Minimal-Edit Instruction Tuning for Low-Resource Indic GEC

TL;DR

This work targets grammatical error correction for low-resource Indic languages by eschewing augmentation in favor of an augmentation-free, instruction-tuning approach. Leveraging a 12B Gemma3 model with LoRA-based adapters and Alpaca-style prompts, the authors implement a two-stage pipeline: Stage 1 instruction fine-tuning and Stage 2 deterministic inference with a lightweight normalizer, guided by a deterministic error classifier. The method achieves competitive GLEU scores on Hindi and Malayalam under BHASHA protocols, highlighting that minimal-edit prompts, deterministic decoding, and surface-normalization can rival augmentation-heavy pipelines. The study provides thorough error analysis, robust prompt design strategies, and explicit limitations, offering a practical path for reproducible, computation-efficient Indic GEC in truly data-scarce scenarios.

Abstract

Grammatical error correction for Indic languages faces limited supervision, diverse scripts, and rich morphology. We propose an augmentation-free setup that uses instruction-tuned large language models and conservative decoding. A 12B GEMMA 3 model is instruction-tuned in bnb 4-bit precision with parameter-efficient fine-tuning (PEFT) and Alpaca-style formatting. Decoding follows a deterministic, constraint-aware procedure with a lightweight normaliser that encourages minimal, meaning-preserving edits. We operationalise inference, subsequent to instruction fine-tuning (IFT), via a fixed, language-specific prompt directly synthesised from a deterministic error classifier's taxonomy, label distributions, and precedence ordering computed on the training corpus. Under the official untuned GLEU evaluation, the system scores 92.41 on Malayalam, sixth overall, and 81.44 on Hindi, third overall. These results indicate that classifier-informed prompt design, adapter-based instruction tuning, and deterministic decoding provide a reproducible and a computationally efficient alternative to augmentation-centred pipelines for Indic GEC. The approach also motivates future work on stronger morphosyntactic constraints and human-centred evaluation of conservative edits.