Proving the genetic relationship between the Japanese and Korean languages

C. Cancagliani  Claudia A. Ciancaglini, How to prove genetic relationships among languages: the cases of Japanese and Korean p. 289 RIVISTA DEGLI STUDI ORIENTALI NUOVA SERIE VOLUME LXXXI Fasc. 1-4 (2008)

 

Different hypotheses on the formation of Japanese and its relationship with Korean We have already said that, as far as similarities among languages are concerned, at least in principle contact is not an alternative explanation to a genetic one (see Section 5 above). For example, the so-called “southern theory” holds that only a relevant lexical stock of Austronesian origin entered into Japanese, whereas Japanese morphology (considered of Altaic origin) remained unchanged. In these terms, this is only a variant of the Altaic theory; only structural features are involved in the individuation of genetic links and, consequently, if we claim that Japanese morphology is Altaic, we are also claiming that Japanese is genetically an Altaic language. To compare this case with another example, from the genetic point of view English is indisputably a Germanic language, notwithstanding the enormous lexical stock of borrowings from French and Latin. A more confused formulation of the same theory affirms that Japanese would involve an Austronesian substratum and an Altaic superstratum (e.g.Murayama 1984; cf. Shibatani 1990: 108). The vagueness arises from the nontechnical use of the terms “substratum” and “superstratum”, which are obsolete and polysemic labels. Traditionally, “substratum” indicates a language that has been substituted by another one, but whose influence on the new language appears in vocabulary items and in some minor phonological details. In modern terms, this situation is called interference through language shift in normal transmission (Thomason and Kaufman 1988: 111). The term “superstratum” refers to the stock of lexical borrowings that a language has taken from the language spoken by a dominant population; superstratum phenomena do not attain phonology or morphology. For example, it is generally admitted that in Gaul the Celtic language exerted a substratum influence on Latin and on the derived Romance varieties (French, Occitan or Provencal, several north-western Italian dialects etc.), consisting in a number of Celtic place names and nouns, in the vigesimal system of numeration (cf. French quatre-vingt ‘eighty’), and in some phonetic tendencies, like the outcome [y] from Latin u (cf. Lat. luna ‘moon’ > French lune [’lynŒ]) or the outcome -it- from the Latin internal cluster -ct- (see, for example, Lat. lactem ‘milk’ > French lait). As regards superstratum, Italian, for example, underwent the influence of the Germanic and Arabic superstrata, consisting in groups of foreign words borrowed during their periods of domination (Ostrogothic and Longobard dominations, 494-555 AD and 568-774 AD respectively; the Arabic domination in South Italy lasted more than two centuries, from 827 onwards). What is even more important is that the obsolete terms “substratum” and “superstratum” do not involve changes in genetic identity: French, notwithstanding its Celtic substratum, is a Romance language; Italian, notwithstanding the influence of the Germanic and Arabic superstrata, genetically remains a Romance language, not a Germanic or a Semitic one. So, to return to our topic, the theory that Japanese involves an Austronesian substratum and an Altaic superstratum says nothing about Japanese genealogy, namely the linguistic layer lying between the two strata, but the scholars holding this theory do not seem well aware of this fact. I believe that we have here a great confusion about what qualifies as linguistic genealogy and what does not, a confusion which is also induced by the use of obsolete terms and concepts. We need to add that historical linguists generally tend to view with suspicion any substratum explanation of language change, especially because substratum supporters have frequently disregarded the methodological prerequisites needed to advance similar hypotheses, the first of which is that we need to know the structures of the two languages in contact before the language shift, namely the structures of both the abandoned and the target language. For instance, as far as the Celtic substratum is concerned, already Bloomfield (1933: 386) dismissed as “mystical” all the explanations involving this substratum influence on changes in Romance or Germanic languages, especially because these changes occurred hundreds of years after all the speakers of Continental Celtic had shifted to Vulgar Latin or incipient Romance, or to Germanic (see Thomason and Kaufman 1988: 112); moreover, what we know about Continental Celtic phonology is very little. There are, however, contact hypotheses about Korean and Japanese that are (or seem to be) alternative to a genetic one. According to the most up-to-date view (Thomason and Kaufman 1988), the main consequences of language contact are of three kinds: a) borrowing, when a language changes by incorporating foreign features but is itself maintained – typically, borrowing interference begins with lexical loanwords: the first foreign elements to enter the borrowing language are words; if the linguistic contact is deep and prolonged, eventually structural features may be borrowed, such as phonological and syntactic ones, more rarely morphological ones; b) substratum interference, resulting from imperfect learning during a process of language shift35 – substratum interference does not begin with lexical loanwords, but with phonological and syntactical elements, and may include morphological items: typically, in fact, the target language borrows but few words from the shifting group’s original language; c) creolization, pidginization etc. Neither borrowing nor substratum interference affect the genetic identity of the languages involved, though substratum interference may be so heavy as to disrupt the original typology of a language. Pidginization and creolization, on the other hand, are instances of shift without normal transmission or, in other words, of non-genetic language transmission.36 For example, Polivanov’s theory (1924) seems to be connected with the abovementioned third type of language contact: he claims that Japanese is an Austronesian-Altaic hybrid or mixed language. However, we need to notice that also Polivanov is not particularly clear about what constitutes linguistic genealogy and what does not. In fact, he first claims that Japanese is genetically linked with Austronesian languages and that a part of its linguistic features is inherited from a common ancestor with the Austronesian languages. On the same page (Polivanov 1924: 151), however, he adds that Japanese shows great differences from the Austronesian languages, because it is a language of hybrid origin, an amalgam of insular southern Austronesian elements and of western elements from the mainland, common to Korean and to other Altaic languages of Asia. In a long footnote, Polivanov (1924: 152-153, n. 7; see Shibatani 1990: 103) lists a number of “external similarities” shared by Japanese and Austronesian languages, some of which are very interesting, as they are marked features which are difficult to attribute to mere chance. We may ask, at this point, if the obsolete terminology involving terms like substratum, superstratum and even mixed language, in reality hide some more up to date concept, for example some kind of non-genetic language transmission. In fact, Shibatani (1990: 107), referring to the fact that Murayama (1973) agrees with Polivanov’s hybrid or mixed-language hypothesis about the formation of Japanese, adds that Murayama’s use of the term “substratum” is not consistent with his opinion: “Though Murayama keeps using the term Austronesian substratum, his view of a mixed language is more than having a large number of foreign words integrated into another language as a substratum or a loan-word component – a situation exceedingly common. What Murayama has in mind is a language whose morphology involves elements deriving from two (or more) different languages – a kind of language that Meillet (1925) declares not to have been found”. Meillet’s opinion is explainable if we consider that Meillet was referring to Indo-European languages and their peculiar morphology, especially involving bound morphemes (case endings, verbal endings, apophonic vowels etc.). Nonetheless, if we look at the various languages of the world and if we take into consideration all kinds of morphological systems, including those in which grammatical relations are expressed by means of unbound morphemes, word order, agreement, dependence, prosodic features etc., we may affirm that the cases in which a language shows a morphology involving elements deriving from two (or more) different languages are not rare at all. The question of the possibility of morphological borrowing in situations of language contact has been widely debated, and only in recent years have scholars interested in language contact phenomena, starting from Weinreich (1953), become inclined to accept that it is a plausible scenario.37 In recent years Thomason and Kaufman (1988: 35) have underlined that “it is the sociolinguistic history of the speakers, and not the structure of their language, that is the primary determinant of the linguistic outcome of a language contact […]. Linguistic interference is conditioned in the first instance by social factors, not linguistic ones”. Nevertheless, Thomason and Kaufman (1988: 37 ff.) warn that, in the case of the normal genetic transmission of a language, morphological borrowing is not exceedingly common; it occurs only in particular socio-cultural conditions and is submitted to a number of structural constraints, many of which have already been singled out by Sapir, Jakobson, Weinreich and others.38 Turning back to Murayama (1973 and 1984; cf. Shibatani 1990: 107), his hypothesis seems to offer a real alternative to a genetic one, given that he holds that Japanese verbal roots are both of Altaic and Austronesian origin, whereas the inflectional endings are of Altaic origin (here Altaic is used in the sense of Tungusic and Korean) and are added to verbal roots of both Altaic and Austronesian origin.39 Murayama also cites as an instance of a language showing a similar mixed morphology the Aleut-Russian hybrid language spoken on Mednyj (or Copper) Island in the Bering Sea off the Kamchatka peninsula. The Aleut language of this island, as a result of the influence exerted by Russian speaking hunters (many of whom married Aleut women), replaced its inflectional patterns in the finite verbal conjugation with Russian ones, whereas its noun and non-finite verb morphology did not change.40 The Mednyj-Aleut example is at the same time ambiguous and well chosen. In fact, according to specialists in language contact, it is a very extreme case. On the one hand, in Mednyj-Aleut the mechanism of interference clearly represents borrowing that arose in a context of deep and mutual bilingualism, and not of imperfect learning in language shift, because Aleut language was maintained, notwithstanding the dramatic change in its finite verbal inflectional patterns. On the other hand, the massive grammatical replacement that took place in this morphological subsystem seems to suggest rather the non-genetic development of Mednyj-Aleut.41 To conclude this section, we would like to emphasize again that the great majority of contact phenomena – which surely occurred in the course of time between Korean and Japanese, given their geographical and cultural proximity – does not affect the genetic identity of the languages involved and do not result in non-genetic language transmission or language shift. Only in the case of non-genetic language transmission are we dealing with hypotheses alternative to genetic ones. Extreme cases such as Mednyj Aleut, where no sharp boundary between borrowing and non-genetic development is discernible, are rare. The hypothesis advanced by some scholars (for example Murayama and Polivanov) that Japanese is a “mixed language” or a descendant of an ancient creole is difficult to prove, because – as underlined by Trask (1996a: 319) – in its earliest records Japanese does not seem to be a creole languages any more than does its modern outcome. This implies that the hypothesised creole has to be located in a time so remote as to render almost impossible a reliable historical and linguistic demonstration of the thesis itself. The same can be said of similar hypotheses about Korean. Nevertheless, Korean and Japanese (and perhaps some of the “Altaic” languages) share a number of structural features. I believe that it may be worthwhile to attempt an explanation in terms of linguistic area. In prehistoric times, the languages in question, whatever their genetic origin may have been, structurally converged with the lapse of time in a linguistic area or Sprachbund (to use the term first employed by Trubeckoj to describe phenomena such as the one that occurred in the Balkan area). There, genetically unrelated neighbouring languages (namely, Bulgarian, Macedonian, Romanian, Greek, Albanian, Serbian, Turkish), in contact for centuries, developed a number of shared morphosyntactic features (but not phonological ones), regardless of their respective genetic affiliation. This hypothesis can be linguistically supported only by internal reconstruction; in other words, it will be necessary to recover the chronology, or at least the relative chronology, of the structural changes that produced the shared features. In technical terms, one would need to know the relative chronology between two main kinds of isoglosses, namely unitary and differential (but not monoglottic) isoglosses.42 Given a group of compared languages, once we have specified a number of marked isoglosses, we call unitary isoglosses the structural features shared by all the compared languages and differential isoglosses the structural features shared by two or more languages of the group, but not by all of them. The monoglottic differential changes are not informative, because it cannot be proved that they are not individual innovations. The Sprachbund hypothesis would be strengthened only if the differential isoglosses can be proved to be older than the unitary ones or, loosely speaking, if diachronically the languages became more similar than they previously were. 7. Conclusions As we have seen, the principal genetic theories advanced about Korean and Japanese, among which we especially have the theories connecting Japanese to Korean or both to the Altaic family, are to be dismissed as not scientifically proved. No scholar, not even Starostin, has been able to prove this theory in a scientific way. The conclusion that Japanese and Korean are isolated languages from the genetic point of view and that all the attempts of scholars to demonstrate a genetic link between Japanese and Korean, or between these two languages and the Altaic group, have been unsuccessful does not depend (only) on the inability of scholars to correctly apply the comparative method. In fact, the failure rather depends on many historical and structural reasons, many of which we have already mentioned above and that we may summarize as follows: – Korean and Japanese are attested too late and in a logographic writing system; – the presumed date of their split from Proto-Altaic is too old to allow reconstruction; – the impossibility to prove cognacy in lists of compared lexemes casts doubt on any phonological correspondence deduced from these cognate sets; this impossibility may be due, at least in part, to the lexical typology of both Japanese and Korean; – the impossibility of finding paradigmatic correspondences in closed sets of grammatical morphemes casts doubt on any genetic hypothesis concerning Japanese and Korean. Under these conditions, the comparative-reconstructive method cannot be applied; as a consequence, from the genetic point of view, both Japanese and Korean cannot be considered anything but isolated languages.

Footnotes:

34 It is a well known fact that both related and unrelated languages may be in contact; on this topic see also Section 6 below

37 As regards some Syriac verbal structures surely borrowed from Persian cf. Ciancaglini (2006); on morphological borrowing in general see Ead. (2008: Section 8)

38 The socio-cultural conditions in which morphological borrowing is possible are as follows: the contact between the two languages must be deep and prolonged in time; the source language must exert a strong cultural pressure on the other one, playing the role of the prestige language; finally, there must be a large number of bilinguals. The main structural constraints are as follows: the morphological borrowing is always preceded by a great number of lexical borrowings; it is not an isolated phenomenon but belongs to a group of morphological loans and calques; the structures of the two languages must be at least partially congruent; a morphological borrowing that produces a simpler and unmarked grammatical structure is more probable; the borrowing language must already have a function corresponding to the borrowed element, especially when a bound morpheme is involved. See Weinreich (1953: 33 ff.); Thomason and Kaufman (1988: 18). 39 See also Murayama (1984). 40 The study to which Murayama (1973) refers is Menovš©ikov (1969); see now Golovko (1994), and in general Thomason and Kaufman (1988: 233-238). 41 See Thomason and Kaufman (1988: 233-238 and 107-108).

42 This argument is usually employed by genetic linguists for establishing whether a group of compared languages is genetically related or whether they converged in the course of time; it also serves to individuate subgroups of languages within a genetic family; see Lazzeroni (1997: 103 ff.).

…

Aspects of the genetic relationship of the Korean and Japanese languages / by Barbara E. Riley

I offer evidence from a variety of fields in order to strengthen the hypothesis that Japonic and Korean are linguistically genetically related to one another. Non-linguistic evidence supports the hypothesis that the Japonic language was introduced into the Japanese Archipelago approximately 2,500 years ago over a thousand year period, where a culturally and technologically advanced group began migrating into the Japanese Archipelago from the Korean Peninsula through Northern Kyushu. A constant and steady influx of Continental culture, language, and people, resulted in the near-complete extinction of the original language. The linguistic evidence comes from Middle Korean texts, written in the Silla-descended language of the 15th century-the kingdom that overwhelmed the Puyo, Koguryo, and Paekche territory and languages, thought to be more closely related to Japonic-and 8th century Old Japanese texts. I hypothesize that there were two “thalossocracies”: one with lzumo and Silla, and the second with Yamato and Paekche/Kaya Japonic elements were incorporated into the Silla language when Silla folded Kaya and Paekche into the new kingdom. In the same way, Yamato incorporated Silla-type elements into itself when Yamato overtook Izumo. I introduce evidence that supports Serafim’s Labiovelar hypothesis; i.e. MK k : OJ p, reconstructing PKJ *kw1. I also found a “reverse” correspondence set: that is, MKp : OJ k, for which I reconstruct *kw2. I hypothesize that this reverse correspondence is due to dialect borrowing. When Silla conquered the Korean Peninsula, it incorporated into itself Kaya, Paekche, and Koguryo, which were closer in genetic relationship to Japonic, and therefore would have (*kw > ) p. As these three languages were overcome, dialect borrowing likely occurred, which means that words with p instead of (*kw > ) k were borrowed into Silla, sometimes replacing and sometimes forming doublets with words retaining k. The second posited case of dialect borrowing occurred when Yamato overtook lzumo; since Silla had close contact with lzumo, words with (*kw > ) k were borrowed into Yamato, replacing, and sometimes forming doublets with, some words with p. Further research will surely lead to more understanding of the measurable effects of dialect borrowing and Proto-Koreo-Japonic.

[Lee Wha Rang, Who are the Japanese?: “Riley employed well established linguistic methods to correlate Japano and Proto-Korean. She went back into a period of time far beyond historical documents to dig out the prehistoric languages. For example, many borrowed terms can be eliminated by using non-cultural vocabulary such as body parts, natural objects, plants, animals, pronouns, and lower numerals. In contrast, technological vocabulary such as words for horse trappings and farm instruments are most likey borrowed from an alien language.Riley concludes “.. evidence from a variety of fields in order to strengthen the hypothesis that Japonic and Korean are linguistically genetically related to one another. Non-linguistic evidence supports the hypothesis that the Japonic language was introduced into the Japanese Archipelago approximately 2,500 years ago over a thousand year period, where a culturally and technologically advanced group began migrating into the Japanese Archipelago from the Korean Peninsula through Northern Kyushu. A constant and steady influx of Continental culture, language, and people, resulted in the near-complete extinction of the original language.”

Riley examined 8th Century Japanese texts and 15th Century Korean texts, expanding upon the work of Samuel Martin who published in 1966 the first comprehensive reconstruction of Proto-Koreo-Japonic, including sound correspondences and a list of reconstructed proto-segments. Silla was founded by the Saro tribe of Chinhan (a substate of Go-Chosun) and the Saro dialect became the official language of Silla (the Middle Korean language), which is extant now but can be reconstructed from place names and surnames in the archives of Silla. The number of cognates found for the Middle Korean and the 8th century Nara dialects is significant

Other readings:

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. (and other articles in thisissue)

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

Lewin, Bruno. 1976. “Japanese and Korean: The problems and history of a linguistic comparison.” Journal of Japanese Studies 2.2, 389-412

Robbeets, Martine. 2005. Is Japanese related to Korean, Tungusic, Mongolic and Turkic? Wiesbaden: Otto Harrassowitz

Kanazawa, Shōsaburō. 1910. The Common Origin of the Japanese and Korean Languages. Tokyo: Sanseidō

Robbeets, Martine. 2004a. “Belief or argument? The classification of the Japanese language.” Eurasia Newsletter8. Graduate School of Letters, Kyoto University.

See also Martine Robbeets, The historical comparison of Japanese, Korean and the Trans-Eurasian languages 261

Tranter, David N (ed.) The Languages of Japan and Korea. London: Routledge 2 The relationship between Japanese and Korean by John Whitman

Starostin, Sergei A., Anna V. Dybo, and Oleg A. Mudrak. 2003. Etymological Dictionary of the Altaic Languages, 3 volumes. Leiden: Brill. (Also: database version.)

Robbeets, Martine Irma (2005): Is Japanese related to Korean, Tungusic, Mongolic and Turkic? Wiesbaden: Otto Harrassowitz

Martin, Samuel E. 1966. “Lexical evidence relating Korean to Japanese.” Language 12.2, 185-251

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: Bril

The genealogy of a language: Tamil and Japanese by Susumu Ohno

Benedict, Paul K. 1990. Japanese/Austro-Tai. Ann Arbor: Karoma.

Japanese and Austronesian: An archaeological prospect

Japanese, Austronesian and Altaic A study of possible connections by Gunnar Holmberg