(Book presentation paper for BMBF workshop "Foundations of Excellence in East Asian Studies")
Odds on the Odes —
how Old Chinese became an (almost) natural language in 1992
If there ever was such a thing — and blessed we are that there is not! — as a Gallup (or even an Allensbach) poll conducted among sinologists interested in philology or historical linguistics and asking for the single most important work "in the field" this century, chances are that Bernhard Karlgren’s Grammata Serica (1940, Recensa 1957) would still come off "top o’ the heap". Indeed, the "GSR”, a rather unwieldy and arcane dictionary of Chinese characters encountered in Archaic and Ancient texts and inscriptions, ordered according to a principle of shared phonetic elements in sets of related characters, which only became tangible by a cryptic procedure of phonetic reconstruction, was certainly a true milestone of sinology. It was an awe-inspiring achievement, not only as a dictionary — the only dictionary of pre-modern Chinese, as some would say [1] — but also as a theoretical work on the organisation and stratification of the Chinese lexicon. A work, which still seems to be irreplacable after more than half of a century, so irreplacble in fact, that several rearranged and computerized versions of the book have been compiled during the late nineties. "Je moderner die moderne Welt wird, desto unvermeidlicher werden die Geisteswissenschaften" ——
Odo Marquard ("Über die Unvermeidlichkeit der Geisteswissenschaften", 1985)What then, was the motor behind its enormous success, the scientific and sociological mould out of which it grew? Could its author serve as a "role model" of excellence in the field of Chinese historical linguistics? Well, at first glance, hardly so, since Claes Bernhard Karlgren (1889-1978) appears to have been the quintessential negation of all those garish personal qualities that fill today’s academic job-ads and evaluation criteria catalogues. A scholar, who worked in self-sufficient isolation during most of his lifetime, entertaining hardly any direct communication, let alone cooperation, with other scholars in the field, but being, on the contrary, strongly inclined towards ad personam-arguments and rancorous hostilities directed at his very few competitors. An author, who published more than 90% of his work in an un-reviewed Swedish Hausorgan in excruciatingly laconic English prose and, more often than not, close to illegible layout. A man, without any teaching responsabilities to speak of, whose only confidante in real life was his very secretive secretary, and whose major (but socially still indecorous) pleasure was to write criminal stories under a pen-name! [2] I do not intend to wax nostalgic about Karlgren, the scholar, or, for that matter, equally towering and crotchety figures among his sinological contemporaries like Ritter von Zach (1872-1942), the "headsman of sinology" in Weltevreden (!), knjaz Budberg/Boodberg (1903-1972), the Pnin of Berkeley, or L.E.J.M. de la Vallée Poussin (1869-1938), the catholic Belgian Buddha — to name but a few of the more obvious quixotic cases. It is self-evident that chances of replicating their mode de travail are well nigh non-existant in any contemporary educational system. Yet I do wish to stress, that much truly excellent work in my own field was and continues to be produced in seemingly bizarre and rapidly obsolescing academic niches by enthusiastic, and, occasionally, slightly whimsical individuals, under conditions which could neither be predicted nor generated by smart research programs.
This is not to suggest, of course, that Karlgren came out of the void as some kind of inscrutable prodigy of sinological learning. He was firmly rooted in a long and glorious tradition of Chinese phonological, semasiological, and evidential research, which, in the form of Duan Yucai’s (1735-1815) Gu shiqi bu xiesheng biao [Table of phonophoric series in the old seventeen rime groups], had even produced a fairly well-known structural model for the compilation of the GSR. Adding to the "empirical" work of these Chinese predecessors the ideal of regularity of sound-change propagated by the Leipzig neo-grammarians, the principles of textual criticism emphasized by his teacher August Conrady (1864-1925) and the Pelliotian school of French sinology, as well as the notational devices of Swedish dialectology, Karlgren managed to create a methodology sui generis for Old Chinese reconstruction, which survived basically unchallenged until E.G. Pulleyblank’s deservedly famous revision and critique of 1962 [3]. For although there was no lack of attempts to refine, criticize, and improve the results obtained by Karlgren ever since the publication of his Études de la Phonologie Chinoise [4] and the Analytic dictionary of Chinese and Sino-Japanese [5], eventually leading to such influential revisions as Lu Zhiwei’s An outline of discussions on Old [Chinese] pronunciation [6] or Li Fang-kuei’s "new" reconstruction of 1971 [7], Karlgren’s "method", as well as the textual basis underlying it, survived basically unchallenged. This was by and large the state of affairs during the late seventies, an almost stalwart atmosphere perceived as a curious form of internecine scholasticism from the outside, and criticized, even recently, as an "endless rehashing of the same old data". Moreover, one might add with a sigh, it is a situation which still pertains in many quarters of East and Central Asian Studies dealing with Old Chinese data until the present day.
I will try not to overburden you with the manifold techincalities surrounding Middle and Old Chinese phonology and reconstruction. In order to understand the changes which occured during the eighties and early nineties, and to appreciate some of their far-reaching consequences, it will probably be enough to know, that the major ingredients for Old Chinese reconstruction are pronunciations recorded in Middle Chinese rhyming dictionaries, the rhymed parts of the Book of Songs (i.e., the oldest edited collection of Chinese poetry), the phonetical relationships obtaining between homophonophoric characters, and, to a much lesser degree, Modern Chinese dialects, Sino-Xenic transcriptions, or early loans to and from adjacent languages. Theoretical and empirical problems of Karlgren’s reconstruction procedure and corpus notwithstanding, any general linguist looking, for instance, at the vowel system posited by him for the Old Chinese stage (his "Archaic Chinese" of, roughly, the Book of Odes period), would immediately object that it shows a very skewed distribution of no less than fourteen main vowels along five vowel heights. Moreover, these fourteen vowels are supposed to be modified by three medials (-i-, -y-, -w-), based on the fallacious assumption that each and every rhyme category recorded in the oldest completely extant Middle Chinese rhyming dictionary (i.e., the Qieyun of 601 a.d.), would have to be reconstructed with a phonetically distinctive main vowel. A system like this is, to say the very least, rather unnatural, it runs against our "Proto-Sprachgefühl" (J.A. Matisoff), our gut feeling of "pronounciability", and inevitably results in the type of unpleasant throat disease, everyone who has ever attempted to read aloud an Ode from the Book of Songs in Karlgren’s reconstruction will be painfully familiar with. On a more formal plane, it is fairly easy to statistically demonstrate the oddity of the GSR-system today: Only 2,2% of the phonemic inventaries available in the so-called UCLA Phonological Segmentory Inventory Database (UPSID)database of 317 natural languages [8] have fourteen phonemic vowels, while none has a distribution of these vowels faintly similar to that envisaged by Karlgren. Suffice it to say, that similar problems of "unnaturalness" obscure the realm of reconstructed consonants as well.
Although many of these deficiencies have been known for a very long time, and perfectly feasable alternatives have been proposed, especially for the Middle Chinese (Karlgren’s "Ancient Chinese") stage, by E.G. Pulleyblank in a number of seminal articles, an important theoretical monograph [9], and a even a handy dictionary of Middle Chinese Readings [10], the fact that no concurring compendium of Old Chinese reconstructions in the form of a bilingual dictionary has been produced to date, explains to a large extent the continuing and widely unqestioned popularity of the GSR, faute de mieux.
The situation only changed with the publication in 1992 of William Hubbard Baxter’s A handbook of Old Chinese phonology [11], a massive tome of 922 pages, which was not intended to replace the GSR as a dictionary, but offered a systematic new reconstruction of the phonology of Old Chinese and, above all, the rhyming of the Book of Odes. The book is divided into ten sections and three appendices, which deal with methodological issues (1), the phonology of Middle Chinese (2), a probabilistic technique of analyzing Old Chinese rhyme data (3), the history of research on Old Chinese rhyming in China (4), a new reconstruction of Old Chinese (6-8, 10), the nature of the evidence represented by the Book of Odes and contained homophonophoric characters (9), an overview of sound changes between Old and Middle Chinese (App. A), and the rhymes and rhyme words of the Odes (App. B, C), respectively. It is the most explicit and readable of three independently produced new reconstructions of Old Chinese, strikingly similar in many crucial aspects, the other two by the Russian scholar Sergej A. Starostin [12] and the PRC phonologist Zheng-Zhang Shangfang [13]. Starostin’s book goes beyond Baxter in the sense that it includes an analysis of other rhyming periods after the Book of Odes stage and adds much controversial comparative Sino-Tibetan data, which are important for the reconstruction of the system of initials. However, it lacks an important element, which makes Baxter (1992) epistemologically interesting. This is a probabilistic model and computer simulation, used to test assumptions about the distribution of rhyming characters throughout rhyme groups. Until Baxter’s first studies in this direction during the early eighties, the question, crucial for reconstruction, of how (or even if) a given Shijing character rhymes, had always been decided by recourse to "common sense", i.e. by reference to a presumably auhtoritative reading tradition, which declared certain lines as rhymed and others as, not by simple fiat. Although it turned out to be rather difficult to include all relevant parameters such as corpus size, line and verse length, stock-phrase interference etc. into an integrated probablistic model of rhyming, which allows to draw reliable inferences about the regularity of rhyme occurences, and thereby the stability of rhyme classes, Baxter’s model eventually led to a reclassification of the rhyming characters in the Book of Odes, which is not only coherent in terms of the theory of phonological naturalness, but also statistically testable, i.e. "falsifiable" against the background of the actual Shijing corpus. It resulted in the confirmation of a six-vowel system for Old Chinese, and, along with the less controversial reconstruction of the initial system, made Old Chinese look like a perfectly natural language for the first time in the history of modern research on it. To be sure, there might be other, equally consistent and parsimonious accounts of the same data, as for instance those proposed by E.G. Pulleyblank, and neither in terms of theory, nor in terms of the selection of underlying Old Chinese data, the Handbook will be the last word on Old Chinese phonology. But the main attractiveness of Baxter’s work clearly stems from the integration of a notion of testability into reconstruction.
The immediate result of the ‘new naturalness’, which made Old Chinese phonologically similar to many modern spoken Tibeto-Burman languages, presumed to be genetically related, was a renewed interest in the genealogy of the "Sino-Tibetan" language family. It also provoked some radically new vistas on the interrelationships of Old Chinese with neighboring language families, traditionally thought to be unrelated. Based on comparisons in the framework of the six-vowel system, wider extra-Sino-Tibetan affiliations of Old Chinese have been proposed with Proto-Austronesian [14], and with North-Caucasian and Yeniseian [15]. Within "Sino-Tibetan", comparison to Himalayan (Bodic and Kiranti) languages has led to a radical reshuffling of the Sino-Tibetan Stammbaum [16]. And, beyond all genetical alignments, a fresh consideration of loanword relationships between Old Chinese and Tocharian, the easternmost Indo-European language in what was traditionally known as Chinese Turkestan, has been attempted [17]. If true, and to the extent where proposals are not mutually exclusive, these new linguistic genealogies would, of course, drastically alter our perception of Chinese ethnogenesis, with its corollary of such trifling matters as the emergence of Chinese civilization and the state during the late neolithic and eary bronze age, the question of the earliest Chinese contacts with the Ancient West and Near East, and, a fortiori, the roots of what might have constituted the "Chineseness of China" (if there ever was such a thing at all!).
Equally important within the field of Chinese linguistics, yet somwhat more "down to earth", and slightly less prone to nationalist abuse, has been the impact of Baxter (1992) on the study of Old Chinese morphology. Few myths about "the" Chinese language have been as successful and, one might well say, disastrous for its perception by linguists, philosophers of language, and the general public alike, as those about its alleged lack of word classes, its supposed total absence of morphology, and its presumably "isolating" or "analytic" monosyllabic structure. The origins of this seemingly inextirpable complex of linguistic clichees can be traced back to the tradition of missionary linguistics of the Jesuits during the 16th century, which entered the philosophy of language through Leibniz (1646-1716), and became more widely accepted since the work of Johann Christoph Adelung (1732-1806) in Germany and Jean-Pierre Abel Rémusat (1788-1832) in France in the early 19th century [18]. It reached its now classical epitomization in Wilhelm von Humboldt’s (1767-1835) ideas of a linguistically determined Weltanschauung, which in turn directly influenced the formulation of the Sapir-Whorf hypothesis of linguistic relativity [19]. It managed to excercise a decisive role in American ethnolinguistics, and was recently resurrected once more, more sinico, in Shen Xiaolong’s pertinacious agenda for a "Chinese cultural linguistics" (Zhongguo wenhua yuyanxue). Even outside the narrow confinements of linguistic theories and fashions, few western interpreters of Chinese culture, philosophy, history and behaviour from Hegel through Max Weber, from the French ethnosinologists since Granet to the neo-pragmatism of Chad Hansen, from Durkheim’s social psychology up to the current neuropsychological discourse about a genetically determined alterité of Chinese "cognitive styles", have been able to resist the temptation of basing audacious relativist claims on the presumed "deficiencies" of a language lacking more familiar patterns of wordhood and morphology. Given this momentous background, the decisive role of the new reconstruction of Old Chinese phonology in the study of its morphology can not be underestimated, and the first results based on it are already extremely promising [20]. Viewed from this new vantage point, Old Chinese would have had a quite intricate system of derivation by affixation and possibly even remnants of paradigmatic inflection or infixation [21]. In other words, Old Chinese became a somewhat more "natural" language in terms of morphology as well.
Before I come to a conclusion, let me briefly sketch what I perceive as some of the conducive factors behind the major success of Baxter’s book. First, and foremost, it displays an openness to view old data with new techniques or methods that radically transgress the boundaries of the traditional humanities. In fact, it is reminiscent of many approaches to "messy data" typically encountered in the natural sciences, quantitative psychology and sociology. The methodological proximity is especially striking in the case of evolutionary biology, where similar historical problems exist even in the same domain of application, i.e. the reconstruction of a part of human phylogeny or the taxonomy of species. One easily tends to forget that historical linguistics and comparative philology have grown out of the same champ épistemologique as natural history and modern biology (as well as the theory of political eonomy) during the 19th century [22]. It is thus no sheer coincidence that much of the recent interest in historical linguistics does not come from within linguistics, but from ancient DNA research in molecular biology and, to a lesser degree, archaeology. This is an interesting development which makes one feel hopeful about potentials of cross-fertilization in the not too distant future.
Another major element behind the success of Baxter’s work, as I perceive it, was simply time. Baxter had been working on his system already during the preparation of his dissertation submitted in 1977 [23], and he had published the core of his new system of reconstruction already in 1980. He cooperated on his system with his former thesis advisor, Nick Bodman, and their reconstruction was for a long time known as "the Bodman-Baxter system". One can feel on every page of the book that Baxter took his time in discussing, explaining, and finally producing it, much to the benefit of the non-sinological readership.
Third, some major ideas behind Baxter’s reconstruction were inherited from the great Russian sinologue and linguist, Sergej E. Jaxontov (*1923), who also influenced Starostin and Zheng-Zhang. It can not be stressed enough, therefore, how important a good command of an important secondary reading language like Russian or Japanese can be, and I am sure that there a still many sinological treasures to be recovered in the former Soviet Union.
Fourthly, a solid training in in general linguistics was certainly an important prerequisite for some sections of the book as well, but, incidentally, this training did not have to include close familiarity with the dominant schools of contemporary American phonology (i.e. generative, especially non-linear, autosegmental, and metrical phonology, and even less, more recent developments such as optimality theory). Berthold Laufer (1874-1934) once said, "The more theories will be smashed, the more new facts will be established, the better for the progress of our science" [24]. And although this might be somewhat exaggerated, his caveat against a precedence of theory over fieldwork and philology is still fully applicable to many institutionally dominant frameworks of linguistic inquiry today. The balance, found in Baxter (1992) between a theory-neutral yet theory-inspired approach to phonology, between very solid philological foundations on the one, and the strict application of a highly formalized staistical procedure on the other, strikes me as an especially happy one.
The consequences emerging from this example for the administration of long-term research proposals are, I believe, amply clear. "Conditions of excellence" are the possibility to work continuously and over a considerable period of time as an individual, or in small groups, if necessary, independent of what might be current trends, fashions or "discourses" in research, and without permanent interference of applicability and profitableness-demands bequeathed to institutions of higher learning from the increasingly insolent champions of neo-liberal economies. Interdiscplinarity of scientific inquiry can not be generated out of preestablished epistemological preferences. It does easily transgress traditional boundaries between C.P. Snow’s "two cultures", yet it will not necessarily lead to the establishment of non-subject specific overarching methodologies and hermeneutics in every case. On balance, then, it is my contention that "excellence" in the humanities is still best guaranteed by upholding the sublime utopia of polymathy, fully conscious, of course, of its inherent impossibility. Successful research in the humanities has always been rooted in an internationally oriented scientific culture, which in turn depends on an excellent command of foreign languages not belonging to one’s narrower field of study. And, in the case of Asian Studies as much as anywhere else, it will have to be anchored in an intimate knowledge of texts, artefacts and institutions of the particular culture studied, which can not dispense with philology and traditions, not even rigidities. Jede Bildung ist ein Gefängnis ... aber das Resultat ist eine wirklich gewonnene Freiheit (J.W. Goethe).
October 1999, Wolfgang Behr
Notes:
[1]
Cf. D.B. Wagner, A classical Chinese reader, Richmond : Curzon, 1998, introduction.
[2]
Cf. N.G.D. Malmqvist, Bernhard Karlgren: ett forskarportrett [Svenska Akademiens Minnesteckningar], Stockholm : Norstedt 1995.
[3]
"The consonantal system of Old Chinese", Asia Major IX: 58-114, 206-265.
[4]
4 vols., Leiden : E.J. Brill, 1915-1926.
[5]
Paris : Geuthner, 1923.
[6]
Guyin shuo lüe, written on scrappaper in prison, and published as Yanjing Xuebao Monograph 20, Peking 1947.
[7]
"Shangguyin yanjiu" ["Studies in Archaic Chinese"], The Tsing-hua Journal of Chinese Studies, n.s., 9 (1971) 1-2: 1-61.
[8]
On which see Ian Maddieson & Sandra Ferrari Disner, Patterns of sounds, Cambridge : University Press, 1984.
[9]
Middle Chinese: A Study in Historical Phonology, Vancouver : University of British Columbia Press, 1984.
[10]
Lexicon of Reconstructed Pronunciation in Early Middle Chinese, Late Middle Chinese, and Early Mandarin, Vancouver : University of British Columbia Press, 1991.
[11]
(Trends in Linguistics; Studies and Monographs; 64), Berlin & New York : de Gruyter, 1992.
[12]
Rekonstrukcija drevnekitajskoj fonologicheskoj sistemy,Moskva : Nauka, 1989.
[13]
"Shanggu yunmu xitong, he si deng, jieyin, shengdiao de fayuan wenti" ["The archaic system of rhymes and the problem of the origins of the four divisions, medials and tones"], Wenzhou Shifan Xueyuan Xuebao (1987) 4: 67-90.
[14]
L. Sagart, "Proto-Austronesian and Old Chinese Evidence for Austronesian", Oceanic Linguistics 33 (1994) 2: 271-308.
[15]
S.A. Starostin, "Gipoteza o geneticheskix svjazjax sinotibetskix jazykov s enisejskimi i severnokavkazskimi jazykami", in: Lingvisticheskaja rekonstrukcija i drevnejshaja istorija vostoka, pt. 4: 19-38, Moskva : Nauka, 1984.
[16]
G. van Driem, "Sino-Bodic", Bulletin of the School of Oriental and African Studies 60 (1997) 3: 455-488; "Black Mountain Conjugational Morphology. Proto-Tibetan Morphosyntax and the Linguistic Position of Chinese", in: Y. Nishi, J.A. Matisoff & Y. Nagano eds., New Horizons in Tibeto-Burman Morphosyntax (Senri Ethnological Studies; 41): 229-259, Ôsaka : National Museum of Ethnology, 1995.
[17]
A. Lubotsky, "Tocharian Loan Words in Old Chinese: Chariots, Chariot Gear, and Town Building", in: V. Mair ed., The bronze age and early iron age peoples of Eastern Central Asia (Journal of Indo-European Lingustics Monographs; 26), Washington, D.C. : Institute for the Study of Man & Philadelphia : University of Pennsylvania Museum Publications, 1998.
[18]
Cf. G. Ineichen,"Historisches zum Begriff des Monosyllabismus im Chinesischen", Historiographica Linguistica 14 (1987) 3: 264-282.
[19]
See, recently, J.E. Joseph, "The immediate sources of the ‘Sapir-Whorf hypothesis’", Historiographica Linguistica 23 (1996) 3: 365-404.
[20]
W.H. Baxter & L. Sagart, "Old Chinese word formation", in: J.L. Packard ed., New Approaches to Chinese Word Formation. Morphology, Phonology and the Lexicon in Modern and Ancient Chinese (Trends in Linguistics, Studies and Monographs; 105): 35-76, Berlin & New York : Mouton de Gruyter, 1998; L. Sagart, The Roots of Old Chinese, Amsterdam & Philadelphia : J. Benjamins, forthcoming.
[21]
For an interesting alternative view, not dependant on Baxter’s system of reconstruction, cf. also E.G. Pulleyblank, "From Archaic Chinese to Mandarin", in: G. Booij et al. eds., Morphology: a Handbook of Inflection and Word Formation, Berlin : de Gruyter, forthcoming.
[22]
Cf. Michel Foucault, Les mots et les choses : une archéologie des sciences humaines (Bibliotheque des Sciences Humaines), Paris : Gallimard, 1966.
[23]
Old Chinese origins of the Middle Chinese chongniu doublets: a study using, multiple character readings, Cornell University, Albany, 1977, unpublished.
[24]
In his review of R.H. Lowie’s then famous Culture and Ethnology (New York: D.C. McMurtrie, 1917), American Anthropologist, n.s. 20 (1918): 91; repr. in: H. Walravens ed., Kleine Schriften 2.2: 1202, Wiesbaden 1979.