CoPiT: Cognitive Pivot Translation for Digraphic Low-Resource Mongolian in the Traditional Script

2026-07-07

AI Summary

This study addresses the challenges of machine translation for Traditional Mongolian, which suffers from data scarcity and orthographic ambiguity. The authors propose a cognitively inspired pivot-based translation framework that leverages Cyrillic Mongolian as an intermediary script. Their approach explicitly resolves script-level ambiguities in Traditional Mongolian prior to translation and mitigates low-resource constraints through synthetic parallel data generation. This work is the first to incorporate cognitive motivations into bidirectional script translation, integrating an ensemble of multiple backbone models with script-aware data augmentation strategies. Experimental results demonstrate substantial improvements over direct translation across multiple target languages, with significant gains in BLEU scores and 1.5–1.6× higher COMET metrics. Notably, their open-source models match or surpass GPT-4.1 in performance, and they release the first large-scale, multi-script Mongolian parallel corpus.

Abstract

Low-resource languages remain challenging for machine translation, and Mongolian is a representative case. As a digraphic language, Mongolian is written in both Cyrillic and Traditional scripts, which exhibit a severe imbalance in data availability. While the Cyrillic script is relatively well-resourced, the Traditional script remains extremely data-scarce and orthographically ambiguous, leading to substantial performance degradation in direct translation. We propose CoPiT, a cognitively motivated pivot-based translation pipeline that exploits this internal resource hierarchy by routing translation through the Cyrillic script. The pipeline explicitly resolves script-induced ambiguity in the Traditional script before translation, enabling more stable and accurate meaning transfer. Across multiple backbone models and target languages, CoPiT consistently outperforms direct translation, achieving substantial absolute BLEU improvements together with consistent 1.5-1.6x COMET gains. These gains allow strong open-source models to match or outperform GPT-4.1 under comparable evaluation settings. Beyond inference-time improvements, CoPiT enables the construction of synthetic parallel data directly from Traditional-script text, mitigating data scarcity in realistic low-resource scenarios. We release a new multi-script parallel dataset covering Mongolian in both scripts alongside English, Korean, and Russian. All datasets and code are publicly available at https://anonymous.4open.science/r/anonymous_project-76C7.

Problem

• low-resource
• Mongolian
• Traditional script
• machine translation
• digraphic

Innovation

• pivot-based translation
• digraphic language
• low-resource machine translation
• script disambiguation
• synthetic parallel data