THE GENEALOGY OF THE JAPANESE LANGUAGE

- Tamil and Japanese -

Susumu OHNO

Introduction

In search of languages genetically related to Japanese, linguists over the last one hundred years have compared Japanese with almost every other language in the world-not only those of neighbouring peoples such as Ainu, Korean and Indonesian, but even Greek; yet none of these efforts have succeeded in establishing any kind of kinship.

It was more than ten years ago that interest in the Dravidian languages of South Indian began to spread among some Japanese researchers. Similarities between Japanese and Dravidian had been first pointed out in the mid-nineteenth century. In his major work, A Co~nparative Grammar of the Dravidian or South-Indian Family of Language, the English missionary R. Caidwell, cites resemblances and discusses the connection between the two languages.

The Japanese-Dravidian connection was studied in Japan for the first time by Susumu Shiba, who approached the subject from the point of view of religion. His findings were presented in “Kodai ni okeru Nihonjin no shikO” (Ways of Thinking of Ancient Japanese), which appeared in 1970 in the journal Jinbun ronsã (No.18, Kyoto Women’s University), and in a later study, “Dravida-go to Nihongo” (Dravidian Languages and Japanese), published in the same journal (No.22-23, 1973-74).

Comparative linguist Akira Fujiwara, began publishing the results of his research on Dravidian in 1974. In 1981 he put out a book entitled Nihongo wa doko kara kita ka (Whence the Japanese Language? Tokyo: Kodansha). His extensive comparisons of lexical items, comparing a number of words, including some particles and auxiliaries, were impressive. However, bececaue he took on the Dravidian family as a whole, his methodology was rather clumsy, and he failed to sufficiently demonstrate a kinship with Japanese. Another problem was that he did not take ancient Dravidian languages into consideration.

Prof. Minoru Go, who has been engaged mainly in research on Japanese genealogy, with a focus on Altaic for several decades, has also kept an eye trained on Dravidian, although he has not published anything on this subject. I got my start in this direction when he suggested that I study Telugu, one of the Dravidian languages. I became the fourth Japanese to undertake the genealogical study of Dravidian and Japanese.

Earlier, I had done comparative research on Korean, Ainu and other languages, and published Nihongo no kigen (The Origins of the Japanese Language) (Tokyo: Iwanami, 1957). This experience led me to decide to confine my research only to one branch of Dravidian, at first Telugu, and then to Tamil, especially classical Tamil. I chose Tamil for the following reasons:

First, it is a language spoken by a large group of over 48 million. Second, it is a very old language; 2,500 Cañkam verses, written in ancient Tam ii between 200 B.C. and 200 A.D., are preserved. Third, a detailed grammar of ancient Tamil survives, the Tolkãppiyam, written around the third century B.C. (Among other Dravidian languages, literature in the Kannada language can be traced back only as far as 1100 A.D., and that of Telugu, to 1200 A.D.,).

Fourth, while dictionaries in other Dravidian languages are small and simple, there is a large Tamil lexicon. Published in 1936, the Tamil Lexicon consists of seven volumes compiled over a period of twenty years by a special committee at the University of Madras. It contains 1,04,000 word entries, giving ancient usages, indicating dialects, and detailed definitions. An additional reference I have used is A Dravidian Etymological Dictionary, edited by T. Burrow and M.B. Emeneau, which came out in 1960. A revised and enlarged edition appeared in 1984 (hereafter abbreviated DEDR).

These are the tools that make possible the accurate comparison of Tarn ii and Japanese as far as the meaning of words and grammatical features are concerned. The geographic neighbors of Japanese suffer from a dearth of documentary sources going back to earliest times. The oldest extant documents in Korean were produced in the fifteenth century, and those in Monoglian in the thirteenth century. The Ainu language does not have a writing system. This paucity of documents recording the ancient forms of the languages in the vicinity of Japan has been a major stumbling block in the study of the genealogy of Japanese. Tamil is extremely important in that its very old forms are known to us.

I travelled to South India in 1980 to continue my research, receiving invaluable aid from Ms. Rama Lakshmi and Ms. V.N. Balambal. On New Year’s Day the following year, I showed Prof. Jaroslav Vacek of Charles University in Prague a list of the word correspondences I had collected for Tamil and Japanese. He kindly took time out of his busy schedule to check over the list with great care. For one year beginning in the fall of the same year, I studied the reading of classical Tamil at the University of Madras under Prof. Pon. Kothandaraman. During the winter break I visited the Trichi district, his home village, and was able to observe the old Tam ii New Year’s celebrations.

In March 1983, Prof. Arunasalam Sanmugadas, linguist at the University of Jaffna in Sri Lanka, and his wife Manonmani, came to Japan on a Japan Foundation grant, one of their purposes being to assist me in my research. They had grown interested in the Tam il-Japanese connection after hearing a lecture I gave at the 5th International Conference/Seminar on Tamil Studies held at Madurai in the Tamil state of India in

1981.

As guest researchers here at Gakushuin University, Mr. and Mrs. Sanmugadas studied classical Japanese literature and are now working on a translation of the Man ‘yoshu into Tamil. They have meanwhile continued to give me invaluable assistance in my study of the Tamil language. They themselves are Tamils, and have taught me much not only about their language but about Tam ii customs as well.

Note: (Loga) There will be some inaccuracies in the phonetic symbols. Please refer to the original for an accurate rendering

The Phonetic Systems of Japanese and Tamil

a. Vowels

The oldest writings preserved in Japan, which go back to the eighth century, tell us that (1) ancient Japanese had eight vowels, that (2) there was no distinction between long and short vowels, that (3) diphthongs were strictly avoided, (4) all syllables ended in a vowel, and that (5) the eight vowels were divided into two groups.

Group A: a, i, u, o:

Group B: e. e:, i:, o

The vowels in group A were found in 85 percent of all vowel usage, and those in group B in only 15 percent. Word roots and the initial parts of words used group A vowels, rarely those of group B. The same vowel in group A could be repeated with a consonant in between to form a word, such as kata (hard), kimi (millet), ko:to: (matter)and turu (crane). But this was never the case for the vowels in group B.

The vowels in group B are believed to have resulted from the merging of two vowels, as follows:

ia> e, ai> e: ui>i: o:i > i: ua > o

From all these, it can be hypothesized that the vowel system prior to the eighth century was made up of four vowels, or those in group A. I compare these four with Tamil vowels.

The old Tamil vowels were: a, a; i, i; u, u; e, e; and o, o: . By comparing these with Japanese vowels, I ascertained the followingcorrespondences

Japanese Tamil

a a,a:,o,o:

i,i,e,e:

u u u:

o: u,u:

b. Consonants

Neither Japanese nor Tamil have (1) clusters of consonants coming at the beginning of words, (2) double consonants in the middle of words, (although they occur rarely in Tamil words). Japanese has no cerebrals.

Japanese

consonants at head of word

k-, s-, t-, n-, F-, m-, y-, w­

consonants mid-word

-k-, -s-, -t-, -n-, -F-, -m-, -y-, -w­-

-r- ,-g- ,-z- ,-d- ,-b-

Tamil

consonants at head of word

k-, c-, t-, ii-, n-, p-, m-, y-, v­

consonats mid-word

-k-, -c-, -t-, -n-, -p-, -m-, -y-, -v-,

-t-, -n-, -r-, -l-, -L- -z- , -R- -G-

-n~c-, -nt-, -id-, -mp­-

Consonant correspondences are as follows:

word-initial

Jap./Tam. Jap./Tam. Jap./Tam.

1.k- :k- 1.-k-: -k-,-kk- 2. -“g- : -nk­

3. s- : c- 3. -s- : -c-, -cc- 4. -z- : -n~c­

Jap./Tam. Jap./Tam. Jap./Tam.

5. t- : t- 5. -t- : -t-, -U- 6. -nd- : -nt­

7. -t- : -t-,-tt- 8. -nd-: -nt­-

9.n- : n~ -,n- 9.-n-: n_, -nn_-

10.-n-: -N-, -NN-

11.F- : p- 11.-F- :-p-,-pp- 12. -mb-: -mp-

13. F- : v- 13. -F- : -v- 14.-~mb-

: -v-, -vv­

15. m- : m- 15. -m- : -m-, -mm-

16. y- : y- 16. -y- : -y-, -yy- 17. -r- : -r­

22.w-.: p- 18.-r-: -I­

23.w- : v- 23.-w- : -v- 19.-r- :

24. -s- : -t-,-tt- 20. -r- : -I­

25.s- :t- 25.-s- :-t-,-tt- 21.-r- :-r­

26. s- : zero

27. zero: c

28. ya: a_, a, e:, e

Note : Zero signifies no corresponding consonant.

Note: (Loga) There will be some inaccuracies in the phonetic symbols. Please refer to the original for an accurate rendering

Word Comparison

In comparing Japanese and Tamil words according to the rules of sound correspondence, comparison is confined to word roots or stems. In Japanese, verbs have the most, clearly defined stems. Saku (to bloom), the most common type, is a verb with five conjugations in the 8th century, as follows:

sak-a, sak-i, sak-u, sak-e:, sak-e

As this shows, the five forms of the verb saku share the same root, sak-, which expresses the basic concept of saku. This stem functions in actual usage, when it is followed by one of various suffixes, -a, -i, -u, -e: and -e linking it to the next word.

Nouns, too, may conjugate. Take the word kaze (wind) for example. When combined with another word to make a compound noun, kaze sometimes become kaza, such as kaza-Fana (“windflower”, meaning “snow or rain falling like flowers in an early winter wind”) and kaza-maturi (“wind festival”, or ritual for warding off storms). kaz- is the root of kaze.

The most basic unit of a word, whose further subdivision would deprive the word of its core meaning, is what we call the “word-root”. For Japanese verbs and nouns, the initial part of the word, made up of a consonant-vowel-consonant (CVC-) sequence, is the word-root.

Word-roots in Tamil, too, consist of a CVC- sequence, as is well known in the world of linguistics. In comparing Japanese and Tamil words, therefore, I focus on these CVC- roots, observing the phonemic rules strictly, and only then considering similarities in meaning.

My research has shown very close phonemic correspondences between Japanese and Tamil words, in a comparison of 400 pairs of words, but because of space limitation here, let me give a sample of the correspondences for Japanese “F” and Tamil “p” and “pp” below. Most of the Japanese samples are words in the ancient language, and the Tamil samples are those found in Cangkam verses. The entire list is given in Sound Correspondences between Tamil and Japanese (Tokyo: Gakushuin University, 1980) and in Nihongo izen (Before Japanese) (Tokyo: Iwanami Shoten, 1987).

Table 1. Word Correspondence (Jap. F: Tam. p., pp)

J. Far-u (to swell, expand)

T. Par-u (to swell {DEDR 3972])

J. Far-ara (to be broken off)

T. par-i (to be sundered [DEDR 3962])

J. Far-uka (to be far off)

T. par-a (to be far, wide [DEDR 3949])

J. Far-a (the ocean)

T. par-avai (sea (DEDR 3949])

J. Fat-akë (field for cultivation)

T. pa~-ukar (rice field [DEDR 3856])

J. Fat-u (to end, perish, die)

T. pat-u (to perish, die {DEDR 3852])

J. Fir-o (wide, great)

T. per-u (great, large [DEDR 4411])

J. Fo:k-u (to eulogize, praise)

T. puk-aJ (eulogy [DEDR 4235])

J. Fot-o (time)

T. pot-u (time [DEDR 4559])

J. Far-u (to become bulky)

T. par-u (to be bulky [DEDR 3972])

J. Far-e (to be diffused, as clouds, gas)

T. par-a (to be diffused, as clouds {(DEDR 3949])

J. Far-a (field of sky)

T. par-am (heaven {T.L.2499])

J. Far-aFu (to exorcise)

T. Par-avu (to exorcise [T.L.2503})

J. Fat-u (first, new of the season)

T. pat-u (to appear for the first time (DEDR 3852])

J. Fat-u (to stay [ship])

T. pat-u (to stay in a harbour [T.L.244])

J. Fin-a (rustic)

T. pin (rear place [DEDR 4205])

J. Fuk-asu (to smoke, steam)

T. pok-ai (to smoke, vapor [DEDR 4240])

J. För-ö (cloth cover)

T. por-vai (covering [DEDR 4590])

J. For-u (to desire)

T. pur-i (to desire [DEDR 4287])

J. Fut-a (cover, lid)

T. put-ai (to bury, hide, cover [DEDR 4509])

J. Fut-o (to be bulky)

T. pu~-ai (bulkiness, protuberance [DEDR 4253])

J. Fur-c (village)

T. pul-am (village, place [DEDR 4303])

J. Fun-c (ship)

T. puri-ai (raft, boat [DEDR 4321])

J. FOr-ö (tumor, abscess)

T. purr-u (scrofulous, scurby one [DEDR 4336]

J. taF-uru (to die)

T. tap-u (to perish, die (DEDR 3068])

J. öF-ö (big, to flourish)

T. upp-u (to become big, bloat [DEDR 666])

J. aF-u (to meet, be fit)

T. opp-u (to agree, be fit [DEDR 924])

J. köF-u (to beg)

T. kupp-u (to join hand as in worship [DEDR 1894])

J. tuF-a (spittle)

T. tupp-al (saliva [DEDR 3323])

J. suF-u (to suck)

T. cüpp-u (to suck, sip [DEDR 2621])

The sound “F-” shown above is pronounced “h-” today. It is widely accepted among Japanese linguists that this “F-” was ‘-p-‘ in prehistoric Japanese. “Japanese F-: Tamil p, pp”, therefore, is the same thing as “Japanese -p : Tamil p, pp”.

A comparison of kinship terms in the two languages also indicates a connection. Many kinship terms which are not mentioned in the oldest extant documents (8th century) and which have not been found in mainstream Japanese since then have been preserved in dialects spoken in the northeastern end of Honshu and the south western tip of the Japanese archipelago. The reason for this is still unknown, but it has been discovered that corresponding kinship terms existed systematically in old Tamil. It is, therefore, difficult to dismiss the similarities as accidental.

Table 2. Japanese and Tamil Kinship Terms

Japanese Dialects Old Tamil

Tohoku (Northeast) Ryukyus

Region

Father accha Iwate, Aomori chan (Amami, accan

(Shimokita) Kakeroma Is.)

acha Okinoerabu,

Yoron, Yaeyama Is.

acha, aja Aomori (Nishi- Kikai, Tokunoshima,

Tsugaru), Akita Okinoerabu, Yonaguni

(Hiraka) Is.

53

aya Aornori (Tsugaru). Ishigakijima ayyã

Iwate (Kokonoe)

tanda Akita, Iwate, tantai

Yamagata, Niigata

Mother ãya Aomori (Shimokita)

ayã Okinawa (Shun) ãyãl

aya Aomori (Shimokita),

Akita, Yamagata,

Niigata

aecha Aomori (Taugaru) asse; Amami accaJ

ata>ada Yamagata (Mogami) attal

appa Aomori (Tsugaru), Tanegashima,

Akita (Kazuno), Okinawa,

Iwate appa (grandmother):

Yaeyama, avvai

Iriomote

aba Aomori (Tsugaru),

Akita

amma Fukui, Ishikawa, Amami, Okinawa, ammai

Mie, Köchi Yaeyama, Iriomote,

amrnã Okinawa, Amami,

Yoron, Kume,

Yaeyama

Elder annyã Fukushima, Yama - anna

brother gata, Niigata,

Jshikwa

Elder anne Iwate, Fukushima, annai

sister Niigata, Ibaraki

Grammatical Correspondences

The following are some of my findings through typological comparison.

1. Nouns do not decline.

2.Subject is followed by predicate.

Examples

Tamil : veeniL pooyiRRu

Japanese : Haru sarinu. (Spring has gone.)

Tarnil : Katal peritu.

Japanese : Urni hiroshi. (The sea is vast.)

3. Adjective comes before noun.

Tamil : ven tiGkal

Japanese : siroki tuki (white moon)

Tamil : cern malar

Japanese: akaki hana (red flowers)

4.Adverb comes before verb..

Tamil : Mella nata.

Japanese : Yukkuri aruku. (Slowly walk. [Walk slowly.])

Tarnil : Enrum aruLal veeNtum.

Japanese : Tune-ni ataFu besi. (Always give should. [(You) should always give.])

5.Object comes before verb.

Tamil : Kallin naatpali uutti.

Japanese : Isi ni sasagemono o situ. (Stone on offerings put. [(I) put the offerings on the stone.])

6.There are no relative pronouns.

Tam ii : Avar irunta en nenjcu.

Japanese : Kare sumu waga kokoro. (He lives my heart. [My heart, in which he lives.])

7.Auxiliary verb comes after the verb and at the end of sentence.

Tamil : Enntuum pariyal veeNdaa.

Japanese: Sukosi mo doojoo subekarazu. (Not at all, sympathize should not. [(You) should not sympathize at all.])

8. Auxiliary forms follow a specific order. For example: 1. verb, 2. causative,

2. passive, 4. aspect, 5. negative, 6. tense and 7. interrogative, as in the following sentence.

1 2 3 4 5 6 7

Tamil : Nata-tta-ppat-tat-anr-um-kollo.

Japanese : Yuk-ase-rare-tara-zara-mu-ka. (Go make be have been not may? [Have I not been made to go?])

9. Particle comes after noun and verb.

Tamil Arul urn anpu urn aRan urn

Japanese: Megumi mo ai mo gimu o hatasu hito mo

(Favor too, love too, duty fulfill person too [(favor, love and persons who fulfill duties

Tamil Entai vantu uraittanan.

Japanese : Watasi no titi ga kite katatta. (My father came and said.)

10. Interrogative form has interrogative particle at the end of a sentence.

Tamil : Yaatu cevaan-kol.

Japanese : Nani suru ka. [What do you do?])

Tamil : oori kolloo, allan kolloo

Japanese : Ori ka hoka no hito ha.

(Ori? another person? [On, or another person?])

11. Unlike in Japanese, the personal suffix comes at the end of a Tamil verb, but this was not always the rule at a time when CaGkam poems were written, and never the case in the Malayalam language. These facts seem to show that the use of the personal suffix was a later development.

12. Japanese demonstrative pronouns - “ko” (indicating objects “near”), “so” (“middle”), “a” (“far”), and “idu” (“when, where”)-correspond to Tamil pronouns, “i”, “u”, “a” and “e”. The list below details the correspondences.

Japanese

near middle far indefinite

ko: so: ka(a) i

thing ko:-re so:-re ka-.re idu-re

(a-re)

place ko:-k: so:-ko: ka-siko id-uku

(a-soko)

direction ko-ti so-ti (a-ti) idu-ti

(ko-nata) (so-nata) (a-nata)

relation ko:-no: so:no: ka-no:

(a-no)

Tamil

near middle indefinite

u a e

thing i-tu u-tu a-tu e-tu

place i-Gku u-Gku a-hku e-Gku

direction i-Gke: u-Gke: a-Gke: e-Gkee:

i-vvitam u-vvitam a-vvitam e-vvitam

relation i-nta u-nta a-nta e-nta

Note

1. Because of consonant correspondence (26) and vowel correspondence (6), the Japanese ~‘so” corresponds to the Tamil “u”.

2. The pronoun “a” that began to appear in the Heian period (794-1192) may have been a result of sound shift from “ka”, but it may also be that “a” had been in use since much earlier and appeared for the first time in the Heian-period documents. If the latter is the case, the demonstrative pronoun indicating objects far was the same between old Tamil and Japanese.

3. Because of the vowel correspondence (4), the indefinite pronoun “i” in Japanese corresponds to “e” in Tamil.

3. From these, we can say that the stems of the “middle”, “far” and “indefinite” demonstrative pronouns were very similar between Tamil and Japanese.

Particle and Auxiliaries

Below is a list of particles and auxiliary verb correspondences between the two languages.

Table 3. Particle and Auxiliary Verb Correspondences

Japanese Tamil

Particles (postposition)

1. Case indicator particle

2.

i. Follows the noun to link it with tu atu

another noun. no in

ga aka, akam

ii. Follows the noun to link it with a verb ni in

to o~:u

3. Conjunctional [?] particle

4.

Follows the verb to link it with another verb.

te tu

3. Adverbial particle Fa vay

mo: urn

Denotes a preceding word to be the topical ka ku:, kol

about which something is said. Essential ya ya *ya:> e:. aa

for making a sentence. Not related to case.

Auxiliary verbs

1. Makes the verb transitive and causative. asu ttu

Makes the verb intransitive and passive. aru ar, ir

2. Perfect voice

Transitive verb tu tt

Intransitive verb nu nt

Conjunctional form an ir

3. Tense

Past k . . . .

Future mu urn

4. Other

Obligation, necessity be:si ve:Nd

(I have shown the examples of corresponding sounds and usages of these particles in “Nihongo Izen” (Before Japanese, Iwanarni 1987, pp.247-328).)

Some of the correspondences shown above may be difficult to accept. Some Tamil particles and auxiliaries begin with a vowel, but their Japanese counterparts do not. They are:

Tamil Japanese

atu tu

in no.

aka, akam ga

in_ ni

otu to:

um rno:

um mu

The reason for this may be explained as follows. Throughout the history of the Japanese language, the last syllable of a word invariably ends in a vowel. So, if a particle following it began with a vowel, a diphthong would have occurred. Diphthongs, however, were strictly avoided in ancient Japanese. When a vowel was combined with another, a consonant might be put between them, or one of the vowels dropped. The later was common. That is why almost no Japanese particles begin with a vowel. (The only exception is the particle “i”, but it may have been pronounced “yi”.)

In Tarnil, there are many words which end in a consonant. They can be easily followed by a particle that begins with a vowel. When a particle that begins with a vowel follows a word that ends in a vowel, “v” or “y” is often inserted between the vowels. Alternatively, either the vowel at the end of the preceding word or the vowel at the beginning of the following particle is dropped. For example:

Pu: in_ (of flower) pu:vin [“-v-” inserted]

Ce:mpu in (of plant) ce:mpin [“-u-” dropped]

If we assume that when a noun or a verb was followed by a particle the vowel at the beginning of that particle was always dropped, we can say that the Japanese particles correspond to Tarnil particles, taking a form that has dropped the initial vowel. The consonant/vowel correspondences between Japanese and Tamil in the particles and auxiliary verbs listed above are all supported by the consonant-vowel correspondences in the word roots of noun, adjective, and verb.

Critiques of the Ohno Hypothesis

The possibility of a genealogical relationship between Japanese and Tamil suggested by the data I have cited in part above came under vociferous attack in Japan in 1981-82. Critics against me claim that my findings are false and my research is riddled with errors. Among them is Muneo Tokunaga, one of the very few Japanese who understand the Tamil language. He wrote (my translation):

I have studied Prof. Ohno’s lists of correspondences, but believe they reveal his ignorance of the phonetic system peculiar to Tamil, lack of attention to Dravidian sound systems and word structure, and the phonemic changes unique to southern Dravidian languages, as well as the misuse of A Dravidian Etymological Dictionary (DED). He selected words arbitrarily from the DED, distorted their meanings, and misunderstood their English translations. His work disregards the achievements of Dravidian linguistics research conducted over the last century. As a Tamil specialist, I find absolutely no scholarly value in the Ohno theory. If Professor Ohno thinks my argument alone is not enough, I advise him to ask leading Dravidianists overseas for comments (most important of whom are Bh, Krishnamurti, Hyderabad; K.V. Zvelebil, Utrecht; and M. Andronov, Moscow. They should not include scholars in the Tamilnadu state, who are so eager for attention from overseas.)

(Bulletin of the Kokusai Gengo Kagaku Kenkyujo, Kyoto Industrial

University, 2-1, March 1981, p.⁹.)

With the exception of Tokunaga, most of my critics have little knowledge of the Tamil language and their knowledge of ancient Japanese is superficial. None of the native Tamil speakers who cooperated in my research have doubted any basic connection between Japanese and Tamil. Below I would like to mention three Western scholars who commented on my theory, two Europeans and an American. Their comments were made on the basis of my publications in English, which consist thus far of two books and two papers as follows:

Sound Correspondences between Tamil and Japanese (Gakushuin University, 1980)

A Study on the Relationship between Tamil and Japanese (I.J.D.L., Vol. XII, No.2, 1983).

“The Loss of Initial C in Tamil and S in Japanese” (Uyaröyvu, University of Madras, 1983)

“Worldview and Rituals among Japanese and Tamils” (Gakushuin University, 1985).

Kamil V. Zvelebil was the first person to give serious attention to my work an extended encouragement to me. He kindly sent me his books and papers on the Dravidian Languages, and also gave me much advice. In his essay “Tamil and Japanese- Are They Related? The Hypothesis of Susumu Ohno” (Bulletin of the school of Oriental and African Studies [B.S.O.A.S.], Univeristy of London, Vol. XLVIII, part 1, 1985 he says:

One general remark at the outset: a distinction must be made between evidence and proof Is there any valid evidence at all for a (genetic?) relationship between Tamil and Japanese? This question in my opinion, expressed with utmost caution, may be answered in the positive. On the other hand, if we ask about proof of such relationship, there is, so far, none. However, the evidence-in matters of deep grammar, lexicon, and probably even in phonology-is such that the positing of some kind of non-accidental connexion between Japanese and Tamil (Dravidian) is not intrinsically ruled out. It would be premature, sweepingly to dismiss such a hypothesis as impossible and fantastic.

Going into more detail, Zvelebil writes:

In Sound Correspondences Ohno also dealt briefly with phonology; but more importantly, he has discussed two problems pertaining to phonology in the two papers mentioned above. Although the paper on the loss of the initial affricate/sibilant in Tamil/Japanese (April, 1982) may be somewhat lacking in philological sophistication, the phenomenon itself is striking; we must not forget, however, that this tendency is in fact confined to South Dravidian and is strongest in Tamil-Malayalam. It points rather to parallel but unconnected developments in the two languages or groups of languages. Nevertheless, even a common tendency, though not a proof of genetic relationship and a ‘special connexion’, points to shared trend or direction in phonological development and should not be dismissed altogether, particularly in the light of other cumulative evidence. The August 1982 paper on intervocalic -p- is thought-provoking indeed; according to Ohno, intervocalic -p- actually did exist in old Tarnil, at least in a few relic forms, and it corresponds to Japanese -F- which developed from earlier -p-. This Japanese -F- is ‘voiceless’ and ‘bilabial’. Some Dravidianists (Emeneau, Krishnamurti) do not reconstruct *...p.. even for the proto stage but according to D.W. McAlpin, for example, it seems best to maintain it since the contrast helps separate -v- ( -~p-) from a possible *v. and from other shifts. In mymanual of comparative Dravidian phonology I discussed this problem at some length and tended rather to maintain an intervocalic *..p... Now Ohno cites a Japanese correspondence for the Tamil lapu ‘to kill’ (which he correctly locates in the old Tamil grammar Tolkãppiyam, aithogh he greatly antedates the work into the fifth century B.C), viz, taFu-. If we accept this correspondence, it would support our hypothesis of the reconstruction of a pre-Tamil *..p.. for Dravidian. According to Ohno whereas the contrast of -p-; -v- was lost in Tamil, it has been preserved in Jap. -F-; -b-.

Prof. Vacek, too, refers to my work in “The Dravido-Altaic Relationship” (Archly Orienta7nl2 VOl.55/1987 ACADEMIA PRAHA). He has some reservations about the semantic correspondences, expressed as follows:

On the whole, Ohno’s work is an interesting attempt which will obviously be subjected to further revisions, but it seems that the sum total of the sound correspondences makes their accidental appearance impossible. Some of his etymologies could also be enlarged by Mongolian parallels,....

But he also says:

Personally we consider as most persuasive such etymologies in which the relation is direct-verb to verb, noun to noun -with a relatively exact semantic agreement. Etymologies in which in one language we have a verb and iii the other a noun are possible, it is true, but at this stage of research into this subject they are less persuasive....

Prof. Roy Andrew Miller of Washington University, in the United States, attacked Zvelebil’s cool appraisal in a severely critical essay, “Tamil and Japanese?” (B.S.O.A.S., Vol. XLIX, part 3, 1986), dismissing my findings altogether. He says, “Ohno has studded his 1980 book, and indeed all his books and papers, with hundreds of alleged Japanese linguistic forms that are entirely imaginary, ‘words’ that are attested for no known stage of the language, ‘words’ that can neither be cited nor documented- forms that are, most simply put, lexical ghosts”.

I wrote a response to this salvo and sent it to the editor of the B.S.O.A.S., This respected journal, however, apparently does not wish to follow through the debate. Although I feel that pursuing the debate is not nearly as important as getting on with my research and that it will take a very long time before this theory can be sufficiently tested, I wish to defend myself on several accounts.

1. Miller claims that I have encountered criticism in Japan for my findings presented in Nihongo to Tamirugo [Japanese and Tamil] (Tokyo: Shinchosha, 1980). I was frankly astonished by much of the hubbub at that time for I realized that my critics were for the most part neither scholars of old Japanese nor of Tamil. As they are unfamiliar with the ancient literature, they could not comprehend the forms and meanings of the words I was citing. At the time, no university in the country was actively studying the possibility of a Tamil-Japanese connection, but as a result of my probing the issue, the University of Tokyo as well as Gakushiin University where I teach, opened up an inquiry into the matter. There was suddenly a surge of media coverage of the subject, and I found myself, although at the beginning of my research, in the center of a nationwide debate. I am still avidly pursuing my research into the matter, and a verdict has not yet been reached concerning the extent of the correspondence between Tam ii and Japanese.

2. Since January 1983 I have been presenting my findings in Kaishaku to kansho: (Interpretation and Appreciation), one of the leading scholarly journals in the field. I have so far presented over 300 words in each language, and precisely explained the form, meaning and source of each correspondence. Some of them may be off the mark - scholars of the future will sort out the facts and build on my findings to make further revelations - but Miller’s claim that out of 12 words he considered from my list only 2 (17%) are possibly accurate is a gross miscalculation that harms the credibility of my research. Below I will answer his charges directly. As each of the examples with their technicalities takes up a lot of space to explain, I will here limit myself to four of the words.

i) Although Miller claims that “Jap. Kaer- <KaFer- is properly ‘to turn over, go upside down, not capsize’ “, there is clear evidence that it does mean ‘to capsize’. One assumes that he is familiar, as is any student of Japanese literature, with the eighth century poetry anthology, Man ‘yoshu, in which appears the following example:

Ofobune wo! kogino susumi ni/ifa ni fun! kaferaba kaferel imo ni yoritefa (#557). (If I cannot marry her, my big ship which I row along may capsize by running against a great rock).

As I did not write the words “ship” and “row” for the above poem, I hope Miller will concede that my presenting KaFer-’s meaning as “capsize” is not “juggl(ing) the evidence to make it appear to be much better than it is”. I can understand, however, that as he is probably only familiar with modern Japanese, this example, known by any graduate in Japanese literature, appears to Miller to be one of my “ghosts”. Miller also quotes the Vocabulario da lingoa de laparn corn a declaraqão em Portugues(Nagasaki, 1603), giving the impression that he uses the volume. Why did he then fail to note that in the supplement clearly appears the entry:

Cayeri, eru, etta.

Funega cayeru. (A ship capsizes) Virarse, ou emborcarse a embarcação. (p.

338 V).

ii) Miller states that “kara “stone” (is) a form totally unknown to any Japanese, or in Japanese dictionary”, and elsewhere that “kar “stone” does not, and never did, exist”. Allow me to contradict him with the evidence, taken from valid sources read by all in the field. The twelfth century A.D. dictionary, Myogisho (Tokyo: Kazama Shobo, 1962), gives 4 Chinese characters which mean “stone mortar”; Japanese kana characters (romanized below) are presented for each Chinese equivalent:

~ kara-usu (p.598) ~ kara-usu (p.603)

~ kara-usu (p.602) ~I11 kara-usu (p.607)

We can see that in all 4 characters above is included the radical for “stone”, “.~“. Moreover, in all standard and dialectal Japanese we have the following doublets, given in Zenkoku hogenjiten (Tokyo:Tokyodo, 1951):

dialectal standard meaning dialectal standard meaning

gani kani crab gabu kabu stump

gama kame tortoise gasu kasu dregs

gama kama pit, hole gara (kara) stones

According to the above example, the dialectal form “gara” corresponds to the standard form “kara” (p.204). “Gara” once again appears as a dialectal form meaning “stone” in the highly authoritative Nihon kokugo daijiten Vol.5 (Tokyo: Shogakukan, 1974), p. 176, and its usage is widespread, found on Sado Island, and in the Niigata, Mie, Kyoto, Nara and Toyama prefectures. One can see from the above how an untrained student in the field, ignorant of ”gara” and unaware of the dialectal correspondence between “k” and “g” might be led to mistakenly believe that I invented “kara�