Review by
John Makeham
China Review International (Spring 1999)

This review is here presented as edited by Thomas Carlson with the permission of the reviewer; it contains some additions [in brackets] and some omissions. The review also considered the Ames translation, of which the reviewer had a lower opinion; we omit that section here.

Few books - especially those already available in multiple translations - have the distinction of being translated four times within a two-year period. [Chichung Huang (1997), Simon Leys (1997).] The Lun yu, or Analects, is such a work. Of course, being a historical text, the Lun yu has been garnering commentary and interpretation for two thousand years. What distinguishes the two most recent translations-cum-studies of the Lun yu - the subjects of this review - is the effort the authors have made to move beyond the constraints of traditional interpretation. The first of these, The Original Analects: Sayings of Confucius and his Successors, by Bruce and Taeko Brooks, is the most exciting study of the Lun yu yet published in a Western language. Its potential implications are monumental, ranging from a rewriting of our understanding of Warring States texts and classical philosophy generally to major revisions in our understanding of early Confucianism and the nature of intellectual transmission in early China.

The authors have been working on this and other Warring States (479-221 BC) texts for a quarter of a century, developing a sophisticated developmental chronology in which there emerges a pattern of intellectual interaction on a scale previously uncharted. Their findings have been honed as the result of the input and criticisms of an international network of scholars (largely based in the United States) who are affiliated with the Warring States Working Group Project (formally inaugurated in June 1993). The implications of their work to date demonstrate that the simplistic identification of a text with a single author and a single compositional date is, now more than ever, untenable. As for the Lun yu: it "contains only a core of sayings by the historical Confucius, to which have been added layers of attributed sayings and conversations invented by his successors to update their heritage, and to address the new needs of changing times" (p.vii). The theory used to explain how and why texts developed is an accretional theory, according to which texts represent the accumulated repository of changing advocacy positions of sponsoring groups over extended periods of time. The accretional growth of the Lun yu is further complicated by interpolational growth. In coming years, findings based on the application of the accretional theory to other Warring States texts will be published. As such, the study under review provides a test case for the appropriateness of this theory.

This book is not for the faint-hearted. Here I am referring not to its profound iconoclastic and revisionist implications, but rather to the demands it places on the reader. One must pay undivided attention to its details if the many nuanced threads of argument (developed in the author's commentaries, reflections, and appendices) are to be pursued and appreciated. This requires carefully following up the multiple cross-references to other Lun yu passages and to the all important Appendix 1. Unlike other English translations, this is no tome to take to bed - unless you are well armed with the Harvard-Yenching concordance and an array of other translations and commentaries for comparative consultation.

The translation follows the order of the chronology the authors propose for its accretional composition: Lun yu 4, 5, 6, 7, 8, 9, 10, 11, 3, 12, 13, 2, 14, 15, 1, 16, 17, 18, 19, 20. This order is held to represent "a consecutive record of the elaboration of Confucius' thought, and its interaction with other modes of thought, from the time of his death to shortly before the founding of the Chin [Qin] empire" (p.1). Authors/compilers are identified for each chapter (pian), as are dates of composition, ranging from 479 BC to 249 BC. Of the total 530 passages (zhang) that the Brookses identify in the text (compared to the 501 in the Harvard-Yenching text), 142 passages are judged to be interpolations. These interpolations are all given a date of composition. Formal evidence for identifying the interpolations is presented in Appendix 1. Generally, each chapter includes some introductory remarks, a running commentary, identification of section divisions within the chapter, identification of paired sayings, and a concluding set of reflections. The provision of the inter-passage running commentary is especially welcome. At last the reader is provided with a context in which flesh is added to the very spare bones of the text.

Appendix 1 sets out the accretional theory, in which the authors review earlier theories on the "evolutionary" nature of the text. Common to these other theories and the author's own theory is the identification of the chapter, either singly or in clusters, as the basic unit of accretion. The theoretical core of the book is presented in this appendix, and the more serious reader would be well advised to read it even before reading the translations. Here the formal criteria used for identifying paired sayings, interpolated passages, dating, and authorship are explained and defended. Appendix 2 identifies other developmental patterns in the Lun yu, the gist of the argument being that changes in the distribution of certain terms and ideas provide an independent set of criteria to confirm the authors' proposed chapter sequence because these changes could not have occurred "in the opposite direction" (p.257). Topics presented include those changes that evidence the gradual "aggrandizement" of Confucius, material and social developments, and conceptual and textual references. Appendix 3 provides a third set of criteria to confirm the developmental picture by identifying "historically meaningful conjunctions" (p.257) between the Lun yu and other Warring States texts: the Guan Zi, Mo Zi, and Zhuang Zi. Appendix 4 is a detailed and useful summary of information culled from early sources relating to Confucius, his disciples, and his ancestors and family successors. Appendix 5 presents chapters 1-4 in the received order, including interpolations, to show how this arrangement influences the way the reader will understand the text by privileging certain contents. In sum, the authors have constructed a powerful argument in which the Lun yu emerges not as the records of the school sayings of Confucius, but as a "history of early Confucianism" (p.1).

The authors succeed admirably in reducing a welter of complex material into an orderly, accessible format. References are made to a wide range of sources: sinological, biblical, historical, scientific, and literary. The referencing system used in the body of the translation is economical and minimizes distraction. Physically, the book is handsomely produced, enhanced by a selection of black-and-white reproductions of early Chinese artifacts, characters are interspersed in the text (but the original text of the Lun yu is not reproduced). Typographical and editing errors are few. The translations are generally competent, and the interpretations are frequently original. [ ]

The real meat of the book, however, lies not in its contribution to translation, but in the structuring that is identified as the framework for the accretional theory. If any single feature of the Brookses' study were to be singled out as the most significant it must surely be the identification of this structuring. Yet it is also this same feature that undermines the accretional theory in its proposed form. Given the implications of this theory - not only for our understanding of Confucius, the Lun yu, and early Confucian traditions but also for our understanding of early Chinese thought in general - it must be subjected to the closest scrutiny; the authors would readily concur. A review article is not the appropriate forum to conduct such a project; nevertheless, even a limited examination of the arguments underpinning this theory reveals problems. In what follows I am going to assume the role of the devil's advocate. In posing counterarguments my purpose is not to discredit the view that the Lun yu is composite or accretional. Rather, my aim is to question whether the structures identified by the authors and which support their particular accretional model are, in fact, persuasive. Different readers will read the text differently and will by no means agree with all of the points I make. If, however, even only a reasonable proportion of the criticisms I make are valid, then it seems that the theory, as it stands, is brilliant, but unpersuasive. Put in more dramatic terms, just as the author's tight-knit theoretical structure lends their interpretation such strength, so, too, the vulnerability of that same structure to even minimal recalibration exposes its greatest weakness: if even a few links are removed, the entire edifice threatens to collapse. The authors are well aware of this - hence Appendixes 2 and 3. Yet even these rely on the foundational support of the accretional theory argued in Appendix 1.

It is at this point that I must beg the reader's indulgence, for to follow my argument adequately will require access to both the Harvard-Yenching concordance and the book under review. The second requirement is, of course, all the more reason for readers to purchase a copy of the book if they have not already done so. The following examination is limited to the first five chapters of the reconstructed text, this sample being sufficient to identify structural deficiencies that are generic to the authors' overall thesis regarding the text.

The authors propose a three-tiered structure of the Lun yu text: (1) that pairing of sayings "characterizes the entire text." (2) "Complementing the paring principle is the section principle: pairs of sayings (sometimes ending with a single unpaired saying) tend to be arranged in sections with a thematic or other coherent identity... As far as we know, we are the first to propose that sectioning, like the pairing of sayings, is a pervasive structural device in the Analects" (p. 207). (3) Nearly all chapters are composed of 24 passages.

It is these three structural elements that lend the accretional theory tremendous intuitive appeal, as they seem to provide a solid frame from which to gauge compatibility and deviation.

Pairing and Sections

The single most consistent shortcoming with regard to the identification of paired sayings is that too often the basis for such identification owes more to the wording of the English 'synoptic paraphrasing' of individual passages than to the evidence of the Chinese text. In the following examples, I have listed only the most questionable examples, this does not mean that I endorse all of the pairings not here listed.

Chapter 4

4.5/6. Unconvincing pairing. "Desire" and "passion" are identified as the basis of pairing. This identification owes more to an association being made between these two English words than to the Chinese text. If anything, the theme that is shared in common is ren [rvn] (humaneness), yet, as a cluster, 4.1-7 all involve this theme. Thematically, 4.5 shares most in common with 4.1 and 4.2 (ren, abiding); 4.6 shares most in common with 4.3 and 4.4 (ren, like and dislike).

4.10. This unpaired saying lacks a thematic (the "way") or other coherent identity sufficient to mark this as a section.

4.11/12. Tenuous pairing. Passage 4.11 contrasts the gentleman with the little man and has a better match in 4.16; 4.12 simply warns against seeking personal advantage.

4.13. This unpaired saying does not evidence any particular thematic continuity with 4.11 and 4.12.

4.14/16. Tenuous pairing. Passage 4.14 addresses themes of office and reputation; 4.16 addresses characteristics associated with the gentleman and the little man. Formally and thematically, 4.16 has more in common with 4.11.

4.18, 19, 20, and 21. These are more readily identified as constituting a cluster of four thematically related passages rather than two separate pairs.

The findings above confirm not only the author's admission that pairing is "often based on trivial features" (p.207), but also the extent to which their rigid commitment to pairing as a pervasive structural feature compels them to make forced and subjective links between passages. In sum, there is no consistency in the way pairing is used to structure the Lun yu text. This being so, as a general principle, pairing per se cannot be appealed to as justification for identifying interpolations. While acknowledging that there is evidence to support the identification of pairing and periodic thematic clustering, neither is on a scale or pattern that supports the authors' thesis. The failure to show that pairing is a feature that "characterizes the entire text," in addition to the many unconvincing examples of unpaired section final sayings, also undermines the formal criteria employed by the authors for identifying sectioning as a "pervasive structural device."

The Twenty-four Passages Argument

Having identified putative interpolations in each chapter, the authors argue that nearly all chapters are composed of twenty-four passages. Those that do not are Lun yu 4 (sixteen passages), 8 (four passages), 1 and 16 (twelve passages each), 18 (five passages), and 20 (three passages). In the case of 4 and 8, later extensions are said to have been used to make up the standard complement. In the case of 1 and 16, it is suggested that "they may be an intentional halving of the then normal 24-passage form" (p.248). The shortfall of the passages at Lun yu 18 is explained by identifying nineteen (!) passages that the author of 18 had interpolated into other chapters. Thus the only real anomaly is Lun yu 20, which is explained as being "preliminary to the intended chapter form."

When were interpolations introduced into different chapters? The earliest examples are said to date from around 370 B.C., when the compiler of Lun yu 10 (identified as Zisi) is said to have begun making additions to 7 and 9 "as a way of maintaining consistency in a text which up to that point had simply accumulated, any internal conflicts being allowed to stand" (p.65). [Note: The skeptic might note that Zisi would only have seen Lun yu 4 in its sixteen-passage version and Lun yu 8 in its four passage version (8A).] From then on, interpolation became established practice. Yet if this is so, one wonders how the 24 passages formula was preserved if it was continually being hidden. Even assuming that there was some transmission mechanism that preserved a knowledge of the formula, why were earlier accretions not removed? Consider also the case of 9.1, which is treated as an interpolation and dated circa 360. Many ingenious attempts have been made to reconcile the claim made in this passage that Confucius rarely spoke of ren with the 108 references to the term in the standard text. The Brookses provide a ready solution: "The accretional theory of the text obviates such ingenuity by noting that rvn (ren) is common in the 05c layers but vanishes from LY 10-11. 9.1... is an attempt by the LY 11 people (for whom rvn (ren) was obsolete) to protect their li-based theory of Confucianism by denying the rvn basis of the original school" (p.76) Yet given that these same "LY 11 people" were the transmitters of the accretional text to that point, why would they not rather have simply expunged all or most references to the term? It would, after all, have been hard to deny the common occurrence of this term in the text at that point in its proposed transmission.

Chapter 4

The authors maintain that this chapter consists of only 16 original passages. While this in itself is inconsistent with the 24 passages argument, it is further claimed that 8 passages were later interpolated to achieve the requisite number. The formal criterion used to distinguish 4.1-17 from 4.18-25 (a block of passages treated as a "later extension") is that the focus of the former is said to be official while the latter is said to emphasize domestic and personal virtues. Yet this often rests on little more than subjective emphasis. Why, for example, could not passages 4.1-7 unproblematically be interpreted as being concerned with personal virtues as applied to interpersonal relationships? Similarly, why should passages 4.22-24 relate only to the domestic/personal domain of the gentleman as opposed to his public persona? There are certainly notable thematic clusters in this chapter, but none sustains the blanket demarcation between 4.1-17 and 4.18-25 underpinning the original-extension distinction.

[The 24 passages argument in chapters 5 and 6 is analyzed here in the review.]

Chapter 7

7.11 and 7.31. These are identified as interpolations on the grounds that they "are anecdotes with narrative change of scene, for which there is no earlier precedent" (p.214). Yet there is as much a narrative change of scene at 5.8, 5.19b, and 6.4 as there is at 7.11. More to the point, however, the argument that it is justifiable to remove a passage on stylistic grounds if there is no earlier precedent is both self-serving and inconsistent. The authors are arguing for an accretional model; thus, on the one hand, those innovations that are found in, say, chapter 5, but not in chapter 4, serve to confirm difference and hence accretional growth. Yet, on the other hand, we have the same rationale being used to argue that a lack of precedent is sufficient grounds for a passage's removal. In the case of this chapter, the cost is high: no less than 15 of the 39 passages that the authors identify in this chapter are classed as interpolations.

7.36 and 7.37. These are dismissed on no better ground than that they do not form a pair. Actually, using the author's own criteria, it would be quite easy to identify these as a pair on formal grounds, given that both are concerned with contrastive types. Of course, if they were to be recognized as a pair, this would contravene the Brookses' prescriptive formula that a chapter can only have 24 passages.

[Analysis of the 24 passages argument in chapter 8 omitted here].

Clearly, it is not the case that "24 sayings is the full complement for a standard chapter" (p.216). The Procrustean frame of 24 passages is unconvincing. This being so, then, in the chapters that I have examined, as a general principle, there is no justification for identifying a passage as having been interpolated simply on the basis of its being descriptive rather than a saying. Another problem is the way in which the authors date those passages they identify as interpolations. One of the texts unearthed at Guodian tomb no.1 in 1993, named "Yu cong," includes a passage that matches Lun yu 9.4 (although where the received text reads yi bi gu wo, the "Yu cong" reads yi gu wo bi). The tomb has been dated circa 300 B.C. (give or take a few years), so the "Yu cong" material cannot be later. The Brookses, however, assign a date circa 262 B.C. to the 9.4 passage. If the "Yu cong" passage is, in fact, a Lun yu quotation, this would leave them wide of the mark, even on the most conservative estimate, by forty years.

Appendix 2

I also have difficulty in concurring with some of the "contextual" evidence that is mustered in Appendix 2 in support of the argument that the authors' arrangement of the chapters is confirmed by "items of intellectual and material culture whose developmental direction is either intrinsically plausible of archaeologically attested" (p.249). This is a potentially powerful methodology, but as presented in Appendix 2, it is insufficiently/inadequately developed.

Confucius' Rank

Here a sequence of passages is adduced to evidence that early chapters portray Confucius as an "unsuccessful statesman," while later chapters show his career developing at "increasingly higher levels" (p.250). Counterexamples might include the following. (In citing them my purpose is not to actually argue that Confucius did hold office, but merely to show how the authors' developmental thesis is not without difficulties. Again, I will limit counterexamples to chapters 4-8.) On the evidence of 6.4 and 6.5, Confucius did hold significant office. At 5.8 Confucius is consulted by Meng Wubo about the appropriateness of three of his disciples for office. At 6.8 Confucius is consulted by Ji Kang zi about the appropriateness of three of his disciples for office. If Confucius was an "unsuccessful statesman," why would he have been consulted by these powerful figures? [ ]

Ren versus Li

The authors propose a developmental sequence in which li becomes more important than ren. The evidence adduced includes the identification of a cluster of passages on ren at 4.1-7 (which the authors date ca. 479 B.C.) and the two clusters of passages on li at 10.1-19 and 3.1-23. Yet this clustering is more suggestive of deliberate editorial arrangement of related themes. For the accretional theory to be persuasive, it would require a more random distribution of li in later chapters. It is also claimed that it is only in the fourth century B.C. that "an entirely new ethos, based on li, suddenly supervenes" (p.254). Yet li is specifically mentioned six times in chapters 4 through 8: 4.13, 6.27, 7.18, 7.31, 8.2, and 8.8 - as are a range of li related activities. (For reasons already outlined, 7.18, and 7.31 cannot be dismissed on appeal to interpolation.)

Courage: Too few examples are cited to be meaningful.

The Shr [Shi] and the Yi.

Regarding the argument "No text is cited or mentioned in the original sayings of LY 4, or in those of LY 5-6, which have [sic] some claim to have been written within the living memory of Confucius" (p.255), I am not quite sure what the point is. The argument from absence is, in any case, questionable, as there are "late" chapters that also make no mention of texts. Significantly, texts, reading, and the recitation of texts are referred to and cited in the "early" chapters 7 and 8, at 7.18, 8.3, and possibly 8.8. Some would also argue that 7.17 includes a possible reference to the Changes. The authors treat 7.18 as an interpolation on the grounds that it is descriptive, but, as I have shown, this alone provides insufficient justification for identifying interpolations.

In conclusion, I believe that the biggest methodological gamble taken by the authors is their insistence that the basic accretional unit is a chapter. An alternative would be to regard individual passages (zhang), rather than chapters (pian), as the basic unit of transmission. In the process of transmission, individual passages, pairs of passages, and clusters of passages would have been added, modified, excised, combined, and recombined according to the needs and interests of different advocacy groups and editorial hands at different periods. Nor would all of these individuals and groups necessarily have had to identify themselves as ru. (It also remains an open question whether the ru only transmitted material that was consonant with ru teachings.) This model would also help to explain the inclusion of identical or nearly identical passages and also the irregular nature of pairing and thematic clustering that is in evidence in the received text. Such an alternative model accommodates a far greater degree of fluidity in the transmission process than is allowed in the Brookses' rigid structuralist model. It also requires minimal, if any, rearrangement of the received text. As for the received text, it simply represents one particular combination of zhang and pian. [Note: for an argument that it was not until around 150-140 B.C. that the Lun yu came into existence as a book and that this book was based on a number of early "collected sayings" of the Master, see my Article, "On the Formation of Lun yu as a Book" Monumenta Serica 44, (1996) 1-25.] I cannot prove this hypothesis, but neither, as I have argued, have the authors proven theirs.

Yet even if one is finally left unpursuaded by their reconstruction, one cannot fail to be impressed by the tremendous intellectual creativity and passion that distinguishes this study as a tour de force of sinological virtuosity, both conceptually and in its execution. The rich and diverse possibilities their work opens up promises a sea of change in Lun yu studies.

John Makeham
University of Adelaide, Australia.

Back to Responses Index Page

22 March 2006 / Contact The Project / Exit to Publications Page