Eurasian eureka

« previous post | next post »

After reading the the latest series of Language Log posts on long range connections (see below for a listing), Geoff Wade suggested that I title the next post in this series as I have this one.  If there ever was an occasion to do so, now is as good a moment as any, with the announcement of the publication of Chau Wu's extraordinary "Patterns of Sound Correspondence between Taiwanese and Germanic/Latin/Greek/Romance Lexicons, Part I", Sino-Platonic Papers, 262 (Aug., 2016), 239 pp. (free pdf).

Before proceeding, here are the posts in the current series on Eurasian connections and Old Sinitic reconstructions:

Though it adopts a different approach than the recent spate of Language Log posts on ancient objects and the Old Sinitic reconstructions of the words for them, Chau Wu's paper complements these LL posts that draw comparisons between words in Indo-European languages and words in Sinitic for the same objects or customs.

Chau Wu's paper follows in the tradition of Tsung-tung Chang's well-known "Indo-European Vocabulary in Old Chinese" which was published as Sino-Platonic Papers, 7 (January, 1988), i + 56 pp. (free pdf).  Since Chang's paper appeared nearly thirty years ago, much progress has been made in archeology, Sinitic historical linguistics, ethnography, and other related fields, Chau Wu has more and better resources at his disposal, so we would naturally expect that his results would be more exact and more abundant.  Still, Tsung-tung Chang deserves to be recognized as a modern pioneer in the comparison of IE and Sinitic vocabulary.

It must be pointed out that, so far as I know, neither Chang nor Wu nor myself posits a cognate relationship between IE and Sinitic, a position that some earlier scholars such as Edwin Pulleyblank took.  Rather, what must have occurred is extensive borrowing that started no later than the third millennium BC, when cultural exchange was occurring across the Eurasian steppe.

It must also be noted that, although great strides have been made in the techniques and standards of reconstruction, we should not take the various proposals too literally as reflecting what Old Sinitic actually sounded like at any given period in the ancient past.  All the more, we should not expect that recurring sound correspondences between IE and Sinitic can be identified because, in my view, Sinitic and IE are unrelated.  Indeed, that is not what is at issue, since I, at least, am not attempting to prove a genetic relationship between the two families.  It would be better to think of Old Sinitic reconstructions as formulaic representations of phonological properties within an integral system that evolved through time but that theoretically remained consistent and was supposedly internally coherent — at least in the eyes of those who subscribe to the concept of a single Sinitic family that has lasted for more than 3,200 years and is adequately recorded by the Chinese writing system (i.e., the Chinese characters).

Some reconstructions of OS involve elements, such as "laryngeals", that are clearly abstractions.

For OS there is the added problem that character-forms and character-readings are the vast bulk of the evidence adduced to frame the problem. So what we're really getting is a formulaic representation of the writing system, meaning mostly the writing system as standardized in retrospect beginning in the Han.  Moreover, some skeptics would hold that the division between Old Sinitic (OS) and Middle Sinitic (MS) constitutes a break in the system, so that any projections backward in time from MS are perforce untrustworthy.

[MS — roughly 3rd-13th cc. AD, before that is OS, after that is Modern Sinitic]

Wu obviates the need for reliance on OS reconstructions by comparing words in IE languages, especially Germanic, with Taiwanese (Hoklo / Holó), an extant Sinitic topolect that preserves numerous old features.  Wu's paper constitutes a vast cornucopia of IE-Holó comparisons.  I myself do not necessarily agree with every single one of them, but if even one half or one quarter of them can be sustained, then Wu will have made a monumental contribution to early East-West linguistic exchanges.  What is even more stunning is that Wu backs up his claims based on language with evidence from culture, ethnography, archeology, genetics, and other fields.  Equally exciting is that his methods for the analysis of data are drawn from the hard sciences, in particular molecular biology.  Wu is well versed in traditional Chinese philology as well as in modern, Western linguistics.  He weaves all of these strands together into an impressive, compellingly patterned fabric that combines threads and colors from IE and Sinitic.

Regardless of the reliability of any or all of the existing OS reconstructions, there is plentiful evidence of cultural exchange between eastern and western Eurasia dating back to the third millennium BC and likely even before then.

Brian Spooner, a specialist on Central Asian societies and cultures, states:

I have been convinced for a long time that what we now call the Silk Road was an open highway from very early on — I would even say before the Bronze Age (see what I wrote in "Investment and Translocality", especially parts 2 and 3, starting p. 15 (point your browser here [pdf]).  I suggest that trade was an important factor in the growth and proliferation of cities east and west from Mesopotamia from as early as the 6th millennium, gradually at first, but accelerating, and leading to the development of writing (for recording transactions, to begin with) in the late 4th millennium.

I think that the Steppe was a corridor for traffic and trade from the early Bronze Age, starting already in the third millennium if not before.  See Andrew Sherratt's remarkable posthumous "The Trans-Eurasian Exchange: The Prehistory of Chinese Relations with the West," in Victor H. Mair, ed., Contact and Exchange in the Ancient World (Honolulu:  University of Hawaii Press, 2006), pp. 30-61.

Cf. Barry Cunliffe, By Steppe, Desert, and Ocean: The Birth of Eurasia (Oxford:  Oxford University Press, 2015)


A closing note:

Since I mentioned Ho Dah-an's review of Baxter and Sagart (2014) in this post,

Guillaume Jacques thought that Language Log readers might want to see his paper "On the status of Buyang presyllables: a response to Ho 2016" (submitted to JCL), where he raises some issues with one particular topic discussed by Ho Dah-an, the reconstruction of a presyllable in lù 鹿.

[Thanks to Tsu-Lin Mei, David Branner, Chris Button, and Maria Fasolo]


  1. ardj said,

    September 12, 2016 @ 7:04 am

    Thank you for (as so often) a fascinating introduction to aspects of language I knew nothing about.

  2. Sally Thomason said,

    September 12, 2016 @ 9:39 am

    "Wu obviates the need for reliance on OS reconstructions by comparing words in IE languages, especially Germanic, with Taiwanese (Hoklo / Holó), an extant Sinitic topolect that preserves numerous old features."

    — Wu's methodology looks dubious. A Sinitic lect that "preserves numerous old features" can hardly be assumed to be archaic in all respects (unless of course Sinitic is really really different from languages that have been intensively studied historically, including reconstructions); so the claim that Wu's reliance on "Taiwanese (Hoklo/Holo)" "obviates the need for…OS reconstructions" doesn't create confidence. And Germanic is not a reliable proxy for Indo-European as a whole, among other reasons because Germanic notoriously has a large percentage of vocabulary that can't be (or anyway hasn't been) traced back to Proto-Indo-European.

  3. Victor Mair said,

    September 12, 2016 @ 10:31 am

    The problem, Sally, is that none of the OS reconstructions are any more reliable than Taiwanese.

  4. JS said,

    September 12, 2016 @ 11:20 am

    If Wu finds current reconstructions to rest on shaky (i.e., not-really-comparative) foundations, which is not an unreasonable position to take, it would seem advisable for him/her to at least think in the direction of "Common Dialectal Chinese." But it turns out that s/he finds connections between European languages and Taiwanese that require specifically Taiwanese data (i.e., "L[atin] vir 'man' > Tw bîn 民 'man, people' [v- > b-;-r > -n]"), and rejects widely-accepted Sinitic etymologies in favor of European-language derivations (i.e., Tw peh-sìn 百姓 'common people' is not a transparent Sinitic compound ["the hundred families"] but from Latin "plēbī scītum 'the populace' > simplified plēb-scīt" etc.) This means that, for example, while Tw 女 'woman' is from "IE *dhugh(ə)ter 'daughter' > *du-", Tw hū-lú 婦女 'lady' is unrelated, coming from (or to be compared to?) "O[ld] N[orse] frú 'mistress, lady'." The ideas regarding cultural and genetic connections also relate specifically to Taiwan. I am thus very interested to know what specific historical scenario Wu has in mind; it's hard to square the meat of the study with the claim that Taiwanese is being used as a living proxy for OC or that the hypothesized contact occurred at the OC stage via the Eurasian steppes.

  5. JS said,

    September 12, 2016 @ 11:22 am

    Ah, I now see Wu plans to address the specific migration he envisions in a later paper.

  6. January First-of-May said,

    September 12, 2016 @ 2:33 pm

    Wu's claims as JS describes them are… I don't know… well I think "folk etymology" means a different thing, so maybe crank etymology.

    (That kind of ludicrous comparison is typical for attempts to prove that all languages derive from a single one – Wu just seems to do it the other way around.)

  7. J.W. Brewer said,

    September 12, 2016 @ 3:10 pm

    Can Prof. Mair or someone else explain the claim on page 6 that "Thus, of all the topolects of Sinitic, Taiwanese should be considered closest to Old Chinese." Leaving aside the vexing question of what an appropriate metric for "closeness" is in such situations, it follows a discussion of the claim that it is more distant (synchronically) from a bunch of other current Sinitic topolects than they are from each other. Which is consistent with the claim (which I have seen elsewhere) that the Min group of topolects "branched off" significantly earlier than others, i.e. before the ancestral forms of most non-Min topolects had diverged from each other. But I'm baffled as to how you get from current closeness/distance from other languages also descended from Old Sinitic to closeness/distance to that common ancestor.

  8. Michael Watts said,

    September 12, 2016 @ 4:49 pm

    Moreover, some skeptics would hold that the division between Old Sinitic (OS) and Middle Sinitic (MS) constitutes a break in the system, so that any projections backward in time from MS are perforce untrustworthy.

    If we subscribe to "the concept of a single Sinitic family that has lasted for more than 3,200 years and is adequately recorded by the Chinese writing system", projecting MS readings back must be a reliable indicator of OS. Even if we don't, projecting MS readings back must be a reliable guide to something; they would highlight whatever language was spoken by the linguistic ancestors of the MS-speaking people.

    What kind of break in the system do these people posit? Spelling reform, where OS characters get reassigned to unrelated MS words? Language replacement, where third-century China, speaking a non-Sinitic language, is conquered by a group speaking a Sinitic language, but the old writing system persists through the change?

    I've read an explanation of Japanese as more-or-less a linguistic isolate that is weakly related to Korean, going like so:

    1. The Korean peninsula diversifies from one into three broad linguistic communities, north, central, and south.

    2. South Korea colonizes Japan. The current Japanese people and language descend from these colonists.

    3. Back on the peninsula, North Korea conquers central and south Korea, and standardizes the language on its own dialect, obscuring the connection with Japanese.

    Is there a model for something like that happening with Old/Middle Sinitic?

    Is there continuity of, say, grammatical structures between Old and Middle Sinitic?

  9. Will M said,

    September 12, 2016 @ 5:03 pm

    The reconstruction of 八 pat < *hat < Germanic aht- (sec. 4.8) using Japanese hachi as "evidence" strikes me as especially peculiar. Japanese is unique among the borrowers/inheritors of 八 in having h- as the initial (MSM, Korean, Vietnamese, Lao, etc., in addition to Taiwanese, all have b- or p-), and the alternation among h-/b-/p- in Japanese is reflected in the writing system (は / ば / ぱ) and quite a few morphemes. Directionality also suggests a labial reconstruction.

  10. Eidolon said,

    September 12, 2016 @ 6:50 pm

    I would like to see Wu's work receive a careful review by other scholars. It strikes me that both the internal logic of using modern Taiwanese as a proxy for Old Sinitic and the specific derivations he makes would not be well accepted by the larger linguistic community. But I'd welcome arguments to the contrary.

  11. David Fried said,

    September 12, 2016 @ 7:23 pm

    I have no qualifications in Chinese or in phonology for that matter, and I recognize that most of the article details phonological correspondences that may be very convincing to those who do. detailed However, only skimming the article, I was brought up short by the derivation of a Taiwanese word meaning "smart" from Latin "genius," which of course means not "genius" in the English sense but "tutelary spirit" and derives from Latin "gens," a clan. That doesn't inspire confidence.

    And why does Wu constantly compare Taiwanese, not to proto-Germanic or to Old Gothic, the easternmost of German languages and the first to be written, but to North and West Germanic–English, German, Norse and Dutch? Of course, I'm reserving judgment on the idea that Chinese and the Germanic languages are particularly linked until I see Prof. Wu's case in his next article.

    And whatever the derivation of the Taiwanese words, how surprising is it to find toponyms anywhere that take the form "X-farm" and "X-town"? That's not meant to be a rhetorical question–it should be answerable. And that leads to a larger point. In my cursory read, I saw far too much argument of the sort "Well, there's no way that could be a coincidence." I remember reading once that about 30 words can be expected to roughly coincide in form and meaning in any two unrelated languages. And it's not surprising if they undergo parallel semantic development as well. A random example: "Boor" in English and "bor" in Hebrew both begin with the meaning "peasant farmer" and come to mean "uncourth yokel." Prof. Wu seems to be excessively impressed by this sort of thing.

    But I'm fully aware how cool it would be if he is right. . .

  12. Elessorn said,

    September 13, 2016 @ 12:13 am

    The University of Pennsylvania is home to one of the world's leading experts on Indo-European historical linguistics. Was Don Ringe unavailable for comment?

  13. Matt said,

    September 13, 2016 @ 4:18 am

    The reconstruction of 八 pat < *hat < Germanic aht- (sec. 4.8) using Japanese hachi as "evidence" strikes me as especially peculiar. Japanese is unique among the borrowers/inheritors of 八 in having h- as the initial (MSM, Korean, Vietnamese, Lao, etc., in addition to Taiwanese, all have b- or p-), and the alternation among h-/b-/p- in Japanese is reflected in the writing system (は / ば / ぱ) and quite a few morphemes. Directionality also suggests a labial reconstruction.

    Yes, it is about 99% certain that this sound *was* /p/ in Old Japanese before–when the word written by 八 was borrowed, the Japanese pronounced it with an initial /p/, which only lenited to /h/ much later (like, centuries-to-a-millennium later). That is the explanation for the "regular PSC between Jpn h- and Tw p- in sinograph pronunciation" as the paper puts it: the /p/ was the source of the /h/, not the other way around.

  14. Victor Mair said,

    September 13, 2016 @ 12:00 pm

    Quite unexpectedly, this came to me from one of the world's most distinguished living Chinese historical linguists:

    I offer you apodictic proof that Chinese and English are genetically related. First, English swallow (N., a bird)/swallow (V., “I swallowed my tongue”) corresponds to 燕( a bird)/嚥 (V., to swallow). Second, Chinese 设定 “setting” corresponds to “setting” as in “the setting on my TV”. Third English king/queen corresponds to 乾 qian < gjan / 坤kun. I trust there are passages in the 易经 Yi King which say that the male/female opposition extends to all that is under Heaven, including the British Isles and the Germanic tribes. The first is due to Harold Shadick, the second Ting Pang-hsin, and the third to a paper submitted to BIHP but rejected.

  15. Chris Button said,

    September 13, 2016 @ 4:31 pm

    Pulleyblank’s name is mentioned at the beginning in terms of a possible genetic relationship between IE and OC (coupled with some language contact) as opposed to one solely through language contact as Chau Wu is suggesting. In this regard, it is probably worth stressing just how different Pulleyblank’s approach is from that of Chau Wu.

    Pulleyblank’s take (at least as I see it) seems to have been that if IE goes back to a “vowelless” analysis of syllabic schwa with an ablaut variant /a/ (albeit often obscured by people writing /i/ and /u/ for the sonorants /j/ and /w/ because they have a transcriptional flexibility that other sonorants do not) and if OC also goes back to a similar “vowelless” analysis (albeit obscured by by people often confusing surface phonetics with underlying phonology) then perhaps there is a genetic relationship between the two. Given that Pulleyblank was heavily influenced in his thinking by the “vowelless” analyses being done by people like Sidney Allen and Aert Kuipers on Northwest Caucasian languages (and also looked at vowelless analyses of other far-flung languages), it is unclear whether he would have thought all of this leads back to an even greater proto-language family or not. Unfortunately as a relatively green MA student struggling to keep up with anything Pulleyblank was throwing at me, I never asked him in detail about this and it is sadly too late now.

    My own take is that the most basic element of language is the syllable. A distinction between vowels and consonants is an abstract concept that we are taught rather than something we innately know, and is really something for phoneticians dealing with surface phenomenon rather than phonologists. The syllable is manifested by schwa which may be overtly noted between, for example, two voiceless obstruents, or is something inherent in a sonorant. As a result, the fact that IE and OC were both originally “vowelless” does not mean there is necessarily any connection between the two since this is ultimately the case for all language families if we go back far enough and don’t try to force the data to fit preconceived models.

  16. Will M said,

    September 13, 2016 @ 7:16 pm

    Matt: Thanks! It's probably painfully obvious that Chinese (or Japanese) historical linguistics isn't my field, but I think I'm on the same page with you.

    I imagine you would agree, then, that it's highly unlikely that "At the time of Proto-Holó receiving this word, it may have been thought to have a throaty sound *hat" (previous paragraph), and that Japanese hachi is wholly inadequate as evidence of this. Rather, this is another of the "widely-accepted Sinitic etymologies" JS refers to.

  17. ardj said,

    September 13, 2016 @ 7:17 pm

    @Victor Mair
    I am quite unqualified to comment on this, but would just note that the English verb swallow and noun meaning a bird appear to have entirely different etymologies / sources. Of course that may also be true of the two Chinese equivalents quoted.

  18. Eidolon said,

    September 13, 2016 @ 7:43 pm

    For the record, Chau H. Wu used to believe that Hoklo/Taiwanese culture and language came from Old Europe, specifically Germany, and gave a lecture as such in 2006, which you can still read bits an pieces of online. You could still detect a bit of this in the statements he makes in the paper, such as when he says that an European, perhaps Germanic, people must have brought language and culture to Asia, and specifically to the Hoklo Taiwanese.

    Personally, I find any deep genetic relationship between Chinese and Indo-European, such as those advanced by Pulleybank, an extraordinary claim in need of extraordinary evidence. I think we can all agree that the Chinese did not, as a population, come from Europe or the region in which the early Indo-European/proto-Indo-Europeans are said to have lived. As such, even though early East-West cultural exchanges are certain, that is not enough for the sort of language shift that would be required for Chinese to be a lost Indo-European language. You'd need, as Chau alludes to, some sort of actual Indo-European or proto-Indo-European ruling class in China with the influence to replace the native languages with their own. This requires a lot of archaeological and anthropological evidence, which we simply don't see.

    To this end, while I do believe that there must have been early loan words in Old Chinese from West Asia and Central Asia, including but not limited to Indo-European, and which corresponds to the spread of ideas and technologies that happened around this time and indeed throughout history, I do not think it goes much further than this. It sure would be nice, in a way, for the first and second most popular language families in the world, covering over 66% of humanity, to have been genetically related. But then again, not if the Chinese are led to believe that their ancestors were colonized by Indo-Europeans over 3000 years ago, as that'd just fuel the Western imperialist narrative even more.

  19. Chris Button said,

    September 13, 2016 @ 8:29 pm


    The etymologies of 燕 "swallow (bird)" and 嚥 "swallow (verb)" are indeed unrelated in Chinese as well as English.

    The bird is attested back in the oracle-bone inscriptions and is an early pictograph.

    The verb 嚥 reconstructs as *ʔjan and is the ablaut of 咽 "throat, swallow" which reconstructs as *ʔəɲ (palatal nasal) from earlier *ʔjən. Both of these in turn are related to 讌 "feast" *ʔjans via a semantic extension analogous to the relationship of "swallow" with Norwegian "svelg" (throat) and German "schwelgen" (feast). Ultimately all three are related to 淵 "abyss" *ʔʷəɲ from earlier *ʔʷjən with another IE parallel in Old English "swelg" (gulf, chasm). It should be noted that "swallow", "svelg" "schwelgen" and "swelg" are not genetically related to the Chinese word family in any way, but the semantic extensions are similar because humans tend to make sense of the world in the same way wherever they are.

  20. RP said,

    September 14, 2016 @ 5:37 am

    I'm pretty sure the distinguished Chinese historical linguist quoted by VM is speaking satirically (i.e. he or she is extremely sceptical), so there's no particular need to debunk his evidence.

  21. J.W. Brewer said,

    September 14, 2016 @ 10:07 am

    Perhaps it was unfortunate that Koxinga drove the Dutch out of Taiwan in the 1660's, thus preventing an opportunity for a reconciliation and cultural synthesis between Hokkienophones and their long-lost Germanic-speaking cousins?

  22. Victor Mair said,

    September 14, 2016 @ 1:59 pm

    We'll know more about Chau Wu's understanding of the time and place when IE and Sinitic were in contact when he publishes the second installment of his work, but he could not have been thinking that his Minnan ancestors would have met Germans or Dutch on Formosa around the 1660s because the bulk of them didn't arrive until the 18th and 19th centuries. I'm sure that he will say the time of encounter was much earlier and the place was to the north.

  23. J.W. Brewer said,

    September 14, 2016 @ 2:32 pm

    Yes, the extent to which ethnic-Han people in general would have immigrated to Formosa had it remained under Dutch rule (compared to what happened in reality over the course of the 18th and 19th centuries) is obviously speculative, but if the Dutch had wanted to develop the island economically and the indigenous ("aboriginal") population had seemed inadequately numerous for those needs (and/or too difficult to get to do what the colonizers wanted …) they would have wanted/needed a supply of immigrants from somewhere, and its unlikely it would have been Europeans so the obvious candidates would have been: a) from Dutch-ruled islands farther south; b) from right across the straits in Mainland China and interested in getting out from under Manchu rule. Although I guess a more numerous "aboriginal" population with a smaller-but-economically-significant Han population (as was the case in e.g. the Philippines under Spanish rule) would have been another plausible scenario.

  24. Brian said,

    September 16, 2016 @ 2:53 pm

    From a mathematician's point of view: What is the likelihood that so many cognates could be found between two completely unrelated languages? It depends on how strict "cognate" is. Here it seems that a "cognate pair" means a word in Taiwanese, and another word in any one of English, Latin, Old Norse, Old English, etc., whose first syllable (usually) bears some resemblance to a syllable of the word in Taiwanese. This resemblance can be quite weak (e.g. vir – lin, man – lang) and is not constrained by the usual hypothesis of "regular correspondences," a hypothesis which drastically reduces the likelihood of false coincidences. I suspect one could find this many cognates in any dictionary…

    It does not help that the author seems to be blissfully unaware of widely accepted historical pronunciations — the Latin "genius" was probably pronounced with the "g" of "get," not the "g" of English "genius," and past commenters have mentioned similar errors on the Chinese side.

  25. David Marjanović said,

    September 16, 2016 @ 4:50 pm

    I'll say it out loud: what Wu is doing is pseudoscience. It's no different than the usual Hebrew-Nāhuatl, Basque-everything, Hungarian-everything and Sumerian-everything comparisons, or the Hopi-Tibetan mirror. Instead of reconstructing down to the appropriate historical time depths, Wu picks and chooses from random modern Germanic languages and Latin-pronounced-in-English on the one side and modern Hoklo and modern Japanese on the other, merrily running sound changes backward* or ignoring their regularity altogether.

    I do agree with our esteemed host that 1) there were contacts between IE and Sinitic in hoary antiquity, and 2) as people keep looking, they'll keep finding more of them; the Sinitic "honey" word is transparently Tocharian, so who knows what else is, and indeed who knows if all the words traveled in the same direction! Unfortunately, Wu isn't doing anything to advance this research.

    I further agree that the Min varieties, which are after all not descended from Middle Sinitic, are a mostly untapped source of information for reconstruction of Old Sinitic. Before Baxter & Sagart (2014), they were ignored altogether, and OS was reconstructed from textual evidence alone! Unfortunately, again, Wu isn't doing anything to advance this research either. He just uses Hoklo instead of OS. This is like back in the 18th century when the philosopher Schlegel believed Sanskrit was PIE – only worse, some 3500 years worse to be precise. Of course Min in general and, I'm sure, Hoklo in particular preserves OS features that MS had already lost; at the same time, Min in general has its own innovations, as does every language. Off the top of my head, the three-way contrast of voice and aspiration in syllable-initial consonants is preserved in Wu and "Old Xiang", but not in Min; the glottal stop at the end of "shǎng tone" syllables is preserved in some Wu varieties, but not in Min either.

    * Concerning the Japanese p > h change, the Portuguese missionaries consistently spelled that phoneme as f, the phonetic intermediate – in all positions, not just in fu.

    And Germanic is not a reliable proxy for Indo-European as a whole, among other reasons because Germanic notoriously has a large percentage of vocabulary that can't be (or anyway hasn't been) traced back to Proto-Indo-European.

    The size of this vocabulary has been shrinking for decades; it's pretty small now. But of course the point stands that even Proto-Germanic wasn't PIE and isn't a particularly good proxy for it – let alone for a more eastern ancient IE variety that could have been in contact with Sinitic.

  26. Chris Button said,

    September 17, 2016 @ 7:36 am

    Another pitfall for long-rangers is the notion of being able to set up regular sound laws. When conducting comparative work on two genetically related languages, anything that deviates from regular sound laws automatically suggests some kind of external influence (unless there was wholesale mass borrowing from a certain language at a certain point). Using the "swallow" family mentioned above, the following OC-IE correspondences could be suggested as not merely coincidental:

    嚥 *ʔjan "swallow". English "swallow"

    咽 *ʔəɲ (from *ʔjən) "throat, swallow". Norwegian "svelg" (throat)

    讌 *ʔjans "feast". German "schwelgen" (feast)

    淵 *ʔʷəɲ (from *ʔʷjən) "abyss". Old English "swelg" (gulf, chasm)

    Of course, it isn't down to coincidence because similar semantic developments creating chains of similar word families are common across unrelated languages. However, the idea that this implies any sort of phonological correspondence (genetic or through contact) between the two families is wrong.

  27. Chris Button said,

    September 17, 2016 @ 8:19 pm

    @ David Marjanović

    "I further agree that the Min varieties, which are after all not descended from Middle Sinitic, are a mostly untapped source of information for reconstruction of Old Sinitic. Before Baxter & Sagart (2014), they were ignored altogether, and OS was reconstructed from textual evidence alone!"

    I’m gonna challenge this point a little bit. Firstly this can only really be applied to Min since the other Chinese languages separated off a lot later and can all be accounted for by current reconstructions. Secondly, while the Min evidence is important, I would say the evidence from the earliest inscriptions is far more important. Unfortunately most scholars working on Old Chinese have very limited knowledge, if any, of the earliest inscriptions and so use them incorrectly, haphazardly, or not at all – but very rarely correctly and sufficiently. Furthermore, while Min might help us with the Old Chinese initials, the real problem (as I keep repeating over and over) is with the rhymes. Current reconstructions of Old Chinese simply ignore hundreds upon hundreds of connections across the Old Chinese lexicon because the monosyllables represented by individual characters are considered not to rhyme on the basis of an artificially imposed reliance on vowel-to-vowel correspondences in spite of the fact that rhyming in Chinese manifestly does not work in that manner.

RSS feed for comments on this post