Parents who want to expand a toddler's vocabulary are usually given two kinds of advice: read more, and worry about the wrong things. A study published this year in the journal Child Development, with 3,150 children, is unusually good at telling those apart. It measured what a child hears at home, in detail, and then measured how many words they say.
The short version: the amount of talk a child hears in a given language shapes how many words they produce in that language. The variety of accents doing the talking does not. That second finding matters for a lot of French families, where a grandparent, a childminder or a parent speaks the language with an accent, and where someone has usually told them it will hold the child back. Below: what the study did, what it found, and what actually works at home between 11 and 34 months. For the years that follow, we set out the practical version in our guide to helping a child build language.
What actually makes a toddler's vocabulary grow?
Two things, repeatedly confirmed: how much language a child hears directed at them, and how rich that language is. Quantity means the number of words addressed to the child, not words in the room. Quality means variety of vocabulary, complete sentences, and above all turns: the child says something, an adult answers that thing.
This is why background television and audio playing in a room do very little. Words that are not addressed to a child, and that do not wait for a reply, are close to acoustic wallpaper for a two year old. The unit that matters is the exchange, not the exposure.
It also explains why the same book read in two ways produces different results. Reading a story straight through is exposure. Stopping to ask what the character is doing, waiting, and building on the answer is an exchange. The technique has a name, dialogic reading, and research collected by the American Speech-Language-Hearing Association links it to stronger vocabulary growth in toddlers.
What did the study of 3,150 children find?
That accent variety in a child's input had no measurable effect on vocabulary size, at any age tested. Bilingual children did produce fewer English words than monolingual children from around 18 months, which replicates decades of earlier work, but children hearing several accents of the same language were indistinguishable from children hearing only the local one.
The design is worth understanding, because it is what makes the result usable. Families were interviewed for ten minutes about exactly who talks to the child and in what, then caregivers filled in the MacArthur-Bates Communicative Development Inventory, the standard checklist of words a child says. Children were sorted into three groups: monolinguals hearing one accent, monolinguals hearing several, and bilinguals with 30 to 70 percent exposure to each language. You can read the paper in Child Development directly.
One detail makes the sample unusually clean. The study ran in the Greater Toronto Area, where hearing several languages and accents is not a marker of low income or low parental education. In many earlier studies those things travel together, so it is hard to know what is causing what. Here they did not.
The honest limits: vocabulary was reported by parents rather than counted by researchers, English was the only language measured, and one urban Canadian region is not the world. What the study rules out is a large effect of accent variety. It cannot rule out a small one.
Does a parent's accent slow a child's vocabulary?
On this evidence, no. Children whose input included substantial non-native accents grew their vocabularies at the same rate as children hearing the local accent almost exclusively. Young children turn out to be well equipped to handle variation in how words sound.
This is the finding most worth passing on, because the opposite belief is common and it changes behaviour. A grandmother who is told her accent will confuse the baby speaks less. A parent who is told to use the school language instead of their own switches to the language they are least fluent in, and their sentences get shorter and simpler. That trade is a bad one: the child loses richness of input, which does matter, to avoid accent variety, which does not.
The same logic runs through the wider bilingual literature. Work at the University of Washington on how exposure to a second language affects infants' learning found that what drives uptake is lively, interactive, socially rich speech, not accent purity.
Do bilingual toddlers really know fewer words?
Fewer in each single language after about 18 months, yes. Across both languages combined, typically not. The gap in this study appeared only when counting English words, and words the child knew in the other language were not counted at all.
The mechanism is arithmetic rather than cognitive. A child hearing 60 percent French and 40 percent English gets less English than a child hearing 100 percent English, and vocabulary tracks input. Nothing is being lost; the total is simply distributed. We looked at this in more detail in whether learning two languages delays speech.
Practical guidance for bilingual households has been stable for years, and the summary from the NCT on raising a bilingual child holds up: speak the language you are most fluent in, be consistent about who speaks what, and prioritise conversation over correction.
What can I do at home this week?
Increase the number of turns, not the number of words you say. A child who is answered learns faster than a child who is talked at, and four small habits carry most of the effect.
| Habit | What it sounds like | Why it works |
|---|---|---|
| Expand | Child: "dog." You: "a big brown dog, running" | Adds words on top of a meaning the child already has |
| Narrate | "I am cutting the apple. Now I am putting it in the bowl" | Attaches words to something visible and happening now |
| Wait | Ask, then count silently to five before filling the gap | Toddlers need several seconds to assemble an answer |
| Read with stops | "What do you think he will do?" then listen | Turns a book from a broadcast into a conversation |
Two things to drop rather than add. Correcting pronunciation directly tends to reduce how much a child says; repeating the word back correctly inside your own sentence does the same job without the cost. And background audio, whether radio, television or an audio player, does not count towards input and can mask the speech that does.
When is a small vocabulary a reason to ask for help?
When milestones are missed rather than when a number looks low. The markers worth watching, set out in the ASHA developmental milestones, are babbling by 12 months, single words by 16 months, and two-word combinations by 24 months, plus any loss of words already acquired at any age.
Intelligibility is the other useful check, and it is more objective than word counts. According to ASHA's guidance on speech development, roughly half of what a two year old says should be understandable to someone outside the family, about three quarters at three, and around nine tenths at four.
Being bilingual is not an explanation for missing those markers. Bilingual children reach early milestones on the same schedule as monolingual children; if they do not, the bilingualism is not why. For a sense of the normal range at each age, see how many words a four year old should know.
Where does a talking toy fit into this?
Only as extra turns, and only after the human ones. Everything above points the same way: the mechanism is exchange, so anything that produces genuine back and forth can add to a child's practice, and anything that only produces sound cannot. That is the line between an audio player and something a child talks to.
Ted sits on the second side of that line: an interactive plush that holds a real time conversation, invents stories with a child and answers their questions, with the parent choosing the language, the topics, the tone and the limits from the app. There is no camera, and it is a single purchase at 129 euros with no subscription. What it is not is a substitute for the adult who answers a two year old fifty times before lunch. No talking toy has been shown to grow a vocabulary, the effect that is documented belongs to responsive people, and a toy is at most a supplement to that.
Frequently asked questions
How many words should a 2 year old say?
Most two year olds combine two words and have a spoken vocabulary in the low hundreds, but the normal range is very wide. Missed milestones matter more than a specific count, and a child who understands well and communicates by other means is in a different position from one who does neither.
Will my accent confuse my toddler?
No. In a study of 3,150 children, those exposed to at least 40 hours a week of non-local accents had the same vocabulary size as those hearing only the local accent. Speak the language you are most comfortable in, because fluency gives your child richer sentences.
Does reading to a toddler help more than talking to them?
Both work, and books help mainly because they generate conversation. Reading with stops and questions does more for vocabulary than reading straight through, and a story that ends in a discussion counts twice.
Do audiobooks or podcasts build vocabulary?
Not much on their own. Recorded speech is not addressed to the child and does not wait for a reply, so it lacks the exchange that drives word learning. It can be a useful shared activity when an adult listens too and talks about it afterwards.
My child hears three languages. Is that too many?
Not in itself. Each language grows in proportion to how much of it the child hears, so three languages means slower progress in each one, not a delay overall. If milestones are being missed, ask a doctor or a speech and language therapist rather than dropping a language.
