The Mystery of Babies’ First Words

It’s nearly impossible to discern when an infant’s babbling turns into a fully formed word.

a small girl talking on a phone
Maite Pons / Getty

One Friday in 1977, a 1-year-old named Nathaniel living in Leiden, in the Netherlands, said “mawh,” which his English-speaking parents enthusiastically greeted as his first word. It came with a pointing gesture, and all weekend, his parents responded by giving him what he pointed at, because mawh, they thought, clearly meant more. But when they got home from work on Monday, their Dutch-speaking babysitter excitedly told them about Nathaniel’s first word, the Dutch word for “pretty,” mooi, and that whenever he said “mawh,” she agreed with him, “Ja, ja, dat is mooi!” Yes, yes, that’s pretty.

After Monday, the baby was silent. Those nine hours with the babysitter, his mother later wrote, “either confused or discouraged Nathaniel sufficiently that he stopped using the word completely, and in fact failed to acquire any replacements for several months.”

“A full day of not getting ‘more’ was enough to cause him to reconsider this whole language thing,” his mother, the Harvard education psychologist Catherine Snow, told me. She noted that he was a late talker but “has made up for it since.”

Snow related her son’s woes with mawh in a 1988 essay about a problem faced by parents and scholars of early child language alike: There’s no bright line between baby babbling and first words. Rather, wordlike forms wriggle one by one from the phonological mush like proto–land animals crawling from Cretaceous seas. More might sound like mawh, light might sound like dai, and all done might sound like a-da. As a result, a baby’s true first word can be hard to pin down. To grant a wordish form any status, you have to account for children’s control of their tongue, lips, and jaw, but also what they think words do. They might say something consistently in a certain context even if it doesn’t sound like anything adults would recognize as a word, so does that count? What about something mimicked? What about a name?

“A lot of kids have this disconcerting all-over-the-placeness with their early vocalizations,” says Michael Tomasello, a developmental psychologist at Duke University who studies the emergence of language and communication in babies and primates. There’s a gradualness to early words, he says. “Even things that someone would call a word, kids still use them in situations that are a bit baffling.”

The messy wordishness of early language makes it less of a definitive milestone than some of kids’ other developmental moments, like first steps or sexual maturity. Some Western parents may jot down first words in baby books. The earliest American baby books, dating to the 1880s, provided spots to write first words, says the Rutgers emerita historian Janet Golden. But not every culture gives them attention. For example, among the Kaluli of Papua New Guinea* (as the linguistic anthropologist Bambi Schieffelin noted in the 1980s), children are not considered to be using language until they say two specific words, those for mother and breast—even if they’re already saying other things. It’s as if the Kaluli deal with the fuzzy vagueness of early utterances by waiting for specific ones. No culture has rituals or ceremonies to mark a child’s first words, according to the Utah State anthropologist David Lancy. This makes sense; how can you celebrate what you can’t discern?

Though parents may insist that their children’ first words are important to them, and though they may prize children’s verbal fluency, first words pale as a cultural institution, especially compared with the big language milestone at the other end of life. Last words appear as Trivial Pursuit clues. Biographies standardly rely on them as motifs. They have been anthologized in multiple languages for centuries, which earned them a subject heading in the Library of Congress classification. But apart from a few children’s books (such as Mo Willems’s Knuffle Bunny and Jimmy Fallon’s Dada) and sitcom appearances, first words barely register on the broader cultural landscape. Many people don’t know their own first words, probably because most first words are banal and forgettable.

Child-language researchers found their solution to the problem of wordishness: Let parents handle it. After all, they are experts on their children, who say more in everyday contexts than they ever would for a stranger in a lab. In the 1980s, a team headed by Elizabeth Bates, a UC San Diego researcher, developed the Communicative Development Inventories, or CDI, a checklist of hundreds of nouns, verbs, adjectives, and pronouns that parents tick off if their children say or understand them. Different versions have been designed for children eight to 36 months old. Parents also note how their children use gestures, parts of words, and grammar. The CDI asks: Does your child tend to say “doggie table” or “doggie on table”? Does your child say “blockses”[instead of “blocks”]? Since it became widely available, around 1990, the CDI has been adapted for several varieties of English, Spanish, Hindi, American and British Sign Languages, and nearly 100 other languages, from Arabic to Yiddish. (As a joke, the list of adaptations includes Klingon.)

The CDI allowed researchers to start to understand the full range of kids’ early vocabularies, how they grow, and how they are tied to other language abilities. An early CDI study, published in 1994, of 2,000 24-month-olds showed that at that age, “normal” vocabularies range from fewer than 50 words to 600 words, with the median at 300 words. Everyone knew there was variability, but that much variability “was big news,” says Virginia Marchman, a Stanford research scientist who serves on a nonprofit board overseeing the CDI.

In 2014, a Stanford professor, Michael Frank, approached Marchman. He told her he had a bunch of CDIs from a previous study taking up space in his filing cabinet. She did too. They decided they wanted to build a tool that would make all that information easily searchable and accessible to other researchers and the public. The result is Wordbank, which now consists of more than 82,000 CDI reports in 29 languages and dialects. An initial analysis of Wordbank data was published online in January.

If the CDI showed how variable children’s early vocabularies are, Wordbank reveals that those vocabularies also have consistent themes. Seeing these themes makes first words more interesting as a phenomenon than as any single instance. Infants tend to talk about more or less the same things, no matter what languages they learn. Across 15 languages, they prefer to say and tend to understand words about sounds, games and social routines, body parts, and important people in their life. Words learned early in one language tend to be learned early in other languages. In American English, the 10 most frequent first words, in order, are mommy, daddy, ball, bye, hi, no, dog, baby, woof woof, and banana. In Hebrew, they are mommy, yum yum, grandma, vroom, grandpa, daddy, banana, this, bye, and car. In Kiswahili, they are mommy, daddy, car, cat, meow, motorcycle, baby, bug, banana, and baa baa.

One reason for this consistency is that such words rank high in a trait researchers call “babiness,” which simply means they’re words that have to do with babies, their immediate surroundings, and important, concrete things. They are often words babies hear frequently.

But another reason for the consistency is that babies tend to learn words that help them interact with their parents and caregivers. “Kids want to share things; they want to be part of the social mix,” Frank told me. Hi is the first word for a lot of kids. No is also a frequent first word. (In an earlier study, Frank found that no was more often a first word for younger siblings than firstborn children.)

Early words in each language do reflect cultural norms and parenting practices—sounds (like vroom), body parts, and games and social routines are unusually frequent in English, while babies whose families speak Kiswahili and Kigiriama often learn words for places to go and words about outside. Then there are patterns that are difficult to account for, such as the high proportion of words for vehicles, clothing, and animals learned by infants speaking northern European languages and Korean.

It also appears that 1-year-olds in most languages tend to say and understand more nouns than verbs, and use many fewer function words (such as the, and, and also), even though they hear function words frequently. Two exceptions are Mandarin and Cantonese, where children say more verbs, probably because those languages allow speakers to use a lone verb (run) to stand for clauses that in other languages require subjects or objects (he runs).

There are some interesting demographic differences. According to Wordbank, in 25 of 26 languages, girls under 3 years old produce more words than boys in that age group. There are also gender-related differences in the kinds of words babies tend to say. Boys seem to say words for vehicles and objects associated with stereotypically male activities, such as sports, earlier than girls; girls seem to learn words for genitals and clothing earlier than boys. Also, earlier-born children said and understood more words than younger siblings, perhaps because (as child-language researchers suspect but haven’t definitively shown) parents address more speech to firstborn children.

Once kids get older, there are fewer discernible patterns in which words they acquire. While early words are quite alike across languages, later learned words begin to differ, likely influenced by kids’ environments and interests. As Frank writes in Wordbank, “as acquisition unfolds, the features that make languages (and cultures) different from each other play an ever-increasing role in driving acquisition.”

Yet the overarching theme of Wordbank is variability, no matter the language. This suggests that no culture, no family structure, and no social environment has some special sauce that will turn out speakers or signers of a particular type. Everywhere, kids are “taking different routes to language,” as Frank puts it.

The father of two, Frank finds this liberating. “Parents tend to assume that variations they observe in their child’s language are due to specific parenting decisions that they’ve made. But children vary so much that small variations in parenting will usually come out in the wash.” Major differences in language input will still be consequential, but others, like reading one book or two before a nap, will barely register.

Even though first words are so similar, many American parents still put the first word on a pedestal, just as first steps are a big deal even though the baby will likely go on to become bipedal like most everyone else. But communication doesn’t begin with a fully formed word—there is so much that comes before.

On their way to learning language, children often make vocalizations known as “proto-words,” which do wordlike work but sound nothing like adult words. About eight years ago, I eagerly tracked my infant son through his structured babbling, naively expecting a crisp adult-like English word to one day flutter forth. What emerged, at about 11 months, was “ka,” which came along with a pointing gesture. This was not the arrival of his personhood that I’d anticipated, but what ka lacked in profundity it made up in perplexity.

Maybe it’s car, my wife surmised, because he said it while pointing at trucks in a book. But then he aimed ka at a bicycle. Backtracking, we wondered whether it might be a label, not for a specific thing, but for a category of vehicles. After all, he used ka with a wheelchair, a barbecue grill, and a shopping cart. That hypothesis died when a Ganesha statue on a shelf prompted a ka as well.

Such early utterances have a lot of social work to do—they’re more about enabling an interaction than about referring to something specific. So it seems as if ka was less an act of naming than the on-switch for a shared experience. Essentially, I think he was saying, “Here’s a cool thing; we should look at it together.” That’s when I realized that an earlier sound he used to make, something that sounded like eh, accompanied by a beckoning gesture, was likely a way of communicating too. I would paraphrase its meaning as “Hey you, over there; I am over here looking at you.” It’s hard to imagine writing eh in the baby book or throwing a party to celebrate its appearance, but I insist on calling it his first word.

The truth is that by the time he said his first adult-sounding word, “wheel” (pronounced “whee-oh”), we had already communicated so much with each other via smiles, eye gaze, waving, and pointing that words felt superfluous. I realized that before every first word is a proto-word; before every proto-word, a gesture; before a gesture, what?

When I interviewed Mike Frank via Skype, he was sitting on a couch in his home while his newborn son slept in a bassinet nearby, and he was in the process of telling me how, before he had kids, he too focused on discrete emergence of things like first words—then the baby squawked.

“Hey dude,” Frank cooed, “you okay there?”

The baby was silent, but this was its own kind of communication. He was fine; Frank and I resumed our conversation.

*This article originally misstated the island where the Kaluli live as Samoa.