Introduction

In the 1950s a new way of doing cultural anthropology emerged in the United States, variously called ‘ethnoscience’, ‘ethnolinguistics’, ‘ethnosemantics’, ‘cognitive anthropology’ and ‘the New Ethnography’ (Sturtevant 1964; Colby 1966). It aimed at bettering ethnographic analysis by extending the techniques of structural linguists, like those of Bloomfield (1933) and Sapir (1921), beyond the realm of phonology and grammar to the fundamental ‘codes’, or ‘grammar’, of culture. First employing rigorous elicitation procedures in the controlled questioning of native speakers, the New Ethnographers then sought formally to analyse and evaluate the recorded data for those distinctive mental features that account for human cognition of the world and drive social activity (Conklin 1962; Black 1963).

Although the native's mental categories could not be directly observed, the ethnoscientist would thus infer their existence. Implicit in the technique of using distinctive features to differentiate and assemble atoms of linguistic behaviour was a key assumption of behaviourism (the dominant Anglo-American philosophy of the time, but still close to the original empiricist tenets of Locke and Hume). These techniques would assumedly mimic the way that natives build their conception of the world: using relatively few perceptual criteria of‘similarity’ and ‘difference’, the native supposedly associates those perceptual stimuli that ‘naturally go together’ into percepts; then, by an effort of abstraction, the native likewise forms the concepts that are appropriate for grouping and extending those percepts into thoughts.