Origins of the Japanese Language (original) (raw)

The problem of the origins of the Japanese language is a vexing one, and it was a subject of a long debate between many competing hypotheses starting roughly in the mid-19th century, although one of these hypotheses, namely the Japanese-Korean hypothesis (which is still very influential today), was first proposed much earlier at the end of 18th century. Besides Korean, Japanese has also been compared to macro-“Altaic,” Austronesian, Uralic, Basque, Indo-European, and even to Maninka in Africa. Most of these short-lived pseudolinguistic proposals are not worth discussing here. Although some specialists consider some of these hypotheses to be more plausible than others, such as the Japanese-Korean hypothesis, the bottom line is that any such genetic relationship must be demonstrable, and since it is universally considered impossible to demonstrate the nonrelationship, the burden of proof for such a demonstration lies on the shoulders of those who support the relationship. However, in actuality both the hypotheses claiming a possible phylum relationship (Koreo-Japonic).1 and those bordering on science fiction (Maninka-Japonic) occupy the same gray area of nondemonstrable relationships, albeit tinted with different shades of gray.

The very notion the “Japanese language” is somewhat misleading. In trying to define what it is, we will inevitably run into another old problem, namely defining what is a language and what is a dialect. For all practical purposes, the approach adopted here is based on the principle of mutual comprehensibility: If neighbors in the dialect chain from A to Z can understand each other, we are dealing with a dialect continuum, but when they cannot, we have a language boundary on our hands. From this point of view (which certainly has its own pitfalls), there are two Japanese languages: Japanese proper, which spans from the Southern Kyūshū to the Northern Honshū, and in more recent times to Hokkaidō as well, and Hachijō, a practically moribund language, still spoken by a few elders on Hachijō Islands about 160 km into the Pacific Ocean from Tokyo Bay.2

1. Japanese and Ryūkūan

These two Japanese languages are relatively closely related to four or five Ryūkyūan languages, which are spoken in the Ryūkyūan Archipelago, which consists of a chain of islands much smaller than the main Japanese islands, stretching roughly from Southern Kyūshū to Taiwan. The now generally adopted name for this small language family including both Japanese and Ryūkyūan languages is Japonic (proposed by Leon Serafim in a series of lectures and talks at the University of Hawai’i at Mānoa in late 1990s, which, unfortunately, mostly remain unpublished, but see Serafim, 1998). This term is now accepted by the majority of Western historical linguists or Japanese linguists writing in Western languages (Bentley, 2008, p. xiii; Pellard, 2011, 2015, p. 14; Shimabukuro, 2007, p. xv; Shinzato & Serafim, 2013, p. xviii; Vovin, 2010), although there are also dissenting voices preferring the older designation “Japanese” or “Japanese-Ryukyuan” as the name for the whole family in question (Frellesvig & Whitman, 2008). In Japan the term Nichiryū sogo (日琉祖語‎) ‘Japanese-Ryūkyūan protolanguage’ is increasingly gaining popularity over the older term Nihon sogo (日本祖語‎) ‘Japanese protolanguage’ as well.

There is much greater linguistic diversity in the Ryūkyūan Islands than on the main Japanese islands, and given the fact that the oldest sources on the Ryūkyūan languages date back only to the 15th century as compared to the 7th century for Japanese, we still have much more gaps in our current knowledge about the Ryūkyūan language history than about the Japanese one. Nevertheless, the genetic relationship even between modern Tokyo Japanese and Shuri Ryūkyūan is quite transparent and is accepted today by virtually all specialists working in the field of the Japanese historical linguistics. Table 1 demonstrates some basic common morphology shared by modern Tokyo Japanese and Shuri Ryūkyūan.

The consonant correspondences are for the most part trivial, but one can easily notice that both Tokyo /i/ and /e/ correspond to Shuri /i/, and that both Tokyo /u/ and /o/ correspond to Shuri /u/. Roughly speaking. This means that the regular correspondences on both sides are productive, but they are completely predictive only in the case of predicting Shuri form on the basis of Tokyo form, but not completely from the opposite direction, because of the merger /e/ X /i/ and /o/ X /u/ in Shuri. However, as we will soon see, the Tokyo distinction between /e/ and /i/ can be predicted in a limited number of cases on the basis of Shuri data.

Isolated morphological markers can be borrowed, so it is important to demonstrate not only the identity of separate markers, but also their actual combinations in a string, in other words, shared paradigmatic morphology. This also can be easily done in the case of Japanese and Ryūkyūan with the following example of comparative sentences in Middle Japanese and Shuri:

We can easily see that causative and finite suffixes occupy exactly the same slots and are identical or nearly identical in Middle Japanese and Shuri Ryūkyūan. This certainly cannot be due to borrowing, let alone to a mere accident. Next we look at the regular productive-predictive correspondences in vocabulary between Tokyo and Shuri, illustrated in Table 2.

As the reader can see, -r- in Shuri undergoes elision if we have the correspondence Shuri i : Tokyo i, but is retained if the correspondence is Shuri i : Tokyo e. Therefore, we can predict Tokyo i or e on the basis of the absence or presence of -r- in Shuri forms.

Shared morphology, especially inflectional morphology, is the best proof of a genetic relationship. The second-best evidence, based on shared basic lexical items, together with the productive-predictive regular correspondences, is presented in Table 3.

Therefore, one can see that Tokyo Japanese and Shuri Ryūkyūan share basic vocabulary based on regular productive-predictive phonetic correspondences. The same picture can be demonstrated for any pair of Japanese or Ryūkyūan languages or dialects. Japanese and Ryūkyūan thus have the same origins, both of them going back to the Japonic protolanguage, which we can call proto-Japonic.

2. Insular Japonic and Peninsular Japonic

Roughly no later than thirteen centuries ago, and probably much earlier, the languages related to Japanese and Ryūkyūan were also spoken in the center and south of the Korean Peninsula. Starting from the later part of the seventh century these were gradually replaced by the Koreanic languages. Unfortunately, our knowledge of these Japonic languages is only fragmentary, as they are predominantly preserved as glosses in the earliest extant Korean historical and quasihistorical chronicles, namely the Samkwuk saki ‘The Historical Records of the Three Kingdoms’ (三國史記‎, 1145 ad) and the Samkwuk yusa ‘The Remaining Deeds of the Three Kingdoms’ (三國遺事‎, 1287 ad), in the Japanese chronicle Nihonshoki ‘Annals of Japan’ (日本書紀‎, 720 ad), and in various Chinese chronicles such as Sānguó zhī ‘Chronicles of Three Kingdoms’ (三國志‎, ca. 290 ad), Wèi shū ‘The Book of Wei’ (魏書‎, 551–554 ad), _Zhōu shū_‎ ‘The Book of Zhou’ (周書‎, early 7th century ad), Liáng shū ‘The Book of Liang’ (梁書‎, 635 ad), and others.

Given the paucity of our data on the Japonic languages of the Korean Peninsula, it is not possible to offer a plausible taxonomy of these languages vis-à-vis Japanese and Ryūkyūan. Consequently, I provisionally adopt the traditional classification based on geography, according to which there are two sugroupings: Insular Japonic, represented by Japanese and Ryūkyūan, and Peninsular Japonic, comprised of the Japonic languages of the Korean Peninsula. It is not clear how many Peninsular Japonic languages might have been spoken there, but again on the basis of geography and in descending order according to the number of attestations of glosses, those are the Silla Japonic language Koguryǒ,5 Karak, and Paekche, and the Japonic language of Chejudo. There is an enormous body of literature dealing with these languages, of which the following are only the major highlights: Yi [Lee] (1963, 1981), Murayama (1963), Kōno (1987), To (1987–2000), Mabuchi (2000), Beckwith (2004), Unger (2005), Vovin (2005a, 2007, 2013), and Ceng (2011).

The genetic relationship between Insular Japonic and Peninsular Japonic is universally accepted in modern scholarship except by the majority of scholars in both South and North Korea, where, unfortunately, the general attitude towards the issue remains politically motivated rather than based on scholarship. This, regretfully, leads to linguistic acrobatics, such as, for example, trying to explain that the pseudo-Koguryǒ numeral *mit ‘three’, transcribed with the Chinese character 密‎ (EMC *mit) reflects a word cognate not with proto-Insular Japonic *mi-tu ‘three’, but with Middle Korean :seyh < proto-Korean *sekiK (Kim, 1978, pp. 65–66).

Before we proceed to the comparison of Insular Japonic with Peninsular Japonic, three important observations have to be made. First, since we are dealing here with very fragmentary materials from Peninsular Japonic, the reader must bear in mind that, unlike in the case of Insular Japonic involving Japanese and Ryūkyūan, not all productive-predictive regular phonetic correspondences between Insular Japonic and Peninsular Japonic can be perfectly demonstrated, simply because we do not have enough necessary data. So, the approach may be much more based on intuition in certain cases, although I have not included any doubtful or speculative comparisons here. Second, because we do not have even a single text recorded in Peninsular Japonic, but merely lexical glosses, most of the comparisons are lexical. There are very few rare cases where we have a glimpse of the morphology. Third, for the most part, the comparisons between the Insular Japonic and different Peninsular Japonic languages are not overlapping, and therefore, I have decided to present the comparisons of Insular Japonic with various Peninsular Japonic languages in separate tables. All these comparative tables go in consecutive order, listing gloss, Old Japanese or Middle Japanese (WOJ if unmarked, or EOJ if WOJ is unknown, or MJ if either of the two OJ varieties lacks an attestation), proto-Insular Japonic, and then finally a corresponding Peninsular Japonic language as shown in Table 4.

Table 4. Insular Japonic and Pseudo-Koguryǒ Comparisons

It was first proposed by Unger (2005), principally on extralinguistic grounds, that the original territory of the Silla kingdom in the Southeastern Korean Peninsula was originally Japonic-speaking, and I have presented linguistic proof for this proposal elsewhere (Vovin, 2007, 2013), which is summarized in Table 5.12 The evidence for the Silla Japonic language comes mostly from the place names recorded in the 34th volume of the Samkwuk saki (Vovin, 2013) and partially from the Cin-han.13 words found in Chinese chronicles (Vovin, 2007). In spite of the fact that these place names do not have glosses like pseudo-Koguryǒ place names do, it is quite clear that many of them can be read in Japonic without any hindrance.

Table 5. Insular Japonic and Silla Japonic Comparisons

Kōno Rokurō has suggested that there was a bilingualism in the Paekche kingdom, involving the language of the nobility and the language of commoners (1987). While this proposal is essentially correct, and although certain words from the nobility’s language can be identified as Koreanic, the words from the commoners’ language have so far resisted positive identification. Unless this language was identical to the pseudo-Koguryǒ language, which seems to be more than probable on historical grounds, as already mentioned, we have only two Paekche words attested in Chinese chronicles that can be more or less safely identified as Japonic, as listed in Table 6.

Table 6. Insular Japonic and Paekche Japonic Comparisons

To Swu-huy identified 14 Karak words (To, 1985, pp. 72–78), out of which three can be more or less safely identified as Japonic. These are presented in Table 7.

Table 7. Insular Japonic and Karak Comparisons

Finally, in Table 8, I also present four words common to Insular Japonic and a Japonic language once spoken on the island of Chejudo. Some Chejudo place names seem to be of Japonic origin. For example, a place in Antǒkmyǒn in Chejudo called Kamsan (柿山‎, ‘Persimmon mountain’) has the old spelling 神山‎ ‘deity mountain’ (Pak, 1988, p. 210). The first character 神‎ ‘deity’ does not have the reading /kam/ in Korean. The most likely explanation is that it reflects OJ kamï ‘deity’ or, to be more precise, some Japonic form that is cognate to OJ kamï. The old name of Chejudo, T’amna (聃羅, 儋羅‎), is a contraction of still earlier Tanmura (

牟羅‎) (Kwen, 1994, p. 167). While Tanmura is meaningless in Korean, it has a transparent Japonic etymology: Tanmura < tami.16 mura (民村‎) ‘folk village, people’s village’ or, less probably, < tani mura (谷村‎) ‘valley village’. The vulgar word for ‘mouth’ in the Chejudo language is kulle, as opposed to Common Korean akari. The former is likely to be connected to Japonic *kutuy ~ *kutu ‘mouth’ (Vovin, 2010, pp. 24–25, 2013, p. 237).

Table 8. Insular Japonic and Chejudo Comparisons

Therefore, it appears that combined Insular and Peninsular Japonic once was a much more varied language family than it is today and that at its height it was spoken both on the Korean Peninsula and on the Japanese islands. It is also universally accepted today that Japonic speakers probably migrated from the Korean Peninsula to Japanese islands around 700–300 B.C. and eventually assimilated the local aboriginal languages. Roughly one thousand years later, the Peninsular Japonic languages shared the same fate, being gradually assimilated by the Koreanic languages.

3. Beyond Insular and Peninsular Japonic

If we begin to search for much more distant relatives of Japanese, we start to enter an increasingly gray area. The most widespread hypothesis is that Japonic and Koreanic are genetically related, but as of today this hypothesis is not universally accepted. The best recent presentation in its favor is that of Whitman (2012), whose work represents further improvement and enhancement of his Ph.D. dissertation (1985). It is full of bright and innovative ideas, and by and large represents a very serious attempt to tackle this old problem. This work is fundamentally different in quality from the near-amateurish attempts to compare Japanese with so-called “Altaic” which I will discuss, but let us see how it fares in light of the comparative method and its requirements.

As I have already mentioned, the best proof for a genetic relationship between languages A and B is shared paradigmatic morphology.17 Neither Japanese and Korean hold world records for the complexity of their verbal morphology, but they do come close. There have been many attempts to find common morphological elements between Japanese and Korean (e.g., Martin, 1968, 1990; Whitman, 1985; Vovin, 2001; Frellesvig, 2001), but all have failed to demonstrate the existence of paradigmatic morphology beyond listing certain isolated morphological parallels. The problem may be partially connected with the fact that, unlike proto-Koreanic, proto-Japonic has very little primary verbal morphology, most of its complex verbal morphology, especially as found in Middle (but not Old) Japanese, being the result of secondary developments. Nevertheless, even with the following isolated morphological markers presented, I trust that the evidence of relationship has not been really cogently presented. I first survey below some of the items from Whitman’s table 2.16, “Proto-Japanese-Korean verbal affixes and postverbal particles” (2012, p. 34, altogether nine [!] items), which is presented here as Table 9, followed by my comments on their plausibility.

The comments concerning this table go line by line:

(1)

pJ *-i is reconstructed correctly, but Whitman footnote #17 (p. 36) makes little sense, since it mentions #37 and #38, not present in this table (the first line is numbered as #39). Whitman is right in claiming that *-i adverbializes not only adjectives, but also verbs, in EOK and LOK texts. However, I am unfamiliar with any infinitive marker *-i in these texts or in any other variety of Koreanic. The infinitive is -e ~ -a in MK, but it only appears as -a in EOK (cf. MK kesk-e ‘break and’ vs. EOK KEsk-a ‘id.’ (折叱可‎, HK II.4). A comparison of pJ infinitive *-i with EOK and pK *-a seems to be somewhat of a stretch from the viewpoint of phonetics. Also, monophonemic comparisons are always inherently dangerous.

(2)

I am unaware of the existence of a pJ clausal nominalizer *-ro that would be universally accepted. I can suspect that either OJ and MJ adnominal form of vowel verbs -uru is somehow behind it, but seeing it as *-oro would be highly speculative, as the internal evidence seems to be limited to a not completely clear hapax legomenon attestation in EOJ (-ôrô in MYS 14.3419), or on Whitman’s perception of -rô in sirô- ‘white’ and kurô- ‘black’, on which more below. There are two additional problems here. First, it is a metathesis, which should be used only as a last-resort explanation. Second, there is a functional problem: MK -ú/ól(Ɂ) is not neutral: It is an irrealis/future clausal adnominalizer, a function clearly not observed even in the hypothetical Japonic cognates.

(3)

There are two other problems here. First, as Whitman correctly notes in footnote 18 (p. 36), pJ *-ku is mostly limited to adjectives. Second, and a much more serious problem, is that OK and MK have only -kwó for this gerund. To the best of my knowledge, -kwu is predominantly a EMK and MdK form, appearing for the first time in the first edition of the TSEH (15.28). But even this isolated MK example can be seen as a sporadic case of progressive vowel assimilation, as it is found in the word form twu-kwu ‘place and’ after the verb twu- ‘to place’.

(4)

On the Koreanic *-é/-á see the comments to (1) above. pJ *-a as an infinitive in irrealis conditional is not a universally accepted reconstruction. Certainly the OJ and MJ form -amba is meant here, but the division of it into *-a- (infinitive) and *-mba (conditional) would create a new problem: Where does the irrealis function come from then? It seems much more likely that we are dealing here with a contraction of *-am-u pa, TENT-ATTR TOP, where the tentative takes care of an irrealis function, and the topic that of a conditional. In any case, segmenting a gerund in a single form while it is not found independently or before any other forms is quite problematic, as this clearly violates the principle of Occam’s razor by multiplying unnecessary entities.

(5)

There are multiple problems here. First, the OJ form is clearly -as-, the allomorph -ös- is secondary and has an internal Japanese explanation (Vovin, 2009a, pp. 842–843). Second, the distribution here is the greatest problem: The form is not attested anywhere in Japonic except in OJ (Old Okinawan attestations are doubtful), which was greatly influenced by OK. Third, I may be mistaken, but to the best of my understanding of EOK and LOK Hyangka texts, as well as of the Kwukyel verbal forms, there is evidence for an *-ʌsi- form, but no evidence for an *-ɨsi- form (Song, 2004, pp. 1042–1043, 1075; Hwang et al., 2009, pp. 159ff). Thus, the latter may be a later development, and the OK *-ʌsi- form looks like a perfect loan source for OJ -as-. Fourth, it is also unclear why -i in OK *-ʌsi- and MK -ʌsi- should be an adverbializer, given that this suffix is followed by other verbal suffixes.

(6)

There are two problems here. First, MK -no- [-nʌ-] is not really a processive aspect, but the present tense (Yi, 1997, p. 268). Comparing a present tense suffix with a perfective aspect marker may be problematic. Second, perfective auxiliary *-n- in both OJ and MJ suspiciously looks like the verb in- ‘to go [away]’, because it shares with it a mixed idiosyncratic consonantal/vowel verb paradigm. Since it always follows the infinitive -i, we can suspect the simple crasis i+i > i, because no vowel clusters were allowed in Old Japanese. Thus, we have here a clear case of grammaticalization, which with a further functional discrepancy completely rules out a comparison with a Korean suffix.

(7)

Opinions differ on the exact function of OJ prefix i-: While Yanagida and Whitman prefer to see it as a prefix of active alignment (Yanagida & Whitman, 2009), I have previously characterized it as the prefix of a directive-locative focus (Vovin, 2009a, pp. 561–569). Even if I am wrong, one major problem remains for Whitman’s comparison: Nominal and verbal morphology in Japonic and Koreanic do not mix, so we would not expect such a mixture in the protolanguage either.

(8)

I do not quite understand what Whitman means here by a “complementizer.” Presumably it is an OJ and MJ so-called quotative particle tö [tǝ]. But this analysis is problematic, given the OJ gerund form tö-te, which demonstrates that it must have been originally a verb. Moreover, OJ tö follows finite forms of verbs, while MK tó follows adnominal forms. So, it is quite clear that while the latter is historically a noun, the former is not.

(9)

There is a serious functional problem with this comparison, which eventually translates into the problem of the regularity of correspondences. OJ ka is predominantly used for WH questions, while OJ ya is used for general questions. In MK ka is used for general questions, but kwo for WH questions. Thus, we should compare OJ ka with MK kwo, but this leads us to the problem of the regularity of correspondences.

Therefore, we can see that none of Whitman’s verbal morphology comparisons is unproblematic. Given the fact that these severe problems arise even with isolated verbal morphological markers, the whole theory of the Japanese–Korean genetic relationship must be seriously questioned.

Whitman further presented three comparisons from the nominal morphology (table 2.17, p. 34), which I reproduce below as Table 10 with my subsequent commentary:

As I have already mentioned in comment (7) on the previous table, nominal and verbal morphology do not mix in either Japanese or Korean. The first two examples offer us exactly this mixture: Japonic forms are nominal, while the Korean ones are verbal. Second, as was already noted in the comment (2), metathesis should be a last-resort explanation, but we face it again in the first of these comparisons. Third, in the case of a normal genetic relationship we should have similar core case markers, such as genitive, accusative, and dative, but expect more discrepancies among the peripheral ones, such as different types of locatives, instrumental, and comitative. In addition to all other problems, Whitman’s scheme does not conform to this general principle: It makes no sense to have just a genitive and comitative, especially since MK tus ~ tos has no comitative function whatsoever. Finally, coming to an old (and odd) comparison of Japonic *pa to Koreanic *pa, I am afraid that we again run into a functionality problem: While Koreanic *pa, which appears exclusively after adnominal verbal forms, is essentially a nominalizer, having nothing to do with topicalization, Japonic *pa is indeed a topic marker that appears freely after nominal parts of speech (unlike Koreanic *pa) and nominalizations as well, while it has nothing to do with a nominalization per se.

Let us now move to Whitman’s lexical comparisons, because longer phonemic strings would allow us to test the regularity of the productive-predictive correspondences much better than do much shorter morphological elements. Since it is impossible to survey all of Whitman’s (2012) comparisons here, I have chosen three tables of his, namely 2.10, 2.11, and 2.13, the latter in abbreviated form (pp. 30–31). His table 2.10 is reproduced here as Table 11.

There are two initial problems that should attract our attention immediately. First, what is the internal Koreanic basis of reconstructing both MK hó- ~ hóy- ‘do’ and MK hóy- ‘white’ as *hjǝ- with a metathesis? The sole motivation seems to be to explain *s- in Japonic. But this is a reconstruction “from above” that has no inherent value in comparative linguistics. Even if we were to accept it, we would still have a great difficulty explaining why pKJ *hjǝ- results in pJ *se- in the case of ‘do’, but in pJ *si- in the case of ‘white’. This is clearly not a regular productive-predictive correspondence. Moreover, the case of the third etymology suffers from the additional vice of long-range comparison. There is no root *ka- in Japonic with the meaning ‘many’ or ‘great’. Therefore, analyzing Japonic *kasa ‘bulk’ as hypothetical *ka- plus nominalizer -sa (incidentally not attested as simply a nominalizer in OJ) leads to the creation of nonidentified segments with putative meanings. In addition, the phonetic change from pKJ *h- to pJ *k- is highly improbable since it involves two consecutive fortitions: *h > *x > *k, which must be somehow motivated.

Second, let us look at Whitman’s table 2.11, presented as Table 12 (I include only the first three items).

Further below Whitman notes: “Regarding pK *c, Table 2.11 presents evidence that it corresponds to pJ *s before high vowels and *t elsewhere” (2012, p. 31). But it remains unexplained how the feature [+high] can be responsible for the assibilation *c > *s. In addition, we get three very different pJ reflexes *-ns-, *-s-, and *-t- all corresponding to pK *-c-. Also, pJ *u corresponds to three different vowels in pK: *ɨ, *o, and *ǝ. By no means are there regular productive-predictive correspondences in any of these cases, and it does not matter that the pKJ reconstruction offers three different reflexes: Reconstructions are always speculative abstractions that can be easily tailored or manipulated; real language data cannot. On the top of these correspondences problems we are presented with two unaccounted-for segments in pK: *-ɨrk and *-k, which is not methodologically admissible in comparative linguistics.

Finally, I will partially present Whitman’s Table 2.13 as Table 13, dealing with proto-Korean medial consonant lenition, but limiting myself only to the cases of the *-k- lenition.

First, there is no internal Koreanic evidence that pK *-k- has ever lenited to MK -0-. Quite to the contrary, there is solid evidence that it lenited to MK -h- or -ɣ- (Vovin, 2003, pp. 101–102). On the other hand, the only consonant for which we have uncontroversial evidence of lenition before *i, as in the first example by Whitman, is *r, as supported by OK *nirim ‘lord’ > MK :nim, OK *YEri > MK :yey ‘Japanese’, OK *NUri ‘world’ > MK :nwuy, and so forth.

Second, in addition to the same problem, both “height” and “inside” comparisons involve metathesis, which, as I have already mentioned, has very low explanatory force in comparative linguistics.

Third, *tǝr cannot be a pK reconstruction, because EOK word is attested as *TɅrari (月羅理‎, HK IV.2, XIII.5). In addition to the internal reasons mentioned above, this makes the presence of an additional syllable *-kV- in pre-pK even more unlikely.

Fourth, pJ *-nk- corresponds in pK to *-0- in two cases, but to *-n- in another case. Again we here have the problem with regularity of productive-predictive correspondences.

Therefore, as demonstrated on the basis of these small selections, the genetic relationship of Japanese to Korean still remains non-demonstrable due to the lack of the predictive-productive correspondences and presence of unaccounted segments, not to mention the complete impossibility of offering valid morphological comparisons, especially paradigmatic morphological comparisons. I doubt that the relationship can ever be proven. Although I said earlier that a shared lexicon represents much weaker proof than shared paradigmatic morphology, in order to add a final touch to the discussion of genetic relationship between Japanese and Korean, I present Table 14, which demonstrates that the alleged Koreo-Japonic family behaves very differently from uncontroversial language families such as Indo-European. I take three modern IE languages and the reasonably well-attested historically oldest forms in Japanese and Korean in order to place the latter two in an advantageous position.

Table 14. Some Basic Vocabulary in Indo-European and KJ20

As the reader can see, Russian and French share all seven basic vocabulary items, and both of them share six items with German. Meanwhile, OK and MK share none—a very bizarre situation for a genetic relationship.21

4. Japanese and “Altaic”

The recent attempts to prove that Japanese is related not only to Korean, but also to the “Altaic” languages fare even worse (Starostin et al., 2003; Robbeets, 2005). In spite of the devastating critique that has been leveled at these quasischolarly publications (Georg, 2004, 2008; Miller, 2004; Stachowski, 2005; Vovin, 2005b, 2009b; Knüppel, 2006), they still continue to sprout like mushrooms after the rain (Starostin & Dybo, 2007; Robbeets, 2013, 2015), greeted, of course, by yet another round of devastating critique (Alonso de la Fuente, 2016). It seems that the best approach to these is not to treat them seriously but rather with a certain degree of humor, which is the maximum they deserve. Here then I will only present three examples of flawed Japanese–Altaic comparisons, as I have already said all that needs to be said about such comparisons elsewhere (Vovin, 2005b, 2009b)—the interested reader should consult these references, and I have little to add other than the critique of additional etymologies. First, even if one ignores the irregularity of vowel correspondences and semantic differences between Tungusic *gusin ‘maternal uncle’ and WOJ kasö ‘father’, before reconstructing proto-Altaic *gusa (Starostin et al., 2003, Vol. 1, p. 575), Starostin and his colleagues should have been aware that WOJ kasö is a loan from Paekche. They should also have sufficient elementary reading skills in Chinese to understand that MJ kánzó ~ kànzó, glossed as 父母‎ in RMGS (Ichizuki, 1974, p. 148), means ‘father [and] mother’ rather than just ‘father’, and, given the prenasalization as well, that it is probably a different word altogether. The last MJ attestations of kaso ‘father’ are in WMS 1:115b and 2:14a. Given their lack of understanding of Japanese historical linguistics (see, e.g., Dybo & Starostin, 2007, pp. 218–219) it would seem that trying to make G. Starostin and A. Dybo understand the difference between pJ primary *e and *o and secondary /e/ and /o/ in Japanese would be as futile a task as explaining the same concept to kindergarten pupils. As for Robbeets, she still cannot come to grasp with notions of transitivity and intransitivity in the verbal word formation and its implications for the comparisons. Inventing new terms such as “manipulative” certainly does not help (2015, pp. 214ff). We do not see there anything but a religious zeal to prove the Japanese–Altaic hypothesis (a dogma?), one which should entertain serious scholars no more.

In conclusion, Japanese is a member of a Japonic family of languages. It has no demonstrable relationship with any other language or language family on the planet.

Abbreviations

Grammatical Terms

Languages

EMJ

Early Modern Japanese

OJ

Old Japanese (both WOJ and EOJ)

OK

Old Korean (both EOK and LOK)