In three experiments, speakers were interrupted in the middle of language production in order to try to find out (1) whether bookkeeping processes and representations for message level planning in language production can be demonstrated empirically, (2) how such representations are architecturally tied in to higher- and lower-level cognitive processes such as executive planning and memory, and (3) the applied question of whether particular interruption types are more or less disruptive to speakers. Experiment 1 implemented a semi-natural dialog in which a participant was interrupted at predetermined narrative points by a confederate participant. Interruption types were manipulated, and resumption difficulties were measured in order to address the above research questions. Experiments 2 and 3 were more tightly controlled experiments that investigated whether similar bookkeeping processes exist for syntactic planning during language production. The study begins the investigation of two corners of a potentially large research space on the topic of interruptions in conversation. |