Researchers find that good night’s sleep enhances language learning

Scientists at the University have demonstrated that sleep has an important impact on improving the ability to learn language. The researchers found that sleep improved the ability of students to retain knowledge about speech produced by a computer, even when the students seemed to forget some of what they had learned during the day before a night’s sleep.

Published in the journal Nature, the paper, “Consolidation During Sleep of Perceptual Learning of Spoken Language,” was prepared by researcher Kimberly Fenn; Howard Nusbaum, Professor in Psychology; and Daniel Margoliash, Professor in Organismal Biology & Anatomy.

“Sleep has at least two separate effects on learning,” the authors wrote. “Sleep consolidates memories, protecting them against subsequent interference or decay. Sleep also appears to ‘recover’ or restore memories.”

Scientists have long hypothesized that sleep has an impact on learning, but the new study is the first to provide scientific evidence that brain activity promotes higher-level types of learning while we sleep.

Although the study dealt specifically with word learning, the findings may be relevant to other learning, Nusbaum said. “We have known that people learn better if they learn smaller bits of information over a period of days rather than all at once. This research could show how sleep helps us retain what we learn.”

In fact, the idea for the study arose from discussions Nusbaum and Fenn had with Margoliash, who studies vocal (song) learning in birds. “We were surprised several years ago to discover that birds apparently ‘dream of singing,’ and this might be important for song learning,” Margoliash said.

“Ultimately, our discussions stimulated a research design first proposed by Kim Fenn. The interdisciplinary nature of the research and the free exchange of ideas between animal and human work is also very exciting for us,” Margoliash added.

For their study, the team tested college students’ understanding of a series of common words produced in a mechanical, robotic way by a voice synthesizer that made the words difficult to understand. They first measured the students’ ability to recognize the words. They then trained them to recognize the words and then tested them again to measure the effectiveness of the training.

None of the students heard the same word more than once, so they had to learn how to figure out the pattern of sounds the synthesizer was making. “It is something like learning how to understand someone speaking with a foreign accent,” Nusbaum said.

The team tested three groups of students. The control group was tested one hour after they were trained, and they recognized 54 percent of the words, as opposed to the 33 percent they recognized before training.

The scientists trained the second group of students at 9 a.m. and tested them at 9 p.m., 12 hours later. During that 12-hour interval, the students had lost much of their learning and only made a 10 percentage point gain over their pre-test scores.

A third group was trained at 9 p.m., allowed a night’s sleep and then tested the next morning at 9 a.m. Those students improved their performance by 19 percentage points over their pre-test scores.

The second group was then re-tested after being allowed a night’s sleep, and their scores improved to the same level as the students in the third group.

“We were shocked by what we found,” Nusbaum said. “We were particularly intrigued by the loss of learning the students experienced during the day and then recovered.”

Researchers have yet to determine if the reduction in performance was because students had forgotten what they learned, had listened to other speech or had thought about unrelated issues during the day.

“If performance is reduced by interference, sleep might strengthen relevant associations and weaken irrelevant associations, improving access to relevant memories,” the authors wrote. If information was forgotten, sleep might help people restore a memory.

Margoliash said, “Although these initial results cannot explain what is lost during the day, the question is very amenable to follow-up experiments.”

Fenn added, “We are currently considering an FMRI study to investigate brain activity at the end of a day’s learning compared with activity patterns after a night’s sleep.”