Accommodation Goes Both Ways: Studying Linguistic Convergence Between Humans and Language Models

2026-05-27

AI Summary

This study investigates bidirectional linguistic convergence and its symmetry between large language models (LLMs) and human interlocutors in multi-turn dialogues. Drawing on the WildChat corpus, the authors employ an asymmetric convergence metric across eight languages to analyze stylistic accommodation at both function-word and open-class lexical levels. The research reveals, for the first time, that LLMs consistently over-converge toward users’ linguistic styles, whereas human speakers exhibit convergence toward LLMs at levels comparable to those observed in human–human baseline interactions. This asymmetry indicates a systematic imbalance in current LLMs’ adaptive behavior, offering empirical evidence that informs efforts to enhance naturalness and interactional parity in human–AI communication.

Abstract

As LLMs become increasingly integrated into daily life, understanding how their presence will shape human linguistic behavior is an open question. We present a large-scale study of linguistic convergence in human-LLM dialogue, examining how humans and LLMs accommodate each other's linguistic style during multi-turn conversations. Using an asymmetric convergence metric on WildChat, a corpus of real-world ChatGPT transcripts, we find that while LLMs significantly overconverge toward their users on both function word and open-class features across eight languages, human convergence rates in this setting are broadly consistent with human-human baselines. These findings suggest that accommodation in human-LLM dialogue is asymmetric: while LLMs dramatically overfit to their users' style, humans linguistically accommodate LLMs no differently than they would another person.

Problem

• linguistic convergence
• human-LLM interaction
• language accommodation
• large language models
• dialogue

Innovation

• linguistic convergence
• large language models
• asymmetric accommodation
• human-AI interaction
• dialogue analysis