Mooré (à)yìibú “two”, beside e.g. Kusaal (a)yi.
This particular form seems to be confined to Mooré within Oti-Volta, but it’s striking that several Oti-Volta number words do appear in variant forms with mysterious final added CV elements present in one language and absent in another, like Mbelime tāātē “three” beside e.g. Moba -tāā, and these CV elements do look like plural noun class suffixes (here, the “leaf” gender plural *dɪ.)
They never actually function as such: synchronically, Oti-Volta number words differ from all other nominal words in that they inflect for class-based “gender” with prefixes rather than suffixes.
This can hardly be a common retention between Oti-Volta and Bantu, but it may be a sort of parallel, at least.
Can be still broken down into a few different grades; Uralic etymology and reconstruction got, broadly up to the 70s, quite a bit of mileage of “identify vowel correspondences but don’t bother debating too much about how to reconstruct them exactly”. Predictably a bunch of errors too but much less than it could’ve been (and many of them more towards excluding comparisons where something weird happens conditionally). Chadic might benefit from something like this if it’s currently too difficult to tell if a given /CeCeCe/ should be taken back to *CaCʲaCa or *CeCaCa or *CaCaCaj or what.
Another institutional problem with anything like AA or NC or Sino-Tibetan reconstruction, I think. The field is not in a state for anyone exept maybe Semiticists to start putting out mature etymological dictionaries, so in the meanwhile there’s nowhere specific to reasonably check for what “negative results” might be known, about things like Awjila iša coming from something with earlier *ns and thereby not readily well-compareable with forms like ʔVs-, sVV-. I figure reconstruction–etymological enterprises that are serious about getting things done should strive to have something like a common wiki for putting their results on, to cut off bad comparisons copied from older literature and to have knowledge of newer research easily findable.
A decent bunch of that issue though, to repeat, can be also fought just by demanding decent within-branch attestation: no citing just Awjila forms without at least half a dozen other Berber cognates too, no citing just Mao and claiming that it is thereby a Proto-Omotic etymon, no citing just two or three West Chadic languages and claiming that it is thereby Proto-Chadic…
That’s the Moscow School Attitude to Synonymy. Supposedly it doesn’t matter if we find multiple conflicting etyma even for some core vocabulary meaning, that just means that all except one of them must’ve independently drifted from meaning “A” to “B” multiple times (something like the case of PIE *mēh₁n̥s ‘moon’ but *louksnah₂ ‘shining’ > ‘moon’). I.e. maybe *sər- actually meant ‘pair’ and then areally turned into ‘two’! Or inversely, maybe #biri did that. (Conceivable, I’m sure, but would need much more discussion.)
I figure reconstruction–etymological enterprises that are serious about getting things done should strive to have something like a common wiki for putting their results on, to cut off bad comparisons copied from older literature and to have knowledge of newer research easily findable.
This is obvious once you say it and should definitely happen.
In Niger-Congo work, there seem to be a lot of people who are highly committed to their own bad comparisons, and who would be unlikely to play by the rules (and indeed, agreeing the rules would be the hard part.) I suspect this is also true of other semi-long-range projects like AA, where the time-depth is much greater than in Indo-European or Uralic and there is controversy over whether some entire subgroups actually belong at all.
With “Nilo-Saharan”, the problem is even worse, as the most active finders-of-etymologies tend to be those with the most tenuous grasp of how comparative linguistics should actually operate.
African comparative linguistics may be an extreme case, though. Greenberg casts a long shadow.
Sanity is gradually being restored, though: I was encouraged to read the section 1.1 “Gur and ex-Gur languages” in Jeffrey Heath’s grammar of Tiefo-D, which as is his wont (may the blessing of Panini Himself be upon his head) he makes available online:
https://www.academia.edu/72836986/A_Grammar_of_Tiefo_D_of_Daramandugu
The “ex-Gur” languages are of course all related, but “Gur” itself is not a thing, at least in its traditional sense. Heath is not a comparativist himself, so it’s all the more encouraging that word seems to be getting round. At last.
But there wouldn’t need to be any rules, would there? You just put your (alleged) results on the wiki and people can use them as they please.
I don’t see how you can “cut off bad comparisons” without some sort of rules.
However, even if the wiki ended up with a very low signal-to-noise ratio, I suppose it would be good to have the better stuff at least freely available in among the dross.
The approach me and colleages are following for this in our slowly burgeoning Finnic etymology wiki is primarily research-historical–bibliographical: as long as someone has managed to publish a comparison, etymological analysis, or commentary, it can and should be added there, and indeed one may not censor any published results from the record (there may be after all something to be learned also from what kind of errors have been made in the past). However, conflicting proposals should be explicitly noted to be in fact in conflict, and no additional speculation can be added willy-nilly in the tally (at most perhaps still in a discussion section by the editors). I.e. the wiki is a record-of-research, not the research itself. Thus no one needs to abide to rules about what’s allowed be compared; it’s seeing multiple conflicting etymologies that would already highlight if some of them look good and some do not. Any cutting-off would then happen instead in later work by people who have checked what the wiki records. Or in other words, the setup would hopefully encourage explicit criticism-in-print of bad proposals, not merely “oh this paper looks like junk / does not look compelling”, which does not really constitute an argument that could be cited in future work.
The initial collection of whatever has been proposed on something like Afroasiatic etymology would rake in a lot of noise I imagine, but, hopefully also, at least some comparisons that turn out to complement each other. Say Abu compares something Cushitic + Chadic, Ben compares the same Cushitic with something Semitic, Cali provides reasons to think the Cushitic data isn’t even native at all (maybe it’s a loan from something different Semitic); but then turns out the Chadic + Semitic would be good comparanda for each other anyway.
Even IEistics would still benefit from such a thing. It never ceases to amaze me, as I read semi-randomly through academia.edu, how insanely splintered its literature is. It is normal to find two cutting-edge papers that are a few years apart but the newer one doesn’t cite the older one or anything like it, meaning (in the vast majority of cases, I’m sure) the author, the reviewers and the editor were wholly unaware that the paper contradicts published literature in an interesting way that somebody should look into.
Again, peeps, you are rewarded for getting papers published, not for them being correct. Indeed, if they are flat wrong you have a much better chance of getting them into Nature as “ground-breaking”.
there may be after all something to be learned also from what kind of errors have been made in the past
“I know I at least keep a few “Hungarian is too a Turkic language” type works around for this purpose. The intended main thesis is not going to pan out; but any data cited to this end could prove to be regardless still valid. Usually anything of this sort mostly relies on word comparisons (appeals to typology are strangely rare), and these might remain valid as etymologies of any imaginable type… not just Turkic loans in Hungarian, but maybe also old Hu. loans in Tk.; Hu. cognates of Khanty or Samoyedic loans in Tk.; common loans from some third source like Iranian or Yeniseian or Mongolic; some could even end up being evidence for a general Turkic–Uralic relationship. None of this is a priori ruled out, and in this way it may well be possible, with patience, to find meaningful building blocks even within theories that don’t hold up in their entirety.”
(you, previously, as part of this post)
I’ve found several previous works on comparative Oti-Volta very useful even when their methodology is extremely questionable, for the very reason that they contain a great deal of data in the form of supposedly cognate sets, most of which, of course, really are cognates. Assembling potential cognate sets is where the hard work comes in: testing hypotheses about how they are related is the fun part, but pointless without lots of laboriously collected data.
Assembling potential cognate sets is where the hard work comes in
Arguably the hard work, at least in still-extant languages, is the initial (fieldwork) assembly of the words (and grammar) from which the cognate sets are to be built in the first place… of course Uralic was very good on this as well (for the languages that were still extant at the time, anyway, which many weren’t).
Yeniseian, AFAIK, was the linguistic equivalent of a rescue dig, whereas it was realized that some of the relevant languages had probably already gone extinct and most of the rest were very close to extinction as well, and as much evidence as possible was attempted to be collected from what was still around. (And it’s a good thing it was, because some of those varieties had gone extinct within decades after that, and by now we’re down to just Ket and records of others.)
Of course in much of Africa even that stage is hardly done. Fortunately (AFAIK) at least in the Oti-Volta area the relevant languages mostly aren’t particularly endangered, so there’s (probably) some time for fieldwork still left.
I’ve recently read a paper on a Sino-Tibetan language where the introduction was like “by the time we found out that there is actually a separate language in there, it apparently already had no L1 speakers remaining, but there were about a hundred L2 speakers and we managed to get a lot of data from those”. IIRC some West Pacific places are even worse than that.
assembly of the words (and grammar)
“And grammar” indeed; significant errors in Oti-Volta comparative work have arisen from thinking that it’s enough just to compare lists of lexemes.
Quite apart from the fact that morphological matches are of core importance in establishing relatedness, all the languages involved are both heavily suffixing in both flexion and derivation and prone to extensive sandhi changes which conceal underlying word-internal structure.
Oti-Volta is obviously far from an unusual case (this is of course one of the major reasons why Mass Comparison is wholly inadequate as anything but a preliminary stage of analysis.)
Fieldwork is, as you say, the foundation of everything (and the excellent fieldwork that’s been going on in West Africa since Manessy’s floruit is why his work on “Gur”, remarkable in its day, is now in need of major revision.)
Fieldwork is not drudgery though, at least not in the same way as patently going through dictionaries trawling for potential cognates or parallel morphological formations. That is when you start thinking “what I could do with, is a nice trainee to do this part for me.”
(If you can’t be bothered with the drudgery, what happens is that you end up cherry-picking one or two cases which “confirm” your preconceived theory and you miss all the counterexamples. Niger-Congo comparative work is full of that sort of thing.)
“Turn every page,” as Robert Caro says.
Sure, but the reviewers and the editors have no interest in publishing sloppy work.
Sure, but that’s irrelevant to almost anything I was thinking of. Very few IEistics is of wide enough interest to end up anywhere near Nature.
Different strokes for different folks. I’ve read thru good double-digits of comparative dictionaries by now “just for fun”, without even being paid for it. As it happens I’m right now some 400 pages deep in Hudson’s Highland East Cushitic Dictionary which, together with a handful of other works, I’m thinking of taking as a base starting point for building something on assessing Cushitic etymology in more general, maybe some unknown number of years down the line. Though yes e.g. the lack of detailed Oromo or Somali dialectology remains a major hurdle which I’m also not really equipped to address.
One does hope there to be some correlation between the two.
Different strokes for different folks
True enough.
Or in other words, the setup would hopefully encourage explicit criticism-in-print of bad proposals
Only if there was some sort of reward for doing so, and preferably a separation between authors and critics. One of the reasons why the public criticism of literary fiction is that it’s done by authors, who form a back-scratching cartel.
Sure, but the reviewers and the editors have no interest in publishing sloppy work.
I doubt if anyone was ever denied tenure because they let an obviously bogus paper through the gate. In the case of reviewers, there is no incentive at all to do a good job, except in the sense that the death penalty (where it still exists) is an incentive not to commit murder: retribution comes much too late and much too infrequently.
One does hope there to be some correlation between the two.
That’s why economics is called the dismal science. It predicts that to a first approximation everyone is a lazy entitled asshole (in agreement with even Arminian theology, actually), and that’s hard to bear. Although when Carlyle coined the phrase, what he said was dismal was that classical economics provided no support for, and indeed argued against, what he considered an essential element of civilization, namely slavery. (Speaking of entitled assholes….) I’m sure he would have felt the same way about linguistics as practiced by the Boas Totem.
One of the reasons why the public criticism of literary fiction [is/does what?] is that […]
Sorry. “Is bad”. This is not true in general of so-called genre fiction.
I disagree. Ninety percent of criticism of literary fiction is bad because of Sturgeon’s Law. There are just as many good critics as there are good writers; the trick is to find them.
And once you have found them, how do you pay them for doing a good job?
Only if there was some sort of reward for doing so
Yes, namely getting your views cited in the etymology-bibliography wiki. Note how ten seconds earlier we were worried about people wanting to “play by their own rules”.
Research-historical priority re: refuting someone’s mistakes is probably mostly not be a sufficient reward for getting people to work them into a whole peer-reviewed journal article, but then that’s not strictly required anyway, just having people publishing that in some form; say, in a literature review or in a footnote to an article about something else. (Wiki admins could in principle even decide to allow citing things like people’s blog posts.)
Of course there’s also a number of researchers in most fields who don’t care about sloppy work and lack of making progress on understanding their topic of study, but then also, the aim here is not to build infrastructure that patiently ropes these kind of people to play nice, but rather, infrastructure where one does not need to spend extra effort on playing with them at all. A long game, but a probably necessary one.
Well, in the case of reviewers, there’s no incentive of that sort to do a job at all. If you decline all review offers, hardly anyone will complain, and it’s not likely anyone will even find out.
In the case of the breadless sciences, moreover, people who don’t to some extent actually care about the science don’t get in; they go work in the real world instead of becoming academics and reviewing papers for free. That may be different in economics, I suppose – I didn’t think of that.
Back-scratching cartels exist, but they’re all very small. There have been “peer-review rings” where people managed to populate the editorial position and the review pool sufficiently densely with each other that they were in a position to get each other’s garbage published, but that’s rare, limited to the moneyed sciences (I’m thinking of those with obvious medical applications), and treated as a sizable scandal whenever discovered.
…and finally… I have no idea if any hiring committees have ever noticed, but there are nowadays ways to track how much a scientist has reviewed.
Is that explicitly forbidden? I’ve cited blog posts (and tweets even) in peer-reviewed papers of mine. The better ones are really no different from a paper that wasn’t peer-reviewed (because, say, it came out in mid-20th-century Europe where peer review was unknown), and they’re in open access.
Self-published sources (online and paper):
“generally” would seem to leave some space for “destroy all monsters, ignore all rules“…
And once you have found them [good critics], how do you pay them for doing a good job?
Send them a check or voucher. If they’re dead, send it to the organization set up to administer a prize or stipend in their name. If all else fails, keep the money as a reward for being economical with your appreciation.
This last might explain the appeal of writing about famous dead people. You owe them nothing.
“generally” would seem to leave some space for “destroy all monsters, ignore all rules“…
You haven’t met the Wikieditor Mafia, have you?
Actually, I have. Just this month I got a message that there’s a discussion posted that I should participate in. When I arrived, it was already officially closed, and the article was deleted. The discussion whether to delete it was only up for a week. Clearly they assumed everyone who isn’t negligible checks in every day.
I did leave a comment under the official discussion, but at this point I’ll be surprised if anyone reads it.
Oh we absolutely wouldn’t want to put Wikimedia Foundation in charge of anything like wiki-based bibliography; already besides usual bureaucracy problems, the institutional twin mistakes of “no original research no synthesis” and also “everyone including anonymous IPs should be by default free to contribute even if it’s vandalism and/or complete nonsense” have been too deeply ingrained in the minds of hardcore Wikipedia editors over the last 20 years. My gold standard examples for how to run a wikiesque crowdsourced database are instead entertainment social media platforms like RateYourMusic, MusicBrainz, An Archive Of Our Own… all of them at least somewhat more strictly moderated.
(Come to think of it, talking with the MetaBrainz foundation instead, who are in charge of the 2nd, could be eventually more helpful.)