Area of inquiry (group 1): classifying the world practice questions.

Durkheim and Mauss argued in 1903 that the schemes by which societies sort the world — dividing the heavens, the seasons, animals and plants into groups — reproduce the divisions of the societies themselves, so that a people organised in two intermarrying halves classifies the world in two corresponding halves. The most serious objection raised against their argument is that

• societies with no divisions of any kind have since been documented in several regions of the world
• no society has ever been recorded that sorts animals and plants into named groups at all
• the evidence cannot show the direction of the relation, since the two schemes are read off one another
• classification is a purely individual matter and no two members of any society use the same categories

Answer: the evidence cannot show the direction of the relation, since the two schemes are read off one another

The correspondence they document is striking; what the material cannot establish is that the social divisions came first and the cosmological ones followed. The lasting contribution is the demand for a social account of classification, not the specific claim about priority.

Bowker and Star examined classifications built into forms, registers and databases: the international list of causes of death, nursing work categories, and the racial classification imposed under apartheid in South Africa, under which reclassification could be applied for and granted. Their central point about such systems is that

• they record distinctions that were already firmly established before the system came into use
• they are technical instruments whose categories have no consequences beyond the records they organise
• they can be corrected simply by improving the definitions attached to each of the available categories, after which the difficulty largely disappears
• once embedded in procedures they allocate resources and identities, and become hard to see or to contest

Answer: once embedded in procedures they allocate resources and identities, and become hard to see or to contest

A classification written into a form stops looking like a claim and starts looking like the world: it decides what can be counted, who is entitled to what, and which cases fall into the residual box. The apartheid case is the extreme instance, but the argument is about ordinary infrastructure.

Sahlins asked in the 1970s why cattle and pigs are eaten in the United States while dogs and horses are not, given that the flesh of all four is nutritious and that dogs and horses are eaten elsewhere. His explanation was that

• the four animals are ranked by their closeness to persons, and edibility follows that classification
• the prohibition follows from the greater cost of raising horses and dogs for meat in that setting
• the animals not eaten carry a documented risk to health that the others do not carry
• the pattern is an accident of history with no order to it, as the variation between countries shows, which no single explanation could be expected to cover

Answer: the four animals are ranked by their closeness to persons, and edibility follows that classification

Dogs and horses bear personal names and are spoken to; cattle and pigs are counted. The order is confirmed by the vocabulary, since the edible animals are renamed as beef and pork on the table while the inedible ones keep their own names, and by the fact that the same reasoning marks the parts of an animal as edible or not.

Berlin and colleagues compared the naming of plants and animals in a number of unrelated languages, including their own work with Tzeltal speakers in Chiapas. They found that such vocabularies are organised in a small number of levels, that the most frequently used level often corresponds closely to what a biologist would call a species, and that the content of the categories varies from place to place. This combination supports the conclusion that

• folk classifications of living things are approximations that improve as scientific instruction spreads
• the levels at which people classify recur widely, while what is placed in them depends on the local world
• classification of living things is entirely determined by the practical uses to which each species is put
• the classifications of unrelated languages have nothing in common once the vocabulary of each has been recorded in detail

Answer: the levels at which people classify recur widely, while what is placed in them depends on the local world

The finding cuts both ways and that is its interest: there is recurrent structure, which pure relativism did not expect, and locally determined content, which a single universal taxonomy could not accommodate. Practical use turns out to explain some of the naming but not the levels.

Douglas examined in 1966 the animals prohibited as food in Leviticus and noted that those excluded are the ones that fail the scheme of their own domain: creatures of the water that lack fins and scales, land animals that chew the cud but do not have cloven hooves. Her general claim about such prohibitions was that

• what is prohibited is what does not fit the classification, so the rules defend the categories themselves
• the prohibitions record early observations about which animals were most likely to carry disease in a hot climate
• the prohibitions were adopted in order to mark one people off from its neighbours and bear no relation to any scheme of classification
• prohibitions of this kind are arbitrary, so that the search for a pattern behind a list of forbidden foods is misplaced from the start

Answer: what is prohibited is what does not fit the classification, so the rules defend the categories themselves

The reading turns on anomaly rather than on hygiene or danger: the animal is not unclean in itself, it is unclassifiable. Douglas rejected the hygienic reading explicitly, and a later confirmation by germ theory is exactly what her argument denies.

Tambiah studied the animal categories of a village in north-east Thailand in the 1960s. He found that the animals people would eat, would not eat and would not keep near the house lined up with the way the house was divided into spaces of differing seclusion, and with the rules about whom a person might marry. His conclusion was that

• the food rules had been carried over unchanged from religious teaching and stood quite apart from the arrangement of the house and from the rules governing marriage between villagers
• villagers avoided the animals whose meat spoiled soonest in the local climate and gave those avoidances a symbolic dress afterwards
• one scheme of categories was at work across the domains, so that the sorting of animals, of house spaces and of possible spouses could be read off one another
• categories of animals vary so much from village to village that no general account of classification can be given for a region of this size

Answer: one scheme of categories was at work across the domains, so that the sorting of animals, of house spaces and of possible spouses could be read off one another

Tambiah found one gradient of social distance ordering marriageable kin, the spaces of the house and the edibility of animals, so each classification can be read off the others. This extends Douglas's approach from anomaly within one scheme to correspondence between schemes; an explanation from spoilage would have to predict the plan of the house as well.