Communication exercises for couples, and the 21 minutes that beat them

The best-evidenced thing anyone has found for an ordinary struggling couple takes twenty-one minutes spread over a year, is done alone, is written rather than spoken, and is never shown to your partner. It is the opposite of every exercise on every page about this.

The genre is settled. Mirroring, or say-it-back. “I” statements. The speaker-listener technique with an object you hold to show whose turn it is. A weekly check-in. No phones at dinner. Every one of them is a thing you do out loud, together, in front of the person you are currently annoyed with.

It is worth knowing that the centrepiece of all of it has been tested, and it did not go the way the pages suggest.

What happened when somebody looked for active listening

In 1998 Gottman and colleagues published a study of 130 newlywed couples recruited around Puget Sound, followed for between three and six years. In the lab, each couple talked about their own disagreement for fifteen minutes while video, heart rate and a rating dial ran. Seventeen of the couples later divorced.

The paper tested seven different models of what distinguishes couples who last from couples who do not. Its own abstract:

Support was not found for the models of anger as a dangerous emotion, active listening, or negative affect reciprocity.

Gottman, Coan, Carrère & Swanson (1998), Journal of Marriage and the Family

The reason is more interesting than the result. They went looking for the behaviour and it was barely there. Paraphrase-and-validate almost never happened in real conflict, so they widened the definition — any moment where one partner expressed something negative and the other responded with interest, affection, humour or validation. Even on that generous reading it came to a handful of seconds in a fifteen-minute conversation, including among the happiest couples in the sample. Something that rare cannot be what separates the groups.

Two pieces of fairness, because this is usually reported as more damning than it is. Gottman was measuring what untrained couples spontaneously do, not whether teaching it helps — and teaching it does produce measurable change, with modest effects on how couples communicate and smaller ones on how satisfied they are. What cannot be defended is the claim the pages actually make: that paraphrasing is the foundation, the thing a good relationship is built from.

Why it collapses in the actual argument

The best sentence on this is not Gottman’s. It is a therapist he quotes inside the same paper, Daniel Wile:

It is impossible to make ‘I-statements’ when you are in the ‘hating-my-partner, wanting-revenge, feeling-stung-and-needing-to-sting-back’ state of mind. At such a moment you cannot remember what an ‘I-statement’ is, and, frankly, you do not care.

Wile (1995), quoted in Gottman et al. (1998)

Gottman’s own comment on it is the quiet demolition: these techniques are unnecessary when things are going well. The model asks, in his phrase, for a form of emotional gymnastics from people who at that moment are somewhat emotionally disabled by conflict.

The design flaw, in the author’s own words

And then, in the same paper, an observation that has almost nothing to do with statistics and is the most useful thing in this entire subject.

Active listening did not come from couples work. It was lifted out of client-centred therapy — where the person doing the empathic listening is a third party, and the client is usually complaining about somebody who is not in the room. The therapist empathises with the client against an absent third person. That is a comfortable position to occupy, and it is why it works.

Yet in the marital arena, the spouse, even though the target of these complaints, is expected to empathize with her or his partner.

Gottman, Coan, Carrère & Swanson (1998)

He calls it, accurately, a more confrontational model than the one it was copied from. All six of the standard exercises inherit that shape.

And the assumption underneath all of it

Every page on this subject rests on something nobody states: that how you talk to each other causes how you feel about each other. In 2016 three researchers tested that directly, observing newlywed couples four times over three years and comparing their communication with their own satisfaction reports.

Reliable communication-to-satisfaction and satisfaction-to-communication associations were identified, yet neither pathway was particularly robust. These findings raise important doubts about theories and interventions that prioritize couple communication skills as the key predictor of relationship satisfaction.

Lavner, Karney & Bradbury (2016), Journal of Marriage and Family

Both arrows exist and both are weak. Which allows for a possibility the genre cannot afford: that talking well is as much a symptom of a relationship that is going well as a cause of one — and that the exercises may be teaching the symptom.

The twenty-one minutes

Here is the study that ought to be at the top of every page about this and appears on none of them.

A hundred and twenty married couples in the Chicago area, married on average eleven years, were followed for two years and checked in with every four months. For the whole of the first year, everybody did the same thing: describe your most significant disagreement of the last four months, sticking to behaviour rather than feelings, and rate how distressed you are about it.

Then half of them, chosen at random, were given an extra seven-minute writing task at three of those check-ins — months twelve, sixteen and twenty. Three sessions. Twenty-one minutes in a year. Private, written, never shown to the other person. The prompts were:

1. Think about that disagreement from the perspective of a neutral third party who wants the best for all involved — a person who sees it from a neutral point of view. How might they think about it? How might they find the good that could come from it?
2. Most people find that perspective hard to hold during an actual disagreement. What gets in the way of it for you?
3. Over the next four months, try to take that perspective anyway. How might you manage it? How might it help?

The mechanism, as far as the study could trace it, is not that they argued less. It is that they were less distressed about the arguing — and that change in distress is what carried the change in quality. Twenty-one minutes, alone, in writing, about a fight that had already happened.

So what should you actually do

Four things, chosen because the evidence behind them is real and because between them they cover the parts of a relationship the exercise genre ignores.

Where inbetween fits

We did not design this app from that 1998 paper — we found it afterwards. But it describes what we built more precisely than anything we had written ourselves.

The exercises are a three-person structure with the third person removed. inbetween puts one back. You talk to it, they talk to it, and it is the only one who hears both sides, so the empathic listening is not being demanded of the person who is currently the target. Nothing you say crosses to them unless you allow it, which means the version you write while furious can exist without being delivered — and the seven-minute private write-up, which is the intervention with the best evidence on this page, is a thing you were going to have to do alone in a notebook anyway.

The honest limits. It will not calm you down in the middle of a fight — nothing will, and the flooding finding says so. The best trials of guided self-help for couples all included calls with a human coach, and we do not have those. And if the relationship is in real trouble, the right way to find a therapist is a more useful page than this one.

The thirty-six questions are the gentler version of the same mechanic, and the games are for an evening that does not need to be heavy.