Imperative Interference: Social Register Shapes Instruction Topology in Large Language Models

TL;DR

This paper investigates how social register influences instruction interpretation in multilingual large language models, revealing that imperative and declarative instructions are processed differently across languages due to social conventions.

Contribution

It demonstrates that social register mediates instruction topology in multilingual models and shows how rewriting instructions can significantly alter cross-linguistic interactions.

Findings

Rewriting declarative instructions reduces cross-linguistic variance by 81%.

Changing imperative blocks shifts instruction topology from competitive to cooperative.

Models process instructions as social acts, not just technical commands.

Abstract

System prompt instructions that cooperate in English compete in Spanish, with the same semantic content, but opposite interaction topology. We present instruction-level ablation experiments across four languages and four models showing that this topology inversion is mediated by social register: the imperative mood carries different obligatory force across speech communities, and models trained on multilingual data have learned these conventions. Declarative rewriting of a single instruction block reduces cross-linguistic variance by 81% (p = 0.029, permutation test). Rewriting three of eleven imperative blocks shifts Spanish instruction topology from competitive to cooperative, with spillover effects on unrewritten blocks. These findings suggest that models process instructions as social acts, not technical specifications: "NEVER do X" is an exercise of authority whose force is language-dependent, while "X: disabled" is a factual description that transfers across languages. If register mediates instruction-following at inference time, it plausibly does so during training. We state this as a testable prediction: constitutional AI principles authored in imperative mood may create language-dependent alignment. Corpus: 22 hand-authored probes against a production system prompt decomposed into 56 blocks.