Children learn language much faster than adults and teenagers. This learning advantage is not due to differences between children or adults but rather differences in how people talk to them.A team of researchers devised a method that allowed them to evaluate experimentally how parents use the information they have about their children's languages when speaking to them. The researchers found that parents had very precise models of their children’s language knowledge and used these models to adjust the language they use to communicate with them. These results can be found in an online advance publication of Psychological Science."We know for years that parents communicate with children in different ways than they do to other adults," stated Daniel Yurovsky (assistant professor of psychology at Carnegie Mellon University). "This stuff helps young children get a foothold in language. But we don't know if parents alter the way they talk to children as they acquire language. Giving children input that is 'just right,' it's helping them learn the next thing.Children tend to be more patient and speak at a slower pace than adults. They use exaggerated repetition, exaggerated enunciation and simplified language structures. To gauge the child's understanding, adults often ask questions during communication. Adults use more complex sentences and structure as the child improves their language fluency.Yurovsky compares it to the path a student takes when learning math at school.Yurovsky stated, "When you go to school you start with algebra and then you take plane geometry before you move onto calculus." People talk to children using the same structure, without even thinking about it. They track how much language their child understands and modify how they speak to make it easier for them to understand.Yurovsky and his team wanted to find out how caregivers adjust their interactions to suit their child's speech development. The team created a game in which parents assisted their children to choose an animal from a group of three animals. This game is something toddlers (ages 15 to 23 months old) and their parents often play in their daily lives. The matching game featured half of the animals children are familiar with before they turn 2. The other half of the animals in the matching game were animals that children typically learn later, such as cow, cat, and leopard. peacock, leopardAdvertisementResearchers asked 41 adult-child pairs to play the game in a laboratory setting. The researchers measured how different parents spoke about the animals they believed their children knew compared to what they thought they didn't know.Yurovsky stated that parents have an in-depth knowledge of their children's language as they have seen them learn and grow. These results demonstrate that parents can use their knowledge of children's language development to refine the linguistic information they give.Researchers discovered that caregivers used many techniques to communicate the "unknown" animal to the child. The most popular approach was to use additional descriptive words that were familiar to the child.Yurovsky stated that this [research] approach allows us to confirm experimentally ideas we have developed based upon observations of parents and children engaging in the home. "We discovered that parents used what they knew about their children's language skills before the study. However, if they were wrong about something -- for instance, their child didn’t know the word ‘leopard’ -- they changed how they talk about the animal the next time.Each animal was a target at least twice during the 36-part experiment. Participants were similar in race to the United States (56 % white, 27% Black, and 8% Hispanic).AdvertisementThese results are representative of western parenting perspectives and caregivers with higher education than the average country. Researchers did not measure children's knowledge about each animal. This study could not distinguish whether children learned new animals during the game.Yurovsky believes that the results could be of some use to researchers in the field machine learning.He said that these results could help him understand machine learning language systems. "We train language models right now by giving them all the language data that we have. We might be able to give them the right data at exactly the right time and keep it at the level of complexity they need.Yurovsky was assisted by Ashley Leung from the University of Chicago and Alex Tunkel from The George Washington University School of Medicine and Health Sciences. The James S. McDonnell Foundation provided funds for this project.
