LATA: A Tool for LLM-Assisted Translation Annotation
This work addresses the problem of improving translation annotation for researchers in computational linguistics, though it is incremental as it builds on existing LLM and human-in-the-loop methods.
The paper tackles the challenge of constructing high-quality parallel corpora for structurally divergent language pairs like Arabic-English by introducing an LLM-assisted interactive tool that reduces the gap between scalable automation and expert precision, resulting in a system that balances annotation efficiency with linguistic accuracy for complex translation phenomena.
The construction of high-quality parallel corpora for translation research has increasingly evolved from simple sentence alignment to complex, multi-layered annotation tasks. This methodological shift presents significant challenges for structurally divergent language pairs, such as Arabic--English, where standard automated tools frequently fail to capture deep linguistic shifts or semantic nuances. This paper introduces a novel, LLM-assisted interactive tool designed to reduce the gap between scalable automation and the rigorous precision required for expert human judgment. Unlike traditional statistical aligners, our system employs a template-based Prompt Manager that leverages large language models (LLMs) for sentence segmentation and alignment under strict JSON output constraints. In this tool, automated preprocessing integrates into a human-in-the-loop workflow, allowing researchers to refine alignments and apply custom translation technique annotations through a stand-off architecture. By leveraging LLM-assisted processing, the tool balances annotation efficiency with the linguistic precision required to analyze complex translation phenomena in specialized domains.