關於臺灣高山族的社會人類學研究，作者在本文中所檢討的幾個要點如下：
(1)在民族誌的整理方面，多執著於固定不變的方式，例如親屬稱謂的分類，家庭結構的類型和構成份子之統計，缺乏因族而異的彈性。因此不能掌握各族的主題意識，例如布農族的家庭成員觀念和親屬稱謂在社會制度上的意義。
(2)曾被用於處理高山族親屬分類的Harvey-Liu Notation System,大膽地將婚姻關係“還元”爲繼嗣聯繫的關係，在邏輯上有商榷的餘地。
(3)在親屬理論方面，國內有關臺灣高山族的研究雖有衞惠林的世系論架構，但在概念上頗多謬誤或前後不一致之處。在處理民族誌材料時也偶陷於主觀性。
(4)王崧興等以非單系理論爲基礎批評前述之世系論，但並未解決根本的問題，尤其是大而化之的kith-based group概念和模糊的宗敎和經濟關係概念。
(5)關於羣體構成法則的問題，有許多不同類型的中心主義者(衞惠林的世系中心論和王崧興所代表的constitution觀念），消極論者(如非單系說）和社會全體論者（如中根千枝對雅美族研究的觀點），這些均未能充分反映社會事實。
(6)在討論到社會羣體構成時，對質與量的問題混淆不淸，例如有關系性的討論，泰雅、雅美和阿美可能是程度的問題，而排灣、魯凱和卑南却與系性無關。
(7)向來的髙山族民族誌著作太過於格式化，未能適當地整合社會生活的各個層面，尤其是在社會關係和儀式行爲之間。
(8)過去雖有統合觀念(syntagmatic view)的功能論者討論宗敎信仰和儀式行爲的社會意義。但最豐富的置換觀念（paradigmatic view),或所謂象徵的（metaphoric or symbolic)層面則尙未被發掘。
(9)此種討論架構也可用於批評和期望目前已經式微的物質文化和原始藝術之研究。

In this article, originally read at the Symposium on Taiwan Aborigines (August 24—September 1, 1974, Academia Sinica, Taipei), the author reviews strictly the studies of Formosan Aboriginal social organization done by Chinese and Japanese anthropologists after Second World War. Some essential results are summarized below:
1)In processing ethnographical corpus, a few of Chinese works stuck unnecessarily to stereotypes initiated by former students. Yi-fu Ruey’s classification of kinship terminology, Chi-lu Chen's types of family structure and Shao-hsing Chen’s table of family components were among the most popular. The obsession on these stereotypes may have prevented the later students from penetrating the fundamental problem of each tribe. The Bunun studies are taken as examples to illustrate the situation. In the study of family structure, native’s normative concept of the membership is more relevant than the statistical or normal distribution of different family types, but the former concept received little attention. The Bunun’s Omaha trend of kinship terminology elegantly demonstrated by Toichi Mabuchi also eluded the perception of a specialist’s confusing study on Bunun kinship.
2)The Harvey-Liu Notation System of kin types, proposed .mainly by Pin-hsiung Liu and applied in another monograph on Paiwan tribe, boldly reduced conjugal relationship to parent-child link. Its logical processes are criticized in the article.
3)Hwei-lin Wei’s works on kinship organization of Formosan aborigines are comprehensive, but lack rigidity and consistency in his theoretical considerations. He was criticized for his subjectivism and lineage-centrism in the analyses of ethnographic materials.
4)Wei and his followers’ lineage-centrism had been criticized by Sung-hsing Wang from the standpoint of non-unilineal theory. However, with indefinite concept of kith-based group and elusive terms of religious relationship and economic relationship, he did not solve the fundamental problems of Atayal’s and Yami’s principles of group formation.
5)Some anthropologists of structural-functional school inclined to search a constitution of social organization for a society. However, it seems that not allsocieties must have a supreme constitution of this kind. The classification of any societyi nto a lineage, unilineal, bilineal or non-unilineal type was more often the anthropologist’s bias rather than the native’s reality.
6)Some discussions of group formation confused the concepts of range and substance. For example, difference of lineality among Atayal, Yami and Ami is perhaps of range, not of kind. But the succession among Paiwan，Rukai and Puyuma is irrelevant to sex which is the only criterion of classification of lineality. Anthropologists used to neglect social features other than sex in the studies of kinship organization.
7)Some ethnographic works of Chinese students patterned vapidly after a pre-designed inventory without proper integration. Thus they presented identical framework. Every aspect of social life received equal treatment separately. It seems more similar to a patchwork rather than a holistic view. After reading such works we still know very little about the nature of the society investigated.
8)In the studies of domestic and communal rituals, some authors are inclined to adopt functionalistic view. But lack of considerations of social context has weakened their arguments. The more productive paradigmatic (metaphoric or symbolic) dimension still awaits further development. In this article, I illustrate the point with the Yami case.
9)This framework of review can also be applied to the studies of material culture and arts which prospered several years ago but is now considered passé.