The Weight of Words

A few weeks ago, a friend sent me a clip from a 2015 event at De Balie in Amsterdam. Nawal El Saadawi — the late Egyptian feminist writer, physician, and activist — is on stage. A Palestinian woman in the audience begins a question by referencing “the Middle East.” Saadawi interrupts before the question is finished.

“The minute I hear ‘Middle East,’ I become upset. Because this is a colonial language.”

The woman tries to continue, but Saadawi stops her twice more: “Not Middle East. Just a second.” She will not let the term pass.

Her argument, in full: Egypt was called the “Middle East” relative to London; India was the “Far East” — also relative to London. Both were British colonies. The naming convention is Eurocentric by design. Then the reversal: when she goes to London, she says she is going to the “Middle West.” When she goes to the United States, the “Far West.” People laugh. But nobody laughs when someone says “Middle East.” For Saadawi, this asymmetry is proof that we have internalized the colonial framing so deeply that the original terms feel neutral while the reversed terms feel absurd. She proposes alternatives: North Africa. West Asia. North West Asia. “We need to decolonize the language,” she says. Then: “Go on.”

Saadawi was an Egyptian feminist writer and physician who knew this history firsthand. But her argument illustrates something that reaches far beyond this one exchange — how language bundles description and judgment, and how that bundling shows up everywhere from geopolitical discourse to the dinner table.

1. The Colonial Language Trap

Let me be precise about what Saadawi does in this clip.

She starts with a descriptive fact: the term “Middle East” originated in a colonial context. Britain colonized the region. The naming convention is Eurocentric — it locates the region relative to Europe. These are historical facts, and Saadawi — who grew up in British-colonized Egypt — knew them firsthand.

But she does not stop at description. She moves from “this term has colonial origins” to “this term is colonial language,” and from there to the implicit claim that using it is bad, wrong, or complicit in colonization. The word “colonial” shifts mid-argument from descriptor to judgment. The term is no longer a geographic convention with a known origin — it is a weapon, and anyone who uses it is complicit.

This is the move I want to examine. It is not unique to Saadawi. It is everywhere — in academic departments, on social media, in political rhetoric. And, crucially, in our closest relationships.

2. How This Works, Structurally

Saadawi’s move has a specific structure:

1. Identify a descriptive fact (X happened).
2. Attach a moral valence to that fact (X happening was bad).
3. Treat the fact and the valence as inseparable.
4. Accuse anyone who acknowledges the fact of endorsing the valence.

Step 4 is the mechanism that shuts everything down. If I say “the term Middle East has colonial origins,” I am stating a fact. If you respond “so you admit it’s colonial language,” you have already bundled the fact and the judgment. I can no longer discuss the origin of the term without appearing to defend colonization. The conversation is over before it began.

Saadawi’s reversal — “Middle West,” “Far West” — works because it exploits an asymmetry of convention. People laugh at “Middle West” because they have never heard it. They do not laugh at “Middle East” because they have heard it thousands of times. This is not evidence of colonialism. It is evidence of how language works: familiar terms feel neutral, unfamiliar ones feel strange. Try calling a “telephone” a “distant-speaker” in English and see if people don’t laugh.

The term “Middle East” persisted not because Britain enforced it at gunpoint for two centuries, but because the West — Britain and later the United States — became the primary geopolitical frame through which global events were interpreted. This is Eurocentric. Yes. But that is a description, not an indictment.

The deeper problem with Saadawi’s move is that it perpetuates exactly what it claims to oppose. By insisting that colonial history permanently contaminates the language we use, it ensures the past is never left behind. This is not decolonization. It is a self-renewing grievance loop.

3. The Dinner Table — Emotional Language and the Same Trap

Now I want to move the same lens to a different arena.

A friend and I were talking about relationships — I had recently ended one; she was in one. The conversation turned to conflict, and specifically to what happens when one person says:

“You made me feel this way.”

I have come to believe this is one of the most destructive things you can say to a partner. Not because expressing feelings is wrong — it is essential. But because the sentence does the same thing structurally that Saadawi’s argument does. It bundles a description (“something happened”) with a causal judgment (“you caused my emotional state”) and presents both as a single, indivisible fact. The same four-step operation: identify a fact, attach a valence, fuse them, and treat anyone who questions the fusion as hostile. Saadawi did it with colonial history; couples do it with who-left-the-dishes.

Marshall Rosenberg, who developed Non-Violent Communication (NVC), spent his career teaching people to untangle these two things. NVC structures communication around four components: observations (what you see or hear, without judgment), feelings (your emotional state, not something done to you), needs (the universal human need connected to the feeling), and requests (a specific, doable action). The formula is: “When I observe X, I feel Y, because I need Z. Would you be willing to do W?”

The first component — stating observations without judgment — is exactly what “you made me feel” fails to do. It jumps from observation to causal attribution in a single breath.

The person on the receiving end now has two options.

Option 1: Defend themselves. They point out that you, not they, are responsible for your emotional state. Your reaction is mediated by your history, your mood, your interpretation of events. But the moment they say this, you hear dismissal. You feel invalidated. You double down. They double down. The spiral begins.

Option 2: Accept the accusation. They absorb it. They apologize. They take responsibility for your feelings. This preserves the peace, but at a cost: they have surrendered their version of reality to yours. Rosenberg would call this a “life-alienating” response — one that disconnects a person from their own experience by prioritizing compliance over honesty. Both people lose, even if one appears to win.

If the sender tends to own their reactions and attribute them carefully, Option 2 may work for a while. But if they do not, Option 2 becomes a slow poison. The receiver accumulates misattributions they did not consent to. Eventually, the injustice becomes unbearable, and they switch to Option 1 — perhaps explosively, perhaps after months of silent resentment.

This dynamic is not just intuitively familiar. It has a formal structure, and it is worth understanding directly.

4. A Brief Primer on Repeated Games

What I just described is what game theorists call a repeated game.

The starting point is the Prisoner’s Dilemma: individually rational choices produce collectively worse outcomes. Two players can each cooperate or defect. Defection always gives a better individual payoff regardless of what the other does — so both defect, and both end up worse off than if they had cooperated.

That is the one-shot version. When the game is repeated, the math changes. You are no longer optimizing for a single round; you are optimizing for a relationship. The winning strategy in Robert Axelrod’s famous tournament was tit-for-tat: start by cooperating, then mirror your partner’s last move. It is nice (never defects first), retaliatory (punishes defection), forgiving (returns to cooperation immediately when the other does), and clear.

Tit-for-tat has a vulnerability: if one player defects first — even by accident — the relationship can cycle through alternating defections and retaliations that never fully recover. This is why the socially optimal equilibrium is not tit-for-tat starting from a random position. It is mutual cooperation from the very first move, sustained by trust built over repeated interactions.

Apply this to relationships. The emotional equivalent of cooperation: the sender avoids accusatory language, the receiver hears the feeling rather than the accusation. Defection: the sender makes a causal claim about the receiver’s responsibility (“you made me feel”), the receiver defends instead of listening.

A one-shot game would incentivize Option 2 every time — absorb the accusation, end the immediate conflict. But relationships are repeated games. If the sender consistently misattributes emotional responsibility, the receiver will eventually stop cooperating. This is not a moral failure. It is rational. No strategy survives indefinitely against a partner who never reciprocates.

The desirable equilibrium is mutual cooperation from the start: both people commit to separating observations from judgments, owning their interpretations, and extending each other the benefit of the doubt. This is not naive. It is the strategy that produces the best long-term outcome for both people.

Rosenberg understood this implicitly. NVC’s framework — observe without judging, express feelings without blaming, identify needs rather than grievances, make requests rather than demands — is a cooperation protocol for a repeated game. Communication is not a one-shot transaction. The protocol determines whether the exchange converges to understanding or spirals into blame.

5. The Sender, the Receiver, and the Message

A game-theoretic model tells us why defection leads to defection, but not where the breakdown happens. For that, we need to look inside the message itself.

Every message has three components: sender, message, receiver. Noise can enter at any point. The sender carries a full internal world — history, mood, unspoken assumptions. Some of that context is inevitably lost in forming the message. The message itself is imperfect: language is ambiguous, some emotional states resist articulation. The receiver decodes through their own filter — their history, defenses, interpretation of tone, expectations about what the sender “really means.” In a high-stakes emotional conversation, the noise is deafening.

A statement like “you made me feel this way” is noisy for two reasons.

First, it makes a causal claim that is almost certainly false, or at least incomplete. You did not make me feel anything. Something happened. I interpreted it. My interpretation produced a feeling. You were involved in the first step; the rest was me. A spark does not make a forest burn; dry wood does. My emotional state is the forest, and it was dry long before you arrived.

NVC addresses this directly. Rosenberg insisted that feelings be expressed as your emotional state, not as something done to you. Its formula — “When I observe X, I feel Y, because I need Z” — separates observation from feeling from need, leaving attribution to the side. This is not evasion. It acknowledges that my emotional reaction belongs to me, even when your action triggered it.

Second, the statement forces the receiver to choose in real time between two interpretations: “my partner is expressing a feeling” or “my partner is accusing me.” Pick wrong — hear accusation when feeling was intended — and cooperation becomes defection.

The solution is not to stop expressing feelings. It is to separate the expression from the attribution. “I felt hurt when that happened” is better than “you hurt me.” The first leaves room for a conversation about why I felt hurt. The second closes that conversation before it starts. Rosenberg would point out that “you hurt me” is not actually a feeling — it’s a thought about what someone did to you. The feeling is hurt. The attribution is optional.

The receiver, meanwhile, has their own job. When someone says something that feels like an accusation, the natural response is to defend — to correct the account, to explain your intentions. But the other person did not come for a fact-check. They came with a feeling. Responding to the feeling by debating the facts communicates that the feeling does not matter until the facts are settled — and the facts may never be settled.

NVC’s concept of empathic listening is the receiver’s discipline: hear what the other person is feeling and needing before you agree, disagree, fix, or defend. “Are you feeling hurt because you need to be heard?” The goal is to connect with the feeling beneath the words — even when the words contain an accusation. This is not surrender. It is discipline. It means deferring the debate about attribution until the emotional temperature has dropped.

This is hard. It is harder when the sender has a track record of misattributing. If your partner has consistently held you responsible for emotional states you did not cause — a defection pattern, in game-theoretic terms — your trust erodes. You hear accusation in every sentence. You choose Option 1 more quickly. The spiral tightens.

Trust is not granted once and then enjoyed forever. It is renewed — or depleted — in every interaction.

6. The Philosophy of Facts — Why This Is Harder Than It Looks

Everything I have said so far assumes we can distinguish facts from interpretations. But this distinction is not as clean as we might wish.

Descartes tried to doubt everything and found he could only be certain he was thinking. Cogito, ergo sum. Everything else — the table, his body, the existence of other minds — could be an illusion. He was not endorsing radical skepticism; he was trying to find a foundation it could not erode. The exercise reveals something important: what we treat as facts rest on assumptions we rarely examine.

Consider the cold beer. Two people hold the same beer at 5 degrees Celsius. One finds it cold; the other does not. The statement “this beer is cold” sounds factual, but it is a subjective report filtered through physiology, expectation, and context. Specifying the temperature — 5 degrees Celsius — moves us closer to a fact, but not all the way. Celsius is a measurement system invented by humans, relying on properties of water at specific pressures, assuming a particular scale, defined by boiling and freezing points that shift with altitude and impurities. Celsius is enormously useful. But it is not an unmediated description of reality. It is a model, built on assumptions, that happens to work.

Facts are thick. They carry assumptions and conventions within them. This does not make them false. But it should make us humble about how we wield them — especially against other people.

NVC’s observation component implicitly acknowledges this thickness. Rosenberg knew that even observing is theory-laden — an observation is a selection of what to notice, framed by what you consider relevant. This is why NVC insists on concrete observations: “I noticed you arrived at 7:15, and we had agreed on 7:00” rather than “you are always late.” The more concrete the observation, the less the other person can dispute it, and the more space there is for the feelings and needs that follow.

In an emotional conversation, “you made me feel this way” is not a fact like “5 degrees Celsius.” It is a construction — built from observation, interpretation, emotional state, personal history, and a dozen other variables. The error is not in having the construction. The error is in presenting it as raw fact and demanding the other person accept it as such.

This is the same error Saadawi makes. She observes a historical fact — colonial origins — and presents her interpretation as if it were the fact itself. The colonial origin is real. The condemnation she attaches to it is a construction. Bundling them makes it impossible to discuss the origin without appearing to defend colonization. The same epistemic mistake, operating at different scales.

7. The Two Errors — And Where the Middle Ground Sits

The mistake I am diagnosing has a mirror image. Both are wrong.

Error 1: Naive objectivism. The belief that facts are self-evident, that my perception of reality is reality, and that anyone who disagrees is ignorant or dishonest. This is what leads someone to say “you made me feel this way” and mean it as a closed, non-negotiable statement. It is also what leads someone to dismiss all criticism of “Middle East” as politically correct nonsense without engaging the history.

Error 2: Radical subjectivism. The belief that facts do not exist, that everything is interpretation, that your truth is your truth and my truth is my truth and neither can be questioned. This is what leads someone to say “you made me feel this way” and expect you to accept it without scrutiny because all emotional reports are equally valid. It is also what leads someone to claim that because a term has colonial origins, it is irredeemably oppressive, and no discussion of its utility is permitted.

Both errors destroy communication. Naive objectivism destroys it by denying the other person’s perspective. Radical subjectivism destroys it by denying the possibility of shared ground.

The middle ground: we can be reasonably certain about many things, based on axioms or principles we agree to accept. We can reserve the possibility that those axioms are wrong without paralyzing ourselves. Science works this way — we do not need absolute certainty, we need shared methods, transparent assumptions, and the willingness to revise when evidence demands it.

NVC lands in this same middle ground. It does not say “there are no facts” or “all interpretations are equal.” It says: state your observation concretely so the other person can see what you saw. Then express your feeling and need, so they can understand what that observation triggered in you. Then make a request. The entire framework bridges two subjective experiences without denying either.

8. The Shared Framework — Why It Must Be Shared

There is a temptation to treat this middle ground as something each person can adopt individually. I clean up my side; you clean up yours. We assume we will converge.

This does not work.

The middle ground is not two parallel individual practices. It is a shared model — a common framework for interpreting and responding to each other. In the language of philosophy of science, it is a paradigm: a set of shared assumptions, methods, and standards that enable a community to make progress. It only works if both people are operating inside it.

The science parallel

Scientists can disagree productively because they agree on a great deal first: what counts as evidence, what methods are valid, what standards a theory must meet. These are conventions, not facts about nature. But without them, science collapses. Two researchers with incompatible standards of evidence cannot resolve a disagreement; they cannot even have the same disagreement.

This is why Thomas Kuhn’s paradigm crises are so disorienting. The old methods stop working. The community fractures. Progress halts until a new shared framework takes hold. The crisis is not that the facts are uncertain. The crisis is that the community no longer agrees on how to interpret them.

Relationships work the same way.

The “communication is key” problem

“Communication is key” is said the way “eat healthy” is said — everyone agrees, almost nobody does it, and nobody specifies what it means. Talking more? Listening better? Hiding less?

What people rarely say is that communication requires a shared protocol. You have to agree, explicitly or implicitly, on how conflict will be handled, what counts as an accusation, how repair works. Couples who converge on a shared model thrive. Couples who don’t fall back into the pattern from Section 3: bundled statement, defense, spiral. They are not bad people. They are trying to communicate without agreeing on the rules of communication.

NVC’s value is not its specific techniques. It is that it provides a complete protocol — a shared paradigm two people can adopt. It tells you what a well-formed emotional expression looks like, what counts as a fair response, how to repair. Two people practicing NVC are not just communicating better. They are communicating inside the same paradigm.

This is not the only possible paradigm. But the paradigm must be shared.

Not all frameworks are compatible

Here the radical subjectivism trap re-emerges. It is tempting to say: you have your style, I have mine, we respect each other’s. This sounds enlightened. It is a recipe for drift.

If you have one framework and your partner has another, and they are incompatible, no amount of goodwill bridges the gap. You are speaking different protocols. The noise will be constant.

Science faced this problem. When two theories make contradictory predictions, they cannot both be true. The community must choose. The same is true in relationships. If one partner believes emotional expressions should be accepted without scrutiny and the other believes they carry falsifiable causal claims, those frameworks are incompatible. One must give way, or the relationship gives way.

This is structural, not personal. A physicist cannot accept Newtonian and quantum mechanics as complete descriptions of the same phenomenon. You can use each in its proper domain, but you cannot use both to answer the same question.

The practical implication: couples need a foundations conversation. Not mid-fight. But early and deliberately: how do we handle disagreement? What do we owe each other when one of us is upset? What counts as fair? Meta-communication — communication about how we communicate — is arguably more important than any specific communication that follows.

Some styles are better than others

I am not saying NVC is the only valid framework. I am saying some frameworks are better than others, and the criteria are not mysterious.

A good framework separates observations from judgments, encourages ownership of emotions, makes vulnerability safe, provides a repair path, and survives imperfection. A bad one bundles observation and judgment, attributes emotional states to the other person, punishes vulnerability, has no repair mechanism, and collapses the first time someone makes a mistake.

These criteria are derived from the structure of the problem. Communication is a repeated game; a good protocol makes cooperation easy and defection obvious. Frameworks that punish cooperation or reward defection will fail — regardless of how much the participants love each other.

What effectiveness without fairness looks like

A framework can be effective without being good. We have to acknowledge those frameworks, because doing so prevents us from proposing replacements that are morally superior but practically useless.

Consider the traditional patriarchal marriage. Our grandparents had fifty-year marriages. People ask them for advice. The longevity was not magic. It was a shared paradigm: the woman stayed home, obeyed, swallowed ambitions; the man worked, provided, bottled up emotions, never complained. The rules were unambiguous. There was no negotiation about emotional authority. The framework was brutally effective at resolving conflict. We rejected it because its costs were unacceptable — removing agency from one party, emotional range from the other — not because it was ineffective.

Consider “Happy Wife, Happy Life.” The man concedes complete emotional authority to the woman. Her satisfaction is the metric. It resolves the ambiguity about emotional responsibility: she does not have to own her feelings, he does not have to defend his. Effective. But his interior life becomes invisible. He becomes a function of her happiness.

Consider the couple that sweeps conflict under the rug. Disagreements are not addressed. Things are “fine.” This works — for years, sometimes decades — because both have silently agreed not to probe. The shared paradigm is avoidance, and it is stable precisely because it is shared. It breaks only when one party stops avoiding. I would not recommend it, but I cannot deny it is effective while it holds.

All of these work because they are shared paradigms. Predictability — not love, not vulnerability — resolves the repeated game. The problem is not that these frameworks fail. The problem is the trade.

The cost criterion

A good communication framework is effective at low cost. Cost is measured in agency, emotional range, honesty, and fairness. A framework that works by silencing someone or suppressing a dimension of human experience is not good. It is merely stable.

NVC satisfies both conditions: it resolves the repeated game, and it does so without requiring either party to surrender agency or emotional range. Both people express vulnerability. Both people have needs. Both people make requests and can say no. It demands discipline — not submission.

The patriarchal husband was disciplined. But his discipline served an unfair structure. NVC’s discipline serves one designed to be fair.

This is the standard: prefer a framework that is both effective at resolving conflict and fair to both parties. Effectiveness without fairness is tyranny. Fairness without effectiveness is a therapy session. Neither sustains a relationship.

If you find a framework that satisfies these criteria better than NVC, use it. The door is open.

9. What This Requires

If you accept the argument so far, the practical implications are demanding.

For the speaker: Separate observation from judgment. Own your feelings. Avoid language that assigns causation. “I felt hurt when that happened” is honest expression; “you hurt me” is a verdict.

For the listener: Extend charity. Resist correcting the facts before hearing the feeling. Treat the other person’s emotional report as valid even when the causal attribution is wrong. Empathy before education. The attribution can be discussed later.

For both: Build trust over repeated interactions. Cooperate from the start. Trust that the speaker will be careful with their attributions and the listener will hear the feeling behind the words. This is not a one-time agreement. It is a practice.

None of this is easy. My last relationship failed in part because, despite my efforts, we could not establish this equilibrium. My partner could not consistently own her emotional reactions, and I could not consistently absorb the misattributions without eventually defending myself. The spiral was cumulative — slow, quiet, and, in the end, structural.

10. Language Is All We Have

There is no escape from language. Every word carries history. Every sentence bundles description and judgment. Every conversation is mutual interpretation under uncertainty.

The question is not whether we can make language perfect. We cannot. The question is whether we can be honest about its imperfections — and build relationships, both geopolitical and personal, that account for them.

Saadawi was right to be suspicious of the language she inherited. “Who named this? For whose benefit? What does the name assume?” Her grandmother asked these same questions, barefoot and illiterate, facing down a mayor with a gilded Quran: “Who told you God is a book?” The error is not in asking. It is in answering too quickly and closing the door behind you.

The questions are worth asking. The frameworks — NVC, the repeated-game model, the philosophy of facts — are just attempts to answer them well.

If you take one thing from this article, let it be this: the next time someone says something that feels like an accusation, or the next time you feel yourself about to say one, pause. Ask what is observation and what is judgment. Ask what is feeling and what is attribution. Ask what framework you are both operating inside — and whether it is one you chose, or one you inherited.

That is the work. And it is harder than it sounds. But the alternative — letting language use us rather than the other way around — is far more costly.