How do speakers tailor lexical choices according to their interlocutor's accent?

When conveying a message to an interlocutor, speakers need to code concepts in lexical expressions, a process known as lexical retrieval. There is evidence that speakers can take into account the dialectal background of their interlocutor to tailor their lexical retrieval; for instance, our pilot experiment showed that participants, when asked to guess the defined expression according to a definition, were more likely to produce American English (AE) expressions (e.g., apartment instead of flat) when the definition was spoken by an AE interlocutor than a British English (BE) interlocutor. It is possible that such an interlocutor effect arises from a top-down interlocutor model, from bottom-up accent details, or a combination of both. This registered report aimed to address this issue with two experiments. In Experiment 1, we observed that the interlocutor effect on lexical retrieval did not differ regardless of whether a definition was presented by the interlocutor via speaking (with accent details) or writing (without accent details), suggesting that the effect arises from top-down interlocutor modeling instead of bottom-up accent details. Experiment 2 interleaved two interlocutors with different dialects, one providing a spoken filler definition and the other a written target definition. Participants used more AE expressions for target definitions given by an AE interlocutor than a BE interlocutor, suggesting they maintained concurrent models for interlocutors and used them to retrieve dialect-congruent expressions. These findings highlight top-down influences of the communicative context in language production. (PsycInfo Database Record (c) 2025 APA, all rights reserved).