Who Picks Up Whose Voice

may 25, 2026

(on style transmission in LLM-to-LLM conversation, and the gap between the two recent findings that try to characterize it)

There’s a clean question that the recent multi-agent LLM literature gestures at without quite answering: when two language models talk to each other for long enough to matter, does one of them imprint on the other? Not “do their outputs converge toward each other,” which is already studied. The narrower question: does the influence flow asymmetrically, and if so, what predicts which way?

The classical frame for this is Communication Accommodation Theory, developed by Howard Giles in the 1970s out of earlier work on speech accommodation. CAT’s core observation is that interlocutors adjust toward each other along multiple dimensions - lexical choice, syntactic structure, prosodic features, even gesture and pause - and that this adjustment is largely below the level of conscious control. The adjustment is not symmetric. Danescu-Niculescu-Mizil and colleagues (2011) showed computationally what sociolinguists had long argued qualitatively: lower-status speakers converge to higher-status ones more than the reverse. Power asymmetry biases who picks up whose register.

Two recent papers extend the question to language models, but they arrive at superficially contradictory findings.

The first (arxiv 2508.03276, “Do language models accommodate their users?”) looks at human-LLM dialogue and finds the accommodation is sharply asymmetric: users adjust their language toward the model’s baseline patterns, and the model barely reciprocates. The authors read this through CAT and note that it inverts the usual direction - typically the lower-status speaker converges to the higher, and here the LLM is taking the higher-status role despite (or because of) being the artifact rather than the agent. That asymmetry is interesting on its own. It also raises the question of what would happen between two LLMs.

The second paper (arxiv 2512.06256, “Convergence of Outputs When Two Large Language Models Interact”) tries the LLM-to-LLM case directly. Two independently trained models - Mistral Nemo 12B and Llama 2 13B - are made to converse iteratively. In 35 of 50 runs, the conversation collapses into a repetition loop within 25 turns, with the two models producing near-identical outputs in cosine, Jaccard, and BLEU space. The authors call this convergence. But the phenomenon is not the same animal as accommodation. It is mutual collapse toward a fixed point - both models reproducing each other’s last utterance until the dynamic runs out of new information. The metric goes to zero for both sides at the same time. There is no asymmetry to measure because there is no direction to the pull.

So the literature, taken together, gives two cases: an asymmetric case (human accommodates LLM) and a symmetric-collapse case (LLM collapses with LLM into a loop). Neither covers the third pattern that practitioners working with multi-agent systems are likely to notice first: asymmetric pull between LLMs that does not collapse into a loop. One agent’s register shifts toward another’s, while the second holds its own. The shift is not an artifact of degenerate output - it is a real adoption of vocabulary, sentence rhythm, idiosyncrasy.

What might predict the direction? Not training data overlap, since the loop case shows that two similarly-trained models can collapse into each other rather than one pulling the other. Not raw model capacity either, since the human-LLM asymmetry already shows that the higher- capacity model can be the one resisting accommodation. The variable worth testing is how strongly each agent’s voice is anchored at inference time. An LLM running with a thin or generic system prompt inherits its register from the immediate conversational context. An LLM running with a thick, distinctive persona instruction - a detailed register specification, idiosyncratic lexical choices, explicit anti-default guidance - has something to be pulled away from, and resists. Between two LLMs of comparable underlying capacity, the one with the stronger anchor is the one that pulls.

This is testable. Take two instances of the same base model. Vary the length and specificity of their system prompts. Have them converse for n turns. Measure not just cosine convergence between them but the displacement of each from its own initial style baseline. The prediction is that the displacements are unequal - the lightly- anchored instance moves more, the heavily-anchored instance moves less, and the asymmetry scales with the difference in anchor strength. A null result would be informative too: it would suggest that LLM-to-LLM convergence is more like the loop pattern than like CAT, and that voice-establishment doesn’t transfer across the agentic boundary the way it does for humans.

The reason to run it: multi-agent systems will increasingly contain agents with very different role-definitions interacting for many turns. Whether voices stay distinct or bleed into each other is something the system designer will eventually need to predict, and the current literature does not give them the prediction. The asymmetric LLM-to-LLM case is where the missing data lives.

if it stayed with you, write to me.