Via Fiasco da Gama’s MeFi post: Auslan, the majority Australian sign language, has a visual dictionary with three recognised signs for ‘holiday’, of which the second is noteworthy. The first comment, by prismatic7:
Oh yeah. […] you think hearing Aussies are sweary, cynical, and irreverent, you’re gonna shit meeting Auslan speakers.
I have a very dear friend whose Auslan name is essentially “short but big tits”, and she has a friend whose name translates as “the guy who put his dick in a vacuum cleaner”.
Auslan is wild.
Sounds like my kind of sign language, and the dictionary is great.
The WP link tells me that Auslan is closely related to British Sign Language, and reminded me of the pleasing fact that the not-mutually-comprehensible American Sign language is, instead, derived from French Sign language.
The Auld Alliance!
…as is Austrian but not German sign language.
I wonder if anyone has produced one of those nice world maps of language families, done for sign languages instead of spoken?
Right there.
Yes, just found one myself. This Google thing is quite handy.
https://www.theodysseyonline.com/6-important-asl-deaf-culture
A better image of the same map:
https://external-preview.redd.it/WNcbFjwYRop–xFCwKsqziHu1pfUMv0HB6Pkss3yrS4.png?auto=webp&s=ba2914c1b6172cc15616fdc0d8550529455e18f8
Official information on Northern Ireland’s two sign languages written in its other three languages.
Japanese Sign Language adapted FSL/ASL fingerspelling to create a full syllabary.
https://faroutliers.wordpress.com/2007/07/04/fingerspelling-from-alphabet-to-syllabary/
Somewhat derivative from JSL, Taiwanese Signers rely less on fingerspelling than “airwriting” big hanzi. I don’t know if they write them backwards for their audience’s orientation.
Korean Sign Language relies on fingerspelling with finger shapes that resemble Korea alphabetic characters.
Don’t have any idea how irreverent JSL/KSL/TSL signers may be.
The JSL sign for “Friday” (“gold-day” in kanji) is the same as the ASL sign for “F” (thumb and forefinger in circle with remaining three fingers raised).
From Bonny Sands’ paper Africa’s LInguistic Diversity:
Madagascar sign language is related to Norwegian sign language (NSL)
!!!
Knowing almost nothing about sign languages, I have long thought it unfortunate that a chance for universality seemed to be missed. We could all have recourse to an Esperantoid version, ugye? Dreams of an outsider.
It appears that the Young People of Today are increasingly borrowing ASL vocabulary into their Auslan; perhaps universality will come about through the remorseless Americanisation of the world, rather than Zamenhöflich Utopianism.
It may be time to promote Maganar Hannu, to avoid the colonisation of our minds.
We will add your linguistic distinctiveness to our own. Resistance is futile.
Madagascar sign language is related to Norwegian sign language (NSL)
Must be due to the prevalence of lude fiske in Antananarivo and the large herds of sifakas in Bergen.
Y: Since Protestantism entered Madagascar before Catholicism did (and indeed Protestants outnumber Catholics in Madagascar today), I am going to take a leap and guess that the fact that Madagascar Sign Language (MSL) is an outgrowth of Norwegian Sign Language (I share your surprise) is a consequence of some Norwegian missionaries having founded the first school for deaf pupils in Madagascar (Can any Hatter confirm or deny this?)
(Hmm. Here is a trivia question for Hatters: is Madagascar the only country on earth whose dominant spoken language was transplanted from a distinct continent within historical times, and its dominant sign language from yet another -third-continent? Off the top of my head I can think of no other).
Noetica: Plains Indian sign language was used as a lingua franca over a rather large area of the United States, and I strongly suspect other sign languages must have played a similar such role in other parts of the world (David Eddyshaw: I would be very surprised if this was not the case in some of the more linguistically diverse parts of Africa: you wouldn’t happen to know of (any such) such (a) case?), so the vision of a sign language Esperanto as a planetary lingua franca is not that utopian.
Etienne: I’m absolutely sure that’s what happened in Madagascar. WP says that the language came to be c. 1950 and that “Seven deaf schools in Madagascar are being sponsored by Evangelical Lutherans.”
The flip side of your African question is, how did a sign language come to be used as a lingua franca in the Great Plains, rather than some existing spoken language, or a creole, as elsewhere in the continent?
ノルウェー手話とマダガスカル手話の予備的比較研究
I would be very surprised if this was not the case in some of the more linguistically diverse parts of Africa: you wouldn’t happen to know of (any such) such (a) case?
I don’t think it’s a thing in West Africa, perhaps because well-established lingua francas have long been common there, even before the European invasions. Maybe elsewhere … or maybe Lameen knows of such things.
In the highly multilingual area I actually know of first hand, people who aren’t content with simply farming their own fields are usually either multilingual themselves or at least conversant with a local or regional (spoken) lingua franca. In the 1990’s in Bawku, that was Hausa or Mooré, with Kusaal itself fulfilling the lingua-franca role locally for the many Bisa people living in the district, and the occasional Bimoba person. (Unlike all the other local languages, Bisa is Mande rather than Volta-Congo, and it’s extremely unlike its neighbours, despite a good few Mooré loanwords.)
I once had a memorable consultation with a patient from western Burkina via a chain of three interpreters, English-Mooré, Mooré-Dyula, and Dyula-Whatever-it-was-that-the-patient-spoke (I never found out.)
“The linguistic diversity represented by Africa’s sign languages alone is greater than that which is widely assumed for the continent’s languages as a whole.”
“Widely assumed” is an accurate formulation. (the treatment of the notion of “diversity” is less accurate of course:). The excerpt sounds interesting.
Per Wikipedia, Norwegian SL itself < Danish SL < French SL.
@drasvi:
It’s an interesting paper, but I don’t seem to be able to find a downloadable pdf of it online any more. Still, that should be no problem for a Russian …
It’s not particularly iconoclastic, at least not from my point of view; the only unequivocal deviation from Greenbergian orthodoxy on Niger-Congo is that she thinks Mande is unrelated, and I think this is nowadays the communis opinio. She points out that “Atlantic” is extremely diverse even internally: “Estimates of [lexical] cognacy range from 3% (Pozdniakov 2007) to 8% (Childs 2003)” and also goes into the fact that Kordofanian is plainly not itself a unity.
Like most people, she’s not sold on Nilo-Saharan: “The growing consensus is that Nilo-Saharan represents a minimum of four distinct lineages. For the present, Nilo-Saharan is best considered as a referential grouping, and not a language phylum or family.”
She herself is an expert in “Khoisan”, and like all those with any actual firsthand knowledge of the group these days is briskly dismissive of Greenberg’s construct:
“Specialists in long-distance relationships”. I like that. In other words, “authorities on making something from nothing”.
Y: My suspicion (no more than that) is that Plains Indian sign language as a general lingua franca arose/spread (?) as a result of the introduction of the horse, bringing into close contact, at a very quick pace, a score or more of structurally distinct and mutually unintelligible languages, none of which enjoyed prestige over the whole area: as a result I suspect Plains Indian sign language became a lingua franca “by default”: one was needed, no spoken language had the prestige to take on that role, not even in pidginized form, and so sign language took on that role.
Yes, Madagascar was (and still is) especially important for the Norwegian Missionary Society, The first missions were established in the mid-19th century, soon after the ban on Christian lay preaching was removed in 1842. The mission in Madagascar focused from the start on schools and healthcare, especially the fight against leprosy (not entirely coincidentally, Mycobacterium leprae was identified as the cause of the disease by Armauer Hansen at the leprosy institute in Bergen in 1873).
Anecdote 1: I took a class on African history in university back in the nineties. The lecturer told a story of a former student of his, a young woman from Madagascar, who had graduated from Norwegian missionary schools and came to Norway on a scholarship from the Society. She was first enrolled at the Norwegian Missionary Academy in Stavanger, the alma mater of all her teachers, and had (in my lecturer’s retelling) been envisioning her arrival at the Spiritual Center of Christianity. Instead she met a thriving international city in the middle of an oil boom and Norwegian students who may have been conservative Christians from a Norwegian POW, but they were young women (and men) who didn’t accept traditional authority or male superiority without a fight and who had very little regard for the public morality of their parents and grandparents. Her first term was a shock, but she soon came to admire the independence of her fellow female students, and after graduation she went on (IIRC) to take a degree in education from the University of Trondheim.
Anecdote 2: I have a friend who grew up in a missionary family in Madagascar. She tells about a similar culture shock when returning to Norway for good in her early teens. In Madagascar they were well integrated in their community, and whatever their friends did, so did they. Once the family made a trip to a bigger city, her younger brother who was about five at the time (and whitish-blond), spotted a busload of French tourists emptying into the town square. So he did like the kids back home and pointed and shouted (as only a five-year-old can) in Malagasy “look at all those white people”.
Addendum on the topic of the early spread of Protestantism in Madagascar (Ah, being a reader + participant within the Hattery does lead to unexpected places…): if the description of this book is any guide, the first missionaries were not only British, but British missionaries whose relationship with the Celtic periphery of the British Isles had an impact upon their vision of the culture of Madagascar:
https://librairie-7ici.com/6684-les-premiers-missionnaires-protestants-de-madagascar-1795-1827.html
Specifically, the Welsh part of that periphery, by the look of it:
https://en.wikipedia.org/wiki/David_Jones_(missionary)
Doesn’t go into how the Welshitude coloured their vision of Madagascar, unfortunately. The WP article seems to have been rather inexpertly edited by someone with some sort of family connection to Jones, and who hasn’t quite worked out how Wikipedia functions.
Auslan.
It is NOT tame.
“look at all those white people” – https://en.wiktionary.org/wiki/vazaha#Malagasy
“vazaha
1. white person
2. foreigner
This word comes from the time of Queen Ranavalona I, who, in her desire for “Malagachisation” had foreigners listed (in Malagasy “list (make a census)” is called voazaha). Further linguistic etymology is unknown, with no known Austronesian cognates.”
It’s kind of frustrating how much contradictory information there is about sign language families. From Wikipedia I’ve found out that:
* Swedish and Danish SL are mutually intelligible
* Swedish SL may be related to BANZSL
* Danish SL may be related to French SL
…and of course BANZSL and French SL are well-known to be completely unrelated. Go figure!
Historical comparative work must present major challenges with sign languages. (The Japanese paper drasvi linked to above gives an idea of this; it’s in English, BTW.) The criteria for cognacy, and what actually constitutes a plausible historical change in the case of signs, as opposed to words made of phonological segments – all that would have to be worked afresh out almost from the ground up, to avoid the whole exercise just being impressionistic.
New Zealand Sign Language online dictionary — the signing for holiday seems way up-tight compared to the Aus link @Hat provided.
Allegedly British, Aus, NZ SLs are tightly related — up-thread somebody mention BANZSL — see family tree[wp].
NZSL is an officially recognised language here — notionally equal status with English and Te Reo Māori. There’s a Gerrit Van Asch [Dutch] castigated for suppressing NZSL at schools for the deaf. The school bearing his name is still going in a suburb of Christchurch.
“Auslan” immediately reminded me of Pte J McAuslan, the Tartan Calaban, genius loci of Glasgow.
https://en.wikipedia.org/wiki/The_General_Danced_at_Dawn
“A man of principle, McAuslan believes that being ordered to enter the pillow fight at the Army’s Highland Division Games is an illegal order.”
The Russian expression is “to throw-self with-pillows”. As in “let’s ….”.
Both times it is recognised as a genre – but in Russian it is a verb, while in English it is a Named genre.
Also in English a fight is implied.
Also in Russian it is about hurling projectiles (which per se is an occupation, that involes aiming, hitting or missing the target and of course flying thinngs all around) while a fight does not necessary implies hurling anything.
Which does not mean of course that a Russian child won’t (or I don’t) hit his/her neighbour with a pillow without releasing it.
That’s how Chinook Jargon came about. Before European sea trade, people knew the languages of their neighbors, but not much further than that.
I don’t know of any case other than Plains Indian Sign Language where a sign language took up such a role. I don’t know enough about Australia, though.
I see your Starostin 2003 (a lexicostatistical work) and raise you Starostin 2007 with actual nontrivial sound correspondences.
I see your Starostin 2007 and raise you a quote from Tom Güldemann, in the recent The Languages and Linguistics of Africa: “Some recent lexical studies like Starostin (2013) effectively conclude with the abandonment of the all-comprising genealogical hypothesis.” That refers to Языки Африки. Том I. Методология. Койсанские языки, Moscow (which I haven’t seen myself).
Grigory Starostin can’t tell if his Peripheral Khoisan (including Ju and Tuu above) is a group. He can tell what is a promicing direction for comparative work.
“Still, that should be no problem for a Russian”
if this Russian’s name is Alexandra Elbakyan and she’s a Kazakhstani Armenian…
Güldemann himself is not hostile to the idea that future work might demonstrate further relationships among the various “Khoisan” groups, e.g.
https://www.researchgate.net/publication/319528342_Khoisan_linguistic_classification_today
Curses! Foiled a…
It’s here, but I never took a look inside because I haven’t found the time to read that much Russian. 🙁
GT from Grigory Starostin’s book,
I copied a random line from the book:
dog *ʰoe 1 ɕeama *ǂUiŋ 1 *ǂqʰa- 1
—-
is a Unicode character from private use area (underlined ǂ), 1 is subscript. So it’s a clean pdf with the original text layer (I downloaded it from the library genesis, not academia).
ǂ̱?
A universal and indispensable condition for justifying linguistic kinship should be the common basic vocabulary of the languages under consideration, with regular phonetic correspondences defined on it.
A necessary condition, sure; a sufficient condition, no. In particular, if you can’t produce any morphological evidence, your theory is weaker (like it or not), and if your lexical evidence is not overwhelming, you may no longer be able to make a plausible case for a genetic relationship to anybody not already predisposed to believing in it anyway. You will also need to be able to explain why there is no morphological evidence (e,g, because of phonological attrition, of a kind that you have independent evidence for.) It’s no use just declaring that the issue doesn’t matter.
but only within the framework of chronologically shallow families – using morphological comparison to justify deep kinship, with rare exceptions, is rather futile
Quite so: to establish truly deep kinship, we must learn to do without the really sound evidence.
grammatical morphemes can, apparently, be considered on an equal footing with “medium stable” elements of basic vocabulary
“On an equal footing with medium stable elements” is nonsense. The major difficulty with establishing genetic kinship, apart form sheer chance resemblances, is separating similarities due to contact* from those due to inheritance. Morphology certainly can be borrowed, but it is simply not true that this happens more often than with core vocabulary. If your theory of long-distance relationship depends on this, then there is something wrong with your theory.
Incidentally, much the best evidence that Afro-Asiatic really is a genuine genetic entity is morphological. The lexical evidence is much, much weaker than the Ehrets and their like make out. It’s precisely for this reason that despite the almost certainly dizzying time depth, Afro-Asiatic is accepted (Omotic apart) by splitters as much as lumpers.
* Which, note, can easily result in many numerous absolutely regular sound correspondences in core vocabulary. Sambiéni reconstructs amaa “but” to Proto-Eastern-Oti-Volta by perfectly respectable comparative methods; and much English vocabulary corresponds entirely regularly to French.
Yeah, what DE said. If you’re trying to downplay the importance of morphology, you’re basically justifying sloppy methodology so you can more easily show what you want to “prove.”
“prove”
@LH, where exactly G.S. says that he has “proven” something or “shown” something? If you don’t know, then why these quotemarks? His style is like this:
Если для (гипотетического) правост.-кхой. языка реконструкция *ábà однозначна, т
….
При условном допущении реальности «общекойсанского» или «макрокойсанского» таксона
….
Напротив, «предварительная» нетривиальная система соответствий между, например, сев.- и юж.-койсанскими языками, описанная в работе [Starostin 2008a], в рамках которой одному сев.-кой. кликсу могут соответствовать несколько юж.-кой. и наоборот, в ходе «ручной коррекции» сознательно не была отражена, т. к. обнаружение и проверка таких соответствий — задача уже следующего этапа работы.
……
В ситуации с хадза все достаточно наглядно: проценты совпадений настолько ничтожны, что говорить даже о сугубо гипотетической возможности генетического родства хадза с другими «койсанскими» языками не имеет смысла до тех пор, пока аналогичные сопоставления не будут проведены между хадза и хотя бы некоторыми (желательно — всеми) из остальных языковых групп африканского континента.
….
В любом случае ситуация такова, что говорить о «макро-койсанском» родстве (даже без хадза) можно будет по-настоящему только после того, как (а) этимологический анализ подтвердит — или опровергнет? — гипотезы более низкого уровня («периферийно-койсанскую» и «сандаве-кхой-квади»); (б) будет проведено предварительное автоматическое и откорректированное вручную сопоставление данных «койсанских» групп с материалами по другим языковым группам Африки.
…
Таким образом, предварительный анализ 50-словников, составленных для всех «койсанских» праязыков 1-го уровня и языков-изолятов, не дает позитивных свидетельств в пользу «макрокойсанской» гипотезы, но позволяет свести девять койсанских «мини-узлов» к трем:
…
Необходимо учитывать, что эти цифры не отражают результаты строгого этимологического анализа на основании сравнительно-исторического метода, а представляют собой сугубо предварительный результат, на базе которого впоследствии будет производиться более тщательная обработка данных для определения хронологически более глубоких таксонов и реконструкции прасписков базисной лексики «2-го уровня».
It doesn’t matter whether you call it proof or условное допущение, the point is you’re making things easier on yourself by arbitrarily removing a source of difficulty. Morphology doesn’t become less important because you say it is.
He didn’t say that. He divided the core (“basic”) vocabulary into three groups from most to least stable (all of them being more stable than the rest of the vocabulary), and said grammatical morphemes (no mention here of morphology as a system) are “apparently” as stable as the second group of the core vocabulary…
…and that’s “stable” against all sources of change put together, not only specifically against borrowing.
So which subset of “core” vocabulary is more stable than grammatical morphemes; and what does the group of “grammatical morphemes” consist of? (From the fact that you’ve raised the subject, I presume that this must include a lot more than flexional and derivational affixes.)
This still seems like dodging the question of the role, specifically, of morphology in justifying proposals of genetic relationship, and the fact that if you can’t reconstruct any morphology then your proposal is going to be a lot less compelling to any disinterested judge.
I must admit that in Oti-Volta, the various bound preverbal particles that mark tense, polarity and mood are quite strikingly variable. This is the case even within Western Oti-Volta: Farefare and Mooré, for example, have no sign of the distinction between indicative and irrealis preverbal negative particles which appears in other WOV languages, but have generalised (oddly enough) the irrealis form. (On the other hand both nominal and verbal morphology are pretty easy to reconstruct for Proto-WOV.)
And reconstructing verb flexion for Proto-Oti-Volta is extremely difficult* (Manessy simply gave up.) On the other hand, there is a great deal of reconstructable nominal morphology, and a significant amount of reconstructable derivational verb morphology.
* But I think I can plausibly explain why it’s difficult: the basic imperfective flexion in the protolanguage was *-u, and the phonological distinction between non-root /i/ and /u/ was independently lost after non-velars in most OV branches (a development clear on other grounds); with the result that each branch had to create new ways of marking the fundamental perfective/imperfective distinction in flexion.
@DM still I would love to see the argumentation.
Yes, here morphemes are compared to average/median CORE vocabulary items (possibly not even in the list of 100 most stable word, but in the list of 50 words).
But (1) morphemes also must differ in stability (2) why?
“…and that’s “stable” against all sources of change, not only specifically against borrowing.”
True, though these are different concerns. Borrowings and other sources of convergence diminish the role of existing correspondences in your estimate of the likelihood of a link. Other sources of change just make relatedness less salient. There is no symmetry between borrowings and other sources of change.
I don’t see how arguments about what is or isn’t typically stable help in figuring out language relationships, long distance or otherwise. You take what you get. Sometimes you have lots of lexical resemblances but the grammar has been restructured beyond recognition, or old stable grammar but a reinvented lexicon (I’m saying ‘grammar’ to cover isolating languages with no “morphology”.) Each case is different.
OK, I somewhat take it back. If you have a bunch of likely meaningful resemblances and you’re trying to (or you have to) short-cut through the comparative method to figure out if the resemblances are due to contact, then such statistical methods are useful. For example, Heggarty’s work on showing that Quechua-Aymara lexical correspondences occur mostly in the non-core vocabulary and hence are likely due to old contact.
Starostin has a bunch of recent papers on lower-level reconstruction in Khoisan, published in JOLR. In a 2018 paper he says that Ju and ǂHoan are related (using lexical data), at the level of Fenno-Ugric (is that still a thing?) or Kartvelian.
Actually, assuming that Starostin the Younger is in fact referring to “grammatical” morphemes other than affixes, I have some sympathy with his proposition that such things are less stable than you might have thought, and can readily come up with examples.
This was (for example) what I was on about myself in saying that the famous noun class system of “Niger-Congo” is a good bit less solid* as open-and-shut evidence for genetic relatedness than commonly supposed.
Though my take on it is not that it shows that such things aren’t important in demonstrating common descent, but on the contrary that exactly because the markers seem not to have been actual affixes in the protolanguage, but bound words with a position that was not fixed relative to the head noun (and moreover with known cases of borrowing between some languages) their probative value is reduced. So there needs to be better evidence from lexicon than would otherwise have been needed, to get to a similar level of reasonable belief in the genetic link.
In other words, in undermining (probably justifiably) the idea that grammatical words are especially stable, Starostin is making his long-distance proposals that much harder to substantiate.
* Still solid enough to be going on with for Volta-Congo, though. At least three classes are reconstructable as a bare minimum: human sg *ŋu pl *ba, non-human sg *lɪ pl *ŋa, liquid *ma. All the major Volta-Congo branches have evidence of those, and that doesn’t look like something that can plausibly be explained by chance or even by borrowing.
“Starostin is making his long-distance proposals” – his proposal is the same as Güldemann’s.
he says that Ju and ǂHoan are related (using lexical data), at the level of Fenno-Ugric (is that still a thing?) or Kartvelian
Kartvelian is surely nowhere near as diverse as that? Even Svan is not that remote from Georgian, say, once you undo its Old-Irish-like habit of missing out every other vowel for the lolz. Sure, the lexical matches are fairly low, but that is a great illustration of the limitations of lexicostatistics when it comes to telling you about language relationships. In fact the contrast seems rather neatly to prove my point about the relatively high value of morphological evidence for demonstrating relatedness.
Brute lexicostatistics is not going to get around Güldemann’s points about borrowing within the South-African Sprachbund either; particularly not if it’s done by comparison of surface similarity of words in modern languages without prior establishment of historical sound changes relating the forms. It’s exactly that sort of thing that can distinguish loans from common inheritance. Or not. If you don’t bother with it.
If we’re going to trust lexicostatistics to tell us anything other than the degree of relationship between languages already shown to be related (which we should not, but never mind) the numbers look odd, in terms of relatedness. I count about 50% cognates in core vocabulary between Mampruli and Mbelime, two Oti-Volta languages which are barn-door-obviously related, with many equally obvious cognates in nominal flexion and in verbal derivational affixes. This number is boosted a good bit by the fact that I have traced a fair number of far-from-obvious historical sound changes too (like my favourite Mbelime wuonu = Mampruli kuliga “river.”)
The words “depending on the degree of strictness required from phonetic correspondences” in the summary do not inspire confidence.
Oops …
Güldemann’s paper that I linked about does actually regard ǂHoan as definitely related to Ju, as the two main branches of what he calls Kx’a, one of his five distinct families within Greenberg’s “Khoisan.” So not long-range at all, chez Güldemann. In fact, “lower-level reconstruction in Khoisan”, exactly as Y said.
[Heh: Ninja’d by drasvi. Though I still stand by my tirade about lexicostatistics, on principle.]
If Starostin fils really is not making any more distant proposals than Güldemann these days, I welcome his conversion to the One True Splitterism (the more so as I’ve found some of his papers on what lexicostatistics can and can’t achieve very sensible.)
I didn’t read the summary of Starostin’s paper carefully enough, either.
I called him Grygory above – he is Georgy, of course. Sorry:(
@DE, no.
Muscovites are fellow travellers. We understand “reconstruct smaller units, then reconstrcut larger units”. We are dumb. Незатейливые увальни we.
Starostin:
[NK ǂHoan] [ǃKwi Taa] – “Peripheral Khoisan”
[Khoekhoe Kalahari Kwadi] [Sandawe]
[Hadza]
Where NK is Ju, Kalahari is Tshu–Khwe, groups confidently demonstrated by lexicostatistics in brackets.
He may yet be proved right, but that goes well beyond what Güldemann regards as demonstrated at present. (Assuming that you mean that Starostin regards everything you’ve put on the same line as genetically related.)
You can’t demonstrate the existence a genetic group by lexicostatistics at all. In applying lexicostatistics, you are already assuming that the languages you’re comparing are related. Among languages already reasonably held to be related, it may help you towards subgrouping (and it can’t do even that reliably by itself.)
Anyone who thinks that lexicostatistics can (for example) show that Ju and Taa are related has already made the assumption that Khoisan is a genetic group.
Easy things first…
Kinda. For a long time it was taken for granted that Uralic consisted of Samoyedic on one side and F{e|i}nno-Ugric on the other, largely on lexical grounds actually. Recently, more regular sound correspondences were discovered, and more cognates between Samoyedic and the rest of Uralic were found as a result; also, much of the distinctive Samoyedic vocabulary has identifiable sources (e.g. “4” is Turkic), and given geography there’s plenty of reason to think that much of the rest is borrowed, too. Then, some people wondered if the almost perfectly asymmetric tree (a “Hennigian comb” as biologists call it) traditionally assumed for Uralic, which assumes a steady spread from east to west, is actually misrooted and there’s an East Uralic branch instead that unites Samoyedic with Ugric. The phonological innovations postulated for East Uralic have even more recently been strongly questioned, and the latest paper I remember that needed a tree for Uralic worked best with Samoyedic as the sister-group to all the rest, i.e. Finno-Ugric.
In short, we’ll probably see fairly soon.
On to the harder stuff…
I haven’t read any such papers in years, but the Moscow School has been making detailed statements on the comparative stability of Swadesh-and-other list entries for decades, at least some of them based on actual research, e.g. Yakhontov’s extension of the Swadesh-100 list to 110 with 10 meanings he found to be particularly stable in his fieldwork in southeast Asia.
…Oh, here’s one I haven’t read yet, by Starostin Jr himself. Abstract:
The Moscow School talks very little about grammar, but you’ve probably guessed right what the “grammatical morphemes” are.
Similarly, the Moscow School seems to have collectively given up on grammar decades ago. I do agree that’s going too far; if the evidence is there, of course you should use it. I think there’s an overreaction going on against “if you can’t reconstruct a complete IE-style flexion system, your hypothesis is automatically wrong, wrong, wrong” (or, as someone once put it to marie-lucie, “language relationships are either immediately obvious or forever unknowable”) – keeping in mind that “a complete IE-style flexion system” has not in fact been quite reconstructed for PIE either.
Tellingly, the paper on comparative Dené-Caucasian grammar comes not from the Moscow School, but from Bengtson, and the work on comparative Altaic grammar, other than a short list of suffixes, doesn’t come from the Moscow School either, but from Team Robbeets (keyword insubordination).
But this doesn’t negate the point of treating hypotheses of language relationships as scientific hypotheses. You don’t need to prove them beyond reasonable doubt before you’re allowed to explore them and, say, look for a system of regular sound correspondences (that you need to cross-check the lists of potential cognates anyway, both for lexical and for grammatical morphemes).
The concept of confidence in hypotheses as a matter of degree is particularly well illustrated in Starostin Jr’s lexicostatistical work on Nilo-Saharan-in-constant-scare-quotes, where you can watch his confidence fade out from east to west from “fair enough” over “why not” all the way to “sorry, no evidence, no statement possible”.
But this doesn’t negate the point of treating hypotheses of language relationships as scientific hypotheses. You don’t need to prove them beyond reasonable doubt before you’re allowed to explore them and, say, look for a system of regular sound correspondences (that you need to cross-check the lists of potential cognates anyway, both for lexical and for grammatical morphemes)
Agreed, absolutely.
The concept of confidence in hypotheses as a matter of degree
Yes. That’s where my idea comes in, that the lumper/spiltter “dichotomy” can owe as much to the researcher’s personality type as to their linguistic competence. (And to the linguistic culture they grew up in, come to that: Russians are clearly much more well-disposed, ceteris paribus, to lumpery than Americans are. Let alone cynical Brits.)
@DE, no, sorry for confusion.
These two groups are “testable” (his classification : “confident”, “testable”, “uninterpetable”).
Sounds very reasonable.
other similar observed situations, e.g. for the Greek language, though not with others (Icelandic)
am i missing something, or is there something distinctly odd in that Starostin ben Starostin abstract’s seeming belief that rates of lexical change through language contact over several millenia can be meaningfully compared among (1) the imperial language of a territory ranging from large to humongous; (2) a language with periods of imperial status, periods as a regional lingua franca, and periods as one local language in a quite multilingual region, sometimes with a wide but often diffuse scattering of diasporic speakers; and (3) the overwhelmingly dominant if not sole language of an isolated island in a part of an inhospitable ocean that nobody has a good reason* to go to.
if what you’re aiming to measure is at heart about the effects of language contact, comparing languages that have radically different social and geographic environments, leading to radically different conditions, degrees, and kinds of contact, seems like a solid contender for the worst way to go about it. but is there something that i’m not seeing here by not having read the rates-of-replacement literature?
.
* the commercial whale fishery was a reason, but not a good one. likewise colonizing greenland.
the rates-of-replacement literature
There hasn’t been a lot of work on the subject. The old glottochronologists came up with a rate, allowed that it was variable, and left it at that. More work has been done recently, but it’s still pretty sparse. My understanding is that 1. the Swadesh “typical” rate (86% retention per millenium) is a pretty good mean estimate; 2. There is a lot of variability, which Bergsland and others have used to successfully but unfairly besmirch the whole enterprise as crackpottery, and it is still presented so in textbooks, because people don’t know what an error bar is; 3. Some semantic domains are more prone to replacement than others, as Starostin and others have pointed out; 4. Most importantly, nobody has ever tried to explain where that 86%±/millenium rate comes from, and, clever as I am, I don’t have an idea how to explain it either.
seeming belief that rates of lexical change through language contact over several millennia
I think Starostin ap Starostin is moving (has moved?) towards the much more plausible position that lexicostatistics has more of a role in comparing the relative closeness of the branches of an already-established family, with not even speculative absolute time depths being posited. I have no quarrel at all with that in cases where we have nothing better to go on.
And that actually sounds more dismissive than I intended: in the African contexts I am most familiar with, the gold-standard tests of non-trivial common innovations are often difficult to apply in practice – and, worse yet, can be actively misleading. There are plenty of common non-trivial innovations in the Oti-Volta languages spoken in the Atakora, for example, and it’s only by the happy accident that one of these languages (Boulba) is demonstrably not closely related to any of the others genetically that I can be sure that a great many of these common non-trivial innovations are actually areal phenomena and not evidence for subgrouping at all.
And in fact lexicostatistics bolsters this conclusion, as it supports the idea that, overall, the other Atakora languages are as distant from one another as Yom/Nawdm is from Western Oti-Volta.
Moreover, two of those languages, Ditammari and Nateni, show definite common innovations in morphology, and lexicostatisitcs obligingly agrees in making them a subgroup within the Atakora languages (though still far from as close to each other as the Western Oti-Volta languages are.) It all helps in making that particular subgrouping more solid: one piece of evidence supports another.
So even where we do have other criteria, lexicostatisitcs can be helpful. It’s only a problem when people expect more from it than is logically possible, or imagine that it can produce all the answers all by itself; and Starostin ibn Starostin seems to be very well aware of this. I’m well prepared to accept him as an honorary Good Guy. (I expect he’ll be mightily relieved.)
Ibn. I’ll quote the first part of my comment on this:
It doesn’t matter whether you call it proof or условное допущение, the point is you’re making things easier on yourself by arbitrarily removing a source of difficulty. Morphology doesn’t become less important because you say it is.
Just to claim priority:)
(I wrote this yesterday, but still haven’t finished the second part).
It does matter that G.S. says “preliminary”. Corrected GT:
But what I wanted to write is just what I think about this all (rather than just grumble), and this is where I got stuck:)
I appreciate Christopher J. Henrich’s Narnia reference.
Ha, I didn’t catch that till now!
Bonny Sands:
Starostin (some old artcile in Russian, google translate):
I think Starostin has got better since then. (If not, he’s just wrong.)
It is just some article written between 2003 (where he expressed more careful views and made some preliminary observations) and 2013 (where his method found macro-Khoisan uninterpretable. With the following note:
‘On the contrary, a “preliminary” non-trivial system of correspondences between, for example, North Khoisan and South Khoisan languages, described in [Starostin 2008a], within which one North Khoi. klix may correspond to several southern. and vice versa, during the “manual correction” it was deliberately not reflected, since the detection and verification of such correspondences is the task of the next stage of work.’).
I am only quoting it because Sands attributed reconstructions to Starostin, and Starostin attributed macro-Khoisan to Sands:-)
I won’t correct “clixes”. I don’t like this borrowing, because “click” in Russian is “щелчок” – and there is no need to introduce kliks (to me it looks like borrowing for the sake of borrowing). Note, кликсы – here Russian borrows the plural English form. I don’t know why it happens, but Russian constantly borrows Englihs plurals.
Čipsy are Very Late Common Slavic…
On Hadza, there’s this work from 2008 saying its core vocabulary is actually Afro-Asiatic, so the Khoisan features must be (substrate?) borrowings. Fun, eh?
Yes, and it is in this paper Starostin attributes strong support for macro-Khoisan to tghe book by Sands.
Here the whole article google translated (scroll down, below… below).
FWIW, I heard Bonny Sands give a talk some time in the early aughts. I remember someone asking in the Q&A how many independent language families Khoisan consisted of, and she said three to five, depending on who you ask.
(She pronounced clicks with very enviable ease.)
WP: “Starostin (2013) gives the following classification of the Khoisan “macrofamily,” which he considers to be a single coherent language family.[12] However, this classification is not widely accepted.”
No, it is not true (sorry for flooding here with G.S.) I quoted the paragraph above for fun (becauise of how they speak about each other), but Starostin is not expressing personal views about MK here or elsewhere. I am not even sure he cares about classification – I think he hopes to reconstruct something. If so, I wholeheartedly agree. His proposal is indistinguishable from Güldemann’s above (either he simply agrees with him like Alice Werner agreed with Meinhof’s agreement with Westermann – or he has independent reasons for this. I don’t know).
Also he published results of test based on a simple algorithm from 60s, with manual corrections and he presented the test in the following way (google translate): “That is why we will resort to automatic data analysis only as a means of objectively constructing primary, purely preliminary hypotheses regarding high-level kinship, and we will call the trees obtained with the help of such an analysis “pseudo-trees”, in order to emphasize not even the unreliability, but the impossibility their use for further historical research until the relevant classification schemes are confirmed or refuted at the next stage.”
Having this said, WP is even more brutal with others: “Despite their shared clicks, the Khoisan languages diverge significantly from each other. Traill demonstrated this linguistic diversity in the data presented in the below table.“. And then several words that are different in 5 languages follow.