First Language Acquisition: How Children Learn to Talk

A two-year-old who can barely use a spoon can also string words together, ask questions, and argue about bedtime. Give that same child another three years and she will be handling tenses, embedded clauses, and sarcasm—in a grammar so intricate that engineers still cannot get machines to match it. First language acquisition refers to this near-universal feat: the way humans pick up their native tongue without lessons, worksheets, or anyone sitting them down to explain a rule. It sits at the heart of linguistics and developmental psychology, and the more we learn about it, the stranger it gets.

Why Acquisition Is So Astonishing

Think about the raw problem. A baby hears a continuous ribbon of noise with no gaps between "words"—nothing in the audio signal corresponds to the white space on this page. Out of that stream, the child has to carve out individual units, attach each one to a meaning pulled from a world of rabbits, aunts, microwaves, and feelings, and then crack the hidden combinatorial rules that let those units form sentences. On top of that, they pick up the pragmatic side of conversation—whose turn it is, what counts as rude, how to drop a hint.

All of this happens in roughly four years, mostly from overheard chatter and casual back-and-forth, with no curriculum and no grammar book. The same pattern shows up in Seoul and Helsinki and a tiny hamlet in Papua New Guinea. That kind of cross-cultural reliability is why researchers treat language acquisition less like a subject children are taught and more like a biological program running on schedule.

Listening In Before Birth

Language learning gets underway while the learner is still in the womb. By the final months of pregnancy the fetus can hear, with the mother's voice arriving through bone and amniotic fluid. Low-frequency information travels best, so what reaches the fetus is mostly the music of speech—its rhythm, its melody, the rise and fall of phrases.

Newborns give away how much of that has sunk in. Only hours out of the womb, they suck harder on a pacifier to hear their mother's voice over another woman's, and they prefer a passage their mother read aloud during pregnancy to one they have never heard. DeCasper and Fifer's 1980 experiment is the classic demonstration, and later work has stacked plenty more evidence on top of it.

Within days of birth, babies also sort languages by rhythm. Infants born into Spanish-speaking homes perk up for Spanish over Japanese; babies raised around English show a preference for English over French. The stress-and-timing fingerprint of a language seems to be the very first thing the infant brain learns to recognise.

Babbling and the First Speech-Like Sounds

Somewhere between six and eight months, infants start babbling—chaining consonants and vowels into runs like "dadada" or "gigigi." Linguists call this canonical babbling, and it is a genuine milestone: the baby is voluntarily producing the kinds of syllables real speech is built from.

Babble is not acoustic filler. It has shape, and that shape starts bending toward the sound patterns around the child. By around ten months you can often hear the difference: an Arabic baby's babble, a Korean baby's babble, and a Brazilian Portuguese baby's babble drift apart in the consonants they favour and the intonation they sit on top of.

The clincher is what happens in deaf households. Babies raised with sign language babble with their hands—rhythmic, repetitive gestures that mirror vocal babbling in timing and syllable-like structure. Babbling, in other words, is not really about the mouth. It is about the brain priming itself to produce a linguistic system, whichever channel happens to be available.

The Arrival of First Words

The first recognisable word usually shows up around a child's first birthday, though they understand far more than they can say for several months before that. Early lexicons lean heavily on what children see, touch, and care about: caregivers ("mama," "dada"), pets, snacks, body parts, vehicles, toys. Abstract nouns and verbs come later.

For a while, progress is slow—maybe fifty words by eighteen months. Then, for many children, the dam breaks. The vocabulary explosion (sometimes called the naming spurt) can bring in several new words a day. By the end of the second year a toddler typically commands 200 to 300 words, and by the time that same child starts school she is working with a vocabulary somewhere around 10,000 to 14,000 words.

How does any of this mapping actually work? This is the word learning problem. Your father points at a golden retriever and says "dog." How does a toddler know "dog" names the animal, not its tail, its fur colour, its wagging, or the whole scene of a dog-on-a-lawn? Researchers have found children lean on several biases at once: the whole-object assumption that a new label goes with a whole thing rather than a part, mutual exclusivity that says each object already has one name so a new word probably refers to something else, and social cues like eye gaze and pointing that reveal what the speaker is actually attending to.

Putting Two Words Together

Between roughly 18 and 24 months, word pairs appear: "daddy shoe," "more juice," "kitty sleep." These pairings are not thrown together at random. They map onto recognisable relationships—who is doing what, what is acting on what, where something is, who owns what—hinting that a rudimentary grammar is already in place under the hood.

Linguists sometimes call this telegraphic speech because it leaves out the small grammatical bits—articles, prepositions, auxiliaries, tense and plural endings—and keeps only the words that carry the most meaning. Context fills the rest in. "Daddy shoe" can just as easily mean "that shoe belongs to Daddy," "Daddy is putting on the shoe," or "Daddy, pass me my shoe," and caregivers usually understand without missing a beat.

The Grammatical Growth Spurt

From the second birthday to around the fourth, grammar takes off. Utterances stretch. Function words slot in. Morphemes show up in an order that is eerily consistent from child to child: Roger Brown's landmark 1973 research traced English-speaking toddlers acquiring the progressive -ing, the plural -s, the possessive 's, and the past-tense -ed in roughly the same sequence, regardless of personality or upbringing.

By three, it is normal to hear a child produce sentences with clauses inside clauses ("I want the cookie that grandma made"), proper question formation with subject–auxiliary inversion ("Where did the cat go?"), and negation that matches adult patterns ("I don't like broccoli"). The sheer pace and orderliness of this development is hard to account for without some fairly muscular learning machinery—whether you locate that machinery in innate grammar, powerful statistical learning, or some mix of the two.

Overgeneralization: Useful Mistakes

Some of the best evidence that children are building rules, not just imitating, comes from a certain kind of error. A toddler who happily used "went" and "caught" for months will suddenly start saying "goed" and "catched." Nobody taught her those forms. She has never heard an adult use them. What has happened is that she has extracted the general past-tense pattern—add -ed—and is busy applying it across the board, including to verbs that are supposed to be exceptions.

The trajectory is often U-shaped: the correct irregular form, followed by an overgeneralised form, followed eventually by the correct form again. A child saying "goed" actually knows more about English grammar than one who memorised "went" as a stand-alone chunk—she has spotted the rule. The detour through "goed" is, paradoxically, a sign of progress.

This pattern is not just English showing off. German-speaking children overapply plural suffixes. Russian-speaking children overapply case endings. Hebrew-speaking children overapply verb templates. Wherever there is a regular rule and an irregular exception, children the world over will flatten the exception for a while. That consistency says something important about the mechanisms all humans bring to the job.

Competing Theories of Acquisition

The Nativist Account

Noam Chomsky's position is that children come into the world already equipped with a Universal Grammar: abstract structural principles shared by every human language. Learning a specific language, on this view, is really a matter of setting a handful of parameters inside a pre-built framework. The main push behind this idea is the poverty-of-the-stimulus argument—the claim that the speech a child hears is simply too patchy and ambiguous to explain the grammar she ends up with.

The Behaviorist Account

B. F. Skinner's 1957 proposal treated language as a learned behaviour like any other—picked up through imitation, reinforcement, and conditioning. Parts of vocabulary learning do fit that mould, but Chomsky's famous 1959 review tore the account apart when it came to grammar, pointing out that imitation cannot explain novel sentences children have never heard and cannot explain the very creative errors they do produce.

The Usage-Based Account

Michael Tomasello and colleagues argue for something in between: children build grammar up out of the input using general-purpose cognitive tools—pattern detection, analogy, tracking statistical regularities, and reading other people's intentions. On this view, grammatical structure is an emergent property of accumulated experience, not a blueprint the child was born with.

The Social-Interactionist Account

Jerome Bruner, Lev Vygotsky, and others in this tradition put the spotlight on what happens between caregiver and child. Adults scaffold language learning by simplifying their speech, repeating, asking questions, and embedding words in predictable routines like bath time and book reading. Language, in this frame, is as much a social achievement as a cognitive one.

Is There a Deadline? The Critical Period

Eric Lenneberg argued in 1967 that acquisition runs on a biological clock. During a critical period—roughly childhood into early adolescence—the brain is ideally tuned for soaking up a language. Miss that window without adequate exposure, his hypothesis says, and native-like mastery becomes extremely difficult, maybe impossible.

The evidence, sadly, often comes from children who were cut off from language altogether. "Genie," found at thirteen after years of isolation and abuse, eventually picked up a sizeable vocabulary but never got complex grammar working, even after years of dedicated intervention. Deaf children whose first real language exposure (spoken or signed) is delayed until late childhood show similarly lasting gaps in their grammatical abilities compared with children exposed from infancy.

The idea has clear implications for second language learning, where adults rarely reach fully native-like accents or grammatical intuitions, even after decades of immersion. Vocabulary, interestingly, is the one area that seems relatively unaffected by age—people can keep adding new words across the whole lifespan.

Why Input Matters

Whatever your stance on innate knowledge, no one disputes the basic fact that children need to hear (or see) language to acquire one. There is no recorded case of a child developing a full language from nothing, with no linguistic community around them.

Child-directed speech—often called "motherese" or "parentese"—is the distinctive way adults tend to talk to babies and toddlers: higher pitch, swooping intonation, slower pace, shorter sentences, lots of repetition, and a pared-down vocabulary. That style appears to help, probably because it makes speech easier to notice, easier to chop into chunks, and easier to map onto what the child is looking at or doing.

Both how much and how varied the input is turn out to matter. Children who are talked to more often, and with a wider range of words and structures, tend to build bigger vocabularies and more flexible grammars. The well-documented "word gap" across socioeconomic groups reflects real differences in input quantity and predicts measurable consequences for later reading and school performance.

Growing Up With Two Languages

Children raised with two languages from the start learn both at once, and they hit the main developmental milestones on roughly the same timetable as monolingual children. Bilingual toddlers are not confused—well before their second birthday they already track which language goes with which adult and switch accordingly.

The vocabulary in each individual language may run a little smaller than a monolingual peer's, but add both languages together and the total is usually matched or larger. Any small delays in hitting early milestones tend to level out by the time formal schooling begins, and in many tasks bilingual children show advantages in cognitive control and perspective-taking.

When the Path Looks Different

Not every child moves through these stages on the typical timetable. Developmental Language Disorder (DLD, historically called Specific Language Impairment) affects somewhere around seven percent of children; they have persistent trouble with grammar and vocabulary despite typical hearing, intelligence, and social development. Autism often reshapes language development, especially on the pragmatic and conversational side. Undiagnosed hearing loss can delay spoken language substantially, which is why newborn hearing screening and early intervention are such a big deal.

Knowing the typical sequence is what makes atypical acquisition recognisable in the first place. The work of psycholinguistics and applied linguistics feeds directly into the screening tests, therapy approaches, and classroom supports used by speech-language pathologists, teachers, and clinicians around the world—turning a basic research question ("how does a child learn to talk?") into practical help for the kids who need it.