For decades, researchers studying the cognitive science of language acquisition have wrestled with a fundamental question – how important is experience in learning a language? And how does that experience affect language development?

The role of experience is difficult to study because people typically process the input data they experience –  the language they’re exposed to through daily life, interacting with others, watching videos, listening to conversations and media, etc. – and create language through internal computation simultaneously.

Barbara Lust, Ph.D. ’83, professor emerita in the Department of Psychology, and colleagues Suzanne Flynn at the Massachusetts Institute of Technology and Ahyoung Alicia Kim at the University of Wisconsin-Madison, encountered a natural experiment that helps provide answers. They published their findings in the journal Frontiers in Human Neuroscience in November 2024.

Lust studies how children acquire language from a theoretical and experimental cross-linguistic perspective. In the Cornell Language Acquisition Lab, she has conducted experimental and naturalistic studies of young children's acquisition of early syntax and semantics. Since the 1970s, she and colleagues have collected data from children acquiring more than 20 languages. Currently this work continues through a Virtual Center for Language Acquisition.

As part of this work, they were assessing the acquisition of a second language by a young Korean boy, referred to as MJ in the article, acquiring English in an English-only nursery school in the U.S. MJ returned to Korea over the summer, effectively pausing English language input for several months. This natural experiment gave Lust and her colleagues the opportunity to see how this sustained experience without language input affected MJ’s English ability. 

They hypothesized that if language data input and language acquisition are directly related, MJ would regress over the period without input. If they’re directly related but acquisition doesn’t require continual reinforcement through experience, MJ’s language ability would remain stable. And if they’re indirectly related and language acquisition requires internal computation in addition to that required to process input data, MJ’s English language ability might improve even while he wasn’t experiencing English.

They found that MJ’s English language acquisition significantly improved, continuously developing both before and after his time in Korea.

“The standing belief among scientists and educators is that language acquisition directly depends on input data,” said Lust. “AI models and machine learning now attempt to replicate human language acquisition by training with more and more data. Educators are often taught that more data, more repetition of language, or special forms of language data (e.g, the Baby Talk register) is essential to advanced language acquisition. 

“But now we see that it is really the child mind, what it does with the language data it has, that is a major source of that advancing acquisition, not only how much data the environment provides. There is a factor that’s as important – maybe even more important – that many scholars are overlooked.” 

Lust, Flynn and Kim used an enriched case study methodology, combining repeated naturalistic observations of MJ with periodic standardized and experimental tests. They argue that while this method is time-intensive and requires flexibility and a strong relationship between the researcher and the child, it’s a promising approach for studying the internal computation that goes on in a child’s mind when they acquire language.

Lust said that this paper also demonstrates the importance of studying bilingualism.

“We couldn’t have done this study and made this discovery just by studying monolingual children alone,” she said. “Furthermore, this child continued to use and acquire Korean in Korea. Does this experience in one language enforce or advance the acquisition of another?  If so, how?  This is an interesting question we don’t have the answer to yet.”