This will provide an overview of the debates surrounding the textual composition of the Liji. For those wanting a more in-depth look at this issue, the best source in English is Jeffrey Riegel's dissertation, “The Four ‘Tzu ssu’ Chapters of the Li chi.” James Legge also retells the traditionally accepted account of the composition of the Liji in his 1885 translation of the text.1 In contemporary Chinese and Japanese, the lengthy prefaces to Hong Ye's Liji Yinde 《禮記引得》, Wang Meng’ou's Liji Jinzhu Jinyi 《禮記今註今譯》, Jiang Yihua's Xinyi Liji Duben 《新譯禮記讀本》, and Takeuchi Teruo's Raiki 《禮記》represent the state of the field previous to 1993.2 While there are dozens of premodern commentaries on the Liji, the three that I have found the most helpful are Kong Yingda's Liji Zhengyi 《禮記正義》, Wu Cheng's Liji Zuanyan 《禮記纂言》, and Wang Fuzhi's Liji Zhangju 《禮記章句》.3 In combination with contemporary scholarship, these commentaries provide glimpses into the study of the Liji at different points throughout Chinese history. Kong Yingda (574–648) wrote in the Tang Dynasty and included much of Zheng Xuan's (127–200) commentary, which is otherwise lost, but is the earliest known commentary on the Liji. Wu Cheng (1249–1333) wrote in the Yuan Dynasty and saw himself following after the great Neo-Confucians of the Song. Wang Fuzhi (1619–1692) wrote in the transitional period between the Ming to the Qing Dynasty—a period characterized by the call for a return to a purer understanding of the classics, which from Wang's view had been corrupted by mistaken commentators for over a millennium. I also found collected commentaries such as the Liji Daquan 《禮記大全》and the Liji Jishuo 《禮記集說》 to be of great assistance.4
Since the Guodian discovery, numerous articles and several books written by Chinese scholars have taken up the project of reexamining the composition of the Liji.5 The preeminent scholar in the field is Wang E 王鍔, whose Liji Chengshu Kao 《禮記成書考》takes into account the recent archeological finds.6 His work periodizes each chapter of the Liji by looking at things such as linguistic patterns, appearances of portions of the text in other texts with known dates of composition, (p.220) and archeological evidence. This book easily provides the best summaries of modern and premodern accounts of the compilation of the Liji, and presents a contemporary snapshot of the field. While his general conclusion is a little specious—he claims that the entire text was written, but not redacted, before the Qin—he represents a renewed excitement in the study of the Liji that is eager to challenge contemporary perceptions of early China in light of recent discoveries.
Debates about the textual composition of the Liji focus on two issues: the question of source material (What was the source material for each chapter? Who composed it and when?), and the question of redaction (How was the source material obtained? How was it edited? When did all of this occur, and who was involved?). The general consensus, excluding a few outlying opinions, is that the text has been relatively stable since the late second century CE.
Arguments over the first issue range from claims that the entirety of source material was written in the Six Dynasties (220–589 CE) to claims that the entirety of source material was written in the Warring States and late Chunqiu periods (770–221 BCE). Most scholars take up a position between these two extremes, seeing portions of the source material written in the Warring States, and portions of it written later in the Han.
Regardless of the position one occupies on this spectrum, the theory supporting it must account for several pieces of information. Internally, some chapters in the received text purport to be the recordings of particular historical events. The “Ai Gong Wen” 哀公問 chapter, for instance, records an appearance of Confucius before Duke Ai, who presided in the state of Lu from 494–466 BCE. Other chapters in the current text contain implicit claims of authorship. The seventh chapter, “Zengzi Wen” 曾子問, for instance, presents a series of exchanges between Confucius and his disciple Zengzi. Since the authors of texts in early China rarely identified themselves, it is not always clear who composed a text. However, all interpreters must confront the prospect that in the case of chapters such as “Zengzi Wen,” the text itself claims to be written by the figures described in the text.
External information that must be accounted for includes the fact that other early texts speak of a text, or texts, called Liji 《禮記》(or sometimes Li 《禮》), or they refer to recorded discourses (ji 記) on the topic of li 禮. Some of the texts that mention a text named Liji date from as early as the Chunqiu period (770–476 BCE), and a few of the references parallel chapters or passages in the received text. This problem is further exacerbated by the emergence of several texts bearing the term li 禮 in their title—all compiled in roughly the same time period as the Liji. Among the most prominent are the Da Dai Liji 《大戴禮記》, the Yili 《儀禮》, and the Zhouli 《周禮》.7 When early texts mention a text named Liji, determining which of these texts it refers to (if any), and whether or not they were discrete entities, is no simple matter.
(p.221) Additionally, any theory must take into account earlier theories—in particular, the claims of early commentators who may have also assisted in the process of redaction. Ma Rong 馬融 (79–166 CE), who some scholars believe appended the “Yueling” 月令, “Mingtangwei” 明堂位, and “Yueji” 樂記 chapters in the received text, for instance, attributed authors to certain chapters in his description of the Liji; oftentimes without providing reasons for so doing. Other difficult theories to account for include claims from some early scholars that a guwen Liji 古文禮記 was found in the wall of Confucius's home during the Han Dynasty and incorporated into the received text. It is also difficult to account for material from later commentators claiming to quote earlier commentators whose work is no longer extant.
Last but not least, any contemporary theory must account for recent archeological evidence. The Guodian find in 1993 and the bamboo manuscripts acquired by the Shanghai Museum, both dating to the late third century or early fourth century BCE, contain texts that closely resemble chapters of the Liji. The most complete example is the “Ziyi” 緇依, which is a part of both the Guodian and Shanghai collections. The received edition has character variants, a few extra passages, and a different ordering of the passages, but otherwise this provides conclusive proof that the “Ziyi” chapter of the Liji is indeed a pre-Qin text.8 Other texts in the Shanghai manuscripts such as the Rongchengshi 《容成氏》 and Minzhifumu 《民之父母》—titles assigned by modern-day editors—are contained in chapters of the received Liji. The “Wen Wang Shizi” 文王世子 and “Kongzi Xianju” 孔子閒居 chapters, respectively, are comprised of these texts (with slight variation) in addition to other material not found in the Shanghai manuscripts. The recently acquired Qinghua manuscripts are rumored to also contain a number of texts explicitly treating the topic of li 禮.
As far as the issue of redaction is concerned, scholars generally agree on the activities involved in the process of editing the current forty-nine chapters of the Liji. Interpreters, however, disagree as to which activities were involved in redacting each text. Wang E provides a useful list in this regard. He claims that redacting the text involved the following activities:
1. Selecting texts from original source material.9 The Hanshu 《漢書》, for instance, mentions 131 ji 記 in the category of li 禮.10 It otherwise makes no mention of any text entitled Liji, leading many interpreters to suppose that the 131 ji 記 were the source material.11
2. Adding initial or closing lines to chapters. The “Kongzi Xianju” chapter, for instance, has seven characters that begin the text and seventeen characters that close the text, which the Minzhifumu manuscript from the Shanghai Museum collection lacks.
3. Adding passages from other texts. The “Jiaotesheng” 郊特牲 chapter, to use an example, quotes at length from the Yili with practically no variation.
(p.222) 4. Combining similar texts into one chapter. The literary structures of the “Xiangyinjiuyi” 鄉飲酒義 and “Wen Wang Shizi” chapters, to provide an illustration, suggest that they are each made up of several sub-texts.
5. Combining passages that were previously considered commentary into the text. Everything following the first portion of the “Daxue” 大學, for instance, is a commentary on the first section.
6. Mis-transcribing the source material. The “Yanyi” 燕義 chapter quotes from the Zhouli but leaves out the last few characters in the sentence, leaving the line extremely difficult to interpret without presuming that it was mis-transcribed.
7. Adding new material, perhaps written by the redactor(s), in order to incorporate ideas otherwise not present in the text. The yinyang 陰陽 and wuxing 五行 theories that appear in the “Liyun” chapter are commonly thought to be such additions.12
While some of the examples employed in this list are debatable, the general idea that redacting the text involved these activities is widely accepted. The problem lies in determining what activity took place in which chapter; and since the Liji has been commented on for over 1800 years, there are about as many different interpretations as there are possible combinations of these activities in each chapter. Oftentimes, scholars will reach the same conclusion about a particular chapter or section of the Liji, but arrive at that conclusion by means of different arguments. This is particularly true in comparing contemporary interpreters with premodern interpreters.
Most scholars attribute the process of redaction to Dai Sheng 戴聖, a scholar that lived sometime in the first century BCE.13 Dai Sheng and his second cousin Dai De 戴德 came from Liang 梁 and were scholars (boshi 博士) who served in the Han court's Office of Ritual (Liguan 禮官).14 They were both the students of Hou Cang 后蒼 (73–49 BCE), a ritual specialist thought to be an authority on the Yili. Dai Sheng is purported to have participated in the great debate at the Stone Conduit (Shiquge 石渠閣) in 51 BCE, which sought to make clear several issues regarding ritual. He eventually went on to become the Grand Protector (taishou 太守) of Jiujiang 九江. Zheng Xuan, in his text Liuyi Lun 《六藝論》, which is preserved in only fragmentary form, is the first to attribute a text called the Liji (with 49 chapters) to Dai Sheng.15 While some scholars have questioned whether or not the text Zheng refers to is in fact the received Liji, the majority of scholars tend to accept Dai Sheng as the redactor. The received Liji is even sometimes referred to as the Xiao Dai Liji 《小戴禮記》 in the centuries that follow (since Dai Sheng was the younger of the two cousins).16
Not all scholars believe that Dai Sheng was the redactor of the Liji. While this is the minority opinion, there are good reasons to challenge the theory that Dai Sheng was the redactor.17 The problem, however, is coming up with an alternative theory that is more persuasive than the widely accepted position. In this light, some scholars have suggested that Ma Rong or Lu Zhi 盧植 (d. 192 CE), rough contemporaries of Zheng Xuan, were the redactors.18
(p.223) To provide an example of the kinds of conclusions that can be drawn from the information and theoretical assumptions discussed above, Wang E claims that fourteen chapters were composed in the late Chunqiu or early Warring States periods. Some of these texts, in his opinion, were even authored by Confucius himself. Nineteen chapters were composed in the mid–Warring States period, seven chapters in the mid– to late Warring States, and three chapters in the late Warring States period. This source material was then redacted sometime between 51 and 21 BCE by Dai Sheng. Wang claims that the process of redaction is easily seen in most chapters.19
The above discussion was meant to serve as an overview of the debates surrounding the issues of the textual composition of the Liji. My purpose in going through this was threefold. First of all, I sought to describe the major factors that go into a theory of the textual composition of the Liji. Second, and related to the first, was to demonstrate the parameters of interpretation over the past two thousand years. It should be obvious here that there is no conclusive position regarding most of these issues. Last, and perhaps most important, it was to demonstrate that my reading of Liji was not performed unaware of these issues, although they rarely came up in the project. I do not believe that my conclusions—as far as their relation to textual composition are concerned—ignore these parameters. As a matter of fact, I would say that the factors involved in a theory of textual composition are an implicit part of the underpinnings of this project. These factors demonstrate that early Confucians were concerned with a large number of texts on a variety of interrelated topics, all of which were written over several centuries. These texts became chapters in one central text around a time when similar efforts resulted in grand claims of creating a comprehensive world view—evidenced in texts such as the Lüshi Chunqiu 《呂氏春秋》 and the Huainanzi 《淮南子》. The Liji, in stark contrast to these texts, does not claim to present one guiding cosmological vision with which to understand the world. Rather, it presents a complex web of interrelated concepts, which were constructed and written about in a multitude of texts. These concepts were meant to create an ordered world; yet precisely how they fit together is often unclear, and indeed seems to be less of a concern for the redactor(s). The Liji, perhaps, represents a conscious choice not to create a grand unifying vision of the world. (p.224)
(1.) James Legge, trans., The Sacred Books of China: The Texts of Confucianism, parts 3–4, The Li Ki (Oxford: Clarendon Press, 1885), 3:1–14.
(2.) Ye Hong 洪業 et al., Liji yinde 《禮記引得》(Beiping 北平: Hafo Yanjing xue she yinde bian zuan chu 哈佛燕京學社引得編纂處, 1937). Meng’ou Wang 王夢鷗, Liji jinzhu jinyi 《禮記今註今譯》 (Taibei 臺北: Taiwan shangwu yinshuguan 臺灣商務印書館,1969). Yihua Jiang 姜義華, Xinyi Liji duben 《新譯禮記讀本》(Taibei 臺北: Sanmin shuju 三民書局, 1997). Teruo Takeuchi, 竹內照夫, Raiki 《礼記》 (Tōkyō 東京: Meiji Shoin 明治書院, 1971–1979). (p.261)
(3.) Xueqin Li 李學勤, ed., Liji Zhengyi《禮記正義》in Shisanjing zhushu《十三經注疏》 (Beijing 北京: Beijing daxue chubanshe 北京大学出版社, 2000), vols. 12–15. Cheng Wu 吳澄, Liji Zuanyan《禮記纂言》 (Jinan 濟南: Qi-Lu shushe chubanshe 齊魯書社出版社, 1997). Fuzhi Wang 王夫之, Liji Zhangju《禮記章句》 (Shanghai上海: Shanghai guji chubanshe 上海古籍出版社, 1995–1999).
(4.) Guang Hu 胡廣, Liji Daquan 《禮記大全》 (Taibei 臺北: Taiwan shangwu yinshuguan 臺灣商務印書館, 1983). Shi Wei 衛湜, ed., Liji Jishuo 《禮記集說》(Taibei 臺北: Taiwan shangwu yinshuguan 臺灣商務印書館, 1983).
(5.) Takuji Kudō 工藤卓司 provides a wonderful overview of the Japanese scholarship on the Liji over the last 100 years in “Jinyibainian Riben Liji yanjiu gaikuang 1900–2008 nian zhi huigu yu zhanwang”〈近一百年日本《禮記》研究概況－－1900–2008 年之回顧與展望〉, Zhongguo wenzhe yanjiusuo tongxun《中國文哲研究所通訊》19.4 (2009): 53–101.
(6.) E Wang 王鍔, Liji Chengshu Kao 《禮記成書考》 (Beijing 北京: Zhonghua Shuju 過華書局, 2007).
(7.) Although the Zhouli did not incorporate the term li 禮 into its title until the Tang dynasty.
(8.) For more on the “Ziyi,” see Edward L. Shaughnessy, Rewriting Early Chinese Texts (Albany: State University of New York Press, 2006).
(9.) Wang's original list includes six activities. I have added this one to his list.
(10.) Ye Hong 洪業 et al., eds., Hanshu ji buzhu zonghe yinde 《漢書及補注綜合引得》(Beijing 北京: Hafo Yanjing daxue tushuguan yinde bian zuan chu 哈佛燕京大學圖書館引得編纂處, 1940), 30.6b, page 577.
(11.) The Suishu 《隋書》notes that the Da Dai Liji contained eighty-five chapters. Many scholars have supposed a connection between the 131 ji 記 mentioned in the Hanshu, the eighty-five chapters of the Da Dai Liji noted in the Suishu, and the forty-nine chapters in the received text of the Liji. They claim the received text of the Liji to be a further redaction of Dai De's initial redaction of the 131 ji 記 into the eighty-five chapters of his book. See Huaixin Huang 黄怀信, “Guanyu Da Dai Liji yuanliu de jige wenti” 关于《大戴礼记》源流的几个问题, Qilu Xuekan 齐鲁学刊 184.1 (2005): 17.
(12.) These can all be found in Wang, Liji Chengshu Kao, 20, although I have modified number six.
(13.) Much of what is known about Dai Sheng comes from the Rulinzhuan 《儒林傳》; translated in Riegel, “Four ‘Tzu ssu’ Chapters of the Li chi,” 6–8.
(14.) Parts of this narrative are problematic as pointed out by Riegel, “Four ‘Tzu ssu’ Chapters of the Li chi,” 9.
(15.) The Liuyilun《六藝論》 is quoted throughout the Liji Zhengyi 《禮記正義》. The relevant quote is found in Li, Shisanjing zhushu, 12:10.
(16.) Dai De, or Da Dai 大戴, is also credited with composing a Liji—the Da Dai Liji 《大戴禮記》, which survives today in fragmentary form. Dai De 戴德, Da Dai (p.262) Liji《大戴禮記》 (Taibei 臺北: Taiwan shangwu yinshuguan 臺灣商務印書館, 1983). For the relation between the Xiao Dai Liji and Da Dai Liji, see note 11 above.
(17.) For a summary of these theories, see Wang, Liji Chengshu Kao, 283–299.
(18.) Wang, Liji Chengshu Kao, 299.
(19.) Wang gives a summary of his argument for these dates in Liji Chengshu Kao, 19 and 321–324.