How do speakers coordinate? Evidence for prediction in a joint word-replacement task.

We investigated whether speakers represent their partners' task in a joint naming paradigm. Two participants took turns in naming pictures; occasionally the (initial) picture was replaced by a different picture (target), signaling that they had to stop naming the initial picture. When the same participant had to name the target picture, he or she completed the name of the initial picture more often than when neither participant had to name the target picture. Crucially, when the other participant had to name the target picture, the first participant also completed the name of the initial picture more often than when neither participant named the target picture. However, the tendency to complete the initial name was weaker when the other participant had to name the target than when the same participant went on to name the target. We argue that speakers predict that their partner is about to respond using some, but not all, of the mechanisms they use when they prepare to speak.