Humans are pretty good at filling in the blanks. One experimental paradigm that tests our brains’ ability to predict language has to do with verbal shadowing. “The task is to listen to someone speaking and repeat what they say as soon as possible after they say it,” says Liberman. “There used to be people who would go on variety shows because they could do it almost as fast as the person was talking. They hardly seemed to be behind them at all. But everyone can do this to some extent with a lag of a few tenths of a second.”
As the speech becomes more unpredictable—or what Liberman calls “word salad—just random words spoken in sequence—the shadowing lag is very long compared to semantically incoherent but syntactically well-formed, nonsense material.” The shadowing rate gets better and better as the structure and the content of the speech become more coherent.
Theories about how our brains prefer predictability show up in music research, too. “What we know from a biological perspective is that the best surprise is no surprise,” says musicologist David Huron. “Large parts of your brain are oriented toward predicting what’s going to happen next. There are excellent biological adaptive reasons why brains should be so oriented toward what’s going to happen. An accurate prediction is rewarded by the brain. It’s one of the reasons why in music we have very predictive rhythms. The thing to say about music is that it’s incredibly repetitive.”
Emberson tested the idea that halfalogues distract us more than dialogues or monologues do by asking people to listen to half of a cell phone conversation while performing a task that required paying attention. To make the cell phone conversations as realistic as possible, Emberson and her colleagues gathered Cornell undergrad roommates, brought them to the lab, and recorded them chatting to one another on their phones. Then the researchers asked them to sum up the conversations in monologues. This provided the researchers with halfalogues, dialogues, and monologues to play to listeners.