A probably nutty idea about the origin of Japonic

[WeepingElf]

The Japonic languages are usually ascribed to the Yayoi people who brought farming to the islands, while the Jômon people spoke a family of which Ainu is the last survivor. Now, a research paper says that there was a third migration wave, the Kofun people, after the Yayoi. Apparently, the Yayoi people came from southern China, perhaps Taiwan, and the Kofun from Korea.

My idea, which may be utter bullfrogs, is now that the Yayoi people did not speak Proto-Japonic but a non-Malayo-Polynesian Austronesian language (an Austronesian substratum in Japanese has been proposed long before), while the Kofun people spoke a language typologically related to Korean, which was transformed into Proto-Japonic by the influence of the Yayoi substratum.

How much sense does that make to you?

[linguistcat]

I like the idea but as far as I'm aware, the 300s AD was after or about the time that Old Japanese was starting to become distinct from Proto-Japnoic before it and the Ryukyuan languages. It may have even been splitting into the distinct Eastern and Western dialects by that point. Perhaps Proto-Japonic came over earlier and it was only in the 300s that it became the most prevalent language, pushing the Jomon speakers northward (if they are in fact the ancestors to the Ainu), and superseding the Yayoi speakers.

I don't know how the above would be testable from any other situation where the Yayoi might have come from Austronesia, because if that were so, it would result in the islands farthest from Japan being less "Japanese-like" as the islands of Japan's language became more Japanese-like, with the islands in between having more influence from one or the other side of the spectrum. This would be true whether Japanese evolved strictly from whatever the Yayoi spoke with influence from the Kofun people's language, or if the Kofun speakers took over with influence on their language from the Yayoi.

Maybe someone else would see a detail I'm not seeing though. It would be interesting if this were a testable hypothesis even if it turned out incorrect. Also I may be remembering wrong and the split from PJ to OJ may have been thought to have happened slightly later, or much before. I think that would also speak to how likely this would be as a viable theory.

[WeepingElf]

I have to admit that I know close to nothing about Japonic diachronics, such as the degree of diversity in the family. According to Wikipedia, Old Japanese was spoken in the 8th century AD; the conventional dating of Proto-Japonic to 700-300 BC AFAIK is because at that time, the Yayoi people arrived, so arguing with that date would be circular. Does a Proto-Japonic at about 300 AD make sense? Or another possibility would be that Japonic began to diversify before it came to Japan, and the main islands and the Ryukyu islands were colonized by two different branches of Japonic, but developed similarly because of related substrata (kind of like how Brythonic and Goidelic developed the Insular Celtic type in parallel perhaps because of related substrata in Britain and Ireland).

[Ketsuban]

[T]he conventional dating of Proto-Japonic to 700-300 BC AFAIK is because at that time, the Yayoi people arrived, so arguing with that date would be circular. Does a Proto-Japonic at about 300 AD make sense?

Depends how you evaluate the attestation in the Book of Wei of a polity named Yamatai in the 3rd century CE.

The confirmation of a Kofun-era migration is nice but uncontroversial, since it was already proposed as the cause of changes in Japan's material culture—Unger, for example, suggests that it consisted of speakers of "para-Japanese" whose ancestors didn't migrate to Japan during the Yayoi period as the Proto-Japonic speakers did. While we can't be decisive, the evidence for the Japonic family originating in Korea (and being related to Korean) is good enough that I think any proposal of a relationship with another language family would need to contend with it.

[Moose-tache]

One issue with this theory is that Wa, Yamato, and Pimiko would need to be Austronesian, rather than Japonic. Wa is one syllable, and has no obvious etymology in either family. But the other two are more interesting.

Yamato (transcribed Yamatai in early Chinese sources, but there is some evidence this is supposed to be Yamato, and in fact there is a third form Yamaichi, so apparently scribes like wine or something), at least in its first element, has a potential Japonic etymology, and no obvious Austronesian one. Apparently there is some debate over whether Proto-Austronesian had /j/, but the Formosan languages have it, so they might have a word like "yama," but even if they do it won't be as good a match as a wide-spread Japonic word attested in numerous placenames.

Pimiko is somewhat complicated. The second element has been folk-etymologied into the -ko on feminine names, but the Chinese wrote Japonic k with k, while Pimiko is written with /x/. The first element, however, is more interesting. Hime means princess, but it is not clear if this is a back-formation, a real etymology, or even a coincidence. In any case, the situation is even worse if we try to match it to any Austronesian names. Like Japonic, Formosan languages mostly lack any obvious candidate for the /x/ in the third syllable, and I can't find any reconstructible term that the Chinese would transcribe pjemje.

Also, unless the Kofun changeover happened extremely quickly, we would expect to see some evidence of Austronesian in the much denser Chinese records of the third century. An invasion large enough to replace the language of the archipelago in a couple of generations would leave more of a mark than the subtle changes we see in the mid third century. Everything we have from archaeology to later histories points to a small invasion, probably a knock-on effect of the expansion of Baekje into what is today Jeolla province.

In other words, none of this is impossible, because speculation rarely is. But I don't see any strong evidence that pre-Kofun Japanese were speaking Austronesian. If this theory is an attempt to explain the handful of Japonic words that look Austronesian, it is aiming a cannon to swat a fly.

[Rounin Ryuuji]

I had always (apparently mistakenly) understood the Wa to be a Sinitic loan, but the use as 倭 was apparently simply from borrowing the character from Sinitic use, and the on'yomi of 和 is apparently used for its conicidental resemblance. As far as Japonic elements go, it's likely simply the same as in ware, wagahai, watashi and other words meaning "I, me, we, us".

The -to in Yamato is tempting to connect to 戸 "door, gate; narrows", but the Old Japanese kun'yomi was usually */to₁/, probably */t^wo/, and Old Japanese Yamato was usually */jamato₂/ (the exact value of */o₂/ is debated, but it was probably something like *[ə] or *[ɵ]. Other Japonic elements surviving into the modern with */to₂/ are と ("if"), which is unlikely; and 十 ("ten"), which is also unlikely, and often so in compounds (note the fictitious place names Yasogami, Yasoinaba). My general thought is that it's probably a sporadic mutation, and that Yamato origianlly was etymologically "Mountain Pass, Mountain Narrows". The name apparently referred originally to the area around Nara, where there are some big mountains, so it would make sense as a descriptor of the area.

The name Pimiko, modern Himiko, is probably etymologically Solar/Imperial-Exalted-Woman. The element hi-, probably the same as in "sun", appears not only in Pimiko > himiko, but also 姫 (pime > hime - "princess"), 彦 (piko₁ > hiko - "boy, prince") — note the family is mythically descended from the sun goddess Amaterasu, which seems to be composed of elements menaing "heaven-shining". The pitch accent suggests this is not identical to the element in hito ("person"), which might be the same as hi, hito ("one"), but also probably isn't. The element mi appears also in words like miya ("palace", literally "exalted house"). The element probably originally meant "god, deity, spirit", but is probably not connected with the -mi in modern Kami (which was archaically */kami₂/, where mi was */mi₁/; note also the former form has variants in kamu-, kan- in compounds). In Himiko, it might also be an element meaning "woman" (which seems to have been */mi₁~mje/), note that 女 (onna "woman") was (womi_1na in Old Japanese). The element wo- also appears in the archaic reading of 男 ("man"; modern otoko, Classical wotoko). The wo- probably meant "young".

The element tai would not have been characteristic of a native Old Japanese word, but an archaic form of 手 (te - "hand") might have had a proto-Japonic form */taj/, noting the root can be /ta/ in a few compounds, however Yama-te ("Mountain Hand") doesn't make nearly as much sense as "Mountain Pass, Narrow Place in the Mountains".

In my mind, the compelling evidence is that they're all somehow Japonic.

[WeepingElf]

Thank you for the discussion. I have abandoned the idea again - it is always hazardous to draw linguistic conclusions from archaeogenetic studies, as genes and languages often travel together, but often enough not, and archaeogenetic studies can themselves easily be skewed by the small sample sizes (ancient bones from which DNA can be obtained are not exactly common). Also, as I have already admitted, I know hardly anything about Japonic diachronics, so this was doubly hazardous.

[Rounin Ryuuji]

I realise I need to correct a few misstatements — the hi in 人 (hito) does have the same pitch pattern as 日 (hi), but the element hi- as a standalone element meaning "person" is not attested; the word seems also somewhat conflated with the differently-accented readings in 一 (hito) "one", 一つ (hitotsu) "one" + native counter word, 一人 (hitori) "sole, single, one person". It might be possible that hito is an ancient compound meaning "sun-person", I don't think I've encountered the idea before, and think it's more likely that it's a variant form of the root meaning "one", or possibly a back-formation from "hitori", with the element tori or ri (apparently an ancient counter for persons) having "person" as its original meaning.

An etymology for Yamato I hadn't considered could be a syncopated form of an earlier *Yamapi₁to₂ > *Yamato₂, originally referring to the people who lived around Nara. If the more unlikely case given with pito₂ meaning "sun-person", the etymology becomes further simplified to mean "mountain-person" or "mountain-people", but the semantic complication caused by the sun-person etymology for pito₂ pushes me further towards the first supposition.

I've also just encountered that old spellings sometimes use the kun'yomi of the characters 山跡 Yama-ato₂ — "mountain track". Old Japanese tended to assimilate two adjacent vowels into a diphthong, or simply delete one of them, so 山跡 would probably have sounded like Yamato₂. This is also very plausible as an etymology.

The element hitori is possibly interesting to note. Old Japanese apparently had a variant hidari (some words do alternate between -o and -a, notably shiro/shira "white" and kuro/kura "black", note also yo — usually yon in modern Japanese[/i] — "four" and ya "eight", also i/u alternation in mi "three", mu "six"; both in hito "one", futa "two").

[WeepingElf]

Also, the Austronesian substratum in Japanese is about as vacuous as the Semitic substratum in Insular Celtic. All that Japanese has in common with many Austronesian languages is its restrictive phonology (few consonant phonemes, simple syllable structure), but there are of course many languages with such phonologies, many of which don't have anything to do with Austronesian.

[Moose-tache]

Many of the numbers look Austronesian as well, but that's not very telling in a language where we know that foreign numbers were borrowed at least once.

[Richard W]

Many of the numbers look Austronesian as well, but that's not very telling in a language where we know that foreign numbers were borrowed at least once.

It is significant if one is just looking for adstrates. I wonder if English (from some Brythonic) and Welsh (from Irish) sheep-counting numerals are relevant comparanda.

[Raphael]

I wonder if English (from some Brythonic) and Welsh (from Irish) sheep-counting numerals are relevant comparanda.

Now I'm intrigued - any basic information on those sheep-counting numerals?

[bradrn]

I wonder if English (from some Brythonic) and Welsh (from Irish) sheep-counting numerals are relevant comparanda.

Now I'm intrigued - any basic information on those sheep-counting numerals?

See here: https://en.wikipedia.org/wiki/Yan_tan_tethera

[Raphael]

Thank you, that's fun!

[Seirios]

Depends how you evaluate the attestation in the Book of Wei of a polity named Yamatai in the 3rd century CE.

The element tai would not have been characteristic of a native Old Japanese word, but an archaic form of 手 (te - "hand") might have had a proto-Japonic form */taj/, noting the root can be /ta/ in a few compounds, however Yama-te ("Mountain Hand") doesn't make nearly as much sense as "Mountain Pass, Narrow Place in the Mountains".

Yamatai is almost definitely Yamato, because 臺 is reconstructed as *dɯː in Old Chinese and *dʌj in Early Middle Chinese. 3rd century CE was the times of Late Old Chinese, and the pronunciation must have been something in between. Hence it's a good candidate to transcribe OJ *jamatə (the feature of distinction for OJ stops was prenasalization instead of voicing)