essayy

variance in American kinship:

implications for cultural analysis

SY LVlA JUNK0 YANAGISAKO-Stanford University

If there is a single key or core issue that encapsulates both the substantive and
theoretical issues raised by David Schneider’s (1968) cultural analysis of American
kinship, it is perhaps Schneider’s conclusion that, at least at one level of analysis, all
Americans share the same kinship system.’ While granting a wide range of variation a t the
behavioral and normative levels of American kinship, Schneider (1968:112) claims that a t
the cultural (symbolic) level the system of distinctive features that define the person as a
relative constitutes a “firm, fixed core” around which variation occurs. The distinctive
features that define the cultural universe of relatives for all Americans are relationship as
natural substance (blood) and relationship as code for conduct. These features are in turn
elements from two major cultural orders: the order of nature and the order of law
(Schneider 1968:27). Out of the conjunction of these distinctive features are constructed
three salient classes of relatives: relatives in nature, relatives in law, and relatives by blood
(Schneider 1968:28).

Not surprisingly, Schneider’s claim for a single system of American kinship has been
received with considerable skepticism. American anthropologists, I suspect like natives
everywhere, find it irritating-to say the least-to have someone t e l l them how they think.
Furthermore, given the current academic climate, in which it appears to be universally
accepted that words are polysemic, that symbols have multiple referents, that cognitive
diversity not only lives but may be a societal prerequisite (Wallace 1970)’ and that the
omniscient informant i s either a misguided fool or a foolish misconception of the
anthropologist (Gardner 1976), an analysis that dares to posit for Americans cultural
unanimity, in any domain or a t any level of abstraction, i s bound to elicit emphatic native
pro test.

The question o f variance in American kinship provides a rich basis for a
discussion o f several c r i t i c a l conceptual, theoretical, and
methodological issues embedded in the cultural analysis o f kinship or
any other cultural domain. The paper begins with an examination of
the extent to which David Schneider’s cultural account of American
kinship represents and explicates the symbolic system of
secondgeneration Japanese-Americans. This leads to the major
theoretical problem: the way in which we formulate heuristic levels o f
analysis and construe their interrelationships, An explication of the
theoretical consequences o f Schneider ’s scheme o f the behavioral,
normative, and cultural systems and his articulation o f the “pure” and
“conglomerate” levels o f the cultural system compels us to reassess the
goals o f cultural analysis and suggests the kind o f theory of meaning
and action that will prove most instructive in such an endeavor.

special section: American kinship 15

Aside from the question of whether Schneider’s cultural analysis represents the
symbolic system o f every American anthropologist, there is the question of variance
among salient social categories and groups in American society. As Wallace noted in his
review of American Kinship :

granting the worth o f describing culture as a system o f cognitive structures or symbols and the
necessity o f constructing ideal-type systems for this purpose, there remain certain issues of
sampling by class o f utterances, as well as by class o f speaker, which cannot be begged without
opening the door to a number o f alternative systems, which are equally elegant and equally
valid and which differ either because they are based on a different sampling o f types of people
and types o f statements or because a different frame was chosen, or both. Does this study of
American kinship equally well represent Wasps, Negroes, American Indians, jews, Puerto
Ricans, Poles, etc? . . . Do all American groups maintain this symbol system? And if not, which
ones? One cannot know, and no procedures are indicated in this work for finding out (Wallace
1 9 6 9 : 1 0 6 ) .

I agree with Wallace that the issue o f sampling i s a critical one and that the answer to
the question of diversity among American groups cannot be found in Schneider’s book. I
disagree, however, with Wallace’s statement that there are no procedures indicated in
Schneider’s work for finding out. By combining the definitions, concepts, and the
(admittedly somewhat vague) methodology presented in American Kinship with the
detailed fieldwork procedures outlined in the supplemental monograph of the American
Kinship Project (Wolf n.d.), one can indeed formulate a research strategy for discovering
whether different classes of speakers share, at any level of analysis, the cultural system
described by Schneider.

In this paper, 1 examine the extent to which Schneider’s cultural account of American
kinship can be said to represent and explicate the symbolic system of one particular
section of the American population. The class of speakers whose statements form the
basis of this examination are second-generation Japanese-Americans (Nisei) residing in
Seattle, Washington.’ By employing Schneider’s strategy of differentiating the cultural
system of kinship from the social system o f kinship (the actual patterned interactions
between people), I have extracted the system of symbols and meanings embedded in
informants’ normative statements about kinship relationships and then compared the
native’s own units of meaning, the way in which these units are defined and
differentiated, and how these units form an integrated order or classification (Schneider
1972:38) with Schneider’s account of American kinship.

The purpose of this presentation, however, is not limited to answering the question
“do Japanese-Americans share the cultural system of kinship outlined by Schneider?”;
rather, the point i s to use the question of variance in American kinship as a starting point
for examining several critical conceptual, theoretical, and methodological issues
embedded in this substantive question. As I will show, the question of variance leads us to
a major theoretical problem; namely, the construction of heuristic levels of analysis and
the manner in which we construe their interrelationships. Specifically, we are led to
scrutinize Schneider’s scheme of the behavioral, normative, and cultural systems and his
articulation of the “pure” and “conglomerate” levels of the cultural system. This in turn
compels us to reassess the goals of cultural analysis and guides us toward the kind of
theory of meaning and action that will prove most instructive in such an endeavor.

the cultural system of Japanese-American kinship

In my research, I discovered several topics in which the statements made by
second-generation (Nisei) Japanese-Americans did not fit well with Schneider’s analysis.

16 american ethnologist

For the purpose of this discussion, however, I will focus on only one area of contrast that
initially appeared congruent with Schneider’s model, but which, on closer scrutiny,
demonstrated significant structural variation as well as complexity. This subject, which
lies a t the core of Schneider’s analysis of the cultural system of American kinship,
includes the cultural domain of relatives and the distinctive features that define the
person as a relative.

Following the field procedures outlined in the American Kinship Project monograph
(Wolf n.d.), I asked each informant to “list for me all the people you consider to be
related to you.” 1 then proceeded to elicit from informants their reasons for including or
excluding kin types and specific individuals, and I explored generally the cultural domain
of “relatives.” The data appeared initially to mirror perfectly Schneider’s account of the
domain of relatives. That is, the rule according to which a person is included in the
category of relative i s : “a person i s a relative if he i s related by blood or marriage and
provided he i s closely enough related (or i s not too distant)” (Schneider 1968:62).
Distance here refers to either (or both) the degree to which two persons share biogenetic
substance and the magnitude of the claim on diffuse, enduring solidarity (Schneider
1968:65).

Everything my Nisei informants said indicated complete agreement with Schneider’s
(1968:69) statement that “the decision as to who i s a relative i s made by and about a
person.’’ In other words, there are no rules that set formal degrees of genealogical dis-
tance for inclusion in the category of relatives. Instead, ego decides to include an in-
dividual in this category depending on her assessment of the closeness of the relationship
existing between herself and alter. Thus, ego may include a specific individual from a kin
category but a t the same time exclude other individuals from the same category. Finally,
as Schneider sagaciously notes, the elements of substance and code for conduct are not of
equal value, and

their different values alone and in combination, along with “distance,” account for much of the
variance i n the system at the level o f the person, both as decisions about concrete individuals
and as normative constructs. Substance has the highest value, code for conduct less value, but
the two together (that is, the “blood” relative) have the highest value o f all (Schneider
1968:63).

The Nisei appeared to differ from Schneider’s informants only with regard to their
treatment of a category o f affinal kin that Schneider elsewhere (Cottrell and Schneider
1963) has claimed his middle-class Chicago informants do not generally think of as
“relatives.” This i s the category of ego’s consanguine’s affine’s consanguines, which
Cottrell and Schneider refer to as C.A.C. kin. According to the authors, “informants do
not generally think of C.A.C. kin as relatives or as being related to them in some special
sense of this idea” (Cottrell and Schneider 1963:3). Thus, among Schneider’s informants,
C.A.C. kin (and more distanct affines such as C.A.C.A. kin and its reciprocal, A.C.A.C.
kin) appear only erratically within the meaning of the word “relative.”

In contrast, my Nisei informants commonly listed C.A.C. kin as relatives. A third o f
the informants included in their kin listings one or more C.A.C. kin, the most common
types being ego’s sibling’s spouse’s siblings, ego’s sibling’s spouse’s parents, ego’s child’s
spouseJs parents, and ego’s child’s spouse’s siblings. Several other informants who did not
l i s t C.A.C. kin as “relatives” referred to them as “shirt-tail relatives” or “coat-tail
relatives” and categorized them as people who are not quite relatives, but who are more
than friends, and to whom you have obligations. A few Nisei also included as relatives
people in the C.A.C.A./A.C.A.C. category (for example, sibling’s spouse’s sibling’s spouse)
and the A.C.A.C.A. category (for example, spouse’s sibling’s spouse’s sibling’s spouse).

special section: American kinship 17

For Schneider, however, a difference in the rate of inclusion of a particular
genealogical kin type does not necessarily reflect a difference at t h e cultural level. Indeed,
the variable inclusion of C.A.C. kin by the Nisei, along with the variable inclusion of
other kin types, only supports his thesis that the decision as to who is a relative i s made
by and about a person. The higher rate of inclusion of C.A.C. kin by the Nisei as
compared to Cottrell and Schneider’s informants, therefore’ can be dismissed as a
statistical difference a t the behavioral level, which does not reflect any difference
between the two groups in the definition of cultural units (persons), in the rules for
inclusion of these units in the domain of relatives, and in the symbolic structure
underlying these definitions and rules.

The seemingly easy resolution of this issue was soon undermined, however, when I
moved on to examine another topic within the Nisei domain of kinship, namely, the
exchange of mortuary offerings (koden) in the Japanese-American community. The
Japanese custom o f giving koden at funerals has persisted in most Japanese-American
communities (see also Johnson 1974), although there have been some modifications in
the rules of exchange as they operated in the past in rural Japan (see Nakane 1967;
Embree 1939; Fukutake 1967). For example, in contrast to rural Japan, wherekoden was
given in the form of fixed quantities of rice, in Japanese-American communities koden
has taken the form of relatively fixed quantities of money. As in Japan, however, a l i s t of
all persons who have given koden a t the deceased’s funeral i s kept by the deceased’s
spouse or, if there i s no surviving spouse, by the deceased’s eldest son. The keeper of the
koden list is responsible for reciprocating koden on the death of any contributor or any
member of a contributor’s immediate family (that is, his children, spouse, parents, and
siblings) .

Since the beginning o f t h e Seattle Japanese community, koden reciprocity has
involved friends and acquaintances as well as kin. The perpetuation of reciprocal
exchange relationships established by their parents (the first-generation Issei) is viewed by
the Nisei as a manifestation of the persisting solidary strength of t h e community. Koden
reciprocity i s thus associated with both the cultural domain of the community and the
cultural domain of kinship. The relevance of koden reciprocity to both of these cultural
domains provides a useful example o f the interpenetration of cultural domains-a topic
that will be discussed in the final section of this paper.

Because koden reciprocity i s conceived of as an exchange between family units, it i s
considered inappropriate for the members of the deceased’s immediate family to give
koden upon his or her death. The immediate family of the deceased includes his or her
parents, children, siblings, and spouse. These kin are considered the receivers of the
koden, although only one of them usually receives the money, which i s then used to
defray the cost of the funeral. Disagreement exists among informants as to whether the
deceased‘s grandchildren, parents’ siblings, first cousins, nephews, and nieces are also “too
close” to the deceased to give koden. Several informants said these kin are “too close,”
both genealogically and socially, to the deceased and that monetary contributions from
them are a form o f intrafamilial aid rather than koden. Other informants disagreed.
Regardless of the variance in the determination o f who i s a “close” relative and part o f
t h e deceased’s immediate family, the cultural units involved in koden reciprocity are
“families” rather than “persons.” Not only are the deceased‘s spouse, parents, children,
and siblings considered “too close” to give koden, but so are any of their spouses and
dependent children. Thus, the determination of which relatives are “too close” to
participate in koden exchange is made on the basis of the genealogical and social
connections between “family” units rather than between “persons.”

18 american ethnologist

While “close relatives” do not give koden, “distant relatives” are expected to do so.
Since relatively few Nisei have genealogically distant consanguineal kin in the Seattle
area,’ the category o f “distant relatives” i s filled primarily by affinal kin. These affinal
kin are members o f family units that are connected to the deceased’s family by an affinal
t i e only (no consanguineal tie), and they contain the deceased’s C.A.C., A.C.A.C./
C.A.C.A., and A.C.A.C.A. kin-in other words, just those categories of affinal kin that the
Nisei variably included as “relatives” in their kin listings. Figure 1 depicts the affinally
related family units4 that participate in koden reciprocity.

F G H I

A 8 C D E

– – ,‘ -\
Family to whom ,‘ O = A ‘, Family to whom

1 ,I ego’s family does
ego‘s family \ \
gives koden —/ ’ not give koden

*See footnote 4 for a discussion of the domestic cycle o f family units.

Figure 1. Family units engaged in koden reciprocity.

In the figure, ego (13) i s a member of family C along with her spouse (14). On the
deaths of individuals 11 , 12, 20, 15, 16, and 22, family C does not give koden since it i s
connected to families B and D by a sibling tie. Thus, even at the death of ego’s spouse’s
sibling’s spouse (an A.C.A. kin), ego i s considered too “close” a “relative” to give koden
because she is a member o f a family connected to the deceased’s family by a sibling tie.
Neither does ego give koden on the deaths of individuals 3, 4, 5, and 6, because they are
members of families (G and H) connected to ego’s family by a parent-child tie.

On the deaths of individuals 9, 10, 19, 17, 18, and 23, ego’s family gives koden
because neither families A nor E have a consanguineal tie with ego’s family. They are
instead connected to ego’s family (C) only by an affinal tie. The same i s true of families F
and I.

Families A, E, F, and I contain members of t h e categories of affinal kin (C.A.C.,
A.C.A.C./C.A.C.A., and A.C.A.C.A.) that Cottrell’s and Schneider’s informants did not
consider “relatives,” but that one-third of the Nisei informants included as “relatives.”
None o f the Nisei informants listed more distant affines (for example, A.C.A.C.A.C.),
that is, members o f families o f more than one degree of affinal removal (in Figure 1 the
sibling o f individual 9 would be an A.C.A.C.A.C. kin o f ego).

special section: American kinship 19

Because Nisei informants, in their discussion of koden reciprocity, agreed that
members o f affinally related families are “distant relatives” a t whose death one is
obligated to give koden, I was initially puzzled as to why only a third of them included
such affines as “relatives” in their kin listings. An examination of this seeming
inconsistency in the determination of who i s a “relative” revealed an important feature of
Nisei kinship ideology and behavior. It also explained why C.A.C. kin appear more than
erratically in Nisei kin listings.

The contrast between the vuriubie inclusion of C.A.C. kin (as well as
A.C.A.C./C.A.C.A., and A.C.A.C.A. kin) as “relatives” in Nisei kin listings and the
invariable inclusion of these kin as “distant relatives” to whom koden must be given can
be understood only by explicating the cultural domains to which the Nisei assign these
two separate phenomena. Koden reciprocity i s conceptualized by the Nisei as a
“Japanese” custom rooted in a “Japanese” system o f ascribed moral obligations. Within
this system of ascribed obligations, personal relationships are considered irrelevant. One
has a moral obligation to give koden because of the formal social identity relationships
one has with others, regardless of how one feels about the individuals filling these
positions.

For the Nisei, then, the relevant units in “Japanese” kinship relationships are
“families” and not “persons.” Consequently, they feel obliged to give koden to affinally
connected family units on the occasion o f the death of any member of that family. The
expectations people have of the relationships that should obtain between affinally
connected family units are not restricted to koden reciprocity but operate as well in areas
such as participation in kin gatherings for holidays and ritual celebrations. At wedding
ceremonies and receptions family units that are connected to either the bride’s or the
groom’s families by an affinal t i e (for example, t h e bride’s sister’s husband’s parents and
siblings) are included as “relatives” (Yanagisako 1975a:236). Thus, C.A.C. kin are
“relatives” in the sense that they have definite rights and duties to each other as members
of affinally related families.

In contrast, the act o f constructing a list of one’s “relatives” was interpreted by my
informants as being an “American” phenomenon in which the appropriate units are
“persons” rather than “families” and that entails choice rather than ascription. Only after
I had completed the interviews did it become clear that the procedure of eliciting the kin
listing encouraged this conception. Following the American kinship monograph field
procedure (Wolf n.d.), I asked each informant to “list for me all the people you consider
to be related to you.” According to Schneider, these instructions avoid asking for a l i s t of
“persons” (that is, concrete individuals) and allow the informant to l i s t family groupings,
kin categories, and households, as well as individuals. Since this mode of elicitation was
designed to avoid contamination by implying any specific action-system or context,
Schneider assumes that i t i s contextually neutral.

Yet on closer scrutiny, the request for a l i s t o f people can be seen to encourage the
listing of specific persons as opposed to groupings of people. Moreover, the instruction to
“list for me all the people you consider to be related to you” contains two critical
implications. First, it implies that individuals are the units in the relationship; that is, it
asks for a l i s t of people related to you rather than to your family or household. Second, i t
implies that there is some consideration and, by implication, choice involved in the
matter. Consequently, the Nisei informants perceived the kin listing to be an “American”
procedure involving individuals and choice; one chooses which “persons” are one’s
“relatives.” Within this procedure the greater the degree of genealogical distance, the
greater i s the significance of the “personal” relationship (socioemotional distance) in
determining whether alter i s a “relative” of ego. For A.C., C.A., and A.C.A. kin, the close

20 american ethnologist

consanguineal tie between ego and alter’s families tends to override the nature of the
“personal” relationship, and these categories of kin are generally included in the kin
listing. Members o f affinally related families, such as C.A.C. kin, however, are included
only when the informant feels that she and alter have a “close” personal relationship.

The contrast between the variable inclusion of C.A.C. kin as “relatives” and the
invariable exchange of koden with C.A.C. kin reveals that the Nisei utilize both the
“person” and the “family” as units in the determination o f who i s a relative. Which unit
they employ in any instance depends upon the perceived nature of the context. I f the
context i s perceived to be an “American” one (as in the case of the kin listing) the
relevant units are considered to be “persons.” If the context i s perceived to be a
“Japanese” one (as in the case of koden reciprocity) the relevant units are considered to
be “families.” This usage o f two different units in reckoning kinship relationships i s
neither inconsistent nor contradictory; it reflects a cultural structure that pervades all
levels of Nisei kinship ideology and behavior. This structure can be readily abstracted
from the statements the Nisei make about a wide range of subjects pertaining to the
family and relatives.

For example, the Nisei construct o f marriage and the ideal conjugal relationship can
only be fully understood within the context of their notions of “Japanese” versus
“American” marriage. To the Nisei, the ideal white-American marriage is based on
“romantic love,” that is, on the emotional and sexual attraction of two unique
biopsychological beings (persons). “American” marriage is, accordingly, viewed as having
i t s basis in “feeling.” In comparison, “Japanese” marriage i s viewed as rooted in “duty”
(giri), that is, in one’s social and moral commitment to a contractual relationship. Within
this contractual relationship, how one “feels” about the “person” i s considered irrelevant
to the fulfillment of t h e contracted obligations. In the Nisei view, the parties to a
“Japanese” marriage are not “persons” but social identities (nonpersons). A wife i s loyal
to her husband not because she i s attracted to his unique “personhood,” but because he
stands in the social identity relationship of husband to her.

While the Nisei have rejected “Japanese” marriage as a model for their own conjugal
relationship, they have not adopted without qualification the ideal “American” marriage.
Instead, the ideal conjugal relationship of the Nisei i s a synthesis of these two contrastive
constructs. “Love” and “feeling” have been brought into the conjugal relationship, but
“duty” has not been discarded. The ideal marriage is one that includes both these
elements. Within this ideal construct the emotionality of “romantic love” i s balanced by
the constraint o f moral duty. Thus, one has an enduring commitment (duty) to “love”
(feeling) one’s spouse (the social identity rather than the “person”).

Because the Nisei construct of conjugal love mediates between their notions of
“American” marriage and “Japanese” marriage, any particular statement by a Nisei may,
in isolation from the whole range o f statements made with respect to marriage, give
expression to only one of t h e components. Informants sometimes made statements about
marriage as if it were based solely on “romantic love” and “natural atrraction.” Other
informants a t times stressed “duty” and never mentioned “love.” This alternative
emphasis on only one o f the components did not reflect different ideals of marriage,
however, since in the course o f several interview sessions all informants eventually gave
expression to the other conponent.

The same opposition between “American” and “Japanese” constructs is made by the
Nisei in their discussions o f t h e parent-child relationship. Informants invariably contrast
their relationships with their own children with the parent-child relationship in their natal
families. The Nisei view their relationship with their parents as having been based on strict
discipline and fear and as lacking “feeling,” “love,” and “companionship.” They view

special section: American kinship 21

their relationships with their own children as more “open,” “expressive,” and “loving.”
Considerable emphasis is placed on “communication” within the family and Nisei parents
say they encourage their children to discuss personal problems and a wide range of topics
with them. Nisei fathers are particularly proud of their ability to discuss with their sons
controversial or “sensitive” topics such as sex and politics, which they could never have
brought up with their own fathers. Shared family activities are also valued as a means of
developing close emotional ties and family unity. Consistent with the emphasis on
“communication” i s the Nisei parents’ value of compliance based on “reason” rather than
“fear.” Parents should explain their decisions to children rather than rely on fear-evoking
physical force or disciplinary measures.

The Nisei have not, however, totally accepted what they construe as the “American”
parent-child relationship. The “Japanese” way of discipline and respect for the parent i s
also valued. Again, the ideal relationship l i e s somewhere between “American” and
“Japanese” extremes. While espousing the values of “open communication,” informants
admitted to having reservations about “too much freedom” and “lack of discipline.” One
informant expressed the following reservations:

Kids are more relaxed these days, b u t that’s what’s wrong with Japanese families these
days. . . that’s what’s wrong with this assimilation stuff. When we were children, you could
always see a big difference i n the way Japanese children behaved and the way hakujin [ w h i t e ]
children behaved. Now you can’t tell them apart. There have to be some limitations; you should
be able t o decide who’s the parent and who’s the child.

The fundamental contrastive pair emerging in all these areas of kinship i s the
“American” versus the “Japanese” cultural orders. That Japanese-American ideals are
syntheses of the Nisei’s notions of the “American” and “Japanese” orders reveals their
underlying conception o f their social identity. Within this conception, t h e
Japanese-American is a synthesis of “Japanese” and “American” traits, tendencies, values,
contexts, and behavioral patterns. Being a Japanese-American means integrating the
contrastive elements of these two cultural orders within oneself. An overemphasis of
components from either order to the detriment of the other threatens one’s identity as a
Japanese-American. If too much precedence i s given to the “American” components one
ends up being indistinguishable from a hukujin (white). If the opposite components are
regnant, a person i s too “japanesey” and “old-fashioned.”

An individual’s identity as a Japanese-American i s further affirmed by his ability to
alternate between these contrastive orders. Depending on the prevailing context, elements
from either one of the contrastive orders may be selected to construct and legitimize
normative ideals and concrete behaviors. An event, action-system, or institution may be
perceived as belonging to the “American” cultural domain, as in the case of the kin
listing. In these contexts, one’s performance should be constructed out of, and will be
evaluated in terms of, the premises, values, and motives of the “American” order. On the
other hand, when one performs in contexts construed as “Japanese,” as in the case of
koden reciprocity, elements drawn from the “Japanese” order are considered appropriate.
Finally, in “Japanese-American” contexts a synthesis of elements from the two orders i s
constructed.

The Nisei construction and recognition of “Japanese-American” contexts,
action-systems, and institutions, along with their conception of “Japanese-American”
identity, reveals that while the “Japanese” and “American” orders are isolable and
contrastive a t the symbolic level, they are not isolated at the level of action or at the level
of normative expectations. Indeed, neither the corpus of statements made by an
aggregation of Nisei nor those made by an individual Nisei can be fully comprehended
until one views them as combinations and blendings of elements from these two symbolic

22 american ethnologist

categories. Even then the task o f identifying and categorizing these symbolic components
i s neither a simple nor mechanical one, because their selection and combination in any
particular setting does not appear to be governed by standardized formulas or procedures.
The ambiguity of meaning that prevails in most concrete situations permits individuals
creatively to manipulate the symbolic structure by advocating a definition of the context
that facilitates and legitimates idiosyncratic goals, strategies, and styles.

That Nisei kinship norms and behaviors are structured a t the cultural level of meaning
by an opposition between the “American” and “Japanese” orders can hardly surprise us.
Having matured in a domestic and community environment supervised by lssei parents
who transmitted to them Japanese conceptions, values, and interactional styles, the Nisei
must have been impressed by the contrast between this system and the European-Ameri-
can system dominating the public school and extracommunity environments. Yet the
Nisei did more than merely observe the differences between these contexts and cultural
systems. Instead they formed, consciously or unconsciously, a unitary conceptual scheme
out of the differences that they perceived to exist between these contexts and cultural
systems.

Interestingly enough, in constructing the contrastive categories o f the “American”
cultural order and the “Japanese” cultural order, the Nisei appear to have committed t h e
same analytical error of which social scientists are so often guilty; that is, in formulating
this dualistic scheme they lifted cultural elements out of context. For example, in the
cultural system of traditional Japanese kinship, both “duty” (giri) and “feeling” (ninjo)
were present as contrastive cultural elements. Yet the overt precedence placed on “duty”
in the Japanese normative system, when contrasted with the relatively greater emphasis
on “feeling” in the American normative system, appears to have.moved the Nisei to select
only one element as representative of each cultural system. Consequently, “duty” came
to represent t h e “Japanese” system and “feeling” the “American” system. This
decontextualization of elements appears to have been a necessary subprocess in t h e
formation o f a dualistic conceptual scheme. In effect, it enabled the Nisei to reduce each
of the former schemes to one of i t s elements, which could then be paired in opposition
with the reduced element from the other scheme. In the process of forming such a
unitary conceptual scheme, the Nisei developed a system of symbols and meanings that
lent order not only to their social identity as Japanese-Americans but to all areas of social
action and all cultural domains, including the domain of kinship.

a t w h a t level variance?

While it must now be readily apparent that some degree of variance exists between
Japanese-American kinship and the cultural system of kinship Schneider has claimed for
all Americans, a more critical question remains; namely, at what level of analysis does this
variance lie? Schneider, it should be recalled, has never argued that variance i s absent in
the American kinship system, but rather that “it i s possible, a t one distinct level o f
cultural analysis, to discuss and describe a single [American] kinship system, and a t
another to define and describe both alternate and variant forms” (Schneider 1968:18).
Thus, while variance in the rate or frequency of occurrence of behaviors, in alternate
norms or alternate forms, in variant forms or norms, and in the outcome of strategy
decisions that individuals make are all admittedly found in American kinship (Schneider
1968:16-17, 112-1 14), variance of a culturd order, that is, in the system of distinctive
features that define the person as a relative, i s claimed to be absent.

Here I must disagree with Schneider, because the difference between the cultural
order of Japanese-American kinship and Schneider’s model of the cultural order o f

special section: American kinship 23

American kinship cannot be handled adequately by this typological scheme of variance.
Although Nisei informants and Schneider’s informants seem to agree, a t least in some
contexts, on the distinctive features that define the person as a relative, it does not
necessarily follow that they attach the same symbolic meanings to either the distinctive
features or the cultural unit o f the “person” itself. For while the “person” i s a salient
cultural component of Japanese-American kinship, the structural position of this unit in
the cultural system of Japanese-American kinship i s quite different from i t s position in
the cultural system described by Schneider. In Schneider’s scheme, the “person” i s the
cultural unit by which and about which decisions pertaining to inclusion or exclusion in
the category of relatives are made and to which other “persons” direct their expectations
and actions depending on their mutual (but not necessarily symmetrical) perceptions of
the socioemotional and genealogical distance obtaining between them. In my cultural
analysis of Japanese-American kinship, however, the “person” i s merely one kind (or
mode) of cultural unit that may be employed in a range of contexts that
Japanese-Americans construe as, or wish to have other people construe as, “American”
domains. The Nisei’s alternate usage of the cultural unit of the “person” and the cultural
unit of the “family” in different culturally defined contexts, and their assignment of
these units to the “American” and “Japanese” cultural orders, therefore, cannot be
dismissed as mere variance in the rate of a behavior or as an alternate or variant norm or
form. Neither can it be viewed as variance in the personcentered system.

I hasten to point out here that I am not claiming that Schneider’s analysis overlooks
the importance o f the “family” as a cultural unit in American kinship. Schneider is quite
aware of t h e significance of the American construct “family” and has presented an
intriguing analysis of i t s symbolic structure. Neither am I arguing that Japanese-Ameri-
cans do not share some, if not all, of the symbols and meanings of the “family” identified
by Schneider. I do, however, disagree with the notion that because t h e Nisei and
Schneider’s informants attach some very similar, if not isomorphic, meanings to the
cultural units “person” and “family,” they necessarily share a single cultural system of
kinship. I base my opposition to this claim for a single cultural system of American
kinship on the additional, but no less important, meanings that the Nisei attach to
cultural units such as the “person” and “family”-meanings that are not shared by all
other Americans. As my analysis o f not only the domain of relatives but also the con-
struct of marriage and the parentchild relationship demonstrates, Japanese-American
kinship cannot be fully comprehended with an explication o f the contrastive “American”
and “Japanese” cultural orders underlying Nisei normative statements and concrete
actions. This i s not to say that the order of nature and the order of law are not also
important symbolic components, nor are they any more fundamental, essential, or pure
components of Japanese-American kinship than are the “American” and “Japanese”
cultural orders.

Japanese-American kinship, therefore, can be subsumed within Schneider’s model of
the cultural system of American kinship, but only if we isolate the definition of relatives
and the rules for the inclusion of persons as relatives from the corpus of other meaningful
acts and statements made by Japanese-American informants. However, such a step i s
hardly a justifiable procedure in cultural analysis. For one, it assigns a priori precedence
to the act of defining and listing one’s relatives, assuming it to be contextually neutral
and, consequently, more central to what American kinship i s about than all the other
meaningful acts and statements made by informants. Yet the decision to include or
exclude persons as one’s relatives has been shown to be a meaningful act that i s as
embedded in (and contaminated by) a culturally defined and specified context as any
other act we might arbitrarily select-including sending Christmas cards, inviting people to

24 american ethnologist

a wedding, or giving koden a t a funeral. Such a procedure also entails the unwarranted
decision to consider only variance in the distinctive features that define the person as a
relative as variance in the “cultural order” or kinship. Why, we may ask, should we be
unwilling to consider variance in the person-centered system as well as variance in the
definitions of household, family, marriage, and divorce as variance in the cultural order of
kinship? Why should the cultural structure of American kinship be equated with the
symbolic significance of the distinctive features that define the person as a relative?’ The
justification for such a reduction of the cultural order of American kinship cannot be
found in either the statements made by American informants or in Schneider’s
conceptual framework. Yet by adopting a synecdochical strategy that equates the totality
of the cultural order with one of i t s many context-bound domains, we limit ourselves to a
straight, but narrow, path of cultural analysis that ignores the contours of the cultural
landscape.

At this point it would appear fruitful to replace the question of whether “it i s possible
a t one distinct level of cultural analysis to discuss and describe a single [American]
kinship systemJJ (Schneider 1968:18), with the question of whether it i s analytically
productive to do so, and for what end?

If our goal i s to explicate native cultural constructs so that we may eventually
understand the manner in which symbolic systems articulate with the actual state of
people’s lives, then it seems neither productive nor desirable to define a cultural system in
terms of a minimally shared set of distinctive features.6 Even if we accept Schneider’s
more immediate and restricted goal of describing culture as an independent system and
analyzing it in i t s own terms, that is, as a coherent system of symbols and meanings
(Schneider 1968:8), synecdoche does not appear to be a very productive analytical
device. If we carve up the corpus of statements Japanese-Americans make about kinship
relationships to extract from it the minimal set of features that it shares with the
statements of all other Americans, we detach and isolate these features from both their
social and their symbolic contexts and thereby discard everything that renders these units
meaningful for Japanese-Americans. By focusing on minimal definitions of the cultural
order and by parsimoniously extracting one unitary symbolic system from t h e statements
of a widely diverse s e t of people, many of whose actions, beliefs, and symbols are
embedded in vastly different social, economic, and historical contexts, we lose the
richness that makes cultural analysis more than a sterile exercise in misplaced
concreteness, much like the genealogically based methods of analysis that Schneider
wisely eschews.

implications for cultural analysis

From the preceding discussion of the problem of variance in American kinship, two
key implications can be derived for the strategies we might develop to pursue fruitfully
the cultural analysis of kinship (or any other cultural domain) and the kind of theory of
culture that will prove most instructive in such endeavors. Both of these implications
pertain to the manner in which we construct heuristic levels of analyses and construe
their interrelationships.

First, we must eschew any a priori premises as to whether the cultural significance
(that is, the symbolic meaning) of any word, statement, or action i s less specified and
more detached from concrete situations than are the normative aspects of such
phenomena. Because Schneider adheres to the Parsonian notion that norms are more
specified, more tied to particular contexts, and not widely shared outside a particular

special section: American kinship 25

social field, while the cultural meaning of a unit is less restricted in scope and context
(Schneider 1976:202), he can admit variance at the normative level without admitting i t
at the cultural level. Yet this premise has nowhere been adequately supported with
cultural data. To the contrary, my analysis of Japanese-American kinship strongly
suggests that the symbolic meanings people attach to cultural units such as the “person,”
“relative,” and “family” may be just as restricted in scope and context and just as tied to
a particular set of social fields as are normative rules for action. For the Nisei, the
symbolic meaning attached to the unit “family” may differ depending on whether the
social context i s perceived to be “American,” “Japanese,” or “Japanese-American.” Thus,
whether we can claim greater generality and detachment from specific cultural contexts
for the symbolic aspects of units requires further examination. To accept uncritically the
notion that the symbolic significance of utterances and actions are less embedded in
particular contexts than the norms from which they are abstracted is to confuse the
degree or height of analytical abstraction with i t s generality.

Second, we must make central to the pursuit of cultural analysis an investigation of
the interpenetration of cultural domains and t h e relationships among the levels of analysis
we identify for heuristic purposes. This strategy i s necessary if we are to avoid the danger
of hastily trapping ourselves in theoretical positions that fit poorly with our general
knowledge o f culture and social action. Thus, for example, we should be wary of the
unexamined theoretical consequences that Schneider’s (1 968) treatment of the behav-
ioral, normative and symbolic systems and his differentiation of the “conglomerate level”
from the “pure cultural level” (Schneider 1969, 1972, 1976; Schneider and Smith 1973)
entail for a theory of meaning and action. According to Schneider the variance we
perceive to exist in American kinship, for example between the lower class and middle
class, lies not at the “pure cultural level,” which i s abstracted from the normative matrix
and consists o f the distinctive features, or “primitive irreducible elements” o f kinship
(Schneider 1972:41), but a t the behavioral level and the “conglomerate level,” which
“consists in units made up o f elements from different pure systems” (197242). So
Schneider situates the differences in American kinship a t the level at which components
from many different “pure” cultural systems (such as the “pure” kinship system, the
“pure” sex-role system, and the “pure” age or generation system) combine to form the
conglomerate systems for action. If we accept this formulation we are left with some
problematic theoretical implications.

If we admit a great degree of variance in the behavioral and conglomerate aspects of
American kinship but insist upon an absence of variance in i t s pure symbolic aspects, we
appear to commit ourselves to an untenable view of the asymmetrical relationship
between meaning and action. Schneider’s scheme allows only a unilateral flow of
information from the “pure” to the “conglomerate” level (and presumably on downward
to concrete action). The combination or mixing of components i s restricted to the
conglomerate level, and the various “pure” symbolic systems remain uncontaminated by
their contribution to the emergence o f this integrated action-oriented system. This
scheme places a serious limitation on the pursuit of cultural analysis. By denying the
possible and, I would argue, likely flow of information from the conglomerate level to the
pure symbolic level, such a scheme hinders our understanding of the interpenetration of
symbolic domains. It does not help us to recognize nor aid us in explaining how it i s that
the cultural structure of Japanese-American kinship i s pervaded at all levels by people’s
conceptions of their social (ethnic) identity. Neither does i t provide us with the
framework for interpreting the relevance of koden reciprocity to both the cultural
domain of kinship and the cultural domain of the community. If we insist on the
imperviousness of “pure” symbolic systems, despite the fact that these “pure” systems

26 american ethnologist

are merely abstractions drawn from a conglomerate level oriented toward action, we
cannot hope to explain or even recognize the cultural processes through which people
construct integrated (through not necessarily perfectly coherent) systems of meaning out
of their designs for action and their actual experience in differentiated settings.

This obfuscation of the interpenetration of cultural domains, in fact, runs counter to
Schneider’s (1 969) perspicacious observation o f the similarities among the symbolic
structures of the domains of kinship, nationality, and religion in American culture.
Indeed, Schneider has gone so far as to suggest that “at the level of the ‘pure’ domain,
religion, nationality and kinship are all the same thing ( ~ u l t u r a l ) ~ ’ (1969:124). Yet the
isolation of a “pure’’ symbolic level that remains unaffected by variance a t the
conglomerate level impedes our understanding of the processes and mechanisms by which
different cultural domains interpenetrate.

The conception of the “purity” of higher-level structures of meaning, therefore,
generates a dilemma by failing to spell out how an enormous mass of people (such as the
universe of all Americans), who have a wide range of social experience and a diverse set of
normative expectations adjusted to these differing experiences, come to assign the same
symbolic significance to social phenomena, whether these are material objects, norms,
actions, symbols, or entire social institutions. The way out of this dilemma i s blocked
unless we adhere to a mechanistic model of learning in which people magically absorb,
through some sponge like acculturative process, available cultural units and their given
meanings. Yet the statements made by Japanese-Americans, and I would suspect those
made by many other salient categories and groups of people (both ethnically and
nonethnically defined) in the United States, clearly demonstrate that people do not adopt
symbolic systems wholesale, but instead construct systems out of their own experience.

We appear then to have no satisfactory alternative but to question the theoretical
underpinnings of a cultural analysis oriented toward the isolation of “pure” symbolic
systems. Its failure to explain how people can combine constructs a t one level yet keep
them distinct and uncontaminated a t another level suggests a confusion o f analytically
differentiated levels of analysis with real compartments in people’s heads. The absence of
explication tends to reify the merely heuristic isolation of symbolic, normative, and
behavioral levels and to obfuscate the productive path of examining the dialectical pro-
cess through which social experience shapes, and i s shaped by, systems of meaning.

notes

’ I am grateful t o Jean-Paul Dumont, Michelle Zimbalist Rosaldo, Renato Rosaldo, and David M.
Schneider for reading and commenting on earlier drafts o f this paper.

’The data for the present discussion o f Japanese-American kinship were collected in 1973 and
1974 through intensive interviews with thirty-one second-generation Japanese Americans (sixteen
males and fifteen females) residing in Seattle, Washington. For a social and demographic profile o f the
informant sample and an assessment of their representativeness o f the Seattle Japanese-American
population see Yanagisako (1 975a:306-313). A description o f the historical background and
contemporary character o f the community, along with analysis o f various aspects o f
Japanese-American kinship, are found in Yanagisako (1975b and 1977).

’Although many o f the lssei (first-generation Japanese immigrants) males in Seattle immigrated
with, or joined, close consanguineal k i n in the United States, this was n o t the case for the lssei women
who came to join husbands or to marry lssei men from their natal prefectures in Japan. Moreover,
since a majority of the relatives of the lssei returned t o Japan during the early settlement period o f the
Seattle Japanese community, the Nisei were usually without aunts, uncles, cousins, or more distant
consanguines in the United States (Yanagisako 1975a:108-109). The absence o f kin in the local
community n o doubt stimulated the development of prefectual clubs ( k e n j i n – h i ) and other voluntary
mutual aid associations i n Japanese-American communities.

special section: American kinship 27

‘The family units shown in Figure 1 are static representations of units whose composition changes
over time with the unfolding of the domestic cycle. For the purposes of koden exchange, children are
included in their parents’ family unit until they marry and establish a separate household. Those who
do not marry and continue to reside with their parents remain members o f their parents’ family unit.
Single adults who reside separately from their parents generally are considered part o f their parents’
family unit, but this varies with the age o f the individual and the extent to which he or she participates
in community activities. In the increasingly rare cases where a married couple resides with the wife’s or
husband’s parents, they are considered one family unit. However, when an elderly couple or widowed
parent moves into the established household o f a married child, they may continue to a c t as two
separate family units in koden exchange.

’Scheffler (1976) has similarly faulted Schneider for treating the question “what i s a relative?” as
though it were one and the same as the question “what is t h e meaning of kinship in American
culture?” (1976:85). According to Scheffler this confusion i s produced by Schneider’s use o f the word
“meaning,“ which

leads him [Schneider] to write o f the distinctive features o f a category as part of the ‘meaning’
o f that category. This leads.. . to a confusion between signification and significance, or be-
tween the meanings o f words and the forms of significance attributed to the objects they
denote and the categories they designate (Scheffler 1.976:85).
‘Gardner (1976:459) has made the same point with regard to studies of cognitive sharing. For an

illuminating exposition o f the limitations o f parsimonious definition o f both native and anthropologi-
cal constructs see Rosaldo 1975.

references cited

Cottrell, Calvert B., and David M. Schneider
1963 Some Aspects o f American Genealogy. Paper presented at the Sixty-second Annual Meeting

o f the American Anthropological Association.
Embree, John

Fukutake, Tadashi

Gardner, Peter

1939 Suye Mura: a Japanese Village. Chicago: University o f Chicago Press.

1967 Japanese Rural Society. R. P. Dore, Trans. Ithaca: Cornell University Press.

1976 Birds, Words, and a Requiem for the Omniscent Informant. American Ethnologist 3:446-
468.

Johnson, Colleen
1974 Gift Giving and Reciprocity Among the Japanese Americans in Honolulu. American Ethnol-

Nakane, Chie
1967 Kinship and Economic Organization in Rural Japan. London School o f Economics Mono-

ogist 1 :295-308.

graph on Social Anthropology No. 32. London: The Athlone Press.
Rosaldo, Renato

1975 Where Precision Lies. In The Interpretation o f Symbolism. Roy Willis, Ed. London: Malaby
Press. pp. 1-22.

Scheffler, Harold W.
1976 The “Meaning” of Kinship in American Culture: Another View. In Meaning in Anthropol-

ogy. Keith H. Basso and Henry A. Selby, Eds. Albuquerque: University of New Mexico Press.
pp. 57-91.

Schneider, David M.
1968 American Kinship: A Cultural Account. Englewood Cliffs, N J : Prentice-Hall.
1969 Kinship, Nationality and Religion in American Culture: Toward a Definition of Kinship. In

Forms of Symbolic Action. V. Turner, Ed. Proceedings o f the 1969 Annual Spring Meeting of
the American Ethnological Society. pp. 116-25.

1972 What i s Kinship All About: I n Kinship Studies in the Morgan Centennial Year. Priscilla
Reining, Ed. Washington, D.C.: The Anthropological Society o f Washington. pp. 32-63.

1976 Notes Toward a Theory o f Culture. In Meaning in Anthropology. Keith H. Basso and Henry
A. Selby, Eds. Albuquerque: University o f New Mexico Press. pp. 197-220.

Schneider, David M., and Raymond T. Smith
1973 Class Differences and Sex Roles i n American Kinship and Family Structure. Englewood

Kinship and Family Structure. Englewood Cliffs, NJ: Prentice-Hall.
Wallace, Anthony F. C.

1969 Review of American Kinship: A Cultural Account. American Anthropologist 71:lOO-106.
1970 Culture and Personality. Second edition. New York: Random House.

Wolf, Linda
n.d. Anthropological Interviewing i n Chicago. American Kinship Project Monograph #l. Chicago:

University o f Chicago. Mimeo.

28 american ethnologist

Yanagisako, Sylvia j u n k o
1975a Social and Cultural Change in Japanese-American Kinship. Unpublished Ph.D. dissertation.

1975b Two Processes of Change in Japanese-American Kinship. Journal of Anthropological

1977 Women-Centered Kin Networks i n Urban Bilateral Kinship. American Ethnologist 4:207-

Department of Anthropology, University of Washington.

Research 3 1 : 196-224.

226.

Date of Submission: May 10, 1977
Date of Acceptance: June 6,1977

special section: American kinship 29