This work comprises eight interconnected essays which address the main arguments, findings and implications of my research on Nuaulu animal classification conducted over a period of 22 years. Included also are 13 appendices which systematically list most Nuaulu animal terms in order of their phylogenetic glosses. The exigencies of academic publishing have prevented a more detailed treatment of each category and their immediate interconnections here, though this can be found in a companion volume, Nuaulu ethnozoology: a systematic inventory, produced by the Centre for Social Anthropology and Computing at the University of Kent at Canterbury. Wherever statements require further ethnographic expansion or elucidation in the text which follows, reference should be made to this volume. I have not generally cross-referenced between the two works since the possible ways in which the companion volume might be used as a back-up are potentially so numerous.

Some of the material included in this volume has previously appeared elsewhere, though it is generally used here in a revised and modified form.

Introduction

Over more than a decade analyses of the various connections between cognition and collective representations, mind and culture, and between ‘mundane’ and ‘symbolic’ classifications, have all received a certain degree of prominence in the professional anthropological literature. In contrast to some writers [e.g. Bloch, 1977; Bloch, 1985], my own view is that the interrelationships between these apparent opposites, as evident in particular substantive cases, are often far from clear [Ellen, 1979a; c.f. Harris and Heelas, 1979]. In large part, the present book is an attempt to vindicate this view in relation to a specific body of data. Having said as much, it is clear that some confusion has arisen from a failure to distinguish clearly instruments (means or agents) of cognitive process from the medium of belief and cultural representation. A generation of anthropologists, most distinguished of which are perhaps Claude Lévi-Strauss, Edmund Leach and Mary Douglas, have tended to conflate cognition with collective representations, but as Bloch [Bloch, 1985: 30] has insisted, we cannot treat cognition as some arbitrarily-imposed scheme. The kinds of cognitive process which I have outlined are apparent in the social construction of categories across the complete range of human experience, ‘mundane’ no less than ‘symbolic’.

Introduction

Language is never static and much less so when it is unwritten. I wish to argue in this chapter that much work on classification has assumed – at least for the purposes of description – the reverse to be the case. In other words, it has ignored historical change. This has been partly because it has been thought irrelevant in a discourse largely underpinned by assumptions of synchronicity, and partly perhaps because those societies for which we have detailed empirical reports are those least likely to provide the kind of evidence required for an examination of such change. All this has reinforced the view of classifications as composed of reified categories, rather than classifying as a rule-bound process ever responsive to new situations. This is particularly curious since it would appear that inferences concerning the direction of evolutionary changes in classification – about which much has been written – rest upon an assumption of inherent flexibility in the arrangements of categories. If we start, explicitly, from a demonstration of the extent of such flexibility then our models and reconstructions are not only more consistent, but also more powerful [Dwyer, 1976b: 442].

Introduction

Despite some determined efforts to eliminate him, must published reports on ethnobiology still list folk categories and classificatory constructs on the basis of the hypothetical omniscient ‘speaker–hearer’. Much the same is true for the ethnographic analysis of any kind of category. This non-existent figure has the distinction of possessing a maximum knowledge of the corpus under examination, the sum total of the knowledges of all persons consulted, without reflecting the practical (and generally partial) knowledge of individuals or indicating the dimensions and degree of variation. Yet if data are assembled using this heuristic device, they will always be in excess of those pragmatic competences common to all individuals. Moreover, they will not reveal unique abilities or those shared by only part of the population. These are matters which may be of considerable sociological significance. So, since there is a large area with an uneven range of abilities, and since no techniques of elicitation are ever completely exhaustive, variation must be expected from any corpus, even if the data provide no means of revealing it.

There are a few early references in the ethnobiological literature to variation between individuals and between groups [Henderson and Harrington, 1914: 8].