The short answer, in five lines

An only child does not lose Japanese the way a sibling pair loses it

If you have read our piece on siblings who speak English to each other, you know the mechanism that does most of the damage in households with two or more children. Once two siblings settle on English, English becomes the working language of the child layer of the household — the layer where play, conflict, jokes and secrets happen — and the parent has no way inside it. The pair outweighs the parent, because the pair is together far more often and because what they do together is more interesting than anything a parent is offering. In a sibling household, the single biggest determinant of whether Japanese survives is usually not what the parents do. It is what the children do to each other.

Your household does not have that layer. There is no dyad to flip. Nobody is standing next to your child all day modelling the idea that Japanese is the language you use with the grown-ups and English is the language of real life. This is a genuine structural advantage, and it shows up in exactly the way you would predict: parents of only children quite often report that their eight- or nine-year-old still answers in Japanese without being asked, at an age when a comparable sibling household has already gone quiet. If that is your experience, you are not imagining it, and it is not a reward for better parenting. It is the absence of a competing force.

But the same structure produces a different shortfall, and it is harder to see because it does not look like loss. Your child’s Japanese is not thinning out. It is staying exactly where it was built — inside one relationship, about one set of topics, with one listener. In a sibling household you can hear the problem, because the Japanese stops. In a one-child household, the Japanese keeps going, and what is missing is everything the child has never had to say.

What research actually says predicts output — and it is not only hours

The study most worth knowing here is Place and Hoff’s 2011 paper in Child Development. Twenty-nine bilingual two-year-olds, mothers keeping diary records of exactly who spoke which language to the child and for how long, with vocabulary and grammar measured by parent inventory. The headline is unsurprising: the relative amount of exposure predicted outcomes in both languages. Amount matters, and nothing in this article says otherwise.

The part worth sitting with is what came next. Two further properties of the input were, in the authors’ words, unique sources of variance in the children’s skills: the number of different speakers from whom the child heard the language, and the percentage of that input which came from native speakers. “Unique” is the load-bearing word. It means those two things were not simply standing in for hours. Two children with identical exposure totals, one hearing the language from four people and one hearing it from one, were not in the same position.

Be clear about the limits before you lean on it. The sample was twenty-nine children; they were toddlers; the language that gained from speaker variety in that dataset was the community language, not a heritage language being maintained against the pull of a majority one. So this is a mechanism to reason with, not a measured effect on Japanese maintenance abroad. But as a mechanism it is very easy to believe, because of what a single speaker actually is.

One speaker is one idiolect. One vocabulary range, shaped by what that person habitually talks about. One set of sentence frames. One speed, one accent, one level of patience. And — this is the part parents of only children almost never consider — one listener who already knows what the child means. You can decode your own child’s half-finished Japanese from two syllables and a gesture. You fill in the particle they dropped. You answer the question they were trying to ask rather than the one they managed to ask. Every one of those acts of love removes the one thing that makes a child’s language grow under pressure: the need to say it again, differently, until it works. A child with one Japanese interlocutor is a child who has never had to repair a sentence.

Child speech is its own input type, and your household has none of it

For decades the research on language input was research on adults talking to children. In 2025, Johanna Schick and Sabine Stoll published an analysis in Open Mind that went somewhere the field had mostly not looked: a community in the Peruvian Amazon where child-directed speech is far less common than it is in industrialised households, and where children grow up in large mixed-age groups with older children supervising younger ones. Twelve children, longitudinal naturalistic recordings, from the Shipibo-Konibo corpus.

Two findings came out of it. First, child-directed speech was still the best predictor of what the children produced — so talking to your child directly is not obsolete, and never was. Second, and this is the one that matters for a one-child household: across both the speech directed at children and the speech merely happening around them, the single input type that best predicted children’s production frequencies was speech from other children.

Again, the honest framing. Twelve children. A rural Amazonian community whose social structure is nothing like a flat in Frankfurt or a house in Seattle. Monolingual first-language acquisition, analysed at the level of word frequencies — not a study of heritage-language maintenance, and not a study of Japanese. What it establishes is narrower and still important: child speech is not a weaker substitute for adult speech. It is a separate source, with its own statistical shape, and that shape is closer to what children themselves produce.

Which is precisely the slot a one-child household contains none of. And it is worth being blunt about why you cannot fill it yourself. When you speak Japanese to your child you produce caregiver speech, because you are a caregiver. You simplify, you scaffold, you ask questions you already know the answer to, you grant turns. Even if you consciously tried to speak to your child the way another nine-year-old would, you would still be you — the person who is not going to take the toy, not going to say the plan is stupid, not going to stop listening. There is no amount of effort that converts a parent into a peer. This is the one gap in the environment that is structurally outside your reach, which is exactly why it is the one worth spending money and logistics on.

The register map: four kinds of Japanese, and the one you can actually supply

“Speaks Japanese” is not one skill, and this is where the one-child pattern becomes legible. Think of four different registers, each learned in a different kind of relationship.

Put your own household on that map and the shape of the problem appears immediately. An only child abroad with one Japanese-speaking parent typically has register 1 at or near age level, register 2 at close to zero, register 3 limited to whatever happens on family visits, and register 4 wherever you have had the energy to take it. If you have been worried and have responded by speaking more Japanese at home, you have been pouring the one resource you have into the only column that was already full.

And notice what this does to how the child looks. A child with strong register 1 and no register 2 sounds fluent to you and strange to a Japanese child. They talk like someone’s child, because that is the only role they have ever had in Japanese. If you want a single sentence for the whole problem: your child has learned to be a son or daughter in Japanese, and nothing else.

What it looks like at eight, at ten, and at thirteen

The same deficit presents differently as a child gets older, and the later it is identified the more it costs to address.

At eight, it looks like shyness, and almost every parent misreads it as shyness. The child is comfortable with you, manages five minutes on a video call with a grandparent, and then meets a Japanese child at a park or a wedding and goes silent. Asked what happened, they say they were embarrassed. What actually happened is that someone said something like 何して遊ぶ？ and they had no frame to answer with, because nobody has ever proposed a game to them in Japanese. It is not nerves. It is a blank where the sentence should be.

At ten, you can see it in the vocabulary, if you know where to look. The child can tell you お風呂入って and ask what is for dinner, and cannot produce the lexicon of negotiation — ずるい, 順番, 貸して, 先にやってた, ちょっと待って, 入れて. These are not advanced words. They are the first hundred words of being a child among children, and a child who has only ever been a child among adults does not have them. Parents frequently mistake this for a general vocabulary gap and buy a vocabulary book, which does not help, because the words were never missing from a list. They were missing from a relationship.

At thirteen, the narrowness itself becomes the reason to stop. We have written separately about what changes at that age and why refusal splits into three kinds with opposite remedies — see teenager refuses to speak Japanese. The only child’s version is almost always the capability-shaped one. They can speak Japanese to exactly one person, so every new Japanese situation is an occasion to discover a limit in public. A thirteen-year-old will avoid that, and will describe the avoidance as not being interested. If you are hearing “I don’t want to” from a child whose only Japanese relationship is with you, read it as “I can only do this with you” before you read it as a loss of motivation.

One more thing to watch, which cuts against the usual advice. Because an only child keeps the Japanese frame of their sentences longer, their mixing of Japanese and English tends to look like the benign kind — English nouns inside a Japanese sentence with the particles still attached. That reading is correct as far as it goes. It is also not reassuring about the thing this article is about, because range and frame are different measurements. A child can have an intact Japanese frame and still be unable to use it with anyone but you.

Why one-to-one tutoring is the wrong default here

When a parent of an only child decides to act, the first purchase is almost always a private tutor or a one-to-one online lesson. It is the most heavily marketed option, the easiest to book, and it feels like the serious choice. It is also, for this specific problem, close to the least effective thing you can buy.

Look at what it actually adds. It adds hours — which, as we have just established, is not the binding constraint in a household where the child already hears Japanese every day. It adds one speaker, which is real: going from one Japanese interlocutor to two is a meaningful change, and the research above says speaker count carries its own weight. And it adds register 3 and register 4 — teacher Japanese, the language of being taught. All of that has value.

What it does not add is register 2, in any quantity whatsoever. And it leaves the deeper problem untouched, because a tutor is an even more accommodating listener than you are. They are paid to understand. They are trained to simplify, to wait, to rephrase. They will never walk off mid-sentence because the child took too long. Every professional instinct a good tutor has is an instinct to remove exactly the communicative pressure that makes a child reach for a word they do not quite have.

There are situations where one-to-one is the right call, and we would be doing you a disservice not to name them. A child who needs to learn to decode kana and kanji needs instruction, not conversation, and a group will not deliver it — that is the territory of our piece on children who speak Japanese but cannot read or write it. A child preparing for a specific test needs targeted work. And a child whose speaking is far enough behind that a peer group would simply talk over them often needs a few months of one-to-one first, so that the group becomes survivable. If you are in any of those three cases, buy the tutor.

Outside those cases, hold the distinction in your head: a tutor is an excellent way to teach a child Japanese and a poor way to give a child someone to speak Japanese to. Those are not the same purchase, and the second one is what a one-child household is short of.

Count speakers, not minutes: the audit that takes one week

Before you buy, change or arrange anything, spend seven days measuring the thing that actually varies. This takes about two minutes a day and it is the most useful thing in this article.

For one week, keep a list of every human being who spoke Japanese to your child and had to be answered in Japanese. Four rules make the count honest:

Most one-child households abroad come out at one, occasionally one and a half, sometimes two if a grandparent calls weekly. Almost nobody comes out at three. And almost nobody has anything at all in the child-to-child column, which is exactly the finding the audit exists to produce.

Our working target — and we want to be explicit that this is our rule and not a research finding — is three distinct Japanese interlocutors in a normal week, of whom at least one is a child, and at least one cannot switch to English. Three is not magic. It is the smallest number at which a child stops having a single Japanese relationship and starts having a Japanese social world, however small.

The three slots you are trying to fill

Slot 1: finding a child who cannot switch

Here is the thing nobody tells you about the advice to “find a Japanese friend”: it usually does not work, and the reason is predictable. Put two bilingual children together abroad and they will find their shared strongest language within about ten minutes, and that language is almost always English. Two heritage speakers meeting at a community event are not a Japanese-speaking environment. They are two children who both find English easier, discovering that fact about each other. Parents do this, watch it fail, and conclude their child is resistant. Their child is being efficient.

What actually produces child-to-child Japanese is narrower than “a Japanese friend”, and it comes down to two conditions.

The other child must not have English available as an escape. That means a child living in Japan, or a child abroad whose English is genuinely weaker than their Japanese, or a setting where English is simply not on the table. This is why age-matching matters far less than parents assume, and why ability-matching matters far more. A seven-year-old in Japan and your ten-year-old abroad can have a real conversation, because the Japanese-only constraint is doing the work that a shared age would not.

And they need something to do, not something to discuss. “Say hello to Yuki” produces four seconds of Japanese and a long silence. A game with rules, a drawing they are both contributing to, a building task, a quiz, a shared goal — these produce negotiation, and negotiation is register 2. If you take one practical instruction from this article, make it this: never arrange a conversation. Arrange an activity and let the conversation be the by-product.

Ranked by how reliably they deliver those two conditions, the realistic options for a household with no local Japanese children are: a small online group class with children in Japan, grouped by what each child can say rather than by age; a recurring call with cousins in Japan that has a job attached to it; a local Japanese family with a younger child, because the younger child’s weaker English removes the escape route; and a supplementary school, if one is genuinely within reach.

Slot 2: an adult who is not you

This is the cheapest slot to fill and the one most often wasted. Most one-child households already have a candidate — a grandparent, an aunt, an uncle, a family friend in Japan — and the relationship produces almost no language because of how the calls are structured.

Three changes make a relative call into an actual speaking occasion.

Leave the room. As long as you are sitting beside your child, you are the interpreter, the prompter and the safety net, and the child knows it. They will look at you when a word is missing, and you will supply it, because you are their parent. The conversation is between your child and the other adult only when you are not in it. If your child is young enough that this feels abrupt, start by moving out of eye line rather than out of the house.

Give the call a job. “Tell Grandma about your week” is an exam question, and children answer exam questions in three words. “Grandma is going to teach you how to make something and you have to write down the steps” is a task, and tasks generate clarification, repetition and repair — which is the entire point. Teaching a card game, comparing weather, planning what to do on the next visit, reporting on a shared television programme: anything with an outcome.

Short and frequent beats long and rare. Fifteen minutes twice a week does far more than an hour once a month, because the child has to re-enter the relationship often enough for it to stay warm. We have written about the version of this that has gone wrong already — where a child and their grandparents can no longer hold a conversation — and the single biggest difference between households where that happened and households where it did not is frequency, not duration.

One honest caution. A bilingual relative who switches to English the moment the child hesitates is filling half a slot, and sometimes less. It is worth asking them directly, once, to stay in Japanese and to let silences last. Most relatives are relieved to be told, because they had assumed switching was the kind thing to do.

Slot 3: a situation where Japanese is the only way to get the thing

The third slot is the one parents try to create with rules, and rules are the wrong tool. A Japanese-only hour, a Japanese-only dinner, a sticker chart — we have looked at why these backfire in two other articles and will not repeat the argument here beyond the conclusion: a rule makes Japanese a condition the child has to satisfy for you, which converts the language into a chore with your face on it.

What works is a situation, and a situation has one defining property: if the child could get what they want by switching to English, it is not one. Apply that test and most of the usual suggestions fall away.

Situations that pass, in rough order of how easy they are to arrange from abroad:

And the test cuts the other way too. A Japanese-speaking parent who understands English perfectly cannot create a passing situation by pretending not to. Children see through it immediately, and what they learn is that Japanese is a game you play with your mother, which is the opposite of the lesson you wanted.

What does not work, and why parents of only children keep trying it

Every item on this list is well-intentioned, widely recommended, and will not move the thing this article is about. They share one feature, which is worth naming before the list rather than after: they all increase supply, and none of them increases the number of people who have to be answered.

If a supplementary school is within reach: what the numbers say about peer supply

A Japanese supplementary school is one of the few places abroad where register 2 happens without anyone engineering it, and if you can get to one, it is worth real inconvenience. It is also worth going in with accurate expectations about how many children will be there, because the arithmetic surprises most parents.

MEXT’s figures for overseas Japanese education, as at 15 April 2024, give 247 supplementary schools worldwide with 20,822 pupils at the compulsory-education stage, alongside 94 full Japanese schools with 16,134 pupils. Divide the first pair and you get roughly 84 pupils per supplementary school — spread across nine year groups, which works out at fewer than ten children per year group on average.

That number is the reason to calibrate. A supplementary school is a genuine source of Japanese-speaking peers, and it may also contain only a handful of children of your child’s age and ability, several of whom will speak English to each other in the corridor because they live in the same English-speaking country your child does. Both of those things are true at once. Go, if you can get there — and do not assume that attendance by itself fills slot 1.

Two further honest points. A supplementary school’s curriculum is weighted towards written Japanese and the national course of study, not towards conversation; it is a school, not a conversation club. And it costs a Saturday morning every week, which is a price that gets much harder to pay once a child hits adolescence and has their own commitments. Neither is a reason not to go. Both are reasons to be clear about what you are buying.

Age by age: what to prioritise

Preschool to seven. Speaker count is the whole game, and this is the cheapest window you will ever have, because a child this age will talk to almost anyone without self-consciousness. Prioritise a slightly younger Japanese-speaking playmate if one exists anywhere in your orbit, and set up recurring short calls with relatives now, while the child has no opinion about it. Do not worry about register 2 being sophisticated; worry about it existing.

Eight to ten. The target is register 2 specifically, and this is where group settings beat one-to-one decisively. Introduce a task-based peer setting and expect the first three sessions to be quiet. A child whose entire Japanese life has been one relationship needs several sessions to learn that a different kind of Japanese conversation is even possible. Parents who pull out after two sessions because “it is not working” are, almost always, stopping just before the thing starts.

Eleven to thirteen. Ownership becomes the issue, and the mechanics change enough that we treat it separately in the article on adolescence. The short version for a one-child household: the child has to choose the group, and interest-led beats language-led. Something taught in Japanese about a thing they already care about will survive; something framed as Japanese practice will not.

Fourteen and over. The realistic goal shifts from fluency to leaving the door openable, and for an only child that has a specific meaning. It means ensuring there is at least one Japanese relationship in their life that is theirs and not yours — a friend, a cousin they actually message, a teacher they liked. A young adult with Japanese that only works with a parent will put it down when they leave home. A young adult with one Japanese friendship of their own will not.

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

For a household with no local Japanese children, an online group class is the only option on the list that reliably fills slot 1, and it is worth comparing it honestly against the two alternatives rather than simply recommending it.

Read the highlighted column honestly. There is no playground in an online class, no walk home, no incidental talk in a corridor — and corridor talk is a large share of where register 2 actually gets built in a physical school. If you have a reachable supplementary school, it gives you something an online class cannot. An online class also depends on a fixed weekly slot surviving the family calendar and on a time difference you can live with, and a child placed in a group well above their speaking level will go quiet rather than talk, which is why grouping by what a child can actually say matters more than grouping by age.

What it does, that nothing else on the list does for a family with no local Japanese community, is put a child in a room with other children who do not have English available. That is the one slot a parent cannot fill, and it is the reason this option exists at all.

Who this is not right for

Four situations where the advice above is the wrong priority, and pushing on it would cost you time you do not have.

What it costs if the range never widens

We are deliberately not going to tell you that something terrible happens, because we could not find research that supports a specific claim about only children and heritage-language outcomes, and we are not going to invent one. What we can show you is two concrete consequences, one measurable and one not.

The measurable one concerns returning to Japan. MEXT reports 84,759 children in Japanese schools requiring Japanese-language instruction, of whom 11,446 hold Japanese nationality — a group that includes returnee children. For students requiring that support, progression to university or equivalent runs at 41.2 per cent against 75.0 per cent for all students. Those figures describe children already in Japanese schools who have been identified as needing support; they are not a prediction about your child, and a child with strong home Japanese is in a very different starting position. But they are the clearest public indication of what the Japanese school system does and does not absorb easily, and the register gap this article describes is part of what gets identified. If a return is on the table at all, our piece on moving back to Japan covers how to spend the time before it.

The unmeasurable one is the one parents actually mind. Register 2 — negotiating, disagreeing, joking, backing down — is the register adult friendship runs on. It does not get replaced later by a more advanced register; it is the foundation the later ones sit on. A person who can speak Japanese to their mother and to nobody else has a language with a population of one. Most of the parents we hear from are not worried about test scores. They are worried that their child will stand in a room full of relatives and be unable to join in, and that is a reasonable thing to be worried about, because it is exactly what the one-speaker pattern produces.

A ninety-day plan for a one-child household

What to measure at the end, in order of usefulness: how many people had to be answered in Japanese this week; whether any of them was a child; whether your child initiated a Japanese conversation with somebody other than you even once; and whether you heard them repair a sentence — try again, differently, because the first attempt did not land. That last one is the single best sign that something real has changed, and it is the one you will never hear at home, because at home you understood them the first time.

Frequently asked questions

Sources used for this article

If you want the rest of this series, the hub page for keeping Japanese alive while living abroad collects every article in it, including the ones on children who understand Japanese but will not speak it and on households where only one parent speaks Japanese.

A second person, and then a third

An only child with no Japanese-speaking peers does not need more hours. They need somebody other than you who has to be answered in Japanese. TOMODACHI JAPAN puts children growing up abroad into small online classes with peers in Japan who do not speak English, grouped by what each child can actually say rather than by age. The school guide is free to read.

Visit TOMODACHI JAPAN