Casey Lew-Williams
Ricardo Barros
How Children Learn: Q&A With Casey Lew-Williams

CHILDREN LEARN THE COMPLEXITIES of sounds, words, sentences, and communication in a remarkably short period of time. At the Princeton Baby Lab, which was established in 2014, researchers are trying to figure out exactly how that happens, working with babies and young children from central New Jersey. Casey Lew-Williams, co-director of the lab (with Lauren Emberson) is an assistant professor of psychology. PAW spoke with him about what research tells us about learning language, both as babies and adults.

How did you end up doing research about babies and children?

I worked at a preschool when I was in college at UC-Berkeley. I watched kids learn and communicate, and that started my pursuit of developmental research. I’m simultaneously interested in questions about how learning gets off the ground in infancy and how language is learned. Those are very compatible questions to be asking. Language is a wonderful test case for how human cognition works.

What do you study at the Baby Lab?

Most of our studies use language as a window into understanding how babies detect, process, and learn from the structure in their environment.

Do we know how children actually acquire language?

Humans are equipped to learn language in that they have general learning processes that allow them to break into the very complex structure of language. In the first six months of infancy, two main signatures of human cognition interact: our species’ capacity for detecting and remembering patterns, and our interest in communicating with others.

Infants are bombarded with millions of bits of data at every moment. They have the perceptual capacities to take it in: an advanced visual system, an advanced auditory system, advanced motor systems. They begin noticing what is said, and with whom, and when. Over months and years they put the pieces together and begin to understand sounds, words, and sentences. They understand the value that they have in communicating with other people. Those basic cognitive processes can give rise to the complexities of languages.

How do you study such young subjects?

The basic question we ask is “How do babies learn?” So we have to be creative with our methods, given that babies can’t tell you what they’re thinking. They can’t perform sophisticated motor movements or engage in sophisticated communications. So we take advantage of their very simple, yet complex, behaviors. For example, they turn their heads to look at things to pay attention, then get bored with things and look away.

We use methods that capitalize on these movements. An eye tracker can tell you where they’re looking at a certain point in time. We might teach them several words in the course of a few minutes, showing them one novel object at a time. Then we put two novel objects side by side and tell them to look at one of them. We can see the speed at which they get to the correct object by using the eye tracker. What’s great about eye tracking is that it’s been used in many labs and in many studies, and it’s very easy to implement. Babies can sit on Mom or Dad’s lap and passively look or listen.

Why is the speed at which a baby looks at the right object important?

Our ability to process language early on can predict outcomes later. Disparities in language learning are evident even in infancy.

What does that suggest about when to intervene with kids who are poor or otherwise at risk?

As a society we invest a lot more in the older ages, like our investment in preschool for 3- to 5-year-olds. And clearly our government funds elementary school and beyond. Not enough attention is devoted to the first 12 to 24 months of life as a critical time for setting the stage for what happens later.

So what do you think about some of the public-health campaigns urging parents to talk to their babies?

Interventions aimed at supporting early learning are becoming more rigorous with time. They draw from the basic science emerging from the field. So here’s my opportunity to promote basic science! While some of our findings aren’t immediately applicable to helping children learn, they are an important step that can eventually lead to concrete information for parents, teachers, and health-care professionals.

Should parents be narrating everything to their babies all the time?

No, I don’t tell parents to talk all the time. I’d rank quality of speech first. Quantity comes second. The easiest message to convey to parents is that play is important. If you’re sitting on the ground with a child playing, you’re using language — probably in creative ways. The more play that happens, the greater likelihood that the child’s learning will accelerate. You want to follow your child’s lead, to expose him or her to ideas, to interact meaningfully. The byproduct of that is exposure to language.

What about reading to your kid? Is that just as good as chatting or playing?

Reading is another fantastic way to expose a child to language. Ideally you’re not just reading the pages in a book. You’re pausing to engage with the child: How does this relate to his or her life? Children’s books are more diverse in terms of vocabulary and grammar than speech. So there’s an extra value to reading, because it gets parents outside their own natural tendencies or conversational topics and into the language and ideas of an author.

What else are you learning from your research at the Baby Lab?

Our studies collectively suggest that babies are very good at hacking the statistical structure of language, and Ph.D. students in the lab have recently completed studies that approach this idea from several angles. For example, we recently found that 7-month-old babies don’t learn any new pattern that comes their way. Instead, they learn better if they think a pattern will be useful for communicating. And we study how early pattern recognition helps with two important challenges in language.

First, Jessica Schwab [GS] recently showed that toddlers can best learn a new word if parents say the word in a few sentences over a short window of time. Second, we study how the ability to learn patterns gives rise to the ability to predict patterns, such as predicting what words come next. This is important. If your predictions are confirmed, they’re reinforced. If they’re erroneous, it becomes a learning moment so you can revise your view of how things work in light of new information. Tracy Reuter [GS] found that babies’ ability to make and update predictions is a key part of early language learning. If babies are good at predicting — as revealed by very quick eye movements as they look and listen — they tend to have larger vocabularies.

What are you learning about bilingual kids?

As a field, we’ve shown that when babies get enough high-quality, high-quantity exposure to two languages, they can learn both simultaneously. They can segregate the languages; they learn that certain sounds are heard with other sounds. When they receive enough exposure to each language, they have the capacity to learn both in most cases.

Is their learning inhibited, since they’re learning two languages at once?

Generally, the science on early bilingualism finds that bilingual children are less confused than we might expect. They’re not more likely to have language difficulties or delays or disorders. They don’t need neatly packaged information segregated by person or by space. Parents don’t need to avoid mixing languages. In fact, being able to flexibly switch back and forth between languages is a hallmark of many bilingual cultures that babies are raised in.

So is it beneficial to go out of your way to get your kid to learn a second language?

My general answer is that it does appear that bilingualism has some advantage for cognition. But the effects are small, and not high-stakes.

Exposure to music might have the same effect. Other forms of enrichment could have similar effects. When children experience enrichment defined in almost any way that makes sense for them or their family, they benefit. And bilingualism is one example of a type of experience that allows you to flexibly communicate with people, to learn about two cultures, to learn to read.

Is there a difference in how people learn different languages?

The same learning processes are applied to all languages. They’re learned differently only in that they offer different cues to their structure. For example, in tonal languages like Mandarin, rising or falling tones carry semantic meaning. In English, that’s not the case. But babies can detect which cues are the most important and reliable for navigating their particular environments.

Is it impossible for adults to learn a new language very well?

The best summary is that learning is pretty much always better earlier in life! If you’re 45 and want to learn Japanese, better to do it now than when you’re 80. But there’s no definitive threshold. There’s the notion that puberty is when you lose the ability to speak like a native speaker, but that’s unlikely to be accurate.

Another consideration is that adults don’t really have the bandwidth to immerse themselves in a new language the way a baby does. Even if they’re living in a country where the second language is spoken, they have much more to do than look and listen to a new language.

Interview conducted and condensed by Katherine Hobson ’94

VIDEO: Fostering Successful Learning in the First 1,000 Days of Life, by Casey Lew-Williams

Courtesy World Economic Forum