Talk: LLMs syntactically adapt their language use to their conversational partner

It has been frequently observed that human speakers align their language use with each other during conversations. In this talk, I present an analysis on whether large language models (LLMs) exhibit the same behavior of conversational adaptation. The findings show that LLM agents end up making more similar syntactic choices as conversations go on, confirming that modern LLMs adapt their language use to their conversational partners in at least a rudimentary way.

The presentation will go deeper in the analysis that was used to find alignment throughout long conversations. I will further explain a more in depth analysis that further looks into the rate to which syntactic adaptation takes place. I will try to make the explanations as comprehensive as possible to encourage a critical discussion of the results in the end (if there is enough time for discussion).

Info

Day: 2025-05-16
Start time: 14:00
Duration: 00:30
Room: GWZ 4.216
Track: Computational Linguistics
Language: en

Links:

Concurrent Events