The American and Iranian delegations sat down in Islamabad using the same words. They meant different things by them. And because the words sounded shared, everyone treated the failure that followed as a failure to agree, when it was something stranger and more permanent: a failure to mean the same thing.

This is the part of the Islamabad talks that the coverage missed. The story was reported as a breakdown over terms, as if two parties had examined the same proposal and one had refused it. But they were never looking at the same proposal. They were looking at the same vocabulary, and the vocabulary was doing opposite work on each side of the table. The determining variable in the collapse was not a disputed clause. It was that the key words, the words the entire negotiation ran on, had already been redefined by each side to mean what that side needed them to mean. Change a single clause and the talks might have moved. Change the meaning of the words and there was nothing left to negotiate, because the negotiation was settling a dispute the language had already decided.

Twenty-one hours, three rounds, no shared dictionary

The facts are not in question. On the eleventh and twelfth of April 2026, for the first time since the war began, American and Iranian delegations met face to face in Islamabad. The United States sent a large delegation led by Vice President JD Vance, with the envoys Steve Witkoff and Jared Kushner alongside him. Iran sent a team led by the parliamentary speaker Mohammad Bagher Ghalibaf and the foreign minister Abbas Araghchi. Pakistan mediated, its prime minister, its army chief, and its foreign minister all in the room. The session ran about twenty-one hours across three rounds, the first indirect, the next two direct.

No agreement was reached. By most accounts the two sides had in fact agreed most of a ten-point ceasefire and broke on only two points, the Strait and the nuclear program. Vance announced it as Iran refusing Washington's terms, which is itself the first instance of the pattern this piece is about: a narrow breakdown over two clauses, rendered to the world as a flat refusal. Within hours, Trump posted, in capital letters, that he was ordering a naval blockade of the Strait of Hormuz. The talks had lasted most of a day and produced nothing but the next escalation.

Hold the shape of that collapse for a moment, because it tells you what kind of failure this was. Twenty-one hours of careful diplomatic language, three rounds, two hundred staff, the most senior people two governments could send, mediators shuttling between rooms. And the entire edifice resolved, in the end, into a single word shouted in capital letters on a social network. The negotiation had been an attempt to find a sentence both sides could sign. It ended in a word neither side had to interpret, because BLOCKADE means only one thing. When the shared words failed, the parties did not fall silent. They fell back to the one vocabulary that cannot be redefined: force, which needs no dictionary.

The mediators were in an impossible position, and their position reveals the mechanism better than either antagonist. Pakistan was not translating only between English and Persian. It was translating between two incompatible meanings of words that, in both languages, sounded like agreement. A mediator can bridge a gap between positions. It cannot bridge a gap inside a word, because there is nothing to carry across: each side hears its own meaning in the other's sentence and believes, briefly, that progress is being made, until the moment the meanings are forced to touch and the sentence splits in two. Pakistan brokered the room. No one could broker the dictionary.

The temptation is to ask which side was unreasonable. That is the wrong question, because it assumes the two delegations were running the same operation in opposite directions. They were not. They were running the same operation, the redefinition of a shared word into a private meaning, and they were both doing it at once. Look at the words.

The word that meant surrender: "deal"

Every American statement used one word. Iran must accept the deal. Iran refused the deal. No deal. The word arrived so often it stopped being noticed, which is exactly when a word is doing its most important work.

A deal, in ordinary use, describes two parties arriving at terms each can live with. It implies negotiation, exchange, the possibility that both walk away with something. But Washington had arrived with a single non-negotiable demand, that Iran commit permanently never to build a nuclear weapon, and everything else was secondary to it. Iranian state media rendered the American posture in a single sentence: you have been militarily defeated, just concede. When one party fixes the outcome before the conversation starts, the word "deal" is no longer describing a negotiation. It is concealing its absence. The language of commerce is made to perform the work of capitulation, and because "deal" sounds voluntary, like something one chooses, the surrender disappears inside the word.

George Orwell named this mechanism eighty years ago. Political language, he wrote, is built to make capitulation sound like choice and to give "an appearance of solidity to pure wind." The word "deal" is solidity laid over pure wind. It makes a demand sound like an offer. It lets the stronger party describe an ultimatum in the grammar of mutual agreement, and it lets the world report a refusal to surrender as a refusal to bargain. Nothing was lied about. The facts were all on the record. The work was done entirely by the choice of noun.

And the word does most of its work outside the room, which is the point usually missed. "Deal" was not aimed only at the Iranian delegation. It was aimed at the global audience watching the talks, because once the frame is "deal," every headline writes itself in that frame, and Iran's refusal to accept an ultimatum is automatically rendered as a refusal to compromise, as intransigence, as the unreasonable party walking away from a reasonable bargain. The noun pre-writes the coverage. By the time Iran said no, the word had already arranged the world to hear that no as proof of bad faith. A single well-placed word can lose you the argument in the press before you have made it at the table.

The word that meant impunity: "sovereignty"

Iran was performing the identical operation from the other end of the table, and its word was "sovereignty." Tehran arrived defending it, invoking it, refusing to compromise it. And like "deal," the word is doing concealment work, because it takes a set of specific, contestable, negotiable choices and files them inside a container that sounds like an inviolable right.

What did sovereignty mean, concretely, in Iran's hands at Islamabad? It meant refusing IAEA inspectors access to its enrichment sites. It meant maintaining enrichment capacity the rest of the region regarded as a threat. It meant asserting the right to levy tolls on the world's most important oil passage. These are positions. They can be argued, traded, conceded, or defended. But folded inside the word "sovereignty," they stop being positions and become a principle, and a principle cannot be negotiated without the negotiator appearing to surrender something sacred. The word converts a bargaining position into a matter of dignity, and dignity does not come to the table. Two delegations. Two words. The same machine, pointed in opposite directions, each converting a negotiable demand into a non-negotiable absolute.

And like "deal," the word faces in two directions at once. Outward, it tells the world that Iran is defending a right, not protecting a weapons program, which is a far more sympathetic posture in every capital that resents American pressure. Inward, it does something even more binding: it tells the Iranian population that any concession at the table is a concession of national dignity, which means the negotiator who softens on enrichment is not adjusting a position but betraying the nation. The word builds a wall behind its own delegation. Having called the stakes "sovereignty" to its own people, the regime cannot then trade them away without paying a domestic price it has just raised on itself. The vocabulary does not only describe the negotiating position. It removes the negotiator's freedom to move, which is sometimes precisely why it is chosen.

The word that meant "not here": "ceasefire"

There was a third party not in the room whose vocabulary mattered as much as anyone's, and a third word. On the eighth of April, the day the ceasefire formally took effect, Israel launched a large operation against targets across Lebanon, striking densely populated areas within minutes of the truce beginning. Reported casualty figures varied widely between agencies, as they do in the first hours of any strike, and the divergence itself is part of the record. What was not in dispute was the timing: a hundred targets, on the day the guns were supposed to fall silent, not before and not after.

Pakistan said Lebanon was covered by the truce. Netanyahu said it was not. Washington did not dispute the exclusion. The ceasefire was not broken. It was retroactively redefined. The word "ceasefire," which sounds total, was quietly made selective, a truce that covered some borders and not others, and the operation was launched not in defiance of it but in the space created by editing what it covered. A single word was made to mean "fire ceases here, and continues there," and once the word was edited, no agreement had technically been violated. The redefinition did the work that violating the ceasefire would otherwise have required, and at none of the cost.

This is the place to be concrete about what a redefined word costs, because here the abstraction touches the ground. When "ceasefire" was quietly edited to exclude Lebanon, the edit did not stay on paper. It landed, within minutes of the truce taking effect, on specific neighborhoods, in roughly a hundred strikes, on people who had just been told the war was over. The casualty figures differed between agencies, as they always do in the first hours, but the order of magnitude was in the hundreds. Those people were not killed by a broken ceasefire. They were killed in the gap between two meanings of the word, in the space one party had opened by deciding, unilaterally and after the fact, what "ceasefire" would be taken to cover. The semantic operation and the physical one were a single act. A word was edited, and the editing had a body count. This is the part that the elegance of any analysis of language can quietly hide: the fight over a word is never academic, and the people who pay for a word's ambiguity are almost never the people who chose the word.

The damage outlives the single strike, because the redefinition is reusable. Once it has been established, and accepted, that "ceasefire" can mean "on these fronts but not those," the word is permanently degraded for every future truce in the region. The next ceasefire now arrives pre-hollowed, its central term already proven to be selectively applicable, so that no party can rely on it without first negotiating, in advance, the meaning of the very word meant to end the fighting. A word that cannot be trusted to mean one thing cannot end a war, and once one party has shown the word can be edited after signing, it has poisoned the instrument for everyone, including itself the next time it wants a truce to hold.

The word that erased the authors: "market"

The fourth word is the one that hides the most, because it hides the existence of anyone at all. While the delegations argued, the most consequential blockade of the war was already in force, and it had not been imposed by any navy. Within forty-eight hours of the opening strikes, the London marine war-risk underwriters reclassified the entire Gulf as a conflict zone, war-risk premiums on a single tanker transit leapt by an order of magnitude, into the millions per voyage at the peak, and traffic through the Strait collapsed. The insurance market closed the passage before Iran's navy could, and the institutions that priced that risk overlapped, through public shareholdings, with the institutions that profit from the weapons, the oil, and the disruption itself.

That mechanism deserves its own full account, and it has one. The point here is narrower and concerns a single word. When the Strait effectively closed, every description used the same construction. The market responded. The market priced in the risk. The market closed the Strait. The grammar is worth slowing down on, because it is doing something the other three words do not. "Deal," "sovereignty," and "ceasefire" each hide a contested meaning behind an agreed word. "The market" goes one level deeper: it hides the existence of any actor at all. It is not a euphemism for a meaning. It is a euphemism for agency.

Compare two sentences describing the identical event. "The market closed the Strait" and "a committee of London underwriters closed the Strait" report the same fact, but they place it in two different universes. The first is weather, a force of nature, something that happens to everyone and is authored by no one, that can be regretted but not blamed, observed but not lobbied, suffered but not regulated. The second has a room, a table, names, a decision, and therefore a door through which responsibility and politics can enter. The work of the word "market" is to keep that door shut. It takes a structure made of identifiable committees and traceable decisions and dissolves it into a passive verb, and once the decision has become weather, no one need explain it, defend it, or answer for it.

This is why "market" is the most efficient word in the entire negotiation and the one no party contests. The American and Iranian delegations fought over "deal" and "sovereignty" because each wanted its meaning to win. No one fights over "market," because its authorlessness serves everyone at once: the actors it conceals prefer to remain weather, and the antagonists who suffer it prefer a neutral force to a named adversary, since a force of nature implicates no one and demands no response. It is the rarest kind of redefinition, the consensus one, a word every party is content to leave undefined because each profits from the disappearance it performs. The other three words decided what the talks would mean. This one decided that the largest actor in the room would not appear to be in the room at all.

Two ways to make a thing disappear

Lay the four words side by side and they are not four versions of one trick. They are two tricks, and the difference matters more than the similarity. The first three bend a meaning. The fourth deletes a doer.

"Deal," "sovereignty," and "ceasefire" all still point at something real. They simply point at it misleadingly. "Deal" points at a negotiation that is actually an ultimatum, "sovereignty" at a bargaining position dressed as a sacred right, "ceasefire" at a truce quietly hollowed of part of its territory. The word keeps its job of referring; it just refers crookedly. You catch this trick by asking what the word actually means here, and measuring the gap between that and what it pretends to mean.

"Market" does something different. It does not point at the wrong thing. It stops pointing at anyone at all. And it is not alone. It belongs to a family of words that exist precisely to let an event happen without a hand behind it: the international community, the experts, intelligence sources, the economy, the system, stakeholders. Each one takes a set of specific, nameable, answerable people and dissolves them into a faceless abstraction that can be quoted but never cross-examined, blamed but never summoned to the table. "The market closed the Strait." "The international community expects." "Experts agree." "The economy demands." In every case a decision made by identifiable people is repackaged as a condition of the weather.

These two tricks call for two different questions, and the second question is the one most readers never think to ask. Against the first family you ask what the word means. Against the second family that question is useless, because the word means nothing, that is its whole function. The right question is the one the word is engineered to prevent: who disappeared when this word was used. When a sentence tells you the market decided, or the experts agree, or the international community expects, the most important content of the sentence is the name it is not saying. Find that name and the weather turns back into a decision, made by someone, for reasons, who could in principle be asked to account for it. The first three words hide a meaning you can recover by translation. The fourth hides a person you can recover only by refusing the word.

How a word becomes a fact

Step back from the four words and the deeper mechanism comes into view, the one that makes this a structural finding rather than a complaint about spin. A negotiation is supposed to be the place where a shared reality is built between parties who disagree. It runs on the assumption that the words on the table point at the same things, so that the disagreement is about what to do, not about what the words mean. Islamabad violated that assumption at the root. The parties did not bring different proposals. They brought different dictionaries, and a negotiation between two dictionaries cannot resolve anything, because every apparent agreement is two different agreements wearing the same sentence.

This is why the failure was structural and not personal. The philosopher Michel Foucault spent his career on a single observation: that power is rarely the thing that wins an argument, it is the thing that decides beforehand which terms the argument may be conducted in, which descriptions count as serious and which are dismissed before they are heard. Whoever controls the vocabulary controls the outcome before anyone speaks, because the loser of a negotiation is usually the side forced to argue in the winner's words. Whoever owns the words owns the outcome, and whoever owns the outcome wrote the words. The fight that looks like a fight over terms is really a fight over whose dictionary the terms will be read in, and that fight is settled long before the delegations arrive. The United States fought to make "deal" the frame, so that any Iranian refusal would read as intransigence. Iran fought to make "sovereignty" the frame, so that any concession would read as betrayal. Neither was trying to win the argument. Each was trying to win the dictionary, because the dictionary decides the argument in advance.

And there is a reason this guarantees failure rather than merely tilting it. The philosopher Jürgen Habermas drew a line between two ways of using language. In one, communicative action, the parties use words to reach a shared understanding, and the words must mean the same thing to both or the project collapses. In the other, strategic action, the parties use words to achieve an outcome, and meaning is subordinate to advantage. A negotiation is supposed to be communicative action with hard bargaining inside it. Islamabad was pure strategic action wearing the clothes of communication: every word deployed to fix a result rather than to be understood. Once both sides are using language strategically, the shared meaning that agreement depends on is not strained, it is structurally absent, because neither party is even attempting to mean the same thing as the other. The talks were not a communication that failed. They were the absence of communication, conducted for twenty-one hours in the costume of a conversation. The vocabulary had already done what the negotiators were pretending to attempt.

Who taught the words to mean this

It would be a mistake to imagine that the delegations invented these private meanings in the room. They arrived already fluent in them. The vocabulary that failed at Islamabad was written years earlier, in places far from any negotiating table, by the institutions that train the people who eventually sit at them.

This is the part the spin-detector misses. A diplomat does not decide, on the morning of the talks, that "deal" will mean capitulation. He has spent a career inside a policy ecosystem, think tanks, briefings, doctrine, press lines, in which "deal" already carries that weight, in which Iranian enrichment is already "the threat" and American demands are already "terms," and he reaches the table speaking that dialect as naturally as his mother tongue. The Iranian official has spent his career inside a mirror ecosystem in which "sovereignty" already means non-negotiable and Western pressure already means humiliation. Neither is lying. Each is fluent. They are fluent in different languages that use the same words, and the fluency was manufactured upstream, in the slow machinery of how each side reproduces its own experts.

The press completes the circuit. Each major outlet tends to adopt the vocabulary of its own side's foreign-policy establishment, so that by the time talks begin, an entire information environment has already been trained to hear "deal" as reasonable and Iranian "intransigence" as the obstacle, or, in the mirror outlets, to hear "sovereignty" as dignity and American "bullying" as the aggression. The words are not only inherited by the negotiators. They are pre-installed in the audiences that will judge the negotiators, which means each delegation faces a domestic press already fluent in its own dialect and primed to read any concession as defeat. The dictionary is enforced from below as well as trained from above.

That is why the mechanism needs no conspiracy and survives any individual. Replace every negotiator at Islamabad and the new ones would arrive speaking the same incompatible dialects, because the dialects are not personal choices but institutional products, the output of two systems that each train their people to hear one set of meanings as obvious common sense and the opposing set as bad faith. The redefinition of the words did not happen at the negotiation. The negotiation was merely where two pre-written dictionaries were finally laid side by side and discovered to share no entries. The talks were the place the incompatibility became visible, not the place it was created.

The dictionary has decided talks before

This is not unique to Iran, and seeing it elsewhere is what turns it from an observation into a law. The most consequential security dispute of the past forty years runs on exactly the same mechanism, and it turns on a single phrase. In 1990, as Germany reunified, Western officials gave Soviet leaders assurances about NATO. Whether those assurances amounted to a binding promise that the alliance would not move "one inch eastward," or were merely informal remarks about the status of East German territory, has been argued ever since, because the two sides took two incompatible meanings from the same words and each built a grand strategy on its own reading. Russia has treated subsequent NATO enlargement as the breaking of a promise. The West has treated it as the exercise of a freedom never formally surrendered. Decades of confrontation, and ultimately a war, grew in the gap between two meanings of one phrase that was never pinned to a shared definition.

The pattern is older still, and at its extreme it has killed. Historians have long argued over a single Japanese word, mokusatsu, used by Tokyo in response to the Allied surrender ultimatum in the summer of 1945. The word can mean "to reject with contempt," and it can mean "to withhold comment for now," and the question of which meaning Tokyo intended, and which Washington received, has never been settled. On one reading, an ambiguity in one word, translated into the harsher of its two meanings, helped clear the path to the atomic bomb. The case is genuinely disputed, and the dispute is the point: even whether a word's ambiguity changed history is itself an argument about what the word meant. The stakes of the dictionary are not academic.

The closer pattern is "peace with honor," which described a withdrawal as a victory, and "appeasement," once a respectable word for a reasonable policy before it became the name of a catastrophe, and "security," which in the mouth of any state routinely means its own safety purchased with its neighbor's insecurity. In each case the structure is identical to Islamabad: a word that sounds shared, doing opposite work for each party, with the disagreement displaced from the dictionary, where it actually lives, onto the events, where it cannot be solved. The negotiators always believe they are arguing about the world. They are arguing about the words, and losing, before the world is ever discussed.

The deal that proved the point

The strongest confirmation came two months later, and it came in the form of an apparent success. In mid-June the two governments announced a framework: a ceasefire extension, a path to reopening the Strait, a signing ceremony. The headlines called it a breakthrough. It was, in fact, the Islamabad language problem ratified rather than solved, because the two sides immediately described the same document in incompatible terms.

Washington presented it as the road to ending the war, a durable settlement in progress. Tehran presented it as a ceasefire that applied on all fronts at once, with particular emphasis on Lebanon, and that could be revoked if violated on any one of them, with the Strait moving to a joint Iran-Oman administration rather than the open international passage of before. Washington then disputed that the Iranian account of the terms was accurate at all. So here was a signed framework that the two signatories could not describe in the same language, could not agree on the scope of, and could not even agree had the contents one of them published. The words had not converged. They had simply learned to coexist on the same page while still meaning opposite things, which is what a deal looks like when the dictionary problem is never addressed, only papered over. Islamabad failed loudly. June succeeded quietly into the identical condition.

This is what the signing ceremony is actually for. A ceremony performs agreement at exactly the moment when the meanings have not converged, and it performs it precisely because they have not. The handshake, the flags, the signatures supply the appearance of a shared understanding that the words themselves cannot deliver, and they let each side go home and tell its own population that the other side accepted its meaning. The ritual is not the celebration of a settled meaning. It is the substitute for one. A framework that genuinely meant the same thing to both sides would not need a ceremony to convince anyone it existed; the oil would simply move. The more elaborate the performance of agreement, the safer the bet that the dictionaries underneath it still do not match.

What a real negotiation would have to do first

The diagnosis points at its own remedy, and the remedy is so rarely attempted that its absence is itself evidence of how the system works. If the load-bearing words mean different things to each side, then a negotiation that begins from positions is building on sand, because every position is stated in words the other side will hear differently. The repair would be to negotiate the dictionary before the positions: to agree, explicitly and in advance, what "deal" will and will not mean here, what specific commitments "sovereignty" does and does not cover, exactly which fronts a "ceasefire" includes. Settle the meanings first, and only then open the substance.

This almost never happens, and the reason it never happens is the whole point. Defining the words in advance would strip them of their utility as weapons. The United States does not want "deal" pinned down, because its power lies precisely in the word's ambiguity, in being able to call an ultimatum a deal. Iran does not want "sovereignty" defined, because the word's reach is exactly what it is defending. Each party's advantage depends on keeping its key word undefined, which means each party has a direct interest in never doing the one thing that would make agreement possible. The negotiation cannot fix the language problem because the language problem is not a bug in the negotiation. It is the strategy of both sides, and a flaw that serves everyone at the table is never on the agenda.

So the talks proceed as theater, each side performing a willingness to agree while protecting the verbal ambiguity that makes agreement impossible, and the mediator shuttles between two rooms carrying sentences that mean different things in each. The structure guarantees the outcome. You can predict the failure of such a negotiation not from the hostility of the parties but from a single test: did they define their terms before they argued their positions? If they did not, they were never negotiating a peace. They were performing one, in two languages, on the same stage.

Why this is not a story with a villain

It would be easy, and it would be wrong, to end this with a culprit, to say that one side weaponized language and the others were its victims. The discomforting finding is that the mechanism runs in every direction at once. The United States redefined "deal" into capitulation. Iran redefined "sovereignty" into impunity. Israel redefined "ceasefire" to exclude its other wars. The financial system redefined "market" into a structure with no authors. Four words, four actors, one operation, and no innocent party among them. The redefinition of language is not a tactic one side used against the others. It is the medium they were all swimming in, the default behavior of power under pressure, requiring no conspiracy and no coordination because each actor reaches for it independently, the way water finds the lowest path.

That is what makes it durable. A villain can be named and removed. A shared habit of bending words cannot, because there is no one to remove, only a practice that every party adopts the moment the stakes are high enough. The framework of negotiation assumes that disputes can be resolved between the parties who created them. Islamabad revealed the flaw in that assumption: the dispute was not really between the parties. It was inside the words, and the words belonged to no one and to everyone.

The strait does not read

Twenty-one hours did not prove that peace is impossible. They proved something more precise and more useful. They proved that peace cannot be built in the language of victory, because the language of victory is built to prevent it, each side's words engineered to make its own position sound principled and the other's sound like surrender. When the words fail in this way, the dispute does not pause. It moves to the narrowest physical point in the system, a strait twenty-one miles wide, and it squeezes, in tolls and premiums and collapsed traffic, until the pressure that could not be released in words is released in oil.

The negotiators have gone home. The framework has been signed and disputed in the same breath. The Strait is still there, indifferent to which delegation believed it had won, because the water does not read the press releases and the premium does not parse the vocabulary. A word can be redefined. A chokepoint cannot. And when a generation of negotiators has trained itself to win the dictionary rather than the argument, the dictionary is the one thing that will never deliver a peace, because a peace requires two parties to mean the same thing by the same word, and that is the single thing none of them was willing to do.

They did not fail to understand each other. They understood each other perfectly. That was the problem.

Evidence Map

Facts, interpretations, forecasts, and disconfirming signals.

Core claim. The Islamabad talks failed not over terms but over meaning: each party redefined the shared words the negotiation ran on ("deal," "sovereignty," "ceasefire," "market") into private, incompatible meanings, so the negotiation could settle nothing. The determining variable was control of the vocabulary, not the disputed clauses, and the June framework confirmed it by being described in incompatible terms by both signatories.

Evidence level. Facts (high): the 11-12 April 2026 Islamabad talks, delegations (Vance/Witkoff/Kushner; Ghalibaf/Araghchi; Pakistani mediation), ~21 hours over three rounds, failure, the subsequent Hormuz blockade announcement; the April ceasefire and the dispute over whether Lebanon was covered; the collapse of Gulf tanker traffic under war-risk reclassification; the June 2026 framework and the two sides' divergent descriptions of it. Interpretation (marked): that the redefinition of shared words is the operative mechanism of the failure, and that it operates identically across all parties.

What would confirm this. Future negotiations in this conflict continue to break down on incompatible definitions rather than on specific clauses; signed agreements continue to be described in mutually exclusive terms by the signatories.

What would disprove this. If a settlement is reached in which both sides publicly describe the same terms in compatible language, and the agreement holds, then the dispute was about clauses after all and the language thesis does not carry the weight placed on it here.

Watchlist. Whether the June framework's signatories converge on a shared description or continue to contradict each other; the next round of talks and whether they founder on terms or on meaning; the gap between the announced "peace" and the war-risk premium on a Gulf transit.

Jerry van der Laan writes The Manifest Archive. He traces the structures beneath the events.