Merged kinship terms

Why do many societies call a father and his brother by one word while splitting cousins into sharply different kinds?

English speakers find two things strange at once in a great many kinship vocabularies. A man's father and his father's brother are called by a single word. His cousins, all equally close by blood, are split: some called by the same words as his own brothers and sisters, the rest by a wholly separate term.

Most people assume the speakers are being imprecise — that "father" there means something loose, like "father-figure". Suppose instead the vocabulary is exact and English is the loose one. What would a language have to be tracking for those groupings to be the precise ones?

Notice what the merging is not doing. It is not collapsing by generation, since mother's brother stays firmly distinct from father. Nor by closeness of blood, since parallel and cross cousins are equally related and treated oppositely. So a line is being drawn, and it separates father's brother from mother's brother.

Trace the descent lines and it appears at once. Suppose descent runs through fathers, so each person belongs to their father's group. Father's brother is in that group; mother's brother is not, belonging to the group your mother came from. The single word for father and father's brother means senior man of my own descent group in the generation above me; the separate word for mother's brother names a senior man of the group that gave me my mother. Not two flavours of one relationship — opposite social positions.

The cousins then follow without further assumption. Father's brother's children are in your group, so they are called siblings; mother's brother's children are in the wife-giving group, so they get their own term. This collection's walk-through of cross-cousin marriage runs the marriage consequence of the same structure. The fixed point of difference is that this is the lexicon, not the behaviour: how relatives are named need not track whom you marry, and the gap between the two is the subject of the whole modern argument.

Anthropologists label the pattern bifurcate merging. Radcliffe-Brown made the underlying principle explicit in the 1920s as the equivalence of siblings: brothers are socially interchangeable, so their children stand in the same relation to you as you to them. This is not merely verbal. If a man's father dies, his father's brother typically assumes the jural role — authority, guardianship, obligation toward bridewealth, a claim on his labour. Under the levirate, a widow marries her late husband's brother, and the substitution is literal. The word tracks a position a set of men occupy in turn.

Now push it. If terminology tracks descent-group membership, three predictions follow, testable against the ethnographic record.

The first is a mirror. In a matrilineal society descent runs through mothers, so mother and mother's sister should merge and father's sister be the marked outsider, the whole pattern reflected. That is broadly what is found.

The second is stronger. Some systems collapse generations entirely on one side. In the type named after the Omaha, found with patrilineal descent, mother's brother and mother's brother's son take the same term — generation ignored, because the wife-giving lineage is treated as one undifferentiated unit relative to ego. The Crow type mirrors it in matrilineal societies, where father's sister's son may be called "father". A vocabulary ignoring generation to preserve group identity is hard to explain as sloppy genealogy and easy to explain as a naming system for groups.

The third is sharpest, because it is negative. Where there are no corporate unilineal descent groups — where property, residence and obligation attach to households rather than lineages — there should be no merging at all. English is exactly that case: one word "uncle" for both uncles, one word "cousin" for every kind, no term merged with "father" or "brother". The system radiates outward from ego in degrees of distance rather than sorting people into groups, as you would expect where being a father's brother confers no jural office.

The refuting observation, then, is a society with strong corporate patrilineages using a purely distance-based terminology, or a thoroughly bilateral society with full bifurcate merging. Honesty is required: such mismatches exist. The correlation between descent organisation and terminological type is real and strong across large cross-cultural samples, but statistical, not exceptionless.

The commonest misreading is that these people "cannot tell" their father from their uncle. They can, and where it matters they say so — most such languages can specify the actual genitor. The terminology is not a theory of biology; it is a set of address terms for social positions, and classifies by position because that is what it is for.

The second misreading is subtler, and it is the live scholarly question. Morgan, who first assembled these systems in the nineteenth century, and Rivers after him, read terminologies as fossils of past marriage practice, a merged term recording a rule now abandoned. Kroeber attacked that in 1909, arguing that terminologies are linguistic and psychological classifications reflecting how a language cuts up a field — by generation, sex, lineality — and need not encode jural institutions at all. Rodney Needham later pressed the same separation, insisting that a prescriptive terminology is a system of categories and that inferring marriage behaviour from it is an unwarranted leap. So the account above is the structural-functional reading, contested at precisely the joint where terminology meets behaviour.

What survives the dispute is the negative finding, the strongest part of the case: merging patterns are not randomly distributed but cluster with descent organisation far more than chance allows. Whether terminology causes, records, or merely co-varies with it is what remains argued.

• Why kinship terminologies persist for generations after the descent groups that shaped them have gone.

Every term the collection defines is gathered in the glossary.