As I said a few years ago, Tocharian is one of the Indo-European languages I’ve found most intriguing; now Ali Jones at Phys.org writes about a very promising project called TheTocharianTrek:
The research is helping to pin down where the Tocharians were located in the period between 3,500 BC, when they may have left their ancestral home, and their first written history in 400 AD. In sum, the initiative is mapping the migration route from the PIE homeland all the way to China.
Through the journey, the Tocharians brought their dialect of PIE into contact with people speaking different languages. This influenced and changed the way the Tocharians spoke until finally their recorded languages evolved. Archaeological and genetic evidence suggests that the Tocharians first moved to southern Siberia.
[Professor Michaël] Peyrot and his research colleagues have sought to provide a linguistic assessment of this route. Their work reveals that, indeed, some of the quirkiest features of the language fit very well with tongues spoken in southern Siberia.
“Languages preserve precious information about their prehistory through the effects of language contact,” said Peyrot. “Observing the effects of language contact, such as borrowed words, enables us to draw conclusions about the proximity of the speakers of different languages and at which point in time the contact took place.” As an example of a borrowed word, he cited a term for sword in a language strand known as Tocharian B: “kertte” was taken from “karta” in Old Iranian.
The research team has concluded that the Tocharians arrived in the Tarim Basin in around 1,000 BC—later than was previously thought. As result, their window of influence in the Tarim Basin has narrowed and the Tocharians are being assigned a more muted role in the prehistory of the area than they have traditionally been given.
Instead, the project has found a strengthened role for Iranian languages and peoples in the area, especially Khotanese, its relative Tumshuqese and Niya Prakrit. All influenced Tocharian.
The project is also piecing together which languages left the PIE community first and when. As their work enters its final phase, the researchers agree with the theory that the Tocharians may well have left the PIE family second and certainly well after the Anatolians, a group of ancient languages once spoken in present-day Turkey.
The piece goes on to discuss weather terminology and says:
The ultimate goal is to create an atlas that maps where the words were used and when. The completed atlas is due to be available on the university’s website beginning in late 2023.
Exciting stuff — thanks, Dmitry!
This study by Peyrot, from 2019, argues for contact between Proto-Tocharian and Proto-Uralic, on typological grounds. I would be curious to know what Uralists think about it.
The general gist of Uralic influence seems to hold up, I haven’t seen anyone outright question it, though many details remain debated.
Tocharian loanwords in Samoyedic remain probably underresearched; there’s been a decent number of recent proposals from TTT member Abel Warries and I’ve got a handful of unpublished finds myself around too. So far no finds of loanwords in the opposite direction though, interestingly enough! but maybe more lexicographic work will help with that.
Something getting from Tocharian even into Ugric (or into Hungarian, Mansi, Khanty separately) could be possible, has really not been checked systematically by anyone; there seems to be nothing already into Proto-Uralic though.
@Y: Thanks!
The uralicist hath spoken. I’ll just for the sake of discussion say that it’s interesting, but it’s based on internal reconstruction and should be backed up by other evidence. Another thought is that this looks like a Sprachbund with Yukaghir as a more peripheral member. One objection would be that it’s hard to imagine especially Yeniseian going through a Sprachbund phase without clear traces of syntactic borrowing or morphological leveling.
I vaguely recall some scholarship which suggests that Yeniseian was spoken considerably further to the south in prehistory. In that case, there would be no need for it to be immediately adjacent to pre-Tocharian and pre-Samoyedic during their time of contact.
No, between Pre-Tocharian and “an early form of Samoyedic”. (Open access! I read the paper a few years ago and recommend it.)
When the grammar is just too different, such things may not happen. Basque and the surrounding Romance languages have basically been spending the last 2000 years approaching their sound systems to each other’s. Grammar? Nope. OK, Basque has been emphasizing “and” (ta) over the comitative case in the last 3 or so centuries, but if that’s all…
I haven’t seen that outside of attempts to tie to, specifically, Burushaski and often also to the archaeological Karasuk culture.
David M.: When the grammar is just too different, such things may not happen. Basque and the surrounding Romance languages have basically been spending the last 2000 years approaching their sound systems to each other’s. Grammar? Nope.
Point. It also occured to me that if Yeniseian instead is a common substrate, carrying the sound system over would be expected, bringing morphology along well nigh impossible.
More on Afanasievo and Tocharians, starting with the comments:
Indo-European and the Yamnaya
They Perished Like Avars (a comment I meant to leave in another thread)
Son of Yamnaya (obviously)
A study from last year that I hadn’t seen:
Kumar et al: Bronze and Iron Age population movements underlie Xinjiang population history, Science (2022)
The paper isn’t easy to follow, or trawl for details on single elements, but it seems that the Afanasievo, the Okuneva (who were their neighbours up north in the Minusinsk Basin) and the Chemurcheck (who took up nomadic herding on the Western Mongolian Plateau after learning animal husbandry from them) may have formed an economic and political community, with the Afanasievo as an initially dominant minority element. This multi-culture could well have become linguistically Afanasievan (Indo-European) with an Okunevan (e.g. Yeniseian) substrate. They spread out in a wide region around the Altai searching for pastures and presumably copper ore. We find them in the Early Bronze Age in the fertile Ili Valley in Western Sinkiang. With the advent of the Iron Age about 1000 BCE they seem to have been replaced or subdued by the Saka, and I don’t where they went after that, but they leave written documents along the northern rim of the Tarim basin, just across the Tian-Shan range, from around 400 CE.
I learn that recent archaeology in the Ili valley divide the metal age into an Andronovo era (c. 1900-1000 BCE) and a Saka era (after c. 1000 BCE). That might weaken my assertion about continuity from the early Bronze Age, but I’ll note that there’s some debate (as always) over whether the Andronovo phase is echt Andronovo or a local development influenced by it. It may be more scathing that there’s no evidence of metal work older than c. 1900 BCE. But absence of evidence etc.
Zhi, Festa: Archaeological Research in the Ili Region: A Review, Asian Perspective (2020)
Wang et al: Copper metallurgy in prehistoric upper Ili Valley Xinjiang China, Archaeological and Anthropological Sciences (2019)
On the upside, it’s in open access, unlike the last few Science papers in this thread!
Remarkable amounts of Anatolian Farmer ancestry – up to 43%.
Yes, in some. I don’t have the paper in front of me, but I think those are believed to be Sintashta (or maybe Andronovo) Indo-Iranians. I gave up trying to unite all data for each individual.
Me: We find them in the Early Bronze Age in the fertile Ili Valley in Western Sinkiang. With the advent of the Iron Age about 1000 BCE they seem to have been replaced or subdued by the Saka, and I don’t where they went after that,
No, that’s not it. Except from this single skeleton which they say is unadmixed Afanasievo, the other finds that are archaeologically Afanasievo or Chemurcheck are concentrated up north. These cluster together in the PCA plot. I’m confused because the “pure” specimen is rather like Sintashta. But in the admixture bars they all look the same. I guess there are things going on that isn’t shown in the plots.
Looking at the PCA plots I’m leaning towards the old suggestion that the Tocharians are the Wusun. They show up at the eastern end of the Silk Road, together with the Yueshi, a couple of centuries BCE, when the Han start pitting the steppe peoples up against eachother. Maybe this is the old alliance of Afanasievo and Chemurcheck?
Anyhow, the Wusun and the Yueshi start fighting eachother, and then moving westwards in turns under pressure from the Xiongnu. The Yueshi first took the Ili Valley from the Saka. Then the Wusun took it from the Yueshi. The Yueshi then went on to Bactria and founded empires, while the Wusun stayed and eventually were allowed to form a sort of buffer state between the Xiongnu and Han China, sinking gradually into obscurity. This is exactly the period when the Tocharian languages are attested.
This Wikipedia page has a “Chemurchek culture and contemporary cultures and polities” map, useful for those of us who have a hard time keeping in mind which culture was where. (Can’t link to the map directly — all the labels disappear.)
Or maybe the Tocharians are both the Yueshi and the Wusun, if the fallout and breakup were messy, if either A or B is a Yueshi stay-behind group, or if either represents the local delegation of the Kushan empire.
That history of the Wusun and Yuezhi* peoples is just my summary of (mostly) Wikipedia. Ideally, this is where John Emerson would join in with all the Yuezhi details from Chinese, Mongolian, and Tibetan sources.
* Did I write ‘Yueshi’ up there? So I did. I devoice on your lousy Pinyin!
Aren’t the Yuezhi more likely the Actual Tocharians, the ones who wrote an Iranian language in the Unknown Kharoṣṭhī script?
…in… the… Son of Yamnaya thread.
The Kushan Script Deciphered this July. I was just pondering how to think about that.
Oh. Remarkable how I half-remembered that and mixed it up with the knife reported on LLog.
(Can’t link to the map directly — all the labels disappear.)
Yeah, wp seems to have a real problem with maps. The thumbnails on the page are too small to read. Magnifying in-situ is temperamental[**]. (I got it to work eventually.) Clicking gives you a full-screen map with no labels — useless.
[**] And randomly magnifies other wp pages you might have open.
Chrome, at least, and maybe Firefox, have the feature that the zoom level is shared between pages with the same host name. I’m not sure if “wp” here is Wikipedia or WordPress, but it would certainly be a thing for separate blogs hosted under wordpress.
com. (Wheras blogs with their own boughten domain name would not share zoom levels, even if actually hosted on wordpress.\code/>com).__________
(*) so called
Ha, my trick worked (writing
wordpress.<code/>comto avoid autolinking), but I misspelled it the second time and forgot to check inside the edit window….Oops, sorry, that depends on how considerable “considerable” is. The evidence from hydronyms is clear that Yeniseian drifted north in historical times; but the names don’t extend as far south as where the Karasuk culture once was, IIRC.
See now “Tocharian Bilingualism, Language Shift, and Language Death in the Old Turkic Context,” by Hakan Aydemir (Sino-Platonic Papers 337 [Dec. 2023], open access):
Thanks @Hat for linking that. I have questions … (about relying on soundalikes for random seemingly everyday words, rather than specialist vocabulary like horse equipment or thole pins). But also I have ‘form’ over there which makes me reluctant to even ask.
uTexas announcement (a year+ ago) of Dr. Aydemir’s research plan.
Publication in the Sino-Platonic Papers does not inspire confidence: it is the home of some really quite extraordinary nonsense, and edited by a man who has a very large and very evident axe to grind, and who lacks any comprehension of the basics of historical linguistics.
Still, the paper deserves to be looked at on its own merits. (An evident proto-Oti-Volta cognate did catch my eye, though …)
… the paper deserves to be looked at on its own merits.
Sadly, it seems the paper that deserves looking at is not this one, but another “ This period, together with other linguistic data, will be covered in another article. ” [p. 9 — all too ‘Deep Thought’] So those wanting posited cognates can inspect only the fewer than a dozen single-syllable roots pp 7 ~ 9.[**] (Which anyways date to very early in posited contact; you’d expect borrowings/influences from the different episodes of contact over the several centuries the paper claims.)
Those few pages were what raised my questions above. I assumed at that stage there’d be much more in an addendum or somewhere.
What I’d expect (but tell me I’m being dumb) is comparanda showing this form doesn’t fit the sound pattern of language X, so must be a borrowing from language Y, which it does fit. And/or this form is only in Turkic languages that were in close contact with Toch; other Turkic languages show a quite unrelated form. I don’t even get a strong feel for the direction of borrowing.
There’s a bunch of possible ethnonyms/toponyms in Turkic languages of the Tarim basin possibly showing Toch influence.
[**] Oh, those who hold their breaths until cont. p42 get
Really? The reflex of PIE *uks-en is TochB ‘okso’? That seems all _too_ neat.
I had the same immediate reactions as David E. and AntC, and this morning I just stopped reading after a few paragraphs, underwhelmed by the borrowings in Old Turkic. I’ll take it up again after Christmas, when days are long and slow and I have more patience.
I’ve downloaded the paper, but haven’t had time to even begin reading it.
Does it mention Iranian/Iranic? There must be a decent helping of words from that family that were borrowed into both Turkic and Tocharian.
haven’t had time to even begin reading it
You might do best to start with the Summary pp 74 ff, especially the posited timeline. (Apologies if I’m ‘teaching my grandmother …’.)
Does it mention Iranian/Iranic?
The ‘meat’ being “in a separate study” seems to be a recurring pattern. The guy’s a Turkicist not an Iranianist, so reasonable, I suppose.
That said, there are plenty of mentions in the footnotes of findings/speculations[**] from Iranianists.
What worries me methodologically is the risk of finding a borrowing into Turkic and presuming it’s from Toch (and taking that as evidence Toch was still spoken), without first eliminating the possibility it’s from Iranian.
[**] Indeed many of the mentions of Iranian are suggestions of mis-attributions: what had been thought as an Iranian borrowing into Turkic (or v.v. or into Toch) is more likely Toch into Turkic — note 121 p. 47, for example. Indeed on a closer reading, I tend to think the footnotes carry more ‘meat’ than the text.
a) Long after Toch lost prestige as a Buddhist liturgical language, it continued as a vernacular, gradually losing ‘grammaticalisation’.
b) Although TochA might be the older IE variant, it continued in parallel/in different areas vs TochB and indeed outlasted TochB. the Argu were undoubtedly the descendants of Yuezhi, and their language was TochA. [p. 79]
c) That’s why there was still in C10th enough knowledge of TochA [p. 23] to translate key Buddhist texts (Maitrisimit and Daśakarmapathāvadānamālā) into Uyghur — albeit with some misunderstandings of the grammar.[**]
[**] like modern English misunderstandings of Shakespeare or KJV?
But please explain: if you’re translating Buddhist texts why not start from the Sanskrit?
1. Nobody knows Sanskrit.
2. The Tocharian text is canonical or sacred in its own right.
Compare the Vulgate in the Cactholic tradition.
Compare the KJV in some Protestant domains.
I have read the paper. It’s not so much a research paper as a synthesis hypothesis or an outline of his ongoing work, so the Sino-Platonic Papers may be a good place for it.
Main points:
– The Yuezhi are the Toch A (Yarki). They ruled a state where the population consisted of both Toch A and Toch B groups.
– The historical Tocharians are the Toch B (Tukhar). They ruled the western part of the basin
– The mobile elites of both states were chased out by the Huns and fled first to the Ili Valley and then to Bactria in the second century CE.
– The societies rebound, united by Buddhist religion and a monastic economy,
– The languages took the first blow after the Tibetan conquest of 840 CE. The Tibetans were Buddhists too, so the religion and the monastic economy survived and suffered through the Uighur-Tibetan wars that followed.
– The next blow came with the Uighur conquest of 866 CE. The Tocharian elite may have become Manichaean and Uighur-speaking, but Buddhism and Tocharian spread from below, until the Uighur king converted.
– The Uighur state was bilingual. Uighur was used for official and external communication and Toch A and B in everyday situations and religious rituals.
– The Western basin was conquered by the Karakhanids in 1040, and islamization followed. The Toch B reorganized as a tribe and kept a separate identity within the tribal system for many generations. So did a tribe of Argu, likely Toch A from the Ili Valley. Both were bilingual.
– The late texts in bad Tocharian reflect a period when the language was no longer spoken by the writing class but still used in Buddhist rituals.
– Tocharian finally died out in the Karakhanid realm around 1200 CE, in the Uighur realm before 1300 CE. Paradoxically, Toch B seems to have survived longer than Toch A in the Uighur east, while Toch A survived longer than Toch B in the Karakhanid west.
The case relies entirely on reinterpretation of written sources, and he suggests that genetics, archaeology and thorough lexical work in the rural regions of Xinjiang and eastern Kyrgyzstan and Kazakhstan might clarify the matters.
Thanks very much for that summary! Interesting and plausible, at any rate.
Synchronicity strikes again. Here’s an article putting genetic sequencing for ethnic origin in grimmer light:
Amy Hawkins, The Guardian: Academic paper based on Uyghur genetic data retracted over ethical concerns.
The article mentions lack of informed consent by the participants, but the ethical problems seem much deeper than that.
This isn’t exactly about ancient DNA, but it’s about using genetic anthropology in criminal investigations — and potentially about using genetic screening to isolate ethnic groups — and that makes me very uneasy. Maybe Dmitry can throw some light on the issue.
@Trond The case relies entirely on reinterpretation of written sources, and he suggests that genetics, archaeology and thorough lexical work in the rural regions of Xinjiang and eastern Kyrgyzstan and Kazakhstan might clarify the matters.
Yes, thank you for your ‘due diligence’.
It was not clear whether the “thorough lexical work” is already further advanced than the too few examples in S-PP. Are we about to see another paper in a more ‘technical’ forum?
I appreciate many lines of research would like genetic data wrt Uyghur ethnology — especially before PRC wipe them out completely — but I’m appalled at Elsevier and OUP (alleged). “ help the police identify suspects in cases” my foot! As far as Chinese State police are concerned, if you’re Uyghur, you’re guilty.
I just found this open-access follow-up paper by the abovementioned Abel Warries to Peyrot’s abovementioned 2019 paper on the suspiciously similar vowel systems of Pre-Proto-Tocharian and Pre-Proto-Samoyedic.
I just finished reading it, recommend it, and note it cites two blog posts by Juho Pystynen for support.
Warries’ paper does a great deal to sort out the remaining questions in Peyrot, ending up with near-identical phoneme inventories in the two languages and a seemingly plausible timeline for both. That’s neat, but identical phoneme inventories don’t just happen even in close contact. I’d like to see phonological development in one language explained with the phonology (including phonotactics) of the other.
The suggested loanword Toch B yasa “gold” < PT *wʸəsa < pre-PT *wesa < pre-P-Sam *wäsa “metal” < PU *vaćka is important if it holds out. It would be informative for the timeline of changes in both languages and for the environment where they interacted. But it’s just one word.
Apart from the phonotactics, that happens near the end of the paper – the IE vowel system getting reinterpreted in Samoyedic terms.
As the paper mentions, it’s probably not actually PU, but a Wanderwort circulating between freshly separated Uralic branches – and ought to have spread during everyone’s favorite Seima-Turbino Phenomenon.
By the way, Toch B yasa < PT *wʸəsa betrays initial stress. PT *a gives B stressed ā /a/, unstressed a /ə/; PT *ə gives B stressed a /ə/, unstressed ä /ɨ/. It looks like the word had initial stress, which fails to contradict the hypothesis that it’s a Uralic loan.
(Tocharian stress, although never directly written, is understood pretty well now. I’ll try to look for the paper(s).)
David M.Apart from the phonotactics, that happens near the end of the paper – the IE vowel system getting reinterpreted in Samoyedic terms.
Yes, to a limited degree. What I meant is that a great deal of development is assumed to have happened independently before the languages entered the period of contact with already very similar — and rich — vowel systems. This means that contact isn’t used to explain the changes, only to time them. If instead, say, phonemization of reduced vowels could not be well explained language-internally, but would be a simple reinterpretation of allophones in one language based on distinctions made in the other, the case would be stronger. If that’s the case here, I can’t read it from the paper.
PU *[w]aćka
If PU *vaćka was a borrowing in (post-)PU, it would still have been borrowed into Tocharian from Samoyedic.
Hungarian vas << *waś[k]a would seemingly come about as close, but that’s probably already too far west.
I imagine Warries may have missed by a small margin my also fall 2022 presentation about vowel reduction in Samoyedic alas, which has a few additional observations.
The paper is quite long-winded, detailed and comprehensive and explores alternatives in full. But Table 11 presents the conclusions quite clearly. Most striking is the reinterpretation of (*iH *uH >) *ī *ū as *ij *uw, sequences that are imaginable in Uralic (and must have been present in a pre-Samoyedic stage in the right timeframe) but not in IE (unless much farther-reaching changes to the sound system have already happened) where they would have been **/jː wː/ in a system that didn’t allow long consonants. (The dative singular of the PIE *i-stems, *-ej-j, did not come out as **-eji but as *-ēj, no different from the animate *s-stem nom. sg. *-os-s as *-ōs or the *m-stem acc. sg. *-om-m as *-ōm.) The outcome of *o, even though that may not have been a rounded vowel in Proto-Indo-Tocharian, as something that was able to merge with *ē as Proto-Tocharian *e, without taking *a or *ɨ or anything else along for the ride, may be even harder to make sense of without considering it an adaptation to the Proto-Uralic through Proto-Samoyedic *ɤ (Uralicist tradition: e̮; “Uralic Typewriter Alphabet” and tocharologist tradition: ë).
I also wonder if the split of PU *ɤ into PS *ɤ and *ɯ was motivated the other way around, as an adaptation to the Pre-Tocharian contrast of *o and *ɨ…
I must have had this in mind. (Long and in not quite native German.)
At that link I discover that scaricabile means downloadable. I haven’t encountered a Spanish adjective for that – the locution I know is se puede descargar.
descargar also mean unload. Things gotten from the internet might have fallen off the back of a truck, for all some people care.
“downloadable. I haven’t encountered a Spanish adjective for thatl”
descargable (recognized by the Royal Spanish Academy, widely used, and probably universal in the Spanish-speaking world).
I believe you. It remains a fact that I have not encountered the word.
David M.: Table 11 presents the conclusions quite clearly. Most striking is the reinterpretation of (*iH *uH >) *ī *ū as *ij *uw, sequences that are imaginable in Uralic (and must have been present in a pre-Samoyedic stage in the right timeframe) but not in IE (unless much farther-reaching changes to the sound system have already happened) […]
Yes, I follow the logic of the sound changes, and I probably sounded more critical than I meant to be. Maybe what I miss is a clearer comparison with Peyrot’s timeline, not as sequences of sound changes but as independent versus contact induces changes. That could even allow a discussion of (the development of) the substrate influence and what that might mean for the Samoyed-Tocharian contact situation.
This is relevent because I’m not convinced that the contact process involved language shift from Uralic, i.e. an actual Samoyedic substrate in Tocharian. Other scenarios that might be entertained are a sprachbund with Samoyedic as lingua franca and a Samoyedic superstrate/adstrate with sociolinguistic prestige. That doesn’t matter for the purely linguistic evidence, but it might for timing and for comparison with archaeology.
To be a little more specific: Afanasievo seems to disappear from the Minusinsk Basin around 4500 BP, being replaced by the Okunevo culture. This culture is very different from Afanasievo, instead continuing the customs of the local foragers, but it still seems to have some 10-20% specifically male ancestry from Afanasievo. This does look like a situation where Afanasievo could prevail on an unrelated substrate, but the Trans-Baikal ancestry that will later spread with Seima-Turbino and become diagnostic of Uralic makes its first appearance west of Lake Baikal — just in this region — about 4200 BP.
Calm summer days. I’ll just pretend there are readers and keep going.
I see three reasonably simple substrate-forming events. I could come up with infinitely many more, but they would be more convoluted and hence less likely until the simple scenarios are ruled out.
(1) The Afanasievo settlement in the eastern forest steppe around 5500 BP. For the hypothetical substrate to be pre-Proto-Samoyedic, Uralic must have started differentiating some time before that, say 6000 BP. Untocharianized Samoyedic must have been spoken close by, and the remaining “Core Uralic” not too far away, since I don’t think there’s evidence of other long-distance migrations in this region in this era. When the Trans-Baikalians arrived more than a millennium later, they seima-turbinized Uralic and uralicized themselves before spreading quickly westwards along the forest steppe.
(2) The replacement of Afanasievo with Okunevo around 4500 BP as sketched in my previous comment. In this case, Samoyedic might have kept developing in the closely related cultures around the Irtysh that didn’t become Okunevo but eventually became affiliated with Seima-Turbino. Otherwise, (1) and (2) are pretty similar and not mutually exclusive, except for the nature of substrate (1) and the implied age of Proto-Uralic.
(3) Tocharianization of (a part of) the Seima-Turbino-affiliated culture in the Altai foothills. This must have happened some time after the beginning of Seima-Turbino, so perhaps 3500 BP. In this case Proto-Uralic could be defined by the starting point of Seima-Turbino. Otherwise, again, this is not mutually exclusive with (1) and (2), with the same qualification.
There are weaknesses with all three of them.
As a heavy metal fan I am happy to learn about Seima-Turbino, pretending to have readers is unnecessary effort; thanks, sincerely
I am reading with interest but have nothing to say as of the current moment. And you know what Wittgenstein said.
pretending to have readers is unnecessary effort
Readers read, writers write. Only money talks in the text business.
Yeah, I’m just being cute.
Some scenarios for adstrates/superstrates:
(A1) Situation (2) above, except that Okunevo didn’t become tocharophone. The families with Tocharian fathers kept in contact with the remaining Tocharians, now Chemurchek steppe nomads, whose speech slowly adjusted to the not-longer-primary-language of their settled relatives.
(A2) Situation (3) above, except that there was no language shift to Tocharian. The peoples around the Altai used the language of the metal workers as lingua franca. Speaking with a Seimo-Turbino accent after spending time in Seima-Turbino settlements or taking part in Seima-Turbino expeditions was attractive,
(A3) The emerging Chemurchek culture was a multi-lingual alliance of nomadic peoples that used pre-Proto-Samoyedic as common language until eventually the covert prestige language Tocharian prevailed.
The list is not exhaustive.
What does this mean for the timelines?
If, as argued in 2023, (a part of) Chemurchek was the vehicle that brought Tocharian to the Dzungarian Basin, then a Samoyedic substrate must have been acquired before 4500 BP.
The intitial phases of the development of Uralic would have taken place in the Irtysh-Yenisei region before the Trans-Baikalian prestige group arrived and Seima-Turbino got going.
An adstrate or superstrate in Tocharian could be introduced later, by more mechanisms, but even this window isn’t wide open.
If the onset of Seima-Turbino is the onset of contact between Uralic and Pre-Indo-Iranian, it’s too late for Proto-Uralic, because Pre-II contact appears to have happened entirely with separate branches of Uralic.
Yes, that too. Based on current dates, Seima-Turbino spread from east to west ~4200-4000 BP, exactly contemporary with Sintashta and Andronovo from west to east. If that holds out, it’s hardly a coincidence. Whatever way that happened, I assume that contact with neighbouring Indo-Iranians was established separately for each Seima-Turbino group, and that Uralic started to disintegrate immediately while Indo-Iranian held together for longer.
Language Log has a post about “Yuezhi archeology without concern for Tocharian language” by none other than Victor Mair (https://languagelog.ldc.upenn.edu/nll/?p=65252). I found it long and very hard to follow. Mair’s point appears to be that the Chinese are again trying to rewrite history from a Sinocentric perspective, although I had great difficulty following the chronology and movements. Our own DM wrote a comment that Mair immediately slapped down.
A second post has since appeared: “Rethinking the Yuezhi?” (https://languagelog.ldc.upenn.edu/nll/?p=65255), a guest post by Craig Benjamin.
Our own DM wrote a comment…
A clear, informed, and interesting one. I would be happy to get this kind of comment on a peer-reviewed manuscript.
…that Mair immediately slapped down.
Pissily and rudely, for no reason.
I’m reminded once again why I got tired with Language Log.
When it comes to agendas, I fear it’s more what they choose not to look for and may inadvertently destroy in the process. By sheer coincidence (named Dmitry) I had occasion today to read about the Henan Longshan culture in China. This was a third millennium culture on the middle Yellow River with massive interaction with nomadic peoples — very much including the Chemurchek — during what arguably are the formative years of Chinese civilization. I found reference to one genetic study of this culture, and even if it’s as recent as 2022, it’s of mt-DNA alone, which (as expected) is overwhelmingly local. An isotope analysis from this year OTOH tells of long distance mobility especially among women.
As I noted, Mair appears very concerned that China will use this to rewrite history. For instance, he quotes this from the original Wall Street Journal Article (China Reaches Back in Time to Challenge the West. Way, Way Back): “Asked whether Beijing could use the Yuezhi to make territorial claims, Wang said the notion was absurd because the nomads are a historical people and no one serious would put forth that argument”.
The Wall Street Journal article and Mair’s comments both frame the arrival of Chinese archaeologists in Uzbekistan in terms of a challenge to the West and the rewriting of history, including the dark intimation that his discoveries might lead to new Chinese territorial claims. While that doesn’t seem relevant in this case, Chinese attempts at rewriting history should never be taken lightly. The Chinese claim to the South China Sea is a wholly manufactured one based solely on interpretations of history by Chinese nationalist historians in the ROC period. The recent successful Nantes museum exhibition about Genghis Khan should also give pause for thought. When the Nantes museum attempted to source material from museums in Inner Mongolia, the Chinese side insisted on imposing their interpretation of history on the exhibition. This wasn’t made very clear at the time, but it later became apparent that the Chinese wanted to retitle the exhibition “China’s Steppe Empire”. The narrative that Genghis Khan was “Chinese” (as part of the Zhonghua Minzu) is alive and well in China — even though the Communists’ favourite author, Lu Xun, shot it down almost a century ago.
one of their burial forms, defleshing of the dead
I assume this is what is known with regard to the Tibetans as “sky burials”.
@ Trond
massive interaction with nomadic peoples
There has always been massive interaction with nomadic peoples. The problem is the spin that is put on this. The Chinese narrative is that these nomadic peoples have been successively and successfully absorbed into mainstream Chinese culture. This gives rise to the concept of 民族融合 mínzú rónghé, or fusion of peoples, which takes the general stance that all backward surrounding peoples are destined to be absorbed into China’s superior agrarian, urban civilisation. It is this mentality that Mair is resolutely opposed to.
Originally the absorption of surrounding peoples was thought to be a long-term, natural process that would inevitably unfold. The new view that has been adopted by historians who have influenced Xi Jinping is that this process could and should be accelerated by government policy.
might lead to new Chinese territorial claims …
The Belt & Road Initiative is all about pouring money into impoverished countries, to lock them into Chinese hegemony. With the ‘-stans’ the CCP’s first step is to unlock them from Russian influence — not difficult with Putin distracted elsewhere; and avoid them falling under Islamic influence. Then where else will they have to go? Being landlocked and all … The CCP plays a long game.
Poor dupe. Has he never met Xi Jinping thought?
Yes, when I opened the paper about Shimao one industry, I imagined that it was their exports ware to the pastoralist North… but what surprised to find out how peripheral and unusual was this town with respect to the rest of the Late Neolithic Loess plateau, and how transitional it appeared to be the Steppe. But I couldn’t figure out if there were archeological parallels
Back in the day, countries would cleave to their patrons through instilled ideology (communism, capitalism), on top of the material benefits. Does the Belt & Road Initiative provide any ideology with which to attach people to China?
@Y I would be happy to get this kind of comment on a peer-reviewed manuscript.
Indeed. The site gets little enough informed commentary. (I include my own commentary as mostly ill-informed. Seeking elucidation; rarely receiving much.)
any ideology with which to attach people to China?
Most of the ‘attachment’ derives from bribing heads of government in corrupt (pseudo-)democracies. Pacific Island nations “switching allegiance from Taiwan to China”. “China to attempt to meddle in a national election later this year”
@Dmitry: I haven’t actually read the Shimao paper yet. I had a brief look, decided I needed a better grip of the context, started reading about the cultures of the loess plateau, and realized the relevance for the Afanasievans.
The transitionality is indeed interesting. The Longshan had pigs, chicken and millet and got cattle, sheep&goats, wheat and barley from the west. In about 2000 BCE they retracted to the lowland and started naming dynasties.
@Bathrobe: Fair point. But who are these “historians” giving advice? It sounds more like thinktankery.
I can’t find anything that will introduce you to the public intellectuals who have been pushing for “second-generation ethnic policies” in China, but I have seen articles mentioning the people involved. One that sticks in memory is a gentleman called Ma Rong. Ma is of the Hui nationality but is critical of earlier ethnic policies (see this article: https://www.readingthechinadream.com/ma-rong-ethnic-regional-autonomy.html). He precedes Xi Jinping.
There is quite a bit around the Internet about the “second-generation ethnic policies” of China, which can be quickly found through a Google search.
I think he saw I was criticizing a few selected parts of his post, thought I was criticizing the post as a whole or even him as a whole because apparently normal people think in such ways, correctly saw I wasn’t offering any critique of the whole thing (let alone of him as a person), and got pissy for that reason.
I don’t begin my peer reviews with the exact words “yes, but”, I make the introduction longer; still, how common is peer review in sinology? Are the SPP peer-reviewed…?
I’m not surprised by the second sentence; but the first? They were so isolated in their bubble that they thought this would fly in Europe?
Are the SPP peer-reviewed…?
Not by Mair. (Indeed in some cases I wonder if he’s even read them.) Clearly some authors get their Ps reviewed before publishing at SPP. Others not so much.
This is sad for the diligent authors: SPP is of such variable quality, it’s too easy to dismiss anything published there.
I only recently (around June) learned that the Chinese wanted to retitle the exhibition “China’s Steppe Empire”, but now I’m unable to find the source. Google doesn’t help. I’m pretty sure I haven’t misremembered it. It’s a pretty breathtaking emendation but I wouldn’t put it past Xi Jinping, who seems to me to the ultimate Han chauvinist. I think that’s one reason Mair loves to pick up evidence of Xi Jinping’s illiteracy in Chinese. (After all, he was banished to the loess plateau in his youth so his education was sadly lacking, although those dusty days don’t seem to have diminished his enthusiasm for the glories of Han Chinese civilisation.)
I started living in China in 1993 and, hearing the term Zhonghua Minzu, assumed that it has always been part of China’s official vocabulary. I was surprised to discover, during my searches, that the term was actually coined by Liang Qichao in 1902 and much beloved of the KMT but was expunged by the Communists, especially Mao, who rejected Han chauvinism. It was only revived by the ethnologist Fei Xiaotong in 1988 and has now become central to Xi’s thought and policies.
(See also this paper: https://www.fiia.fi/wp-content/uploads/2019/04/bp260_sinification_of_china.pdf)
And then it got worse… AntC asked a few direct, incredulous questions, and instead of answering any of them, Mair helpfully informed the world that this was “ostentatiously otiose”. He seems to believe he’s surrounded by trolls!
That’s obviously how he feels. He’s an extreme example of professorial poisoning — he’s so used to students drinking in his every word and never raising objections that he expects everyone to defer to him similarly. More to be pitied than censured, perhaps, but I censure him anyway.
He did it again (on the same earlier post), reacting to DM’s interesting and clear excursion.
I’ve had professors like that. They would ruin any sort of academic attraction to their classes by insulting students whose intonation they didn’t like when they asked questions. Who wants to see a 19 y.o. be humiliated in public?
He once simply deleted a comment of mine that he didn’t like, on account, I presume, of my having had the gall to point out (rightly) that the source he was using was very unreliable. (I blame myself for trying. I should really know better by now.)
The comment was then given its own post by Mark Liberman, and later undeleted in situ.
He seems completely unable to admit to either error or ignorance. It would be easier to take if he were less prone to error and ignorance himself. His atrocious manners are amusing, given his constant baseless attribution of rudeness to others.
Or it’s a cultural difference, due perhaps to Mair having lived in “guess cultures” for decades.
Are you suggesting that he never learned how to guess accurately, after all those decades ?
DM AntC asked a few direct, incredulous questions,
Thank you David for answering my incredulity. In this case Mair is not in error or ignorance; I am. Never the less a straight answer giving the evidence would be appropriate (I’d already said the evidence is paywalled — if that’s where it is); not abusing my persistence.
surrounded by trolls
He often posts on tidbits I send him — usually wrt Taiwan. So he knows I’m not a troll(?) Or perhaps he thinks I suffer from split personality(?)
——–
And BTW, if the PIE for wasp/bee is unknown, how do we know the Tocharians had a word, and where did it come from?
I think he is just getting old. Personally I think his heart is in the right place but unfortunately his crotchetiness is increasingly detracting from his credibility. We need his scholarship and sharp insights on China, just not delivered in such a cranky (in the sense of petulant and annoyed) way. He is just alienating fans and providing grist for the mill of those (particularly the Chinese government) who would like to sideline him entirely. It’s easy to do when he acts like this. FWIW, I’ve always found him easy to deal with. What I find annoying about LL is the stodgy Americanness of it all, including the commentators. There seems (to my sensibility) to be a stolid wall of conventional and boring academic thinking and earnest Americanness in the contributions and comments. It’s almost like an industrial construction site. More perspectives from non-American commenters would be nice. And that’s precisely what commenters from here provide.
(This purely my personal feeling. I’m sure many Hatters would disagree.)
the stodgy Americanness of it all, including the commentators.
emmm? the far-and-away stodgiest commentator is a Brit, as they frequently remind us. As to their academicness, I’m perplexed: an unevenly-learned amateur?
The Brit guy is LL’s own Colonel Blimp. He is so eccentrically removed from stodgy Americanness that he is a joke.
(I didn’t realise the creator of Colonel Blimp was a Kiwi.)
WRT to the “bee” issue on that LL thread:
Hatters will no doubt be pleased and relieved to hear that the word for “bee” is reconstructable in proto-Oti-Volta. That can scarcely have anything like the time-depth of PIE, though it does have some cognates farther afield in “Central Gur”, which is pretty much a baggy-monster group and probably is of near PIE-level time depth. It’s not related to anything in Bantu, though.
Difficult to say whether “bee” or “honey” is prior in “Gur”: proto-Oti-Volta *ʃɪ̀m-fʊ̀ “bee” is evidently derived from *ʃɪ̀-tʊ “honey”, but several non-Oti-Volta Gur languages just call honey “bee grease”, so who knows?
Hatters will no doubt be pleased and relieved to hear that the word for “bee” is reconstructable in proto-Oti-Volta.
I for one had been losing sleep over it, so I thank you.
… and just as we share frustrations over Victor Mair, he posts this review: Magisterial German translation of a neglected monument of ancient Chinese literature, Mu Tianzi Zhuan.
I forgot to add the link to the open access pdf, found at a link provided by commenter Andreas Johansson.
Why “wasp”? That’s a separate issue, and unrelated to honey.
Anyway, here in the sidebar there is “A Dictionary of Tocharian B (with etymologies)”. “Bee” is in it:
Then follows a whole paragraph of etymological proposals (one of them from a PIE word that would have meant “yellowish ~ golden ~ brownish” and could also be ancestral to honey), followed again by “uncertain.”
There is no proposal that Tocharian and Chinese might have exchanged a word for “bee”; it’s just “honey”.
Thank you @DM, ok I’m being dumb.
Why “wasp”?
Because Chinese ‘Feng’ 蜂 covers both wasps and bees. Wasps are yellow-feng 黃蜂, bees are honey-feng 蜜蜂. If it took the Tocharians to introduce them to honey, what was the term for bees before that? And if the Chinese already knew about bees, surely they knew about honey? (I’m still incredulous — given that Chinese eat everything and anything. Surely they noticed bears helping themselves at hives?)
it’s just “honey”.
Just one word? (and a very short word at that) Then why is Mair being all supercilious? I’d expect a whole vocabulary to do with beekeeping, wax, comb/cell, equipment to extract and strain the honey …
Or … Evidence that Old Chinese already had a word for honey and ‘mi’ supplanted it. IOW to rule out the possibility that the Chinese discovered honey for themselves; invented a word for it; that coincidentally sounded like the Tocharian.
Here’s my take on the Tocharians.
(Sorry for all the names of cultures. Sorry I can’t link them all. And sorry that even with all these names, I’ve emitted a lot.)
It seems clear that Afanasievo had long distance trade with both BMAC and the Neolithic Yangshao culture (c. 5000-3000 BCE) of the Yellow River. When Minusinsk Basin Afanasievo dissolved, (pre-Proto-)Tocharians kept their pastures in the Mongolian Plateau and the Dzungarian Basin as the Chemurchek culture (c. 2750-1900 BCE). Afanasievo and Chemurchek may have been separate entities to begin with, but they eventually merged. Chemurchek kept regular contacts with the Okunevo in the Minusinsk Basin, BMAC, and the successors of Yangshao, the Tibeto-Burman farmers of the Majiayao culture (c. 3300-2000 BCE) of the Upper Yellow River and the Upper Longshan (c. 3000-2000 BCE) on the Loess Plateau. The proto-cities of Shimao and Taosi belong to the later phase of the Longshan.
The Chemurchek mediated Bronze Age innovations, helping to give the Shaanxi region the military and technological edge that would turn it into a center for unification of the Chinese heartland, starting already in the early second millennium BCE.
Maybe pressed eastwards by the Andronovo, many of the Chemurchek settled in the Hexi(/Gansu) Corridor and formed the Xichengyi culture (c. 2000-1600 BCE), probably with a strong local substrate of old trading partners from the Majiayao. The Xichengyi interacted closely with Upper/Middle Yellow River cultures like the Qijia culture (c. 2200-1600) of the eastern Gansu and the Erlitou culture (identified with the semi-mythological first Chinese dynasties),
The Chemurcheks on the Mongolian Plain may have established contact with the newly formed Seima-Turbino groups in the northern Altai foothills, learned wagon technology from the Andronovo, and transformed into the Munkhkhairkhan culture (c. 1800-1600 BCE) of the Western Mongolian Plateau. As long distance traders east of the Andronovo, they might have found a niche trading directly with the emerging Chinese state, but eventually they were forced out or assimilated by the Saka, who kept ruling the Western Mongolian Plateau until the formation of the Xiongnu Empire in the 3rd C. BCE.
The settled/semi-nomadic Xichengyi became the Siba culture (c. 1600-1300 BCE) who gave room to two independent but related cultures in the Hexi corridor, Shanma (c. 900-200 BCE) in west, and Shajing (c. 800-100 BCE) in east (I’m not sure what to think about the hiatus in the archaeological dates, but it ends about the time of the Mu Tianzi Shuan). Both cultures were heavily influenced by the Saka, who now dominated the eastern steppe. The location and the end dates fit well with the Xiongnu and Han conquests of the Hexi Corridor and the expulsion of the Yuezhi and the Wusun. The Yuezhi were the first to leave, so they may have been the Shanma. I think both were Tocharian, but I have no clear idea which was A and which was B.
(An alternative story is that the “Saka” of the Western Plain were the Yuezhi, and that one or both of Shanma and Shajing were the Wusun, but I think this is less likely.)
(The formation of the Chemurchek is obscure. It seems to have involved some very different populations.)
Much of this will no doubt be solved — and complicated — by genetics. As for now, the genetic data from China are to sparse to tell.
@AntC: Both trade and imported prestige culture may lead to replacement of good native words.
There’s nothing impressive about a two-phoneme word like mi, but it’s part of a set of suspected IE loans related to Bronze Age prestige culture. It’s reconstructed as Old Chinese *mit, very similar to attested Toch A mit. Old Chinese was probably the language of the first Chinese dynastic states, i.e. of the Shanxi and Shaanxi in the early second C. BCE, just when the Tocharians were mediating prestige goods and culture from what would become the Silk Road.
@Trond It’s reconstructed as Old Chinese *mit,
Yes, I realise there was a coda, so we’ve more than two phonemes. Still hardly impressive for a soundalike. What’s “prestige” about honey? I guess the Chinese had sugar cane for sweetening: “S. sinense was a primary cultigen of the Austronesian peoples. … at least 5,500 BP”
Chris Button chipped in on the LLog thread, so that’s more substantive; nevertheless, this alleged ‘gold standard’ etymon is from the same Prof Mair who thinks five-phoneme ‘kumar(a)’ soundalike is absolute evidence in the absence of any other vocab nor any anthropological artefacts whatsoever. (I’m not denying there was contact and cultural exchange between Tocharians and OC; merely I’m asking why this soundalike is of especial mention.)
… replacement of good native words.
Is there evidence of a native word that got replaced? Is there evidence or absence of evidence of Chinese native beekeeping/honey-gathering before Tocharian contact? Or are people just making stuff up, like the miraculous voyage that brought kumara to Pacifica?
Ah, the old strategy of tying your current argument to one you previously lost.
I always tie my arguments to each other. It’s called consistency. Winning and losing are for people who seek instant gratification. I’m in it for the long haul.
Me: The Chemurchek mediated Bronze Age innovations, helping to give the Shaanxi region the military and technological edge that would turn it into a center for unification of the Chinese heartland, starting already in the early second millennium BCE.
Maybe pressed eastwards by the Andronovo, many of the Chemurchek settled in the Hexi(/Gansu) Corridor and formed the Xichengyi culture (c. 2000-1600 BCE)
There’s another possible — or an additional — story buried between those paragraphs. The Chemurchek people settled in the whole Gansu/Shaanxi region around 2000 CE, maybe as specialist soldiers with bronze weapons, maybe also bringing in bronze workers from the Altai or the Minusinsk Basin. In Shaanxi they were assimilated to what would be the ruling elite of the Chinese state. In the Xichengyi culture of upper Gansu they would form stable societies for two millennia. In the region in-between, the Qijia culture would become the Rong peoples that the first Chinese dynasties were at constant war with.
@AntC: What “gold standard”? As I said, the honey word isn’t impressive in itself, but it’s part of a semantic field with several suggested Tocharian loans. But I think it’s fair to say that most are tentative, not at least because the reconstruction of Old Chinese is so shaky.
Oh, here’s an interesting paper!
Rasmus G. Bjørn: Indo-European loanwords and exchange in Bronze Age Central and East Asia
Six new perspectives on prehistoric exchange in the Eastern Steppe Zone, CUP 2022.
The maps look like they might shake the foundations of my story above. I’ll have to read it!
What “gold standard”?
(I’d better start by apologising to our host for leading ‘The Tocharian trek’ even further off course.)
I’m quoting Prof Mair; in which ‘honey’/’mi’ is given as the first/most secure piece of evidence.
Thank you very much for that Bjørn 2022 paper. I see a lot of handwaving: OC *m(j)it is so close to some early IE language, one of them must be the source, though we can’t nail it down, and all of the known ones require some slightly odd sound changes.
As I understand it, that date is far too early to be through Tocharian contact. Since (contra Ryan) I am eager to be lead by evidence, we can compromise that the word is maybe cognate with some IE-or-the-other, but probably not Tocharian and certainly not reaching ‘gold standard’ as Tocharian. The date would line up with Zhou Dynasty:
OTOH there’s maybe reason to doubt that paper[**]: it’s the same author/around the same date/same material (but more clearly organised IMO) as what I already cited at LLog. It illustrates [photo 3, which is some but not all of ‘cute’ photo 5 in that paper] “a series of characters suggested to be “feng” (note the stinger)” from oracle scripts, already pooh-poohed by Christ Button on the LLog thread, on grounds those characters are for locust. (But this version of the paper clearly acknowledges those other illustrations are locusts.) Counter-Chris note the illustrated characters clearly have something protruding from their rear parts, which the locust illustration Photo 2 does not. Locusts don’t sting.
If those oracle characters represent wasps/bees at 3,000 BP, we still have a huge gap not mentioning honey until Zhou dynasty. Too bucolic to merit recording in Imperial annals?
[**] I won’t test your patience with ‘Apiculture in China’ 2015 but note the specific date, note the Prof’s alleged name!
There is a big difference between gathering wild honey and systematic beekeeping. Evidence of familiarity with bees and with honey (which I would expect to be practically universal) is compatible with either; the transition to beekeeping, on the other hand, would be a great moment to borrow associated vocab. We might even expect to find cases where the old word for honey lingers on in reference to special wild honey, while the new one becomes the default for honey as a trade good.
Me: I must read the paper!
Done. I’ve even thought about it for a couple of hours while chopping wood in my backyard.
The paper is a list of six supposed wanderworts and an attempt to explain how each of them got from Indo-European and into different language families. Since each word got there in its own way, with its own gaps of unknowns to bridge, everything is essentially special pleading. It does substantiate the claim of wanderwort status, but adds little to the understanding of actual cultural transmission beyond what was already known from archaeology. That’s unfortunately how loanword studies must be, until patterns start to emerge from the special pleadings.
What I really would want to see is a treatment of the lexicon of Chinese/Sinitic/Sino-Tibetan metallurgy and animal husbandry. Chinese bronze metallurgy started around 2000 BCE. It’s widely accepted that it was transmitted from Inner Eurasia to the Upper Yellow River. Sheep and cattle breeding started in the Gansu-Qinghai region around 2000 BCE.
This means that archaeology alone tells about a major cultural change in the region of the middle/upper Yellow River. Archaeology also tells about cultural continuity with the preceding cultures of the Yellow River Neolithic. which should make us assume that it wasn’t all about massive population replacement, but about arriving groups settling among the native population (by any mechanism on a scale from slave import to foreign conquest) and spurring synthesis cultures and cultural exchange. It’s highly likely that the newcomers brought the technical vocabulary with them and transmitted much of it to the native populations.
So what languages would Central Eurasian cattle breeding metallurgists of 2000 BCE speak? Very likely Indo-European, but it’s just a tad too early for Indo-Iranian. The Afanasievo are known to be Indo-European by culture and genes. They lived as settlers in the Minusinsk Basin and as sheep and cattle breeders in the mountains and steppes to the south. They are known to be founding members of the bronze-producing Okunevo culture of the Minusinsk Basin and of the nomadic Chemurchek culture of the Dzungarian Basin and the Western Mongolian Plateau. They are also known to have disappeared from the eastern steppe when the Iranians expanded. Thus an Afanasievo hypothesis for a Chinese adstrate. That doesn’t make it Tocharian, but we also know that two named peoples (one or both) likely to be the immediate ancestors of the Tocharians show up in Chinese written history in the Hexi Corridor in the last millennium BCE, and that these can be identified with archaeological cultures, with reasonable cultural continuity inbetween.
Thus, there’s all reason to expect an early 2nd millennium BCE adstrate in Chinese, and there’s pretty good reason to expect that it was some sort of archaic Indo-European. There’s also good reason to expect later adstrates, maybe especially horse and iron vocabulary from Iranian — but maybe also from Turkic.
@AntC: If the “honey” word was part of one of the probable Indo-European adstrates is not of great importance, but it would be bad historical linguistics not to take a plausible Indo-European source into consideration.
Another take-out of the Bjørn paper is that it’s apparently mainstream to suggest that Okunevo was Proto-Uralic. That was a message I hadn’t received, but it’s compatible with substrate hypotheses (1) and (2) and the corresponding adstrate hypothesis (A1) as well as (A3).
Me: Me: I must read the paper!
Misquoting myself. Seems trustworthy.
Me (further up). I haven’t actually read the Shimao paper yet. I had a brief look, decided I needed a better grip of the context, started reading about the cultures of the loess plateau, and realized the relevance for the Afanasievans..
That’s what I’ve been doing until now. So how does “the Shimao paper” fit in?
Hua Wang et al: Craft Specialization in the Highland Longshan Society: Perspective from the Bone Needle Workshop on the Central Mound at the Shimao Site, Shaanxi, China, Journal of Anthropological Archaeology 2024
Shimao was a fortified proto-urban center at the northern edge of the loess plateau, bordering on the Mongolian Plateau. It was built around 2200 BCE and reached its peak around 2000-1900 BCE. It was a ritual center with an impressive temple mound and large scale human sacrifice. The top of the temple mound was also a specialized production center where bones (mostly from sheep) were turned into needles for garment production. The economy of the production is not understood. It was obviously under central command by the religious authorities, but there’s little that says that this actually was a commercial enterprise serving a larger region with needles and bone tools. The city must also have been a center for production of leather and leather garments, but no tanneries and sewing workshops have been found yet.
That doesn’t mean it didn’t happen. With the formation of the first Chinese states (semi-mythical dynasties), needle production was centralized in large workshops under royal authority. This is a direct continuation of an industry and a mode of organization that started at Shimao, at a main interaction point between nomads and settled farmers.
What does this have to tell about the Tocharians? It tells that the technological and cultural influence was much more than just copying fashions from neighbours. In the Upper Longshan (Chinese Neolithic), very early in the 2nd millennium BCE, products stemming from the steppe were deeply integrated into the economy and central elements of ritual life — just as the Chinese state system was born.
Most likely just feng.
The periodic famines in southern China are a much, much more recent phenomenon. I’m afraid your incredulity is entirely anachronistic.
More later, I hope.
Trond,
What exactly do they mean by highly diagnostic loanwords associated with the Bronze Age? Diagnostic of what? And how do any of those characteristics apply to ‘seven’?
I’m also baffled by this passage:
> The phonology and extensive agglutinative nominal system of Tocharian have a deviant typology and have played a pivotal role in charting its development (Peyrot, Reference Peyrot, Kloekhorst and Pronk2019b; Warries, Reference Warries2019; Bednarczuk, Reference Bednarczuk2015 ; Krause, Reference Krause1951); a similar change is reflected in the culture that appears as Buddhist, although the conversion can only be few centuries old .
This seems like apples and Volkswagens. What am I missing?
Starting to think “a similar change” means nothing more than “change that happened somewhere in the same multi-millennium period
Ryan: What exactly do they mean by highly diagnostic loanwords associated with the Bronze Age? Diagnostic of what? And how do any of those characteristics apply to ‘seven’?
I read that as loanwords denoting objects or concepts that spread with the culture and economy of the Bronze Age. It’s suggested that one of those concepts was a standardized system for counting high numbers as we know it today. The word for “seven” might have spread with the system, also into Indo-European from “a source akin to” Semitic. I don’t know the reason, but a simple suggestion is that the former counting base was five or six, and when the Indo-Europeans started trading with people that counted longer, they adopted their simple word for one of the most used numbers above. Same story when Indo-Europeans traded with other peoples further east.
By sheer coincidence, this paper is just out:
Ialongo & Lago: Consumption patterns in prehistoric Europe are consistent with modern economic behaviour, Nature Human Behavior 2024
The full article is unfortunately paywalled, but here’s the article in Videnskab.dk that made me aware of it, including rave comments by Kristian Kristiansen. On that caveat, the main finds are:
– Metal fragments — essentially all metal fragments in all Europe — were cut up in pieces on a standard base of 10g.
– The size distribution of the fragments follow a log-normal curve that should be expected when the pieces are money, and the size of the deposits show a distribution that should be expected when metal money is used by all layers of society.
I’m starting to think that the explosive trans-continental spread of metallurgy and mining is a direct result of this system, in a feed-back loop where increased access to metals made the system all-encompassing by the high point of the Bronze Age. Maybe the proto-Indo-Europeans sat in a lucky spot at the right time, just as the Spaniards did with New World gold in the 16th C, or maybe they sat close by and could (strive to) monopolize the new global trade, like the Dutch or the British.
Ryan (again): apples to Volkswagens
Yes. I read right past it, thinking it meant that Tocharian society developed independently of the Andronovo/Scythian sphere, but now I’m not sure. When I write something like that myself, it’s usually because I’ve edited it to pieces.
(Too late for the edit window)
I forgot to say that Ialongo & Lago have worked on this for some time. What is new now is the thorough documentation by statistics.
The 10g unit is an adaptation of the Mesopotamian shekel. We have touched on that before in other contexts.
A standardized money system would promote a standardized counting system.
I think “promote” is an inappropriate word to describe the relationship, because it posits a temporal or conceptual order. It seems to me that the one arises simultaneously with the other – as “standardization” becomes a thing.
A standardized money system is merely one kind of standardized counting system. It’s a good thing that the British money counting system with pounds/shillings etc didn’t “promote” counting everything that way. Otherwise hostesses would have a hard time figuring out if they had laid enough plates for the expected guests.
Everything is a differential equation. Everything that lasts is the result of a feedback loop.
Rocks too ?
In a metal money system, the regions that end up with all the metal are those that produce a surplus of other goods and don’t need (or manage) to use it to finance the import of other useful goods. In early modern world trade, the gold from America ended up in South and East Asia, where it made temples and palaces shine without the average South and East Asian being much better off for it. In the Bronze Age, the metal seems to have ended up in peripheral places like Scandinavia and Sichuan.
It’s very tempting to link numeracy and trade. However, there are large areas, like Australia and Amazonia, where languages where no high numerals exist, and where trade surely existed as much as anywhere. Is there an in-depth study of this issue somewhere?
A prepublication version of Ialongo and Lago’s paper is here. Ialongo has a bunch of other papers in that vein.
Of course, you don’t necessarily need number words to count reliably.
Australian languages famously have few number words, but I gather that there are often complex finger-counting schemes available.
The idea that you have to have words to think coherently is one of those hardy linguistic delusions.
I haven’t seen any studies on this, but then I haven’t really thought about the Bronze Age as the Money Age before yesterday night.
What I have seen – and accepted – is that numeral systems develop when people need them. In the Eurasian Bronze Age, the linguistic innovation could be the standardized counting system rather than just counting, and it could be driven by the proto-monetary system rather than trade in itself. But now that I think more about it, it could even be managing credits and (proto-)bookkeeping. I really wonder how they did that without quipu.
Oh, I don’t expect Ialongo & Lago to have anything to say on the hypothetical spread of higher numerals in the Bronze Age. Their work is archaeological and on metal fragments as money. It’s me who should be blamed for putting the two together.
@Y: Thanks for the link. I’m going to follow up on it, but maybe not today.
I got to think about how in Southern California, while everyone had numeral words, they were a lot more prone to change and borrowing than recent European ones, and I wonder if that had any connection with the spread of the Channel Island shell bead trade.
@DE: Not all trade is based on counting. You can trade an armful of this for a handful of that, rather than five of this for a dozen of that. Anyway, somewhere there must be surveys of trade systems from many different angles. Anthropologists (especially of the older schools) love that kind of thing.
Re: Australia and counting without words: If trade is just an exchange of two or thee commodities then it’s just as easy to trade using barter equivalency of standardized units of measurement, like a bushel per crate, a pile for a bunch. It’s only when the variety of the trade goods becomes too big and their value is understood to be numeric (like smaller and larger metal objects with values linked to weight), then the barrel-per-crate rules grow too complex, and counting and money becomes practical.
I imagine that natural common currency systems like shell beads and cowrie shells and squirrel hides are less likely to induce standardization of the counting base for higher numerals, unless they too are managed in standard bundles of different sizes.
in Southern California, while everyone had numeral words, they were a lot more prone to change and borrowing than recent European ones
Proto-Oti-Volta certainly had numbers up to the thousands, but once you start wider comparison things get a lot more limited. With “Central Gur” you can probably get up to “ten”, with a bit of imagination, but that’s about it.
As far as proto-Volta-Congo goes, “two, three, four, five” seems to be it. (“One” varies a lot, but that’s not too surprising.)
There are a lot of base-five system in West Africa, i.e. ones with “five, five-and-one, five-and-two, five-and-three, five-and-four, ten.” On the other hand “three” and “four” look like cognates all over, even in languages which show precious little other evidence of being related. But then again, *naa “four” is just the kind of shape you’d expect to prove pretty durable over the millennia.
Yélî Dnye has borrowed all its number words from Austronesian (the more remarkable, as it seems to have borrowed very little else.) The Rossel Islanders are apparently famous in the anthropological literature for their highly complex traditional money system. Something to do in the evenings when you’re not working on how to make your language even more complicated, I suppose. (You had to make your own entertainment before television came along.)
In the middle of the night I sometimes wonder who invented multiplication, and FWIW suppose it might have been the Indus Valley folks and their friends in Mergarh who provided them with bricks.
I believe multiplication was invented in the bible, long before bricks: “Be fruitful, and multiply, and replenish the earth, and subdue it”.
I’ve read the Ialongo & Lago paper. I didn’t have the willpower to try to understand the statistics, so for whatever it’s worth: Things are less clear than they seemed, but there are still things to learn.
The main thing to learn is actually that there’s a clear and apparently sudden change in the composition of metal deposits after the introduction of the scale weight (Italy c. 2000 BCE, Central Europe c. 1500 BCE, “Atlantic sphere” c. 1200 BCE). Before the scale weight, there’s mostly complete objects and only few fragments, and there’s no system in the sizes of the fragments. After its introduction, there’s mostly fragments, with most weights distributed narrowly around 10g, and blips in the curve for multiples of 10g. Complete objects show no systematic distribution, before or after the scale weight, so if they’re standardized it’s on something other than weight. Scale weights similarly are narrowly distributed around 10g (except for those that are made for significantly larger weights).
What do I think about this?
My bright idea that metallurgy could have spread because of the demand for metals for transaction doesn’t look so bright anymore. The scale weight was invented in Mesopotamia in the late 3rd millennium and spread slowly in Europe from 2000 BCE, so it’s well after the onset of the Bronze Age. I don’t expect it to have reached Inner Eurasia much earlier.
I find it convincing that bronze objects were cut down to standard pieces that were used for buying and selling goods. I don’t think the fragments primarily were counted as the main check in the transaction. That came with the officially stamped standard tokens called mint or coins*. The fragments were weighed, but the quasi-standard fragments of 10g made it easier to get to roughly the right weight before adding smaller chips. There were probably strategies to do transactions without a scale weight, and they would also be easier starting with standard fragments.
The fact that the composition of the hoards changed with the introduction of the scale weight doesn’t mean that the reason for owning the metal objects suddenly changed. I assume they were kept for their usefulness in trade also before the scale weight, but the operations in trade were different, since the concept of a standard weight didn’t exist.
So what about the idea that a standard base for counting high numbers spread with the concept of money? I don’t think it matters if the transaction units were still understood as weight units and the pieces of metal as a handy way to achieve that, Counting would still be part of the process, and maybe much more so in those transactions that were done without a weight and where both parts should agree on a value. But the introduction of the standard weight unit matters for the timing.
* I actually remember this kind of tokens being used in my youth.
To a mathematician, a number word is secondary to the element of the natural numbers that it names.
I once saw the idea that tally sticks came before larger number words; on the other hand, many humans are able to “see” numbers up to six or nine without explicit counting, like when pouring out seven tablets for my weekly medicine sorting exercise, and I wouldn’t be surprised if that kind of cardinality had names before people had tens of goats to keep track of. (Maybe not tablets, nuts or whatever you wanted to ration).
The Egyptians apparently had an elaborate system of units for grain distribution bureaucracy, cf eg
https://scripturecentral.org/archive/media/chart/egyptian-hieroglyphs-grain-measurement ,
and I read somewhere that in some Central American cultures (Aztecs?) used the cacao bean as a unit of currency; their astronomers – or maybe the Mayans – certainly knew how to count…
As I said above, I don’t think the claim is that the number seven spread with the concept of counting but that the word spread with a standard system for counting in the Bronze Age.
America might actually be a nice comparison. How similar are the number systems of the major civilizations or cultural areas?
Maybe I misunderstood you, but Bronze Age is too late for the assumed exchange of the “7” word between PIE, Semitic, or wherever the word originated. And even for higher numbers – 100 can be safely reconstructed for PIE, and perhaps even 1000 – at least, *g’heslo- is a Greek – Indo-Iranian – Italic isogloss.
Trond will answer for himself, but I took him to mean the possible spread across Eurasia that is the subject of the paper cited a few days ago.
It’s a pretty interesting question when and where Semitic might have given PIE the 7 word. It’s certainly enticing to think there’s a relationship. Hittite for seven is sipta. Assuming that’s the expected form for a word from the root that gave us Sanskrit and Greek hepta, rather than a simultaneous loan into HIttite and it’s sister clade, then the borrowing is awfully early. Could an ancestor or relative of Akkadian have been up in the Caucasus in the early 4th millennium BCE?
Kartvelian almost matches (the promising shvidi in Georgian but shk’viti in Megrelian and shkvit in Laz). Could ‘k’ be epenthetic? Or a regular development from a Semitic sibilant? Or is there a k in a reconstructed Semitic seven word? (No.)
Northeast and Northwest Caucasian seem to have a different etymon as does Sumerian (imin).
You can imagine some sort of trade route, but “imagine” seems all too accurate for the thought experiment required.
For that matter, Akkadian sabe doesn’t seem to account for the d/t sound shared by PIE and Kartvelian.
Another question is whether Semitic is the source or just another recipient of a wanderwort.
Are Hurrian šinta and šitta ‘7’ related? IDK.
See discussion in Björn, Foreign elements in the Proto-Indo-European vocabulary (here), p. 124. He thinks it’s Semitic, following Blažek.
Akkadian sabe doesn’t seem to account for the d/t sound shared by PIE and Kartvelian
Personally, I think all this is just the awesome power of sheer coincidence*, but if you were determined to see an origin in Semitic it wouldn’t be surprising if it went back to the form used with masculine nouns, which has the feminine ending -t Because Semitic.
Sumerian “seven” is formed from “five” plus “two”: the numbers work like the common West African setup that I mentioned. No doubt it all goes back to the Fulɓe conquest of Mesopotamia.
* Björn’s argument is all “probably” and “doubtless.” I note that he cites a source which attributes the Chinese form to the same origin (via Tocharian … someone tell Victor Mair. Though Chinese “ten” is obviously a loan from Fulfulde.)
the awesome power of sheer coincidence*,
* Björn’s argument is all “probably” and “doubtless.”
Thank you for saying that out loud. Some of Bjōrn’s ‘substantiations’ appeal to even weaker evidence. (Eight is a dual for fist/closed hand because 4 fingers, for example.)
I get it that early IE is a lot of speculation. Is this why Mair can call *m(r)it ‘gold standard’?
Personally, I think all this is just the awesome power of sheer coincidence
Same here.
It’s a little hard to understand why one or two number words would be picked up and not others, particularly others numbers higher than the sequence that is borrowed. The idea that higher numbers were rarely used in a pre-trade society might be plausible, but why would IE borrow 7 and hold onto a pre-existing 8 etymon? That seems to prove that higher numbers were already useful and well known. You could maybe understand it if 7 was important in some way to the trade system or cognate with a measurement. Were there cultural meanings to 7 in Semitic cultures other than/before the Judaic creation story?
I’m aware that there are examples in other language families of what is explained as this kind of selective numeric borrowing. Is there ever a more detailed explanation of how and why that happens?
I am sympathetic to the awesome power of sheer coincidence. But that in turn would be more enticing if the etymons for the word 7 didn’t show such similarity across four language families — Semitic, IE, Kartvelian and Hurro-Urartian.
One is anecdote, two coincidence, three a pattern… But four? We’re nearing “God put the fossils there 6,000 years ago because He wanted to make it look like evolution” levels of explaining the evidence away. Maybe, but at four, you can understand why the pagans think there might be science rather than coincidence behind it.
Eight is a dual for fist/closed hand because 4 fingers, for example
Conceivably, proto-Oti-Volta *nii- “eight” actually is a sort of plural of *naa- “four.” The number words often turn up with what look like fossilised noun-class plural suffixes, not always present in every language, and three noun-class pairs had a plural suffix *-i which in fact does cause umlaut of preceding root /a/ to /i/ in some branches of the family (e.g. Kusaal naaf “cow”, plural niigi.)
I can’t think of any way of confirming or disproving this suggestion, though. I reckon it’s fated to remain forever on the list of “what an interesting thought!” (Or not, as you may think.)
Kusaal pisi “twenty”* looks at first sight like a plural-standing-for-two-exactly of piiga “ten”, but, disappointingly, it seems actually just to be from pis “tens” and yi “two” with a bit of smushing** (“thirty” is pis tan’ etc.)
Some Oti-Volta languages do have locally hand-crafted words for some of the numbers six to nine, with still-discernable origins in things like “one less than ten” and so forth, but nevertheless the whole series looks reconstructable to proto-Oti-Volta. But then, it’s not always possible to be sure one is dealing with actual cognates rather than old loans. Some of them might be proto-Algonquian-whisky words after all. (And the word for “money” seems to be really a loan from Western Oti-Volta in all the other branches, on morphological grounds, even though you can reconstruct a POV form for it without breaking any Neogrammarian rules. If counting does go with money historically, that could be a hint.)
* The identical Mooré word pisi “twenty” also means “hundred-franc piece”, because French West Africa.
** Technical phonological term. Goes back to Ladefoged, I believe.
Eight is a dual for fist/closed hand because 4 fingers, for example.
On ‘eight’ in Indo-European… Just for the curious among LH readers, an Avestan passage (Videvdat 13.30) in which the measure ašti- ‘breadth of four fingers extended’ (‘a palm’?) is used. Ahura Mazda is asked how Zoroastrians should deal with a dog that bites without barking or growling:
I am not sure how this muzzle would look exactly.
On ‘eight’ in Indo-European…
The cite there is to this same Bjørn. (Gets himself about a bit. Changing the diacritics on his ø don’t fool me.)
And _is_ it related to four?
What they have in common is kʷ. Any more alike than tʷo cp thʷee?
@Hans: Maybe I misunderstood you, but Bronze Age is too late for the assumed exchange of the “7” word between PIE, Semitic, or wherever the word originated.
Yes, sorry. As Ryan says, I meant the supposed transmission from IE to languages in Eastern Eurasia (of the “seven” word and/or the standard counting system).
David E.: Personally, I think all this is just the awesome power of sheer coincidence*
* Björn’s argument is all “probably” and “doubtless.”
For the record: I agree. Probably.
AntC: Some of Bjōrn’s ‘substantiations’ appeal to even weaker evidence. (Eight is a dual for fist/closed hand because 4 fingers, for example.)
“Substantiate” was my word. By that I meant that he took words that until now have been floating around looking enough alike that actual card-and-no-tin-foil-hat-carrying linguists have suggested that they are borrowings or wanderworts, and he tried to explain how they could actually be related. That adds some substance to the claims — something that others might work with, or at least will be able to evaluate. My own layman’s evaluation is that it’s all special pleading and adds little about the contact situation beyond what’s already known from archaeology. Still, I do think it’s worthwhile to start pulling the threads.
“Eight” as a dual of four is not his invention. The etymology of numbers is mostly internal reconstruction. That’s a risky game under the best of conditions, and much more so with number words, since they’re inadvertently prone to be adjusted to the rhythmic and phonetic environment in the “counting chant”. But again, it’s a worthwhile exercise. Sometimes things turn up that gain wide acceptance (like “hundred” <- "(ten) of tens").
Ryan: It’s a little hard to understand why one or two number words would be picked up and not others, particularly others numbers higher than the sequence that is borrowed. The idea that higher numbers were rarely used in a pre-trade society might be plausible, but why would IE borrow 7 and hold onto a pre-existing 8 etymon?
One is anecdote, two coincidence, three a pattern… But four? We’re nearing “God put the fossils there 6,000 years ago because He wanted to make it look like evolution” levels of explaining the evidence away. Maybe, but at four, you can understand why the pagans think there might be science rather than coincidence behind it.
“How could just that number possibly have spread” vs. “How could that areal similarity arise by chance”. It’s essentially two improbabilities stacked against eachother. It may be coincidence, but not the easy kind of coincidence, It needs being addressed, and readdressed, under new evidence and new methods.
(That said, some of the sevens could just as well be removed from the list of suspects for now. Bjørn’s Chinese was far from convincing.)
Me: One is anecdote
Trond: “How could just that number possibly have spread” vs. “How could that areal similarity arise by chance”. It’s essentially two improbabilities stacked against each other.
Agreed. That’s why I put those in the same comment, but I think the sharper rhetoric of “God wanted to make it look that way” made my position seem less equivocal.
But things might be more interesting. Some reconstructions of proto-Semitic and proto-IE six are also pretty similar. And Wiktionary’s Hurro-Urartian word list gives šeše for six. It’s harder to see how Kartvelian six would fit.* But three languages with pretty similar 6 and 7 lexemes is interesting, and the Hurrians might be in the right place to connect early Semites** with the south Caucasus part of the genetic cline that was recently reported to be behind the early IE gene pool. If that’s a trade route… I can see how 1 through 5 and “two fours” might be more resilient lexemes than six and seven among hunter gatherers moving into a higher level economy.
— — — — —
* — The mere presence of s/š in Kartvelian six words isn’t meaningful. But someone with knowledge of the languages involved might have more to say – there is a pattern that suggests prefixing and an epenthetic k/g on a root that is something like si / tsi /ši. So maybe? I’m also using a Reddit version of numbers in four Kartvelian languages. I know less than nothing about the family myself and have no idea whether Reddit is accurate here.
** — Well, not that early unless we believe the original Semitic expansion came from Mesopotamia, but anyway.
One might argue that the words for six are so short that coincidence is uninteresting.
But consider instead that the six/seven sequence across these 3 language families (maybe 4) has a string of five phonemes that one might reasonably say were linked. Could still be coincidence, but it’s much more remarkable, especially when considered in light of the South Caucasian cline and IE.
Here is an wildly interesting article from the excavator of the Hurrian capital Urkesh. He strings together what he concedes is a “fragile chain of inference”* to make the argument that the foundation of Urkesh was in the mid-fourth millennium, that it was already recognizably Hurrian, that it controlled a hinterland in the highlands that was also the Hurrian homeland. He connexts them to the Early Transcaucasian culture whose burnished ware spread from north of the Caucasus through Anatolia to Syria and Palestine.
This at minimum puts Hurrians ain the right place at the right time with the right scale of development to transmit a few trade-related words from the Amorites in the west or the Akkadians to the PIE people in the northern foothills of the Caucasus.
—-
Fragile, he confesses, but not that fragile. He is convincing on points that Urkesh is Hurrian by the early mid 3rd mill. That the city was established a thousand years rarlier. And that there are significant ritual continuities across that period that seem connected to Hurrian mythology and identity.
Does anyone ever try to estimate the probability of such resemblances arising by chance? I’m not saying it would be easy.
There was once a question at Language Log about whether there was any connection between Yiddish goy, Romany gajo, and Japanese gaijin. Victor Mair added Albanian huaj (Arbëresh goj) ‘foreigner’ and Swahili mgeni ‘foreigner’ (ugeni ‘strangeness, foreignness’). Chris Button added Cantonese gweilo. I’d add Latin gentes ‘foreign nations’, later ‘foreigners’ (Lewis’s Latin Dictionary for Schools), and related words. And I’d add the “gn” in “foreign”, obviously the second morpheme of a compound. Some of those are rather a stretch, but I think “gajo”, “gaijin”, “goj”, “-geni”, “gentes”, and “foreign” make a nice set.
Of course I’m distorting things—in particular, gentes apparently has many other meanings—but if you imagine that our knowledge of those languages consisted of just the right fragments, and that the languages were spoken closer to each other, I think you’d wonder whether it was just coincidence.
Edit: The “g” in “foreign” comes out of nowhere, maybe through the influence of “reign” and “sovereign”, according to etymonline.
Also Nateni cãnn “stranger.”
Actually, it would be nice if proto-Oti-Volta *cà̰n-wà “stranger” were actually cognate with the proto-Bantu *mʊ̀-gènì behind Swahili mgeni, but I can see no way of making it so. Apparently there is a *mʊ̀-jènyì variant in NW Bantu, but I don’t think it helps much.
Does anyone ever try to estimate the probability of such resemblances arising by chance? I’m not saying it would be easy.
Yes:
http://www.zompist.com/chance.htm
https://languagelog.ldc.upenn.edu/myl/Ringe1992.pdf
That zompist page is (of course) excellent:
Thanks, David Eddyshaw and languagehat. I think I probably saw a reference to Ringe’s paper at Language Log but forgot about it.
I’m not entitled to an opinion on “seven” or “honey”, but to the limited extent that I’m entitled to one on evidence of borrowing or relations between languages, I strongly agree that you need to consider, in some statistical way, the probability of coincidence. Yes, easy for me to say.
Ryan: Were there cultural meanings to 7 in Semitic cultures other than/before the Judaic creation story?
The Hebrew word, sheva` in modern Israeli Hebrew, also means ‘oath’, and the two meanings are said to be connected, as at https://www.balashon.com/2006/05/sheva.html . I don’t know whether there’s anything like that in other Semitic languages, or whether some people believe the meanings aren’t really related.
I just searched “Basque” in this post and found only one mention, so: sei ‘six’, zazpi ‘seven’.
I just searched “Basque” in this post and found only one irrelevant mention, so: sei ‘six’, zazpi ‘seven’.
I just searched “Basque” in this post and found three mentions, so soon we’ll have six or seven.
Berber six and seven seem to adhere to the pattern, but I don’t know enough about Berber to know which of the several forms I find to present. I’m waiting for Lameen.
The Wiktionary entry for Proto-Kartvelian *šwid- ‘seven’ is exemplary in that it is fully referenced, with links to scans of all of the references. The summary of the etymology is thin, though.
And of course Egyptian (per Wikipedia):
6: sjsw or jsw (?) (masc.) / sjst or jst (?) (fem.)
7: sfḫw (masc.) / sfḫt (fem.)
It’s interesting if the numbers can (and especially if they can’t) be reconstructed for PAA.
>Berber six and seven seem to adhere
Of course, so do Coptic soou and sasfe. Berber, Coptic and Semitic may share descent from an Afro Asiatic root rather than showing a trade route wanderwort.
[Edited – my Coptic source was Omniglot and I didn’t pause to think about era, so it must be modern, and Trond’s Egyptian etymons are more relevant.]
I’d have guessed if there was any Basque relationship, it was borrowed during Roman times. (And I’m still in the IF camp on all of this.) But it’s still startling to see the Coptic and Basque together.
And truly startling just how many sibilants there are in six and seven roots around the Mediterranean/Middle East/Black Sea, in different language families. We’re at IE, Afro-Asiatic, Hurro-Urartian, Kartvelian, Basque.
I mean whatever. It still may be coincidence. But it is startling.
And then Etruscan. I raise my hand for the duodecimal interpretation.
I’m not gonna mention that the proto-Semitic 3 is a t and a liquid. That way madness lies.
>I raise my hand…
I raise both my 6-fingered hands.
But wherever we go, Václav Blažek was there before us and left his name on Wikipedia’s list of references.
I mean whatever. It still may be coincidence. But it is startling.
OK, it’s starting to startle me.
On the other hand, with however many digits, I like the way the Etruscan word for 1 is cognate to our word for 2.
>On the other hand, with however many digits, I like the way the Etruscan word for 1 is cognate to our word for 2.
Not sure of your level of sarcasm here, or whether you just mean you like how it happened that the Etruscan 1 word looks like an IE 2 word. Cognate is a technical term implying a real relationship, which seems more than unlikely. Despite the phrasing in that section of the wiki Trond linked to, I’m not even sure those researchers believe the words are cognate. I can’t imagine someone postulating that seriously without a really solid chain of evidence. Not that much is even known at all about Etruscan, let alone Etruscan etymology.
The whole IE-derived Etruscan numerals section of the wiki looked less like crackpottery than hoax. For the Hattery’s amusement:
Conversely, other scholars, including F. Adrados, A. Carnoy, M. Durante, V. Georgiev, A. Morandi and M. Pittau, have posited a “perfect fit” between the ten Etruscan numerals and words in various Indo-European languages (not always numerical or with any apparent connection), such as θu ‘one’ and Sanskrit tvad ‘thou’, zal ‘two’ and German zwei ‘two’, ci ‘three’ and Iranian sih ‘three’ (from proto-Indo-European *tréyes, which is not a match to Etruscan [ki]), huθ ‘four’ and Latin quattuor ‘four’, etc.[10][11][12]
I love the dry wiki editor who added “not always numerical or with any apparent connection.”
We’re at IE, Afro-Asiatic, Hurro-Urartian, Kartvelian, Basque.
Lakota šákpe and šakówiŋ. 😄
Let’s face it, “six” and “seven” are obviously phonaesthetic words.
Kusaal ayɔpɔi “seven” has a /p/ in it. QED.
[And the -y- actually comes from *d (long story.) And as we recently established in another thread, the change *s > *d is Totally a Thing.]
Not sure of your level of sarcasm here, or whether you just mean you like how it happened that the Etruscan 1 word looks like an IE 2 word.
I was seriously startled, but I just like the way the 1-2 thing happened, and I suppose the temptation for someone on the lunatic fringe to think the words were really related. I hadn’t read the theory mentioned in the Wikiparticle that the Etruscan for 1 is related to “thou”, but I’m sure I could have come up with some sarcasm.
> Lakota šákpe and šakówiŋ
In contrast to their Hocaąk cousins. Clearly only the Lakota established trading contacts with the Phoenicians.
Berber six and seven seem to adhere to the pattern, but I don’t know enough about Berber to know which of the several forms I find to present. I’m waiting for Lameen..
Most Berber varieties use Arabic loanwords, but sḍis and sa are representative for those that still use reflexes of the proto-Berber forms. They are a little too similar to proto-Semitic, and have been suggested as possible very early loanwords.
Arabic surra سرة “navel” must surely be a loan from Waama surufa (plural surusu) “navel.”
David E.,
I’m curious whether you’d offer any criticisms of this paper on “Numeral Complexity in Hunter Gatherer Languages”.
The paper looked at HG languages in Australia, Africa, South America and the California Great Basin region. They found that 92%, 41%, 61% and 0% respectively have number terms ending at 5.
Um, I thought to ask you because of your interest and expertise in African languages, but maybe all speakers of Oti-Volta languages are agriculturalists or herders. Well, today they’re more likely doctors, truck drivers and short order cooks, but anyway. Maybe you’re still interested.
Is there someone here who has enough familiarity with HG languages to comment? Or who is interested enough to look at the paper and knowledgeable enough to find its flaws?
It seems to offer support for the idea that a wanderwort might have filled a gap, supporting the idea that the “continental scale consonance” in the words for six and seven in much of Eurasia and North Africa might be more than coincidence.
Many of the families with the putative wanderworts provably, and others likely, were in contact with each other at the moment that as hunter gatherer populations, they became connected to the economic zone expanding out of Mesopotamia and Anatolia.
By my count, from the pillars of Hercules to the Caspian, the language families that don’t fit the paradigm are
– Sumerian, NW and NE Caucasian;
Those that do fit are
– Afro-Asiatic, IE, Kartvelian, Etruscan, Hurro-Urartian and Basque
And one offers a match on seven but not six:
– Uralic
Have I missed any ancient language families. Turkic moved in. Dravidian does not fit, but if the theorized mechanism is a new need for higher numbers as the scale of economic activity increases, Dravidian may have been part of an independent early economic development.
The paper of course isn’t a perfect fit at all for the theory, since many of these HG groups have been in contact with agriculturists for millennia. To pursue that idea, the Australian situation may be most relevant, since the only subsistence pattern was HG. The prevalence of numbers above 5 is very low.
maybe all speakers of Oti-Volta languages are agriculturalists or herders
Indeed they are; and in fact a good many agricultural terms can be reconstructed to proto-Oti-Volta, including “millet” (at least two distinct varieties), “rice”, “granary”, “plant”, “sow” (i.e. by casting, as opposed to planting individual seeds) … and, not least “beer.”
“Cow” is readily reconstructable, too, as is “goat” and a specific verb “look after a herd/flock.” Currently, though, most Oti-Volta speakers are arable farmers rather than herders (the Fulɓe tend to fill that niche, as in many other parts of West Africa.)
“Iron” and “blacksmith” are reconstructable too.
I don’t know about trade: the word for “money” looks like a loanword from Western Oti-Volta when it turns up in the other branches, and “sell” shows some irregular correspondences which also suggest borrowing. “Buy” must go back a long way, but the commonest etymon for it is absent in all of Eastern Oti-Volta.
The numbers up to at least the hundreds must go back to POV: “thousand” looks reconstructable, but again there are some odd correspondences suggesting borrowing.
But looking farther afield, it’s certainly the case, as I said above, that lots of West African languages have clearly made up the words for “six, seven, eight, nine” out of “five plus one” etc. (Fulfulde is one, for example.) And I don’t think there is any prospect of reconstructing those numbers for proto-Volta-Congo.
Agriculture goes back a long way in those parts. I don’t really know anything about hunter-gatherer languages, apart from some forest Bantu groups; they’ve all just kept the Bantu etyma for the numbers anyway (and they also tend to have vigorous trade relationships with neighbouring farmers.)
Have I missed any ancient language families.
Elamite? But this word list at Wiktionary gives numbers only up to five.
Have I missed any ancient language families.
It would only be courteous to mention Hattic.
(Schrijver and Kroonen, I think, believe that some Germanic words are of Hattic origin, from back when the Proto-former was neighbor to the latter. Hattic numerals are not recorded, AFAIK.)
Have I missed any ancient language families.
Do you count Polynesian/reconstructs to Austronesian as ‘ancient’? They’re Hunter-Gatherers of the sea, at least. five, seven, eight. I see no additive structure.
Interesting paper, Ryan: thanks!
I note the bit on the fairly common borrowing of numerals under “five”: I despair of the way that “Niger-Congo” maximalists assume that the evident similarity of (say) Mande or Dogon words for “three”, “four”, “five” to Volta-Congo constitutes convincing evidence of a genetic link. It really doesn’t.
(Even on their own wildly optimistic assumptions, you’re talking about a “Niger-Congo” protolanguage at a time-depth of something like 10,000 years ago or more. Everybody was a hunter-gatherer in West Africa back then. Same as Europe. Ah me! Those were the days …)
How could I skip Hattic? And wasn’t there a language isolate newly discovered in the last year, based on the Hittite ecumenistic urge to write down everyone’s rituals in case theirs was the true god. (After looking it up, nope, that’s thought to be a new Anatolian language.)
Languages where we don’t know their number words aren’t useful for my exercise except as reminders of gaps in the data.
AntC, I think it’s David E. who mentioned additive structure. I’m focused on language families whose Hunter-Gatherer proto-speakers would have first come into contact with an economy of a different scale via the Middle East/Anatolian Neolithic expansions. Those where it could be plausible that six-seven similarities represent extremely ancient wanderworts rather than mere chance. I think it’s 6 families with the paradigm, 3 without, 1 score-draw and 2 games in progress — Elamite and Hattic.
Were the Polynesians really hunter-gatherers? I thought you believed that they domesticated kumara (or is your position just that it grew there and they learned to use the wild plant?) But more important, the organization of Polynesian societies I’m aware of doesn’t seem very HG-like. From the US Pacific NW to the Dnieper rapids, I think there’s an understanding in archaeology that fishing cultures even in the absence of (significant) agriculture can still be very different from hunter-gatherer cultures, partly through the effects of aggregating in larger numbers in stable sites.
I just finished the paper. It is interesting. Most of the methodological questions I had are pretty well answered in the paper, so I think it says what it says with appropriate awareness and humility.
It raises a thousand questions that I’ll never have answers for.
But I do feel that the plausibility of my sixes and sevens theory is supported by this paper. A great many hunter-gatherer languages have no numbers above 5, and the likelihood they do is related to their proximity to other subsistence strategies (the focus of the paper and its statistical approach) but also other economies (something they only mention a couple times in passing, rather than statistically). So it seems reasonable to propose that at some great time depth, the proto-protos of many of our Eurasian language families had no words for six and seven, until they began trading with the network expanding out of Mesopotamia/Anatolia and borrowed the words used in the network.
Plausibility. Certainly not confirmation.
Conversely, the paper makes clear that although many smaller scale agricultural groups do have high-limit numbering systems, many others have low-limit and restricted systems. This makes it seem less likely that mere contact with the Neolithic demographic wave would have had such a dramatic effect on Eurasian speech. So “sixes and sevens” may be a marker of the development of, or contact with, proto-urban trading systems.
While this idea doesn’t contradict anything about the known time-frames of Hurrian, IE, Etruscan, Basque and Kartvelian, it could have fascinating implications for Afro-Asiatic, since the continuity of terms across Semitic, Egyptian and Berber might imply that they were already participating in such an economy prior to their branching off. Otherwise, one might expect higher numeral terms to have dropped out at some point.
Though I guess it’s difficult to differentiate between family-scale cognacy and ancient intra-family borrowing at this time depth. Could we recognize the difference between Berber and Semitic sharing the aboriginal Afro-Asiatic roots for 6 & 7 with Egyptian from B & S picking up the Egyptian terms around 4,000 bce through trade with early Nile urban populations, and then adapting the terms till the evolution of the phonemes is no longer transparent.
Cushitic, Chadic and Omotic do not seem to partake of the putative wanderworts. That’s a twist in the plot for me. But maybe that just adds to the likelihood that these were wanderworts even within AfroAsiatic, long post-dating its divergence. Perhaps the original roots were in Egyptian.
Another possible implication, if we were to accept this hypothesis, is that Hurrians, proto-Kartvelian speakers and proto-IE people shared in that early economy in a way that the speakers of NE and NW Caucasian languages may not have.
It’s also interesting that using some concepts from the number systems paper above, Sumerian doesn’t have lexified terms for 6-9, but instead has transparently derived terms: 5-1*, 5-2, 5-3 and 5-4. At the first link, the terms are derived simply from lower numbers. The writer at the second link seems to be suggesting that the number words had started to lexify, but they are still completely transparent. The authors of the systems paper suggest that transparency is a sign of recency. But it’s also obviously a language-internal development.
And it’s also interesting that in the known languages and language groups, no one seemed to borrow or calque Sumerian higher digits. Could this tell us something about Sumerian participation in the early up-river trade routes?
* – Just guessing, but the difficulty of distinguishing i-as from as may have led to the change to dis for 1, and the dropping of the initial i=5 in the word for 6.
long post-dating its divergence. Perhaps the original roots were in Egyptian.
Weee…ll Something like 2,500 BC in proto-proto-who-knows-what, somebody ordered a stone 4.9 x 1.0 x 0.5m weighing 6t (note the suspiciously round metric measurements) to be brought ~750km from NE Scotland to Wiltshire. I can only boggle at what sort of counting and measurement this entailed.
As for the megaliths at Göbekli Tepe …
No reason to snark. I don’t think anyone claims that wanderwort status is demonstrated beyond doubt, or that it ever will be, but it’s a coincidence on a scale that it would be irresponsible not to have in mind while working on the individual language families or the contacts between them. Identifying loanwords is pretty important.
And if linguists working on single language families need to keep it in mind, it would be helpful to have comparative or comprehensive work on the whole consonance cluster. That would come with a lot of maybes, if-this-then-thats, but maybe that’s what a scholar of Basque needs to be able to say that a preform of Berber-if-borrowed-from-Egyptian would actually help making sense of “seven” but not “six”.
Actually, megalithic engineering and the apparently tightly connected early astronomy, is a field where the need for standardized high numbers might arise, but I’d expect those to be more specialized terminology than colloquial language. But it would, of course, be a complicating factor in the hypothesis that the words instead (or also) might have spread as scientific terms long before seeping into the colloquials with e.g. broad access to trade.
Nearly round is a synonym for I eyeballed it, not I counted out units. And megaliths weren’t being built by hunter gatherers anyway. The idea isn’t that agriculturalists couldn’t develop higher numbers. It’s that hunter gatherer interactions wouldn’t often sustain the need for them, but that HG participation in urban trade networks could generate that need and provide the words to fill it.
Well, the familiar (from the clock dial) system of 12 and 60 is natural, because these can be conveniently divide in 2, 3, 4 and 5 (for 60) parts.
6 could be convenient to someone who deals with halves and thirds of six or halves and tenths of 12 or 60.
7 is extremely inconvenient in the very same sense:)
__
If both were significant one would expect 42 or 420 to occur unusually often in the region.
“The ultimate goal is to create an atlas that maps where the words were used and when.”
I approve. Wish there were more such things.
Trond’s is more accurate about numbers. He says that diverse systems could have been replaced by one.
He is not saying that
– any given system is more convenient for trade than others
– that anyone “does not have numbers above N” (e.g. 5).
Only that when a system “one two two and one two and two hand hand and one hand and two…” clashes with a system “one two three… ten ten and one ten and two…” one of them may replace the other.
____
Though this per se does not mean that users of “hand and five” will be more willing to borrow a word for “seven” than users of “seven”.
“Were there cultural meanings to 7 in Semitic cultures other than/before the Judaic creation story?”
Ryan, I’m definitely not a specialist but the number is just as definitely important in Sumerian (and not only) culture. As someone on Reddit suggests, count its mentions in The Literature of Ancient Sumer. Also https://en.wikipedia.org/wiki/Week#History
The seven-day week seems to go back at least to Sargon in the third millennium:
https://en.wikipedia.org/wiki/Week
Whether you think this antedates the Biblical creation story presumably depends on your religious views …
“Three” (male) and “four” (female) are the only culturally supercharged numbers in the Western Oti-Volta speakers’ cultures. Other numbers are just numbers. (Even now, many Kusaasi don’t grok “week”; we used to get many patients asking us how many days we meant if we asked them to come back in N weeks. Weeks are a Muslim thing around there.)
But then, proto-Volta-Congo doesn’t seem even to have had any unanalysable word for “seven.” (Proto-Oti-Volta did, though.)
Well, I suppose gestural counting means more analysable words (up to one, two, two and one, two and two, hand…). Because you dont utter each too often.
“Elamite? But this word list at Wiktionary gives numbers only up to five.”
Blažek (Dravidian numerals) says that Elamisches Wörterbuch has 1 ki 2 mar 3 ziti 5? tuku 80? barba. Also: “It can also be mentioned that F. König ([1965: 42, fn. 15]) offered to interpret the Middle Elamite word nulkippi as “4 pairs”, i. e. “8”. If his solution were correct, the hypothetical root *nul– could be a cognate of Dravidian *nāl “4”. However, Hinz & Koch ([EW: 1016]) interpret this word quite differently, namely, as a plural form of the noun ‘fertility-maker’.”
“8” and “fertility-maker” is not an unexpected pair of “translations” for a word from an ancinet text:( Especially “fertility-maker”, 8 is profane.
Jan Tavernier in his sketch of Elamite (here, open access) for The Elamite World mentions only ki “1”. He then discusses its another meaning in Achaemenide Elamite, “each” and the vowel change and then swtiches to suffixes for numerals and fractions. Perhaps he finds others problematic.
This is why I left Elamite as a draw. I didn’t assume lack of attestation meant that there were no words for six or seven. We just don’t know what they are or whether they match the potential wanderworts.
2023 Thesis of Chams Benoît Bernard, a student of Peyrot, is now available online:
https://www.universiteitleiden.nl/en/research/research-output/humanities/like-dust-on-the-silk-road-an-investigation-of-the-earliest-iranian-loanwords-and-of-possible-bmac-borrowings-in-tocharian
Bernard adds to the list of known Iranian loanwords into proto-Tocharian, and, from their consistent phonology, concludes that all of them have come from the same old Iranian language, which wasn’t related to the Khotanese (the closest Iranian-family neighbor of the Tocharians). Instead, the mysterious “Old Steppe Iranian” language, probably spoken North of the Tian Shan, shared a lot of phonological innovations with Ossetian (but it had at least one additional innovation not found in Ossetian, and therefore, is unlikely to have been the direct ancestor of Ossetian).
Bernard also uncovers a dozen non-Iranian borrowings into proto-Tocharian. These words largely refer to natural phenomena. The hypothesis is that they came from an unknown language of the BMAC
Exciting stuff!
Proof that the Deka is natural and the Germans are sadly missing out.
Oh, that’s even more interesting.
What’s the Italic reflex?
It is. The evidence is a bit thin – namely, šiptammiya- is thought to mean “7th”, IIRC it’s only attested once, and the cardinal number has not been found spelled out –, but, AFAIK, it would fit perfectly. Skt saptá, Greek heptá and the application of Verner’s law in Germanic *sebun make clear that their last common ancestor was, bizarrely, *septḿ, with unstressed *e and a stressed syllabic consonant. (The stress is probably copied from “8”, which was some sort of *HokʲtóH.) In Hittite, *e preceding the stress comes out as i, while *é is preserved as such and *e following the stress comes out as a, so even the unusual stress fits.
(š is used to transcribe cuneiform ultimately by reading Akkadian with a Hebrew accent.)
It has indeed been noted in the literature that its indeterminate-or-something form *sabʕatum (…right…?) is even more similar… assuming the very voiced *-bʕ- ended up as a PIE *p by contact with the *t after IE invented the zero grade (but before the stress shifted from the first to the last syllable by the abovementioned analogy to “8”).
Oh, that reminds me of silver being a Wanderwort that stretches from the Atlas to the Don and may have a Semitic etymology. Check out section 3.39 in this open-access book chapter. It very much does not have the same geographic distribution as seven, though.
No. The idea was that it’s the dual of an otherwise lost word for “4” that was completely replaced, but more likely the “breadth of 4 fingers” word is derived from “pointed” instead of from any word for “4”. The Anatolian word for “4” is completely different.
Not of Hattic origin; but they think that the ante-IE languages of generic west-central Europe filled the large gap between Basque on one side and Hattic + Minoan on the other.
Nonono – like the Romans, the Hittites assumed all gods were true. Unlike the Romans, however, they didn’t think all gods were fluent in all languages, so they wrote down all rituals they could get a hold of in the original languages – followed by a translation when we’re lucky.
Also unlike the Romans, they didn’t equate similar gods worshipped in different cities. Not even the sun goddesses.
No; and many of the Polynesian agriculture-related words go straight back to Proto-Austronesian. Agriculture was largely abandoned in New Zealand, though, because it’s too cold for tropical agriculture there.
Didn’t we have a thread on speculation (by Schrijver?) of a link between shekel and schilling? Does this theory of the shelel as being at the root of a widespread measuring system make that more plausible?
Here, and that was a Vennemann idea. Schrijver reaches sometimes, but doesn’t fall off the end like Vennemann does.
Lehmann’s Gothic Etymological Dictionary says (edited for style),
Previous discussion of shilling here, addendum here.
What’s the Italic reflex?
Latin mille, which is usually reconstructed as *smiH2-ghsli- or so, i.e., a compound with “one” as the first element (I’m not near my library or computer now, but if you have access to de Vaan, you can check the exact reconstruction there).
Ah yes, I’ve come across that before.