SlangDIT: Benchmarking LLMs in Interpretative Slang Translation
Yunlong Liang, Fandong Meng, Jiaan Wang, Jie Zhou
TL;DR
This work targets the challenge of context-dependent slang translation by proposing SlangDIT, a benchmark comprising slang detection, cross-lingual slang explanation, and translation. It provides a 25k English–Chinese dataset built from movie subtitles and annotated for polysemy, along with a deep-thinking model, SlangOWL, that explicitly reasons through slang term detection, senses, explanation, and translation. Experimental results show that SlangOWL, aided by deep thinking, outperforms vanilla and non-reasoning baselines across detection, explanation, and translation on both general and hard test sets, demonstrating the value of integrating detection and explanation for improved idiomatic translation. The work advances slang understanding and translation with a unified benchmark and reasoning-enabled model, offering practical gains for bilingual NLP systems and highlighting directions for future multilingual expansion and pipeline experimentation.
Abstract
The challenge of slang translation lies in capturing context-dependent semantic extensions, as slang terms often convey meanings beyond their literal interpretation. While slang detection, explanation, and translation have been studied as isolated tasks in the era of large language models (LLMs), their intrinsic interdependence remains underexplored. The main reason is lacking of a benchmark where the two tasks can be a prerequisite for the third one, which can facilitate idiomatic translation. In this paper, we introduce the interpretative slang translation task (named SlangDIT) consisting of three sub-tasks: slang detection, cross-lingual slang explanation, and slang translation within the current context, aiming to generate more accurate translation with the help of slang detection and slang explanation. To this end, we construct a SlangDIT dataset, containing over 25k English-Chinese sentence pairs. Each source sentence mentions at least one slang term and is labeled with corresponding cross-lingual slang explanation. Based on the benchmark, we propose a deep thinking model, named SlangOWL. It firstly identifies whether the sentence contains a slang, and then judges whether the slang is polysemous and analyze its possible meaning. Further, the SlangOWL provides the best explanation of the slang term targeting on the current context. Finally, according to the whole thought, the SlangOWL offers a suitable translation. Our experiments on LLMs (\emph{e.g.}, Qwen2.5 and LLama-3.1), show that our deep thinking approach indeed enhances the performance of LLMs where the proposed SLangOWL significantly surpasses the vanilla models and supervised fine-tuned models without thinking.