13. Becoming Chinese? Ethnic Transformation and Ancestral Tablets among the Puyuma of Taiwan

Suenari Michio

This paper addresses the processes of ethnic transformation that have occurred in Taiwan and the South China region, focusing specifically on a group of native Taiwanese who have gradually adopted Chinese customs.[1] The paper focuses specifically on transformations in the ancestor worship cult among the Puyuma peoples. Over the past few decades the Puyuma have gradually begun to accept the custom of worshipping ancestors by establishing ancestral tablets in their own homes. This custom has been adopted from neighbours who are themselves ethnic Chinese. The data were collected during field research from 1966 to 1968, and again for a brief period in 1983. During the interlude between these two investigations, many changes became obvious. This paper thus discusses the process of ethnic transformation as the Puyuma went through periods of hesitation, resistance, and finally full-scale, accelerated acceptance of Chinese customs.
Today most of the aboriginal peoples of Taiwan have gone through a long process of sinicisation whereby they have accepted many Chinese customs. The processes of sinicisation have never been as clear-cut as many descriptions by historians would lead us to believe. This investigation, along with the work of others, shows that there has been a constant process of transformation and ethnic redefinition over the past several centuries in Taiwan. It is impossible to locate a central tendency or orientation that will explain all of the processes of sinicisation involved. Differences in social organisation among contemporary ethnic groups can be viewed as the results of multiple interactions with the dominant Chinese culture. This particular study looks at only one case and generalisations deriving from this case will have to be used with caution when applied to other groups.
A study that focuses on ancestral tablets has a certain advantage in that we are dealing with concrete and easily observable materials. Ancestral tablets also have the decided advantage of containing written information. In this paper I use ancestor tablets as one means of judging and evaluating the process of sinicisation among the Puyuma.[2]

General background

The Puyuma people are an aboriginal group currently living on the west edge of the Taitung Plain, located on the east coast of Taiwan.[3] The village of Rikavon is eleven kilometers west of Taitung, a prefectural city which developed from a small hamlet of twenty-eight houses in 1872. The Puyuma had been famous for their military bravery and magical power, dominating the plain area in spite of their relatively small population. In 1915 the Puyuma consisted of 6,476 people. By 1964 the population was 6,335, indicating a slow but steady decline. In contrast, the Ami, a neighbouring group that lives on the east coast of Taiwan, doubled their population during the same period (from 1915-1964).
The traditional subsistence patterns of the Puyuma consisted of agriculture, which was done by both men and women, plus specialised hunting by men. Hunting was not economically critical for subsistence, but it became a major activity related with status for men and gave much meaning to the maintenance of community life. The major crop was millet, which is still important for ritual activities. The Puyuma also produced rice, taro, potatoes, sweet potatoes, and beans. Today they cultivate rice in paddy fields, which were introduced at the beginning of the current century. Other major crops include sugar cane, pineapple, banana, and areca nuts.
Puyuma traditionally lived in permanent villages with an average population of approximately 600. The villages tended to be organised around the principles of endogamy and political autonomy. The inhabitants fell into two distinct categories, nobles and commoners. Intermarriage between these two categories was not prohibited. The noble group consisted of members of certain chiefly families, of whom there were several in each village. Each of these chiefly families maintained its own men's house. The household heads of these chiefly families exercised political control. The office of village headman also tended to be held by members of these chiefly families (Mabuchi, 1960: 134-35).
During the Japanese colonial era, Puyuma villages became administrative units and many were relocated to be close to newly established paddy fields. The status and privileges of the chiefly families decreased, even though Japanese officials often relied on them for administrative purposes. At the end of the Second World War, the distinction between commoner and chiefly families was formally abolished by state policy. The vestiges of this cultural system can only be seen today in the consciousness of older people and in their performance of certain rituals.






The ideal family type among the Puyuma was an extended unit, in which daughters usually lived in their natal household after marriage. As a result of division, usually only one daughter remained with her parents. It also happened that a son might remain and succeed in his natal household. Although there have been many changes during the past 100 years, the Puyuma were primarily an ambilineal people loosely organised around family property. The expected form of marriage was uxorilocality, with the husband moving into the household of his mother -in-law. The wife's male kinsmen had more decision-making powers about property and the disposition of agricultural land than did the in-marrying husband. The central significance of the wife's male kinsmen is expressed in the marriage rituals and in the distribution of engagement presents, which is also supervised by the wife's mother's brother.
One of the central features of the Puyuma kinship system is an emphasis on personal kindred. The range of recognised kin extends to third cousins and sometimes even to sixth cousins. While the kindred provides the Puyuma with a functionally important system which operates in everyday life, many people also participate in ambilineal groups which are centred on ritual houses (karomahan) of varying sizes. These ritual houses focus on the performance of certain divinatory rituals, associated with illness. Individual Puyuma can belong to more than one of these divinatory groups, which trace descent ambilineally to the ancestors who are deemed to be causing the afflictions.
Marriage among the Puyuma is primarily monogamous and involves no significant brideprice. As noted above, post-marital residence was primarily uxorilocal. Although marriage between close kin is not encouraged, there are numerous cases in my field data which record marriage among second cousins, the major reason being the desire to keep property within a group of close kinsmen. The strong tendency for village endogamy is no doubt related to the avoidance between affines, who might have become engaged in cross-village fighting over hegemony. Among Puyuma affinal relations are clearly prescribed and reflected in carefully observed rules that regulate the exchange of gifts between affines (Suenari, 1968).
This paper deals primarily with ethnographic data collected in Rikavon Village. Originally this was a community divided into three discrete hamlets. Under Japanese rule, however, these hamlets were joined together and moved to the present location, which was laid out with streets arranged in a grid pattern. In 1968 Rikavon had a population of 925 people living in 150 households.
Chinese ethnic infiltration into the Puyuma territory had begun in the middle of the Qing era. Chinese peddlers and itinerant merchants often took Puyuma women for their wives, on the understanding that this would help provide them with safety and kinship links within the local community. Chinese peddlers who lived in Puyuma communities behaved much like ordinary husbands who married uxorilocally. The Puyuma had no strong sense of ethnic distinctions which set them apart from in-marrying outsiders, as long as these outsiders lived within the village and followed the Puyuma way of life. Although the descendants of these original Chinese peddlers have been culturally absorbed into Puyuma society, they still maintain a "memory" of their Chinese ancestor and his origin. Descendants who claim this link with a Chinese ancestor often establish a Chinese style ancestor tablet, assuming their Chinese ancestors would be happier to be commemorated in the Chinese way. It is to be noted that this does not actively involve boasting of Chinese descent.
After the end of the Second World War, the Han Chinese rushed to the east coast because of the population pressure in the west. This tripled the population of the Han and reduced the aboriginal population in the east coast from majority to minority as is seen in Table 1. This reversal of percentages had a profound effect in the relations between the aborigines and the Han Chinese.
There are many factors which have led to a change in the Puyuma traditional style of ancestor worship. We cannot neglect the influence of the Japanese occupation which occurred from 1895 to 1945. During this time Japanese colonial officers, including school teachers and police who tried to "improve" the life of the people, operated in the Puyuma region. They also were charged with eradicating "superstition" or "bad habits" among the Puyuma. The latter included the vestiges of headhunting, burial within the house and various forms of divination that Japanese authorities defined as unsuitable and unacceptable. The Japanese also encouraged formal education by establishing a system of schools. The cult rituals at the village level of the karomahan (the Puyuma shrine consecrated to ambilineally traced ancestors) were abolished by the Japanese authorities. This had the effect of undermining the unity of local village communities among the Puyuma. Furthermore, there was a Christianisation movement that began in this region soon after the Second World War (see Table 2). A highlight of this new development was the distribution of food and clothes which had the effect of attracting large numbers of "converts" who were then expected to abandon their traditional religious activities (see Table 3). It is clear, therefore, that the process of sinicisation described in this paper has been the result of many other influences besides accommodating and assimilating to dominant Chinese cultural models. One cannot forget the effects of both colonisation by the Japanese government and official ideas imported from the Christian church.[4]

Ancestral tablets among the Puyuma

In 1968 during my first field investigation, about one third of the 150 households could be classified as "Buddhist" and thirty-three of these households had established an ancestral tablet on the altar in the central room of their house.[5] The size and style of the tablets were not uniform, but took two basic forms: (1) a wooden tablet in the form of a plaque, and (2) tablets contained in a box. The former type was further subdivided into tablets that had been painted with black lacquer or were simply plain wood. The names of the ancestors were inscribed collectively on the face of these tablets. The latter type consisted of a box containing slips of wood on which were written the name, age at death, time and date of birth and of death. On the front of the box was written simply "the supernatural seat for the ancestors for every generation of such and such a house". Each box would then contain a number of individual ancestor slips. Because of its convenience for containing many ancestors, the second type (the box) became more and more popular. By the 1980s, almost all families requested this box-type ancestral tablet container when they set up a new ancestral altar.
Most of these Chinese-style ancestor tablets were introduced after the Second World War, particularly following the demise of many Christian converts. Following the procedures of neighbouring Chinese villages, the Puyuma purchased these articles in a nearby town where they would request an experienced Chinese calligrapher to write out the slips that are contained within the box. Today, it is almost impossible to distinguish these Puyuma tablets from those of their neighbouring Chinese, except for the missing details of the dates and the use of non-Han names and Japanese kana inscriptions (i.e., non-Chinese writing).
Investigations of particular tablet boxes indicate that the names of the ancestors actually inscribed on the slips do not always correspond with what people believe to be contained within. There is no regular occasion for Puyuma to open the tablet box periodically to clean and inspect the slips. This paper begins with an exploration of the contemporary Puyuma understanding of who they think is actually contained within these boxes; later I will examine the actual names that are physically written down on the tablet slips.

Custody of tablets

Puyuma ancestral tablets are usually entrusted to those who succeed to the natal house (pahatayan). Traditionally, it was the eldest daughter who was usually supposed to remain in the natal house and continue the traditions of that household. Accordingly, if the Puyuma custom were followed it was the eldest daughter who would succeed to the custodianship of the tablets. This is in direct contrast to the Japanese and the Korean cases in which tablet custodianship is transmitted through the eldest son. Among the Puyuma, the preparation of the offerings for the ancestral rites is sometimes shared out among other descendants. Some descendants of the natal house may also bring their own offerings, as is the case among some neighbouring Chinese.
Table 3 shows who actually succeeds to the custodianship of the tablets. In eighteen out of thirty-three cases investigated, the eldest daughter was the custodian. But, if we eliminate those cases in which there were either no daughters or only one daughter, this leaves fifteen cases: in eight of them the tablets were in the hands of the eldest daughter, and in the other seven they were not. So, nearly half the actual cases deviate from the norm of the eldest daughter succeeding within her natal household.
From the analysis of the cases, we might note that the following factors also affect the custodianship of Puyuma tablets. The kinship relationship by birth is obviously basic. The number of children through which the tablets might be transmitted is also relevant when considering custodianship. Also important is the distribution of inherited property, as well as divination to determine the views of the ancestors.

The cult of the ancestral tablets

In the following sections we will investigate how tablets are actually treated. A key informant among the Puyuma maintains that the Puyuma worship their tablets on the first and the fifteenth of each lunar month. During this time they place incense and tea before the tablets. Ideally, the entire family also gathers before the tablets on four other occasions: on the fifth day of the fifth lunar month and on the fifteenth day of the seventh lunar month as well as the fifteenth day of the eighth lunar month, and guonian (the last day of the twelfth lunar month). In actual practice, however, most Puyuma only worship on the fifteenth day of the seventh lunar month and guonian. Even those Puyuma who have accepted many Chinese customs do not necessarily follow the Chinese festival cycle. Below, we investigate cases of ancestor tablet worship on the occasion of the guonian ritual.
Example 1. Five brothers gather each year at the guonian ritual to worship their ancestors in the natal house (Figure 1). The eldest brother (A) has succeeded to the natal house. His brother, B, a Protestant Christian, does not participate in the worship and only comes to share in the dinner. Brother C has married out uxorilocally. Although he attends, his children do not come with him. The children join their mother as she performs rites elsewhere. Brother E comes with his two daughters, but his wife returns to her natal house for the guonian ritual. A's wife, on the other hand, participates with her husband in this ritual in the natal house. This is because her natal house is in another village. The other four wives of the brothers do not participate in the ritual, even if they do come with their husbands and help with preparing the dinner. This set of brothers are not wealthy and no substantial property has been divided among them. The offerings are simple and the relatives remain to enjoy a party with dinner and drinking.
Example 2. The eldest daughter, A, takes care of the ancestral tablets (Figure 2). Her two sisters B and D have branched out and sister C and brother E have also married out. A's husband's natal family believes in what he refers to as "the religion of the Puyuma" and will have nothing to do with this particular event on this day. B attends with all of her children except her eldest daughter who lives in a distant city. A's adopted daughter is living with her husband in another village and comes back by herself to participate in the worship. Sister D comes to worship by herself because her husband worships the tablets of his own ancestors in her own house. After finishing the worship here, sister D returns to her own house and will worship there with her husband to celebrate his own ancestors. At her own home, sister D kills two chickens and prepares these as offerings for both sets of ancestors, including those of her husband. Her own children are not concerned about these events and only her daughter will attend to help. Brother E worships at his wife's house into which he has married. C has married out and does not come back for the worship. Even though her present house is very close to sister A's, sister C does not attend this ritual.
From the examples above, we might highlight the following general principles. (1) Irrespective of post-marital residence, people participate in the worship of the ancestors in their own natal house. (2) Adopted children worship at the house of their foster parents. (3) The spouse may participate as a courtesy, rather than as a member of the family. (4) One may worship in two places, if this is possible given transportation time. (5) The actual residence of one's own parents and the distribution of inherited property are not always decisive factors in determining who will and who will not worship particular ancestral tablets. It is sometimes the case that people will worship at an altar at which it is designated to be "fitting" and suitable to do so. These features are in contrast to the general patterns of the Chinese ancestor worship cult that operates among neighbouring Chinese people.

Ancestors of the household

Among the Puyuma, the most dominant form of ancestor worship is the veneration of tablets by those who remain within the household. Those who marry in often bring tablets of their ancestors with them if there is nobody to take care of them within their natal family.
Example 3. Figure 3 shows such a typical case. In this case a daughter has the responsibility for worshipping the tablets of her parents, plus the tablets of her father's parents and also those of her husband.
Example 4. In Figure 4, a somewhat different pattern emerges. The focal person A had a sexual liaison with a Chinese mainlander D and left the community. Her ex-husband E retains custody of their son B and continues to live in the house which holds the tablets of his wife's ancestors. The man in question E worships his own ancestors at the house of his elder sister. Her younger brother is a Protestant Christian and does not dare to worship ancestors offering incense. This case demonstrates how the worship of ancestral tablets cannot be fully explained by analyzing kinship relations alone. One must also consider the concept of the house and the choice of residence after marriage as being important factors.

Ancestors of the in-marrying and the adopted

To the ethnographic eye familiar with tablets that conform to the patrilineal Chinese descent system, Puyuma tablets may appear to be rather strange. One finds, for instance, many ancestors of in-marrying people who are not within the strict patrilineal system. One also finds many cases of adoption which are listed openly. The following is a case in point:
Example 5. Ego is a woman born into a Chinese family; she was soon fostered out to a Puyuma family. She fell ill and divination showed that her illness was deemed to result from her deceased, biological Chinese mother who was said to be "hungry". Offerings of food to Ego's Chinese mother soon cured Ego's illness. A Japanese policeman in the village, however, advised her, "You need not worship your ancestors in the Chinese way, since you were adopted by the Puyuma. Even if your ancestor is Chinese you do not need to do this." Therefore, Ego did not make a tablet for her Chinese biological mother. After the delivery of Ego's second son in 1949, her breasts began to swell and she became faint. She went to ask the local Puyuma shaman to find out what the trouble was. She was told to prepare an ancestral tablet for her Chinese mother which she did. By that time, the Japanese police had already left and there was no one to be concerned about such matters.
Example 6. After living with her foster parents for six months after her marriage, Ego moved to look after her foster mother for the next three years. Given that Ego's oldest child had died as the result of a fire, they built a house on the present site which had been that of Ego's biological mother. Since Ego's nephew had inherited the property of her foster parents, she worshipped only the tablets of her biological parents. She paid her respects to her foster parents together with the souls of her husband's parents, on an altar set outside for wandering ghosts, while she prepared an altar to her biological parents inside the house. Had she inherited any property at all from her foster parents, she would have made a tablet for them. Meanwhile, the nephew who inherited the property of Ego's foster parents became a Protestant Christian and hence, the foster parents have no one to worship their tablets.
Example 7. Ego's wife's father's tablet had been deposited in Ego's wife's natal house (see Figure 5). Members of this natal house asked Ego to take this tablet with her to her present house, since she already had enough children. Given that Ego was a member of the Catholic church, she was reluctant to do this. But her sisters had few children and were poor so she felt obligated to bring the tablet with her to her new house. Her children were all baptised into the Catholic church. She confessed to her Catholic priest that she had kept the tablet after building her new house. The priest gave her a dispensation and allowed her to keep the tablet, so that Ego now keeps this tablet in her house near a cross that has been blessed by the priest. She was also told to rewrite the characters on the tablet, changing the last few characters which read shenwei (the seat of the god, a traditional designation) to lingwei (the seat of the soul, apparently more acceptable to the Catholic church). Ego has not yet made this change. This example demonstrates that the Catholic church takes an pragmatic attitude toward indigenous ancestral worship activities and that the tablets are often taken over by those who are members of that particular church. The important factor here seems to be that Ego had many children. Having many children is the symbol of prosperity and they would also be able to continue worshipping the tablet after her own death.
Example 8. Ego is a woman who married virilocally, moving to the household of her husband (see Figure 6). Her biological parents and siblings had died, which means that she brought the tablets of her parents' ancestors with her. These tablets were installed on a separate altar, apart from the altar that holds the tablets of her husband. It is believed that her husband's ancestors would be angry if they found these outsiders on their altar. An interesting aspect of this case is that the wife's tablets appear to be situated in the center, a location superior to that of her husband's ancestors' altar. This would be inconceivable for Han Chinese.

Analysis of tablet data

Out of the data collected from thirty-eight examples, thirty-three households show the following general features. It is noticeable that most tablets symbolise individual, known ancestors; tablets are not focused on ancestors in a general or abstract sense. They represent the specific and not the aggregate. In the thirty-eight cases there are only four tablets denoting "ancestors of every generation." One of these tablets was brought to Taiwan by a mainland Chinese who married into the Puyuma. This tablet commemorates ancestors of his patriline. Another general tablet was made for the sake of good luck as a result of instructions after divination. This tablet is not related to anyone's death. The remaining two cases of a general tablet were both made by a couple who worship their own ancestors separately.
Further analysis demonstrates that the generational depth of these Puyuma tablets is very shallow. This reflects the recent introduction of the custom into the Puyuma community. In most cases, tablets have only been introduced for one or two generations. Four out of five cases in which there are three generations represented on the tablet involve Ego's own generation. During the 1966-68 period, except for one case (see Figure 4), no Puyuma tablets listed more than four generations (except for the general tablet to "various generations". Judging from these data collected in the 1960s, it is apparent that the custom of ancestor worship in the Chinese style was introduced very recently into the Puyuma community.
In reviewing these tablet data, one discovers that it is difficult to actually trace the true relationship between the worshipper and the worshipped by any conventional principle. The conventional classification of lineality does not help much in trying to analyze the Puyuma cases. Among the Puyuma, descent shows no clear inclination either to matrilineality or patrilineality. For example, matrilateral ancestors appear to be traced more often than others, but this does not reflect the fact that that a Puyuma mode of descent is necessarily matrilineal. Instead, this reflects a general pattern of uxorilocal residence which predominates among the Puyuma. In cases where virilocality is practised, the agnatic ancestors are indicated on the tablets but it is rare to find more than two generations of patrilineal ancestors (including their wives) among the Puyuma. It might be noted here that among Chinese in the immediate neighbourhood, it is common to find many generations of patrilineal ancestors on the tablets.
In order to make sense of these data, I would propose that we introduce a new system of classification. Rather than talking about matrilateral ancestors and patrilineal ancestors, I would suggest that we analyze the Puyuma cases in terms of the individual tablet maker's relationship to his or her house. In this new classification system, we can talk about "house ancestors" of the house of residence versus "natal house ancestors" of household members that have married in. This classification system fits the emic perspectives of the Puyuma people themselves. This distinction is also useful in making clear the differences between the Puyuma case and the neighbouring Chinese cases of ancestor worship. The classification system I am proposing also follows more closely the patterns of inheritance. It is illustrated in Table 2. The house ancestors constitute about two-thirds of the total, and most of these are direct ascendant ancestors. As in the case of the Chinese in the neighbouring region, the names of ancestors of the same generation are written on one tablet. In the Chinese cases, however, Ego's worship might be commemorating only direct ancestors, that is to say only a part of the total ancestors recorded in the tablet.
A more conspicuous difference between the Puyuma and the local Chinese population is that one-third of the Puyuma examples commemorate ancestors of the natal house of in-marrying or adopted members. Examples of this type also exist among the Chinese, but the customs are very different. Among the Chinese, such cases are regarded as special and the tablets are put in inauspicious locations, such as the kitchen or near the roof. They are worshipped, but only the main house ancestors receive formal offerings. In contrast to this Chinese pattern, the Puyuma in-marrying ancestral tablets are put either on the same altar as the house ancestors or on a nearby altar in the same room. This is a general pattern among the Puyuma.
Summarising the data collected from ancestral tablets during the 1966-68 investigation, the ancestors worshipped only represent a few generations in depth (see Table 5). Furthermore, considerable numbers of ancestors of those who were in-marrying or adopted into the house are currently worshipped in the house where they resided after marriage or adoption. Even though the eldest daughter is supposed to be the main keeper of the tablets, the order of birth and the sex of siblings are not determining factors. Tablet custodianship is also affected by factors such as economic independence, divination of ancestral views, and conversion to Christianity. Inheritance of property is also a factor but it is not the major factor as it is among neighbouring Chinese. Most of these features, in fact, are in direct contrast to patterns that occur among neighbouring Chinese. We may conclude from these data that the traditional Puyuma pattern of ambilineal affiliation which is evidenced in the traditional karomahan group (see Suenari, 1970), is also reflected in the grouping and custodianship of Puyuma ancestral tablets.

Changes in ancestor worship from 1969 to 1983

Fifteen years after the above data were collected and analyzed, I returned to Taiwan to investigate changes in Puyuma ancestral tablet worship. Following are some of the conclusions of changes that occurred during that fifteen year period. One obvious change was the establishment of new tablet altars among families who had previously resisted adopting this pattern of ancestor worship, which they consider to be Chinese (see Table 6).

Increase in tablets among the Puyuma

Another obvious change that occurred in the fifteen years between the two investigations was an increase in the number of tablets among the Puyuma. This occurred by adding new names to previously established tablets and also by adding new tablets alongside the older ones. Examples can be shown for the following cases (see Table 7).
Example 9. Ego previously worshipped her mother's tablet (see Figure 7), and she made a new tablet for her husband and placed it to the left of her mother's tablet. The box and worship items for her husband are smaller than those of her direct ancestor, but they are on the same shelf and no partition is made to separate them on the altar. Ego maintained that her husband's tablet would be disposed of after his name had been inscribed on the larger tablet of her ancestors. This will occur, Ego says, in the future, after her own death.
This pattern of temporarily establishing ancestor tablets for in-marrying spouses has occurred in six out of seven households that now have more than two tablets. There were also three cases in which no new tablet was made for the dead in-marrying member whose names were put directly on previously existing tablets.

Decrease in tablets of the ancestors of in-marrying members

Another conspicuous change is a decrease in the number of tablets dedicated to the ancestors of in-marrying members (from twelve out of thirty-eight in 1966 to five out of forty-four in 1983). One might conclude from these data that the idea of "ancestor" has become more unilateral among the Puyuma (not necessarily patrilateral, however, because the system does not focus on the sex of the parents but the house in which they live. So, the term "houselateral" might be more pertinent to denote this phenomenon). The emerging Puyuma pattern may appear to be "Chinese" but something rather more complex is happening. I will return to this point in the conclusion.

Sloughing off old ancestors

One pattern that needs to be discussed is a general tendency for the names of older ancestors to be generally forgotten and sloughed off as there has been a change in generation.
Example 10. As outlined in Figure 8, names can disappear when new tablets are created: the person referred to E1 in this Figure was the household head in 1968. After the death of E1, a new tablet was made in 1977 and the names of E1's parent were sloughed off. This omission only becomes apparent when one compares the old tablet (which was still kept in the storeroom) with the new tablet. The villagers were not particularly conscious of this phenomenon.

Division of tablets

It is obviously an interesting issue to determine how tablets are distributed after the division of the household. It is sometimes the case that a tablet can be copied and distributed to another household. Among the Chinese, there is a hierarchical relationship between such tablets, with the duplicated tablet being inferior to the original. This question is addressed in Figure 9.
Example 11. The parents of two sisters E1 and E2 died when the girls were infants. They were raised by their mother's elder sister A. In her lifetime their aunt A was not on good terms with E1. Also after her death, A told E2 through divination to make a tablet for her and said that she would thereafter "eat" in E2's house where E2 had moved after dividing from her own natal house. Her sister E1 still keeps the original tablet and worships both their parents and their biological mother's father.
Table 4 summarises the content of ancestors in the tablets as they existed in 1983. This should be compared to Table 2 which highlights the data as it existed in 1968. Comparing the former with the latter we soon recognise the fact that ancestors on the natal side have radically decreased in number. Also, the pattern of keeping more than two tablets has grown in popularity. The figures indicated in parentheses are highlighted to demonstrate the rise in the number of new tablets over the intervening fifteen years.
Table 5 summarises the dates of the cases so as to make clearer the trend over time. This table shows the tendency for an increase in tablet-making over the fifteen year period between investigations.

Conclusions

In analyzing the data presented above we must begin by considering changes that occurred in the fifteen years between the investigations. Many Puyuma establish a tablet and an altar on the death of a family member or on the construction of a new house. The construction of a new house is an occasion for many who originally resisted establishing altars to do so. There are also some Catholics and Protestants who resisted making tablets. These people have either died or have begun to de-emphasise their Christian connections.
As an incentive for making tablets, we must consider ideas about respect toward the ancestors. Many Puyuma believe that ancestors depend on their own descendants for their welfare in the afterlife as evidenced by traditional illnesses known as mutuha (by which ancestors tell the patient what they want their offspring to do before they cure the affliction). It is not, therefore, difficult for Puyuma to understand the reverence toward the ancestors that is reflected in the Chinese ancestral cult. In fact, a shaman (tumaramao) may sometimes explain to Puyuma how the gods in the Chinese system correspond to Puyuma gods. They do the same for ancestors.
With the increase in the contact among Puyuma and neighbouring Chinese, there has been an increasing tendency to adopt Chinese style ancestor tablets. This has been done among most people, except for those who are zealous Protestants. It would appear that, for most Puyuma, the ancestors are of more immediate interest and importance than Chinese gods.
Even though the keeping of more than one ancestor tablet has increased, most of these cases are for temporary tablets of in-marrying members. In contrast to the Chinese case in which a woman is accepted into the ancestor cult of her husband's group, among the Puyuma the in-marrying household member is often treated differently.
On the other hand, data over the fifteen year period demonstrate that in-marrying household members are participating more and more in the ancestral cult of the family into which they married instead of returning to their natal house. We may interpret this as a change from the traditional ambilineal tendency among the Puyuma which is being transformed into a unilateral tendency under the influence of Chinese culture. But, it should not be interpreted as a change to a patrilineal system. Increase in patrilateral succession reflects simply the increase of virilocal marriage rather than outcome of patrilineal system. The change which occurred is the stress on the house side, i.e. "houselaterality".
The data also suggest that there is a tendency to keep the number of generations constant which means that there is a gradual sloughing off of upper generation ancestors. There seem to be few efforts to write down the names of older ancestors on tablets. If this tendency is general, the Puyuma concept of the ancestor would appear to be rather different than that held by neighbouring Chinese. In the Chinese context there is cumulative knowledge about ancestors that is important and their names must be written and preserved. Among the Puyuma one can rarely find the founding ancestor of a family or a kinship group. This contrasts with the case of the Ami. They can remember the founding ancestor of the group and make a genealogical chart, even though this is rather a new phenomenon. The affiliation of the descent group of the Ami is decided by the residence of the parent at their marriage. The preponderance of uxorilocal marriages causes their descent group to have a matrilineal appearance (Suenari, 1983c).
Reflecting such a radical difference in conceptions of the ancestors, we also note that there is less segmentation of tablets among the Puyuma than is often to be found among the Chinese. There were only two cases of reproduction of tablets, and in these cases, the tablets were copied only because the ancestors complained and would have inflicted illness had the duplication not been carried out. These two cases, therefore, cannot really be considered instances of tablet segmentation following the model common among neighbouring Chinese.
Another important factor to consider in evaluating the Puyuma ancestor worship cult is the rise in relative affluence over the fifteen year period. Puyuma now have access to a cash income by working in cities and in the Near East, or in deep sea fishing. In 1968 they were a relatively impoverished community. By 1983 there had been a boom in the economy and their bamboo houses had been replaced by slab concrete houses. Younger villagers had become migrant labourers and therefore had access to cash incomes and those who stayed at home were able to work in nearby factories. The introduction of this new money has helped to promote the introduction of Chinese lifestyles of a modern type. Gradually, sinicisation has become an accepted mode of life.
One must also consider how the systems of descent and kinship among the Puyuma are affected by the process of sinicisation. The ambilineal descent group appears to have been weakened through the disappearance of ritual houses (karomahan) among the Puyuma. The bilateral kindred is still functionally important.[6] It is nonetheless true that there have been major changes in the factors relating to kinship organisation: a change in residence patterns after marriage, with a preponderance of virilocal marriage, a change in inheritance customs with sons now inheriting property while daughters are given a dowry, plus an increase in Chinese-style rituals of all types.
G.P. Murdock (1949) once proposed that a change in residence patterns after marriage would eventually lead to a change in the descent system. The analysis of ancestor tablets in this paper demonstrates, however, that such changes are not always evident. In my own view we can expect to see a change in the kindred system rather than transformations in the Puyuma descent system. It seems unlikely that the Puyuma will develop a Chinese style descent system without any evidence of a unilateral ideology.
At least two steps would be necessary for the Puyuma kindred to change from bilateral to unilateral.[7] The first step would be the emergence of a distinction within the kindred between the relatives within the house of residence, and those of the natal houses of people marrying in. The change from bilateral to ambilateral is reflected in the increase in the number of tablets of "house" ancestors and a decline in the number of tablets dedicated to ancestors of the in-marrying members. The second step would be for a qualitative distinction to be made between patrilateral and matrilateral ties. In order for this kind of transformation to occur, the Puyuma would have to be subjected to strong, external ideological influences as was the case in the transformation of the Korean kinship system. Another possibility would be the introduction of new, externally enforced regulations regarding inheritance. Given that such external influences are absent from the current Puyuma environment, one cannot expect to find a rapid change in the Chinese-style patrilineal descent system. As long as the Puyuma can keep their community intact, the changes that are occurring will be much more complex processes.
In this paper, I have attempted to demonstrate the gradual and uneven process of sinicisation that has occurred among one group of Taiwanese aboriginal peoples. The Puyuma have over the past few decades become more and more "Chinese" in their ritual life. But this case also demonstrates that sinicisation is not a uniform or uncomplicated process. The Puyuma data indicate that what might appear to casual observers to be "Chinese" cultural evidence, is in fact something rather different. The Puyuma have not accepted outright the Chinese pattern of ancestor worship. They have established ancestor tablets but they have done so in the context of their own changing culture.
The conclusions of this study demonstrate that one must pay close attention to the internal characteristics of people under study when one considers processes of sinicisation. The Puyuma, of course, are only one of hundreds of examples of the process of sinicisation that has occurred in the Chinese cultural universe over the past 1,000 years. One assumes that the process was somewhat similar in cases that have occurred in mainland China as well as on Taiwan. More work of this type, focusing specifically on ritual activities, needs to be done before such general conclusions can be drawn.
The Puyuma constitute an interesting case because they retain important elements of their own traditional culture while at the same time behaving and performing rituals as if they were assimilated Chinese. Can one characterise them as "Chinese"? This, of course, depends upon one's perspective. Some people who consider themselves Chinese would accept them as Chinese, whereas others who consider themselves traditionally Puyuma would accept the behaviour as stemming from traditional Puyuma culture. Ethnic identity, in this case, is in the eyes of the beholder.

Notes

1. The data presented in this paper were collected during fieldwork from December 1966 to June 1968 and during a short survey in 1983. An earlier version was published in Japanese (Suenari, 1983b). I would like to express my gratitude to the villagers of Rikavon who were so kind and generous during my research. I also like to thank Dr. James Watson for his useful comments and assistance in rewriting this paper in English.
2. For a general description of the Puyuma, see Sung (1964), which was the first ethnological report by a Chinese scholar on Taiwan. Chiao (1973) points out that certain elements of the Chinese funeral can be followed separately without performing the whole ceremony, whereas in foreign ceremonies a choice of elements is not permitted. This gives native Puyuma culture more compatibility with the Chinese than with other foreign cultures concerned. This conclusion agrees with my own (Suenari, 1983a, 1983b) regarding Puyuma curing ritual and their adoption of ancestral tablets. On kinship Mabuchi (1960) is still the best concise description. On karomahan see Mabuchi (1960, 1976), Suenari (1970), and Chiao (1972). Kasahara (1980) and Takoshima (1989-90) also write on rituals among the Puyuma.
3. The Puyuma constitute the second sinicised group of the nine so-called "raw" aboriginal peoples of Taiwan. This has been evident since the early period of Japanese control. Kono (1915: 34) notes that there are many Chinese-style houses, utensils and tableware in use among upper class Puyuma.
4. The missionisation of the Rikavon was begun in 1953 by the Protestants, and two years later the Catholics reached the height of their influence (see Table 2).
5. Shih (1975:126) reports that 53 (37.58%) of the total of 141 households in another Puyuma village had established ancestral tablets.
6. Chiao (1989: 132-133) describes this as a change from an ambilineal type to a bilateral one. Even though he and I deal with the same phenomenon, my view differs from his in that I argue that the bilateral type did not simply replace the ambilineal one. In my view, the ambilineal system has simply disappeared, while the bilateral has been in existence for a very long time.
7. I discussed the unilateral kindred at the symposium on "Folk Documents and South China Studies" held at the Hong Kong University of Science and Technology on January 6, 1993, and at the Seventeenth International Symposium on "Korean Studies in East Asian Anthropology", held in Osaka on September 1, 1993. These papers are now being edited for publication.