India: What Gets Negotiated Inside an Arrangement

India is the case where the law runs ahead of lived practice and behind it at the same time, and where the two facts are not a contradiction waiting to be resolved. They are the shape of the country’s argument with itself.

In the autumn of 2018 the Supreme Court of India delivered three judgments in twenty-two days. On 6 September it read down the colonial provision that criminalised consensual same-sex intimacy. On 27 September it struck down the adultery offence, and said in terms that the provision had rested on the premise that a married woman was her husband’s property. On 28 September it held that a temple’s exclusion of women of menstruating age could not stand against the Constitution. Taken together it was as far-reaching a month as any constitutional court has had on the question of who owns a woman’s body and her attachments.

Five years later the same court declined to recognise same-sex marriage and said the question belonged to Parliament. The exception that removes a husband from the definition of rape is still in the statute book, renumbered into a new criminal code, and the Union government has filed to keep it. The law did not travel in one direction and meet resistance. It travelled in two directions at once.

And underneath all of it sits the institution this essay is actually about, which the law has barely touched and which is not disappearing. Arranged marriage in India is not a survival being slowly displaced by love. It is being rebuilt — with platforms, with engagement periods that function as courtships, with a veto that young women now hold and use, with a timeline that is argued over rather than announced. This chamber is interested in it for one reason. It is the only large modern society where a negotiated-marriage system did not collapse when women’s education and earnings rose. It had to absorb them instead.

The three judgments of September 2018 are worth stating precisely, because their precision is what makes the later refusals legible. In Navtej Singh Johar a bench of five held unanimously that the provision criminalising what it called carnal intercourse against the order of nature was unconstitutional to the extent that it reached consensual acts between adults. The provision survives for non-consensual acts, for acts involving minors and for bestiality. It was a reading down, not a repeal, and the difference matters: nothing was enacted, something was switched off.

In Joseph Shine the same court struck down the adultery offence. The reasoning is the part this chamber cares about. The provision made a man liable for intercourse with another man’s wife, with the husband’s consent operating as a defence, and the court said the structure was intelligible only on the assumption that the wife was property. Adultery remains a ground for divorce. What was removed was the criminal law’s stake in a wife’s fidelity.

In the Sabarimala matter the court held, by four to one, that a custom excluding women between ten and fifty from a temple was unconstitutional. The sequel is instructive. In November 2019 a review bench, by three to two, kept the review petitions pending and referred a set of broader questions about religious freedom and judicial review to a larger bench. The 2018 holding was not overturned; the argument around it was reopened and has not closed.

Then the two places the law did not move. In October 2023, in Supriyo, the court declined by three to two to read same-sex marriage into the Special Marriage Act or any existing statute, holding that the right to marry is statutory rather than constitutional and that creating a new status would be for Parliament. And the exception to the rape provision — sexual acts by a man with his own wife, the wife not being under eighteen, are not rape — survived the recodification of India’s criminal law and sits in the new sanhita. In October 2024 the Union filed against striking it down. As this is written the matter is still before the court.

The pattern is not hypocrisy and it is not a lag. Courts move fastest where they can switch something off and slowest where they would have to build something. A bench can remove a criminal provision in a morning. It cannot legislate a marriage, and it will not readily enter a bedroom the legislature has marked as private. Every society has a version of this asymmetry. India’s is unusually visible because its constitutional court says out loud what it is doing and why.

The expectation, held almost everywhere outside India and by a good many people inside it, was that arranged marriage would follow the path Japan’s did: a declining share, a quaint older cohort, eventual disappearance into a footnote. It has not happened, and the best demographic work on the question says something more interesting than that it merely persisted.

Allendorf and Pandian, working with nationally representative survey data across marriage cohorts, found the changes real but modest. Women were increasingly active in choosing their husbands. Fewer met their spouse for the first time on the wedding day. Marriage across caste rose and marriage between close relatives fell. And yet substantial numbers of recent marriages still carried the hallmarks of arrangement, and marriages the woman had arranged for herself without her parents remained a small share. Their conclusion is the sentence this essay is built around: the practice is shifting rather than declining, and rather than displacing their parents in the decision, young women joined them in it.

That is a different object from the one the word arranged conjures abroad. The joint arrangement is not a ceremony performed on a person. It is a process with named participants, a defined sequence and a set of things that can be contested. Who may be proposed. How many meetings before an answer is required. Whether the two of them may speak alone, and for how long, and on whose phone. What may be asked — about income, about a previous attachment, about whether she will be expected to leave her job. When the answer must come. And, decisively, whether no is available, and what saying it costs.

None of that is choice in the sense the word carries in a society organised around self-selection. All of it is negotiation, and negotiation is not the opposite of choice. It is what choice looks like when it is exercised against other people rather than in a vacuum. A woman who can refuse three proposals and set the terms of the fourth has less freedom than one who chooses a stranger in a bar and more leverage. The essay is going to keep both halves of that sentence.

The visible face of the rebuild is the matrimonial platform. It is not a dating app with different branding. The account is often created by a parent or a sibling; the profile is a document about a family as much as a person; and the fields are the fields of an arrangement — height, education, occupation, diet, horoscope details, community, caste and sub-caste, and whether the family will consider a match outside them. A large study of more than three hundred thousand profiles on a major Indian platform used exactly those fields to read attitudes toward marrying across caste, and found younger cohorts more open than older ones, and those open to it describing themselves in terms of individual qualities rather than family ones.

The less visible and more revealing face is the engagement. In North India a roka ceremony marks the point at which two families have agreed on a match. Recent qualitative work in the Delhi region, based on interviews with recently married couples, unmarried young people and parents, found that the period after the roka has become a courtship in everything but name. Couples in arranged matches spend it going out, meeting each other’s friends, keeping and disclosing secrets, having and resolving arguments — the whole repertoire the word love marriage was supposed to own. Sometimes it ends in a broken engagement, which carries a social cost that falls unequally on the woman, and which families in that study were nonetheless increasingly willing to absorb rather than deliver a daughter into a marriage she had come to dread.

And the traffic runs the other way too. Couples who met on their own and formed a relationship privately use the roka to convert it into something the neighbourhood can recognise — a match with parental endorsement, socially indistinguishable from an arrangement. The phrase people use for the hybrid, arranged-cum-love, is not a euphemism. It is an accurate description of a two-way conversion mechanism between the two systems.

Dating apps sit beside this rather than replacing it. They are widely used in Indian cities and they are doing something the platforms do not: producing relationships that have not been pre-cleared by anybody. Research on Indian women’s use of them describes a great deal of unpaid safety work — vetting, staging, sharing locations, keeping a friend on a call — which is the tax on a relationship formed outside a structure that would otherwise have vouched for the man. That tax is the cost of the freedom, and it is worth naming rather than romanticising.

Everything above is real and it is bounded, and the boundary has a name. The negotiation inside an arrangement runs over timing, contact, conduct, work, money and the right to refuse a particular person. It very rarely runs over caste.

The India Human Development Survey, which produced the first direct national estimate of the question, asked married women aged fifteen to forty-nine whether theirs was an inter-caste marriage. Around five in a hundred said yes. The figure varies enormously by state — far higher in parts of the north-east and in some western states, far lower elsewhere — and it has risen slowly since. It is a measurement of behaviour, and the measurement of stated attitude points the same way: in a face-to-face survey of about thirty thousand Indian adults conducted in 2019 and 2020, roughly two thirds said it was very important to stop women in their community from marrying into another caste, and a similar share said the same about marrying into another religion.

Two things have to be said immediately, because the sentence above is the one most easily misused. The first is that this is a description of a structure, not of a people. Endogamy is an institution with a history, an economics and an enforcement apparatus; it is not a disposition that individuals are born holding, and the same survey data show it weakening fastest among the young and the educated. The second is that the strongest opposition to it is Indian. The court that heard Shakti Vahini in 2018 held that the right to choose a partner is protected irrespective of what family, community or clan agrees to, and directed the states to act against the extra-legal village councils that police it. A society that produces both the endogamy and that judgment cannot be described by either one alone.

The law here is also a study in how a route can exist and still be closed. The Special Marriage Act, the civil statute under which people of different religions may marry without conversion, requires a thirty-day public notice during which anybody may object. For a couple whose families are opposed, the legal instrument designed for them is the instrument that publishes their intention to the neighbourhood for a month. A High Court held in 2021 that publication is not mandatory where the parties ask that it be omitted, which is a real remedy and also one that has to be known about, asked for and granted.

And attached to the arrangement, in much of the country, is a price. Dowry has been prohibited since 1961. Economic work reconstructing marriage payments across rural India over the twentieth century finds the share of marriages involving dowry roughly doubling between 1930 and 1975 with the real value of payments tripling, and the real value declining thereafter. A prohibition of sixty-five years standing coexists with the practice it prohibits. That coexistence is the same pattern as the marital rape exception seen from the other side: there, the law declines to enter the marriage; here, it entered and was not obeyed.

The negotiation described in this essay is conducted by people who have something to negotiate with. It is worth being exact about who does not.

India’s most recent national family health survey found that just under a quarter of women aged twenty to twenty-four reported having been married before the legal age of eighteen. That share has fallen a long way — it was around twice as large a decade and a half earlier — and the fall is one of the genuinely large social changes of the period. It is also, in a country of this size, an enormous number of individual women, and a woman married at sixteen did not hold a veto, did not have an engagement period that functioned as a courtship, and was not on a platform.

Then there is the woman for whom the law has decided the question cannot arise. The exception in the rape provision does not merely decline to prosecute; it defines the act out of existence where the parties are married and the wife is over eighteen. Whatever one concludes about the difficulty of proof or the risk of misuse — and the Union’s filing raises both — the structure of the exception is that a particular woman’s refusal has no legal content. Nothing in the rebuilt arrangement touches that.

And the economic story that is supposed to underwrite the whole negotiation is more complicated than it is usually made. India’s official labour force survey records female participation among those aged fifteen and over rising steeply over recent years, from under a quarter to around two fifths. That is a large recorded movement. It is also one that specialists read with care, because much of the increase sits in rural self-employment and in unpaid work in family enterprises, and because what a labour survey counts has itself shifted. Earnings that a woman controls are not the same quantity as work a survey records, and it is the first, not the second, that buys leverage inside a marriage negotiation.

Put the three together and the shape is clear. The room to negotiate is distributed along the same lines as education, urban residence, household income and a family disposed to let a daughter use the veto. The apps and the platforms are a photograph of that group. The women with the least room do not appear in the photograph, and an essay that reads the photograph as the country has made the second of the two errors this chamber is about to name.

The first error is to read arranged marriage as the opposite of choice. It is the error a reader outside India makes almost automatically, because the English word arranged implies that the arranging was done to somebody. What that reading cannot see is the veto, the meeting, the engagement that functions as a courtship, the question about whether she will keep her job, the broken roka, the platform profile that a daughter wrote herself. It also cannot see that the alternative on offer in self-selecting societies is not unmediated choice but a different and unnamed mediation. A reader who holds only this error will find Indian women’s accounts of their own marriages incomprehensible or dishonest, which is a reliable sign that the frame is wrong rather than the accounts.

The second error is to read the system as simply modernised — as having quietly become a slower, better-chaperoned version of dating, with the same freedoms arriving by a different road. What that reading cannot see is that around ninety-five in a hundred marriages are within caste, that the civil route across religion publishes your intention for a month, that a quarter of young women were married before eighteen, and that the exception in the rape provision is still there. A reader who holds only this error will mistake the negotiating power of an urban, educated, employed minority for a national condition, and will produce the genre of writing in which a country is congratulated on a change its poorest women have not received.

Each error is the corrective for the other, which is why this essay refuses to choose between them. The first error flatters the outside reader’s sense of her own freedom. The second flatters the inside reader’s sense of progress. The country is the thing that both descriptions fail to contain, and the honest position is to hold the two failures at once rather than to average them into a moderate sentence that is true of nobody.

Japan ran the other experiment. Its own negotiated system, miai, was the ordinary route into marriage within living memory: the national fertility survey series records something close to seven in ten marriages around 1930 arranged through an introduction, including those made by agencies, with love marriage overtaking it during the 1960s and the arranged share down to roughly one in twenty by the middle of the last decade. Japan did not reform its arrangement. It won the argument against it, and the arrangement lost the name.

What did not disappear was the work the arrangement had been doing. This chamber has already written about what happened next — about choosing a partner becoming a strategy, about a generation of women told to become excellent who did, while the norm about who they should partner with did not move, and about the arithmetic rather than anybody’s character producing the gap that followed. The term konkatsu, coined in the late two-thousands on the analogy of job hunting, named the strategy exactly. It did not restore the seat at the table. It reinstalled the search as an individual project with the effort privatised and the outcome treated as a personal result.

This is the comparative finding, and it is not about which country is freer. It is about the difference between a system that is named as arranged and one that is arranged by defaults nobody admits to arranging. A named arrangement has participants who can be addressed, terms that can be contested, a filter that is visible on the screen, a court case that can be brought about caste, a mother-in-law who can be refused and a veto that has a word attached to it. An unnamed arrangement has none of that, and it is still an arrangement: who introduces whom, which colleagues get invited, which ages read as marriageable, whose transfer decides where a couple lives, whose career bends at the first child. You cannot negotiate with a default. There is nobody to address the objection to, and the person who loses by it experiences the loss as a fact about herself.

This essay declines to rank the two. An unnamed arrangement leaves a woman free of a specific authority, and that freedom is real; nobody can require her to meet anybody. A named one gives her something to push against, and that leverage is real; a family that has proposed a match has entered a conversation in which it can be told no. Neither is freedom and neither is captivity, and a reader who wants one of them declared the better system is asking the wrong question of a mirror.

The earlier essay in this chamber on what religion was actually doing found that the rules were the smallest part of it, and that what receded with them was a script, a community that noticed, and a ceremony at every threshold. The Indian arrangement still supplies all three. That is not an argument for it. It is part of the explanation of why it did not collapse when the economics changed, and of why the thing that replaced it elsewhere has felt, to a great many people, like less rather than more.

This house sells evenings, and it should state the awkward part before the attractive one. Its product is a named arrangement. What is on offer is written down in advance, the terms are agreed before anything begins, a word ends the evening without explanation, and the person providing it has no standing to reinterpret any of that afterwards. The house benefits directly from the fact that almost nothing else in a woman’s life is arranged so explicitly. If the unnamed arrangements around her were negotiable, the room where everything is named would be worth considerably less.

That is the whole of the honest claim. A named arrangement is easier to refuse than an unnamed one, and one evening is a small demonstration that being inside a structure and being able to set its terms are not opposites. Whether that demonstration transfers to anything outside the room is not in the house’s gift and is not asserted here.

What must not be claimed follows from it. Nothing in this essay is a model for a marriage, in India or in Japan or anywhere else. Nothing here is instruction in how to negotiate with a family, and a woman whose refusal carries a real cost is not helped by an essay observing that refusal exists. Nothing here endorses or condemns any named judgment, platform, ceremony or campaign, and nothing here is a contribution to any argument being conducted by Indians about India.

The strongest case against this essay is that it has used a country as a screen. A Library written for readers in Japan has gone looking abroad for a mechanism it wanted to find, has located it in India, and has organised a vast and internally contradictory society around a single distinction that serves the writer’s argument. That charge cannot be fully answered, only mitigated: by sourcing each factual claim, by giving the counter-reading its own section, and by declining the comparison’s natural conclusion rather than delivering it.

A second case is that the word negotiation dignifies an institution whose costs fall on women — that describing a veto is a way of not describing the dowry, the broken engagement that damages a daughter’s prospects and not a son’s, or the marriage that could not be refused. The essay has tried to carry those in the body rather than in a footnote, and concedes that a veto exercised at a price is a thinner freedom than the word suggests. A third is that treating caste endogamy as a structure risks making it sound mild. It is not mild, and the judgment on honour crimes cited above exists because people have been killed over it.

The limits are these. National figures conceal enormous variation across states, religions, languages, castes and classes, and a sentence beginning in India is nearly always too large. The best national data on how marriages are arranged is now more than a decade old, and the country has moved since. The qualitative work on engagement periods describes college-educated households in one metropolitan region and is not a national portrait. This essay does not describe Muslim, Christian, Sikh, Adivasi or Dalit marriage practice separately, and each would change the picture. The legal position on the marital rape exception was pending when this was written and may have moved. The Japanese survey figures on arranged marriage are retrospective self-reports whose categories shifted over eighty years. No claim is made about how often any law described here is enforced, or about any individual family. And the essay is written by an outsider to both of the societies it compares.