Shamanism Theory and the Early Chinese Wu

Shamanism Theory and the Early Chinese Wu Thomas Michael* This article undertakes a reexamination of shamanism in early China, an issue that centers on a religious title (wu) that is consistently mentioned in virtually every major text from the period. For roughly the last fifty years, sinologists have vigorously argued the appropriateness of identifying these wu as shamans. In an effort to bring a deeper degree of clarity to this issue, Parts 1 and 2 of the article explore certain findings from the field of modern shaman studies that can open up new ways of thinking about the wu. Part 3 examines the ways in which sinologists have approached the wu and attempts to show how modern shaman theory can allow us to better situate our thinking on this issue. Part 4 offers a brief case study of one early Chinese text and considers how modern shaman theory can shed new light on our interpretation of the wu. APPROACHES TO THE SHAMAN, SHAMANISM, AND THE WU WU 巫 IS A TITLE THAT REFERS TO A VERY UNUSUAL TYPE OF PERSON found throughout Chinese history and represents one of the earliest envisioned religious figures present from the very beginnings of Chinese civilization. The term is first found on Shang dynasty (1554 1046 BCE) oracle bone inscriptions dating from before the first millennium BCE, centuries before the earliest textual histories describing them *Thomas Michael, Department of Religion, Boston University, 145 Bay State Road, Boston, MA 02215, USA. E-mail: maike966@gmail.com. Journal of the American Academy of Religion, September 2015, Vol. 83, No. 3, pp. 649 696 doi:10.1093/jaarel/lfv034 Advance Access publication on July 5, 2015 The Author 2015. Published by Oxford University Press, on behalf of the American Academy of Religion. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com

650 Journal of the American Academy of Religion came into existence. 1 Following a period of relative silence concerning the wu in the Western Zhou (1046 722 BCE), the Spring and Autumn period (722 481 BCE) marks the beginning of early Chinese writings in which we encounter the first historical depictions of the wu and their activities, primarily in the Zuo Zhuan and the Chun Qiu. The Warring States period (480 221 BCE) witnessed an explosion of writings in which the wu are depicted in a baffling variety of ways; in most of these writings, including the Shijing, the Yijing, the Shujing, the Guoyu, the Mozi, the Lunyu, the Mengzi, the Xunzi, the Zhuangzi, the Lüshi Chunqiu, the Hanfeizi, and the Guanzi, the wu are mentioned almost haphazardly and in passing, but their brief textual appearances still provide tantalizing bits of information on their position and activities. Other writings include substantial sections in which the wu are given center stage, including the Chuyu, xia from the Guoyu, the Chunguan from the Zhouli, and the Jiu Ge from the Chuci. Indeed, these three writings have provided by far the most substantial information about the wu in the Warring States period, and they have all been subjects of in-depth study by modern scholars who have explored the role of the wu in early China. Writings that discuss the wu continued throughout the entire course of Chinese history, from the Qin (221 206 BCE) and Han (206 BCE 220 CE) dynasties to contemporary times, and people bearing the title of wu are still legion in contemporary China. 2 In this study, I primarily focus on the literary representations of the wu in early China (roughly the fifth to third centuries BCE), with recourse from time to time to representations (and scholarly interpretations of those representations) of the wu from earlier periods. In these writings, wu are at times shown to have vital positions in the royal courts next to the supreme rulers, charged with responsibilities vital to the well-being of the court and the country. These include, for example, the proper control of the rains by way of their ritual dancing and communication with the natural spirits and the spirits of the ancestors through flight or possession. Other depictions of the wu describe them as living on mythical 1 For discussions of the earliest forms of the graph wu, see Chen (1936: 536 538), Chang (1994: 11 12), Allan (1991: 77), McCurley (2005: 136), and Li (2001: 44 46). Keightley (1998: 765), however, writes, Because I do not believe that oracle-bone graph, which is often read as wu 巫, referred to shaman or spirit medium, I do not consider the Shang use of the term relevant to the present enquiry. Boileau (2002: 354 355) provides a precise summary of the meaning of this graph, namely as a spirit, a sacrifice, a form of divinization, and a living human being. 2 Mathieu writes, However, shamanism pursued an autonomous existence in the countryside up to contemporary times, as one sees in the studies by de Groot [1910: 1187 1341] and Doré [1911 1938] (1987: 23). For more on the political fortunes of the wu in the Ming (1368 1644) and Qing (1644 1911), see Sutton (2000); for the wu in Communist China, see MacInnis (1989); for the wu in contemporary China, see Anagnost (1987).

Michael: Shamanism Theory and the Early Chinese Wu 651 mountains, collecting magical medicinal herbs to treat illness and confer longevity, and also journeying throughout all corners of heaven and earth in their physical body. Should the wu be identified as shamans? Gilles Boileau writes, The relationship between shamanism and archaic Chinese religion has been the subject of debate for some time. The origin of the controversy is to be found in the article written by Chen Mengjia in 1936... (who), even if he did not use the term shaman, has established the basic framework for later scholarship on Shang shamanism.... Hopkins and Schafer were the first to translate wu by Shaman, but they gave no anthropological reference as to the definition of this term (2002: 351). There is, however, more to say about the origins of this debate; in fact, there are two separate origins (one Chinese and the other Western) that have given rise to two separate approaches to the question of translating wu as shaman (and these two approaches exactly line up with two separate traditions of early Chinese shamanism that I later call bureaucratic shamanism and independent shamanism). Although J. J. M. De Groot (1910) appears to be the first modern scholar to have isolated the wu as an object of scholarly attention, studies of the early Chinese wu began in earnest with three studies in China in the 1930s. 3 Although the terms shaman and shamanism were not at that time part of the modern Chinese vocabulary, a handful of scholars began to turn their attention to the wu, and their writings all shared one central idea that would come to play a dominant role in later studies that were framed by cross-cultural conceptions of shamans and shamanism. This is the idea that the wu were the earliest religious figures and leaders of archaic Chinese religion who, when social organization transformed from matriarchy to patriarchy, 4 became tribal chiefs and, later, the earliest kings. The wu, therefore, were a central component of royal authority, so much so that the kings themselves were identified with the wu; Chen Mengjia succinctly writes, The ancient kings were wu (1936: 535). 5 Kwang-Chih Chang cites Chen (1936), the 1954 study of Li Tsung-t ung, and the 1962 study of Yang Hsiang-k uei when he writes, in fact, 3 Qu Kezhi (1930), Li Anzhai (1931), and Chen Mengjia (1936), cited in Tong (1995: 180 181). Notice that the studies by Qu and Li predate by some five years Chen s study, who has received the lion s share of recognition by Western scholars. 4 Chen writes, The ancient Chinese religious leaders were also the political leaders; when primitive society changed from matriarchy to patriarchy, the males [namely the male wu] took charge of political and military authority (1936: 533). 5 Li writes that the wu developed from private wu to official wu. After they had been given official posts, they developed into the leaders of the community. Along with the increase in their authority, they became chiefs, and finally, kings this is the origin of chiefs and kings (1931: 42). Chen continues this line of thinking: The chiefs of the wu evolved into kings as political leaders (1936: 535).

652 Journal of the American Academy of Religion scholars of ancient China agree that the king himself was actually head shaman (1983: 45), but Chang himself was the first Chinese scholar to apply the terms shaman and shamanism to the wu after the publication of Mircea Eliade s study that appeared in English in 1964. 6 More recently, two studies by Tong Enzheng (1995, 2002) have presented the most comprehensive articulation of this approach to the wu as shaman-kings. In the Western academy, the first attempt to identify the wu as shamans was put forth by L. C. Hopkins (1945). Being an early pioneer in this endeavor to identify the wu with shamans (some six years before the French publication of Eliade s work; he also appears not to be familiar with the Chinese scholarship discussed above), he just did not have a lot to work with. 7 Edward Schafer s (1951) study was next in the line of Western scholarship to translate wu as shaman, but coming in the same year as the French publication of Eliade s work, he was only incipiently aware of what was at stake in that translation. 8 He was, however, very familiar with Chen s (1936) study of the wu and he consistently cited it, but his translation of wu as shaman was, apparently, only for the purpose of giving an English language term (if indeed shaman can be called that) for the foreign Chinese term, most likely to avoid alienating his English readership; he nowhere problematizes what is at stake in that translation. Whereas all of the studies of the early Chinese wu discussed to this point approach the topic first of all from the standpoint of ancient Chinese (primarily Shang Dynasty) religion and politics with special attention to the etymology of the character wu, the 1951 publication of Eliade s work, and even more so the English translation of it in 1964, marked the emergence of an entirely different approach. 9 This approach attends to a Chinese tradition of folk shamanism that only tangentially relates the wu to official positions of rulership and bureaucratic 6 Puett writes, Chang did not indicate which scholarly definition of shamanism he had in mind in making these arguments, but he did occasionally refer to Eliade. Moreover... Chang s interpretation of a shamanistic cosmology is identical to Eliade s (2002: 34). 7 Hopkins writes, The Shaman or Wizard of the proto-historic Orient, his vocation, his reputation, his strange psychosis, and his place in the community, are not all those written in the chronicles of the Works of de Groot and Shirokogoroff? (1945: 3). Nowhere in this work does he explore the idea of the shaman or shamanism or explain why he makes this identification. 8 As Boileau points out (2002: 351), Schafer made one cross-cultural comparison: After the Chou dynasty, the female shaman... was forced into sub rosa channels for the practicing of her magic arts, analogously to the witch of medieval Europe (1951: 134). 9 Actually, Bodde (1961) could be claimed to be the first scholar to have seriously, albeit very briefly, applied Eliade s theoretical apparatus to the early Chinese wu, but as it seems that his ideas only become relevant in another context of this debate, I reserve comments on him for a later section of this study.

Michael: Shamanism Theory and the Early Chinese Wu 653 institution. Eschewing etymological analyses of the ancient origins of the term wu, this approach focuses on the very different kinds of activities associated with the wu of the southernmost early Chinese state of Chu, a tradition that some modern scholars see vividly represented in an anthology of literary poems or songs called the Chuci, in which the wu are seen to go on ecstatic flights to heaven and the underworld. The parts of this collection most readily associated with shamanism are the Jiu Ge, which Arthur Waley translated into English in 1956. He made a passing reference to Eliade about his choice to translate wu as shaman, but he also never problematized this; for him, it was simply a matter of convenience. 10 David Hawkes (1985), in his complete translation of the Chuci, also translates the term wu as shaman, and he in fact does make an effort to problematize this, but he never once mentions Eliade. Furthermore, his stated reasons for this translation do more to obfuscate the possible identity of the wu with shamans than it does to clarify it because he uses the term to mean nothing in particular, and he himself concedes this: The now common practice I have followed in using the word shaman as a substitute for the Chinese word wu... is, I believe, fully justified; but it can occasionally become a source of confusion (Hawkes 1985: 43). 11 Because so many of these modern scholars who attended to the early Chinese wu were, in essence, translating the term wu as the term shaman with apparent abandon and without critical awareness of what was at stake, it took a single sinologist engaging with modern shamanism theory to put the brakes on this. Three seminal works by David Keightley (1983, 1989, 1998) demonstrate his informed criticisms of the early Chinese shaman hypothesis, but it was in the last of these that he brought to bear his most cutting insights: Shamanism, indeed, is much in the eye of the beholder; determining its presence or absence in purely formal terms, accordingly, can become a sterile and circular exercise (1998: 767). 10 Citing Eliade s work, he writes, Indeed the functions of Chinese wu were so like those of Siberian and Tunguz shamans that it is convenient (as has indeed been done by Far Eastern and European writers) to use shaman as a translation of wu (Waley 1956: 3). 11 And he does nothing to dispel this source of confusion : Shamanism is a word invented by nineteenth-century anthropologists from the Tungusic word shaman. Modern anthropologists have exercised their proprietal right to define it in various ways, generally rather narrowly, but in popular usage it has come to apply to a rather wide range of religious beliefs and practices for which oldfashioned terms like animism and sorcery are felt to be inadequate or unsuitably patronizing (Hawkes 1985: 43). One is here reminded of Geertz s famous statement on the insipid nature of the category shamanism (1973: 122).

654 Journal of the American Academy of Religion Despite a formidable and still growing body of scholarship on the appropriateness of calling the early Chinese wu shamans, the issue of shamanism in early China remains as a highly controversial subject for both sinology and religion studies more generally, pertaining as it does to cross-cultural issues of the relation between shamanism, local cultures, and state religion; the development of institutionalized religion; and religious and cultural practice and representation more generally. It is my argument that a comprehensive study of the textual representations of the early Chinese wu shows that their traditions lie at the very heart of any informed understanding of shamanism in this period. I demonstrate that shamanism was a vital element at both the state and popular levels of the religious and political landscape of early China. I limit myself to discussing certain literary representations of the wu that I argue are most coherently interpreted through the application of the category of shamanism. In order to do this, I first attempt to bring a further measure of conceptual clarity to the categories of the shaman and shamanism, and also to the ways that scholars have deployed them in previous studies of early China. The present work examines a selection of literary texts representing several distinct discursive traditions of early China that scholars have used to base their claims for the existence or absence of shamanism for the period. One problem with previous studies on this issue involves the absence of a clear distinction between the shaman, which is primarily an ethnographic category, and shamanism, which is primarily a geographical and theoretical category, and scholars typically seek to establish or deny one by proving or disproving the other; this is particularly the case in modern sinological studies of the wu. This has resulted in a clear lack of consensus among writers on this topic concerning whether or not shamans and shamanism existed in early China: some scholars claim that because there is no definitive evidence for shamans in early China, early Chinese shamanism too is either minimal or nonexistent; these are the positions held most notably by Keightley (1998), Boileau (2002), and Michael Puett (2002). Others scholars claim that since there is an overwhelming body of evidence of shamanism in early China, shamans must have existed; this is the position held most notably by Waley (1956), Chang (1983), Hawkes (1985), and Tong (1995, 2002). 12 More specifically, all of these scholars use the category of shamanism to refer to what shamans do ( primarily the séance), which is quite different from using shamanism as an interpretative tool for the analysis of more general 12 Chen (1936) should head this list, but since neither the term shaman nor shamanism was available to him at the time of his writing, he of course did not make the connection.

Michael: Shamanism Theory and the Early Chinese Wu 655 representations of direct contact (possession) or face-to-face communication (spirit journeys) between humans and bodiless beings in a séance event for the benefit of the community. A strict understanding of shamanism has little or nothing to do with actual shamans and their séances; in fact, there exist no historical records of any actual early Chinese séance, and the descriptions of the séances of the Chuci remain a work of literary genius, not an ethnography. 13 Claims about shamans are of an altogether different order from claims about shamanism, a distinction already clearly noted by Arnold van Gennep over a century ago: The word sorcerer is too evocative of modern or medieval Europe; for the semi-civilized, it is better to use the word shaman. But when it comes to the word shamanism, which does not apply to anything definitive, one would do better, it seems, to leave it out (1903, quoted in Narby and Huxley 2001: 52). More recently, Åke Hultkrantz provides a clear articulation of the difference between the ethnographic shaman and theoretical shamanism; he writes that the shaman is a social functionary who, with the help of guardian spirits, attains ecstasy in order to create a rapport with the supernatural world on behalf of his group members (1973: 34). He continues: Shamanism, of course, is the complex of beliefs, rites and traditions clustered around the shaman and his activities (1973: 36). As for this complex of beliefs designated by Hultkrantz, Mihály Hoppál writes, Shamanism is a complex system of beliefs which includes the knowledge of and belief in the names of helping spirits in the shamanic pantheon, the memory of certain texts (sermons, shaman-songs, legends, myths, etc.), the rules for activities (rituals, sacrifices, the technique of ecstasy, etc.), and the objects, tools, and paraphernalia used by shamans (drum, stick, bow, mirror, costumes, etc.) (1987: 95). I propose that the category of the shaman should be understood to refer to persons who actively perform ethnographically distinct séances, and the category of shamanism should be used as a theoretical construct applied to the interpretation of anomalous records expressing direct relationships between certain humans and non-bodily beings. Problems in the study of shamanism arise when these two usages become mixed, and this is especially so in the debates concerning shamanism in early China. 13 On this note, Humphrey writes, We should try to discover what shamans do and what powers they are thought to have, rather than crystallize out a context-free model derived from images they may or may not use (1996: 192). Although Humphrey was discussing the living shamans available to her at the time of her fieldwork, I am less interested in locating specific shamans than I am in investigating the primary characteristics of the textual representations of the early Chinese wu. But the idea of eschewing any of those essential qualities of the shaman and shamanism that cannot be identified with actual evidence, literary or ethnographic, as the case may be, still holds.

656 Journal of the American Academy of Religion All too often writers who deny the existence of shamanism in China do so on the grounds that the shaman is nowhere to be found; Keightley writes that the character of the wu in the Chuyu, xia does not seem markedly shamanistic.... On the contrary, these religious practitioners are described in terms suited to a perspicacious and reverential sage (1983: 8). About the socio-historical existence of shamans in early China, all that we have at our disposal are the records, literary or otherwise, of something that we may or may not construe as shamanism. 14 Still, one expects that if shamanism existed in early China, it would closely resemble recognized forms of Siberian or Central Asian shamanism known from eighteenth-, nineteenth-, and twentieth-century fieldwork, 15 although it seems reasonable to date shamanism in Siberia to the sixteenth century at the latest. 16 But those are already very late and highly developed forms of shamanism, and for early Chinese shamanism to resemble them, it too must be a highly developed form. 17 The texts that I examine from early China, however, provide us with the earliest written records of 14 It is this absolute absence of ethnographic possibility that allows the study of early Chinese shamanism to be the object of such extremely divergent conclusions. Atkinson writes, Most items by anthropologists are ethnographic and focused on single cultural traditions.... Such scholarship has had insufficient impact on the wider field of shamanic studies, which features general theorizing, model-building, and self-actualization.... Without an ethnographic counterweight this literature slips quickly into unwarranted reductionism and romantic exoticizing of a homogeneous non- Western Other (1992: 308 309). 15 Boileau brings this point home: Arthur Waley also translates wu as shaman, referring directly to Siberian shamanism. He was followed by numerous scholars who used comparatism to demonstrate that a form of religion present primarily in Siberia and still observed today is a reliable explanatory tool for Shang and Zhou civilization (2002: 351). 16 Hutton provides a lengthy excerpt of what appears to be the earliest eyewitness account, written by one Richard Johnson, of a Siberian shamanic séance that is dated to 1557 (2001: 30 32). 17 This notion that the Siberian shaman demonstrates the pure shamanism against which all other shamanisms are to be gauged is, arguably, the most important legacy bequeathed by Eliade. In the dim, confusionistic mass of the religious life of archaic societies considered as a whole, shamanism taken in its strict and exact sense already shows a structure of its own and implies a history that there is every reason to clarify. Shamanism in the strict sense is pre-eminently a religious phenomenon of Siberia and Central Asia.... [Shamanism] has had its most complete manifestation in North and Central Asia, [and] we shall take the shaman of these regions as our typical example.... But this Central Asian and Siberian shamanism has the advantage of presenting a structure in which elements that exist independently elsewhere in the world... are here already found integrated with a particular ideology and validating specific techniques (1964: 4 6). Note here that Eliade has presented and connected two separate ideas about Siberian shamanism that has had a lasting impact on the history of shaman studies, including those by sinologists: first, of all shamanisms found in the world today, only Siberian shamanism demonstrates a pure shamanism; second, Siberian shamanism most closely approximates archaic shamanism and represents the shamanic substratum of archaic religion (1964: 6). For a theoretical critique of these notions of Eliade s Siberian shaman, see Kehoe (1996, 2000). For a corrective study of Siberian shamanism that gives a historical critique of these notions, see Hutton (2001).

Michael: Shamanism Theory and the Early Chinese Wu 657 shamanism anywhere to be found. 18 The question is not how closely these records demonstrate a resemblance to highly developed and very modern forms of shamanism, but why shamanism appears in the archaeological records for the first time in the second millennium BCE and then regularly appears in the classical texts from roughly the middle of the first millennium BCE onward. The modern categories of the shaman and shamanism are, for the most part, constructed rather than native and have been used in different ways to explain many things, leaving us with a great deal of confusion regarding what writers mean when they use these categories. 19 Those who have actively participated in the historical construction of the categories of the shaman and shamanisms include missionaries, philosophers, theologians, archaeologists, anthropologists, psychologists, historians, botanists and other scientists, fiction writers, new age practitioners, and Hollywood screenwriters. 20 Some scholars, treating the categories as natural pieces of discourse able to designate autonomous and independent regions of experience really existing out there, have reacted to this by attempting to curtail the ambiguous nature of these categories through theoretical precision, ethnographic description, or both. 21 The continued ambiguity of the categories, however, has not gone away through simple cataloging. In large part encouraged by the very title of Eliade s work, Shamanism [is] Archaic Techniques of Ecstasy, many scholars argue for the universal presence of shamans in archaic cultures, representatives 18 As mentioned in the previous note, the earliest records of Siberian shamanism can be solidly dated to only the sixteenth century, a far cry from the Stone Age. There are, as far as I know, only two other contenders for the honor of being the earliest records of shamanism: prehistoric rock art exemplified in the cave-paintings of Lascaux and Trois Frères, and ancient Greece. First, prehistoric rock art is not written and, further, there is anything but consensus that the paintings are shamanistic (see, for example, Bahn 2001; Diaz-Andreu 2001; Francfort 2001; Klein et al. 2002; Guthrie 2006). Second, shamanism as a potent force in ancient Greece was presented most enthusiastically in Dodds (1973 [1951]: 135 156), but this shamanic hypothesis has had very little traction; Hadot wants to define shamanism as a certain ritual conduct linked to concrete situations. It is obvious that, in the Greek philosophical tradition, one can find no trace of this shamanic ritual conduct (2001: 399). Puett (2002:83 94) also challenges the hypothesis of Greek shamanism. 19 As Keightley also rightfully points out, There may well be many kinds of shamanism... and, if so, that makes it all the more important to specify in each case precisely what definition one is using. Whether these different shamanisms can then be related to one another in any meaningful way always needs to be demonstrated. And the unexamined use of the term should always be a warning flag, alerting us to the dangers of lapsing into an academic trance as we make our own attempt to communicate with other, unseen realms (2002: 409). 20 Four works stand out for their comprehensive treatment of the historical construction of these two terms: Flaherty (1992); Narby and Huxley (2001); Znamenski (2007); and Tomášková (2013). 21 By far the best resource for such scholarly work is Shaman: Journal of the International Society for Shamanistic Research, launched in 1993.

658 Journal of the American Academy of Religion of a certain type of religiosity that have exerted a profound influence on the religious expressions of later humans; dead shamans of archaic world, however, exist only as virtual ghosts in any historical records. In the effort to come to an acceptable understanding of pre-modern shamans, one could first of all start with modern shamans and trace their history backward all the way even to the Stone Age, taking for granted a linear developmental history based on any static and essentialized definition of the shaman, and Eliade (1964) has provided the model. If, however, any essentialized definition of the shaman that one finds at hand does not exactly fit one s own material, one is also free to modify that definition in the creation of a new essentialized definition so that it will fit the material. 22 In the end, all of this continues to assume the shaman to be a unitary and unchanging category throughout all societies in archaic times (even if the various essentialized definitions do not exactly line up with each other), which is to say hunter-gatherer societies and, to an extent, agricultural societies. If one assumes the universal presence of shamans in archaic societies, then the next step in the developmental path posits the emergence of other religious functionaries directly derived from the shaman, namely shaman shaman-healer healer (such as medicine men and doctors). 23 There is also a more insidious version of this developmental line: shaman shaman-priest priest, or even shaman shaman-king king. 24 The problem with this is that before one can even discuss the various religious and political functionaries that emerged from the archaic shaman, one has to know that archaic shaman as a universally present figure; if we cannot find him, then there is no significance 22 This indeed is Eliade s method whereby he designates what he finds to be the essential traits of the shaman and shamanism (calling, initiation, soul-flight, axis mundi, etc.) and then attempts to locate them in the archaic mind of homo religiosus. Remarking on Pre-Columbianists willingness to create new criteria for shamans whenever Eliade s model fails to work for them (2001: 218), Klein et al. write, Despite Furst s attempt to redefine shamanism so that it could be found in the Americas, the new criteria he produced are as unreliable as Eliade s (2001: 213). 23 Winkelman is a major spokesperson for this view: It is argued that shamanism is found throughout the world because it derives from an ecological adaptation of hunting and gathering societies to biologically based altered-state-of-consciousness potentials. Agriculture and political integration are shown to cause the transformation of the shaman into other types of magico-religious healing practitioners, labeled in this study as shaman-healers, healers, and possession-trance mediums (1990: 308). 24 Although Klein et al. do not specifically mention this, I would argue that the most radical element of these redefinitions of Eliade s shaman found in the scholarly literature is the association or even identification of shamans with kings; Eliade himself never made this identification. This seems to me to lie at the heart of what Klein et al. (2001) call shamanitis in relation to the pre- Columbianists that she takes to task, and this is also what I find to be the weakest, even the fatal, element of Chang s ideas about ancient Chinese shamanism.

Michael: Shamanism Theory and the Early Chinese Wu 659 to any of these developmental lines. Coming to an understanding of the archaic shaman, it is first of all necessary to grapple with the contours and limits of who can qualify for inclusion in the category. Given this, arguments for the socio-historical reality of the archaic shaman, whether or not the early Chinese wu can be included in it, face formidable obstacles. These include the question of what might count as valid evidence for them, the question of language involving local terms used by alien cultures to refer to potential candidates of this category, and, maybe most formidable, the question of the applicability of the various essentialized definitions of the archaic shaman that are commonly used to identify true shamans (but rarely used to weed out imposters). The question of definition, however, does not simply concern the shaman as such, but any specific definition, if it is to be workable, also has to differentiate the shaman from all other religious and political functionaries; any definition that does not attend to this will inevitably be so openended that it will lose all tractability. Cecelia Klein et al. write, The problem with all of this is that if there is no single, universal definition of shaman, it is often impossible, when one is mentioned, to know exactly what that person did or does. Moreover, we cannot accurately relate this individual to persons similarly identified in other societies. One scholar s shaman, in other words, is not necessarily another s (2001: 219). 25 I find it heuristically reasonable to mark off four historical moments in which the term shaman transformed from a referent appropriate to a particular and local time and place to a universal referent applied almost randomly in the current world. The figure of the shaman first of all comes to us from an existing language, Tungus, used by an existing Siberian community to refer to a privileged social participant in their society, and it made the shaman available as an object of study starting in the late seventeenth century. 26 In other words, in a limited sense, this 25 Klein et al. continue: Most importantly, what do these imprecise words really signify today, and why is that meaning so important to so many of us? We suggest that the best way to answer this question is to ask what words and concepts the words shaman and shamanism allow the scholar to avoid. If we perceive certain individuals as primarily religious practitioners, for example, the label shaman allows us to avoid referring to them as priests. Most academics have a general notion of what the distinction between a shaman and a priest theoretically entails (2001: 219). They relate the same separation to the shaman and the magician, the doctor, and the king. What they are trying to get at is, first, the question, not of what a shaman is, but rather who is not a shaman, and, second, is it possible for a shaman to no longer be a shaman; this latter question is quite central for any position, for or against, that sees other religious and political functionaries as emerging from the archaic shaman. This again is a central consideration for any discussion of shamanism in early China. 26 In fact, the first published appearance of the term shaman appears to be found in the autobiography, first published in 1672, of one Avvakum Petrovich, in which he writes of his experiences in exile in Siberia (as pointed out in Hamayon 1993: 4;Pascal 1938 has translated the autobiography into modern French, part of which is excerpted in Narby and Huxley 2001: 18 20).

660 Journal of the American Academy of Religion usage has a concrete ethnographical reality which, however, should not in any way exclude it from the de-essentializing effects of history. Next, the usage of the term shaman was quickly extended throughout other hinterland Siberian societies that came more and more under the gaze of Christian Russian missionaries and intellectuals (who were often banished to those regions) and was used as a general category marker for various of these societies speaking different languages, many of which did not possess the Tungus term shaman in their vocabulary, to refer to a slightly different type of social participant who shared in direct communications with bodiless beings. 27 The late nineteenth and early twentieth centuries, in tandem with the spread of Western colonialism and the birth of anthropology, subsequently witnessed a further extension of this referential field of the shaman to refer to other figures which were seen to participate in strikingly similar performances of direct communication with bodiless beings, but in vastly different societies throughout virtually all parts of the globe (Narby and Huxley 2001: 39 73). Finally, hand in hand with the production of general theories of shamanism, the notion of the shaman was further extended to refer to vastly different types of social actors locatable in virtually all times and places. 28 This lack of conceptual Hutton singles out the 1692 account of a Siberian shaman written by Nicholas Witsen that popularized the term shaman among Europeans (2001: 32). For the further account of the European adoption of the term in the eighteenth century, see Flaherty (1992); Znamenski (2007); and Tomášková (2013). 27 Winkelman points this out: Terms that are etymologically similar to the Tungusic saman (e.g., csaman, sama, shaman, khaman, śam, csam, kam, xam, xamma-, and xamsa) are widely dispersed in the languages of Siberia, as well as in Asia and even in ancient Indo-European languages (e.g., śaman, śramana, samâne, saga, and wissago) (2013: 47). For more on the slightly different types of social participants in these Tungus-related societies, see Shirokogoroff (1935), Dioszegi and Hoppál (1978), and Hutton (2001). Over and beyond the Tungus-related societies, Hutton writes: [Shaman] was not the word which would have been used of such figures by the great majority of native Siberians: among the Turkic-speaking peoples the equivalent term was kam, among the Samoyed-speakers tadibei, among the Sakha oyun, among the Buryats bö, among the Koryaks enelan, and so forth (2001: 47). See also Eliade (1964: 4). 28 To better grasp the range of inclusions in this last phase, I find it helpful to speak of indigenous traditional shamans, indigenous neo-shamans, and non-indigenous neo-shamans. By indigenous neo-shamans, I mean those people in foreign countries with a history of shamanism who cater to Western tourists; by non-indigenous neo-shamans, I refer to those people in modern first-world countries who seek to have shamanic visions through altered states of consciousness with the goal of self-knowledge; Harner (1990) is the classic text for this. The birth of both forms of neo-shamanism can be dated to May 15, 1957, which marks the publication by R. Gordon Wasson, Seeking the Magic Mushroom, in Life magazine; the article documents his journey under the supervision of the indigenous traditional shaman Maria Sabina (excerpted in Narby and Huxley 2001: 141 147). Some twenty years later, Maria Sabina was recorded as saying, It is true that Wasson and his friends were the first foreigners who came to our village in search of the sacred children, and who did not take them to heal from an illness. Their reason was that they wanted to find God. Before Wasson, nobody took the mushrooms simply to find God. One had always taken them to heal the sick (excerpted in Narby and Huxley 2001: 167).

Michael: Shamanism Theory and the Early Chinese Wu 661 clarity has given rise to the very interesting debates in which the intellectual authority to have the term shaman refer in any one of these four ways is presently being argued. 29 Many scholars today with a good awareness of geographical and ethnographical differences who strive for accurate cultural specificity use the term shaman not to refer to an ideal figure, but to particular types of social actors fulfilling socially recognizable roles (to be understood at least partially in the theatrical sense); Laurel Kendall writes, There is irony in Western theater s having sought its reflection in ethnography before anthropology was prepared to confront the subjects of its own musings in theater.... Walter Borgoras s turn of the century account of a Chukchi shaman s séance, a masterpiece of ethnographic description, is attuned to the beats of the drum, to how the shaman modifies his voice as he chants, to the spectators shouts and then the cries and whispers of the spirits as they enter the room.... The shaman as actor has been an uncomfortable notion for a relativistic social science (Kendall 1996: 19). 30 Although worldwide societies outside of Siberia do not have shamans in the strict sense (Eliade 1964: 4), some scholars show that some of these societies have social actors who demonstrate the essential qualities and activities of the shaman, in accord with whichever definition that the author provides. Different societies from all parts of the globe in many different ages have displayed some evidences of the shaman when some of their members fulfill those strikingly similar roles, functions, and 29 Kehoe provides a similar understanding for the concept of shamanism: Shamanism has come to be used to refer to: (1) Its original reference, a religious complex in Siberia centering on the Tungus-Evenki trained practitioner utilizing drum and chant to create an altered state of consciousness believed to enable the practitioner to divine and to negotiate in the spirit world for desired effects.... (2) Religious practice opposed to historical Western religions, featuring ecstatic states (trance, possession) and emphasizing individuals subjective calling by spirits as contrasted to the literate religions formal ordination of practitioners.... (3) A primordial or primeval religion, or type of religious leader, supposed to have persisted since the Paleolithic among primitive huntergatherer/nomadic peoples.... (4) Techniques of altering consciousness, in contemporary Western societies no longer necessarily yoked to religious beliefs.... Contemporary shamanism may be used to heal or for self-expression (1996: 377). 30 Hultkrantz was instrumental in directing scholars of shamanism coming after Eliade back into the actual social history of shamans, because he was among the first (and the best) to approach the shaman in terms of a social actor fulfilling a social role for the members of his or her society (1973, 1996). The works of two other scholars also stand out for their ability to demythologize the shaman, Lewis (1989) and Hamayon (1990, 1996). Although somewhat dated, Atkinson discusses all of these works up to 1992; she writes, Much valuable ethnographic work on shamans is not billed as such but is contained in monographs with titles that give no hint of a shamanic focus except perhaps to an area specialist.... Newer ethnographic writings offer an important corrective by underscoring the connections of shamanic practices to local, regional, national, and transnational contexts. They also call attention to some of the culture-bound assumptions Euroamerican scholars and their audiences have brought to the study of shamanic traditions (1992: 308 309). See also Demanget (2001).

662 Journal of the American Academy of Religion activities. In my opinion, the essential and defining feature of the shaman is that he or she performs a séance event during which direct contact or face-to-face communication between human beings and bodiless beings is believed to occur. In many ways, the shaman, firmly situated in the words and worlds of ethnography, is more complex, more defiantly uncontainable, than the theoretical category of shamanism. Theoretical categories, however, all too often seduce scholars into imagining rational neatness and ideological unity in the flow of temporal events that may be more chaotic than we expect or desire. Shamanism as a general category is useful, allowing us to place a general wall of significance around a messy collection of representations that may or may not be essentially the same, depending upon how strict or flexible we allow our understanding of it to be. The stricter definitions work to exclude at times large pieces of the evidence from inclusion in the category, while the more flexible definitions allow a much wider range of inclusion. The least inclusive definitions of shamanism see it as a very late manifestation of the impulses to cultural survival in the face of aggressive foreign imperialism. 31 In this way, the religious significance of shamanism is placed second to its political genesis. 32 The most inclusive definitions, based most notably on Eliade s conception of shamanism as the foremost of the archaic techniques of ecstasy, see it as the manifestation of a certain universal biological potential commonly referred to as altered states of consciousness or ASC (which is a more technical way to talk about notions of trance or ecstasy, but ASC is such a wide-open idea that it also includes mediumship and possession). 33 The coming together of the idea of ASC with theories of shamanism in modern writings has played a decisive role for many scholars who see in this the answer to the very origins of human religiosity. Some of them have gone on to produce general theories of the evolution and development of 31 In this vein, Taussig (1987) provides the most subtle yet devastating critique of shamanism to be found in print, with Kehoe (2000) coming in at a close second. Both of these works seem to take Geertz s famous comment as their starting point: the individuality of religious traditions has so often been dissolved into such desiccated types as animism, animatism, totemism, shamanism, ancestor worship, and all the other insipid categories by means of which ethnographers of religion devitalize their data (1973: 122). 32 Lewis understands shamanism much along these same lines as protest movements (1989: 26, 107 113), but, since such protest movements are not exclusively tied to imperialism but arise from numerous styles of protest from gender to class to ethnicity, he finds shamanism to be pervasive throughout many different societies. 33 Atkinson writes, ASC... has been the buzzword in interdisciplinary studies of shamanism over the last decade.... Whereas earlier scholars of religion had defined shamanism in ways that incorporated cultural understandings of shamans and their followers, behavioral scientists skeptical about the ontological basis of spirit worlds have epistemological bedrock in the concept of altered psychological states (1992: 310). See Noll (1985, 1987).

Michael: Shamanism Theory and the Early Chinese Wu 663 human civilization in which notions of the decline or displacement of a universal and archaic shamanism are put to work. 34 Societies that are said or seen to possess shamanism are primarily labeled as such because they have social actors who perform séance events, and these social actors are known by many different titles specific to each society. The burden of listing the dozens of titles for such actors can be somewhat relieved by handling them in terms of their cultural types in relation to general geographical regions (and by doing so we begin to move from ethnographic shamans to theoretical shamanism). Of these, the most important include the shaman from Siberia and Central Asia (see Shirokogoroff 1935; Shi 1996), which may also include similar types from North America (see Jones 2008); the paye from South America (see Bacigalupo 2007); the angatkut from the Arctic regions (see Blodgett 1979); and the wu from East Asia (one can, for example, still argue the differences and similarities between the classical Chinese wu, the more modern Cantonese tongji [Elliot 1964; Potter 1974], the Korean mudan [Kendall 1985], and the Japanese yamabushi [Blacker 1975]. Can the claim of a single cultural type of East Asian shamanism be sustained for all of these figures?). The shamans in these regions among themselves can be designated separately from the shamans of every other region because they provide ethnographic types of local shamans that cohere around recognizable cultural traits shared among them (to a greater or lesser degree) within each area (granting that the specific labels for these local shamans can and often do differ within each region, and this is the case even in Siberia). 35 Relying on a general category, in this case a geographical one, in order to organize potentially similar kinds of representations of séance events, 34 Although Eliade did not make the move to see the emergence of separate religious and social functionaries from an archaic shamanism, he often condemned what he found in modern manifestations of shamanic activity and possession as the decadence of shamans (1964: 67 and passim). La Barre is arguably the most well-known spokesperson of the shaman as the source of later religious functionaries; he writes, The world s oldest profession is that of the shaman or first professional, the shaman is ancestor not only to both the modern medicine man or doctor and the religious priest or divine, but also ancestor in direct lineage to a host of other professional types. It would seem odd that both the doctor, the most secular-minded, and the divine, the most sacredminded of modern helpers of people, should derive from the same source. But we can readily understand the seeming paradox when we recognize the basic nature and function of the primitive medicine man of shaman (1979: 7). This position was whole-heartedly embraced by Winkelman (1990) and Harner (1990). Although the combination of the idea of ASC with that of shamanism is pervasive in modern writings that see shamanism actively present throughout the world, McKenna (1992) is arguably the leading spokesperson for this. 35 I am not implying that socially recognized shaman figures from other cultural areas not separately designated here do not demonstrate performances of the séance event, or that they do not have shamanism.

664 Journal of the American Academy of Religion I suggest that there is no easy way out of using the term shaman to comparatively label these culturally distinct shaman figures named above, and the term shamanism as a geographical and theoretical category that allows us to organize certain representations cohering around séance events shared cross-culturally that indicate direct communication between human beings and bodiless beings for the benefit of the community. In a sense, since the application of the term shamanism has become embedded in our way of speaking, it is probably too late to coin new terms such as angatkutism, payeism, orwuism. 36 Doing away with the singular and essentialized notion of the shaman while recognizing differences among even local shamans within each cultural area can allow us a more reflective space in which to use a geographical conception of the category of shamanism for both organizational and comparative purposes. Applying the category of shamanism in this geographical sense gives interpretative significance to evidentiary demonstrations of séance events with as much attention to cultural specificity and difference as possible. Like Ludwig Wittgenstein s family resemblances, these culture-specific shamans who perform them are neither completely identical nor radically distinguished. Applying the category of shamanism in a theoretical rather than a geographical sense (and keeping at bay here the ethnographic shaman in its ethnographical sense) marks off and isolates a wide range of actions, beliefs, and representations that would otherwise be impossible to interpret (at least comparatively), and the application of this theoretical category allows us to have more or less precise criteria of evaluation with which to do this. Shamanism as a theoretical category allows for a delimitation of the range of chaotic representations of direct communications between humans and bodiless beings, while shamanism as a geographical category reveals the pervasiveness of these representations with their own cultural specificity. Confusion in using the category of shamanism arises in part because of a lack of consensus on which features to include in it, and that consensus can only be informed by adherence to culture-specific representations that then can be utilized in wider, cross-cultural studies. Debates over which features to include or exclude often fail to take into account the fact that shamanism, whatever is included in it, is at best only a trace and will never be found to exist in any pure demonstration anywhere at any time; there are only representations that we interpret as 36 The term wuism, nonetheless, was De Groot s favored term (1910); scholars who have written on the Chinese wu since then, however, have refrained from keeping it.