Liminality in initiation

Why do initiation rites so often put the initiate through a spell of isolation, hunger or humiliation?

The obvious answer to why initiations hurt is that the pain is a test: only someone who really wants in will bear it. That account is powerful. But it explains the wrong feature.

Look at what the ethnographies actually describe. The initiate is not simply hurt. He is taken away — into the bush, a lodge, a hut at the settlement's edge. His old name may be dropped, his clothes taken, his hair cut. He is fed strangely or not at all, spoken to in a special register, treated as dead or as an infant, forbidden what his old status permitted. And this goes on: days, months, sometimes years. Then he comes back, is renamed, and is treated by everyone as a different kind of person. Cost alone predicts none of that. A screening test could be one hard afternoon. Why a middle?

Arnold van Gennep noticed the shape in 1909 and it has held up remarkably well: rites of passage come in three phases — separation, a margin, and reincorporation. Victor Turner later took the middle phase and made it the interesting one, calling it liminality, from the Latin for threshold.

So ask what work a threshold does. Consider the alternative: a society that simply announces, one morning, that these boys are now men. What goes wrong?

Two things, worth separating. The first is that everyone must update at once. A status like "adult man" is not a fact about a person; it is a set of expectations held by everybody else — who may speak in council, who may marry, who may be sent to fight. If half the village has updated and half has not, the person occupies contradictory positions at once and every interaction must be renegotiated. What is needed is not that people know, but that they know everyone else knows. A long, public, jointly witnessed process produces exactly that, and the length matters because a slow visible transition is hard to miss.

The second is that the old status and the new one are frequently incompatible. A boy may be indulged, is not held responsible, sleeps in his mother's house. A man may not be indulged. There is no coherent way to be halfway between them while living among people who must treat you as one or the other. So the society removes the person from the situation for the duration — and while he is out of it he is not in an intermediate status, he is in no status. That is what stripping the name, clothes and food is doing: not testing him, but blanking the slate so the old attributions have nothing to attach to.

This predicts something specific. If the margin exists to keep an incoherent halfway state out of ordinary life, its intensity should track how incompatible the two statuses are, not how valuable the new one is — and transitions that merely add to a role, such as a promotion within a rank, get a feast rather than a seclusion.

Turner added one observation about the inside of the margin: initiates there are levelled — identically dressed, identically deprived, family and wealth suspended — which he argued produces an intense bond he called communitas. That explains the company one is initiated with rather than the isolation itself, and the two can coexist.

Now the honest part. This is a structural account: it says what the pattern accomplishes, not how it arose or what it means to the person undergoing it. The quantitative claims made for it are shakier than the structure. John Whiting's mid-century cross-cultural correlation between severe male initiation and particular child-rearing arrangements rested on the Human Relations Area Files, whose entries are coded by outsiders from uneven ethnographies and whose units are not independent. The social-psychology parallel — Aronson and Mills on severe initiation raising liking for a group — is a small mid-century experiment of a kind that has replicated unevenly, and I would not build on it.

The ethnographic record is itself selected: rites performed in public, by men, and legible to a visiting outsider are far better documented than those conducted in secret or by women, so the "typical" initiation in the literature is partly an artefact of what a fieldworker could be shown.

Two refinements connect the steps.

First, this does not compete with the collection's costly-signalling account; it answers a different question. Cost explains why the passage is expensive; structure explains why it has a middle, with a beginning and an end. A rite can do both, and most do.

Second, "no status" is not "no rules". The liminal phase is usually the most heavily regulated stretch of a person's life. What is suspended is his position in ordinary social accounting, not the constraints on his behaviour — which is why Turner's initiates can be at once the most powerless people present and treated as dangerous or sacred.

What would refute the account? It predicts a marked marginal phase specifically where a status change is publicly consequential, irreversible, and incompatible with the one it replaces — and its absence where transitions are gradual, reversible, or additive. So the refuting observation is a society with sharp, irreversible, mutually exclusive status changes handled routinely by announcement, or elaborate seclusion attached to a transition nobody's obligations turn on.

• Why funerals show the same three-phase shape, with the corpse and the mourners each occupying their own margin.
• What happens to these structures in societies with written registers, where a status change can be recorded by a clerk.

Every term the collection defines is gathered in the glossary.