How “speech disfluencies” help toddlers learn words

From a New Scientist interview with Richard Aslin, one of the authors of a recent study published in Developmental Science titled “Toddlers use speech disfluencies to predict speakers’ referential intentions”:

I had always assumed “ums” and “ers” were useless noises, an indication that my brain isn’t working quickly enough. What does your research show?
It’s one of those things your mother tells you: “Speak in full sentences. Don’t um and er.” I think it’s a view that most people have, that disfluencies are not a good thing because they don’t really communicate anything; they are just fillers.

Our latest study shows that disfluencies in speech directed to young children have an interesting benefit. What children have learned, surprisingly early, is when there is an “um” or “er”, the word that follows is almost always one they don’t know. When you are fumbling for the correct word, you are sending a message to the child that they should pay attention. That’s very useful.

How did you find this out?
The methodology is fairly straightforward. We put pairs of photographs of objects in front of a child. One is an object familiar to the child, say a ball or banana. The other is one that they have never seen, say, a wrench. Using an eye-tracking device we can measure precisely where the child is looking. A voice says: “Look at the ball” (in a fluent way). The next sentence is disfluent: “Look at the… er… wrench.”

We found that when there is a disfluency, the baby will look at the object they don’t know, not the familiar one. It helps teach them new words.