Proto Oti Volta.

Academia.org provides us with “Toward Proto-Oti-Volta (A Work in Progress),” by one David Eddyshaw, a name that will not be unfamiliar to aficionados of the Hattery. The preface ends with this perhaps excessively modest paragraph:

In many ways the following work is perhaps better regarded as a set of suggestions (some better supported than others) for future directions of investigation, rather than any kind of compendium of settled conclusions. I hope to take up some of these suggestions myself, but would be even more pleased if I succeeded in encouraging others to do so. I have generally tried to err on the side of providing too much in the wayof raw language data, rather than limiting myself to what is strictly needed to illustrate any particular point. This particularly the case with my fairly extensive summaries of the verb conjugation systems of the major Oti-Volta branches, where I include many details of perhaps questionable relevance in the hope of preparing the ground for a more rigorous attempt to trace the historical developments which led to such a bewildering variety of systems in the modern languages. At this stage far too much is unclear to me about proto-Oti-Volta to attempt anything more than this preliminary account, and it therefore seems appropriate to “show my work” quite extensively.

The list of references includes half a dozen by Urs Niggli, a wondrous name of which DE himself once said “Switzerland is probably full of them.” Thanks, JC!

Comments

  1. Will there be any author events where we can get our copies signed?

    Congratulations!

  2. Also, free from sign-ins, at Zenodo.

    I’m glad that all the insights shared here, bit by bit, are being published properly.

  3. Stu Clayton says

    it therefore seems appropriate to “show my work” quite extensively

    Those quote marks … Something from an Ancient Latin Author, I bet. But what and who ? I show my ignorance.

    DE is a master of modesty. No one is born a master of ceremony. Es ist noch kein Meister vom Himmel gefallen.

  4. David Eddyshaw says

    Thanks, all.

    Objective, rather than modest (at least in this case.)

    I’ve been dithering about actually publishing anything like this for too long, on the grounds that I’m far from happy wirh it. But eventually I realised that if I waited until I was happy with it, it would never appear at all.

    Also, I tend to make ex cathedra pronouncements about Oti-Volta as if they were unassailable verities, and it seems only right to exhibit some of the scanty evidence and shaky chains of reasoning so that people can come to their own conclusions. There’s unlikely to be much progress in this area unless people other than me have a fair chance to undermine my conclusions and come up with better ones.

    Lameen actually deserves credit (but no blame) for spurring me into action a bit. When collaborating on his paper about Gur loans in Songhay I was awkwardly conscious of the fact I was making a lot of assertions about proto-Oti-Volta with nothing that I could point to as already-published evidence. If I’d been reviewing it, my response to much of the Gur stuff would have been: “Says who?”

  5. jack morava says

    Many a man lives a burden to the earth, but a good book is the precious lifeblood of a master spirit, embalmed and treasured up on purpose to a life beyond life.

  6. mazltov, DE! i hope this lights fires in the hearts of people near and far, so that they can become the well-informed peers you deserve for argument and collaboration!

  7. Trond Engen says

    May it be widely read and discussed. Do you plan to turn it into a paper (or a series of papers) for publication or for a conference? Not that it should matter, but I imagine that’s how the peers are awakened.

  8. John Cowan says

    Of course it is DE’s call, but there are a very limited number of gatekeepered (gate-kept?) slots available per annum, and they should be left for people to whom such publication can make a professional difference. I think DE’s work is exactly where it belongs. (Lameen may have something to say as well, since he may need something citable.)

  9. Looks like thorough work to me! Covers more bases than I had been expecting: arealities, proto-grammar, syntax, the works. A small monograph really at ~150 pages.

    The big thing needed after this would really be a comparative dictionary (I imagine some of this already exists, given the lexicostatistics section). The little appendix with Bantu comparison at the end brings home for me how it’s impossible to make comparative progress without data… and it kinda seems to me by now that everyone working on comparative Niger-Congo is forced to do variants of “we compare here X with Bantu” because Bantu has reconstructions available in the literature. (Even pure etymological sets without attached reconstructions might be helpful.)

  10. David Marjanović says

    there are a very limited number of gatekeepered (gate-kept?) slots available per annum

    That’s how it was until the age of the online megajournal. The scientific monograph, previously moribund, was saved in 2015.

  11. David Eddyshaw says

    The big thing needed after this would really be a comparative dictionary (I imagine some of this already exists, given the lexicostatistics section)

    Tony Naden has done an enormous amount on the way of collecting cognates, but only of Western Oti-Volta. He very generously shared all of that with me, but I don’t think he’s envisaging publishing anything soon (though I hope to be proven wrong.) Highly impressive as it is, for the purposes of what I’m doing at present it’s not very useful, because the WOV languages are pretty closely related to one another, and the WOV protolanguage seems to have been pretty similar to what you end up with just by internal reconstruction of Kusaal, especially taking in both major dialects. Proto-WOV doesn’t take you very far back (which is not too surprising, if the current spread of WOV over half of the Oti-Volta teriritory is connected with the rise of the Mossi-Dagomba states in the fourteenth century, as seems plausible.)

    Coffi Sambiéni has extensive comparative lists for the Eastern Oti-Volta languages (other than Mbelime) in his Le Proto-Oti-Volta-Oriental. Unfortunately only Waama and Mbelime have published dictionaries at all, so it’s not easy to go any farther on that front. The more unfortunate, as that’s where most of the internal diversity of Oti-Volta is: “Eastern Oti-Volta” is not at all parallel to Western Oti-Volta taxonomically.

    AFAIK there’s nothing at all for Oti-Volta as a whole; my homebrewed lexicostatistics is all based on sets I’ve collected myself from the dictionaries and published grammars. Almost all of these sets are incorporated somewhere or other in Toward proto-Oti-Volta.: it wouldn’t be too difficult to sieve them out and index them.

    Your point about everyone ending up comparing with Bantu is very valid: it’s a combination of a vast amount of fairly securely reconstructed lexicon, and the fact that the morphology of the protolanguage is also misleadingly easy to reconstruct in detail. (I say “misleadingly”, because there seems increasingly to be an awareness that traditional reconstructions of this have tended to sideline the area of maximum diversity of the Bantu languages, in the far northwest, with the result that they probably really reflect the morphology of an innovating subgroup rather than yer actual proto-Bantu.)

    John Stewart did a lot of solid work on reconstructing proto-Potou-Akanic before he succumbed to the siren song of premature comparison with Bantu (and, worse yet, with Fulfulde.) Although Akan is without doubt actually related to Bantu, he ended up with all the classical errors of multiplying the number of contrasting segments in the presumed common protolanguage to make as many series of comparanda “work” as possible: his own version of proto-Bantu is very remote indeed from what actual Bantuists have reconstructed on the basis of the Bantu languages themselves.

    The case with attempts to compare the Atlantic languages with Bantu is much worse. “Premature” is the operative word …

    I do think proto-Oti-Volta has considerable potential for helping with the ongoing efforts toward proto-Volta-Congo, though. The languages seem to be on the whole fairly conservative phonologically. I don’t think this is merely Teeter’s Law in action.

    And I think proto-Oti-Volta is reconstructable, though probably never to the degree of precision of proto-Bantu, given that there are only thirty-odd languages to work with, and that the groups is at least as diverse internally as Narrow Bantu.

  12. This is wonderful! Congratulations.

    The obvious place to publish this as a printable monograph would be Language Science Press; there should be no difficulty getting it accepted, but the problem is they insist on LaTeX formatting, which is a considerable waste of time better spent on reconstruction. My lab has been working on a similar project accepting Word files, but, for reasons unrelated to format, it’s off to a very slow start.

  13. David Marjanović says

    Are there no automatic converters to LaTeX format?

  14. David Eddyshaw says

    Actually, I see that the Language Science Press guidance for authors says that they provide one. I suspect that that only gets you so far, though. (My own experience with LaTeX is zero.)

  15. jack morava says

    LaTeX is very intuitive. Go for it.

  16. There are lots of automatic converters to LaTeX, but you will have to proofread everything line by line to make sure there are no errors. The converters have gotten a lot better this millennium, but I don’t precisely know how much improvement there has been with the types of formatting that David Eddyshaw has used in his monograph. (My own experience is mostly with trying to convert equations people have produced using other programs. It astonishes me how many biologists, chemist, and even solid state physicists still try to use Mircrosoft products for math formatting!) If there were a completely reliable LaTeX converter that worked well enough to produce essentially final copy from manuscripts in Word and other formats, no publications would still be insisting on receiving LaTeX submissions.

    Part of the point of using LaTeX in the first place (apart from keeping away the cranks who can’t be bothered to learn it) is that it is completely portable, so any two users anywhere should get the same output starting from the same text-based markup code. This differs from a format like Microsoft Word, which depends a lot more on the local installation environment (although a lot less than it used to). Probably more importantly, it is fairly simple to make changes to a LaTeX file (to adapt it to house style) without risking totally breaking the intended formatting. In fact, may journals provide all the style files you need to produce camera-ready output at home.

  17. David Eddyshaw says

    Perhaps time to seek out LaTeX pour les nuls

    My original is actually in odt format, on account of the fact that I expunged Windows from my hard drive as soon as I found out how to install Linux, some time about 1998 …

    But the first thing would be to see if Language Science Press are interested at all. (And the second thing, if they are, would be quite a lot of tidying and elimination of inconsistencies and sheer proofreading anyway. It’s not worth the lives of any real-life trees as it stands.)

    Currently recovering from a family wedding in Valencia. Hispanophones will understand when I say that the event was largely Venezuelan. Also, there were Finns. It’s a wonder I’m able to type at all …

  18. Bathrobe says

    It’s not worth the lives of any real-life trees as it stands

    Dang, I was thinking of printing it out

  19. John Cowan says

    expunged Windows from my hard drive

    Note that both Word and LibreOffice can handle all of .doc (binary), .docx (compressed XML) and .odt (also compressed XML) formats, though not entirely with the same fidelity. Still, I would say that LO deals with documents written using Word at least as well as other versions of Word do. I keep my resume/CV in .doc format and edit it with LO.

    LaTeX pour les nuls

    Googling for “LaTeX for beginners” will supply innumerable tutorials at all levels.

    lives of any real-life trees

    There is such a thing as printer paper made from “recycled 100% post-consumer waste” (i.e. the trees already lost their lives for other causes), though doubtless this is not everywhere available.

  20. Jen in Edinburgh says

    I don’t have any sensible comments to make, sorry, but are we suggesting that webpages are printed on imaginary trees?

  21. David Eddyshaw says

    They’re not?

  22. Imaginary trees with real webpages in them.

  23. “LaTeX is very intuitive.”

    That’s certainly not true. It might be for a few people (maybe particularly people from a programming background), but not for the overwhelming majority of people. I pity anyone who tried to use it under the impression it was actually intuitive or user-friendly — this kind of “recommendation” seems a good way of making sure people give up quick and never come back.

    (I use LaTeX, and usually happily.)

  24. PlasticPaddy says

    My memory of TeX was that it avoided WYSIWYG in the editor and you needed to use a separate “debug’ or “print preview” tool to see output, then go back in to the editor to fix problems…

  25. jack morava says

    LaTeX is indeed not WYSIWG, it makes few assumptions about one’s defaults. Measure twice, cut once.

  26. @PlasticPaddy: For the LaTeX itself, that’s true. However, there are now innumerable frontends that provide WYSIWYG capability (showing output in a separate window or frame). There are browser-based frontends, as well as ones that run locally; the latter are definitely superior for long documents or ones that use a lot of ancillary macro packages or intricate formatting commands. You can find ones that try produce the output instantaneously, or ones that wait for you to click a button to make* the output. Again, if you are doing heavy math or formatting stuff, the latter is definitely better, since as you type, a lot of the time what you have entered so far will not be valid compilable code, and it’s just wasted computational overhead to have the frontend flailing about trying to process it.

    * Make is used here in a specific computing sense. It means processing all the source files (originally, this specifically meant files containing C code) that have been updated since the the last time the compiler/parser/whatever was run, with automatic detection of which files those are. This is thus an automated process of producing a new build, although sometimes build is used in a contrasting sense, meaning compiling/processing all of the source files, regardless of their apparent update status. (Obviously, this kind of completely rebuild is not a very viable approach for large programs today, which often rely on massive, externally maintained codebases.)

  27. УРА!!!!!!!!!!!!!!!!!!!!!!!!

    DE, LH, thank you for sharing!

  28. David Eddyshaw says

    Pʋ’ʋsʋg kae!

  29. @DE, one of two reasons why I haven’t read your grammar yet (the other is unexpectedly slow progress over other items in my to read list…) is that I like to be able to play with a langauge before or at least simultaneously with reading what others tell about it.
    So ideally I need texts and audio.

    It would be silly to ask you to spent two more months of your time on compling this:) Especially, I could just buy a ticket to Ghana or Burkina Faso…
    But maybe you know where I can find it?

  30. AFAIK there’s nothing at all for Oti-Volta as a whole (…) it wouldn’t be too difficult to sieve them out and index them.

    Yes, an appendix like that would be probably already valuable to others.

    I think proto-Oti-Volta is reconstructable, though probably never to the degree of precision of proto-Bantu, given that there are only thirty-odd languages to work with

    Well, thirty-odd from above but also like a thousand for reconstruction from below! Eventually, once there are also earlier reconstruction stages for guidelines, fine-grained etymological work could end up netting quite a bit more detail. (But don’t ask me which century is this going to happen.)

    УРА!!!!!!!!!!!!!!!!!!!!!!!!

    I wud on the counterarie re-commend agaynst retranscrybing anny of the woark yn th’ Ynternationalle Phonetyc Alphabette.

  31. David Eddyshaw says

    So ideally I need texts and audio

    The only readily available text online is the Bible translation, alas; and even that doesn’t seem to exist as a convenient pdf, only an Android app. Though if you search the Google Play Store (or whatever cutesy name it has) you can find some collections of proverbs and tales in the Toende dialect.

    The only audio I know of is the SIL version of the 1996 New Testament version, which you can get via

    https://live.bible.is/bible/KUSTBL/MAT/1

    However, it leaves a lot to be desired (even if you’re keen on the text in question.) The worst thing about it is that it comes with a highly intrusive musical accompaniment, which is terminally annoying even if you aren’t trying to hear the actual words. It’s kinda semi-dramatised.

    The readers mostly read in a very stilted manner, with multiple unnatural pauses. Very few people are used to reading Kusaal fluently out loud, and unfortunately it shows. It doesn’t help that the 1996 text uses a spelling that reflects a pronunciation that was already very old-fashioned at the time, for example in writing the ubiquitous clause-linker particle as n, which practically nobody actually pronounced like that even in the 1990s. The readers often stumble over it. The man they’ve got for Jesus is particularly prone to this.

    The best reader is the bloke they got to read some of Paul’s letters. I’m fairly sure he’s the man who first taught me a bit of Kusaal. He was a local teacher, and had a sort of standing brief to teach new medical staff how to say “What appears to be the trouble, my good man?” in Kusaal.

    I have a collection of literacy materials and a book of tales in Agolle Kusaal that I got when I was resident in Ghana, and a copy of a somewhat evanescent local newssheet in Kusaal, but I don’t think that there is really anything available outside Ghana (and you have to search pretty hard to find much inside Ghana, for that matter.)

    On the Burkina side it’s all Toende dialect. I think there actually is an Android app of someone reading Mark’s gospel in Toende Kusaal, but again, pretty disfluently.

    I’m afraid you’ll just need to go to Ghana, drasvi. (There are certainly worse places to go … Ghanaians spend a lot of time talking about how hospitable they are, and have consequently conditioned themselves into actually being strikingly hospitable. It’s a nice place to be a foreigner in.)

  32. @languagehat: The linked Web version of that poem by Marianne Moore has an interesting formatting issue. The intended peculiarities of the formatting appear to have created another, possibly unintended, peculiarity. Look at the third stanza:*

    eat, elephants pushing, a wild horse taking a roll, a tireless wolf under
       a tree, the immovable critic twinkling his skin like a horse that feels a flea, the base—
       ball fan, the statistician—case after case
          could be cited did
          one wish it; nor is it valid
             to discriminate against “business documents and

    The second line ends with a cut through the middle of “base—ball” that is necessary for the rhyme scheme. However, normally this compound word would be spelled with a hyphen “base-ball” (if it was not just written closed as it normally is today). Yet the poem uses an M-dash, which leaves me wondering whether this was intentional done by Moore (to emphasize the break at the end of the first line of the couplet) or whether it arose through a formatting error (and, if so, at what stage in the poem’s history it firs appeared).

    * Pretend there’s a <blockquote> here. I had to take the actual one out so as not to mess up the key line, which was a little too long to fit.

  33. Definitely a formatting error, and presumably a web-induced one. I was sure of that, but I checked my Collected Poems to be sure.

  34. Trond Engen says

    The map is extremely helpful.

    Comparing it with the genetic tree, there’s a noticeable correlation between subgroups and political borders.

    Eastern Oti-Volta = Benin
    Gurma = Togo (and an extension northwards)
    The Bugum Group = Ghana (and an extension northwards) except that Boulba and Yom/Nawdm seem to have been pushed east by a Gurma bolt.

    Did the north-south colonial borders follow ancient political systems (until a sudden east-west line was drawn with French East Africa), or did some languages or ethnic constellations manage to spread within the borders of the colonies and modern states?

  35. David Eddyshaw says

    The match with the map is partly deliberate: I’ve tried to lay out the genetic tree in a similar order to the actual location of the languages wherever it didn’t result in crossing lines.

    I think any alignments with modern state boundaries are coincidental. Bear in mind that the border between what is now Ghana and what is now Togo shifted east due to the plebiscite in 1956, which reunited pretty much all of the Agolle Kusaasi, previously split between the two, and putting a fair number of Moba/Bimoba speakers on the Ghana side. But there seems to be fair bit of evidence for Western Oti-Volta influence on Moba, presumably antedating the European invasions (which are relatively recent in those parts, after all.)

    There are a good many Konkomba (another, more southerly, Gurma group) in Ghana, too; there was a very nasty conflict between them and the Dagomba in the 1990s. Propagandists on the Dagomba side claimed that the Konkomba were recent arrivals from Togo, but in reality it seems to be an agriculturalist vs pastoralist thing, not a cross-border invasion of any kind. The Konkomba were there all along.

    The Atakora in Benin, along with the adjacent part of Togo, is probably the old homeland of Oti-Volta, judging by the relative linguistic diversity.

    The expansion of the all-fairly-similar Western Oti-Volta languages over half the Oti-Volta area is presumably related to the spread of the Mossi-Dagomba kingdoms since the thirteenth century; both linguistic criteria and traditional history support the idea that the starting point was from about where Kusaal is spoken now; the immediate northern neighbours were the Mande-speaking Bisa (still there after all the intervening centuries, and like the Kusaasi, notably reluctant to be incorporated into anyone’s empire, thank you very much.)

    Dagaaba traditions suggest that they migrated to their current far-western area only a few centuries ago. The push northward of Mooré is relatively recent too:

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

    The major mismatch between genealogy and geography is indeed with the Yom/Nawdm branch of my “Bugum” group. Nobody has seen this as a problem before, because the evidence for a Bugum group hasn’t previously been put together. I don’t know if the Nawdba or the Yoowa have any traditions of migration from a more westerly homeland (the Boulba do, but the evidence for the Western Otii-Volta character of their language has always been clear; it’s not potentially controversial like my “Bugum.”)

    Nawdm is strikingly divergent lexically, which would fit a late-incomer scenario; but Yom isn’t. Indeed Yom has a quite startling degree of lexical agreement with Buli, way over in the west.

  36. Modgame, logtore! (Excuse my broken Mooré. How does one form a vocative?)

  37. the problem is they insist on LaTeX formatting, which is a considerable waste of time better spent on reconstruction.

    FACTS!
    TRUST AND BELIEVE!!!

    (As the children say.)

  38. David Eddyshaw says

    F barka, a Xerîb!

    Though it is highly ungrateful of me to correct the grammar of your kind words, that should properly speaking be f modgame, literally “you’ve made an effort.” The -ame bit is historically a chain of two particles, marking a clause-final positive indicative verb, though I’m not sure if it should be segmented from a synchronic point of view (and it’s always written solid with the verb.)

    Oti-Volta languages don’t do primitive things like case. Logtore is fine.

  39. David Eddyshaw says

    @drasvi: talking of Mooré reminds me that if you decide to switch your attention from Kusaal to the most widely spoken (by a big margin) of all Oti-Volta languages, there are quite a few good films in Mooré you can watch. (I would highly recommend Yaaba, for one.)

  40. Lars Mathiesen (he/him/his) says

    Hurra! (to echo drasvi).

  41. CrawdadTom says

    Beautiful work, Mr Eddyshaw.

  42. David Eddyshaw says

    Alaris, Akɔliginnɔraug Tɔɔmya: m pʋ’ʋsya bɛdegʋ!

  43. Trond Engen says

    Observation:

    Ditammari is a tomato language.

    (Except exceptions, which I guess are edifying. Banally dī- ~ -rì, less obviously kū- ~ (or -V?), possibly with wider implications dī- ~ (or -V?).)

  44. @DE, thank you. Sigh.

    @drasvi: talking of Mooré reminds me that if you decide to switch your attention from Kusaal to the most widely spoken (by a big margin) of all Oti-Volta languages, there are quite a few good films in Mooré you can watch. (I would highly recommend Yaaba, for one.)

    Aha!
    I once began watching it, and since then once in a while I remember it and want to ask you to remind the title!:-)

  45. David Eddyshaw says

    The title means “grandparent” in Mooré (same as the Kusaal yaab.) But the rather schmaltzy suggestion that gives in English is not at all reflected in the actual film.

    I must watch it again myself.

  46. David Eddyshaw says

    Banally dī- ~ -rì,

    You’d think so; but actually the ri in “Ditammari” is part of the stem rather than representing a noun class suffix: it’s the the language of the Batammariba “builders.” In Kusaal the equivalent is tammɛɛdib “builders”, which is a compound of tan “clay, earth” with the plural of the agent noun mɛɛd of the verb “build.” Although the roots are cognate, the agent-noun-deriving suffix r in the Ditammari is not cognate to the d of the Kusaal: the latter goes back to POV *t (which would surface as t in Ditammari, not r), and although this suffix is reconstructable to POV, its use in productive derivation of agent nouns is only seen in the Bugum group and in Waama. The Ditammari r here is cognate to the agent-noun-deriving l of Moba tanmaal “builder” (maa “build out of mud.”)

    The actual class suffix in the word “Ditammari” has eroded away: same as in dīnũ̀ũ̀ “mouth” (= Mooré nóorè.) There’s a lot of that in Ditammari.

  47. “a Xerîb!”
    ..it’s difficult not to read it as Irish vocative, substracting lenition.

  48. David Eddyshaw says

    Indeed. Though the a particle actually is the marker of personal names, in Mooré as in Kusaal. In the Mooré Bible, the fourth Gospel is attributed to a chap called a Zã.

    It works rather like spoken capitalisation. In Kusaasi fables, Abaa is “Br’er Dog” (baa “dog.”)

  49. Awesome David, congratulations. All my efforts to follow along your ‘workings’ over the years meant that … it’s still way over my head. I delayed replying on this thread until I’d at least tried. One thing:

    my sources use different orthographies

    Did you find that the very act of recasting those sources to consistent orthography revealed connections that had previously been obscure?

    Why have different ethnologists used different phono notations? Is it that IPA is inadequate for the tone system? nasalisation? or something? Using apostrophe for anything strikes me as hazardous — but especially for glottalisation.

    Or does everybody want to make their own stamp on the subject?

  50. Oh another thing: adjectives [wot we do not ‘ave].

    You show the adjective getting infixed between the noun stem and its class suffix. How does this go with multiple adjectives: large, white hungry goats ? How ‘heavy’ can compound-initial forms get before switching to some other compounding form, like a relative clause or multiple clauses with deictics?

    You suspect it’s relatively recently that the agreement system has been abandoned. What would that system have looked like? Do you meant that’s Proto-Proto-O-V?

  51. David Eddyshaw says

    Did you find that the very act of recasting those sources to consistent orthography revealed connections that had previously been obscure?

    No; the authors involved are all experts: they just adopt slightly different conventions. The only major alterations I made were to change Kantchoa’s phonologically-correct-but-unusual-for-an-Africanist ɟ j to the usual Africanist j y used by everyone else, and to change Beacham’s very odd q to ɣ. (Beacham also uses a very strange notation for tone, but I left that as is, and explain it in loco.)

    The main odd man out in this is André Prost, whose orthography I didn’t alter at all. He’s the only extensive source for Boulba, and describes an actually different dialect of Nateni from Neukom.

    Why have different ethnologists used different phono notations?

    j y for IPA [ɟ] [j] is pretty standard in (West) Africanist works. (And after all, it’s a case where the creators of the IPA basically screwed up.)
    And in fairness to those who adopted the apostrophe for glottalisation, the analysis of this as a vowel feature is original with me (though definitely correct.) Previous investigators have all taken these glottal vowels to be VʔV sequences, so they are basically using the apostrophe to represent the consonant [ʔ] – according to their own lights.

    Neukom’s sticking of the tilde for nasalisation below the vowel symbol is just so it doesn’t clash with tone marks. Given that it’s not needed in its IPA-compliant role as a glottalised-vowel marker, that is quite sensible, and I even dallied with the idea of transposing the tildes throughout, before deciding it was better not to make grautuitous changes.

    Myself, I prefer to mark nasalisation with a superscript n (which is what Jeffrey Heath does in his grammars), which likewise leaves the way clear for the the tone marks. I can get away with this in the Kusaal because I’m citing my own work, but elsewhere it once again seemed a step too far.

    Overall, my feeling is that variations in orthography don’t really matter very much, unless they are seriously liable to be misinterpreted.

    How does this go with multiple adjectives: large, white hungry goats ?

    You can just repeat the process: adjectives have combining forms, just as nouns do. Constructions with more than one attributive adjective are rare in the wild, though.

    You suspect it’s relatively recently that the agreement system has been abandoned. What would that system have looked like? Do you meant that’s Proto-Proto-O-V?

    Definitely proto-Oti-Volta. The Western Oti-Volta languages are the only ones that don’t have full-fledged grammatical gender, except for Moba, which seems to be in the process of abandoning it under the influence of WOV (the system in Moba has changed even over the past few decades.)

    Moreover, even two WOV languages still have full agreement systems (Farefare and Boulba) and Talni and some Mampruli dialects have reduced but still functioning systems with few agreement classes.

    You could probably reconstruct the system even on the basis of internal evidence in Kusaal alone, whihc preserves numerous fossilised relics of the agreement system it set expressions and the like, even though the system itself is defunct.

  52. David Marjanović says

    (And after all, it’s a case where the creators of the IPA basically screwed up.)

    [ɟ] is not such a common sound worldwide. In Europe, I’m not sure it even occurs outside of Hungarian and Latvian. Most reports of [ɟ] worldwide are [d͡ʒ], [d͡ʑ], [dʲ] or even [gʲ].

    Admittedly, it’s about as common as [c]…

    Is it that IPA is inadequate for the tone system?

    Originally it was, of course; but nowadays it allows both the African notation (accents) and the East Asian one (tone letters). The East Asian one would become very clunky if applied to languages where the tone-bearing unit is the mora instead of the syllable; plenty of languages in Africa are analyzed as having the former.

  53. Albanian <gj>, in some varieties.

  54. David Eddyshaw says

    The map accounts for the usual Africanist convention quite neatly …
    (I daresay that the map for IPA [y] would be pertinent too.)

    EDIT:

    Maybe not so much:

    https://phoible.org/parameters/27E498422242BAF54775C75A8C884CF7#2/39.8/147.3

    although they don’t note much of it in West Africa, at any rate.

    Kusaal actually does have [y], but only as an allophone of /u/: for example, zug “head” is [ʒyg] for a lot of speakers. Historically, /u/ is quite often derived from earlier /i/ via various rounding processes: it’s one of the ways that Dagbani (for example) has ended up with a contrastive series of palatal stops, from an earlier stage in which the palatal stops were just allophones of velars before from vowels. (The original POV palatal stops have merged with *s *z everywhere in WOV except in Boulba.)

  55. David Eddyshaw says

    [before front vowels]

  56. David Eddyshaw says

    Actually, the ɟ-versus-d͡ʒ issue leads me to a more coherent answer to AntC’s question about orthography: the exact phonetic manifestation of a segment is, on the whole, not very important in comparative work, which essentially works by pattern matching. It comes into play if you’re thinking about “naturalness” of sound changes, but a lot of that is fairly question-begging anyway: if a change can be shown to have actually happened, the question as to whether it is “natural” seems pretty moot.

    This is relevant to the question of WOV glottal vowels, too: I’m no Ladefoged, and previous investigators may be quite right that they are phonetically Vʔ(V) sequences, at least sometimes. But they quite definitely pattern morphophonemically as vowels in all the languages in which they occur, and never as VCV sequences. They never constitute two syllables. (In fact, the same is true of Vɦ sequences in Nawdm: Nicole, who wrote a detailed and very useful grammar of Nawdm, gets quite excited about how /ɦ/ is a perfectly good consonant just like any other, contrary to previous descriptions of the language, but he’s not looking at the issue in the right way: basically it has much the same status as the the presumed WOV /ʔ/, regardless of how it is actually realised phonetically.)

  57. John Cowan says

    Similarly, in Lojban /V’V/ is phonemically and morphologically a kind of diphthong contrasting with /VV̯/ and V̯V (which themselves do not contrast), even though phonetically it is [VhV]. So phonemically we have /ai̯/ vs. /a’i/ ([ahi]) but not /a̯i/ on the one hand and /i̯a/ vs. /i’a/ [iha] but not /ia̯/ on the other, and there is no /h/.

  58. David Marjanović says

    Albanian gj, in some varieties.

    That’s often been claimed, and it makes historio-geographic sense, but I wonder if it’s actually the case; do you have an audio sample you can link to?

    et al.

    Ah yeah, the Basque occurrences are probably real because I read of them in sources that also listed numerous other realizations of j. But I suspect the Turkish ones are fake. I’ve heard Turkish with plain velars in front-vowel words, and Turkish with [kʲ gʲ] identical to the Russian ones, but not so far Turkish with actual palatals.

    Kusaal actually does have [y], but only as an allophone of /u/: for example, zug “head” is [ʒyg] for a lot of speakers.

    Interesting. What’s the conditioning factor? Or is this another kind of variation?

  59. David Eddyshaw says

    The conditioning factor is a preceding /s/ or /z/. These often reflect POV *c *ɟ, and as you yourself pointed out a while back, the sound change here is likely to have proceeded via /ʃ/ /ʒ/. Currently, no Oti-Volta language contrasts s/ʃ or z/ʒ.

    Kusaal zug is actually from POV *zù- rather than *ɟù- historically, but it’s quite likely that POV *z was in fact realised as [ʒ]: it regularly becomes [j] in Gurma and the Atakora languages (cf Moba yùl̀ “head”; this is in the *ɹɪ/*ɰa noun class, which probably reflects the position in POV, but the Bugum group languages all have it in the *kʊ/*tɪ noun class instead. Waama, as often, goes its own way, with yuubu, plural yuuna. In most of Oti-Volta, this is the “tree” class. Waama is weird.)

  60. David Marjanović says

    Interesting! That makes sense.

    (Phonetic sense specifically as [ʑu] > [ʑy]. [ɟ] > [dʑ] > [ʑ]…)

  61. David Eddyshaw says

    POV *s has a strong tendency to lose its alveolar/palatal place of articulation too, and end up as /h/ (or zero); and synchronically, too, /s/ often has [h] allophones, at least for some speakers. Kusaal uses /s/ for non-initial /h/ in loans: Alaasid “Sunday”, Dasmaani “(Abdur)rahman.”

  62. I haven’t looked for a recorded sample of Albanian. I checked Wikipedia, which relied on Kolgjini’s 2004 dissertation. The relevant quote from the abstract: “More recently, for many native speakers of Albanian, the palatal stops have changed into palato-alveolar affricates, e.g. c, ɟ > _ ʧ, ʤ. […] this de-dorsalization of palatal stops is especially the case in the Gheg variety of Albanian, and is beginning to be the case in the Tosk variety, as well.” This detailed investigation of Northern Tosk speakers (all ca. 30 y.o.) shows a contrast between apical (ç, xh) and laminal (q, gj) postalveolar affricates.

  63. David Marjanović says

    The paper is interesting. It confirms I haven’t been imagining [ɹ ~ ɻ].

    “The consonants represented by ⟨q⟩ and ⟨gj⟩ in writing are traditionally described as plosives (Newmark 1957; Bevington 1971; Newmark, Hubbard & Prifti 1982; Dodi 1996; Memushaj 2005, 2011; Jubani-Bengu 2011, 2012), though Lowman (1932) classifies them as affricates and Belluscio (2014) reports that they are realised with strong frication. In our data, in fact, they are never realised as plosives and are instead clearly affricated for all speakers. The waveforms and spectrograms of qava ‘I cried’ and gjaku ‘blood’ as produced by speaker S04 are provided in Figure 6. From these it can be seen that the plosive closure in word-initial position is followed by a period of frication of substantial duration.”

  64. The conditioning factor is a preceding /s/ or /z/. These often reflect POV *c *ɟ

    Intrig’ing, sounds like one of those cases where a purely distributional phonemic analysis fails to look particularly naturalistic. There’s little reason why an alveolar /z/ should trigger fronting from /u/ to [y], especially if other alveolars do no such thing. [+front] probably has to be specified somewhere instead of dropping from heaven, all the more if it is diachronically archaic. (Especially if there were to be any people who have [ʒy] and would be able to contrast it with [zu], even if it doesn’t currently exist etymologically?)

    I think I’ve seen some other things of this sort before too, where two alleged allophones “mutually preserve” each other. Slightly adjacently, reminds me at least of the competing analyses of Mandarin Chinese [sɿ ɕi] (+ likewise for affricates) as either /sɿ si/ or /si ɕi/ … here it is at least clear that some form of distinction must be posited.

    I suspect the Turkish ones are fake

    Probably also fake in Veps, where they should also be just /kʲ/ and /gʲ/.

    Definitely not fake in this direction for Skolt Sami where /c ɟ/ (from palatalized *kʲ *gʲ) contrast with /tʲ dʲ/. However, they better might be (already often are) described as affricates /cç ɟʝ/, showing even the Albanian-like assibilation in current youngest speaker generation (but, interestingly, with original /tɕ dʑ/ first getting out of the way by merging into the relatively rare /ɕ ʑ/). Hungarian, too, allows [cç ɟʝ] in variation and there’s quite a lot of literature arguing if they “really” are stops or affricates… seemingly rather stuck on a note of wanting a One True Description of Platonic Standard Hungarian…

    The /c ɟ/ given for Kildin / Ter Sami (besides, as expected, /tʲ dʲ tɕ dʑ kʲ gʲ/) I suspect might be completely hallucinated, these varities have neither the strong velar palatalization of Skolt nor *j-fortition as in Northern / Lule / Pite Sami.

  65. January First-of-May says

    Slightly adjacently, reminds me at least of the competing analyses of Mandarin Chinese [sɿ ɕi] (+ likewise for affricates) as either /sɿ si/ or /si ɕi/ … here it is at least clear that some form of distinction must be posited.

    AFAIK Russian palatalized consonants are a little bit like this across the inventory (are [sa sʲæ] really /sa sʲa/? if yes, why? if no, what are they? doesn’t help that the orthography focuses on the vowel distinction), though IIRC the distribution is such that it’s actually possible to prove that the palatalized consonants are, in fact, separate phonemes [EDIT: because both versions can occur word-finally, for one], while the vowels are usually considered to be allophones.

    (It’s even more complicated with /ɨ/, which is in 99%-complementary distribution with /i/, but, unlike the other palatalization-induced vowel pairs, both versions are technically allowed to occur word-initially, so there are a few straight-up minimal pairs that technically make them separate phonemes.)

    [EDIT 2: why the ch*rp is the edit function breaking my IPA? I’m seeing [ɿ ɕ ʲ æ ɨ] as [É¿ É Ê² æ ɨ] and I’m sure they would be even more mangled after the second edit.]

  66. I’m seeing them fine.

  67. January First-of-May says

    The text of EDIT 2 in the previous comment, as intended, posted without editing:

    [EDIT 2: why the ch*rp is the edit function breaking my IPA? I’m seeing [ɿ ɕ ʲ æ ɨ] as [É¿ É Ê² æ ɨ] and I’m sure they would be even more mangled after the second edit.]

    I still have no idea why it’s mangling stuff like that. It’s really not supposed to, and AFAIK it never did before. At least one of those letters is straight-up European and had been supported long enough to be in CP-1252.

  68. This came up before. It happens after you edit a comment, but only for you. Refresh the page and it’ll be fine.

  69. January First-of-May says

    I’m seeing them fine.

    Huh, works for me now too. Might have been a browser bug. Weird, though; as I hinted in the previous comment, æ isn’t exactly a rare letter (in fact it’s even used in English in some contexts, but in Danish and Norwegian it’s literally part of the alphabet).

    Testing: [ɿ ɕ ʲ æ ɨ]

    Update: [ɿ ɕ ʲ æ ɨ]
    …the ch*rp is a • and why it shows as a square with 00 95 in it in my edit window?
    [EDIT 2: and apparently, whatever this character is, it doesn’t show up in the posted version of the message. I can copy it, though.]
    [EDIT 3: U+0095 MESSAGE WAITING, in the C1 Controls block. Copying it into a Google search bar makes it show up as • where • is 0x95 in CP-1252 but I have no idea where  comes from.]

    Update 2: huh, now it works correctly. I have literally no idea what could have happened.

     
    EDIT 1:
    This came up before. It happens after you edit a comment, but only for you.

    That doesn’t make it less weird.

  70. David Marjanović says

    Everything shows as intended.

    (It’s even more complicated with /ɨ/, which is in 99%-complementary distribution with /i/, but, unlike the other palatalization-induced vowel pairs, both versions are technically allowed to occur word-initially, so there are a few straight-up minimal pairs that technically make them separate phonemes.)

    Not if you interpret word-initial [i] as /ji/!

    (Ukrainian spells this out, e.g. їх.)

  71. David Eddyshaw says

    why the ch*rp is the edit function breaking my IPA?

    It does that for me, too, but reverts to what I actually typed when I refresh the page.
    And it only happens when I edit a comment, not when I first make it.

  72. @J1M, and hard consonants are not allowed before /ɨ/, appear before /e/ only in borrowings (where they too get replaced with soft consonants) and soft consonants are not allowed before /i/.

    I don’t like the analysis “hard and soft consonants and vowel allophones”. Personally I want phonology to reflect what I feel as a native speaker, and as a native speaker I feel something quite different from “soft consonants soften articulation of vowels”…

  73. Lars Mathiesen (he/him/his) says

    @J1M, it actually makes it less weird, because it’s a very common type of error where (to get technical) AJAX responses are sent with a different encoding from the main page, or maybe sent with the same encoding but marked as something else. The strange characters arise when Unicode encoded with UTF-8 is interpreted as Latin-1 or a similar one-byte character set. (The byte value of Latin-1 / CP-1252 Â is a prefix byte in UTF-8, introducing a two-byte sequence for the code point range from 0x80 to 0xbf, and it just so happens that for technical reasons the second byte has the same high bits, 0x80, as the encoded value so you see the original 0x95 there. Ã is the prefix byte you get for lower-case accented Latin letters, 0xc0-0xff, and É covers the range 0x240-0x27f, part of the IPA extensions at 0x250-0x2af).

    TL;DR: For a web hacker, this is just “Oh, they made that specific mistake, let me at the code and I can fix it in half an hour.” Or rather, as this is a new phenomenon that started after an update to WordPress, something else changed in the system — maybe the “render a comment to HTML” function started returning UTF-8 instead of what it used to — and we can hope that the comment editing plugin gets updated soon. In the meantime, refresh is your friend.

  74. David Eddyshaw says

    I’ve just realised that the Swahili iba “steal” is probably cognate to Kusaal zu “steal” (respectively, proto-Bantu *jìb-, proto-Oti-Volta *zùβ-: a pre-POV *zìβ- would very readily become *zùβ-, with plenty of parallels for the rounding of the vowel in such contexts in Oti-Volta.)

    This may displace my previous favourite unobvious cognate pair, Swahili fa “die”, Kusaal kpi.

    Sorry. I just really like improbable-looking cognates. I hope that this shrimp does not buffa too many Hatters.

  75. I just really like improbable-looking cognates.

    You and me both — no need to apologize!

  76. Very nice. French /nu/ and AmE /ǝs/ is the pair I always turn to. /u/ and /ɔgǝst̚/ is nice too.

  77. David Eddyshaw says

    I think French should be banned from this competition to give lesser languages a chance.

    I mean, /ʃwaʁ/ < cadere

    (Though in Y’s example, unusually, it’s not the French form that wins the Improbable Sound Change Cup.)

  78. Lars Mathiesen (he/him/his) says

    Danish has the doublet /dju:s/ ~ /sky?/.

  79. David Eddyshaw says

    Ah, but Danish is The Language Without Minimal Pairs.*

    * One of H P Lovecraft’s lesser-known stories.

  80. Lars Mathiesen (he/him/his) says

    I’m sure we have some, like between /hun/ hun ‘female n.’ and /hun̰/ hund ‘dog’. (This time I’m marking stød as creaky voice pour épater les gens).

    And some of the usual suspects, of course. We haven’t merged /s/ and /k/. Yet.

    And for JC’s lists: Stød doesn’t have stød.

  81. J Pystynen said:

    Hungarian, too, allows [cç ɟʝ] in variation and there’s quite a lot of literature arguing if they “really” are stops or affricates…

    Followed by a mention of Russian palatalized consonants from January First-of-May, so maybe this is the right thread to add something I just learned: Hungarians apparently map Russian palatalized alveolars onto their own patalals, since they spell Vladimir as Vlagyimir, Putin as Putyin, Nikolai as Nyikolaj. (If I understand correctly, the Russian [dʲ] isn’t the same as the Hungarian [ɟ] or [ɟʝ] — though I’m no expert and probably couldn’t reliably distinguish them — but the point is that Hungarians distinguish it from [d].) The Hungarian spelling sztyepp also suggests a borrowing directly from Russian rather than through German, French, or English. The latest Hungarian translation of Anna Karenina is titled Anna Karenyina, although earlier translations spelled it with n instead of ny.

  82. David Eddyshaw says

    An improvement over my previous least-likely-looking pair of Oti-Volta exact cognates meaning “river”: Akaselem ŋkpe, Mbelime wúónú. (Cheating, but only very slightly: Akaselem has secondarily acquired a set of class prefixes, which is what the ŋ- part is; apart from that, every segment corresponds. The words are even both overtly marked as belonging to the same noun class, the singular of the “long thin things” gender. Akaselem saves time and effort by deleting all non-root proto-Oti-Volta syllables completely, unless the preceding syllable was closed. It’s almost as bad as French.)

    I’ve added an appendix listing all the comparative series now, following on from JP’s suggestion. A step towards a proper etymological dictionary.

  83. A bit of googling turns up the Greek/Latin pairs Christ and fricassee on the one hand, and Pepsi and cook on the other.

  84. JC, sounds like you’re hungry.

  85. Christ and fricassee? Don’t believe everything Google shows you. Some anonymous joker on Reddit said that fricassee was derived from the root meaning ‘to smear’ (as in Christ, ‘the anointed’), referring to “the application of sauce”; that was bullshit and/or trolling, as a few seconds in any dictionary will tell you. For one thing, fricassee is not prepared by smearing sauce on, but by stewing cut-up pieces in sauce. For another, all current dictionaries have French fricasser derived from Latin frīgere ‘to fry’, either with a frequentative suffix (Wiktionary) or as a blend with some other verb meaning ‘to break’ or ‘to press together’; some sources say the Latin verb is cognate with Greek and Sanskrit verbs for ‘roast, bake, fry’, but nobody says it’s from the PIE root *gʰrey- meaning ‘to rub, smear’ .

    Christ *is* cognate with grime, at least according to the AHD Appendix.

  86. David E.: I’ve added an appendix

    Appendopexy! Unorthodox, doctor.

    (I just performed suffoplasty, myself. Not sure about the result.)

    Edit: Almost forgot: Good news. Thank you!

  87. Yes, I think I speak for the entire Hattery when I add my congratulations and thanks!

  88. Trond Engen says

    My son notes that suffoplasty can be generalized to affoplasty by the common procedure of prefoplasty.

  89. David Eddyshaw says

    Infoplasty is more challenging.

  90. Chalfuckinglenging.

  91. Trond Engen says

    Unplastydoable

  92. David Eddyshaw says

    Academia.edu has just sent me an email informing me that a “David Eddyshaw” is mentioned in a paper by a certain “David Eddyshaw”; it asks me whether this lucky mentionee is me.

    I’m tempted to answer “no”, but I’m afraid of destroying the logical structure of reality if I do that.
    The enquiry does raise some deep philosophical questions, though. Quite Borges …

  93. “No, I am not Pierre Menard, Author of the Quixote, I am another Pierre Menard, who happens to have rewritten the Quixote. I resent the ongoing confusion.”

  94. Lars Mathiesen (he/him/his) says

    I am cursed by the existence of numerous people named Lars Mathiesen who actually are listed as authors of articles on academia.edu. Presumably (though I don’t have high hopes) this would not be a problem if my namesakes were uploading the articles themselves, but as it is I get emails about once a week asking if some article on organic chemistry or dam engineering was by me. I no longer give them a slice of my time to inform them that I am not a researcher at all, since the first 50 times didn’t seem to have an effect.

    A few weeks ago it claimed to have found my name on an article in “Historical Linguistics” which was pretty intriguing, but I’m sure it’s just lying to make me buy a membership. (Or they are trying to let GPT-4 figure out how to lure me in, in which case it’s not even lying).

  95. I have an uncommon name (first-last combination, but my last name is especially uncommon, well both are — suffice to say when I tell people my first name, they might think it’s a pseudonym if they’re not Bulgarian — they might be more familiar with it from a popular fantasy series). The only other person I know with that combination of first and last name is a sculptor who lives in Barcelona I found on facebook. An ex of mine also shares an uncommon first-last combination name with an infamous Bulgarian ultra-nationalist politician, also rare. Think Marie Le Pen.

  96. David Marjanović says

    There’s Marine, and there’s Marion.

    I get spam from academia.edu, but haven’t been told I mention myself yet; apparently that’s only because I haven’t uploaded anything there. (I’m on ResearchGate. And so is at least one of my namesakes.)

  97. Trond Engen says

    Even I was told my name was mentioned in an africanist linguistics paper just the other day. Strangely enough not long after I opened David’s little booklet.

  98. David Eddyshaw says

    My ophthalmic alter ego is actually the usual cause of spam from academia.edu, despite the fact that I/he has never uploaded anything there at all. I’m often somewhat bemused at the way my slender medical oeuvre is apparently cited in medical fields that I have no knowledge of whatsoever, but have latterly concluded that this is probably more about the desire of the writers to have a good long “References” section to establish that they are Serious Scholars, rather than the seminal character of my work.

    “Index I copy from old Vladivostok telephone directory.”

  99. When I google my name it’s only things I have posted in the ’90s. And the guy who does sculptures.

  100. I think I mentioned that before here, but I get the question whether I’m the author on tons of papers on agriculture and plant genetics. There seems to be a much-quoted guy going as “H.” plus my last name.

  101. I have somehow (I don’t know precisely how) managed to stop these e-mails.

  102. David Eddyshaw says

    I get the question whether I’m the author on tons of papers on agriculture and plant genetics.

    Urs Niggli, worthy describer of Toende Kusaal, Ninkare Farefare and Kassem, presumably has more or less this exact same problem. Practically all the Google hits for “Urs Niggli” lead to the President of the Swiss Institut für Agrarökologie. While it would be pleasing to imagine that U.N. is a true Renaissance Man, with more than one very different string to his bow, one suspects that there may in fact be two distinct individuals involved here …

  103. There seems to be a much-quoted guy going as “H.” plus my last name.

    Out of (my) idle curiosity — no reason you should trouble yourself — have you followed up any of those alleged citations?

    What I find is that rarely is there anybody with my last name. If there is, they don’t match my initial. If they match my initial (hardly rare), they don’t match my last name. I think they’re just making up the whole gig.

    I don’t see how this helps with their ‘business model’.

  104. Trond Engen says

    I think they’re just making up the whole gig.

    Not all of it. I do get occasional e-mails about specific archaeology papers, which seem to be real papers by my hapless namesake the professional archaeologist.

  105. Lars Mathiesen (he/him/his) says

    Following up — they don’t show you the paper unless you pay. And I have no reason to pay otherwise, so I consider the annoyance to be the price I pay for the convenience.

  106. @AntC – yes, the papers and quotations are real. I just don’t have anything to do with them.
    There is a small number of papers that actually do refer to me, where my participation in discussions of draft papers on IE topics is mentioned. That’s why I bother to check every now and then.

  107. AcaDotEdu’s ongoing downfall now also seems to mandate their new Twitteresque timeline for everyone, instead of the previous activity view (and may or may not have zapped discussion sessions while they’re at it).

    Cheers for the POV appension!

  108. David Eddyshaw says

    You’re welcome: I ought to give you a credit for suggesting it …

    (I’ve actually already found it quite useful myself in the ongoing effort of improving the reconstructions.)

    There’ll be an updated version of the thing at

    https://doi.org/10.5281/zenodo.7979828

    on Monday for anyone who is interested but wants (naturally enough) to avoid academia.edu.

    I’ve made quite few changes over the past few months, mostly as a result of paying more attention to data from the Gurma group, where my comfort-zone languages Moba and Gulmancema have turned out to be quite unrepresentative of the whole group in some ways.

    Also more stuff on the external relationships of Oti-Volta as a whole, which are now treated in a somewhat less perfunctory way than before.

  109. I really like the stuff on the external relationships, but I want to take issue with this:

    Although (pace Stewart 2002) it is methodologically suspect to leap over intermediate steps in reconstruction, it is interesting to compare with proto-Bantu, which has been reconstructed in detail.

    That’s on the supposition that intermediate groupings are actually identifiable and that corresponding intermediate levels can be reconstructed in some detail. If not, you work with what you’ve got, without a priori forcing a tree on the data. Thus, if a rudimentary reconstruction of PVC is possible from POV and PB alone, and no other branches are approaching the same level of historical analysis, it should be done. Then one can add other languages and branches with greater confidence, and as those branches find their place in the family tree, intermediate levels can be hypothesized and the reconstruction of the proto-language modified,

    You almost say that in the summary:

    Despite the difficulties mentioned in §7.2, it seems very probable that Oti-Volta and Grusi are, as Manessy supposed, two major branches of a genuine Central Gur genetic group. However, Miyobe may be more closely related to Oti-Volta than Grusi is, and Baatonum may prove to be closer to all three than Koromfe is.

    The relationship between Oti-Volta and Bantu seems to be comparatively close. This may be to some extent an artefact of the fact that it is possible in this case to compare reconstructed protolanguages, but it is strikingly easier to find cognates between Oti-Volta and Bantu than between Oti-Volta and some of the more peripheral Western “Gur” languages. It may well prove to be the case that Volta-Congo languages with class prefixes and languages with class suffixes do not in fact represent two genetically distinct Volta-Congo branches, but that the true genetic groupings will turn out to cut across this division.

    Oti-Volta also appears to be relatively conservative phonologically compared with many other Volta-Congo subgroups. It seems likely that further comparative study of Oti-Volta will be valuable in reconstruction to a deeper level.

    … but you might as well own up to it upfront.

  110. David Eddyshaw says

    If not, you work with what you’ve got, without a priori forcing a tree on the data. Thus, if a rudimentary reconstruction of PVC is possible from POV and PB alone, and no other branches are approaching the same level of historical analysis, it should be done

    That was pretty much Stewart’s own take on the matter. I take your point, but I still feel vaguely sinful about what I’ve done there. Video meliora, proboque; deteriora sequor. It may well often be the only practical way forward for now, but it’s important not to lose sight of the fact that it’s only a kind of temporary fix.

    Stewart himself ended up in some pretty strange places as a result of his strategy. He was particularly an expert on Potou-Akanic, and he ended up projecting some very idiosyncratic features of that group onto his Niger-Congo protolanguage in a way that really didn’t work at all, as he seems to have come to see himself, latterly.

    I understand the temptation very well myself: I’ve had to work at correcting my own Kusaalocentric bias in reconstructing POV, and as I was just saying, latterly discovered that the two best documented Gurma languages actually seem to be (a) quite closely related to each other and (b) not all that representative of Gurma in general. I would have avoided some mistakes if I’d taken more care with reconstructing proto-Gurma (so far as that can be be done) rather than assuming that the two most convenient Gurma languages would do for my comparative purposes.

    (Moba also shows quite a number of signs of having been influenced by contact with Western Oti-Volta, making it even less of a good stand-in for proto-Gurma.)

  111. I agree with all that. I most certainly don’t mean reconstruction based on lookalikes in single languages, but POV and PB are substantiated and reconstructed intermediate levels, and there’s reconstructed morphology and syntax to work with. That’s exactly what make them suitable for reconstruction without trying to add underdescribed branches and unsubstantiated intermediates,

    Maybe my point is that if low-level and high-level reconstruction can take place in parallel, they can inform eachother. Stewart’s reconstruction of PVC may be untenable, but how it breaks down in comparison to the results from other branches could still be very informative for understanding the position of the Potou-Akanic* branch.

    * That took me to Wiktionary’s appendix for Proto-Potou-Akanic reconstructions. I gather that Grimm’s Law took place before the breakup of Scandi-Congo.

  112. *-pĩũĩ is an extremely good word for ‘blow (nose).’

  113. January First-of-May says

    That took me to Wiktionary’s appendix for Proto-Potou-Akanic reconstructions.

    Proto-Potou-Akanic reconstructions previously on LH. TL/DR: there might be some font problems involved.

    EDIT: I was unable to figure out how and where I found the original text of Stewart’s apparently-unpublished 2004 article a year ago; my Google search attempts today didn’t find anything relevant.
    I did find some other articles by him that give the reconstructions, in clearer font; it turns out that ʋ̃ is the consonant and ʊ̃ is the vowel, and whatever OCR was being used for the text of the 2004 article presumably couldn’t quite tell one from the other and assumed instances in the same word to be all the same thing.

  114. David Eddyshaw says

    an extremely good word for ‘blow (nose)’

    True.
    Perhaps a bit too good …

    “Blowing” words beginning with p or f tend to have star-billing in long-rangers’ lists of wannabe cognates. (Potou-Akanic is of course perfectly real, but that doesn’t mean that every proposed reconstruction within proto-Potou-Akanic is real.)

    As it happens, “blow the nose” is one of the tiny number of non-clitic words that might have had initial *f in proto-Oti-Volta (and three of the others mean “blow”, as in “wind.”)

    (The other verb for “blow the nose” attested in most of Oti-Volta is a derivative of the root “nose/snot” itself, with the all-too-clear literal sense “de-snot.”)

    it turns out that ʋ̃ is the consonant

    Makes sense: that’s basically just IPA ʋ. Bad choice notationally, though, as this illustrates.

    (Furthermore, many local orthographies in West Africa, especially in Francophone countries, use ʋ for IPA ʊ; the standard Agolle Kusaal orthography does, too.)

  115. Trond Engen says

    (I forgot to mention that my fingers wrote Poitou in both instances, but unfortunately my eyes noticed it immediately.)

    @J1M: Wikipedia in Potou-Tano languages has a now dead link to a pdf of Stewart’s work,

Speak Your Mind

*