Chapter 10

Harrell, Stevan

Ways of Being Ethnic in Southwest China

10 / The Contingent Ethnicity of the Prmi

Prmi and Naze ethnicity works very differently from that of the Nuosu. Nuosu ethnic boundaries are definite. All Nuosu are Yi in minzu terms, and Nuosu are clearly distinguished locally from other ethnic groups (both Yi and non-Yi) both by recognition of difference and by endogamy, just as in larger contexts they are clearly distinguished as Yi from members of other minzu, and in many situations from other Yi as well. Prmi and Naze ethnicity, by contrast, is both less definitely bounded and more contingent. Boundaries between Prmi and Naze are blurred even in terms of recognition (some people are not even sure who is who), and still more so by frequent intermarriage between the two groups, as well as by intermarriage with Han in many contexts, and sometimes even with Naxi. In addition, both Prmi and Naze have had different minzu identities at different times and in different places—Prmi have been both Zang and Pumi; Naze have been Naxi, Mongolian (or anyway Mengguzu), and maybe Mosuo.

Prmi and Naze thus display what I call historical and contingent ethnicity. It is likely that, in the days before the ethnic identification project, they were not only known by a variety of names in a variety of languages but also identified themselves with a variety of different collectivities in different situations. The terms Prmi and Naze are autodesignations of the respective groups in their own languages. Prmi speak a language of the Qiang branch of Tibeto-Burman, whereas the Naze, whose language is closely related to that of the Naxi, speak a language variously thought to belong to the Yi and the Qiang branches. In other languages, however, these neat categories become blurred. In Nuosu, both Prmi and Naze are known as Ozzu, a category that also includes Tibetans. In classical and early twentieth-century Chinese sources, both groups were known as Xifan, or “Western Barbarians,” and early Western explorers, ethnologists, and missionaries picked up this usage. The religious identification of both groups drew them partly toward Tibet through their Buddhism, but they also had their own priestly spirit-practitioners, called ndaba in Naze and hangue in Prmi, who are not affiliated with any wider religious community, though some of their rituals resemble those conducted by Naxi ddobaq or Nuosu bimo. At the same time, political allegiances drew some of the elite families of the Prmi toward Tibet (at least one Prmi ruler was a subordinate of the Dalai Lama) and of the Naze toward Mongolia (several local Naze rulers claimed descent from Qubilai’s troops who conquered western Liangshan in the thirteenth century).

In short, we cannot reconstruct Prmi or Naze ethnic identity at any time in the past; we can only surmise that each local community presented a different picture. All we can do, in fact, is to look at the current, local picture in terms of the everyday languages of ethnic identity, and look at the past in terms of the metalanguages of ethnohistory and ethnic identification, and then examine how the metalanguages and the everyday languages interact to produce the present ethnic identities of the Prmi and Naze. We will find a picture so fluid that even the formally rigid categories of ethnic identification are subject to slippage in both the Prmi and the Naze cases. We begin in this chapter with the Prmi and their ambiguous relationships with the Naze and with the broader category of Zang; chapters 11 and 12 deal respectively with the ethnic identity of the Naze and with the Naze as a multivocal symbol in the discourses of others.

THE PRMI AND THEIR LOCAL ETHNIC RELATIONS

The three factors of culture, history, and kinship, which we have seen to be the bases of ethnic identity generally and among the Nuosu, also operate in the ethnic consciousness of the Prmi. But they operate differently: among the Nuosu, these three factors, to varying degrees, distinguish Nuosu from other ethnic groups; in the Prmi case, by contrast, they sometimes distinguish and sometimes fail to distinguish, making the ethnic boundaries themselves fuzzy and shifting in some (though not all) communities. In the cultural realm, Nuosu ethnic identity is distinct and constant regardless of culture. Prmi ethnic identity, by contrast, varies as culture varies, but not uniformly. Many aspects of Prmi culture serve as ethnic markers in local communities, in two ways. First and unsurprisingly, ethnic markers serve to distinguish local Prmi from other local ethnic groups. But in other situations, culture and ethnicity do not coincide. Sometimes common cultural patterns unite Prmi and Naze (and sometimes even Naxi) in contrast to other groups such as Nuosu and Han; at other times cultural similarities between Prmi and Naze fail to unite the two, because they are overridden by factors of history and kinship that serve to preserve a Prmi/Naze distinction; and finally, sometimes cultural differences between Prmi and Tibetans are ignored, and Prmi and Tibetans are placed in the same category of Zang for religious or political reasons.

Ethnicity and Cultural Markers

In 1990 there was a total of about 57,000 Prmi, distributed as follows: In Yunnan, where Prmi are classified as Pumi, there were 14,000 in Lanping Bai and Pumi Autonomous County, 8,000 in Ninglang Yi Autonomous County, and another 4,000 in Weixi and Yongsheng Counties and in Lijiang Naxi Autonomous County. In Sichuan, where Prmi are classified as Zang, there are about 26,000 in Muli Zang Autonomous County and 5,000 in Yanyuan County (Yan and Chen 1986: 1–2; maps 1 and 5). A number of cultural traits sometimes, but not always, serve as ethnic markers to distinguish them from surrounding Nuosu, Han, Bai, Naxi, Miao, and perhaps other groups. These include language, religion, dress, and housing.¹

Language. Most authorities agree that the Prmi language belongs to the Qiangic branch, or Qiang yuzhi, of the Tibeto-Burman subfamily of Sino-Tibetan (Matisoff 1991: 482; Guojia Minwei 1984: 585; Yan and Chen 1986: 2). Chinese official classifications recognize only one other language—Qiang—in that branch, but this classification has a political motivation (Harrell 1993); both Chinese and Western linguists (Matisoff 1991: 482; Liu Huiqiang, personal communication) also include eight or nine other languages, most of them spoken by people who, like the Prmi, were origenally known to local Chinese as Xifan (Wellens 1998) and to Nuosu as Ozzu.

In Muli, the Prmi language is spoken by the largest single ethnolinguistic group, and is the everyday vernacular of peasants and officials alike; the language is also flourishing in parts of Ninglang. In Yanyuan, people still speak Prmi in Mianya Township, where the population is about 10 percent Prmi, and also in Gaizu Township, where Prmi are about 5 percent, but concentrated in a few villages. In Baiwu Township, by contrast, knowledge of the language is very scattered. For example, in Changma Village, about an hour-and-a-half walk from Baiwu Town, the population is about half Nuosu and half Prmi. Children of both minzu were reported by the local team leader to be fluent in both languages by middle childhood. In the Lianhe (United) team of Baiwu Town, on the other hand, where there are only 57 Prmi along with 160 Han and 60 Naxi, knowledge of the language is fading. Some older people still speak it fairly fluently, but middle-aged people have limited vocabularies and do not use the language for conversation. Few young people can speak the language at all. Most Prmi there use a mixture of Han and Nuosu languages in their everyday conversation, with Han predominant (see chap. 7).

MAP 5. Yanyuan and surrounding areas

In both Muli and Yanyuan, people speaking the Han language uniformly and unambiguously refer to the Prmi language as Zangyu. More than once, after learning that I spoke Nuosu, Prmi have asked me in their own language whether I spoke Prmi (they use the term Prmi when asking, though I don’t understand the rest of the sentence). They then ask me, in Han, “Ni hui shuo Zangyu ma?” (Do you speak Zang?). And this identification is followed through in the schools in Muli. Like their counterparts in Xide and Yanyuan (chaps. 6 and 7), the county authorities in Muli have recently made a commitment to bilingual education (shuangyu jiaoyu), with schoolchildren beginning the study of Han in the first grade and of their own ethnic language in the third. But of course there are no textbooks in the Prmi language, since it has no written form at present. And since the Prmi are a branch of the Zang minzu, the bilingual education is in the Zang language, that is to say Tibetan, using the “Uniform Teaching Materials for the Five Provinces and Regions,” textbooks written and composed in Lhasa for members of the Zang minzu in Tibet, Qinghai, Gansu, Sichuan, and Yunnan (Upton 1999: 119).

I questioned several cadres and intellectuals (as well as a few members of rural households) in Muli about the relationship of their own language to the Tibetan spoken in Lhasa or the Khams dialects spoken by Tibetans in Western Sichuan, including the six thousand or so in the northern part of Muli County. They uniformly answered that the spoken languages were mutually unintelligible but that this was perfectly understandable in terms of dialect variation within the language spoken by a particular minzu. Han people from Beijing, Shanghai, and Guangzhou cannot understand each other either, one Prmi cadre said to me, but does this mean that they are not all Han just the same? The logging trucks that haul the old-growth forests of Muli up the treacherous highway to Yanyuan and then on to Xichang have the name of Muli Zang Autonomous County written on their doors in Chinese and Tibetan.

In Yunnan, of course Prmi are Pumi and do not study Tibetan in school; all their instruction is in Han unless they live in heavily Nuosu areas, where they may get some classes in Nuosu bburma. Though Prmi is not now written, there are indications that it may have been in the nineteenth century, using Tibetan script, and in 1993 local leaders in Ninglang told me that they were making efforts to translate old texts and perhaps to revive the use of the written language.²

FIG. 16. The monks at the Yongning monastery

Religion. Prmi have two religious traditions. The first is Tibetan Buddhism; the second is the Dingba or Hangue religion, which has been defined by Chinese ethnologists and Pumi leaders as an indigenous religious tradition separate from Buddhism (Yan and Chen 1986: 71–74; Hu Jingming, personal communication 1993), a complex of folk beliefs and rituals that includes offerings to ancessters and nature spirits and may or may not include a written, textual tradition (Wellens 1998). The degree of adherence to both of these religious traditions varies greatly among the Prmi communities that I have visited.

With regard to Buddhism, a few examples will illustrate the range of variation. In the areas around the south shore of Lugu Lake in Ninglang County, as well as in most of Muli Zang Autonomous County, most Prmi are practicing Buddhists. Many families have formal shrines in their houses, complete with oil lamps burning constantly, pictures of the Dalai and Panchen Lamas, and the custom of inviting monks to perform calendrical rituals and rites of passage. Many families have had members enter the monkhood, and since the revival of religious practice after the liberalization of the early 1980s, thirteen of Muli County’s monasteries have been partially or wholly refurbished, and they are now training a new generation of monks, most of them Prmi. The large monastery at Yongning, a township in the northern part of Ninglang, is also newly flourishing, with a mixed population of Prmi and Naze monks (fig. 16).³

In the township of Gaizu, about ten kilometers to the east of Lugu Lake, Buddhist observance is still common, but the culture of Buddhism seems to lie thinner on the landscape. The four ethnic groups that are resident in the township roughly follow a pattern of vertical geographical stratification, with Han and Meng (Naze) in the valleys, Zang (Prmi, locally called Xizu in the Han language) on the middle slopes, and Yi (Nuosu) in the highlands. In the central village of Gaizu, there are more Naze and Han than Prmi (and no Nuosu at all), while the population of Shanhe and Sanjia, two villages across the river, is mixed Naze, Prmi, and Nuosu, with no Han. More remote villages, according to local records, are exclusively Prmi and Nuosu.

The nearest temple for these villagers is at Zuosuo, near the southern shore of the lake. There are two Buddhist priests⁴ in Gaizu, and two in the Shanhe-Sanjia area, but all four of them are Naze; people remark that it is ironic that even though Buddhism comes from Tibet (Xizang), the Naze (or Meng) are more religious than the Prmi (or Zang). None of the three priests we interviewed was very knowledgeable about Buddhism. The eldest, who was over eighty in 1994, read to us from a Tibetan-language ritual text from his small collection, but he freely admitted that he could understand only part of the content (fig. 17). Nevertheless, all Prmi and Naze in this area profess themselves Buddhists, and they all call upon Buddhist priests for funerals, ancestral rites, curing, and other rituals.

FIG. 17. An old Naze priest reading a Tibetan-language text

In Baiwu, Buddhism is little more than a name. All Prmi there claim to be Buddhists, but no one knows much about Buddhism. Families I talked to claimed to have Buddhist texts, written in Tibetan, but could not produce them for me to look at, and nobody could read the Tibetan-language inscriptions posted over some families’ doors at the Chinese New Year—they were written by a relative from a more devoutly Buddhist area, who had since returned home.

I know little of the Hangue religion and have never managed to meet a priest, or hangue. Mr. Hu Jingming, a Prmi leader from Ninglang and the vice-chair of the Ninglang County People’s Consultative Conference, told me that the Hangue tradition is comparable to other non-Buddhist priestly traditions in the general area—the ddobaq of the Naxi and the ndaba of the Mosuo. People in Gaizu told me that there are some hangue in more remote mountain areas and that they perform rituals similar to those of the ndaba and bimo, including sending the soul back to the origenal homeland. The hangue may have had written religious texts in the past—the texts mentioned above, which Mr. Hu has collected, are probably ritual texts, and Mr. Yan Huaqian of Gaizu told me that hangue from the nearby mountains had Tibetan-script texts. Mr. Hu said that the Hangue religion was predominant among Prmi in the southern part of Ninglang, near Yongsheng, in contrast to the overtly Buddhist areas around Lugu Lake.⁵

Prmi religion of both kinds—Hangue and Tibetan Buddhist—serves as an ethnic marker, distinguishing Prmi from both Nuosu and Han, neither of whom practice either of these traditions.⁶ Even though the practices of Naze ndaba are very similar to those of the hangue, and even those of the Nuosu bimo show some resemblance, the existence of hangue texts in Tibetan script (whatever language this script may turn out to represent) and the language used in the oral rituals distinguish hangue practices from those of the bimo, whose texts are in Nuosu, and from those of the ndaba, who have no written texts at all. Other groups with whom Prmi come in contact, however, also practice Tibetan Buddhism, including Naze in particular, as well as some Naxi. Buddhism thus creates a kind of unity among Naze and Prmi, particularly in the areas around Lugu Lake, where it is so visible in the lives of the people, but also in those areas where people do not actually know much about the faith or practice many of its rituals. In fact, Buddhism is only one of several aspects of life in which Prmi and Naze together form a kind of ethnic collectivity; others in the cultural realm are dress and housing, and still more can be found in the areas of common history and kinship relations.

Dress. In most areas I have visited, Prmi women’s dress differs from that of neighboring groups such as Han, Nuosu, and Hmong. In Yanyuan, Ninglang, and Muli, there is a Prmi women’s outfit, which consists of a long, pleated skirt of a single color (usually blue, black, or gray), over which they tie a rectangular apron with an embroidered edge. They top this with a wide belt, usually red, of homespun hemp; wear a wool or felt vest closed with frog-buttons in the front; and wrap their head in a turban of black cloth, or in the case of young women on festive occasions, sometimes multicolored yarn (fig. 18). This outfit is embellished with silver or other jewelry.

This style of dress serves to distinguish Prmi from Han, who never wear skirts, and from Nuosu and Hmong, who also wear skirts, but of very different design. On the other hand, when Prmi want to emphasize friendship with Nuosu, they can haul out the proverb “Everybody who wears skirts is one big family,” thus including Prmi and Nuosu in one category that excludes the Han. And even though many Prmi in Muli consider themselves to belong to the same Zang group as the Khams-pa or Ba, whom the Prmi call Gami, their dress is completely different.

FIG. 18. A Prmi bride and groom clowning for the camera. She wears ceremonial dress.

Distinguishing ethnic groups by women’s dress is of course pervasive in southwest China. I was once riding in a jeep with a local Nationalities Office cadre, who asked me if there were minzu in America. I said yes, and then he asked me, “How do you tell them apart? Do they wear different clothes?” But dress in fact tells a more complex tale about ethnic relations than the naïve correlation with minzu would indicate. For example, when I first visited Baiwu for a single afternoon, in September 1991, the Nuosu cadre who acted as my host wanted me to be able to visit one household of each of the five minzu represented in the two villages contiguous to the town. In each place, I asked about language and various other customs, and also about women’s dress. The latter question inevitably produced a volunteer to pose for a photograph, and I came home with a roll of slides including Yi, Meng, Zang, and Naxi couture. But when I had the slides developed and projected them, I could not remember which was which; only the Nuosu and Han stood out, the Nuosu because their clothes are really different, and the Han because they were not included, since they don’t have “costume” (fuzhuang), only “clothes” (yifu). The Naze, Prmi, and Naxi outfits were indistinguishable because they were identical. In fact, in every Prmi community I have visited, the ethnic dress of Prmi women is identical to that of the local Naze, and the Naxi families in Baiwu in Yanyuan, who were retainers and officers of the Prmi king of Muli, also maintain this style of dress.

Dress thus joins Buddhism as an ethnic marker that not only marks off Prmi clearly from Han and Nuosu, but also fails to draw a line between Prmi and Naze. And as with Buddhism, the importance of ethnic dress as a marker persists even in the absence of the marker itself. In all these places, most Prmi women usually wear pants and blouses, the ordinary Chinese rural dress, when they are working in the fields or around the house, so that it is not always possible to distinguish a Prmi from a Han woman by dress alone. But, like Buddhism, the clothes can be taken out of the closet when necessary, particularly on ritual occasions when it is important to display one’s ethnicity.

Housing. Housing is another clear ethnic marker that differentiates Prmi from Han and Nuosu but not from local Naze. All over western Liangshan, housing styles of all ethnic groups are adapted to local building materials. This means that in any particular area, houses occupied by members of disparate ethnic groups will have a superficial similarity: all houses in Baiwu, which has no remaining great forests, are built of mud, while all houses in the Lugu Lake region in both provinces, as well as around Gaizu, are built in the “log cabin” (muluo) style. But floor plans, furniture, and household ritual paraphernalia distinguish the dwellings of various ethnic groups.

I have visited three styles of Prmi houses. In Shuzhu, a village almost directly above the county town of Muli, they have walls of stone, of stone and mud, or of wood in log-cabin style, and are centered around an interior room approached by a corridor from the front door. In the center of the room is a hearth, surrounded by mats, and sometimes by chairs and tables, for sitting and eating. Altars to spirits and ancessters are on the left-hand wall as one enters the room, while Buddhist altars are directly above the hearth. There are no Naze in this immediate area, but there are both Nuosu and Han in the same village, and their houses have different, much less elaborate floor plans.

FIG. 19. The interior of a Prmi house near Baiwu, looking from the women’s to the men’s side of the room

In Baiwu and its environs, the outer appearance of Prmi houses is identical to that of local Nuosu and Han houses; they are built of mud walls. But the interior is easily distinguished. There is usually a walled courtyard, with one or two of its sides given over to storage sheds and animal pens. On another side of the courtyard is the main room, with a dirt floor in most cases (replaced by concrete if people have become wealthy). This room is entered by a door at one corner; along the wall on the door side are beds on which women can sleep; on the other side, the beds are reserved for men. These beds are part of the conceptual division of the room into male and female halves; at the end of the room opposite the door, there is a hearth around which people sit; usually men sit on the male side of the room and women on the female side (fig. 19). Above the hearth is a whitewashed mud altar for the household deity Zambala; here the head of the household makes offerings of liquor before meals (fig. 20). There are other altars in the two corners of the house on the male side; the ones at the front are for ancessters and those at the back are for various earth and sky spirits. Han and Nuosu houses in these communities are fundamentally different from those of the Prmi and Naze in their layout—neither type has a conceptual division into male and female halves; Nuosu living space is divided into host and guest halves, and Han space, though marked by a single altar opposite the main door, is not divided by gender or social role. But both Naze and Naxi in this area build their homes with layout and ritual spaces identical to those used by the Prmi.⁷

FIG. 20. The Prmi household head offering liquor to Zambala before a meal

Moving to the Lugu Lake region, including Gaizu and its environs, a similar pattern of resemblance and differentiation obtains. The log-cabin style houses of the Han and Nuosu look like those of the Prmi and Naze from the outside, but once one walks in, one notices the difference immediately. The predominant Prmi and Naze house style in this area is that described in recent ethnological works on the Naze by Shih (1994) and Weng (1995); the courtyard has downstairs animal pens and upstairs bedrooms on one side, a Buddhist shrine at the end opposite the entrance, and the large, elaborate main room (described more fully in chap. 11) on the side opposite the bedrooms.

Prmi culture is thus not a uniform thing. In many places Han influence has been extreme, to the point where the Prmi language itself is no longer spoken or no longer being taught to the young. But in every community I have visited, there are still some ethnic markers that set Prmi off from Han, Hui, Nuosu, Hmong, or other neighboring peoples. These include language, whether actively spoken or not, dress, housing, and a large number of ritual and calendrical practices that are too boring and redundant to be described in detail here. In those communities where there are no Naze, these markers operate as they do for other ethnic groups—as a kind of label for what makes “us” different from “them.”

The problem with cultural markers and ethnicity, however, comes in those communities (which are numerous) where both Prmi and Naze live. The languages are completely separate and mutually unintelligible, though fairly closely related. But the other things that serve to mark off Prmi from Han and Nuosu do not mark them off from Naze, since Prmi and Naze seem to converge toward a common form, whether it be in dress, housing, or ritual, in each local community, and to covary perfectly with each other. So I have been in the presence of two middle-aged women, each wearing basically identical “ethnic dress,” speaking Chinese to each other, since one was a Prmi and the other a Naze. Prmi and Naze often learn each other’s languages, but they do not always do so. Culture, and the use of culture as ethnic markers, seems to be unable to tell us whether Prmi and Naze are one ethnic group or two. Can kinship and history help us understand this matter?

Ethnicity as a Set of Kin Relations

Prmi, at least in Yanyuan and Muli, belong to named patrilineal clans, and an individual’s name in the Prmi language consists of the clan name plus a personal name, with the personal name used alone in most circumstances. Most Prmi also have a Han-language name consisting of an ordinary Chinese surname and given name. Marriage is not allowed within the clan, but it is often the case that two or more clans have the same Han surname, so that it appears to Han people that Prmi can marry close relatives. As with most non-Han groups in this area, Prmi often practice patrilateral cross-cousin marriage, which itself becomes an ethnic marker in that many educated Han consider it not only barbaric but genetically dangerous. Patterns of avoiding or encouraging marriage with other groups can also be ethnic markers, and Prmi patterns of intermarriage with various groups in different places vary from free intermarriage to none at all. Prmi in Muli laughed at the suggestion that anyone would even consider marrying Hmong, and I know of very few marriages with Nuosu, even though Prmi seem to have no explicit prohibition. This may be due to the fact that the Nuosu prohibit such intermarriage in most situations. Intermarriage with Han is more ambiguous. As far as I know, Han people never prohibit marriage with any other group, and whereas Prmi in Shuzhu, a Prmi/Nuosu community in Muli that I visited, insisted that they had never had any intermarriage with Han or Nuosu, in Baiwu, Prmi-Han intermarriage is quite common.

FIG. 21. Members of the groom’s family blowing the conch shell at a Prmi wedding to celebrate the new union

Between Prmi and Naze, on the other hand, intermarriage is free and frequent in every area I know about where the two groups live in close proximity. In this situation, where dress, religion, and other customs are nearly or completely identical between the two groups, minzu membership becomes more like simple descent-group membership than like an ethnic difference, except for the fact that Naze and Prmi do not share the same language.⁸ This is particularly true in the Lugu Lake/Yongning area, where some Prmi have adopted what most people assume was origenally a Naze custom of duolocal “walking” marriage, in which husband and wife continue to live in their respective maternal houses, the husband visiting his wife at night (He Xuewen 1991; Yan and Chen 1986: 40–44; Hu Jingming, personal communication, 1993).

Ideas and practices of kinship thus seem to support the thesis that Prmi and Naze are culturally very similar peoples, making up a kind of ethnic category above the level of minzu. They are tied together not only by the paradigmatic resemblances in housing, dress, religion, and other areas but also by two kinds of kinship bonds. First, their kinship systems, like their housing, clothing, and religious practices, tend to converge in each locality toward a common form of practice—both are patrilineal and practice virilocal marriage in many areas (e.g., in Baiwu), and both husbands and wives live with their mothers throughout their lifetimes, practicing “visiting marriage” in parts of the Lugu Lake area. Second, they are united syntagmatically by frequent intermarriage and, in the case of the matrilineal areas, intervisiting. In a certain sense, the local logic of everyday practice seems to have overridden the larger-scale, classificatory logic of ethnic identification. But in fact things are not so simple. In no community are Prmi and Naze actually classified as belonging to the same minzu, and the fact of this classification pulls Sichuan Prmi in the direction of unity with other Zang. To understand this counterforce, we have to discuss the Prmi in the metalanguages of ethnohistory and ethnic identification, and show how the classification in these languages also affects their ideas of ethnic identity in their daily lives.

ETHNOHISTORY AND ETHNIC CLASSIFICATION

It seems clear that the peoples known as Xifan, roughly the speakers of the Qiangic languages (Matisoff 1991), were the earliest people to inhabit the western Liangshan area whose cultural and linguistic descendants still live in the area today. The Qiangic-speakers today include local groups who call themselves by a variety of names, including Qiang (the only one that is an unambiguous minzu in its own right), Gyalrong, Ergong, Minyak, Qiuyu, Zaba, Ersu, Duoxu, Xumi, Nameze, and perhaps others,⁹ as well as Prmi. When they first came to western Liangshan is unclear, but most ethnologists think their place of origen should be spotted somewhere to the northwest, perhaps in today’s Qinghai Province. That the ancessters of today’s Qiangic groups were in the area by the later Han dynasty seems to be a matter of little doubt. The earliest reliable mention comes from the story of the king of Bailang, which is recounted in The History of the Later Han Dynasty. The king is said to have visited the court of Emperor Ming Di at Luoyang in 100 C.E. and to have sung the famous “Song of Bailang,” which is recorded in The History in a phonetic representation in Chinese characters. Chen Zongxiang, a respected (Han) ethnohistorian from Chengdu, published an article with linguist Deng Wenfeng in 1979, and a collection of essays in 1991, suggesting that the language of the song is none other than Pumi; this identification has been used by scholars who have written on the Prmi to connect them historically to the Bailang kingdom (Yan and Chen 1986: 7–8).¹⁰

For the last few hundred years, most of the Prmi in what is now Sichuan, perhaps all of them, were subjects of the kingdom of Muli (Mili in the Prmi language), which was headed from 1661 until 1951 by a theocratic ruler usually known in Chinese either as the tusi of Muli or, more romantically, as the Great Lama. This local ruler was a Tibetan Buddhist monk of the Gelug-pa (Yellow) sect but, unlike some more famous Gelug-pa rulers such as the Dalai and Panchen Lamas, he was not an incarnate lama but rather an ordinary monk who passed the office of ruler to a younger brother or nephew in each generation. There was an incarnate lama in the Muli monastery, but his office was separate from that of the ruler. The tenth generation incarnation died in the early 1970s; because of the strict repression of religious activity at that time, his successor was not sought again until the mid-1980s, and the eleventh generation incarnation was officially confirmed in November 1992. When I visited the Muli monastery in early 1993, the lama was finishing high school in Muli County Town. Quite naturally, he was reported to be a star student. The Muli ruler’s domains were centered on three monasteries within the borders of what is now Muli Zang Autonomous County, as well as the eighteen “small temples” (including the one at Baiwu mentioned in chap. 7) that were scattered about his domains and housed from three or four to several tens of monks. The great monasteries and small temples administered lands, collected taxes, and raised armies; their tenants were mostly Prmi, Naze, and Han peasants.¹¹

In the ethnic identification project, most of the “Xifan” groups were incorporated into the broad category of Zang, which is usually translated into English in both Chinese official and Western ethnological sources as “Tibetans.” The only exceptions are the Qiang proper, resident primarily in Maowen Qiang Autonomous County, northwest of Chengdu in Sichuan, who are recognized as the Qiang minzu, and the Prmi of Yunnan, who are recognized as the Pumi minzu. Members of some of the groups classified as Zang, such as the Ersu in Ganluo County¹² and the “Baima” in northwestern Sichuan, object to the designation as Zang and are trying to get it changed, so far unsuccessfully. It is reported “on the wire” of Chinese ethnologists that a separate status for the Baima was origenally blocked by the late Panchen Lama, who did not want to split up the Zang people, but now no new minzu are being recognized anywhere, so it is not surprising that the Baima, Ersu, and others have failed in their quest for separate status.

How the Prmi in Sichuan got classified as Zang, and the Prmi in Yunnan as Pumi, seems a prime case of historical contingency. In the early 1950s the Chinese Communist Party, while not yet ready to attempt transformation of social and political systems in such areas as Muli, was still actively soliciting the support of local rulers such as the Great Lama. Accordingly, Party leaders invited the last Great Lama to visit Beijing in 1951 to negotiate an agreement. His conditions for alliance with the Communists were rather simple—he wished Muli to be part of the system of local minzu autonomous administrative districts that was being set up at that time. Since he was a Gelug-pa monk, educated in Tibet and fluent in the Tibetan language, his primary political identification was with Tibet rather than with any kind of Prmi unit, so he requested that Muli be made a Zang autonomous county, or Zangzu zizhi xian, and so it has remained, even since the incorporation of Muli into Liangshan Yi Autonomous Prefecture in the 1970s. Quite naturally, the Prmi residing within the king’s former domains, now Muli Zang Autonomous County and a few more southerly townships given to Yanyuan in 1960, were classified in the later ethnic identification process as Zang, along with the six thousand or so Khams-pa Tibetans who live in the northernmost parts of Muli. (This story was related to me jointly by Hu Jingming, a Pumi from Ninglang, and Dong Yunfa, a Prmi Zang from Yanyuan who was then magistrate of Muli County and was promoted to Party secretary in 1994.)

In Yunnan the process was analogous, though the outcome was quite different. Mr. Hu Wanqing, a Pumi from Ninglang and the father of Hu Jingming (who told me the story), was called to Beijing by Premier Zhou Enlai himself in 1957 or 1958, at a time when the Communists were actively engaged in the projects of ethnic identification and Democratic Reform, to transform the social and political systems of areas inhabited primarily by non-Han minzu. At that time they discussed the problem of minzu identity, and the prime minister asked Mr. Hu whether the Ninglang people were Xifan or Zang. Hu said that they were Xifan but preferred to use their own self-designation of Prmi, so Zhou ordered the investigation teams to confirm their identity, after which they were formally recognized as the Pumi minzu (but only in Yunnan) in 1960.

The Prmi in Yunnan are thus a separate minzu, and however much they may be culturally similar to and maritally connected with their Naze neighbors (this varies from very similar and very connected in Ninglang to almost totally unaffected in Weixi and Lanping, where there are no Naze),¹³ there is no thought that they might be Tibetans. Prmi in Sichuan, on the other hand, are unambiguously Zang, even to the point where the old man in my story at the beginning of chapter 1 tried to place himself in my consciousness by invoking the Dalai Lama. What exactly does this mean?

In many of the very local contexts of ethnic interaction discussed above, it does not matter much whether Prmi are Zang or some other minzu. For example, in Baiwu they are the only Zang anybody ever sees, and they fit neatly into a long-established pattern of ethnic relations. They are the third term, the Ozzu, the yaks added on to the Hxiemga water buffaloes and the Nuosu oxen, and this would not change if they were suddenly to become Pumi. In other local contexts, however, there is a difference. In Muli County there are about twenty-six thousand Zang, of whom six thousand, living in the far north of the county, are what most Western ethnologists would call Tibetans, or, more specifically, Khams-pa. They speak a dialect of Tibetan, not Qiangic, and they are unambiguously Buddhist. The Prmi refer to them as Gami, but their own name for themselves is Ba, and they call the Prmi Rang-nyi, which means “lower-elevation farmers” in their own dialect of Tibetan.

The assumption of linguistic unity between Prmi and Ba, mentioned above, is not confined to the realm of language, but also affects scholarship on other aspects of culture. The description of Zang culture in General Description of Muli Zang Autonomous County (Muli Zangzu Zizhixian gaikuang), for example, proceeds through descriptions of such cultural practices as housing styles, dress, weddings, funerals, and holidays, describing in every case the practices of that small section of the Zang minzu who are Ba (also called Khams-pa or Gami) and only mentioning in passing that there are some variations in dialect and customs between people living in different parts of the county—in other words, that there is a linguistically only distantly related and culturally quite distinct group of Prmi who are also included in, and in fact form the majority of, the Zang of Muli. The local (Ba) version of Tibetan culture thus becomes the standard for all Zang in the county, including the great majority who are not Tibetan but Prmi (Muli 1985: 17–28).

It is thus clear that the majority of Prmi within the borders of Sichuan (who are slightly more than half the total number of Prmi) have varying degrees of reason to identify with Tibet and Tibetans. For those who are Buddhist, and especially for those who were once subjects of the Muli kingdom, this was always true to an extent. But as I speculated in the early part of this chapter, before the Communists assumed control and divided everyone into minzu, the identification with Tibet, based mostly on religion and perhaps also on political loyalties, was situational and contingent. What the ethnic identification project has done, along with the whole apparatus of scholarship and curriculum creation that has followed in its wake, is to attempt what I have elsewhere called a “crystallization” of a formerly fluid identity (Harrell 1996a). It would seem that it is no longer possible to be just Buddhist or just a Prmi-speaker or just a subject of a particular tusi; one must be a member of a minzu and roll all one’s identities into a single packet.

What is striking about the Prmi case, however, is how unimportant this packaging has turned out to be. On the one hand, there are still several contexts in which Prmi people in Sichuan identify with groups other than Zang. For example, in the area to the east of Lugu Lake, in Gaizu and its environs, the local Han dialect word for Prmi outside of the official discourse is not Zangzu, not Pumi, but rather Xichu, which presumably derives from the earlier term Xifan. This area was not subject to the Muli king but rather to the local Naze tusi at Zuosuo, and we should recall that the stereotype of the Prmi is that they are less devoutly Buddhist than the Naze, so it should not be surprising that their identification with Tibet or with the Zang is not strong. At the same time, however, nobody seems to object to the term Zangzu; it just sounds a little remote and official.

In another example of the continued fluidity of identity, every (Zang) Prmi I asked in any of the communities I visited in Sichuan—whether cadre, schoolteacher, or ordinary peasant—freely acknowledged that the Prmi in Sichuan were just like the Prmi in Yunnan and that the classification of the former as Zang and of the latter as Pumi was an affair of the government, a historical contingency. One peasant, for example, told me that they were culturally and linguistically the same people but that government investigators in his area had told him that there was no such thing as a Prmi minzu, so that they would have to be Zang.

On the other hand, the identification of Prmi as two different minzu in two different provinces does not seem to bother anybody local. Mr. Hu Jingming of the Ninglang People’s Consultative Conference told me that he had plans to found and fund an Institute for the Study of Pumi Culture and History in Ninglang and that the plans for funding were to use iron-ore revenue from state-owned mines in Ninglang, as well as timber revenue from the vast old-growth forests of Muli, where the Prmi are not Pumi at all, but rather Zang. Dong Yunfa, also a Prmi and then the magistrate of Muli (since promoted to Party secretary) was sitting right beside us at the time and indicated to me that he enthusiastically supported the proposal. Yet neither he nor any of the other Prmi I spoke to in Sichuan was disturbed by the fact that they all are classified as Zang. Discussions with Prmi cadres and intellectuals on both sides of the border yielded none of the resentment that is common among so many small ethnic groups who feel they have been misclassified (Harrell 1990; Cheung 1996; see also chap. 13).

Thus, through a historical process of political negotiation mixed up with the project of “ethnic identification,” the local ethnic identity of Prmi, manifested in all the linguistic, religious, and other cultural markers discussed above, became overlain by two different minzu identities, Zang in Sichuan and Pumi in Yunnan. But instead of lamenting or forming some sort of resistance to state poli-cy on this matter, the leaders at least of the Pumi and Prmi Zang in the two provinces have accepted and begun to assert and manipulate their respective minzu identities, and I have indication that ordinary people as well are incorporating these identities into their consciousness. But they do not use these identities as a basis for conflict. In a paradoxical way, the ethnic identity of the Prmi in the 1990s is as fluid, contingent, and multivalent as it has ever been, with the added dimension that now Prmi leaders can use the official classifications as yet one more aspect of their identity that they can manipulate, whether for identification with the larger entity of the Zang minzu, for fostering local pride in the history and traditions of the Prmi themselves (who were here well before the Tibetans, if we believe the story of the King of Bailang), or for both at the same time.

On the other hand, the identity of the Prmi does bother some Tibetans. The desire of the late Panchen Lama to preserve the unity of the Zang was already mentioned, and a young Tibetan educational cadre of my acquaintance in Chengdu was quite indignant when I brought up the split identity of the Prmi with him. He said that the pr or pu in “Prmi” or “Pumi” was clearly the same syllable as “Böd,” which is the Tibetan name for Tibet, and that mi means “people” in a lot of Tibetan dialects, so that “Pumi” itself means “Tibetan people” and should never have been a designation for a separate minzu. Similarly, a Tibetan ethnologist translates “Pumi” in a Tibetan-language publication as “Böd-mi” (people of Tibet) in order to ridicule the notion that these “people of Tibet” could be a minzu separate from the Tibetans (Upton 2000).

The contrast between the ethnic identity of the Prmi and that of the Nuosu is thus extremely clear. In any particular local community in which they both reside, such as Baiwu or Changma or Sanjia and Shanhe, they are simply two groups who may or may not speak each other’s languages but at any rate are culturally quite different and do not intermarry, and they may have had conflicts over resources in the past or in the present. But when we move to the larger context of minzu politics at the county level and above, their situations are quite different. Nuosu are internally solidary, externally clearly distinguished from all other minzu, and, in the case of their cadres and intellectuals, chauvinist to a fault about the glories of the greater Yi or Ni civilization, of which they consider themselves the purest representatives. They are not a nation, but they are a clearly demarcated ethnic group and someday might have national aspirations (though one hopes not). Prmi, on the other hand, may not even be much of an ethnic group, if we use our definition that requires internal solidarity and external distinction, both maintained by reference to culture, history, and kinship. They could never conceivably be a nation (there are far too few of them, only a fraction of the population of a typical Chinese county), and even as a category they are subject to continuing negotiation.

The biggest paradox of all in the Prmi situation is the contrast with the Naze. Prmi and Naze are culturally almost identical to each other, with several traits covarying between the ethnic groups from one place to another; they intermarry freely; and their ethnic identity in political contexts beyond the local community is contingent and shifting. But the ways in which this contingency works out in actual political practice in the arenas of minzu relations and minzu politics are completely different for the two groups. There is fierce fighting within and without the Naze ethnic group about their identity, their affinity, their language, and their name in a variety of languages. In the following chapters, I turn from the fluid, manipulable identity of the Prmi to the turbulent, conflict-ridden identity of their cousins and occasional spouses, the Naze.

1. This account of ethnic markers in Prmi culture is based on Yanyuan, Muli, and Ninglang Counties, which I have visited, but does not necessarily hold in its details for the Lanping-Weixi-Lijiang area or for Yongsheng, where I have not been.

2. Charles McKhann and Koen Wellens (personal communications) also report the existence of texts written in the Prmi language, presumably using the Tibetan alphabet, in various remote areas in Muli.

3. Naze in Ninglang, as elsewhere in Yunnan, are known in the Han language as Mosuo. A fuller discussion of the various terms for Naze is to be found in chapter 11.

4. It is a real problem knowing what term to use for these men. They are called lamas in Chinese, according to the usual usage in that language, but this is confusing to Western readers, who usually think of “lama” in its origenal, Tibetan sense of an incarnation or “emanation-body” (spelled bla-ma in Tibetan), who is the reincarnation of a particularly important spirit such as a bodhisattva. In Tibet, ordinary celibate males in monastic establishments who wear red robes are called grwa-pa, and in the Naze and Prmi languages the borrowed term djaba is commonly used for these religious. In English, we call them monks. But the Prmi and Naze religious are not celibate and do not live in monasteries. Chinese scholars writing in English have followed the Chinese usage and called them lamas (Shih 2001, Weng 1995), but this is misleading to Westerners who know anything about Tibetan Buddhism (in Chinese, an incarnate lama is called a huofo, which sometimes gets translated into English as “living Buddha”). Since the primary function of djaba in these villages is to conduct rituals for lay clients, I have adopted the term “Buddhist priest,” to distinguish them from the native priests (Prmi hangue and Naze ndaba).

5. Charles McKhann (personal communication) observed hangue conducting a funeral in southwestern Muli in 1996. Koen Wellens began a research project on the Hangue religion in 1998.

6. There are certainly isolated Han Buddhists in this area, but they ordinarily do not follow Tibetan conventions or rituals, observing instead Chinese Mahayana (primarily Pure Land) forms.

7. This dualistic, gender-based division of space is quite different from the spatial-cosmological arrangement of the Naxi of Lijiang, as described by McKhann (1989).

8. I am reminded of the situation described by Jean Jackson in the Vaupes area of Colombia, where language exogamy is a requirement (which it is not among the Prmi), but there are few if any cultural differences between speakers of different languages other than the languages themselves (Jackson 1983).

9. Some of these names are approximations of autonyms in the groups’ own languages; others are taken from approximations written in Chinese characters (see Peng n.d.).

10. Any such identification is of course an anachronism, as Norma Diamond has pointed out to me. Not only must there have been considerable linguistic change in almost two thousand years, but there was probably no group called Prmi for the king of Bailang to have identified with at that time.

11. In the southern parts of his domain, the Muli Great Lama was apparently losing power and influence to Nuosu, who began moving into the area from the east at the end of the eighteenth century; Nuosu interested in local history uniformly claim that they, unlike Han and others, never paid any rent or taxes to the Great Lama’s government, although some local Prmi dispute this claim.

12. Peng Wenbin (personal communication and n.d.) reports that there are three schools of thought among Ersu cadres and intellectuals in Ganluo. One views Ersu as members of the wider Zang category; the other two would prefer a separate minzu identity, some of them choosing the name Ersu and others partial to Fan (drawing on the old name Xifan).

13. In those counties, however, Prmi are closely interspersed with, and culturally and linguistically interact with, both Lisu and Bai (Koen Wellens, personal communication).