Literary Chinese Viewed in the Light of Literary Latin

I. The Nature of a Literary Language

We usually refer to the languages of classical Rome and classical China in their continued use during two millennia in post-classical Europe and East Asia respectively as Literary (or Late) Latin and Literary Chinese, What is meant by a "literary" language? It is first and foremost a written language. Not all written languages can be called literary languages in the sense of literary Latin and literary Chinese, but many of the special qualities which we attribute to such literary languages--their tendency to transcend the limits of national boundaries and time, which confine and ultimately doom ordinary natural languages; their prestige, usually in contrast to a "vulgar" spoken medium; their irregular phonological evolution; their role as vehicle for and symbol of a transmitted cultural tradition--these can in part be explained as characteristic of written languages in general. But there are as well several important respects in which these so-called literary languages are different from other written languages, The most important of these is their reputed resistance to change, which is a central topic of Parts II and III below.

Now to compare or contrast literary Latin or Chinese with written languages in general can scarcely elucidate their special properties, since the nature of written language, or rather of the writer's competence which underlies its production, is itself largely unknown, being an area of neglect in linguistics. Let me first mention what does seem clear.

It is generally accepted nowadays that written language is not the same as spoken language. We don't simply write down what we would otherwise speak aloud. But this has not always been obvious. When some Chinese in the first part of this century erroneously assumed that Western languages were written the way they were spoken, they felt isolated and unique in that their standard written language of 2000 years, wenyan 文言, or literary Chinese, was so different from the spoken tongue that foreigners in China called it an entirely different language.[2] But in this respect the historical development of Chinese is now seen to be different from that of other languages in degree rather than in kind. Many of the features of that written language are characteristic of all written languages.

We lack any general theory which characterizes the difference between spoken and written language (or style), so what follows is necessarily tentative and fragmentary. We usually assume that if a person is literate, his competence as a "native writer" more or less follows from his competence as a native speaker. When we write our native language we base what we write on how we speak, but we depart from speech in two ways, We deem certain spoken expressions inappropriate and suppress them. At the same time, we often use certain other expressions in writing which are seldom if ever used in speech. In English, as examples of the former, verbs like get or stick, or a conjunction like 'course (= of course in the sense of "on the other hand") are systematically paraphrased in writing. Getting harder is written becoming more difficult. Indeed, there are some expressions in speech in all languages which we would be at a loss how to represent in writing. How does one write a-whole-nother ( < another whole), common in some dialects of American English, an expression which defies the conventional morphological analysis of the word another as an + other? As examples of the latter, the written words not usually used in speech, take the words deem and seldom which occurred a few lines above. Written deem alternates with spoken think or consider. The spoken counterpart of seldom, in American English at least, is hardly ever, with rarely falling somewhere in between in stylistic value, and used regularly in both speech and writing.

The area of contrast between spoken and written language is in the case of English primarily confined to lexical items, although in such cases as the alternation of spoken form anymore with written form no longer, other elements in the sentence may be influenced too: doesn't live here anymore becomes in writing no longer lives here. In the case of other languages like Japanese or French, however, it includes areas of morphology and syntax too, such as contrasting sets of verb endings or personal pronouns. For modern Chinese, as for many other languages, the contrast involves a difference in sentence length and complexity. Written Chinese sentences are not only longer and more complex, but their logical structure is also more explicitly marked by conjunctions and other grammatical function words. Written Chinese vocabulary is also uniformly more polysyllabic than spoken Chinese (although this has much to do with the differing nature of the content).

It appears that the direction in which usage restrictions are primarily imposed varies from language to language. Thus in English there are relatively few expressions which may be used only in writing and which are unacceptable even in formal speech.

Those distinctions which there are derive more from the avoidance of spoken forms in writing, For example, while we usually avoid using the word sick in written English, its written counterpart ill is acceptable in formal speech. But in Chinese the opposite is true. While there are, to be sure, stylistic sanctions against spoken forms in writing, as in English, much the greater part of the difference between spoken and written language, even in the case of contemporary Chinese, comes from an overwhelming number of forms permissible in writing but absolutely impermissible in speech. For example, the expression fancheng X 翻成X 'translate into X,' common in speech, is acceptable in writing too, but not nearly so common as yicheng X 译成X 'translate into X,' which may not be used in speech.

How is the ability to discriminate acquired? In the United States most such knowledge is not explicitly taught, and it appears that the mature native speaker achieves his competence in writing substantially as he acquires spoken language itself as a child--namely, through the creative but unconscious act of inferring a set of rules governing, respectively, the speech he hears about him, or, in the case of written competence, the writing he has occasion to read. This two-level competence is clearly related to the ability of the native speaker to use different styles appropriate to different situations. For some languages there are several discernible styles within written language itself which all literate native speakers can control.

We can make certain other observations. It is also true that in the case of written language the ability of the writer to produce well-formed sentences at will--his competence--is less obscured by factors like lapse of memory, interruption, false starts, and the like, which occur in normal speech and make it desirable in linguistic analysis to consider an utterance as if such features of "performance" were not present. Thus a native speaker/writer's competence is closer to realization in writing than in actual speech. Notice also that in the same way that linguistic competence does not significantly vary according to intelligence or educational background, ability to write is more or less uniform (leaving aside the problems of style and discourse-level acceptability) throughout the literate population. That is, if what a person says is acceptable, then what he writes should be equally acceptable. A written language does not have a phonology which ends up in sounds as a spoken language does, but its rules of orthography serve a very similar function. (Chinese is the great exception, as will be discussed in Part III.) In syntax, in the lexicon, the differences are slight for some languages, like English, greater for other languages. Finally, there is an important distinction between spoken and written language in the degree to which an utterance is attached to a communicative context. Spoken language is always oriented to a particular context, whereas written language is not so context-oriented. So it is unlikely that such sentences as "I forgot to give him the bundle over there," or "You've got ice cream in your beard" would occur in written language, except in the dialogue of fiction or written drama, where great pains are taken by the author to construct artificially such a context.

When we view language in historical perspective and say that language is constantly changing, we include written language as well. Internal change can be said to occur in written as well as in spoken language phonologically/orthographically, and grammatically, through the addition and deletion or reordering of phonological and grammatical rules; and lexically, through the simultaneous action of the forces of lexical profusion, such as meaning shifts and extensions, and those of lexical economy, which prune from the lexicon those items which are redundant or no longer essential to communication. External change is primarily the result of borrowing.

Here in the process of change too there are some important ways in which written language is different from spoken language. It is more conservative and slow to change. Orthography, especially, does not change in the same manner as phonology, and is more the product of human intervention. There are few cases in history where a single culture has developed over a thousand years or more while relying on a single instrument of written expression, to serve as objects of study of the stresses on that instrument. But most written languages keep step with, a few paces behind, their spoken counterparts, and hence the language of one generation becomes gradually more obscure to later generations, even unintelligible in a matter of three to four hundred years.

Consider English, for example. Perhaps we are entering a period of literary crisis in English because the language of some of our greatest authors, principally Shakespeare, in the Elizabethan period is, after less than 400 years, becoming too difficult without special philological study. Despite the fact that Shakespeare's writings have been kept alive this long by updated spelling and the especially conservative orthography of English, which make him more comprehensible to our eyes than he would have been to our ears, it appears that even he will become inaccessible to average readers in the next hundred years or so on syntactic and lexical grounds.

The Japanese written language has changed even more rapidly. The great Heian period novel of a thousand years ago Genji Monogatari 源氏物语 (The Tale of Genji), having long ago become comprehensible only to scholars, was translated by the poetess Yosano Akiko in the 1920's into the appropriate Japanese literary medium of the day. Already in the 1970's that excellent translation was no longer enjoyable for young people to read because of its linguistic difficulty. As a result several new translations into modern Japanese have recently appeared.

Thus written and spoken language, distinct though closely related modes of expression, both undergo change. Literary languages, however, are often held to be exceptional, so the study of literary Latin and literary Chinese can be considered in part a case study of the abortion of this normal process of linguistic change, by which written languages are modified in keeping with their spoken counterparts. Such literary languages were originally typical written languages, participating in the normal process of change. But their special characteristic is that at a certain point in their development they seem to have stopped changing, while their spoken counterparts kept on changing. At the same time, they continued to be used as literary languages, not only in the area of their origin and previous use as ordinary written languages, but also in other areas, by peoples who spoke a totally unrelated language but who had no other written language available to them. In either case people spoke one language, which they could not write, and wrote another, the literary language. For both Latin and Chinese the gradual eclipse of their hegemony as literary languages began just when speakers made their first efforts to write down that language which they spoke but could not write. Parts II and III will take up the historical reasons for both their rise and their subsequent decline. I hope that by first reviewing the fairly well-understood history of Latin in Part II, I can lay a comparative foundation for examining the somewhat more obscure history of literary Chinese in Part III.

First let us look at some common features of their existence. In the case of Latin the point of stunted development is held to be the declining years of the Roman Empire in the first centuries A.D. From that time Latin held sway as the sole written language of any prestige over the entire Romance bloc of southwest Europe, Ireland, England, Germany, Hungary, Poland, and the Scandinavian countries until the Renaissance, and even into the seventeenth century or later in some parts of Europe. (Note that Romania, although a Romance-speaking country, lies outside this sphere of literary Latin dominance because of the importance of Slavic through much of its history.) In the case of Chinese, the point when the written medium became a literary language in our sense here, and no longer kept step with the spoken tongue, is placed in approximately the same period--that is, during the Han 汉 dynasty, whose power extended over the same years as that of ancient Rome, from the second century B.C. to the second century A.D. But its use as a literary language in most of East Asia, including Japan, Korea, and Vietnam, continued even longer than Latin--until the nineteenth century, and in China itself until the first decades of the twentieth century. Indeed, it exists even today in a truncated form, as the only written language of consequence in the minds of some Chinese speakers.

Literary Latin and literary Chinese are sometimes referred to as "dead" languages, Here we must clarify the notion of what is meant by "dead," and also restate the importance of distinguishing the history of the literary languages from that of the spoken languages upon which they were once based. In the strict sense, "dead" means "extinct," and neither the Latin spoken tongue nor Old Chinese, the spoken tongue of the Zhou 周 dynasty which preceded the Han, are "dead" languages without any further evolution, in the way that thousands of languages have become extinct, with more being added to the list each year, because their last remaining native speakers died and failed to be replaced by new ones. Both languages continued to evolve, and modern Italian (or Spanish, etc.) is "living" Latin and modern Chinese is "living" old Chinese, in the sense that they are their evolutionary outgrowths.

Still, the anti-historical concern with the death of the old languages continues. For example, given the early specialization of literary Latin and its divorce from the spoken language, and the continuous nature of the evolution of that spoken language, debates over the "problem" of the termination of Latin as a living language, such as carried on by Giacomo Devoto in Current Trends in Linguistics, seem remarkably irrelevant to linguistic scholarship:

At present there is the problem of the termination of Latin as a living language. The two extreme solutions are represented by H. F. Muller, who assigns it to the Age of Charlemagne (in A Chronology of Vulgar Latin, Halle, 1929); while in my Storia della lingua di Roma, I retain the classic date of 476 A.D.[3]

Interestingly, China, which has enjoyed a more or less continuous cultural and political unity which Europe has not, has been spared the confusion popularly created in the West, at least in the English-speaking world, by the historical break between antiquity and the modern era (the so-called "Dark Ages"), and the resulting nomenclature of "Latin" on the one hand for the parent tongue, and "Italian," "Portuguese," "French," etc., on the other hand for the daughter tongues. China's linguistic evolution has been very similar to the Romance area's and over approximately the same era, but no such difference in terminology has arisen and no Chinese has to be taught to hold the historical view of his language as the modern avatar of the classical language. Yet many Americans believe that the languages of classical Rome and, say, modern Lisbon are two separate languages, rather than the same one in different historical stages of development.

On the other hand, it has taken Chinese linguistic science much longer to describe systematically the differences between the Chinese dialects, of a magnitude like those between the Romance languages, since aren't they all "Chinese"? Moreover, to a certain extent under unfortunate Western influence and the notion of "dead" languages, but also for quite astute political reasons, Hu Shi 胡适 and other Chinese intellectuals who participated in the cultural revolution of the early twentieth century often referred to the Chinese literary language as a "dead" language, in contrast to the "living" language that contemporary Chinese spoke.[4] Hu has been severely taken to task by linguists in the People's Republic of China, grinding their own political axes, to be sure, but also quite properly for linguistic reasons, because of Hu's ahistorical view of Chinese linguistic change.[5]

Since the term "dead," strictly defined, does not apply to the spoken Latin and Chinese languages, and it is uncertain whether literary Latin or literary Chinese were ever "living" in this sense, let us then understand it more loosely in connection with the literary languages to mean either a language without native speakers in the conventional sense, or a language whose grammar and structure is fixed for all time, a language which does not live and grow. The former seems to describe these languages well enough. What of the latter? Did these languages not change at all during the millennia in which they held sway? This is far from being the case for either Latin or Chinese. As I will show below in examining their use and development, they were constantly changing and evolving under the influence of the spoken language of each era and in each speech community. The most we can say is this: although their development did not proceed as we might expect on linguistic grounds alone, from that crucial period of the Roman and Han empires they did change and develop, but more slowly and idiosyncratically. During this long period, as languages without native speakers or "native writers" in the sense discussed above, their behavior was not what we would predict on the model of other written languages. It was much more erratic and unpredictable, capable of very sudden and sweeping changes, and sometimes just as sudden reversals. It seems also to have been considerably more susceptible to the influence of individuals than other "non-literary" languages are. How can such aberrant linguistic behavior be explained? I think that two characteristics shared by the users of literary Latin and literary Chinese need further clarification. The first is that, unlike ordinary natural languages, the competence of the users was not uniform but varied. It depended first on whether or not the user's mother tongue was a later version of the spoken language upon which the literary language was itself founded. That is, if the user spoke a language cognate with the literary language, his competence was of a different order from that of a user whose language was not. A Spaniard who wrote Latin had an advantage over an Irishman or a Swede. A Chinese who wrote literary Chinese had an advantage over a Japanese or a Korean. It was easier for the Chinese and Spaniard to learn the literary language because of all the cognate vocabulary and grammar, and consequently they achieved fluency effortlessly.

Their advantage, however, does not necessarily mean that the language of the Spaniard or the Chinese at any given date was closer to the language of the classical texts which gave rise to the literary languages. The very lack of the mnemonic aids to learning spurred the exceptional Irishman or Korean who made the great effort at mastery to achieve a more classical correctness if he desired than the great majority of users in the cognate areas could achieve. This is true for phonology as well. The absence of phonological similarity stimulated a greater conservatism in the non-cognate areas. But the competence of the user, regardless of whether he was in a cognate or non-cognate area, depended heavily on the quality of education in the literary language as well. Note that the dual importance of both the user's native language and the quality of his education in determining acquisition of the literary languages is true for the acquisition of foreign languages today too. There is often a converse relation between fluency and correctness. It is a familiar phenomenon that a Chinese who has really mastered English speaks it more correctly, though probably not more fluently, than a German or Frenchman with a similar confidence in English, who relies unconsciously on cognates in his native language. The nature of literary language education will be discussed in the historical sketches below.

The second characteristic of the users which must be clarified was their intention to write in the literary language as it was received from previous generations, their desire to imitate rather than to invent, if you will. In understanding the language of any document written in such a literary language, both of these characteristics must be considered. We can assume for users in non-cognate areas an intention to write in the received literary language. Here there is no question of a conscious desire to invent or create a new medium. But for users within the cognate areas of modern Italy, France, Spain, and Portugal; or in the areas of China proper where "Han" Chinese languages were spoken--motivation to write in a classical literary language varied at least as much as the actual ability to do so, and quite independently of it. In understanding the historical development of the literary languages it is naturally important to evaluate both characteristics. But for the historical study of the spoken languages themselves within the cognate area, an understanding of the distance between a written document and the spoken language of its time also depends on a consideration of these two--ability and intention.

For there have been men who were able to write Latin which resembled classical Latin and consciously sought to do so, such as the twelfth-century French theologian Abelard, or the fourteenth-century Italian poet Petrarch, just as there were men in China like the historian Sima Qian 司马迁 of the Han dynasty (around 100 B.C.E.) whose prose was so much a stylistic exemplar for men of later times that it is sometimes difficult to decide whether it is he or his models in the earlier classical period of the late Zhou who should be considered the true fountainhead of literary Chinese style; or Han Yu 韩愈 of the Tang 唐 dynasty, who in the eighth-ninth centuries spearheaded a classicist movement to restore the prose style of the late Zhou philosophers.[6] Though outside the cognate area, Ogyū Sorai 荻生徂徕, the famous scholar of classical China in eighteenth-century Japan, should also be counted as one whose intentions were matched by his skill, according to Professor Yoshikawa Kōjirō.[7] Appendix C contains an example of a contemporary effort at a classical simplicity in literary Chinese prose, by the late Professor Shih-hsiang Ch'en (Chen Shixiang). It is a tribute to his colleague at the University of California at Berkeley, the late Professor Peter Boodberg. And of course classical Latin continues to be used today, long after its demise as a literary language, for the sake of tradition or prestige, in pockets of conservatism. For an example from one such pocket in "Novo Portu in Republica Connecticutensi," a Yale diploma, see Appendix A.

But at the same time there were men who, while they tried to write in a pure Latin or pure literary Chinese, simply hadn't the ability or training, and relied heavily on their own native language. Such is the case with the medieval author of the contract dated A.D. 1011 cited by M. Menéndez-Pidal in his Orígenes del espanol:

Ego Gomez Didaz et uxor mea Ostrozia placuit nobis expontanias nostras uolumtates ut conkambiauimus et uindimus nostra billa. Onia cum suas casas et suos omnes abitantes in ea et terras et unineas et ortus et arbusta et totus pomiferos qui in ea sunt et molinos et pescarias . . . .

Antoine Meillet, who quotes this in his Esquisse d'une histoire de la langue latine, notes that one dare not say whether the text is in Latin or Romance. But he feels that the author clearly tried to write in Latin and the Spanish was involuntary.[8] Is such also the case with the Chinese document reproduced in Appendix B and discussed in Part III? Here one hesitates to assert that the author, presumably a Cantonese native speaker employed by the FBI, was even trying to write literary Chinese. But what else is it? It is certainly not Mandarin, either spoken or written, nor is it Cantonese. There is no other explanation for his diction.

Then there were the men who had the competence to write in the classical literary medium but chose not to do so. Thus St. Augustine, who wrote his City of God in a classical style in order to appeal to cultivated pagans in a language that they respected, consciously shaped a style much more influenced by the Vulgar Latin spoken language of his time (the fourth century A.D.) in his works composed to spread the Gospel among the common people. In the same way and about the same time, St. Jerome's translation of the Bible into Latin, the Vulgate, reflects changes that had taken place in his own colloquial language.[9]So does the Latin of Gregory of Tours in the sixth century. In China we see writers making similar choices, Wang Chong 王充 of the Han dynasty, in the first century A.D. did; so did Tang poets like Du Fu 杜甫 and Li Bai 李白. They frequently incorporated elements of the spoken language of their time in their poetry, although they could be more classical if they chose. Particularly revealing in this connection are their "imitations" of previous styles. In the great variety of prose styles, some quite colloquial, attributed to Zhu Xi 朱熹 of the Song dynasty (twelfth century), we see a parallel to the flexibility of St. Augustine or St. Jerome.

Finally there were those, presumably the vast majority of the populations, who had neither the ability nor the inclination to write the literary languages. Whatever efforts they made have, blissfully for the classicist sensibility, not been passed on to posterity, except in a few exceptional cases. But we can assume that some of them were responsible for contributing to the emergence of a new literary language based more closely on the colloquial language. In Europe this was, of course, the Romance vernacular written languages, which in the earliest extant examples in the tenth and eleventh centuries already show a considerable sophistication reflecting a period of prior development. In China, it was the literary style called "plain speech" baihua 白话 which began to appear during the Tang and Song dynasties and already had had an illustrious and varied history as the vehicle for popular literature for over a thousand years before becoming the basis for a new all-purpose written language, which is taking shape only now in the mid-twentieth century.[10]

Since I have raised the question several times of the relation of these literary languages to natural languages in general, and since for both literary Latin and literary Chinese, but especially for Chinese, there has been a great deal of debate whether they were ever spoken languages, I would like to discuss the extent of their artificiality briefly by putting them not just in the context of natural languages, those we speak and write everyday, but widening the scope of the discussion to include all "languages," including machine languages, man-made languages for international communication, and the like. In the wider context it will be clear that literary Latin and literary Chinese are nine parts "natural" for every one part of artificiality.

Brian Higman has tried to frame a definition for "language" which would include all languages, natural languages as well as artificial languages like those used in computer programming and symbolic logic:

. . . a language consists of a set of objects called its vocabulary, which can be combined into linear strings in accordance with certain rules known as its grammar, for communication to a recipient with the intention of inducing activity in the recipient relative to certain specific features abstracted from a general situation.[11]

He goes on to define more narrowly how natural languages may be distinguished from the rest by two broad criteria:

The associations which tie the vocabulary of a language to the features of the situations with which it deals, and thus determine the meaning of a string, are known as its semantics and are usually completely conventional, even though the history of the language may show them to have a natural origin, and to this extent all language is highly artificial. However it is usual to refer to some languages as natural and others as artificial, meaning by the former those which possess the power of growth, and whose present form is the result of much evolution; and by the latter, ones which were created at a stroke and do not have this power of evolution. . . . Alternatively, we may define a natural language as one which is, or was, used somewhere sufficiently for children to acquire its use in the natural course of development, without a conscious learning process. . . . Every language which is used as a means of communication between human beings will inevitably be required, sooner or later, to deal with situations of a hitherto unforeseen nature. . . . Each intercommunicating group will force the language to meet its own needs. . . . in particular each successive generation of children picking up and using a natural language in the second sense, will introduce evolutionary changes and make it a natural language in the first sense.[12]

Under the second criterion, that of unconscious acquisition, our literary languages are similar to natural languages only in the very restricted sense I have discussed, namely, that they have grown out of written languages based in some way on such naturally acquired languages. But they clearly have the power of growth, and it is easy to recall "situations of a hitherto unforeseen nature" which they dealt with by sweeping changes (the first criterion). Such was the change in literary Latin at the hands of the Scholastics; such also was the change in literary Chinese wrought by the influence of Western civilization and China's modernization. Here they are far more similar to natural languages than to languages created at a stroke, like Esperanto (though Esperanto too would change, were it used over a long period of time as a real instrument of communication). And their structure is infinitely more complex than that of even the most sophisticated computer programming languages like FORTRAN, PASCAL, or C, or of the predicate calculus, or even of a reduced form of natural language like I. A. Richards' Basic English.[13]

In one sense the lexicon of these literary languages has resembled the much reduced one of Basic English, since as the language of literate elites, they did not have to deal with all the aspects of daily life that the lower classes did, and thus found it unnecessary to evolve the vocabulary to meet all daily life situations. But when it has been essential to enrich their vocabulary in response to a need, they haven't hesitated to borrow massively from the resources of colloquial language or from other unrelated languages. The Christian vocabulary of ecclesiastical Latin and the Buddhist vocabulary of Chinese, the technical vocabulary in logic, mathematics, and general science of twelfth-thirteenth century Latin, and the vocabulary of modern literary Chinese are examples. Taken as a whole the vocabulary of both literary Latin and literary Chinese is vast indeed, and each is far greater than the richest spoken tongue could be, since each is the accumulation of two millennia of innovations.

Despite the great similarity of such literary languages to ordinary language, the special nature of their development has led some scholars to claim special status for them. Chinese has especially been subject to this approach because of the nature of its script, which has in fact been instrumental in giving literary Chinese its longevity, as will be discussed in Part III. Some, like the Danish linguist Gustav Herdan (in Herdan (1964)), have treated literary Chinese less as a language than as a mathematical puzzle. He deal with the "combinational element in Chinese grammar," that is, the ability he claims for "ideograms" to be ordered in any arbitrary way and still make sense. He also treats the development of Chinese vocabulary as a "gradual approach to efficient coding," and places great importance in the numbers five and eight, as for example in the number of strokes in the phonetic element of a graph. A less extreme, but still anti-linguistic, conception of literary Chinese has been expressed by Herbert Franke. He feels that the Chinese written language is more correctly considered a system of visual symbols than a genuine language. As such, he asks, are the ordinary methods of linguistics appropriate for studying it?[14]

One feature of both literary Latin and literary Chinese is their concentrated sentence structure, their great economy of expression. The Chinese literary language is often so lapidary that it has been likened to a "telegraphese," which could not possibly have ever been spoken.[15] Similar doubts have been raised about classical Latin's viability as a spoken language in its recorded form.[16] In terms of redundancy, which is so necessary to successful oral communication, the literary mediums are indeed quite low--certainly considerably lower than the approximately 50% redundancy of English.[17] One problem with Franke's observation is that the grammar of artificial languages and other such symbol systems, rather than being more difficult to discover than that of natural languages, is much easier. If Chinese grammar seems to defy analysis with ordinary linguistic tools, it is because of its complexity and subtlety, which as with a natural language, is the result of a long evolution over a vast area, which has introduced diachronic and dialectal variation.

The protean nature of literary Chinese has led other scholars in the opposite direction, to claim that literary Chinese is so chaotic and unsystematic that only in taking account of the context of every individual utterance, and matters like rhetoric and sentiment, can it be understood. They lay heavy stress on individual intuition and frown on attempts to capture regularities in the grammar and lexicon. Achilles Fang is representative of this school: "Another fetish of a group of Sinologists who still think Chinese (Classical Chinese) is a "language" in the conventional sense is their firm conviction that a perfect dictionary will smooth the way."[18]

I would like to conclude this section by noting an often repeated observation that just because a language no longer exists we need not abandon the hope of treating its grammar as capable of the same creativity as the grammar of an existing language.[19] The grammar, or actually the grammars representing the competence of users of successive stages, of literary Latin and literary Chinese should be able to generate an infinite number of sentences which would agree with the grammar of the extant corpus. Robin Lakoff and Robert King have separately observed that while the philologist traditionally restricted himself to actual attested forms in scholarly study of the old languages, that actual users felt no such restrictions, and assumed the possibility of creating new acceptable sentences which had perhaps never before occurred. It was, in fact, only this assumption that made it possible for the teacher to correct his pupils' literary Chinese or Latin composition. Y. R. Chao has made a similar observation:

By studying Mencius closely enough, one could produce a generative grammar from which sentences or whole discourses could be generated that would agree entirely with the grammar of the corpus known as Mencius. As a matter of fact, this is very much what has been actually done, since current wenyan [literary Chinese], which is still a living language, is modeled more closely after Mencius than after any other of the classical works.[20]

The general truth of this may certainly be acknowledged, while demurring from the implied extreme conclusion that any of the sentences of modern literary Chinese could be generated by a grammar which was adequate for the language of Mencius, even if differences in vocabulary were ignored. If wenyan lives on today, it is not in a form that Mencius ever knew.

II. Literary Latin: A Historical Sketch

Literary Latin was taking shape in Rome already by the third century B.C. When Livius Andronicus produced his first play on a Greek model in 240 B.C., he translated the archaisms in Greek by using words which were already obsolete in his own time.[21] From this point on recorded Latin begins to differ from the spoken language. From L. R. Palmer we receive the impression of a literary public consciously shaping an instrument of literary expression, striving always for greater felicity, elegance, economy of means, and the like.[22] Palmer emphasizes the fundamental artificiality of the Latin written language from beginning to end. Even the plays of Plautus, about 200 B.C., reflect an intentionally stylized language. The colloquial elements themselves are a conscious part of the highly-worked whole.

During the next 200 years a literary language of great conciseness of structure was created, which culminated in the classical style of the Golden Age, and in the works of Cicero, Caesar, Virgil, Varro, and Sallust. Scholars have traditionally held that this language was once spoken as well, in cultivated circles, among the noble and well-educated. While classical Latin was clearly based on the popular spoken tongue it had already diverged from it to a considerable degree. We will probably never know for certain the extent of this difference, but we can get a glimmer from the nature of the colloquial forms which do occur in Plautus, and to a lesser extent in Terence. Robin Lakoff notes:

In the case of Latin, we must also constantly bear in mind that classical Latin was, in fact, probably never a spoken language. The plays of Plautus, written about 200 B.C., are written in a language that is demonstrably closer to the language of Petronius, who wrote about A.D. 50, and also to modern Romance languages than it is to the language of Cicero and Caesar.[23]

If the classical Latin of Cicero and Caesar was already so removed from the spoken language of Plautus' day, we can imagine then the situation in their own day when the elapsing of almost two centuries had produced even further changes in the spoken language. Notice here that the greater the distance argued to have existed between the written and spoken languages, the less radical the changes in the development of the spoken language from ancient to modern times become.

In the first centuries after Christ, the use of the Latin literary language had been extended over the entire area of the Empire, where it served as a lingua franca to overcome the difficulties in communication which the ever increasing regional dialectal differences in spoken (Vulgar) Latin gave rise to. Among others, Rebecca Posner has given a very lucid account of the evolution and gradual differentiation of the spoken Latin language both before and after the fall of the Empire, and its ultimate transformation into the modern Romance languages of today, while at the same time the Latin written language failed to keep step with this evolution, serving a certain useful role during the transitional period but ultimately proving inadequate as a literary language in the period of clear national differentiation.[24]

Why didn't literary Latin change along with spoken Latin, as it might be expected to? Certainly in part it was because it had been made into such an effective instrument of expression in the classical period. It had an immense prestige as the language of classical culture throughout the Middle Ages, when culture was everywhere in retreat. The spoken language was, conversely, held in a certain amount of contempt as "vulgar." As already mentioned Latin served in communication, especially written communication, as a lingua franca, a koine among individuals from different parts of Europe whose spoken languages were unrelated or, though related, were now different enough that communication was difficult. The phonology of literary Latin was not as static as its grammatical form. Thus in the Romance bloc the letters of a Latin word tended to be pronounced with the value that they had in the speaker's own mother tongue, giving rise to certain national features of Latin pronunciation, such as exist even today, although these differences did not prevent successful oral communication.[25] But by far the most powerful force maintaining the place of Latin as the literary language was the Christian Church. After the decline of the secular schools in the fifth century, the transmission of ancient culture in Europe was in the hands of the Church with its system of ecclesiastical education, exclusively so in the sixth to eighth centuries, and education was in Latin only. Even though the number of those literate at certain periods was extremely small, perhaps only a few thousand--thus much fewer than in China at even her darkest hours--as the sole transmitters of the tradition, the sole bearers of written culture, their influence was very great. There was, after all, no other written language, no other education.

The Church fathers themselves did much to perpetuate the use of Latin. Since Latin was the language of the liturgy and of Church dogma, its prestige matched that of the Church itself. Christian Latin became something of a special language in itself. J. Schrijnen and Christine Mohrmann, who have studied this language thoroughly, have emphasized the importance of the early Christians, beyond what their small numbers would suggest, as a tightly-knit social group which evolved and propagated its own special style.[26]

The role of Latin as an oral lingua franca within the Church itself has been perhaps exaggerated. Although there is a sizable body of extant correspondence in Latin between bishops in far-flung parts of Europe, whose spoken tongues can be imagined to be quite different, already in the early synods there is evidence that a considerable part of the business was conducted in the vernacular tongues.[27]

It would be wise to look more closely at the nature of education in Latin, because it is through this education that users could achieve their competence. Of the three subjects of the trivium and the four of the quadrivium the very first was grammar. The beginner would typically learn a work like Donatus' Ars Minor by heart, then his Ars Maior; this might be followed by Priscian's Institutio Grammatica.[28] By the time the pupil had mastered the basic structure of Latin, he would proceed to develop his own style through several standard techniques.

Roger Ascham, who wrote a popular guide for Latin teachers in the sixteenth century, describes one of these, translatio linguarum or "double translation," in detail.[29] By this technique the pupil would translate a passage from Tully (for this is how the English referred to Cicero) freely into his mother tongue, then after an interval of at least an hour, translate the translated passage itself back into Latin. The teacher would then go over the pupil's work together with the pupil, juxtaposing it with Cicero's himself and pointing out the relative strengths and weaknesses of the two pieces. Imitatio (shaping one's own style upon the best classical models), paraphrasis, epitome, etc., were other techniques, but Ascham, at least, puts his stock in double translation. And the model was almost exclusively Cicero, although to a lesser extent Caesar, Sallust, and Varro were so employed.[30] It was important to be exposed only to the best. Note also that such an education, which laid heavy stress on imitation and discouraged creativity of expression, except for only the most advanced students, tended to keep change to a minimum, at least when education was successful, and to reproduce the competence of classical Latin in each new generation. In its imitativeness and exposure to good models, such an education unconsciously resembled the experience of the small child acquiring his native tongue. Had it permitted more freedom of expression, the history of literary Latin would no doubt show more change over time than it did.

Erich Auerbach, particularly in his two essays entitled "Latin Prose in the Early Middle Ages" and "The Western Public and its Language," has given us a fine view of the history of literary Latin in the years after the fall of the Roman Empire.[31] By the sixth century Latin had become more and more exclusively a literary language--of people who spoke another language which they could not write.

The leading classes of society possessed neither education nor books nor even a language in which they could have expressed a culture rooted in their actual living conditions. There was a learned language, and there were spoken languages which could not be written, there was no language of general culture.[32]

Everywhere a fresh layer of Germanic conquerors had to be absorbed and everywhere elements of the spoken idiom forced their way into the literary language. There was an increasing provincialism everywhere. The secular schools, which had trained young men for the civil service up through the fifth century, were gradually disrupted by barbarian invasions.

Thus from the sixth to the eighth centuries the only educational institutions were Church schools. The clergy, in facing the difficulties of teaching written Latin while at least in the Romance countries the vernaculars were slipping further and further away, made numerous compromises with the spoken language of their time.[33] This is reflected in the many vernacular forms in the liturgy and in written documents.

We can detect two general directions in the style of authors of the early medieval period. Classicism in the Latin world of this time meant a retention of the rhetorical complexity of Cicero or Sallust in contrast with the perceived crudeness of the Christian texts. For example, in the sixth century Cassiodorus, Venantius, and Fortunatus retained a trace of the classical tradition but their writings are very mannered. In contrast there grew up a utilitarian, plain prose style, much closer to the spoken language in sentence structure, tone, and word choice, represented by the writings of Gregory of Tours or Pope Gregory the Great. Their choice of this style was deliberate, and not the result of incompetence. Gregory the Great showed that he still had the ability, required by classical rules, to vary his style according to the subject matter. He wrote his theological works and letters in a more literary style, but for his Dialogues he chose a very popular form.[34] It was his contemporary Gregory of Tours (538?-593), though, who strikingly forged a new style in his History of the Franks, which drew heavily upon the spoken idiom:

Many turns, many word meanings, much of the rhythm, especially in the frequent direct discourse, were unquestionably taken over directly from the language that he heard around him and himself spoke every day of his life. Still, needless to say his written Latin is not identical with the language spoken at the time. For all the wealth of material that historians of Vulgar Latin have found in his work, he could not, in particular, take account of the colloquial phonetics, for that would have required a system of transcription, and besides, it never entered his head to break with the tradition in this respect. Moreover, his material obliged him to employ words having little currency in the spoken language and he sometimes resorts to a concentrated sentence structure, seldom used in speech and in those days probably never. Yet a good deal of the popular language is there. As far as possible, with the means at his disposal, he committed the spoken language of his environment to writing; it is certain that almost everyone could understand what he had written and that his readers felt no appreciable change from their accustomed idiom. From this point of view one can hardly ask for more in a written language.[35]

Auerbach appears here to mean by "their accustomed idiom" the spoken language of Gregory's time. But if we bear in mind the type of colloquial language reflected by Plautus over 700 years before and its characteristics, already suggestive of modern Romance, then we might imagine that Auerbach has been conservative in estimating the changes which had taken place in the spoken language, and thus overstated the closeness of Gregory's unmistakably Latin language to the spoken language.

If Gregory of Tours had been imitated by others, Auerbach feels that perhaps there would have developed a literary language based on Vulgar Latin, a "Romance koine." But no one did, and instead there came the Carolingian educational reforms in the ninth century, with their emphasis on education in correct Latin, whose effect was to cut off literary Latin from the spoken Romance tongues for good. But before examining the far-reaching effects of the Carolingian reforms, let us look at a few of the ways in which Latin itself had evolved by this time.

The influence of the spoken language on literary Latin was two-fold. Authors used forms which were actually part of colloquial usage. They also used fictitious and inappropriate forms in an effort to avoid what they felt to be new spoken forms, and write more correct Latin. Thus, as an example of the former tendency (after the distinction of the classical Latin diphthong ae had been lost), forms like que and eternus occur for quae and aeternus; as part of the latter tendency, there occur many forms like diae, aei, aemitto, prosequaere, quaem, etc. in medieval texts.[36] When much of the Latin case system had lost its effect, the mechanical assimilation of many phrases without understanding of their grammatical construction led to their frequent misuse. For example, the use of aquarumve decursibus, where ibus had clearly lost its effect.[37] Even when an attempt was made to master the classical syntax and rhetoric, as by Lupus of Ferrières (d. 862), a piece can be distinguished from a real classical text on the basis of its language alone. In the work of Lupus there is a more liberal use of personal and demonstrative pronouns, and an unclassical use of several words (not to mention all the Church-related expressions).[38]

From the seventh century there is a definite regional character at times to Latin texts. Hence we often see the preposition da (< de ab) in Italian texts, while in Gaullic texts the preposition apud replaced cum in order to avoid confusion with the homophone quomodo > com.[39] Change was most evident in Gaul, which is just what we would expect from the great distance between Latin and modern French. In Italy, Latin was much closer to classical models for a longer time; but evolution was still visible. In the early centuries the tradition was perhaps most faithfully followed in Spain.[40] In the non-Romance area such as the British Isles, for example in Ireland, the needs of Latin learning were different, primarily ecclesiastical. Latin there had to be learned from scratch, and only with great effort. As we might expect, the earliest Latin texts were on the one hand replete with strange expressions, showing no sensitivity toward stylistic value or proper choice of words, and on the other hand a more scholarly character than in the texts of the continent.[41] Pronunciation, for example, was more conservative. Germany, which otherwise had close ties with the British Isles, borrowed its Latin civilization from Gaul. The Carolingian reforms had the effect of improving education in Latin for use in the liturgy, in administration, and in literary writing. There came a new understanding of the structure of classical Latin, a new desire to correct "corrupt" forms, but with it came a heightened awareness of the gap which now separated the spoken tongue from classical written Latin. No doubt some of the "clerks," newly awakened to classical civilization and disturbed by this gap, tried to write down the vernacular language.[42] Now Latin took on its exclusively learned and literary character, never again lost.

There were economic factors which contributed to the use of the emerging Romance vernacular literary languages: greater prosperity, more freedom of travel and commerce, increasing population, more social life in the cities, and so more participation in intellectual and artistic movements. Each of these trends enhanced the power and prestige of those who used the new vernaculars, and the Latin-using clergy and scholars were increasingly eclipsed. Later, as well, the spread of printing in Europe gave the vernaculars further impetus. No longer was the transmission and dissemination of texts dependent solely on the clergy, and the greater availability of written materials permitted an expansion of the potential audience, to include people of more modest economic means who were much less likely to have received a Latin education. There was a new awakening, a budding European, post-Roman consciousness that expressed itself in reform within the Church, a new art form, the broadening experience of the Crusades, and new developments in philosophy and jurisprudence.[43]

It was the Carolingian educational reforms which made people conscious of the need for a new written medium. Latin was an inadequate medium for the newly emerging popular literature. None of what was written in a vernacular language before the end of the twelfth century was meant to be read. Rather it was to be sung or read aloud before an audience,, What better medium for an oral literature than the "vulgar" spoken tongue? The audience was broad-based, since the enthusiasm for classical purity which dominated Latin education now made it more difficult than ever to master the language. It was composed of all those who did not read and understand Latin with ease, including, in addition to the common people, nearly the whole nobility and much of the clergy itself. The predilection of such people for tales of adventure and love was easily satisfied, and their everyday preoccupations readily found expression in this early Romance vernacular literature, since there was no class educated in the vernacular to impose its elevated tastes. Thus adventure stories, love poems, fantasies, and other matter at odds with the learned character of Latin came to be written in the "Roman" tongue.

In the years after the reforms, despite a decrease in the grossnesses which sometimes mark the Latin of the earlier centuries, many innovations are apparent.

Still it would not be correct to call [Latin after the tenth century] a dead language; even before 1100 this was not the case, and afterwards still less so. Medieval Latin was extremely varied from the first, and from the eleventh century on it produced an abundance of new and living forms. The language was not static except for the spelling and morphology and these only relatively so. The vocabulary, syntax, style level, and versification were so richly varied that one can speak of many different worlds of medieval Latin.[44]

We continue to see regional differences. The meaning of the same word sometimes varied widely from country to country. Miles meant here simply "soldier," there "lord," or again "knight." Words could be borrowed from the Latin of one country into the Latin of another. A form current in Paris was often taken by students to other countries. For example, we see the suffix -agium in the Latin of non-Romance countries: hommagium, linguagium, passagium, villagium, This suffix originated in France where Latin -aticum became -age (hominaticum > O.F. hommage > Med. Lat. hommagium).[45]In the Romance countries scribes, compelled to use current expressions by their subject matter, racked their brains to try to give them a more Latin garb: for old French mesnil (< mansionile) there occur forms like mesnillum, maisnile, mansile, etc., in addition to the correct etymology. In the Latin of the non-Romance countries there are many words of foreign origin. For example, in England: schopa (= shop), daywerca (= daywork), laga (= low), and the like.[46]

Influence was by no means restricted to that of the spoken language on the literary language. Influence in the opposite direction, from the literary language on the spoken language primarily took the form of learned borrowing, and has occurred in all countries at all times. An unusually productive example is the French borrowing from Latin oculus 'eye' and its derivatives. Aveugle 'blind' is a very early borrowing, held to have come in the Merovingian period from a postulated Latin form *aboculus, the origin of which is uncertain but is perhaps from *ab oculis 'without eyes' or again, through dissimilation, from *alboculus ( < albumoculi 'cataract'). Another borrowing took place in the sixteenth century when oculaire 'eye piece, finder (on a camera)' was adopted from Latin ocularium, itself a new creation. Both of these borrowings contrast with the directly inherited word oeil 'eye' which goes back regularly to Latin acc. sq. oculum itself.[47]

As in the early medieval period Latin writers like Fortunatus and Gregory of Tours had represented the opposing poles of classical purism versus pragmatic adaptation to the needs of expression, now in the late medieval period, there were again two main trends. One was Scholastic and dialectical, specialized and "scientific." The other one was pre-humanist and rhetorical, classical in its orientation and broadly liberal. The Latin of the Scholastics was a rapidly growing medium of syntactic simplicity, stylistic monotony, and greatly increased technical vocabulary. The episcopal schools of the eleventh and twelfth centuries had stressed classical literature, but by the end of the twelfth century the schools of Italy, France, and their neighbors had adopted a course of instruction which emphasized logic, mathematics, and general science, as a preparation for professional training in theology, law, and medicine. Grammar was an elementary study, through which the pupil merely obtained the ability to write and speak correct Latin. He had to master a rapidly growing and technical language, thus the fine points of classical style tended to be neglected.[48]

The humanists and their predecessors from the twelfth century on, in contrast, rejected all the Latin tradition after the classical period as corruption. They stimulated a great new interest in antiquity, and with it came a new appreciation for the subtleties of classical Latin. The Latin literature of the twelfth century reached such a perfection that it is sometimes difficult to distinguish examples of skilled stylists like Abelard, Pierre Damien, and Salimbene de Adam from classical examples themselves.[49] But the very success of the Renaissance and its humanism was ultimately a disaster for Latin. The demand for classical perfection meant Latin could only be a language of imitation--and first writers, then scholars in turn rejected it as too limiting. But it took on a new function as the basis of the liberal education, which it was to keep right up to the present. "Thanks to early humanism and up to Petrarch [in the quattrocento] Latin ceased on the one hand to be a specialized language of the schools. Though still restricted to a minority of the elite it tended toward the far more universal scope which the classical literary languages had enjoyed in their prime. It became an organ of human self-cultivation."[50] Later children learned Latin not for erudite or creative ends but as a foundation for self-cultivation. They expressed their culture in their mother tongue.

The first wave of purism, during the time of the Carolingian educational reforms in the ninth century, had stimulated the rise of the Romance vernaculars, which continued to develop under advantageous economic and social conditions in the ensuing centuries. Now as a result of the second wave of purism, culminating in the sixteenth century, writers and scholars were led to embrace these already well-formed vehicles of expression and develop them further. The decline of classicism and the rise of romanticism in the eighteenth century only further stimulated this trend. Romanticism brought with it an enthusiasm for primitive languages, local dialects, and "vernaculars" of any kind.[51]

Yet Latin continued to hold an attraction among scholars as an instrument of international communication. When Renaissance Italians wrote in Latin it was not only, as Jacob Burckhardt said, "because their minds were imbued with humanism," but also because they sought a wider audience. If a man like Blondus of Forli had written in the dialect of his own Romagna he would have consigned his influence to oblivion; but, writing in Latin, he influenced the entire European world of learning.[52] Goethe, writing at a time when few any longer wrote in Latin, expressed nostalgia for the days when Latin enabled the European intellectual community to commune as they no longer could.[53] In his time there were still conservative areas where the authority of Latin was supreme as the language of culture. As late as 1848 Latin was used in the parliament of Hungary, which was remarkable since Hungary was not a Romance-speaking country, not even an Indo-European-speaking country.[54] In our own era literary Latin is long gone as a tool for communication. But its convenience in the medieval world has not been forgotten. The mathematician Norbert Wiener, in The Human Use of Human Beings: Cybernetics and Society in 1950 echoed Goethe's dissatisfaction with the linguistic fragmentation of the modern scholarly world. Ignoring for the sake of his argument all the other linguistic and extra-linguistic factors which contributed to the decline of literary Latin, Wiener places the blame squarely on the shoulders of the purist Latin teachers:

The effects of misplaced grammatical purism are to be seen well outside of the schools. First among these, perhaps, is the way in which the Latin language, like the earlier generation of classical gods, has been slain by its own children. During the Middle Ages Latin of varying quality, the best of it quite acceptable to anyone but a pedant, remained the universal language of the clergy and of all learned men throughout Western Europe. . . . [The] continued prestige of Latin was made possible by the willingness of writers and speakers of the language either to borrow from other languages, or to construct within the frame of Latin itself all that was necessary for the discussion of the live philosophical problems of the age. The Latin of Saint Thomas is not the Latin of Cicero, but Cicero would have been unable to discuss Thomistic ideas in the Ciceronian Latin. . . . with the coming of the Renaissance, the artistic standards of the Latinists became higher and there was more and more a tendency to throw out all post-classical neologisms. . . . the training necessary to wield such a delicate and refined tool was beyond that which would be incidental to a scientist. . . . The result was that the people who taught Latin and the people who used Latin became ever more widely separated classes. . . . In this vacuum [the teachers] ultimately eliminated any function for themselves other than that of specialists, and as the specialty of Latinism thus came to be less and less in general demand, they abolished their own function. For this sin of pride we now have to pay in the absence of an adequate international language far superior to the artificial ones such as Esperanto and well suited for the demands of the present day.[55]

III. Literary Chinese: A Historical Sketch

In China, a written language some distance removed from the spoken tongue developed very early. In the language of China's oldest written remains, the oracle records inscribed on turtle shell and animal bone of the Shang 商 dynasty, at the end of the second millennium B.C., we see a written style undoubtedly close to speech, but already exhibiting the laconic, abbreviated character which was to become the hallmark of all later literary Chinese. Several scholars have pointed out that the great scarcity and expense of the tools of writing, as well as the difficulty inherent in the writing system itself were forces encouraging economy.[56]

When the tribe called Zhou 周 conquered the Shang and established their own dynasty around the beginning of the first millennium B.C., they gradually adopted the Shang culture as their own, and with it the Shang language. It is possible that the Zhou people spoke a language which was no more than a dialectal variant of the Shang language, but we have no reliable textual evidence to determine what it was like. The texts of the Zhou dynasty (eleventh-third centuries B.C.), from the early bronze inscriptions in the Western Zhou (eleventh eighth centuries B.C.), through the Spring and Autumn Period (eighth-fifth centuries B.C.), and culminating in the written style of the classical language in the Warring States Period (fifth-third centuries B.C.) of the late Eastern Zhou, show a continuous development from the language of the Shang with limited dialect variation. Only a few isolated passages in early Chinese texts reflect what may have been the influence of the Zhou rulers' own tongue, or the influence of other early Chinese linguistic substrata.[57]

Just as was the case with Latin, there has been a considerable amount of debate on the relation of the written language of the preserved texts of the classical age to the spoken language of the time. Scholars like Wang Li, Zhou Fagao, Bernhard Karlgren, Henri Maspero, Hu Shi, Zhou Zumo, Ishikawa Tadahisa, and Tai Jingnong, in writing about this issue, have tended to emphasize the fundamental similarity of the spoken and written idiom.[58] Others, like Herrlee Creel, the early Guo Moro, and Homer Dubs, have in contrast emphasized the features which distinguished the late Zhou classical language from normal spoken language.[59] While our understanding of the evolution of the Chinese language remains primitive, there is little hope of being able to resolve the dispute to everyone's satisfaction. It is worth noting, however, that those who argue for a close relation of the written to the spoken language are in effect arguing that a considerable evolution has taken place in the spoken language up to the present, tantamount to the difference between the language of a classical text and a contemporary written text based closely on speech.

On the other hand, the other side, which views the classical written language variously as a "telegraphese," a shorthand in which superfluous redundancies are omitted, or at least a highly stylized, artificial medium as classical Latin appears to have been, has left up in the air the question of what the spoken language of classical China, that is, Old Chinese, looked like. But since the written language is likely to have been more conservative, we can assume that they view the spoken language as having already evolved further on the path toward modern Chinese. We might even have a situation in which the spoken tongue of that time resembled modern Chinese more than it did the artificial literary language, analogous to the situation in Latin that Robin Lakoff suggested.

The latter view receives modest support from studies of earlier stages of the Chinese language, when there was no question as yet of a stylized literary instrument. In the language of the oracle bone inscriptions one is startled to find cognates of a few words in modern Chinese which scarcely appear in literary Chinese. For example, the ancestor of the modern preposition zai 在 'in, at, on (etc,)' sometimes appears before place names and time expressions just as it does in modern Chinese, whereas yu 于 ( 於 ) appears in these places far more frequently in literary Chinese. In addition there is evidence in the language of the oracle bone and bronze inscriptions to suggest that an embryonic copulative verb (then written variously as 隹 , 惟 , etc.) which precedes its predicate, as in English or modern spoken Chinese, was already more fully developed than is reflected in the written language of classical China, which more often used instead a sentence final particle ye 也 to mark copulative sentences, a construction which left no trace in the later spoken language. But Karlgren has scored for the other side with an ingenious observation. He points out that in many of the classical texts the same particle ye 也 also occurs as a proper name marker (like the der in colloquial German der Fritz, he notes), always in direct quotation. There are scores of occurrences and after names in various grammatical functions. Moreover, other than this difference, the quoted passages and the other, narrative and descriptive, passages are entirely consistent in grammar, vocabulary, and other respects. He argues that, if the written language is a highly stylized, reduced form of the spoken language, we would expect this absolutely unessential element to be eliminated too. Since it was not, we can conclude that the texts in the literary language were not reduced or altered in this way, and that "the literary language was a normalized and moderately stylized reproduction of the colloquial in educated circles," entirely analogous to the relation of the language of Jane Austen's Pride and Prejudice to the spoken language of eighteenth-century England "brushed up into a literary or at least a semi-literary style,"[60] The important point here is that the grammatical system and fundamental vocabulary of the spoken and written languages were the same. Karlgren stresses that the far greater phonetic diversity of that period, the abundance of sound distinctions which are no longer made in modern, particularly northern, Chinese meant that the language was much clearer to the ear, less ambiguous, than it would seem from pronouncing it as if it were modern Chinese.

The majority of scholars today seem to favor the position that spoken and written language were close. A major reason they do this is that it is easy to detect a difference in grammatical usage of certain particles between works which derive from different periods or different regions. Their line of reasoning appears to be this: if there had been an artificial literary language already divorced from the spoken language at that time, then we would expect to see instead a uniformity in the use of these particles like that believed to exist throughout the later period, when the written language was indeed artificial. But since there is no uniformity we can conclude that in the classical period of late Zhou spoken and written language were in fact close.

For example, Huang Liuping (Xiang Xia) cited Xunzi's 荀子 frequent use of the conjunction an 案 'then, consequently' (like ze 则 or nai 乃 ) which is totally absent in other works of approximately the same period (third century B.C.E. ). On the other hand, the writers Han Yu 韩愈 (768-824) and Yao Nai 姚鼐 (1731-1815) are separated by a thousand years, but their syntax and their use of such grammatical particles is almost identical (or so Huang claims).[61]Karlgren has concurred that not even in the third century B.C.E. when literary activity was at a peak was there a standard literary language uninfluenced by the author's own dialect. Thus even though an author was familiar with, say, the pronoun usage of si 斯 'this' in the Confucian Lunyu 论语 (Analects), he didn't necessarily adopt it for his own writing.[62]

In this view the regional, as well as colloquial, flavor of the classical period was lost in the standardization and regularization of the Qin 秦 (221-206 B.C.E.) and early Han dynasties. Whereas dialects of late Zhou written Chinese can be discerned, the literary language from this time was polished and perfected into a more or less unified style. The traditional Chinese view seems to have perceived this process of stylistic refinement getting under way quite early. Tai Jingnong traced the steps in the process in texts from the sixth to the third centuries B.C.E. According to him the Lunyu is relatively colloquial, with only a little polishing, and so is Mengzi 孟子 (Mencius); the language of Guoyu 国语 and Zhanguoce 战国策 is somewhat more stylized and refined, and Xunzi is written in a polished literary style of considerable artifice.[63] Clearly he sees the literary language taking a definite polished form somewhat earlier than many others today. In 1926 Karlgren too had spoken of a "general third century B.C. literary language" in which Xunzi, as well as Zhuangzi 庄子 , Lüshi chunqiu 吕氏春秋 , and Hanfeizi 韩非子 were written.[64] With the increasing sophistication in our understanding of the grammar of classical Chinese, it appears he later postponed that development to the beginning of the Han. In any case it was the style of these works of the classical period, especially Mencius and Xunzi, as well as that of the Zuozhuan 左传 which were the most influential models for later literary prose style.

By the end of the Later (Eastern) Han dynasty, around 200 C.E., the spoken language had evolved to the point that there is no disagreement that the written language of the time, which was still based on the models of the fifth to third centuries B.C.E., 500 or more years in the past, had now become a stereotyped and artificial style used by writers who spoke a quite different language.[65] The spoken language had been continuously evolving. The written language, however, had not.

What forces contributed to the arrest in the development of the written language during the Han dynasty? The main ones were as follows: the prestige attached to the classical language as the vehicle of expression for the Confucian canon, which was China's reigning orthodoxy from the Han to the twentieth century; the educational system; the civil service examination system for entrance into the bureaucracy, which was the political bastion maintaining the Confucian orthodoxy; a unique script which permitted a considerable gap to arise between the spoken and written languages without the phonological contradictions between the two becoming obvious; the value of the literary language as a written lingua franca to bind together the reaches of a large empire, far larger than any previous Chinese state, with formidable ethnic, linguistic, and geographic diversity; even the burning of the Confucian texts by Emperor Shihuang 始皇 of the Qin dynasty has been cited as a cause.

Foremost among these is doubtless the tremendous influence of Confucianism, fully as great as that of the Christian Church in the history of literary Latin. Chinese have often characterized themselves as a fundamentally conservative people, who "favored the past and slighted the present" (hougu bojin 厚古薄今 ). One of the chief manifestations of this backward-looking spirit was the reverence for the sages of the ancient past and a great reluctance to depart from the model of behavior, including linguistic behavior, that they had set. As Han Yu said:

In applying myself to guwen 古文 [ancient prose] I am not of course merely attractedby its diction being different from that of today. One cannot make the men of old appear by thinking of them. In attempting to follow the ancient way one needs to be conversant also with their language. In being conversant with their language, one's mind is basically on the ancient way.[66]

Works linked to Confucius himself, or to later Confucians like Mencius, formed a canon of sacred scriptures. The language of the canon itself became sacred. Bernhard Karlgren attributed this power of Confucianism to the famous burning of the books:

But then this natural and promising literary evolution [of the fifth to third centuries B.C.E.] was violently broken by the book burning. After this incident conditions were radically changed. The early literature that was saved and particularly the Confucian classics won the martyr's crown and became the object of the greatest veneration. And as a consequence of this the language of its various documents was imitated by later authors.[67]

It is significant here, though, that not all the texts linked to Confucius were equally influential in the formation of a stable literary language. The various pre-classical texts which Confucius was traditionally believed to have edited -- the Shijing 诗经 , Shujing 书经 , Yijing 易经 , and Chunqiu 春秋 -- had very little influence in the formation of literary Chinese.

The power of Confucianism alone was not sufficient to enthrone any work as a model of style, but those already stylistically sophisticated works of the classical period which bore the ideological imprimatur of Confucianism especially commanded the respect of later writers and compelled them to emulate their language. The force of numbers was also in favor of the later works. There were a great many works written in that proto-literary language of the third century B.C.E. which won the title of "classical." In comparison the handful of texts handed down from the early years of Zhou rule, essentially those enumerated above, could only appear as archaic relics, no matter how weighty their content.

The handmaidens of the Confucian ideology were the examination system and traditional education. Concerning the role of the examination system in preserving the prestige of a literary language based on classical Confucian models, Hu Shi wrote convincingly and from first-hand knowledge.[68] He claimed that already by the time of Emperor Wu 武 of the Han (reigned 140-87 B.C.E.) the literary language was difficult for the minor officials to understand. It was necessary for young men to be selected from around the empire and sent to the capital for a year's special training in this language. Thus was formed a special cadre of privileged literati. Their number grew as time went on from 50, to 100, then 200, 1000, and over 3000 by the time of Emperor Cheng 成 (reigned 32-6 B.C.E.). As long as one could read and write the language of the classics, he could qualify to be an official, to be exempt from forced military and corvée labor service and eligible for all the perquisites of the privileged elite. This practice was later formalized in the examination system, whose single most important demand on the candidate was familiarity with the Confucian texts. With only rare interruptions it was to continue until 1905. Hu feels that this more than anything else was responsible for the continuation of the literary language and the slowness of development of a "plain speech" baihua 白话 style more closely tied to the spoken tongue. As evidence he notes that in the eighty or so years when the Mongol conquerors suspended the examination system in the Yuan 元 dynasty of the fourteenth century, baihua colloquial literature witnessed a vigorous development even among the educated elite, only to be subordinated again to the prestige of the literary medium when the examination system was restored in the Ming 明 dynasty (1368-1644). Thus the examination system was an essential prop for the influence of literary Chinese.

The educational system which fed candidates to the bureaucratic examinations bore many similarities to the instruction in Latin in Europe. Rote mastery of "classics" of style took precedence. Creativity or conscious innovation was undesirable. Young pupils would memorize texts like the Qianziwen 千字文 (The Thousand Character Classic), which presented traditional culture and morality in four-word phrases, while never repeating a word for its entire length (needless to say, its grammar is sometimes so forced as to invite a conclusion such as Professor Herdan's, that anything goes in literary Chinese); or the Sanzijing 三字经 (The Three Character Classic), which similarly represented traditional homilies in lines of three syllables each. The main purpose of such works was to impart a certain basic literacy, the ability to recognize and pronounce a few thousand characters (or logographs), upon which to found a study of the Confucian texts themselves. These texts, in their various combinations like "The Four Books and the Five Classics" sishu wujing 四书五经 , were the heart of the linguistic education. Memorized, recited over and over, chanted aloud and imitated, not only the structure but also the very spirit of the works was absorbed. And this was essential if one was to succeed in the bureaucratic examination and have the door to official life and elite status opened to him. We have no evidence that pupils, at least before the Tang (618-907), were exposed to any other written style than the classical language or its imitations. Even after the Tang such exposure was possible only furtively, and completely outside the educational process.

As a lingua franca literary Chinese played a role similar to that of Latin in Europe. There the spread of the Roman Empire over Europe had resulted in a great linguistic diversity within its boundaries. Latin could serve to overcome linguistic barriers. In China the rapid growth of the empire under the Han dynasty probably produced a similar need. Undoubtedly many areas came under Han control which spoke foreign tongues. Just as in Europe these foreign tongues, even as they became supplanted by the dominant one, Chinese here or Latin there, formed substrata which would influence the development of the dominant language differently where different foreign tongues had been spoken. Thus arose the different languages of the Romance family, each with signs of its own non-Latin substratum, and thus also arose the languages, or dialects, of modern China. In such a situation a common medium of communication was needed to overcome barriers, at least among those who traveled abroad. But in China we must distinguish the oral and written forms of that medium, because the nature of the Chinese writing system has imposed certain limitations on its users as well as granted them certain remarkable freedoms.

Although in Europe of the Middle Ages literary Latin was an oral lingua franca, literary Chinese never served a similar oral role. In the non-Chinese areas, Korea, Japan, and Vietnam, literary Chinese was never a spoken medium.[69] There was little need for such a medium because personal contact with foreigners speaking unintelligible languages was minimal, even while the influence of Chinese culture was extreme. In China itself the need was confined to the "scholar-officials" of the bureaucracy, the counterparts of the ecclesiastics of Europe, who traveled throughout the country and were often appointed to posts in districts where their native dialect did not suffice for communication. What these men did was master a streamlined form of the language of the capital, if they were not already speakers of that language. Since the capital was located in different places at different times in history, the dialect/language which enjoyed this privileged position also shifted about. This language, which the scholar-officials apparently acquired informally rather than as a part of their earlier schooling, could then be used for oral communication with officials from other parts of the empire, as well as with local subordinates, who learned enough on the job to get by.

Beginning in the Jin 金 and Yuan dynasties in the North (twelfth-fourteenth centuries) and continuing through the Ming and Qing 清 (1644-1911) dynasties, this oral lingua franca was in phonology and grammar based on the northern Mandarin dialect of Beijing (Peking), which was then the capital. From about the end of the Ming it was called guanhua 官话 ("officials' speech") since, although many of the growing number of itinerant merchants also acquired the language as they needed it on their travels, the major source of interest was still among those officials born and raised in the provinces. The English word Mandarin derives from the Portuguese word mandarin 'official' (< mandar 'to order') which missionaries used to translate this term guanhua. In the Qing, this language was sometimes called zhengyin 正音 ("correct pronunciation") and demand grew so great that schools, called zhengyin shuyuan 正音书院 , opened to give instruction in it. Handbooks of correct pronunciation were given names like Zhengyin sheyao 正音摄要 ("Getting the Essentials of Correct Pronunciation") or Zhengyin juhua 正因咀华 ("Savoring the Essence of Correct Pronunciation").[70]

Why couldn't literary Chinese have served this function? The answer lies in the genius of the Chinese script, which ultimately gave literary Chinese a staying power far surpassing that of Latin.[71] For literary Chinese had no unified pronunciation, Chinese of different areas pronounced literary Chinese fully as differently as they did their own spoken tongues. Since it was phonological differences which were primarily the cause of mutual unintelligibility among different areas, literary Chinese, which served nicely as a written lingua franca, couldn't surmount the communicative barriers any better than the vernacular dialects, even though it had a relatively unified grammar and lexicon.

The Chinese of 4000 or more years ago invented a "logographic" writing system in which it is morphemes that are represented, and the phonetic shape of the morphemes is indicated in only a subordinate way that makes each logograph phonetically flexible. The meaning of a logograph, or "character," may be understood by the eye, but it is read with varying pronunciations in different places and different times. Thus the orthographic shape of Chinese has not changed fundamentally in over 2000 years while its phonetic shape has been transformed over and over in successive epochs. Even today a modern Chinese can and does read an ancient work in his own pronunciation--in fact he can do nothing else without special linguistic training. A native of Beijing reads in Mandarin; of Shanghai, in Shanghai dialect; of Guangzhou (Canton), in Cantonese; of Changsha, in Hunan dialect, etc. Bernhard Karlgren cited the example of an edict posted about the country to illustrate the effect of the flexible phonetic nature of the script:

Thus an edict issued in Peking could be read and understood everywhere in this vast country, but the Cantonese read it aloud in a way that sounded utter nonsense to the Pekinese. And this is not all. Though a Pekinese read off the edict and pronounced every single word just as it is pronounced in his dialect, another Pekinese, who was listening but not looking at the characters, could not understand a single sentence if the reader followed the text word by word [because of the historical phonetic simplification of syllable types which had taken place].[72]

While this polyvalence is a major advantage of the Chinese writing system, a concomitant handicap is that, even if he wanted to, a speaker could not traditionally write down exactly what he spoke, in distinction with what everybody else spoke, since the writing system did not allow for it. In fact only lexical items which were shared with the literary language could be written at all, and even for shared items there was no way to indicate that the word was to be pronounced in a certain way, The devices used in the baihua vernacular tradition, which began to take shape in the Tang dynasty, for recording speech were usually ad hoc and subject to the discretion of the individual author. They were not standardized until recent years, and even today the conventions for transcribing the nonMandarin dialects, lying outside the baihua tradition, are only in a rudimentary state of sophistication. For speakers of these dialects the appeal of Hu Shi and others during the early twentieth-century cultural revolution to write as one spoke was meaningless, or at least unattractive, if that meant developing an entire system of representation hitherto nonexistent. I believe that the lack of convenient alternatives to literary Chinese is what is behind the greater popularity, even today, in comparable age groups, of the literary language among scholars who are not Mandarin native speakers than among those who are. (This is an impression yet to receive statistical confirmation.)

So as to better understand the importance of the Chinese writing system as a culturally conservative force, let me draw an analogy, albeit a hypothetical one, with Latin. If an American were to use a process similar to a Chinese to read a Latin text, he would look at the word strata but see his old friend "street" instead, and pronounce and understand it exactly like "street," Or he would look at the sentence Gallia est omnis divisa in partes tres but see, and pronounce, "Gaul is omni-divided in parts three. This is admittedly slightly strange, but the approximate meaning does indeed come through. More important, it doesn't look foreign, only somewhat garbled in the transmission. If such a method of reading were available to us for the great majority of Latin words and sentences, it is easy to see that the Latin classics would even today have a modest currency that they have lost. Chinese have had just this ability, and it has played such a prominent role in giving continuity to China's cultural history that even experts in Chinese historical linguistics have emphasized the similarity of different stages of literary Chinese, rather than its evolutionary development. Karlgren provides a typical statement:

The literary language has been an artificial thing for a thousand years and more, and for all its stylistic variations it has been essentially the same throughout the ages. Once a Chinese had succeeded in mastering it, it was the same to him, from the point of view of the language, whether the poem he was reading was written at the time of Christ, a thousand years later, or yesterday; it was just as comprehensible and enjoyable in either case.[73]

Granting that literary Chinese has been "essentially the same" for the last 2000 years, I would like now to examine some of the ways in which it has varied, far more than just in style alone, during that time, and the causes for that variation. As was the case with Latin, we see a reciprocal influence at every stage between the literary and spoken languages. Just as literary Latin showed a considerable capacity to grow, despite the fact that it was a language without native speakers, literary Chinese also grew as it absorbed forms from the spoken language. The users of literary Chinese could no more escape the influence of their own native speaker's competence than could the users of Latin. But the actual nature of the changes which took place in literary Chinese are only now coming under the scrutiny of linguists interested in the history of the Chinese language. Naturally, in China, as was true in Europe, linguists have been more concerned with reconstructing a picture of the spoken language of each period on the basis of hints found in the literary language--innovations, intentional or unintentional, which betray the writer's own native speaker's competence--than with portraying the literary language itself as a system in evolution. But such hints are often only individual slips, and may have no bearing on the regular evolution of the literary language itself, the changes in which came about instead through the participation and conscious acceptance of the great majority of users at any given time.

To the extent that the literary language was ultimately influenced by the underlying changes in the spoken language, the two avenues of approach differ mainly in focus, but even the former has not been thoroughly pursued, so that there is as yet no systematic, detailed account even of the history of spoken Chinese which can serve as a guide. (Professor Wang Li's Hanyu shigao 汉语史稿 is an important milestone on the way.)

As a result, there is a popular belief that the literary language has remained identical through the ages, or at most differed only in style. One had only to choose his model and he could make an exact replica. One hears the dubious claims that the famous late Qing scholar and translator Yan Fu 严复 used the style of the late Zhou philosophers to translate Darwin's On the Origin of Species, or that the early twentieth-century literary conservative Lin Shu 林纾 used the style of Tang chuanqi 传奇 'tales of marvels' to translate the works of Western novelists like Dickens.[74] (Being a conservative he knew no foreign languages, but worked closely with collaborators who did; they would translate into an intermediate colloquial Chinese, from which he would translate into literary Chinese.) No doubt Yan Fu or Lin Shu did try to imitate their models very closely, but their works reflect the modern native speaker's competence in every line. The single most important feature of the history of literary Chinese is its eclecticism. In each period and in each locale some of the spoken forms of that period or locale have become accepted in the literary language. Each of these additions took its place then as part of the common fund of expressions which later writers could draw upon at will. The resources of late literary Chinese are as a result many times the size of the language in its formative period during the late Zhou and early Han, "Hence arose a fearful grammatical confusion," writes Karlgren, " . . . and so all the various grammatical auxiliaries used in the older literature came to be used promiscuously. The great syncretism (in grammar and vocabulary) which characterizes literary Chinese from the Han period onwards was the direct outcome of the book burning and the ensuing sanctity of the earlier texts."[75] It is this "promiscuity" of the language, as Karlgren calls it, which gives the impression of an essential sameness to the works of different periods.