Kateryna Ivanova
It happens at almost exactly the same point for every family. Somewhere around week six, a mother messages me a version of this: "She had about thirty English words. She used to say 'apple', 'shoes', 'more', 'open'. This week I asked her what it was and she just looked at me. It's like the last two months didn't happen."
And then, almost always, comes the conclusion she has already drawn: the method isn't working, my English isn't good enough, she needs a real teacher.
All three of those conclusions are wrong, and the reason is one of the most useful things I can teach a mother in her first year of doing English at home. Toddlers very rarely forget English words. What actually happens is that three completely different things get mistaken for forgetting — and because they look identical from the outside, mothers apply the wrong fix to all three. The most common wrong fix, repeating the word more times, makes two of the three worse.
This guide separates them. You will get a four-question test you can run in your kitchen this afternoon to find out which one you are dealing with, the memory science that explains each, and a specific repair for each — including the rule that changes home English more than any other single thing I teach: a word needs five places, not fifty repetitions.
Quick answer
A toddler who stops saying English words she used to say has almost certainly not lost them. Three different things look like forgetting: (1) the word never encoded properly — she heard it many times but never used it herself; (2) the word is stored with only one retrieval cue — it exists only in the room, ritual or object where it was learned; or (3) the word is stored but loses the race to a stronger, better-practised word in your home language. Only the first is a learning problem. The other two are access problems, and they are fixed by changing where and how you ask — not by drilling the word again. The single most reliable repair is to use the same word in five different contexts rather than repeating it fifty times in one.
What "forgetting" actually means at ages 1–5
When an adult says "I forgot the word", they usually mean it is gone. When a two-year-old appears to forget an English word, something much more specific is going on, and it helps enormously to know that memory has three separate stages that can each fail independently.
Encoding is how a word gets in. Storage is whether it stays. Retrieval is whether your child can get it back out at the moment you ask. Mothers almost always assume the problem is storage — that the word has evaporated. In four years of watching this pattern in families, storage is almost never the problem. It is nearly always encoding or retrieval.
There is a second distinction that matters even more, and it is the one that causes the most unnecessary panic. Recognition and recall are not the same skill, and recognition comes first by a wide margin. Your child can know a word perfectly well in the sense that she will turn her head, point, fetch the object, or laugh at the right moment — and still be unable to produce the sound herself on demand. That gap is not a defect. It is the normal architecture of early vocabulary, and it is why "she can't say it" is very weak evidence for "she doesn't know it".
| What failed | What you see at home | Is the word lost? | What actually fixes it |
|---|---|---|---|
| Encoding — it never got in | Blank look; no reaction even to the object; never used it spontaneously | It was never really there | Change how it goes in: action, choice, emotion, the child's own hands |
| Retrieval cue — filed in one place | Knows it in the bath but not in the kitchen; knows it with her cup but not with yours | No — stored, but only findable in one setting | Five contexts, deliberately varied |
| Retrieval competition — it loses the race | Says the Ukrainian or Russian word instantly instead; sometimes gets it a beat later | No — it is simply slower than its rival | Wait time, forced choice, and English-only moments |
Read that table twice before you do anything else. If you take the wrong row, you will spend three weeks drilling a word your child already knows, which is the fastest way I know to make a toddler resent English.
The four-question test: which one is it?
This takes about ten minutes and needs one object your child used to name in English — let's say an apple. Do it when she is fed, rested and in a good mood. Do not call it a test, do not use a serious voice, and stop instantly if she loses interest.
Question 1 — Recognition. Put the apple next to two other things and say, warmly, "Where's the apple?" If she looks at it, touches it or hands it to you, the word is stored. Full stop. Everything after this is an access problem, not a learning problem.
Question 2 — Forced choice. Hold up the apple and a banana: "Apple or banana?" Choice questions bypass the hardest part of recall — she only has to pick between two options rather than search her whole vocabulary. If she can answer here but not to an open question, you have a retrieval issue, and it is a mild one.
Question 3 — Change the room. Take the same apple somewhere else entirely: the hallway, the balcony, the car. Ask again. If the word appears in one place but reliably vanishes in another, you have found the context trap, and it is the most common single cause I see.
Question 4 — Listen to what comes out instead. When she doesn't say "apple", does she say nothing at all, or does she immediately produce the word in your home language? Instant substitution with the home-language word is not forgetting. It is competition: two labels for one object, and the better-practised one wins.
| Test result | Diagnosis | Go to |
|---|---|---|
| No reaction even to recognition (Q1) | Shallow or absent encoding | Cause 1 |
| Recognises, but only in one room or with one object (Q3) | Single retrieval cue — the context trap | Cause 2 |
| Home-language word arrives instantly instead (Q4) | Retrieval competition | Cause 3 |
| Answers forced choice, not open questions (Q2) | Normal recall lag — needs the prompt ladder | The prompt ladder |
| Understands everything, produces almost nothing, in every language | Possibly a silent period — a different topic | Understands but won't speak |
Cause 1: the word never really got in
This is the only genuine learning failure of the three, and it has one overwhelmingly common source: the word was repeated at the child rather than used by the child.
Psychologists have understood since the 1970s that how deeply information is processed at the moment of learning predicts how well it is remembered far better than how many times it was repeated. Craik and Lockhart called this levels of processing, and the practical version for a toddler is beautifully concrete: the more of her body and her intentions a word passes through, the more firmly it lodges.
Here is the same word, "apple", entering a child's memory at five different depths. The difference in retention between the top row and the bottom row is not small — it is the difference between a word that fades in a fortnight and a word she still has next year.
| Depth | What it looks like | What the child does | How well it holds |
|---|---|---|---|
| 1. Labelling | You point and say "apple" | Listens | Weakest — fades fastest |
| 2. Labelling with attention | You hold it up, she looks, you name it | Looks and listens | Weak |
| 3. Action | "Wash the apple." "Put the apple in the bag." | Moves her body | Good — this is why TPR works |
| 4. Choice | "Apple or banana?" and she gets what she chose | Decides, and the decision has a consequence | Strong |
| 5. Need | She wants the apple and the word gets it for her | Uses the word as a tool | Strongest — often one-trial learning |
Most home English lives permanently at depths 1 and 2, because that is what "teaching" feels like it should look like. Almost all of the durable learning happens at 3, 4 and 5. If your diagnosis was Cause 1, you do not need more exposure. You need to move the same word two rungs down this table, which usually costs nothing and takes no extra minutes in your day.
The most powerful rung is the last one. A word that genuinely gets a child something she wants — the apple, the door opened, one more turn — is learned with a speed that no amount of pointing can match. This is also why the games in learning English through play outperform flashcards so decisively: in a game, the word does something.
Cause 2: the word lives in only one place
This is the cause almost nobody explains to parents, and it accounts for more "she's forgotten everything" messages than the other two combined.
Memory does not store a word as a clean dictionary entry. It stores the word together with the circumstances in which it was learned — the room, the light, the object, your voice, the time of day, what she was doing with her hands. Retrieval then works best when those circumstances come back. Psychologists call this the encoding specificity principle (Tulving and Thomson), and the most famous demonstration involved scuba divers who learned word lists either underwater or on the beach: they recalled markedly more when tested in the same environment they had learned in.
Your kitchen is your child's underwater. If "apple" was learned exclusively at snack time, sitting in the same chair, from the same fruit bowl, then "apple" is not filed under apple. It is filed under snack-time-in-my-chair. Take her to the supermarket and ask her what that red thing is, and the cue is missing, so the word is missing. She has not forgotten it. You have removed the handle it was stored by.
This produces the exact pattern mothers report as forgetting: the word is perfect at home and absent everywhere else, or perfect for weeks and then apparently gone when the routine changes — a new flat, a holiday, dropping the afternoon nap, starting nursery.
The five-context rule
So here is the rule I give every mother in the course, and it is the highest-value single change most families make:
A word is not learned when it has been repeated fifty times. It is learned when it has been used in five genuinely different contexts.
Five contexts give a word five separate handles, so that any one of them can pull it out. Fifty repetitions in one context give it one very polished handle that only works in that room. Here is what five contexts looks like for a single word across an ordinary week — note that this adds no new time to your day at all.
| Context | Where | What you say | What changes |
|---|---|---|---|
| 1. The real object | Kitchen, snack time | "Here's your apple." | Baseline |
| 2. A different setting | Supermarket, park bench | "Look — apples! So many apples." | Place |
| 3. A picture | Book, on the sofa | "Where's the apple? There it is." | Real object → symbol |
| 4. Pretend play | Toy kitchen, teddy's picnic | "Teddy wants an apple. Give teddy the apple." | Reality → imagination |
| 5. A song or rhyme | Car, bath, tidy-up time | Any counting or fruit song with the word in it | Melody as an extra cue |
Do that with ten words rather than fifty repetitions of twenty words, and the difference by month three is dramatic. It is also the reason the method I teach is built on daily rituals across the whole day rather than a single "English time" — a lesson slot is, by definition, one context.
Cause 3: the word is there but loses the race
Your child has two labels competing for one thing. One of them she has heard several thousand times, from everyone she loves, since birth. The other she has heard a few hundred times, from one person, for two months. When she reaches for the concept, both labels start moving, and the stronger one arrives first almost every time.
This is not confusion, and it is not a sign that the two languages are tangled. Bilingual children keep their languages remarkably well separated from very early on. It is simply that access speed follows practice, and the home language has a colossal head start. Researchers describe this in terms of relative exposure and language dominance: the proportion of a child's input that is in a language predicts the vocabulary she has in it, and the ease with which she reaches it.
Three things follow from this, and all three are reassuring:
- Silence is not loss. A word that loses the race is still fully stored. You will often see this half a second later — she says the Ukrainian word, then adds the English one quietly, to herself. That is the losing runner arriving.
- Relearning is astonishingly fast. When a word that seemed "gone" comes back after one or two exposures rather than the twenty it originally took, that is proof it was never gone. Ebbinghaus named this savings, and it is the single best evidence you have that your work is still in there.
- Dominance shifts, and it shifts both ways. A child who starts an English-medium nursery can flip within months. That does not erase what you built — it changes which language wins the race.
The fix is not more repetition. It is removing the competition for short, predictable stretches: one ritual per day where the home-language label is simply not available, because you are playing a game whose rules are in English, or because teddy genuinely only understands English. If your child actively protests at English rather than simply defaulting to the home language, that is a different problem with a different solution — see when your child refuses to speak English.
Why forgetting is the mechanism, not the failure
Here is the idea that changes how mothers feel about all of this, and I wish it were taught in every parenting book.
Memory that is never allowed to fade never gets strong. Since Ebbinghaus mapped the forgetting curve in the 1880s, and across a very large modern literature on distributed practice, the finding has been consistent and slightly counter-intuitive: material revisited after a delay, when it has partly faded, is retained far better than material revisited immediately. Robert Bjork's term for this is desirable difficulties — the effort of retrieval is not a symptom of weak learning, it is the thing that causes strong learning.
Translated for your kitchen: the moment your child pauses, frowns and then produces "apple" is worth more than ten fluent instant repetitions. That struggle is the word being strengthened. When mothers see the pause and rush to fill it — "apple, it's an apple, say apple" — they remove precisely the mechanism that would have made it permanent.
So the goal is not to prevent forgetting. The goal is to keep catching words on the way down. That is what the schedule below is for.
The three-layer system that makes words stick
Everything above resolves into three layers. Most home English does layer one thoroughly, layer two by accident, and layer three not at all — which is exactly the profile that produces a vocabulary that grows for six weeks and then stalls.
| Layer | Job | What it looks like | Most common error |
|---|---|---|---|
| 1. Encoding | Get the word in deeply | Action, choice, need, emotion, her own hands | Labelling and pointing only |
| 2. Retrieval | Make her pull it back out | Questions she must answer; gaps she must fill; wait time | Telling her the word instead of letting her reach |
| 3. Consolidation | Let it settle and revisit it spaced out | Sleep, and planned returns at days 1, 3, 7, 21 | Massing everything into one day and never returning |
A useful self-check: for every ten minutes of English you do today, how many seconds does your child spend producing rather than listening? If the honest answer is under thirty seconds, layer two is your bottleneck, and no amount of extra input will fix it.
How to ask so she has to reach for it
Retrieval practice — the finding that being tested on material beats re-reading it — is one of the most robust results in memory research. But you cannot test a two-year-old, and you should not try. What you can do is climb a ladder of prompts, each of which asks for slightly more, and drop back a rung the moment she struggles twice.
| Rung | Prompt type | Example | Use when |
|---|---|---|---|
| 1 | Model — no demand | "That's an apple." | Brand new, or a bad day |
| 2 | Recognition | "Where's the apple?" | She reacts but doesn't speak |
| 3 | Forced choice | "Apple or banana?" | She recognises reliably |
| 4 | Cloze — the gap | "You're eating an…?" (and wait) | She answers choices easily |
| 5 | Open recall | "What's this?" | She fills gaps confidently |
| 6 | Delayed recall | "What did we buy at the shop today?" | She names things on sight |
Three rules make this work, and they matter more than the ladder itself:
- Wait five full seconds. Count them. Adults typically wait about one second before filling a silence, and young children processing a second language routinely need three to five. Most "she doesn't know it" verdicts are delivered two seconds too early.
- Never ask twice. If the answer doesn't come, drop a rung and turn it into a model: "That's right — it's an apple." No sighing, no "come on, you know this". A second demand converts a memory problem into a motivation problem, and motivation problems take far longer to repair.
- Ask no more than three times in one interaction. Beyond that you have stopped playing and started examining, and toddlers can feel the difference instantly.
The revisit schedule that fits a real day
Spaced repetition is usually presented as an app with a review queue. That is useless with a two-year-old. But the principle underneath it is not, and it maps neatly onto rituals you already have. The plan below tracks a handful of words at a time — five is plenty — and each return costs seconds, not minutes.
| When | What to do | Ritual to hang it on | Time |
|---|---|---|---|
| Day 0 | Introduce at encoding depth 3–5: action, choice or need | Wherever it naturally occurs | 1–2 min |
| Day 0, later | One more use, in a different room | Bath or bedtime | 20 sec |
| Day 1 | Prompt ladder, rung 3 (forced choice) | Breakfast | 20 sec |
| Day 3 | New context — picture book or pretend play | Story time | 1 min |
| Day 7 | Rung 4–5, out of the house | Walk, shop, car | 30 sec |
| Day 21 | Delayed recall — mention it from memory, no object present | Any conversation | 15 sec |
Six touches over three weeks, in five different contexts, totalling under five minutes per word. Compare that to what most families do: forty repetitions in one week, in one room, then nothing. The second approach costs more time and produces less.
How many repetitions does a word actually need?
You will find confident numbers online — "children need 70 repetitions", "it takes 21 days" — and I want to be straight with you about them: those figures are marketing, not findings. When a blog gives you a precise universal number for how many times a child must hear a word, it has invented it, and that should lower your trust in everything else on the page.
What the research does support is that the number varies enormously — from a single exposure to many dozens — and that it is driven by factors you can actually influence:
- Concreteness. "Apple" is far easier than "again" or "gentle". Concrete nouns you can hold arrive fastest; function words and abstractions take longest.
- Depth of encoding. A word that got her something she wanted can be learned in one trial. A word she was shown a flashcard of may need fifty.
- Number of contexts. Varied contexts dramatically reduce the total exposures needed for durable learning.
- Share of exposure. A child hearing English 20% of her waking time will need more calendar weeks per word than one hearing it 50% — the same learning, spread over more days.
- Emotional salience. Words attached to laughter, surprise or strong wanting stick disproportionately. This is not sentimentality; it is how memory is weighted.
So the honest answer to "how many times?" is: fewer than you fear, if you use the word well; more than you can bear, if you only point at it.
The layer almost everyone skips: sleep
Consolidation — the process that moves a fragile new memory into durable storage — happens substantially during sleep, and in the age group we are talking about that includes naps. Research with toddlers and infants has repeatedly found that a nap after learning new words supports retention and, importantly, generalisation: the ability to apply a newly learned label to a new example rather than only the exact object it was learned on.
That generalisation point is directly relevant to everything above, because generalising is precisely what breaks the context trap. Two practical consequences:
- Put new words before sleep, not after. The last twenty minutes before a nap or bedtime is prime real estate. A bedtime story, a quiet naming game, three words in the bath — this is the highest-value English slot in the day, and it is one most families are already awake for.
- A week of bad sleep will look exactly like forgetting. Teething, illness, a time-zone change, a dropped nap — vocabulary regresses visibly and then recovers. Before you conclude the method has failed, check whether the last fortnight was a normal sleep fortnight.
The seven-day rescue plan for faded words
Use this when a specific set of words — say five to eight — has gone quiet. It is deliberately small. Rescuing five words properly beats re-teaching thirty badly.
| Day | Focus | What you actually do |
|---|---|---|
| 1 | Diagnose | Run the four-question test on three of the words. Write down which cause you found. Do no teaching at all today. |
| 2 | Re-enter at depth | Bring each word back through action or need, never through pointing. Five words, five moments across the day. |
| 3 | Second context | Same words, different room. If day 2 was the kitchen, today is the hallway, the car, outside. |
| 4 | Symbolic context | The same words in a book and in pretend play. Real object → picture → teddy. |
| 5 | Rest and sing | No questions today. Songs and stories containing the words. This is not a day off — it is spacing. |
| 6 | Retrieval | Prompt ladder, rungs 3 and 4 only. Five seconds of wait time. Never ask twice. |
| 7 | Out of the house | Use every word once, somewhere that is not home. Note which ones survive the trip — those are genuinely learned. |
Then leave those words alone for two weeks and come back on day 21. The gap is not neglect. The gap is what makes them permanent.
Nine mistakes that cause forgetting
| Mistake | Why it backfires | Do this instead |
|---|---|---|
| Drilling the same word in the same place | Builds one retrieval cue that fails everywhere else | Five contexts, deliberately varied |
| Filling the pause after one second | Removes the retrieval effort that creates strength | Wait five seconds, every time |
| Asking "what's this?" repeatedly | Turns play into examination; invites refusal | Maximum three prompts per interaction |
| Adding new words while old ones fade | Nothing gets the spaced returns it needs | Cap active words at five to eight |
| Testing in front of other people | Performance pressure suppresses production | Never perform your child |
| Re-teaching from zero when a word goes quiet | Wastes the savings effect; signals failure | Assume it is stored; change the cue |
| Flashcard stacks as the main method | Encoding depth 1; no context variety; no need | Games, rituals and real objects |
| Doing English only in one daily slot | One slot is one context, by definition | Spread across morning, meals, walk, bath |
| Judging a fortnight of illness or bad sleep | Consolidation is impaired; regression is temporary | Wait for a normal week before judging |
Holidays, trips home and long breaks
A three-week trip to see grandparents, where the child is immersed in the home language and hears no English at all, produces the most alarming version of this pattern: a child who arrives back apparently having lost half of what she had.
She has not. What has happened is a temporary swing in dominance plus the removal of every English retrieval cue at once. In my experience with families, most of it returns within one to two weeks of normal routine, and it returns through the savings effect — one or two exposures rather than twenty. The practical measures are small:
- Take one portable ritual with you — the same three songs, or the same bedtime book. It preserves a cue that does not depend on your kitchen.
- On return, restart with the oldest, best-established words, not the newest. You want early wins, and the oldest words come back fastest.
- Expect a 7–14 day recovery and do not add anything new until it has happened.
For families raising children away from their home country, this cuts both ways — the same mechanisms apply to keeping Ukrainian strong. That balance is worth its own read: English for Ukrainian children abroad.
When it isn't memory at all
Everything in this guide assumes typical language development in a child who is progressing normally in her strongest language. My background includes speech therapy as well as language teaching, and this is the point where I want to be careful rather than reassuring.
Loss of English words is not, by itself, a warning sign. The following patterns are worth a professional conversation, and note that every one of them is about overall language, not English:
- Loss of words in every language, especially the home language, sustained over weeks
- Loss of skills she previously had — not just words, but gestures, pointing, waving, responding to her name
- Reduced eye contact or a drop in social back-and-forth alongside the language change
- No clear words at all by around 18 months, or no two-word combinations by around 30 months, counted across all her languages together
- Any regression that is sudden and does not recover within a few weeks
If you do seek an assessment, ask one question that changes the quality of the answer enormously: "Which languages did you assess?" A bilingual child evaluated in English alone will often look delayed when she is not. Her vocabulary must be counted across all her languages together — that combined total is the developmental measure, while the English share is only a measure of how the method is going. Confusing the two is the single most common reason bilingual children are wrongly flagged. There is more on this in does bilingualism delay speech? and on measuring progress properly in how to tell whether your child is actually learning English.
Where this fits in Mommy & Me English
If you have read this far, you may have noticed that almost none of the fixes above involve doing more English. They involve doing the same English in more places, asking instead of telling, waiting five seconds, and coming back on day 7 instead of day 1.
That is not a coincidence. It is the whole design principle behind the method I teach. Mommy & Me English is built for mothers of children aged one to five, and it is structured around daily rituals — morning, meals, walks, bath time, bedtime — precisely because rituals solve the two hardest problems in this article at once. They deliver the context variety that breaks the context trap, and they deliver the spacing that consolidation needs, without asking you to find an extra half hour that does not exist.
There are no textbooks and no cramming, because cramming is exactly the pattern that produces six weeks of progress followed by a stall. And your own English does not need to be perfect — the mechanisms above are about how a word is used, not how flawlessly it is pronounced. If that worry is the one holding you back, it is answered directly in teaching English when you're not fluent yourself.
Frequently asked questions
These are the questions mothers ask me most when words start going quiet.
Where to go next
If you keep one sentence from this guide, keep this one: your child almost certainly has not forgotten the word — she cannot currently find it, and the handle you gave it only works in one room. Give it five handles instead of fifty polishes, ask instead of tell, wait five seconds, and come back in a week.
That is the entire repair, and it costs no additional time in a day that has none to spare.
If you would like the whole method rather than the repair — the rituals, the games, the songs and the week-by-week structure that keeps words from fading in the first place — that is what Mommy & Me English is. You can see how it fits together on the Mommy & Me journey page. I am Kateryna: Trinity College London CertTESOL, NILE Young Learners trained, with a background in speech therapy and primary education and over nine years teaching young children — and the mother's-eye view of all of this is the part I care about most. The free starter guide gives you a first week you can begin tomorrow.
And the mother whose daughter "lost" thirty words in week six? We ran the four-question test on a Tuesday. The child pointed to every single object correctly. Nothing had been lost at all — it had simply all been learned at one small table, in one chair, at one time of day. Three weeks later the words were back, this time in the car, the lift, the park and the bath. They have not gone anywhere since.
