Cocktail party effect in infants: Visual information and speech segmentation in noise

Abstract

What are infants abilities to use what they see to segment speech in a noisy environment? Infants often find themselves in situations far louder and more complex than the acoustic isolation chambers of traditional infant testing. The current series of studies used the headturn preference procedure (with video familiarization) to examine 7.5‐month‐old infants abilities to use visual/auditory correlations to reliably attend to and segment a given speech stream in the face of a distracting voice. Results indicated that in contrast to seeing a static face, infants succeeded at segmentation when a dynamic visual display of the face of the talker matched the acoustic passage. That is, when two blended voices were of equal loudness, infants could use visual correspondences to reliably recognize words presented in the matching video. Furthermore, they did so even if the video display was changed to a synchronized oscilloscope pattern, rather than a face. They also succeeded when the video display was simply a syn...