See here for the first part.
A very brief history of the Altaic hypothesis and the Finnish connection.
The Altaic language family was initially proposed during the early 1850s by Finnish scholar, founder of Finno-Ugric studies, M.A. Castrén (1813-52). He himself expressed caution on whether genetic ancestry between the languages could be established. Early on the Altaic family (sometimes referred to as Tartar) was also considered to include what is now treated as the Uralic language family (comprising the Finno-Ugric and Samoyedic branches) however during the first half of the 20th century, the lack of tenable cognate vocabulary led to the distinction of Uralic and Altaic as two separate families; in the 1970s, Pentti Aalto (1917-98) resurrected the Ural-Altaic hypothesis but his assertion was primarily based on structural rather than lexical similarities. The Uralic languages do possess similar grammatical features to the Altaic languages so it is still possible, and at times useful, to discuss in typological terms a Ural-Altaic complex.
The most famous name in Altaic studies, certainly in relation to Korean, is another Finn, Gustaf Ramstedt (1873-1950). Foremost a specialist in the Mongolic languages, though with an extensive command and knowledge of many other, Ramstedt actively acquainted himself with Korean whilst working at the Finnish embassy in Japan between 1919-30. Aside from writing the first English language grammar of Korean (1939), Ramstedt did most to systematically incorporate Korean into the genetic Altaic hypothesis through the comparative method (as opposed to merely speculating on structures and look-a-like words in their modern shapes) and it is primarily his quite seminal albeit tentative and imperfect work, beginning with a short paper Remarks on the Korean Language (1928) and culminating in Studies in Korean Etymology (1949) which has formed the basis for postwar Korean scholars and popular history writers up until today. It was, incidentally, in the former publication that Ramstedt proposed the linguistic homeland of the Altaic languages to have been located, according to his estimate some 4,000 years ago, around the Khingan mountain range (興安嶺) in western Manchuria as opposed to the Altai mountains after which the language family was originally named.
Slightly proceeding Ramstedt, another pioneer scholar whose 1914-16 work is often relied on by Korean Altaicists, is Japanese ethnographer Shiratori Kurakichi (白鳥庫吉 1865-1942); having studied in Europe Shiratori was influenced by Heinrich Winkler’s (1848-1930) Uralaltaische völker und sprachen (1884) and from as early on as the 1890s he had proposed Korean as a member of the Ural-Altaic family, however, in contrast to Ramstedt, his hypothesis was explicitly based on grammatical similarities rather than sound correspondences (the crucial determinant in the comparative method). Ramstedt and Shiratori are certainly not the only scholars to have contributed to the Altaic discourse but relating to Korean, their works have remained by far the most representative and influential.
Are Korean and Japanese related?
Once more: typologically yes, but genetically not so much at all. Attempts to include Japanese in the Ural-Altaic family began as early as the 1850s but, for the same reasons as Korean, were not very successful.
Conveniently for the nationalist sensitivities of Korean Altaicists today, Ramstedt did not so emphatically associate Japanese with the core Altaic languages as he did Korean. Having mastered Japanese sooner, he did in fact explore Japanese-Altaic connections as early as 1924 (“A Comparison of the Altaic Languages with Japanese”) previous that is to focusing on Korean; at this early stage he also took the view that Japanese and Korean were genetically related but his final position seems to have been that he regarded Japanese a distant relative to proto-Altaic (what might be termed “para-Altaic”) whilst Korean he considered an early branch within Altaic. Studies in Korean Etymology (1949) thus includes some suggested genetic cognates with Japanese and Ryukyuan (the language of Okinawa genetically related to Japanese), but far fewer than for Tungusic, Mongolic or Turkic languages.
Perhaps more than with the core Altaic language families, modern Korean clearly exhibits a close structural similarity with Japanese but still only a relatively tiny amount of shared vocabulary: certainly not enough to posit them as genetic cousins split from a common ancestor. Modern (Sino-)Japanese, therefore, is best regarded as belonging to the Japonic language family which also includes the Ryukyuan languages of Okinawa.
By contrast with Korean, Ryukyuan shares a great amount of its basic vocabulary with Japanese and this is what a genetic relationship should look like.
To elaborate a little more, there are two primary processes through which languages are made or evolve: divergence and convergence, the former representing genetic division, the latter aerial contact. Where languages have diverged or split from one another, they will retain the same basic vocabulary; where they have converged they will possess different basic vocabulary but share other secondary vocabulary. These processes are not mutually exclusive and rarely occur in isolation (except in the neat case of uninhabited island settlement), so even as a language is diverging it will most likely at once be converging with others.
Being an island nation the Japonic languages had to have been taken there at sometime and simultaneously something has to account for the typological affinity with Korean. Given Koreanic seems to have been situated in its peninsula homeland for as far back as we can go, though certainly not forever, the most reasonable explanation is that for a significant period of time the two language families co-existed in close proximity on the Korean peninsula, at which time the Japonic language was typologically “Altaicized” and spread to the Japanese archipelago and Ryukyu.
Based on methods of internal reconstruction, linguists are generally agreed that, unlike the Altaic languages, proto-Japonic was a monosyllabic language, that is similar to southeast Asian languages and Old Chinese which are termed typologically as “Sinic” though genetically they were distinct; “Sinic” is not the best term as it implies the dominance of Chinese which at this early date would not have been the case. What is likely though, is that Japonic had continental origins from whence it crossed to the Korean peninsula and later on to the Japanese islands.
The migration of Japonic to Japan may be archaeologically identified with the bronze-iron age Yayoi period (c.300BCE-300CE) which shows both a distinct break and replacement of the prehistoric Jomon culture. By the end of this period, close material connections are known to have existed between the Gaya states on the south of the peninsula and the Japanese isles; additionally there is the now politically sensitive ethnonym of the Wa/Wae (倭 왜) which turns up in early sources such as the Gwanggaeto Stele (廣開土王陵碑 erected 414CE) and may have originated on the peninsula or been based on Tsushima island, independent of any nascent Yayoi polity then establishing itself in Japan proper.
It is during this early stage that Japonic spread across the Honshu island, pushing the Ainu people northwards, and also to the Ryukyu islands. Meanwhile the Japonic left on the peninsula (termed para-Japonic) further interacted with Koreanic whilst becoming itself the dominant language of Baekje, a kingdom historically well attested as having close links with, and even direct involvement in, the establishment of the Yamato dynasty (based in the Kinki region of modern Japan). During this period and in particular following the overthrow of Baekje, there was a further and indeed final migration of para-Japonic speakers, people of Baekje, who brought their closely related Japonic language but with a greater number of Korean loanwords, whilst also leaving Japonic loanwords on the peninsula, which account for what otherwise appear as genetic cognates between Korean and Japanese today. The proof for this explanation is that the Koreanic influence is predominantly found in Old Central Japanese, in the homeland region of the Yamato dynasty, but less so in the dialects which had spread further east and not at all in Ryukyu (Vovin 2010:240)
The evidence for the relationship between Baekje and Yamato is so strong that it has even been acknowledged by the current Japanese emperor, Akihito (although that linked article is quite wrong in suggesting that Japan’s peninsula heritage or Akihito’s lineage is ethnic “Korean” because the very point is that those people did not remain on the peninsula to become Korean), but it is not a topic otherwise popular with the race-based nationalist ideologies of either the two Korean states or Japan today. For the Koreas this is related to post-colonial sensitivities and for Japan, it would undermine their colonial rhetoric of historical superiority and kokutai (國體) myth of exceptionalism pertaining both to race and language which has never been fully revoked.
Whilst on the peninsula para-Japonic would also have influenced Koreanic and perhaps have formed a substrate over which Koreanic ultimately expanded and gained some of its distinctly non-Altaic traits such as pitch accent and honourific speech levels.
Today it understandably remains ethnic anathema for Koreans to consider that a non-Koreanic language, particularly one ancestral to Japanese, was ever spoken on the peninsula, least of all by one of the celebrated Three Kingdoms, albeit the least popular one! The sensitivity originates in particular from the Japanese colonial era (1910-45) claim that early Yayoi contact with the Gaya polity (named Mimana in Japanese) and misidentification of the Wa/Wae people as “Japanese” was evidence of an ancient Japanese presence on and control over the peninsula; thus Koreans feel compelled to refute the notion that there was ever any relationship between early Korean and Japanese polities even when objective reasoning and historical sources tell us that the direction of cultural and linguistic influence would have been from the peninsula to Japan rarely the opposite direction. One might expect Korean nationalist rhetoric to happily reverse the 20th century colonial argument and claim Japan had once been an ancient colony of Gaya and/or Baekje, but rather the thrust of modern Korean historiography has been to emphasize homeogenous origins and maintain territorial and ethnic integrity over any desire to lay claim on Japan; where they exist, Korean irredentist urges are directed towards former continental territory in southern Manchuria.
Shouldn’t the earliest recoverable layer of the surviving Japanese language be from the Jomon period, prior to the Yayoi or Baekje influx?
Not if the Yayoi and Baekje migrations pushed to the margins and otherwise replaced wholesale the indigenous languages such as occurred to the Celtic languages during the Anglo-Saxon settlement of Britain. Indigenous Ainu, for example, (which may not have been related to other Jomon languages) was not absorbed into Japonic but displaced by it; this was a gradual process and Ainu was in aerial contact with Japonic from an early period as it has preserved some loanwords from Japonic in their earlier shapes.
Wouldn’t it be better to refer to the Japanese language family as “Baekje-ic” rather than Japonic?
Or perhaps “Kudara-ic” as a compromise, Kudara being the Japanese name for Baekje and therefore, arguably, closer to its original name. But, no, because language families tend to be named after their surviving representatives; hence Koreanic is not “Sillaic” because it is the people who self-identify, internationally, with the ethnonym “Korean” rather than Silla, who are the attested speakers of the modern language. This can be misleading in more complex language families such as Mongolic where Mongolian was only one branch of a very widely spoken language family which included other major branches, identified later, such as the Khitan language of the Liao dynasty predating Ghengis Khan. Similarly, individual branches of a language family tend to be described in relation to the more, or earliest studied, attested branch and named as “para” languages; consequently the Khitan language may be termed “para-Mongolic” and, quite separately, the dominant language of Goguryeo may be proposed as “para-Jurchenic” but this should not in any way be misunderstood at implying linguistic or ethnic subordination. Thus the language which was left on the peninsula to constitute itself as a separate branch of Japonic, can be termed para-Japonic.
I still need more convincing..
Compelling evidence for the non-genetic nature of the relationship between Koreanic and Japonic has been produced by the Russian-American scholar Alexander Vovin. Aside from the depth of his knowledge, particularly on the side of Old Japanese, one significant aspect of Vovin’s work is that for a long time he was himself an adherent to the Altaic hypothesis inclusive of both Koreanic and Japonic. The motive of his study published in 2010, Koreo-Japonica, had initially been to substantiate the genetic relationships, but in the process he instead confirmed the opposite. The results of his exhaustive examination of 347 proposed Korean-Japanese cognates ultimately rejects 261 of them as being problematic according to the comparative method, it identifies 75 as definite loans leaving only eleven as “possible”, which were further whittled down to just six that “seem to be impeccable in the regularity of their phonetic correspondences”. (Vovin 2010:237) These were words for “fire”, “to fill [with water]”, “to hold/take”, “crane”, “field” and “melon” which do not strongly represent basic vocabulary items. The ratio of loan words to cognates, meanwhile, overwhelmingly bespeaks of aerial interaction.
In other areas, when analyzing Korean and Japanese personal pronouns, for example, the only possible cognate Vovin identifies is Middle Korean first person “I” na (나 still used today) attested in Western and Eastern Old Japanese, but crucially not the other Japonic branch of Ryukyuan; another word for “I”, wa, meanwhile is found in all of the Japonic dialects and not Korean, indicating that wa is mostly likely the Japonic word for “I” whilst na is a secondary borrowing from the language carried by the 4th century Baekje influx (Vovin 2010:65). It should be noted, however, Vovin regards the Baekje language to have been Koreanic with a Japonic substrate; so he considers the peninsula Japonic to have existed as a dominant language only previous to the emergence of the Three Kingdoms which, while not impossible, is perhaps less convincing when considering the nature of the close political relationship maintained between the Baekje and Yamato elites up until Baekje’s mid 7th century demise.
To give another example, in the case of demonstrative pronouns, modern Korean and Japanese share a similar system of i~ (이), geu~ (그) and jeo~ (저) with Japanese ko~, so~ and a~ (“this”, “that” and “that over there”), but Vovin points out they “seem to have been functionally much more different in earlier times than they are now,” indicating once more aerial convergence rather than genetic divergence. And, further, the words themselves are not genetically cognate.
Is the genetic Altaic language hypothesis completely dead?
Not entirely and there are still a few advocates. The main point to consider is that after decades of intensive effort and high level scholarship from both sides of the debate, the Altaic theory remains unproven, both in terms of the genetic relationship between the core languages and the relationship of Koreanic and Japonic, either to each other or to the others, and today a greater number of historical linguists specialized in typologically Altaic languages stand against rather than for it.
There is instead the view that more can be achieved working on internal reconstructions and establishing the nature of relationships, whether genetic or aerial, between some of the individual Altaic language families before expending further efforts on the “long range” Altaic question.
One of the most emphatic arguments used by proponent Altaicists is that any of the Altaic languages in question are notably more similar to one another than modern Greek appears to Swedish, yet Greek and Swedish are accepted as being part of the same genetic Indo-European family; therefore it should not be so unreasonable to postulate the certainty of an Altaic language family. I know neither language, but assuming Greek and Swedish are as different as they say: if only those two languages and no other intermediate Indo-European language were known then, indeed, it surely would have been impossible to establish the Indo-European family and any such hypothesis would have to have remained unproven, as is the circumstance of Altaic. In fact the situation with Indo-European is, apparently, the opposite in that there are too many attested languages and whilst their genetic affinity is more certain, untangling the knot of secondary areal interactions between them is a real challenge.