Stanford, California - Children learn words best when they are used in a context that is coherent and distinctive, new Stanford research shows.

This context is critical for understanding why some words may be easier for kids to learn than others, according to the new study by Michael Frank, an associate professor of psychology at Stanford, and scholars at MIT.  The study, titled, "Predicting the birth of a spoken word," appeared in the journal Proceedings of the National Academy of Sciences.

According to Frank, the study goes beyond simply addressing how often a child is exposed to a word to where and how they hear the word. In doing so, he said, it provides evidence that what really matters for word learning is that words be used in a context that is distinctive for the child so that he or she can more effectively decode what the speaker is trying to say. In sum, words used in distinctive ways or in specific routines were learned and used by the child earlier than words that were said more frequently.

"Our work develops ways of measuring other aspects of the words the child hears, by using measurements of where, when, and in what context the words were used to reveal how they are used in routines like book reading, mealtimes, or diaper-changing," Frank said.

Quality over quantity

Frank was joined in the project by Deb Roy, an associate professor at the MIT Media Lab who led the research project. In the analysis, lead author Brandon Roy, a postdoctoral researcher at Stanford and the MIT Media Lab, led the team in examining a unique dataset that contains dense coverage of a single child's daily life from birth to age 3.

The dataset consisted of audio and video recordings from all rooms of the child's house, capturing 70 percent of the child's waking hours. The researchers identified the unique words that the child heard and uttered and the contexts in which they occurred.

The societal implications of the research involve how early language sets the stage for later academic achievement, Frank said.

Frank said the most surprising aspect of the findings was how consistent they were across the various "distinctive" contexts. The researchers measured spatial distinctiveness, or which words were used in particular places in the home; linguistic distinctiveness, or which words were used in the context of particular other words; and temporal distinctiveness, or which words were used at particular times of day.

Distinctive contexts

"These three were generally very related to one another. This was a real signal to us that the distinctiveness measures were likely picking up on consistent activities in the child's life," he said.

"Although researchers have been trying for years to understand the ways that language input contributes to learning, the problem has always been exactly what should be measured," he said. "Our study joins other recent research in suggesting that quantity is not enough and that we really want to try and quantify the quality of the language that children hear."

Frank pointed to the importance of using larger datasets, as the study does – especially multimedia data – in understanding human behavior.

"This dataset was useful because it allowed us to take a really in-depth look at one child's environment, from many different angles," he said.

Frank noted that much work remains to be done to test the researchers' conclusions on other children. As for trying to alter parents' behavior, Frank and his colleagues suggest embedding words into fun, coherent activities, rather than just generally talking more to children.