Electrophysiological markers of language adaptation in joint production. Evidence from human-robot interaction
Résumé
The present study investigates whether people engage in lexico-semantic processing when performing a language task with a humanoid robot. A human participant and a robot alternated in naming pictures of objects belonging to 15 semantic categories, while participants’electrophysiological activity was recorded. We manipulated word frequency as a measure oflexical access having half of the pictures associated to high frequency names and half to lowfrequency names. In addition, the robot was programmed to give the semantic category name(e.g., tool for the picture of a hammer) instead of the more typical basic-level name (e.g., hammer for the picture of a hammer) for items belonging to 5 categories. Analyses on thepicture-locked activity revealed a comparable ERP associated to frequency both when it was participant’s turn and robot’s turn to speak. Analyses on the response-locked activity show a different pattern for the category and basic-level responses in the first but not in the second part of the experiment, indicating acquisition and adaptation to the lexico-semantic pattern of the robot. Taken together, our findings provide empirical evidence for 1) the involvement of listeners’ production system to predict robot’s upcoming words, and for 2) partner-adaptive behavior supporting comprehension.