Classification of the Japonic languages

The immediate classification of Japanese is clear: it is a Japonic language, along with the Ryukyuan languages. Traditionally, these are considered dialects of a single language isolate. However, more distant connections remain contentious among historical linguists. The possibility of a genetic relationship to the Goguryeo (Koguryŏ) language has the most currency; a relationship to Korean is widely considered but is problematic; an Altaic hypothesis is less widely accepted. A few linguists support the hypothesis that Japanese is genealogically related to the Austronesian languages.

Possible external relations

There are several hypotheses on the relationship of the Japonic family to other language families. These hypotheses are presented below in approximate descending order of their current acceptance (cf. Vovin 2003).

Japanese-Koguryoic

The Japanese-Koguryoic proposal dates back to Shinmura Izuru's (1916) observation that the attested Koguryo numerals, 3, 5, 7, and 10, were very similar to Japanese. The hypothesis proposes that Japanese is a relative of the extinct languages spoken by the Buyeo-Goguryeo cultures of Korea, southern Manchuria, and Liaodong. The best attested of these is the language of Goguryeo, with the more poorly-attested Buyeo languages of Baekje and Buyeo believed to also be related. Supporters of this theory do not include modern Korean as part of that family because it is thought to have derived from the ancient language of Silla and it has been shown that the Korean and Buyeo-Goguryeo languages share only a few lexical items, which are typical cultural loanwords. A recent monograph by Christopher Beckwith (2004) has now established that there are about 140 lexical items in the Goguryeo corpus alone. They mostly occur in place name collocations, many of which include grammatical morphemes (including cognates of the Japanese genitive marker no and the Japanese adjective-attributive morpheme -sa) and a few of which reveal syntactical relationships. The majority of the identified Goguryeo corpus, including all the grammatical morphemes, are clearly related to Japanese. Most discussion of this theory now centers on arguments about the identity of the speakers of the language recorded as Goguryeo, but so far the identification of the language with the Goguryeo people, which agrees with the ancient Chinese accounts, has been shown to be the most secure historically and linguistically (Beckwith 2006a, 2006b).

Korean hypothesis

William George Aston suggested in 1879 in the Journal of the Royal Asiatic Society that Japanese is related to Korean. A relationship between Japanese and Korean was endorsed by the Japanese scholar Shōsaburō Kanazawa in 1910. Some other scholars took this position in the twentieth century (Poppe 1965:137). Substantial arguments in favor of a Japanese-Korean relationship were presented by Samuel Martin, a leading specialist in Japanese and Korean, in 1966 and in subsequent publications (e.g. Martin 1990). Other linguists advocating this position include John Whitman (1985) and Barbara E. Riley (2004), and Sergei Starostin with his lexicostastical research The Altaic Problem and the Origins of the Japanese Language (Moscow, 1991).

The possible lexical relationship between Korean and Japanese can be briefly exemplified by such basic vocabulary items,see tabel below, and Martin 1966.

Comparison with Japanese
Old Japanese	Japanese	meaning	Mid-Korean	Korean	meaning
midu	mizu	water	myr	mul	water
midu	mizu	water	mos	mot	lake
k-u	k-uru	to come	ka-	ka-da	to go
kata-si	kata-i	hard	kut-	kud-yn kut-yn	hard
wi-ru	i-ru	to be	'	i-da	to be
naɸ-u	na-i	not	an	anh	not
mïna	mina	all, everyone	man-ha-	manh-	many
kasa	kasa	hat	gat	gat	hat

The same possible cognates are often observed in other members of the potential Altaic family, especially among the Tungusic languages. Compare, for instance, Nanai muke "water"; giagda- "to walk on foot"; anaa, anna "not" (from Starostin's database).

The considerable number of shared words within the basic vocabulary which have been cited by different scholars can hardly be attributed to mutual borrowing. The hypothesis is also based on a high degree of typological and grammatical similarity, almost obvious to anyone familiar with both languages (Beckwith 2004).

Some critics of this hypothesis (such as Alexander Vovin) claim that there are difficulties in establishing exact phonological laws and that Japanese and Korean have few shared innovations. There are also drastic differences between the native Korean and Japanese number systems.

The idea of a Japanese-Korean relationship overlaps with the extended form of the Altaic hypothesis (see below), but not all scholars who argue for one also argue for the other. For example, Samuel Martin, who is a major advocate of a Japanese-Korean relationship, has only provided cautious support to the inclusion of these languages in Altaic, and Talat Tekin, an Altaicist, includes Korean in Altaic but not Japanese (Georg et al. 1999:72, 74).

Altaic hypothesis

According to its proponents, Altaic is a language family consisting at a minimum of Turkic, Mongolic, and Tungusic. Since the publication of G.J. Ramstedt's Einführung in die altaische Sprachwissenschaft ('Introduction to Altaic Linguistics') in 1952–1957, most Altaicists have included Korean in Altaic. Roy Andrew Miller's Japanese and the Other Altaic Languages, published in 1971, convinced most Altaicists that Japanese should also be included in Altaic. The most important recent work in favor of this expanded Altaic family is An Etymological Dictionary of the Altaic Languages (3 volumes) by Sergei Starostin, Anna V. Dybo, and Oleg A. Mudrak (2003). (See Altaic languages for a more detailed treatment.)

The Altaic family is by no means generally accepted, either in its core form of Turkic, Mongolic, and Tungusic or its expanded form of Turkic, Mongolic, Tungusic, Korean, and Japanese. The best-known critiques are those of Gerard Clauson (1956) and Gerhard Doerfer (1963, 1988). Currently active critics include Stefan Georg and Alexander Vovin.

Evidence for this grouping lies in the fact that, like the Turkic languages and Korean, Japanese is an agglutinative language. Additionally, there are a suggestive number of correspondences in vocabulary, as shown in the following table.

Japanese	Turkish	English gloss	notes
ishi	taş	"stone"
yo	dört	"four"
kura	kürtün	"saddle"
ya-	yak-	"to burn"	Turkish yak- is exclusively transitive ("to burn (it)", "to light (it) on fire"); intransitive counterpart is yan-
ki-	kır-	"to cut"	Turkish kır- actually means "to break; to split, to chop (wood); to fold; to destroy, to break (resistance, pride, desire, etc.); to reduce (price); to offend, to hurt": cf. Turkish kırma "deverbal noun derived from the verb kır-; a pleat, a fold; folding, collapsible; groats; hybrid, mongrel". Turkish kes- is more specifically "to cut".
inu	it	"dog"	cf. Manchu indahŭn, Nanai ida, Ainu seta, Chinese "zodiacal dog" 戌 *zyüt, Jeju "puppy" gaŋsæŋi
kuro	kara	"black"	cf. Ainu kur "shadow", *kur-ne > kunne "black; dark"
kura-	karar-	"to be dark"

These examples come from Starostin's database, which contains a comprehensive list of comparisons and hypothetical Altaic etymologies.

Eurasiatic hypothesis

Joseph Greenberg (2000-2002) argued for the inclusion of Japanese in his proposed Eurasiatic language family. In contrast to Sergei Starostin, he rejected the inclusion of Korean in Altaic. According to Greenberg, Japanese-Ryukyuan, Korean, and Ainu form a separate subgroup within Eurasiatic.

Like other language classifications of Greenberg's, the Eurasiatic family is often attacked on the ground that it is based on "mass lexical comparison"; however, this is a fictitious method. Greenberg's own terminology was originally "mass comparison", which he later changed to "multilateral comparison"; from his first use of it in the 1950s on, it always involved comparison of grammatical formatives as well as of lexical items, along with considerable attention to typologically probable paths of sound change. See Greenberg's Genetic Linguistics (2005) for his true methodological positions.

In contrast to Greenberg, most historical linguists remain convinced that systematic phonological reconstruction is necessary to establish genetic relationship between languages, and consequently view the Eurasiatic hypothesis with skepticism.

Creole hypothesis

The phonological similarities and geographical proximity of Japanese to the Austronesian languages have led to the theory that Japanese may be a kind of creole language, with an Altaic superstratum and an Austronesian substratum, or vice versa. However, the lists of suggested Japanese-Austronesian cognates proposed by different scholars do not agree with each other^{[citation needed]}. Furthermore, the number of words identified as possibly Austronesian is extremely small.

Austronesian hypothesis

One of the less accepted theories is that Japanese is a purely Austronesian language. This is rejected by all mainstream specialists in both Austronesian and Japanese, since the grammar, lexis, and morphology of Japanese are vastly different from those of any known Austronesian language. Proponents of this theory point out examples of convergent lexis, such as Japanese hina "doll" and hime "princess," as cognate with the Māori word hine "girl," or Japanese kaku "to write, to sketch" with the Hawaiian kākau "to write, to tattoo", or Japanese ne "an expression" with the Kapampangan ne "an expression", or Japanese nomu "to drink" with the Tagalog inom "drink."

It is important, however, to note that many totally unrelated languages exhibit chance occurrences of convergent lexis. Furthermore, these alleged "cognates" soon fall apart upon closer analysis. For example, the Japanese word hime is clearly a compound word; modern Japanese /h/ comes from earlier /p/; Hawaiian /k/ comes from earlier /t/. Moreover, the time depths for Japanese and Proto-Polynesian do not match, and Polynesia is far more distant from Japan than Taiwan, the proposed Austronesian homeland. If there were an Austronesian connection, it might be found closer to the Japanese Archipelago. Beyond that, the time depth for Proto-Austronesian, at roughly 6000 years BP, makes it far too old to be compared with Japanese, which may have come to the Japanese islands perhaps 2500-3000 years ago (see the Yayoi page for details on the peopling of Japan). The biggest problem is methodological: the comparative method requires the establishment of regular sound correspondences, and no comparison of Japanese and Polynesian has yet reached that level of methodological adequacy.

Those who propose that Japanese and Austronesian are related suggest that the Austronesian family once covered the islands to the north of Formosa (western Japanese areas such as the Ryūkyū Islands and Kyūshū) as well as to the south. However, there is no biological or genetic evidence for an especially close relationship between speakers of Austronesian languages and speakers of Japonic languages, so if there was any prehistoric interaction between them, it is likely to have been one of simple cultural exchange without significant ethnic mixing. In fact, genetic analyses consistently show that the Ryukyuans between Taiwan and the main islands of Japan are genetically less similar to the Taiwanese aborigines than are the Japanese, which suggests that if there was any interaction between proto-Austronesian and proto-Japonic, it occurred on the mainland prior to the extinction of Austronesian languages on mainland China and the introduction of Japonic to Japan, and not in the Ryukyus. It should also be noted that groups of peoples that inhabited ancient Japan, such as the Kumaso and the Hayato, are hypothesized by some to have spoken an Austronesian-related language, although due to a lack of evidence, this hypothesis may be untestable.

Dravidian hypothesis

A more rarely encountered hypothesis is that Japanese is related to the Dravidian languages, mostly spoken in South India, of which the best-documented from an early date is Tamil.

The possibility that Japanese might be related to Dravidian was raised by Robert Caldwell, the great pioneer of Dravidian studies, in the 19th century (cf. Caldwell 1875:413). A relationship between Japanese and Dravidian has more recently been advocated by the Japanese scholars Susumu Shiba, Akira Fujiwara, and Susumu Ōno (n.d., 2000).

Evidence for this theory is that the Japanese and Dravidian languages are both agglutinating and have similar vocabularies and sound systems. ^{[citation needed]}

An objection that will occur to many people is that the retroflex consonants of Dravidian have no parallel in Japanese. However, "The cacuminals [i.e. retroflex phonemes] of Dravidian are of secondary origin, as appears from the fact that they never occur as initials" (Trombetti 1922:76). This objection, at least, is not fatal to the Japanese-Dravidian hypothesis.

This hypothesis has been criticized by other scholars^{[citation needed]} of Japanese and Dravidian. Ōno was criticized for making errors in history and archaeology, and for various methodological errors in applying the comparative method, such as positing multiple correspondences without giving conditioning factors (for example, Tamil c : Japanese s; Tamil c : Japanese ∅; and Tamil ∅ : Japanese s), and several other shortcomings in data and application of theory.

Possible reconciliations

Not all of the hypotheses enunciated above are mutually incompatible. Very often, they involve different time depths. The more accepted relationships tend to be those more shallow in time, while hypotheses concerning deeper relationships tend to be more controversial at present. Further, some of these hypotheses may involve correct perceptions of relationship, but not resolve the question of whether the similarities noted are due to genetic relationship or areal influence. For example, the following grouping would reconcile all of those above: Japonic-Goguryeo < Goguryeo-Silla < Altaic < Eurasiatic < Nostratic (which includes Dravidian), with an areal influence on Japonic from Austronesian. While only a minority of linguists concerned would endorse such a grouping, it underlines that the difficulties involved in classifying Japanese concern general methodological issues for linguistics, and will probably only be resolved in tandem with these.

Bibliography

Works cited

Aston, William George. 1879. "A comparative study of the Japanese and Korean languages." Journal of the Royal Asiatic Society of Britain and Ireland, New Series 11, 317-364.
Beckwith, Christopher I. 2004. Koguryo: The Language of Japan's Continental Relatives: An Introduction to the Historical-Comparative Study of the Japanese-Koguryoic Languages. Leiden: Brill.
Beckwith, Christopher I. 2006a. "Methodological observations on some recent studies of the early ethnolinguistic history of Korea and vicinity." Altai Hakpo 16, 199-234.
Beckwith, Christopher I. 2006b. "The ethnolinguistic history of the early Korean peninsula region: Japanese-Koguryoic and other languages in the Koguryo, Paekche, and Silla kingdoms." (page 33 ff.) Journal of Inner and East Asian Studies 2.2, 34-64.
Caldwell, Robert. 1875. A Comparative Grammar of the Dravidian or South-Indian Family of Languages, second edition. London: Trübner.
Georg, Stefan, Peter A. Michalove, Alexis Manaster Ramer, and Paul J. Sidwell. 1999. "Telling general linguists about Altaic." Journal of Linguistics 35, 65-98. Cambridge: Cambridge University Press.
Greenberg, Joseph H. 2000–2002. Indo-European and Its Closest Relatives: The Eurasiatic Language Family, 2 volumes. Stanford: Stanford University Press.
Greenberg, Joseph H. 2005. Genetic Linguistics: Essays on Theory and Method, edited by William Croft. Oxford: Oxford University Press.
Kanazawa, Shōsaburō. 1910. The Common Origin of the Japanese and Korean Languages. Tokyo: Sanseidō.
Martin, Samuel E. 1966. "Lexical evidence relating Korean to Japanese." Language 12.2, 185-251.
Martin, Samuel E. 1990. "Morphological clues to the relationships of Japanese and Korean." In Linguistic Change and Reconstruction Methodology, edited by Philip Baldi. Berlin: de Gruyter.
Miller, Roy Andrew. 1971. Japanese and the Other Altaic Languages. Chicago: University of Chicago Press.
Ōno, Susumu. n.d. "The genealogy of the Japanese language: Tamil and Japanese."
Ōno, Susumu. 2000. 日本語の形成. 岩波書店. ISBN 4000017586.
Poppe, Nicholas. 1965. Introduction to Altaic Linguistics. Wiesbaden: Otto Harrassowitz.
Riley, Barbara E. 2003. Aspects of the Genetic Relationship of the Korean and Japanese Languages. PhD thesis, University of Hawaii.
Starostin, Sergei A. 1991. Altajskaja problema i proisxoždenie japonskogo jazyka, 'The Altaic Problem and the Origin of the Japanese Language'. Moscow: Nauka.
Starostin, Sergei A., Anna V. Dybo, and Oleg A. Mudrak. 2003. Etymological Dictionary of the Altaic Languages, 3 volumes. Leiden: Brill. (Also: database version.)
Trombetti, Alfredo. 1922-1923. Elementi di glottologia, 2 volumes. Bologna: Nicola Zanichelli.
Vovin, Alexander. 2003. 日本語系統論の現在：これからどこへ 'The genetic relationship of the Japanese language: Where do we go from here?'. In 日本語系統論の現在 'Perspectives on the Origins of the Japanese Language', edited by Alexander Vovin and Toshiki Osada. Kyoto: International Center for Japanese Studies. ISSN 1346-6585.
Whitman, John Bradford. 1985. The Phonological Basis for the Comparison of Japanese and Korean. PhD thesis, Harvard University.