Bonnie L. Cook, "Ward H. Goodenough, 94, Penn professor", Philadelphia Inquirer 6/15/2013:
Ward H. Goodenough, 94, a longtime University of Pennsylvania professor whose work helped shape anthropology, died Sunday, June 9, of organ failure at the Quadrangle in Haverford. [...]
Born in Cambridge, Mass., he lived in England and Germany as a child while his father studied at the University of Oxford. He became fluent in German by age 4, and his fascination with languages never dimmed.
After the family moved to Connecticut, he graduated from the Groton School in Massachusetts and went on to earn a bachelor's degree in 1940 from Cornell University, majoring in Scandinavian languages and literature.
Although he enrolled in graduate school at Yale University, his studies were interrupted by World War II. He served in the Army as a noncommissioned officer from November 1941 to December 1945.
During the last years of the war, he was assigned to a social science research unit to study certain initiatives. The unit posited that integration of the armed forces was feasible and desirable, and that the GI Bill would meet the needs of returning soldiers and stabilize civilian society.
Dr. Goodenough earned his doctorate in anthropology from Yale in 1949. He was influenced by George Peter Murdock, his mentor while the two did a survey in 1940, and then field work on the Chuuk islands (then known as Truk) in Micronesia for seven months in 1947.
He maintained a lifelong attachment to Chuuk and its people, and was the author and compiler in 1980 of the Trukese-English Dictionary.
Dr. Goodenough did field work later in Oceania, both in Micronesia and Melanesia. An expert on kinship, his best-known early contribution was the development of a method for applying componential analysis to the study of kinship terminology.
Ward Goodenough retired from Penn before I arrived there, so I'll substitute quotations from his writings for personal reminiscences. He flourished in the days when anthropology and linguistics were much more closely in tune than they are now, and even published in Language — "Componential Analysis and the Study of Meaning", Language 1956:
That the methods of componential analysis as they have been developed for analyzing linguistic forms are applicable in principle for analyzing other types of cultural forms is a proposition toward whose demonstration I have for some time sought to orient my ethnographic researches. The results of some exploratory work toward this end have already been published.l Included among them is an analysis of Truk kinship terminology, in which it proved possible to apply some of the principles of linguistic analysis to the problem of deriving the significata2 of kinship terms and of determining which terms went together in what I called semantic systems. I am taking up this material again in order to present a fuller discussion of the method and of its implications for developing an empirical science of meaning.
This is the sort of thing that the post-structuralists were "post", but it also reminds me of classical A.I. in some ways:
The problem of determining what a linguistic form signifies is very well illustrated by kinship terms. In essence it is this: what do I have to know about A and B in order to say that A is B's cousin? Clearly, people have certain criteria in mind by which they make the judgment that A is or is not B's cousin. What the expression his cousin signifies is the particular set of criteria by which this judgment is made.
This is analogous to the problem of determining what are the acoustical criteria which differentiate sick from thick so that we hear them as different linguistic forms instead of one form, as might a native speaker of German. In this case the criteria are a set of acoustical percepts which in varying combinations make up the phonemes of a language.
He applied his ideas in familiar as well as exotic contexts — "Yankee Kinship Terminology: A Problem in Componential Analysis", American Anthropologist 1965:
The kinship terminology analyzed here is the one with which I grew up and with which I continue to operate. It is not shared by all North Americans or by all native speakers of English. I have encountered many fellow Americans who reckon degrees of cousinship differently and others who confine cousins to ego's generation entirely, regarding all collaterals in their parents' generation as their aunts and uncles rather than as their cousins. There are evidently significant subcultural differences not only in the selection of linguistic forms to designate the same relationships — as with the preference in much of the southern United States for daddy over father as a standard term of reference — but in the classes of kin types that a given term may designate.
Prof. Goodenough didn't specify exactly what he meant by "Yankee" — see "It's Yankees all the way down", 12/9/2003, for some of the options. It would be interesting to see a socio-geographical survey of the "classes of kin type" as well as the "terms of reference" used for them.
Considering only the distribution of kin terms with respect to consanguineal kin types, I arrived at a "solution" of Yankee terminology similar to that published by Wallace (1962). It made a basic three-way cut of the universe of kin types in terms of degree of collaterality, with a "lineal" set, a "first-degree collateral" set (including the designata of aunt, uncle, nephew, niece, brother, sister), and a "second or further-degree collateral" set (consisting of the designatum of cousin). The lineal and first-degree collateral sets were further partitioned by considerations of generation removal, generation seniority, and sex of alter in the relationship. The model perfectly predicted the consanguineal kin types that were permissible denotata of the terms, but it bothered me. It didn't feel right to have the major taxonomic level in the system (as represented by degree of collaterality) separating the terms brother and sister from father, mother, son, and daughter, with which I felt they somehow belonged as a distinct subset of terms. Grouping them with aunt, uncle, nephew and niece just didn't sit right.
An analysis that does violence to the informant's subjective feel for the appropriateness of things is presumably suspect, especially if a satisfactory alternative analysis is available with which the informant is more comfortable. That there was objective structural reason for my discomfort in this case became apparent when I undertook an analysis that included affinal as well as consanguineal kin types. The terms brother and sister had to be considered as part of a subset with the terms father, mother, son, and daughter because, like them, they were subject to transforming operations signalled by the affixes step- and -in-law, and at the same time no other terms were subject to these transforming operations. This observation made it immediately apparent that Yankee terminology consists of a set of basic terms and of another set of expressions that are derivatives of the basic terms. The basic terms fall into distinctive subsets according to the kinds of derivatives that can be built on them. By insisting that analysis preserve the integrity of these subsets, I arrived at the componential models presented here. What preserving the integrity of subsets means operationally can be demonstrated readily in the course of analysis when we consider the structure of componential matrices.
In a later work, Prof. Goodenough emphasized variation in "the understandings about things and the expectations of one another that the members of a society seem to share", though it seems that systematic study of such variation has remained the province of sociologists – "Multiculturalism as the normal human experience", Council on Anthropology and Education Quarterly, 1976:
Anthropologists traditionally have acted on the assumption that most societies are not multi-cultural, that for each society there is one culture. They have seen multi-cultural societies as developing only in the wake of urbanism, economic specialization, social stratification, and conquest states.
The view of culture that characterizes societies or sub-societies as wholes is appropriate to problems that involve comparing societies as organized human systems, or that call for the classification of societies according to one or another taxonomic scheme. For these purposes, minor cultural differences from household to household (such as reported for the Navajo by Roberts, 1951) or even from village to village can often be conveniently ignored. But such a macro-view of culture, if I may call it that, is inappropriate for the theory of culture, for any theory of something necessarily considers the processes of which that something is a product and that accounts for the way it changes over time. If by culture we have reference to the understandings about things and the expectations of one another that the members of a society seem to share, then a theory of culture requires us to consider the processes by which the individual members arrive at such sharing. In this regard, the differences among individuals, their misunderstandings, the different ways of doing things family to family and village to village, all become noteworthy.