Under Goodenough's broad mandate, considerable attention became directed to developing "a kind of ethnography in which the methods of description are public and replicable" (Tyler, 1969, p. 20) as integral to the process of discovering "how cultural knowledge is organized in the mind" (D'Andrade, 1995, p. 248). In this section, three such methodologically oriented approaches used in studying the conceptual organization of the illness domain are described.
An early article by Frake entitled "The Diagnosis of Disease Among the Subanun of Mindanao" showcases elicitation and analytic methods applicable to "a more rigorous search for meanings" (Frake, 1961, p. 113) as a step "toward the formulation of an operationally-explicit methodology for discerning how people construe their world of experience from the way they talk about it" (Frake, 1969, pp. 28-29). During his fieldwork, Frake found that: "Their continual exposure to discussions of sickness facilitates the learning of disease concepts by all Subanun." Characterizing the meaning of a disease concept as the "information necessary to arrive at a specific answer" or diagnosis (Frake, 1961, p. 114), Frake limited the terms elicited to the perceptual realm of skin diseases and the culturally specific (emic) diagnostic criteria which serve to define and differentiate them. Noting the "conceptual exhaustiveness of Subanun classification of natural phenomena" (Frake, 1961, p. 131), Frake portrayed the underlying conceptual organization as taking taxonomic form with different diagnostic criteria operative at different levels of contrast and with increasing specificity at lower levels of the hierarchy. While "informants rarely disagree in their verbal descriptions of what makes one disease different from another," the " 'real world' of disease presents a continuum of symptomatic variation which does not fit neatly into conceptual pigeonholes" (Frake, 1961, p. 130). As a "social activity" that involves negotiating the relevance of culturally shared categories to specific instances, the diagnosis of a particular disease may "evoke considerable debate" (Frake, 1961, pp. 129 & 130-131). Thus, while the "analysis of a culture's terminological systems will not, of course, exhaustively reveal the cognitive world of its members . . . it will certainly tap a central portion of it" (Frake, 1969, p. 30).
Aiming for a more comprehensive understanding of the conceptual domain of illness, subsequent researchers (D'Andrade, Quinn, Nerlove, & Romney, 1972; Young, 1978; Young & Garro, 1994) report unsatisfactory results from efforts to elicit taxonomies or "to carry out standard feature analysis" (D'Andrade, 1995, p. 70). Researchers turned to methods, such as the term-frame substitution task, which allowed for directly comparable responses across informants. Constructing a term-frame interview requires a list of illness terms as well as a set of statements about illness. Each illness term is systematically inserted into each question frame and informants are asked whether the resulting sentence is correct or not (see Weller & Romney, 1988, for further details on this and other systematic data collection techniques). From "Categories of Disease in American-English and Mexican-Spanish" (D'Andrade et al., 1972), an
American-English question frame is "You can catch
_from other people." The number of responses analyzed for each individual can be quite large as this is determined by the product of the number of terms and the number of frames. Analytic techniques for such similarity judgments used in a number of studies include multidimensional scaling (MDS) and cluster analysis (see also D'Andrade, 1976; Garro, 1983). MDS provides a visual representation in which items responded to in similar ways are placed closer together in the scaling plot. Cluster analysis groups items together at increasing degrees of similarity in responses.
In work carried out by Young and Garro in Pichataro, a Tarascan town in the Mexican state of Michoacan, the set of 34 illness terms and 43 frames used in a term-frame interview came primarily from wide-ranging informal interviews about different kinds of illnesses and specific illness episodes, with selected statements recast as yes/no questions (Garro, 1983; Young, 1978; Young & Garro, 1994). Although not intended to be exhaustive, representativeness was a goal, both with regard to the range of illnesses covered and in terms of what matters to community members when they talk about and deal with illness. Rather than being a "shortcut," considerable ethnographic grounding is required to construct a suitable term-frame interview. While the responses can be analyzed in a number of ways, including MDS, that converge on similar patterning (see Garro, 1983), organizing the relatively large number of terms and frames according to the results of a hierarchical cluster analyses facilitated the discovery of similarities and contrasts across this rather large number of terms and frames (Young, 1978; Young & Garro, 1994). Further, a formal analysis of the distribution of frames relative to illness clusters served to differentiate the characteristics linked with particular clusters from those that are more general. Through this procedure the findings are tied to patterns in the data and thus are less dependent on what the researcher sees than when the MDS plot is interpreted. At the broadest level, which clustered illnesses into two main groups, there was an overall distinction between external versus internal locus of cause—illnesses resulting from contact with hazardous environmental agents versus those resulting from internally initiated conditions (related to diet and emotion). Other key distinctions were made on the basis of illness gravity and the life stage of the characteristic victim. All the aforementioned were also among the most important con siderations involved in actions aimed at preventing and alleviating illness. Conspicuously absent from this discussion, however, is the "hot-cold" etiological distinction, even though this was the subject of a number of the question frames. This distinction does not play an important role in the conceptual structuring of the illness categories present in the cluster analysis (see also Weller, 1984b). This was perhaps because, in Pichataro, distinctions based primarily on these etiological principles did not necessarily reflect the types of knowledge most significant for purposive action in relation to illness. In contrast with Frake's efforts to discover the defining features relevant to diagnosis (the distinctive features that define illness terms), the use of the term-frame substitution task in Pichataro as well as in D'Andrade et al. (1972) was oriented around discovering the underlying conceptual distinctions that matter most to people in avoiding and dealing with illness in the context of everyday life.
Using different methods, Weller (1984a) tested the cultural salience of concepts reported in the D'Andrade et al. (1972) study with "samples of urban literate women with children living in the United States and Guatemala." In the earlier research, the concepts of degree of contagion and severity were highlighted for English speakers. For Spanish speakers, important concepts included the relative frequency of occurrence in children versus adults and the appropriateness of hot or cold remedies. Weller (1984a, p. 341) measured "agreement among informants to assess the relative cultural salience" of these illness concepts, starting from the assumption that concepts with the "highest agreement are culturally more salient than those with lower agreement." An initial pile-sort task with well-known illnesses provided similarity judgments used to construct a "'conceptual model' of disease" using MDS. A comparison between the Americans and the Guatemalans carried out by restricting the analysis to the subset of illness terms common to both groups found a high degree of correspondence between the two conceptual models. The data used to assess the saliency of the four concepts mentioned above were separate ranking orderings of the illness on each concept obtained from each of the participants (the Americans did not rank illness on hot-cold). Both agreement among individuals and the fit between the conceptual model and the individual rank-order data for each concept were assessed. For the Americans, the results indicated that all three ranked concepts were important in the cognitive organization of illness. For the Guatemalans, the "conceptual structure" is "best characterized by the concepts of contagion and severity." More variation among informants and a poorer fit between the rankings and the model were reported for the age-related and hot-cold concepts. Indeed, the "variation on the hot-cold dimension is so extreme that it seems to indicate that there may not be a culturally shared definition of that concept" (Weller, 1984a, p. 345; see also Weller, 1983).
Was this article helpful?