tal Bookseller . Russell Street DON, W.C. I

^51 {,


Presented to the library of the


from the estate of


















%\ 24 /#





THE chief object of this work is to provide material for study of the affiliations of the Japanese language, and, in so far as philological evidence is of value, for inquiry into the origins of the Japanese race ; but it has been so planned as to be, I hope, of interest to students of general linguistic theory. I trust also that advanced students of Japanese, especially those who wish to read early and medieval texts, will find it useful as a work of reference ; and even those who are concerned only with the modern spoken and written languages will, I believe, find many of their difficulties removed by gaining some knowledge of the development of grammatical forms and the growth of common idioms.

The question of the racial origins of the people now inhabiting the Japanese archipelago has not yet been solved. Recently much attention has been paid to the Polynesian, as opposed to the 'Ural-Altaic' theory, but the philological arguments on both sides have as a rule been based on incom- plete data so far as concerns the vocabulary and grammatical structure of the Japanese language in its earliest known stages. In the following pages an attempt is made to remedy this deficiency, and I have purposely confined myself to a purely descriptive treatment, without conscious bias towards either theory, leaving it to comparative philologists to make use of the material supplied. It was my intention to furnish as an appendix an annotated vocabulary of Japanese in its earliest known forms, but the lists which I had compiled were, unfortunately, destroyed in the great earthquake of 1923. There exists, however, in the Transactions of the Asiatic Society of Japan (vol. xvi, pp. 225-85) a list compiled by Messrs. Chamberlain and Ueda which, I believe, requires but little revision in the light of recent research.

The chief sources used for the following study were the treatises of the great pre-Restoration grammarians such as Motoori and Mabuchi and their annotated texts of the earliest


records and anthologies ; the indispensable studies of Aston, Chamberlain, and Satow, those great pioneer scholars, to whom all Western students owe praise and thanks ; various modern text-books on Japanese grammar ; and compilations made under the auspices of the Department of Education, such as the complete analysis of the vocabulary and gram- matical structure of the Heike Monogatari, published in two volumes, of 1,000 pages each, in 1913.1

Of all these, I am most indebted to the works of Professor Yamada Koyii, whose great thesis on Japanese grammar ( 0 2fc ^ ffr) and studies of the language of the Nara, Heian, and Kamakura periods are amazing monuments of learning and industry.

The examples of Japanese given in the course of the work are taken, in the case of classical and medieval usages, from the best available texts, and in the case of modern usages from the Readers published by the Department of Education or from newspapers and other contemporary documents.

G. B. S. The British Embassy, Tokyo.

1 I regret that I have been unable to make use of recently dis- covered MSS. of the Heike Monogatari, which show that the work as usually known is refashioned from texts in an earlier language.

CONTENTS Preface ....



I. I. Introduction of Writing

2. Further development of the Script

and the

vn xi xv



representation of Japanese sounds

3. Later developments of the language, and divergence between spoken and written forms 51

II. The Substantive . . . .69

The Pronoun, 71. Demonstrative pronouns, 73. Interrogative pronouns, 74. Indefinite pronouns, 75. Historical development of pro- noun, 76. Possessive pronouns, 80. Relative Pronouns, 81. Numerals, 82. Number in the substantive, 85.

III. Predicative Words . . . .88

Verbs and adjectives, and their simple con- jugation . . . . .90

IV. The Adjective . . . . .98

Inflected adjectives, 98. Auxiliary adjectives, 109. Uninflected adjectives, 117.

V. The Verb ..... 126

I. Simple conjugation, 126 : Stem, 129. Pre- dicative form, 130. Attributive form, 133. Conjunctive form, 137. ' Imperfect ' or negative base form, 140. Perfect form, 142. Imperative, 145. Substantival forms in -ku, 147. Development of conjugations, 151. 3270 b


V. The Verb (continued) :

II. Compound Conjugation, 156 : Suffixes de- noting voice or aspect, 158. Suffixes form- ing causative verbs, 164. Suffixes denoting tense, &c, 173. Negative suffixes, 190. Un- infected verb suffixes, 196. Transitive and intransitive verbs, 199.

VI. The Auxiliary verbs aru and suru . . 202

Other auxiliary verbs . . .221

VII. The Particles ..... 223

Case particles, 224. Adverbial particles, 255. Conjunctive particles, 272. Exclamatory particles, 280.

VIII. The Adverb ....

IX. The Formation of Words X. Grammatical Functions

XI. Syntax ....

Appendix. Comparison of spoken and written forms


288 292 304 313 341



IN describing the development of the Japanese language it is convenient to divide it into stages corresponding to periods usually distinguished by Japanese historians ; and this method is particularly suitable because those periods coincide approximately with well-marked cultural phases.

The earliest period to furnish written records of the lan- guage is the Nara period, coinciding roughly with the eighth century a. d., when the Court was at Nara. Works now extant which may be assigned to that period are :

i. The Kojiki, or 'Record of Ancient Matters', completed in a. d. 713. A description of this chronicle, and some remarks on the evidential value of its text as reconstructed, will be found in Chapter I, pp. 15 et seq. Whatever doubts may be cast upon the reconstructed prose text, there is no doubt that the poems in the Kojiki are most valuable material. They represent the language of A. D. 700 at latest, and it is highly probable, since they bear every mark of antiquity, that they had already at that date been preserved by oral tradition for several centuries.

2. The Nihongi, or 'Chronicles of Japan', completed in a. d. 720. Only the poems and a few scattered sentences in this work are of value.

3. The Manyoshu, or 'Collection of a Myriad Leaves', an anthology of Japanese verse completed early in the ninth century a. d. , and containing some poems which go back at least as far as the late seventh century. Not all these poems are directly available as specimens of early forms of Japanese, since they are not all written phonetically ; but by collation with other poems in the same collection, and by reference back to the poems of the Kojiki and Nihongi, it is possible to reconstruct a great proportion of the native verse of the Nara period with a high degree of certainty.

4. The Shoku Nihongi, a continuation of the Nihongi, completed in 797. This work contains certain Imperial edicts in pure Japanese, and their texts can be restored with considerable accuracy. For translation and notes, see T.A.S.J.


5. The Engishiki, or ' Institutes of the Engi Period ', a code of ceremonial law promulgated in 927. This contains a num- ber of Shinto rituals, such as purifications and prayers for harvest, &c, which are evidently of great antiquity. There is strong internal evidence to show that these rituals belong to the Nara period at latest, and it is almost certain that they are among the oldest extant specimens of Japanese prose. For translation and notes, see Satow, T. A. S. J., vol. vii, of 1879.

In addition to the above there are certain family records (R 3fc) and topographical records (Jig, J^ 12) which contain fragmentary material, but altogether it amounts to very little. There is only one stone monument of the Nara period bearing an inscription in Japanese the so-called ' Footprint of Buddha ' (Bussokuseki) near Nara. All other inscriptions of that time are in Chinese. Unfortunately for philologists, so strong was the influence of Chinese learning in the eighth century that all the documents deposited by the Nara Court in the storehouse called the Shosoin, and marvellously pre- served until to-day, contain not more than a few dozen lines of Japanese.

It will be seen from the foregoing account that the material for a grammar and vocabulary of Japanese of the Nara period is scanty, and that the bulk of it is in the form of poetry. Indeed, it is not an exaggeration to say that our knowledge of the earliest forms of the language depends chiefly upon the Manyoshu.

Following the Nara period comes the Heian period, so called because the centre of government was now at Heian-jo, the modern Kyoto. In the three centuries and more (a. d. 800-1186) comprised by this period there is no lack of material (vide Chapter I, pp. 53 et seq.). To it belong several important anthologies of verse, such as the Kokinshu ; romances, such as the Genji Monogatari ; diaries and miscel- lanies, such as the Tosa Nikki and the Makura no Soshi ; and a number of historical works such as the Sandai Jitsu- roku. From these it is easy enough to fix with certainty the forms of written Japanese. What is difficult, however, is to trace, in its earlier stages, the divergence between the spoken and written languages. There is no doubt that it progressed during this period, for there are important differences be-


tween the language of the verse anthologies and the more serious historical works on the one hand, and the diaries, miscellanies, and romances on the other. But it is impos- sible, at least in the present state of our knowledge, to follow step by step the development of more than a few spoken forms. There are in the large mass of written material only occasional passages of undoubted dialogue or reported speech. Moreover, the general tendency of writers has always been to give a literary form to reported speech. This is particularly true of Far Eastern countries, where the written word is held in high respect, and where the system of writing in use is ill-adapted to phonetic recording. Thanks, how- ever, to the development during the Heian period of the kana syllabary, it is possible to discern some differences, which can safely be ascribed to changes in pronunciation. Thus when we find in, say, the Genji Monogatari words hitherto written yoki and utsukushiku appearing as yoi and utsukushiii, we may assume that the latter forms represented contemporary pronunciation ; and further, seeing that the older forms are preserved in verse and in other works of the same date, it is not unreasonable to suppose that the language of the Genji Monogatari was substantially the same as the cultivated speech current in its writer's day.

The Heian period was succeeded by the Kamakura period (i 186-1332), during which the country was controlled by a military autocracy. Here again there is ample material for the study of written forms, but very little exact evidence as to the development of the spoken language. All we can say is that while the Court at Kyoto remained the centre of the ancient culture, the military aristocracy and its adherents developed in another part of the country on other and less conservative lines, and consequently we find, in addition to literature based on classical models as to style and voca- bulary, a number of works, particularly war tales and other romances, which are plainly under the influence of the con- temporary spoken language. Unfortunately, while allowing us to make the general inference that the colloquial had by now considerably diverged from the spoken language, they do not furnish much evidence as to the details of this variation.

Similar remarks apply to the next, Namboku and Muro-


machi periods (1332-1603), though it is probable that by working backwards and forwards from a detailed study ad hoc of its documents a good deal of information could be gained as to the development of modern colloquial forms. The Yedo period (1603-1867), especially towards its close, witnessed a revival of learning, and a return to classical models of the Heian period, but this was artificial and could not survive, though it was not without influence on the written language. The spoken language meanwhile de- veloped apace on its own lines, and by the middle of the nineteenth century the two languages presented almost as many differences as resemblances.

In the following study of the development of the Japanese language, it has been necessary for reasons of space as well as simplicity to concentrate on a description of the earliest and the latest forms those of the Nara and Heian periods and of the present day without paying much attention to the intervening stages.

In compiling a grammar of any Eastern language one is confronted at once by difficulties of classification and nomen- clature. The traditional terminology of grammars of modern European languages, unsatisfactory in itself, is unsuitable and misleading when applied to a language like Japanese, which has grown up under the influence of concepts and per- cepts that do not correspond to those which form the basis of European speech. At the same time one cannot accept without change the principles of the great native gram- marians, who, remarkable as they were by their erudition and industry, knew no language but their own and were therefore ignorant of general linguistic theory. Consequently in the following pages I have been obliged to compromise, by following the Japanese practice where it seemed advan- tageous and eking it out with the categories of European grammars.


Examples taken from early texts are marked as follows








The Imperial Edicts or Rescripts in

the Shoku-Nihongi

Kok. or Kokin.



Taketori Monogatari


Ise Monogatari

G. or Gen.

Genji Monogatari


Heike Monogatari


Makura no Soshi

T. A.S.J.

Transactions of the Asiatic Society

of Japan

INTRODUCTORY § I. The Introduction of Writing

NOTHING is known with certainty as to the origins of the Japanese language. It has hitherto usually been considered to belong to the group variously known as Altaic or Finno-Ugrian, chiefly on the ground of structural resem- blance to other members of that group. It shows a strong structural likeness to Korean, but very little likeness in vocabulary. Recent investigations tend to disclose certain similarities in structure and vocabulary between Japanese and the Malay-Polynesian languages, but the evidence so far produced is not sufficient to establish any theory claiming a Polynesian origin for the Japanese race or the Japanese language.

The only language to which it is safe to assert that Japanese is closely related is Luchuan. Here the resemblance is so complete that Luchuan can be only a dialect of Japanese, and its vocabulary and syntax therefore provide no indica- tion of the origin of either language. A study of Luchuan is, however, of value in building up hypotheses as to the forms of the archaic language from which the Japanese of the earliest known period and the Luchuan variations thereof are both descended.

Apart from such conjectures, our knowledge of early forms of Japanese is derived from writings of the beginning of the eighth century of our era, which will be presently described. There is no trace of any system of writing in Japan prior to the introduction of Chinese books, which may be put approxi- mately at A. d. 400 ; and it was not until the sixth century, with the gradual spread of Buddhism, that the study of Chinese became in any sense general. Once the Japanese became acquainted with the Chinese system of writing it was possible, though not by any means easy, for them to make use of that system to represent words in their own language. For reasons of pedantry as well as convenience, as a rule they preferred to neglect their own language and write in


Chinese, much as learned men in Europe at one time used Latin ; but luckily for philologists they did elect to per- petuate, by using Chinese characters as phonetic symbols, the native form of certain poems, tales, and records which had hitherto been preserved only by oral tradition. It is these texts which furnish us with the materials for the study of archaic Japanese.

For a proper understanding of the extent and accuracy of the information as to early Japanese forms which can be derived from such documents, it is necessary to study in some detail the system of writing developed by the Japanese. Moreover, since the adoption of the Chinese script had a great influence upon both vocabulary and constructions in Japanese, it is important to trace, at least in outline, the growth of that system.

The unit in Chinese writing is a symbol which, through a curious but pardonable confusion of thought, is usually styled an ideograph, but is much more accurately described as a logograph. It is a symbol which represents a word, as contrasted with symbols which, like the letters of an alphabet or a syllabary, represent sounds or combinations of sounds. It is true that the first Chinese characters were pictorial, and that a great number of the later characters have a pictorial element, and to that extent may be said to represent ideas. But in fully nine-tenths of the characters now in use the pictorial element is either secondary or completely lacking, and the phonetic element is predominant. A simple charac- ter like ^ (moon) retains some vestiges of its pictorial quality, and may be said to represent the idea ' moon ', but nevertheless it stands for the Chinese word for moon (how- ever that word may be pronounced at different points in time and space e. g. ngwet in about a. d. 500, and yue in Peking, ut in Canton to-day) . When we come to more com- plex characters, it is clear that their formation not only presupposes the existence of a word, but is governed by the sound of that word. Thus, though ■% Jang, meaning ' square ', may at one time have been ideographic, tfj fang, ' to ask ', is composed of a phonetic element ~}j fang and a sense ele- ment "b, 'to speak', and does not directly represent the idea of 'to ask', but the word fang, which is the Chinese word for ' to ask '. When they wished to construct a character to


represent fang, 'to ask', the Chinese took the sign "jj, which stands for the word fang, 'square ', and to avoid confusion with this and other words pronounced fang, they added the ' radical ' jf , which conveys the idea of speaking.

A Chinese character, as used by the Chinese, is then an ideograph only inasmuch as any written symbol or group of symbols in any language is an ideograph ; but it stands for a word, and for one word only. I have insisted upon this point because, as we shall see later, the Japanese method of using the Chinese characters does at times approach an ideographic use.

Before describing more fully the Japanese method, it is as well to state briefly the problem which the first Japanese scholars had before them when they came to consider how to make use of the Chinese script for recording their native words. A simple example will suffice. The character \ stands for jen, the Chinese word for ' man '. The Japanese word for 'man' is hito, and a Japanese might agree to let the character \ be read by himself and his compatriots as hito, thus establishing \ as the conventional sign for hito. But there would still remain the problem of representing the sound of the word hito, and there were reasons which made it often essential to represent the sound rather than the meaning of Japanese words reasons which may for the moment be summarized by stating that while Chinese was monosyllabic and uninflected Japanese was polysyllabic and highly inflected. To write by means of Chinese characters the sound of a Japanese word, it was necessary to represent separately the elements composing that sound. Now by the fifth century Chinese had become a monosyllabic language, and since each syllable in Chinese was a word, there was a logograph for each syllable, and often of course many logographs for the same syllable. Consequently, when the Japanese wished to write the sound hito, they had in the Chinese symbols a ready means of representing the syllables of which it was composed, and they had no reason to analyse those syllables further into their constituent vowel and con- sonant sounds. This point has a considerable bearing upon the study of early Japanese forms, but it may for the moment be neglected.

To write, then, the syllable hi of hito, the writer must find


a Chinese character standing for some Chinese word of which the pronunciation was the same as, or as near as possible to, the Japanese sound hi. He would find, for instance, the characters , ^, f£, fa, representing Chinese words mean- ing respectively 'sort', 'not', 'grief, and 'ice', but all pro- nounced hi or something like hi.1 Similarly with the syllable to. He could use such characters as JJ, S\-, ^, and many others, all representing Chinese words of different meanings, but uniformly pronounced to. Thus, to write the word hito he could use any of the combinations Jrfc J], Jrb ^, Jfc -V> f? 71) ^B ^£> &c. Therefore in applying the Chinese script to the Japanese language, two methods were available which may be conveniently described as the semantic and the phonetic methods. The first method indicates the mean- ing of a Japanese word, the second method indicates its sound. The modern Japanese system of writing is a com- bination of these two methods, and we must now proceed to trace its development in outline, for, though an account of the script used to represent a language may appear to be out of place in a study of its grammar, the Chinese language was so much more highly developed, so much richer in vocabulary and scope, than Japanese of the archaic period, that the adoption of the Chinese script was naturally accompanied by important changes in the Japanese language.

Though there is some doubt as to exact dates, it is pretty certain that chief among the first Chinese books brought to Japan were the Thousand Character Classic ("f* ^ j£) and the Confucian Analects (!& gg-), followed very shortly by Chinese versions of and commentaries upon the Buddhist Scriptures. The Japanese scholars, when reading the Chinese classics, would no doubt at first be guided only by the sense of the Chinese symbols, which they had previously learned, character by character, from their instructors ; and since the Chinese logograph can convey to the eye any meaning conventionally assigned to it, irrespective of the sound by which it may be known, it would be possible for the Japanese scholar to read a passage of Chinese without knowing how

1 To simplify matters I assume here that the Chinese and Japanese sounds were both hi, though at the period in question one or both may have been pi. The principle under discussion is, of course, not affected by such an assumption.


it was pronounced in Chinese, and without consciously con- verting the Chinese symbols into Japanese words. Thus, to take a simple passage from the Analects : ^^-— -""" , ■■■-- \

child VMSlXr

speak 9c^4

king )CM, J,

child * So vi o\

weight if then not aweJo^l

a Japanese student of Chinese might take in the meaning of the characters without definitely translating them into words, either Chinese or Japanese. But to retain in the mind the meanings assigned to a large number of characters requires a very great effort of visual memory. It is in practice an aid both to memory and to understanding to associate sounds with signs, and therefore it was customary to read Chinese texts aloud, as we may infer from the habit, which persists among both Japanese and Chinese to this day, of reciting to themselves whatever they read, in tones varying according to the individual from a gentle murmur to a loud chant. Consequently it was for practical purposes necessary for Japanese readers to assign sounds to the Chinese charac- ters which they read ; and it was open to them either to use the Chinese sound of the word represented by the character or to say the Japanese word which conveyed the same, or approximately the same, meaning as that Chinese word. If they merely repeated the Chinese sounds, then what they recited was not intelligible to a hearer, because (owing to the great number of homophones in Chinese) the sound alone, without the visual aid of the character, is more often than not insufficient to convey a meaning even to a Chinese, while a Japanese whose knowledge of Chinese was by force of cir- cumstance chiefly derived through the eye and not the ear would be even more at a loss. Add to this the difficulty that the order of words in Chinese indeed, the whole gram- matical structure is in almost every respect the opposite of Japanese, and it is clear that for practical purposes some


arrangement had to be made to facilitate the reading of Chinese texts by Japanese students who, while visually acquainted with a number of Chinese symbols, were not familiar with Chinese sounds and Chinese grammar.

These were the important considerations which guided the Japanese in building up a system by which they could adapt the Chinese characters to their own needs, and they led to results which must surely be unique in the history of language. The problem differed somewhat according to the nature of the Chinese text in use, for in the period just after the introduction of writing into Japan the Chinese books chiefly studied by the Japanese fell into two well-marked divisions. On the one hand they had the Chinese classics works written in pure Chinese, where (as in the specimen from the Analects given above) every character had a mean- ing or at least a grammatical function. On the other hand they had the Buddhist Scriptures, written, it is true, in Chinese characters, but containing a great deal of phonetic transcription of Sanskrit words.

In reading the Chinese classics, the sound did not matter to the Japanese student. The important thing was to appre- ciate the meaning and to convey it to others. Now it must be understood that for one Japanese to convey to another in writing the meaning of a Chinese text was not at that period a question of translation as we understand it. Since the Japanese had no system of writing of their own, for a Japanese to be able to read any writing whatever presup- posed in those days a knowledge of the Chinese written character, and therefore a greater or less knowledge of the sounds and meanings ascribed to those characters by the Chinese themselves. What was needed, then, for the full comprehension by a Japanese of a Chinese text was not a change of the symbols, or the words for which they stood, but rather a rearrangement of the symbols to accord with Japanese syntax. The separate ideas conveyed by Chinese characters were clear enough to a Japanese who had learned them by rote, but he would not understand their aggregate meaning unless he was familiar with the Chinese method of grouping and connecting ideas. Therefore, for the benefit of the less learned, the more learned Japanese (and doubtless their Chinese and Korean teachers) devised a system of


reading the characters by giving some their Chinese sounds and some their Japanese meaning, taking them as far as possible in the order of words natural to Japanese, and sup- plying orally the inflexions, particles, and so on, necessary in Japanese to show the relations between words. Thus, they would take the sentence quoted on page 5 from the Analects and give to its characters the following readings in the order shown the words in capitals being the native Chinese sounds (or, more strictly, the Japanese approximation thereto), those in italics being the Japanese equivalents of the Chinese words, with inflexions added where necessary :

1. £j^ SHI The Master (i. e. Confucius)

2/fi) . iwaku____r__^- says v^i

3' ^ X~KUNSHI a gentleman-^

5. ^ arazareba if there is not 4. jf; omoku gravity

6. $ij sunahachi then 8. jf> narazu is not

7. Jg£ / respected

meaning ' Confucius said : "A gentleman in order to be respected must be serious ".'

It will be noticed that, though the English order of words corresponds closely to the Chinese, the Japanese order in- volves a rearrangement. The substantives in Chinese re- main in their Chinese form (SHI, KUNSHI, and /), but the remaining words, which in Chinese are uninflected mono- syllables whose function is determined by position, are con- verted into inflected Japanese words or particles. The simple negative ^ FU, for instance, becomes the compound verbal form arazareba, a negative conditional. In other words, the Chinese characters give the skeleton of a statement, and it is clothed in an elaborate grammatical robe of Japanese texture, composed of moods, tenses, and other intricacies to which Chinese is so magnificently superior. The process as thus described sounds exceedingly difficult, as indeed it was ; but, making due allowance for the nature of the script, it does not in essence vary much from the method of literal translation followed by schoolboys when construing Latin prose.


The practical objections to such a system are obvious. It was hard for a reader to tell in what order the characters were to be read ; what characters, if any, were to be taken together ; which were to be given the Chinese sound and which were to be converted into Japanese words. To diminish these difficulties as far as possible, Japanese students of Chinese texts resorted to the use of diacritics, combining them with a system of markings (equivalent to the numerals and brackets in the example) to show the order and grouping of the characters. This is not the place to describe these devices in full, but the general principle may be outlined as follows :

Each Chinese symbol is regarded as being enclosed in a square, and certain dots (ten) or strokes at various points of this imaginary square represent, according to a fixed, though quite arbitrary, arrangement, flexional terminations, suffixes, particles, &c, which in reading are supplied orally after the reading of the character. Thus, according to one such scheme, which can be represented diagrammatically :





() TO





if we take the character & ('fear') and fix as its equivalent the Japanese word ' kashikomi' , then

& kashikomite (a gerund)

kashikomu koto (the act of fearing) kashikomitari (past tense)

It is highly probable that this method of dia- critics was suggested by the marks used by the Chinese to indicate the tones of Chinese words.1

1 It is one of these schemes which accounts for the word ' Teni- woha ' , used by Japanese grammarians as a generic term for particles and other parts of speech which are neither nouns, adjectives, or

and so on.


It was a clumsy method, and obviously not fitted for general use, but it survived in a remarkable way, partly because the Japanese language, though rich in forms, was poor in vocabulary, and it was therefore essential to pre- serve a large number of Chinese words which could not be satisfactorily translated into Japanese. The word kunshi ^ ^ is a case in point. In the Analects it had a special meaning 'the scholar-gentleman' which could not be ex- pressed in Japanese, and consequently kunshi was adopted as a Japanese word, one of the forerunners of the multitude of Chinese words which now form the greater part of the vocabulary of Japanese. Nor was the adoption confined to single units of the vocabulary. Many constructions and grammatical devices in Chinese could not be exactly repro- duced in Japanese, and were often borrowed with little or no change, either because it was difficult to find an equivalent or because they were a convenient addition to the gram- matical apparatus of Japanese. The sentence quoted above provides a good illustration. Shi iwaku, ' the Master says ', is a Chinese construction, while the pure Japanese idiom requires a verb like 'to say' at the end, not the beginning, of a reported speech. But the Chinese method was incor- porated into Japanese syntax, and a construction similar to that of shi iwaku, &c, has survived until to-day.

There was another powerful reason for the survival of the diacritic method. Its very difficulty was a merit in the eyes of the learned men who used it, and the leading schools of Chinese studies, as well as some Buddhist sects, each had their own system or systems of markings, which they kept secret and imparted only to their disciples. It is a curious instance of the esoteric habit which prevailed, and is still discernible, in art and letters in the East.

The use of diacritic markings might have continued in- definitely had it not been for the growth of another system

verbs. Te, ni, wo, ha were the four words at the corners of a system called ' wo koto ten ', represented by

Nit fVVO





which was more convenient in many respects. This was the phonetic system of writing Japanese words, which we have already briefly described. The semantic system grew out of the need to convey to the mind of a Japanese reader the meaning of the Chinese work he was studying. But there were a great number of works in reading which it was essential to know the sound of the characters. Chief among these were the Chinese translations of the Buddhist sacred writings, in which there were many Sanskrit names and Sanskrit terms which could be rendered into Chinese only by a phonetic method. The Chinese, in fact, had several cen- turies before the Japanese been confronted with the problem of applying the logographic script of a monosyllabic language to the phonetic transcription of a polysyllabic language entirely different in grammatical structure. How, for instance, were the Chinese to translate from Sanskrit into their own language not only Indian names of places and persons, but also the terminology of the sacred writings which represented religious and philosophical ideas entirely foreign to them ? The phonetic method was the only pos- sible solution, and the history of the development of a system of transcribing Sanskrit letters and sounds by means of Chinese characters is a fascinating one. Here it is not neces- sary to describe it at length, but some acquaintance with the method used is necessary for a proper understanding of the origin and growth of the system eventually worked out by the Japanese.

If we take the great Lotus Sutra as a typical example, we can see at once what difficulties the translator had to surmount. Its very title, Saddharma Pundarika, was difficult to render, and in the first translation extant (Nanjo 136) an attempt is made at a phonetic rendering, by means of the characters g| j| xfc [?£ f ij, which stand for Chinese words pronounced respectively something like sa, dan, pan, do, and li} Reading these characters together, and paying no atten- tion to their meaning, we have Sadan ftandoli, which is a rough approximation to Saddharma Pundarika, but of course conveys no meaning to a Chinese reader ignorant of the original Sanskrit. This was clearly a makeshift method, and

1 These are only approximate, and I do not pretend that they are the correct sounds of the Chinese words at the period in question.


in later translations an attempt was made to reproduce the meaning of the Sanskrit words, by using the characters IE & ^ $£> pronounced Cheng Fa Hua Ching in modern Pekingese, but meaning True Law Flower Scripture.

Coming now to the opening words of the Sutra, which state that 'once upon a time the Buddha was staying at Rajagriha on the Gridhrakuta Mountain with a numerous assemblage of monks', we see further difficulties before the Chinese translator. Place-names like Rajagriha and Gridhra- kuta have, it is true, some meaning, signifying respectively 'The King's Castle' and 'The Vulture Peak', so that it was possible to represent them by Chinese characters standing for Chinese words of approximately the same meaning, viz. 3E & $c King House Fort, for Rajagriha, and ^ UJ Eagle Mountain, for Gridhrakuta. For the Sanskrit word bhikshu (Pali, bhikkhu), usually rendered by 'monk', the Chinese translator might perhaps have invented some equivalent Chinese term, but since monks did not exist in China apart from Buddhism they preferred to adopt the Sanskrit word, which they reproduced phonetically by the two characters Jfc Jr. pronounced in Chinese pi k'iu. So far it might have been possible to find equivalents for the meanings of the words in the Sanskrit text, though it will be noticed that the very appellation of the Buddha himself raises in an acute form the question of selection between translation and transcription. Shall the translator use characters which signify 'enlightened' but may to the Chinese reader have misleading implications, or shall he use characters divorced from their meaning to represent as nearly as possible the sound Buddha ? *

However, when we reach the later chapters of the Lotus, the difficulties of translation become insuperable, and there is no alternative to the phonetic method. Chapter XXI,

1 The translators chose to use the character fjjj>, which in ancient Chinese was pronounced (according to Karlgren) b'jued. But Chinese pronunciation has changed in a way that the translators can hardly have foreseen, and the modern pronunciation in the Mandarin dialect of f$j is/o. The Japanese pronunciation butsu, which represents the Chinese sound at the time when it was borrowed say, a. d. 400 has survived unchanged, and is therefore nearer the Sanskrit original than modern Chinese.


for instance, consists largely of spells or talismanic words (dhdrani) , such as anye, manye, mane, mamane, which cannot be translated any more than, say, abracadabra. Since these incantations were regarded as of great power and value, the translators of the Sutras were obliged to find phonetic equi- valents for them. So, in an early translation, the above words are represented by % H, ■§! UJ, 0 fg, j$t J^ fgj where each character represents a syllable of the Sanskrit words and is used entirely without reference to its Chinese meaning.

We see, then, that some system of phonetic transcription of the Sanskrit alphabet was essential, and that the Chinese were obliged to adapt their own script to this purpose. Had they carried further the process outlined above, they might from these beginnings have developed a simple alphabet or syllabary. This they failed to do, but we must at least give to the Chinese, and not to the Japanese, the credit for the first phonetic use of the Chinese character. Unfortunately, instead of establishing a uniform system of phonetic tran- scription, which might by gradual simplification have led to the formation of an alphabet, the Chinese translators seem to have deliberately chosen not only a difficult and irregular scheme of transcription but also a great variety of such schemes. Stanislas Julien in his masterly work on the sub- ject gives a list of 1,200 Chinese characters which were used to render the forty- two letters of the Sanskrit alphabet, in- cluding the combinations of the consonants with all the vowels and diphthongs, and this list is far from complete. Not only was a given Sanskrit letter represented by more than one Chinese character, but the same Chinese character was used to represent more than one Sanskrit letter. Thus, according to Julien, the Chinese characters ^, fp, fp, $gf, and \$, pronounced in Chinese cha, tsieh, chi, to, and che respectively, were all used to represent the symbol ^ da : while the character JJj, in Chinese che, is found standing for Sanskrit djha, dha, dya, dhya, and cha.

With such models before them, it is not surprising that the Japanese were slow in developing a phonetic script of even relative simplicity. Their problem was not unlike that which had laced the Chinese translators of Buddhist writings, since they had to find Chinese characters to stand for the


sounds in a polysyllabic language. It is hard to say when the first attempts were made by the Japanese to put their own language in writing. The earliest chronicles, such as the Kojiki and the Nihongi contain references to historical records of events in Japan preserved in writing. Thus, in the preface to the Kojiki, the author states that the Emperor Tenmu complained that ' the chronicles of the emperors and the original words in the possession of the various families' were inexact. We may infer from this that written records had existed long before the reign of Tenmu, which began in 673.

In the Nihongi, under the date 403, the appointment of provincial historiographers is mentioned, but the chronology