The short version

Parents measure their own consistency. They count whether they spoke Japanese at dinner, whether they let an English answer slide. Meanwhile the household’s real language economy is being decided by two children in a bedroom, and that negotiation is not about Japanese at all. It is about speed, jokes, and who gets to be funny.

Once two siblings establish English as the language of their relationship, that choice is extremely stable. It is not a habit in the way that brushing teeth is a habit. It is the medium of a relationship, and changing it feels, to them, like being asked to be a different person with their own brother or sister. This is why “speak Japanese to each other” produces a few stilted sentences and then silence, and why the pair reverts the moment you leave the room.

The good news is that sibling language is not fixed by willpower on either side. It is decided by who else is in the conversation. Add a person the pair cannot use English with — reliably, repeatedly — and the pair’s own language shifts, because the shared activity now runs in Japanese. That is the whole mechanism, and almost everything useful in this guide is a variation on it.

Why the sibling pair outweighs you

Do the arithmetic once and the problem stops being mysterious.

A school-age child is awake for roughly thirteen hours. Six or seven of those are at school, in the majority language. Of the hours left, a meaningful share is homework, screens, transport and the bits of the evening where nobody is talking much. Your actual conversational contact — the minutes in which you and your child are genuinely exchanging sentences — is often under an hour, spread thin across a tired evening.

Two siblings in the same house, on the other hand, are in each other’s company for most of the non-school day and talk continuously while doing it: negotiating a game, arguing about a turn, narrating a video, teasing. For most families with two or more children, the sibling conversation is the single largest block of spoken language in the child’s day after school itself.

This also explains a pattern that confuses a lot of parents: the family where the adults are rigorously consistent in Japanese and the children still stop speaking it. Nothing went wrong with the parents. They were simply outvoted by volume.

What the research actually found

The clearest study on this is not about attrition at all. Dimitra Tsinivits and Sharon Unsworth looked at Greek–Dutch bilingual toddlers in the Netherlands and compared those with an older, school-age sibling against first-borns.

The findings are worth stating precisely, because they are often exaggerated.

Two things follow, and the second is the one that matters for you.

First, the older sibling changes the home, not the parents. Parental input was statistically the same in both groups. What differed was how much of the majority language the household contained — because an older child brings school home every afternoon and speaks it to the baby. If you have felt that something shifted in your house without you changing anything, this is it.

Second, and importantly: this study found no heritage-language damage at toddler age. It is not evidence that siblings destroy the home language. The honest reading is that the sibling effect first shows up as a rise in the majority language, and whether the heritage language then declines depends on what happens over the following years — which is where the rest of this guide lives. Anyone telling you that research proves siblings kill the home language is overstating it.

Why the second child usually has weaker Japanese

Almost every family with two or more children abroad reports the same thing: the eldest speaks the most Japanese, and each subsequent child speaks a little less. This is not favouritism and it is not imagination. It falls out of the arithmetic above.

The practical consequence: if you are expecting a second child, or your first has just started school, you are at the highest-leverage moment you will ever have. Setting the sibling language before the pair has settled is far easier than changing it afterwards. If that moment has passed, the rest of this guide is about the harder version.

The moment it flips

Sibling language does not drift. It switches, usually inside a few months, and almost always around one of four events.

What makes the switch so durable is that it is not stored as a rule. It is stored as a relationship. Two siblings have a particular register with each other — the shorthand, the running jokes, the specific way they wind each other up — and all of it is built in whichever language they settled on. Asking them to speak Japanese together is not asking for a translation. It is asking them to be slower, less funny, and less themselves with the person they are least formal with in the world.

That is why the usual intervention fails so reliably. “Speak Japanese to each other” gets you a few flat exchanges performed for your benefit, followed by a drop in how much they talk at all, followed by a return to English the moment the supervision ends. You have not changed the language of the relationship. You have added a tax to being in the same room as you.

What siblings get from the majority language — and why Japanese cannot compete on those terms

To change the pair, it helps to be honest about what they are choosing. Three things, none of which are about language learning.

Speed. Children’s play is fast. Arguments about rules, narration of a game, the exact timing of a joke — all of it collapses if either party has to search for a word. Whichever language is quicker for both of them wins, automatically and permanently. This is why the pair tends to settle on the weaker child’s stronger language.

Vocabulary for the actual content. If what they do together is a game, the words for it are in the game’s language. A child who does not know how to say “respawn”, “shuffle the deck” or “you’re out” in Japanese is not going to invent them mid-play. The content drags the language with it.

Equality. This is the subtle one. In English, the siblings are matched — both fluent, both able to argue. In Japanese, one of them is usually noticeably better, and the gap turns play into a lesson with a built-in winner. Children avoid activities where one of them is permanently the slow one, and they are right to.

What does not work

Families arrive at the same five interventions. Here is an honest ranking, including why the popular ones backfire.

One of these deserves defending. Japanese screen time together is better than its reputation, as long as you are clear about what it does. It will not make your children speak Japanese to each other. What it does is give the pair a shared world with Japanese words attached — characters, catchphrases, the names of things — and shared reference is exactly what a sibling register is made of. Children who have watched the same Japanese series together have something to be funny about in Japanese, which is the raw material everything else needs.

What does work: the third-person principle

Here is the mechanism, stated plainly.

A sibling pair changes language when a third person joins their activity and cannot be spoken to in the majority language. Not a parent — the children know you understand English. Someone outside the family, ideally another child, whose presence makes English useless rather than forbidden.

What happens next is the part parents find surprising. The siblings do not just speak Japanese to the newcomer. They start speaking Japanese to each other, within that activity, because switching back and forth mid-conversation is more effort than staying in one language. The pair’s language is set by the group it is embedded in, and you have changed the group.

Every family who has spent a summer in Japan has seen the strong version of this. Two siblings who speak only English at home arrive at their cousins’ house, and within about ten days are arguing with each other in Japanese without noticing. Nobody made a rule. The group changed, so the pair changed.

This is also why one-to-one tutoring, which helps individual children a great deal with reading and grammar, does almost nothing for sibling language. A tutor works with one child, in a session the sibling is not part of, with an adult who could switch to English if needed. None of the three conditions are met.

Age by age: what actually shifts a pair

The third-person principle is the same at every age. What changes is what counts as an activity worth joining.

The 10–13 row surprises people, so it is worth spelling out. Parents instinctively look for something the children can do together in Japanese. But by this age, a sibling pair with a Japanese gap will reproduce that gap in any shared activity — the stronger one carries the conversation, the weaker one goes quiet, and both conclude that Japanese is where they are mismatched. Giving each child a separate Japanese group, at their own level, removes the comparison. When both come home having had a good time in Japanese, the pair sometimes starts using it between themselves for the first time in years.

When one child speaks Japanese and the other does not

This is the most common version of the problem, and it needs a different answer from the general case.

Typically the eldest, who had the pre-school Japanese window, is reasonably comfortable, and a younger sibling born into a two-language house is not. Parents then do the natural thing: they ask the fluent one to speak Japanese to the less fluent one. It is a reasonable idea and it tends to make everything worse, for a reason worth understanding.

For the older child, speaking Japanese to a sibling who struggles is unrewarding: they have to slow down, simplify, repeat, and they get a worse conversation for the effort. For the younger child, every exchange is a reminder of the gap. The relationship picks the language where they are equals, and it is right to.

What helps instead:

What it costs if the pair never changes

It is worth being concrete about why this is worth any effort at all, because “bilingualism is good for them” is too vague to motivate anybody through a difficult Tuesday.

The research on what happens to families, not just to languages. Lily Wong Fillmore surveyed around a thousand language-minority families in the United States — 690 in the main sample and 311 in a comparison group — about what happened when their children were schooled in English only. As reported at the time, about 51% of parents in the main sample said their home language was being displaced by English, against just over 10% in the comparison group whose children were in primary-language programmes. Around 12% said their children now spoke the home language poorly or not at all, against 1.6% in the comparison group.

Her point was not about vocabulary. It was that when children lose the home language, families lose the ability to talk to each other about anything that matters — which is the thing parents in this situation are actually afraid of, usually without saying so.

And if you ever move to Japan. The Japanese government’s figures make the practical cost measurable. As of 1 May 2025, 84,759 children in Japanese public schools needed Japanese-language instruction, and 11,446 of them were Japanese nationals — overwhelmingly children who grew up overseas and returned. Among students needing that instruction, the rate of going on to university and similar institutions was 41.2%, against 75.0% for students overall. A sibling pair that kept speaking Japanese to each other arrives in a Japanese classroom with a working language. One that did not, arrives with a project.

If a move is realistic for your family, our guides on Japanese public elementary school for foreign children and what returnee children struggle with cover what happens after you land. And if the broader pattern in this article sounds familiar — children who understand everything and answer in English — the companion guide to why children stop speaking Japanese and what brings it back covers the individual version of this problem.

Two related situations have their own guides: if only one parent in your household speaks Japanese, that configuration is the hardest in the research and has its own levers; and if your children can no longer really talk to their grandparents, the grandparent guide covers the call that most families are running wrongly. Everything sits in our Keeping Japanese Alive hub.

How to run a Japanese-only activity at home that survives past week two

Most families cannot buy their way to a third person immediately, and the home version is worth doing anyway. The reason home attempts collapse is almost always the same: the activity was really a lesson wearing a costume, and children can tell within about four minutes.

What survives has four properties.

Concretely, the things that tend to last: cooking a Japanese recipe where the instructions are in Japanese and dinner depends on getting it right; a card or board game whose rules you only ever explain in Japanese; a standing video call with a grandparent or cousin who needs something from the children rather than testing them; building something from Japanese instructions. The things that tend to die: flashcards, reading practice, writing practice framed as a shared activity, and anything with a progress chart on the fridge.

One more rule, and it is the one parents find hardest. Do not use the activity to check progress. The moment a child senses that the cooking is really an assessment, it becomes an assessment. If you want to know whether their Japanese is improving, find out some other way — not by turning the one enjoyable Japanese thing in the week into a test.

What to say when you hear English between them

You will hear it several times a day, and what you do in that two-second window either costs you something or quietly builds something. Six common moments:

The pattern across all six is that you are trying to stay out of the sibling relationship while keeping your own channel wide open. Parents who police the pair lose both. Parents who leave the pair alone and talk a great deal themselves usually find the pair becomes reachable later, when the group around it changes.

Is a full switch back realistic?

Usually not, and it is worth saying so plainly rather than letting you measure yourself against something that rarely happens.

Families who do everything in this guide, consistently, for a year generally do not end up with two children who speak Japanese to each other all the time. What they end up with is a pair that is mixed: Japanese in certain contexts — around certain people, during certain activities, on calls to Japan, on visits — and the majority language the rest of the time. That is a real and durable outcome, and it is enough.

It is enough because of what it protects. A child who can switch into Japanese with their sibling when the situation calls for it has kept the machinery working: the retrieval speed, the willingness to be imperfect, the sense that Japanese is a thing their family does together rather than a subject their mother cares about. That child can rebuild quickly at fifteen, or in a summer in Japan, or in a Japanese classroom. A pair that has gone entirely to English for a decade has to start from a much colder place — and the research on language attrition is consistent that the longer the gap, the more the re-entry feels like failure rather than return.

So set the target honestly: not a flip, but a pair that still has Japanese available to it. Measure against your own baseline from day one, not against a cousin in Tokyo, and treat any unprompted Japanese between your children as evidence that the machinery is still there.

Where an online Japanese school fits — and where it does not

The hard part of the third-person principle is supply. If you live in Leeds, Lyon or Lagos, there may be no Japanese-speaking children within an hour of your house, and the cousins are nine time zones away.

TOMODACHI JAPAN (ともだちじゃぱん) is an online Japanese school built by NIJIN for children growing up abroad, and the parts of its design that matter here are specific:

Who this is not right for

• If your children already speak Japanese to each other. Do not fix what is working. Protect the conditions that produced it.
• If the problem is reading and writing rather than speaking. A hoshuko or a literacy tutor is a closer fit than conversation classes.
• If one child needs an exam — GCSE, HSC, VCE, AP Japanese — use a specialist for that child and treat the sibling question separately.
• If you only have one child. None of this applies; the individual guide linked above is the relevant one.
• If you need something this month. It does not open until December 2026.

A 60-day plan for the sibling pair

Expect the first sign of change to be small and odd: a single Japanese word dropped into an English sentence between them, usually a word from the thing they watched together. That is not a token gesture. That is the register starting to move, and it is the thing to notice out loud — or better, to notice and say nothing about at all.

Frequently asked questions

Sources used for this article

The third person your children cannot speak English to

A sibling pair changes language when the group around it changes. TOMODACHI JAPAN puts each of your children in a class of eight at their own level, with children in Japan who do not speak English — which is the one condition that moves a pair. The pilot opens in December 2026 and the school guide is free.

Visit TOMODACHI JAPAN