Cognitive Daily

Rochelle Newman developed these recordings to assess infants' ability to stream speech: to distinguish relevant language from irrelevant noise. The ability to stream is likely critical to language learning -- studies have found that more than two thirds of all parents of infants report that other household members were "frequently" talking while they spoke to their babies. Add to this other noises and distractions in a typical house -- TV, pets, appliances -- and it may seem a wonder that babies can learn to talk at all.

So how well do infants stream? Newman played four recordings to 5-month-old infants, all of them like the first sample you heard above, with the baby's name being spoken 10 dB louder than the background noise. One of the recordings played the child's name, one played a different name with the same syllable stress, and the other two played completely different names. A light came on next to the speaker where the recording came from, and an independent observer recorded how long the babies looked at the light. Here are the results:

The babies looked at the light significantly longer when their own name was spoken. But when the difference was reduced to 5 dB, the results were markedly different:

Now there was no significant difference in looking time, even between the infants' own name and the non-stress-matched names. With just a 5-dB change in the level of the voice calling its name, the baby is unable to distinguish it from any other name. Newman repeated the task with 9-month-olds, and the results were the same. Only when she tested 13-month-old infants was Newman able to show that babies could reliably distinguish their name being called when it was 5 dB louder than the background noise.

Interestingly, 13 months is also the age when most infants begin to talk -- perhaps there is a connection.

Newman, R.S. (2005). The cocktail party effect in infants revisited: Listening to one's name in noise. Developmental Psychology, 41(2), 352-362.