How Do Young Children Learn Language? With Statistics

March 15th, 2017

Babies are even smarter than we thought. They are natural statisticians with the ability to use their accumulating experiences with the social world to learn language. According to cognitive scientist Michael Frank, an associate professor at Stanford, understanding that capability can help us ensure children get off on the right foot in school and life.

Young children who have rich vocabularies go on to do better in school than their peers. Unfortunately, children from poorer families tend to have less stimulating language experiences than their affluent counterparts, and studies suggest that may be a factor in pernicious achievement gaps across social classes. Policymakers have begun creating programs to help low-income parents facilitate their children’s language development. But for those efforts to be effective, scientists need to understand exactly how young children learn language and what they need.

Frank’s work is helping to answer those questions by examining the specific mechanisms of language development. Early language acquisition isn’t just about the number of words children hear, but about the quality of conversations they have with their caregivers, Frank and other researchers have found. “Words embedded in rich, coherent discourse are much easier for children to figure out,” he explains. One of the reasons is that young children engage in surprisingly complex processes of inference. Frank, studies social statistical learning, or how children use the information they accumulate in social interactions to make sense of language. “A young child has two separate but overlapping tasks” when communicating with an adult, he says. “One is, ‘Right now, I need to know what you are talking about’ and the other is ‘Over the next two years, I’d like to know what all the words in this language mean and be able to use them.’” Frank’s research suggests that learning one helps children learn the other.

Even young children constantly reason about what they hear and why a speaker said it. These inferences about what the people around them are trying to say can be a great way for children to learn the meanings of words. For example, in one of Frank’s studies, he described a dinosaur to children using a novel word: “this is a dinosaur with a dax.” Children reasoned that “dax” was likely to refer to the most unique or informative property of the dinosaur, and were then able to pick out that property in a later test.

Context is also very important for learning language, Frank’s research shows. For instance, an infant can learn a word more quickly if she hears it in a specific place. Working with a one-of-a-kind dataset that recorded one child’s every utterance from the age of 9 months to 3 years, Frank and his colleagues found that the child learned new words faster and more easily when they were almost always heard in the same place and at the same time of day. For example, the child learned the word “moon” very early, and the researchers found that he heard it frequently right before bedtime and in a specific chair when his parents read him the favorite book Good Night Moon. He also quickly learned other words associated with “moon” in the book, like “spoon” and “chair.”

Frank’s research methods range from analyzing data on one individual child to compiling large datasets with thousands of children. One of his favorite projects is a data repository he started called Wordbank. He has invited researchers from around the world to share data they have collected with a common research instrument. The result is a massive dataset with 70,000 data points from 26 languages that any researcher can use to study how language development occurs across diverse cultures and tongues. Frank’s primary motivation for creating the dataset is that research should be “selfless, not selfish,” he explains.

To that end, Frank hopes his findings will inform the programs that aim to improve low-income children’s language exposure and make them as precise as possible. “Different kinds of words are important to hear at different times,” he explains. “When children are just starting to learn the names of things, it’s important to give them as many opportunities to hear the word as possible. But when you are learning different types of words, like verbs and adjectives, you need to hear them in context” – and it is best for that context to be clear and consistent. Caregivers should engage in playful and interesting conversations with their children, use language to describe predictable routines like mealtimes or getting dressed, and read books frequently and with plenty of interaction. Those kinds of strategies can help make smart babies into well-educated kids.

Michael Frank is a recipient of the Federation of Associations in Behavioral & Brain Sciences (FABBS) Early Career Impact Award. He was nominated by the Cognitive Science Society (CSS) to receive the award, which will be given at the CSS annual meeting in London.