"Cocktail Party" Study

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

For many years, psychologist have wondered how we are able to follow the voice of one talker in the midst of competition from other voices. This situation is much like a person at a cocktail party, bombarded with the babble of a multitude of people. Indeed, in 1953, Cherry appropriately dubbed this dillema of separating one voice from another, the "cocktail party" problem.
     Similarly, infants must pay attention to one voice over others if they are to make sense of the babble around them. Just like our hypothetical individual at a cocktail party, infants must be able to relabily segment one speech stream, even in the face of other sounds and distractions. The question is how.
     One possible answer is that the voice the infants are attending to must be louder. Several years ago, Newman and Jusczyk (1996) found that infants were able to pick individual words out of a blended stimulus made from combining the speech of two voices. However, infants were only able to do this if the voice they were listening to was at least 5dB louder than the distracting stimuli.
     Another possibility is that infants can make use of visual cues to aid them in this task. In the current study, we are examining if 7.5-month-old infants can reliably segment the speech stream of a female talker, 5 dB softer than the distracter stimulus, if the female voice is accompanied by a video which corresponds to her voice.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

    Cherry, E. C.  (1953).  Some experiments in the recognition of speech, with one and two ears. Journal of the Acoustical Society of America, 25, 975-979. 

   Newman, R. S., & Jusczyk, P. W.  (1996). The cocktail party effect in infants.  Perception and Psychophysics, 58(8), 1145-1156.