Foreword: The Ancestry of Chinese: Retrospect and Prospect
序言: [译]：中国语言祖先研究的过去与将来
William S-Y. Wang 王士元

Abstract 摘要

These pages result from a two-day Symposium held at the City University of Hong Kong2 in July 1994.

The word “Chinese” in the title of this volume refers to the group of dialects, which project back at least 3500 years. For the even earliest texts inscribed on oracle bones and bronze vessels show a relation to modern Chinese lexically, syntactically, and orthographically. On the other hand, diversification into the extant modern dialects probably took place only after China was unified under a central authority, and southward migrations began on massive scales. The terms “Chinese”, “Sinitic” and “Sino-“ are therefore quite appropriate since they all derive from the name of the dynasty, the Qin3, from which time these events took place.

The word “ancestry”, however, is not as clear-cut. Obviously part of its meaning has to do with determining which languages are related to Chinese. However, if we accept the monogenetic view in linguistics, then all the 6000 some languages of the world are related to each other. Our enterprise becomes interesting only if we refine our goal by requiring grouping the languages into monophyletic units.

A group of languages is monophyletic if and only if all its members are maximally related within the group, and no language outside the group has this property. Such groups of languages can be grouped into higher-level monophyletic units, which become increasingly inclusive as we go up in the hierarchy. In this sense, our goal in studying the ancestry of Chinese is reached only when we reach the highest level of such a hierarchy that we can justify.

The familiar tree diagrams which depict these hierarchies constructed on the basis of successive subgroups go back well over a hundred years, when Charles Darwin used them for biological species, and August Schleicher used them for languages. However, very shortly after the use of trees was introduced in linguistics, it was pointed out that languages typically behave in a way, which species typically do not: languages imitate each other when they come into contact. Thus, the wave theory suggests that languages which are closer geographically also tend to be more alike.

Genetic material comes only from parents. A language, to be sure, also has material inherited from earlier states of the same language; but it also has material introduced via imitations of other languages. The similarity we observe between languages typically derives from both sources. To know the true ancestry of languages, we face the problem of how to sort out one form of similarity from the other, the inherited which come from within from the imitations which come from without. Only then can we begin to quantify degrees of relatedness and justify monophyletic units. The problem is an extremely difficult one since it appears that anything in one language can be imitated by another language, though not all with the same level of facility.

Retrospect

Language study in China has a rich tradition, which reaches back some 2000 years, with brilliant achievements to its credit4. However, the mainstream of this scholarship largely remained focused on the texts of the ruling groups, with relatively little attention paid to the languages of the neighboring minorities. I have looked in vain in the early Chinese literature for discussions on the ancestry of the Chinese language.

Early scholarly writings in Europe on the Chinese language have been reviewed by Watters [1889]. Thus, we find observations on the Chinese writing system as early as Francis Bacon [1561-1626]5. There were a variety of theories on the ancestry of Chinese, some more fanciful than others. One theory held that the language was “invented all at once by some clever man to establish oral intercourse among the many different nations who inhabited that great country which we call China.” According to Watters [p.4], this brand of special creationism was accepted by no lesser a figure than Leibnitz [1646-1717].

Opposed to this view of Chinese being man-made, there were also theories aimed at relating the language to a biblical scenario. An influential essay by Webb, published in 1669, argued for Chinese to be the first language, spoken in the Garden of Eden. Chinese has also been variously identified as the language of Noah, as well as with each of his sons: Ham, Shem and Japhet. Such fanciful proposals would be more understandable if we recall that it was around this time James Ussher [1581-1656], Archbishop of Armagh, was calculating when the world was created by adding the genealogies given in the Bible.

The dominant views of the 19th century, when evolutionary thinking was already in the air, typically held that the Chinese came from the west, from Mesopotamia, though some preferred Egypt as a source. That the ancestry of Chinese should be traced to Babylonia was put in no uncertain terms by Lacouperie, who was then Professor of Indo-Chinese Philology at University College in London, and president of the Royal Philological Society:

“China has received it[s] language (since altered) … from the colonies of the Ugro-Altaic Bak families who came from Western Asia, … which emanated from Babylonia and was modified in its second focus. This general statement is now beyond any possibility of doubt, for the evidence in its favor is overwhelming.” [Quoted by Watters, 1889:11]

Unfortunately, a century and more later and in spite of Professor Lacouperie’s emphatic assurance, we are still very much in doubt as to how to trace the ancestry of Chinese. The Ural-Altaic hypothesis has never garnered much support. But there was no dearth of alternative hypotheses to choose from, some offered by well-known European scholars. None of these hypotheses can be taken seriously now.

These early investigations had no access to many of the advances and increased sophistication in linguistic methodology that have taken in place in recent decades. Neither could they have foreseen that archeology was to reveal many Neolithic sites in China, which date thousands of years before Bishop Ussher’s Garden of Eden. Perhaps Professor Lacouperie can be excused for his touch of intellectual hubris.

Even more importantly, with these advances there has accumulated a much richer body of data from neighboring languages to compare with Chinese, especially with older stages of reconstructed Chinese. The question is no longer, whether Chinese shares some similarities with language X or language Y, as it was seen during the last century.

Rather, we need to know which of these similarities are due to inheritance and which are due to imitation. The criteria for sorting these similarities one from the other, as I noted earlier, are not yet well understood and far from uniformly accepted. And if indeed the similarities are due to inheritance, the next question is whether Chinese forms a monophyletic unit with X, with Y, or with both [if our tree allows nonbinary branching]. Questions of degree arise here, and Baxter’s paper, which begins Part I of this volume, is an explication of the probabilistic reasoning that must underlie hypotheses of subgrouping.

Indeed, anyone who thinks that we have answers to these questions at present, which are “beyond any possibility of doubt”, will not have read this volume carefully. In Gong’s paper, we have further verification of the Sino-Tibetan hypothesis, with strong support from Tangut evidence that has not been incorporated before. The narrow version of this hypothesis which Gong considers here, including just the Chinese dialects and the Tibeto-Burman languages, is the closest we can come to a consensus at present; but see the comments by Sagart in Part II of this volume.

Attempts to posit higher monophyletic units for Chinese, however, do not as yet command nearly the same degree of consensus. These include the similarities to Indo-European, observed by Pulleyblank, the connections to the North Caucasian and Yenesseian languages, posited by Starostin, and the relation to Austronesian, proposed by Sagart.

These hypotheses are debated by Blust, Li, Pulleyblank, Starosta, and Starostin, mostly in Part II of this volume. This section also includes some remarks from Meacham, who provides a useful archeological perspective.

The papers by Pan, You and Zhengzhang all endorse a wider circle of genetically related languages; besides Chinese and Tibeto-Burman, they would include Miao-Yao, Kam-Tai, Austronesian, and perhaps some other yet unaffiliated languages. They explore different methods to arrive at their groupings, from word families, to animal names, to basic lexicon respectively.

Pan and Zhengzhang use the term “Sino-Austric” or “Hua’ao” to designate this far-ranging phylum they posit. If their hypothesis is correct, however, it is doubtful that Chinese would rank high enough in the tree diagram to warrant being included in the name of the root node. Chinese probably split off from the rest of the tree and came to prominence quit late, several millennia from the root. Its success story is not unlike the great spread of English in Indo-European or Bantu in Niger Kordofanian.

Prospec

Which of the connections advocated in the following pages will stand the test of future research, and how do they fit into a hierarchy of linguistic groups? Though it is useful to explore these hypotheses, we are far from any definitive answers at present. Yet answers to these linguistic questions are necessary if we are to proceed to the larger interdisciplinary topics of [1] dating the various stages of linguistic development, [2] placing the various prehistoric communities on a map, and eventually, [3] being able to say something about the culture, the society, and perhaps the mentality of peoples who have passed away six, seven, or even eight thousands years ago

This last topic is one where linguistic reconstruction is particularly well situated to make primary contributions, since the vocabulary an ancient people used in their daily lives can offer a fuller and more enlightening view than just their material remains unearthed by the archeologist. In discussing the reconstruction of the Proto-Indo-European words for their gods, Watkins wrote that, “The reconstructed words *deiw-os and *dyeu-peter- alone tell us more about the conceptual world of the Indo-European than a roomful of graven images.” [1985: xvii]

The point here is that the two disciplines have a great deal to offer each other. Clearly, we can achieve a much more complete picture of the past when both sets of data are taken into account, for complementation as well as for cross-validation.

Interest in dating linguistic divergence was stimulated in the 1950’s when Swadesh proposed glottochronology. Inspired by the discovery that physical objects can be dated by measuring chemical elements that have a constant decay rate, Swadesh’s original insight is the analogy that the basic lexicon in any living language is replaced at a relatively constant rate.

Over these decades, a great deal of progress has been made in refining the original ideas. In particular, several additional numerical methods have become available in the form of statistical software, which can be used on the personal computer, as reviewed by Wang [1994]. Although these methods were initially developed for purposes of biological systematics, their usefulness to linguistics is obvious and their application straight-forward6.

Until recently, all family trees that have been drawn to subgroup languages do not assign any quantitative value to the branches. The new methods allow us to let the length of each branch represent the duration of independent evolution of the monophyletic unit it dominates. Furthermore, the branches are additive in the sense that the distance between any two languages is represented as the sum of the branches along the shortest path between them. As these methods become more widely investigated in linguistics, and applied to many diverse language groups, we may find Swadesh’s original insight to be largely correct, even though his method is defective. If this turns out to be the case, as I suspect it will, then linguists and archeologists can cross-validate each other on dating as well as cultural interpretations.

Alongside of archeology and linguistics, a third partner for future research on the ancestry of Chinese is genetics. The potential for this collaboration was seen by Darwin when he wrote in Chapter 14 of the Origin of Species: “If we possessed a perfect pedigree of mankind, a genealogical arrangement of the races of man would afford the best classification of the various languages now spoken throughout the world.""

Much more is known now about the extinction, mixing, and replacement of languages, and these processes are certainly all abundantly attested in China.

In particular, assimilation to the language of the Hans, i.e., Sinification, has been a powerful force over the past two millennia. It is clear that all languages within the Chinese sphere of influence have assimilated to Chinese, though to varying degrees. Furthermore, it must also be the case that the Chinese dialects are rife with numerous strata of words which are imitations from neighboring languages over the millennia; this fact is unfortunately obscured by these words all being written in Chinese characters7. These factors all taken together, the correlation between genes and languages can not be nearly as straight-forward as Darwin had envisioned.

Furthermore, to get the gene-language correlation to work at all, it is critical that the appropriate genetic markers be selected. The one comprehensive genetic study of China that has been reported so far [Zhao and Lee, 1989], based on Gm and Km allotypes, actually shows an opposite result. It reveals that the Hans are virtually always closer genetically to their non-Han neighbors than to other Hans further away on the map. In other words, no Han unity can be detected on the bases of these two allotypes.

Nonetheless, the gene-language correlation appears to be more consistent on a global scale [Cavalli-Sforrza et al, 1994]. Here we are dealing with a much grander span of time, since the whole world is included instead of just China, and many more genetic markers were included in the study. Whatever the explanation will eventually be, it seems to me we can learn a lot about the past not by identifying linguistic history with genetic history, but by comparing the two and trying to explain the congruence as well as the incongruence.

The fundamental fact is that children in all societies typically learn their first language from their mothers, from whom they inherit half of their genes as well. And the probability is high that the father speaks the same language since one would not choose a mate one cannot communicate with. So a positive gene-language correlation should be the null hypothesis, following Darwin’s remark. The factors, which reduce this correlation, including the ones mentioned above, are evolutionary events that should be studied in conjunction with the language histories.

If the best attempts at uncovering the ancestry of Chinese, as exemplified by the scholarship contained in the following pages, are seen to fall considerably short of the goal, I would like to think it is because the theory of the past we wish to build is still lacking in crucial data. If nothing else, the Symposium has been invaluable in calling forth a variety of new linguistic data to be examined, digested, and integrated into a synthetic framework

But my perception is that we also need ideas and data from the other areas that are concerned with similar questions. I have mentioned archeology and genetics above, others also readily come to mind. Physical anthropology can examine the fossil remains and tell us something about the characteristics of the speakers themselves8. Comparative ethnography can help us group peoples on the basis of their customs, myths, and beliefs.

We need to be cautious, of course, in drawing inferences across bodies of knowledge – people get displaced and languages get replaced. It would be foolhardy, as an obvious example, to believe that the Liangzhu culture situated in Jiangsu and Zhejiang some 4000 years ago has any direct links with the Wu speakers there today. Nonetheless, the ideas and data from each body of knowledge, judiciously interpreted, can provide a unique window on the past. And the combined view from all these windows allows us to complement and cross-validate our perspective, so that ultimately our knowledge of the ancestry of Chinese can be reconstructed on a broad and secure foundation.

Lastly, it remains to acknowledge the various contributions which made this volume possible. The Symposium was funded in part by a grant from the Chiang Ching-Kuo Foundation for International Scholarly Exchange. Benjamin T’sou was most gracious and effective in providing logistic support from the host institution at which the symposium was held.

I thank the twelve authors for sharing their expert knowledge with us in this volume. Most regrettably, You Rujie of Fudan University and Zhengzhang Shangfang of the Institute of Linguistics in Beijing were unable to join the Symposium, due to circumstances beyond their control; their papers were transmitted through the courtesy of Pan Wuyun of the Shanghai Normal University. No doubt the Symposium would have been richer still had they been able to come.

Weera Ostapirat had the responsibility of turning the Symposium papers into a unified volume. The task of reconciling the numerous differences in fonts, scripts, and spelling conventions is a daunting one. Our guideline is to get the monograph published as accurately and as quickly as possible, while the memory of the Symposium is still fresh, even at the cost of a rougher appearance. Thanks to Weera’s tireless dedication, and to the assistance of Shen Rongqiu of JCL and of Shi Feng of Nankai University, this monograph has become a reality.

My hope is that the ideas and the data presented in these pages will do much toward defining the state of the art for research on this question, and toward stimulating and guiding future work. Clearly, the ancestry of Chinese is a question of central importance for any theory of linguistic evolution, for human prehistory in general, and for the genesis of Chinese civilization in particular. Perhaps the next Symposium on this question will broaden the base of discussion by including other disciplines as well.