Hoedemaker, R. S. 1 , Ernst, J. 2 , Meyer, A. S. 1 & Belke, E. 2
1 Max Planck Institute for Psycholinguistics, Nijmegen, Netherlands
2 Ruhr-Universität Bochum, Bochum, Germany
According to the language-as-joint-action model, interlocutors generate predictions of their partner's upcoming utterances using their own language production system. A strong prediction of this model is that there will be no difference in the effects of semantic context in the form of previously produced and previously comprehended words, as both are processed by functionally equivalent mechanisms. In the current experiment, pairs of participants took turns naming a series of pictures containing sets of exemplars from distinct semantic categories. Pictures were assigned so that task partners alternated naming items within a category. We observed within-speaker Cumulative Semantic Interference (CSI); naming latencies increased with the number of self-produced category exemplars. Critically, we also found significant between-speaker CSI; there was an additional increase in naming latencies as a function of the number of other-produced exemplars. These results support the language-as-joint-action model as self- and other-produced context words had similar effects on subsequent picture naming. However, post-experiment item recall was significantly higher for self- compared to other-produced items, indicating that the comprehension and production systems may be functionally equivalent regarding the immediate effects of semantic context on language processing, but differ in the way they affect downstream processes of memory and retrieval.