Edit-level Majority Voting Mitigates Over-Correction in LLM-based Grammatical Error Correction

TL;DR

This paper introduces a training-free inference method using edit-level majority voting to reduce over-correction in LLM-based grammatical error correction across multiple languages.

Contribution

It presents a novel, training-free inference technique that improves grammatical error correction performance without model modifications or extra training.

Findings

Outperforms greedy and MBR decoding in most benchmarks

Yields stable correction quality across different prompts

Effective across nine diverse language benchmarks

Abstract

Grammatical error correction using large language models often suffers from the over-correction issue. To mitigate this, we propose a training-free inference method that performs edit-level majority voting over multiple candidates generated by a single model, without requiring model modifications or additional training. Across nine benchmarks covering English, Czech, German, Ukrainian, Korean, Hindi, and Romanian, the proposed method outperforms both greedy and MBR decoding in most cases. Moreover, it yields stable correction quality regardless of the instruction prompts used. We release two repository supporting GEC datasets loading and LLM inference.