Learning to Decipher Speech in Noise and the Impact of Sleep on Learning


Conversation is an important part of human life. Given globalization and the numerous languages around the world, it is increasingly likely that we will be communicating with others speaking in their second language (L2) rather than their first language (L1). In these situations, communication may require more effort. However, people can become attuned to different speakers and understand others quite well, even in suboptimal listening environments. Previous research has shown that exposure to L2 speech from many talkers of one language background can give the listener experience with a variety of sounds to gain a more generalized understanding of speech sounds. This information allows listeners to better understand new talkers, even if they are of novel talkers from different language backgrounds. In Experiment 1, we explored how brief training in single- vs multiple-talker conditions affected later performance. We found that participants were able to improve with training and that this training allowed for later generalization to novel talkers when tested roughly 11 hours later. Interestingly, the ability to generalize to new talkers depended on individual speaker intelligibility as well as the number of talkers experienced during training. Experiment 2 built on these findings employing a technique known as targeted memory reactivation to explore the role of sleep in this generalization process after training on a low-intelligibility talker. Based on our results, we hypothesize that long and undisturbed sleep may support this type of generalization learning. Overall, this research adds to a growing literature on speech perception helping us better understand the nuances of human communication.

Alternate Identifier
Date created
Resource type
Rights statement