Speak to your baby. Not at them, not about them, not in the high-pitched cartoon register reserved for objects we find cute. To them. Like a person. Tell them what is happening, what you are about to do, what you noticed about their morning. Use real words in real sentences. Pause for their response — they will respond, in coos and looks and gestures, and these are answers if you treat them that way.
This sounds obvious and is almost universally ignored. Walk through any park, any pediatrician's waiting room, any family gathering, and you will hear adults addressing babies in two registers: silent or silly. Silent — handling the baby's body without verbal accompaniment, as one might handle a bag. Silly — the singsong cartoon voice that treats the baby as a pet to be entertained. Neither is talking to them like a human.
The cost of not talking to babies is documented and large. Hart and Risley's research on the so-called word gap measured tens of millions of words of difference in the language input that children received by age four across different family environments. Vocabulary, reading readiness, and academic outcomes correlated with these differences. The methodology has been refined and criticized since, but the core finding survives: babies who are talked to extensively develop language faster and more richly than babies who are not.
The mechanism is straightforward. Language is learned through exposure to language used in context. The baby's brain is doing astonishing statistical work in the first year — extracting the phonemes of their native language from the speech stream, tracking which sounds appear together, building a map of what referents go with what sounds. They need raw material. The raw material is human speech, addressed to them, in real situations.
But the volume is not the whole point. The quality matters too. Babies addressed in actual conversation — sentences with meaning, intonation that reflects real content, pauses for their response — learn language differently from babies who hear the same volume of disconnected adult chatter on the radio. They are not learning to hear words. They are learning to participate in language, which is a different thing.
Beyond language, talking to babies builds the assumption of being a participant in human life. The baby who is regularly addressed develops the implicit knowledge that they are someone who is talked to, whose attention is sought, whose responses matter. This is upstream of language acquisition and persists even when the language work fails — children with language delays who were talked to extensively still develop a stronger sense of social standing than children who were not.
What does talking to a baby like a human actually look like? It looks like narrating. "I'm going to change your diaper now. First we'll take off your pants. There. Now I need to clean you up. Cold wipe — sorry." It looks like asking questions and waiting for the answer, even when the answer is a noise or a look. "What do you think of this new puree? Yeah? Not yeah?" It looks like reporting on the world. "It's raining today. Look — the drops on the window." It looks like commentary on what they are doing. "You found your foot. That's a foot. Very interesting."
Cartoon voice — what linguists call infant-directed speech or motherese — has its place. Exaggerated intonation does help infants parse phonemes. But the high-pitched silly voice is not the same as infant-directed speech, and the content matters as much as the prosody. You can use the natural musical intonation of address to babies while still saying real things in real sentences. You do not have to choose between warm vocal quality and meaningful content.
The opposite — speaking only in baby talk, only in nonsense syllables, only in the diminutive — gives the baby a smaller language environment than they need. Worse, it transmits the message that the people around them think they cannot handle real speech. The baby cannot articulate this objection, but the impoverished input does its damage regardless.
There is also the question of what you talk about. Babies should hear conversations that include them, not just commands and questions directed only at them. When other adults are present, talk about real things while including the baby in the social field. When the baby is awake during dinner, talk to them and around them about whatever you would normally discuss. They are participating in human social life. Let them experience it.
Avoid the trap of talking about the baby in front of them as if they are not there. This becomes particularly acute as they grow older and start understanding more than they can produce. The baby who hears themselves discussed — their challenges, their funny behaviors, their developmental issues — in their own presence learns that their existence is a topic of public commentary. This is corrosive. Conduct the conversations privately when they are about the baby in any non-trivial way.
The Law of Unity here is that being human is being in language. Not because language is what makes us human in some essentialist sense, but because the human social world is conducted in language, and being addressed in language is the primary mode by which one is welcomed into that world. To talk to a baby like a human is to welcome them. To not talk to them, or to talk to them only in reduced registers, is to leave them at the edge of the human field.
The welcome is consequential. Children who have been talked to extensively from infancy show up to language with confidence. They expect to be understood. They expect to understand. They have a stance toward speech as their medium, not as something foreign they are being asked to acquire. This stance persists into adulthood and shapes how they navigate every speech-mediated context — school, work, relationships, public life.
The opposite stance — speech as foreign, as the medium of adults from which one was excluded — is harder to dismantle later. You can give your child the better stance for free. It costs nothing to talk to them. The conditions for doing it are already present. You only need to choose to do it, every day, in the small moments when the default would be silence or the cartoon voice.