How Do Babies Learn Language, From Sounds to Speech

Babies start learning language from the moment they’re born, and the process unfolds in a remarkably predictable sequence. Long before a child says their first word, their brain is already sorting through the sounds, rhythms, and patterns of whatever language surrounds them. This journey from newborn coos to full sentences spans roughly the first five years of life, with each stage building on the one before it.

It Starts With Listening

Newborns can’t speak, but they’re far from passive. In the first weeks of life, a baby’s brain is already distinguishing speech from other sounds and showing a preference for the voice of their mother. By two to three months, babies begin cooing, producing soft vowel-like sounds that signal the earliest experiments with vocalization. These aren’t random noises. Babies coo more during face-to-face interaction, which means they’re already linking sound production to social connection.

Between four and six months, cooing gives way to babbling. Babies start stringing consonants and vowels together, producing syllables like “ba,” “da,” and “ma.” This stage sounds remarkably speech-like, and it serves a critical purpose: babies are practicing the mouth and tongue movements they’ll need later for real words. Babbling isn’t limited to happy moments either. Babies babble when excited, frustrated, or simply exploring what their voice can do.

The Sound-Sorting Window

One of the most fascinating things happening in a baby’s brain during the first year has nothing to do with producing sound. It’s about perception. Young infants can distinguish between virtually every sound used in every human language. A Japanese newborn can hear the difference between “r” and “l” just as well as an American newborn, even though Japanese doesn’t use that distinction.

This ability doesn’t last. Research on Japanese infants found that discrimination of the non-native “r” and “l” contrast begins declining between 9 and 12 months of age. What’s happening is a kind of neural specialization: the brain is tuning itself to the sounds that matter in the baby’s native language and pruning its sensitivity to sounds that don’t appear in everyday speech. This is why exposing babies to a language during the first year has such a lasting effect on their ability to hear and eventually reproduce its sounds. It’s also why learning a second language as an adult feels so much harder. The window for effortless sound discrimination has narrowed considerably.

First Words and the Vocabulary Explosion

Most babies produce their first recognizable word around their first birthday. Common early words include “mama,” “dada,” “hi,” and “dog.” These first words tend to be labels for the people, objects, and actions that fill a baby’s daily life. Growth is slow at first. A child might add only a word or two per week for several months.

Then something dramatic happens. Around 18 months, many toddlers hit what speech-language pathologists call the vocabulary explosion. At 18 months, most children are saying somewhere between 10 and 50 words. By 24 months, the average jumps to 200 to 300 words. That’s a roughly tenfold increase in just six months. The minimum benchmark clinicians look for at age two is 50 words, so there’s a wide range of normal.

Part of what drives this acceleration is a learning shortcut called fast mapping. Starting around age two, children can create a rough mental entry for a new word after hearing it just once or twice. They may not fully understand the word yet, but they store enough information to recognize it again and refine its meaning over time. This ability lets toddlers absorb new vocabulary at a pace that would exhaust most adult language learners.

From Words to Sentences

Somewhere between 18 and 24 months, toddlers begin combining two words into simple phrases: “more cookie,” “daddy go,” “big dog.” These two-word combinations aren’t random pairings. They follow basic rules of meaning, with children consistently placing words in an order that communicates who did what. This is the earliest evidence of grammar, even though it looks nothing like a textbook sentence.

By age three, most children are producing multi-word sentences and starting to internalize grammatical rules. This is where things get interesting, because children don’t just memorize sentences they’ve heard. They extract patterns and apply them creatively. A three-year-old who has learned that adding “-ed” to a verb makes it past tense will start saying “runned” instead of “ran” or “holded” instead of “held.” This pattern, called overregularization, is common between ages three and five. It might sound like a step backward, since the child may have said “ran” correctly at a younger age by imitating adults. But it’s actually evidence of a deeper understanding. The child has figured out a rule and is applying it consistently, even to words that break the rule. The exceptions come later.

What’s Happening in the Brain

Language learning isn’t just about ears and mouths. It depends on specific brain architecture that begins forming in infancy. Research from Harvard’s Brain Science Initiative has found that connectivity patterns in the left side of the brain during infancy can predict a child’s language skills years later, well into school age. In other words, the neural wiring laid down during the first year of life creates a scaffold that supports reading and language comprehension for years to come.

This doesn’t mean a baby’s language future is sealed at birth. The brain remains highly plastic throughout early childhood, and the quality and quantity of language input shapes how these neural connections strengthen. But it does underscore that what happens in infancy matters more than most people assume. The groundwork for literacy begins long before a child picks up a book.

What Actually Helps (and What Doesn’t)

The single most powerful thing for a baby’s language development is live, interactive speech from a real person. Talking to your baby during diaper changes, narrating what you’re doing while cooking, reading books aloud, and responding to babbling as if it were conversation all build the kind of back-and-forth interaction that wires the brain for language. It’s not just the number of words a child hears that matters. It’s whether those words come in the context of a real social exchange, with eye contact, turn-taking, and emotional engagement.

Screen time, by contrast, offers much less. The American Academy of Pediatrics recommends avoiding digital media for children under 18 to 24 months, with the exception of video chatting. The concern isn’t that screens are toxic in small doses. Large analyses of the research find that the negative association between screen exposure and language development in young children is statistically small. One study of six-month-olds found no detectable link between screen exposure and language outcomes at all. The real issue is opportunity cost: every minute a baby spends passively watching a screen is a minute not spent in the kind of interactive, face-to-face communication that genuinely builds language skills. U.S. parents report that children under two are exposed to an average of about 49 minutes of screen time per day, and while that alone is unlikely to cause a language delay, it represents a significant chunk of waking time that could be spent talking, singing, or reading together.

Why Some Children Talk Later Than Others

The normal range for language development is surprisingly wide. Some children say 50 words at 18 months while others are barely at 10. Some combine words at 20 months, others not until closer to their second birthday. Boys tend to develop language slightly later than girls on average, and children growing up bilingual may appear to have smaller vocabularies in each individual language, even though their total vocabulary across both languages is typically on track.

Late talking doesn’t automatically signal a problem. Many children who are slow to start talking catch up entirely by age three or four without any intervention. However, a child who isn’t using any words by 16 months, or who isn’t combining words by age two, may benefit from an evaluation by a speech-language pathologist. Early intervention, when it’s needed, is most effective precisely because the brain is still so plastic during these years. The same flexibility that lets babies sort through phonemes and absorb grammar rules also means that targeted support can make a significant difference when development veers off course.