A recent study published in the Proceedings of the National Academy of Sciences (PNAS) has shed light on a significant milestone in language development among children. Researchers from the University of Chicago, along with other collaborators, have utilized behavioral and computational data to identify when English-speaking children begin to construct sentences independently rather than merely imitating what they hear.
The study focused on determining when children start using language rules to create new expressions. Susan Goldin-Meadow, the Beardsley Ruml Distinguished Service Professor at the University of Chicago, explained their approach: “We pinpointed the moment when we thought each child can do this, and then we tried to model that with a computer.” The research team discovered that children typically begin forming determiner-noun combinations they have never heard before at around 30 months.
The methodology combined computational modeling with behavioral observations. This novel approach provides new insights into how children learn language. Goldin-Meadow emphasized its potential impact: “This novel approach opens new avenues to explore long-standing questions about how children learn language.”
The study also explored children's learning through mistakes, such as overgeneralizing grammar rules by saying phrases like "I eated my dinner." By examining determiner usage—words like "a" and "the"—researchers assessed whether children understood these patterns by observing if they used both determiners for the same noun.
Researchers observed 64 English-speaking children interacting with their caregivers every four months for 90 minutes. They noted that around 30 months, children began using determiners in front of nouns more creatively than what was recorded from their caregivers.
To verify these findings, researchers developed a predictive computer model simulating how a child learns language. Goldin-Meadow described this process: “To test the model, we give it utterances the child produced that contained a determiner, and we block out the determiner. Then the model has to predict the word that goes in the blocked-out space.” The model confirmed that children's ability to produce novel determiner-noun combinations aligns with real-world observations.
Goldin-Meadow believes understanding these productive moments is vital for addressing broader linguistic questions regarding how much input is necessary for learning specific structures. This inquiry is particularly relevant to her research on homesigners—deaf children who develop unique gestural signs without access to established sign languages like ASL.
She noted: “Determiner-noun constructions may be a lot easier to learn than constructions homesigners don’t invent,” suggesting further experimentation with computational models could offer additional insights into language acquisition processes.