# Discourse Dependency: A Continuous Criterion for Translation Difficulty

arXiv: https://arxiv.org/abs/2609.04959 · HTML: https://mlsanity.com/paper/2609.04959/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 22 of its 58 references that are in the map.
- First posted on arXiv 4 Sep 2026.

## Builds on (most cited first, 22 of 22)

- [Lost in the Middle: How Language Models Use Long Contexts](https://mlsanity.com/paper/2307.03172/) (2023)
- [DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence](https://mlsanity.com/paper/2606.19348/) (2026)
- [Multilingual Denoising Pre-training for Neural Machine Translation](https://mlsanity.com/paper/2001.08210/) (2020)
- [Language-agnostic BERT Sentence Embedding](https://mlsanity.com/paper/2007.01852/) (2020)
- [xCOMET: Transparent Machine Translation Evaluation through Fine-grained Error Detection](https://mlsanity.com/paper/2310.10482/) (2023)
- [Tangled up in BLEU: Reevaluating the Evaluation of Automatic Machine Translation Evaluation Metrics](https://mlsanity.com/paper/2006.06264/) (2020)
- [Neural Machine Translation with Extended Context](https://mlsanity.com/paper/1708.05943/) (2017)
- [Evaluating Discourse Phenomena in Neural Machine Translation](https://mlsanity.com/paper/1711.00513/) (2017)
- [Document-Level Neural Machine Translation with Hierarchical Attention Networks](https://mlsanity.com/paper/1809.01576/) (2018)
- [Large language models effectively leverage document-level context for literary translation, but critical errors persist](https://mlsanity.com/paper/2304.03245/) (2023)
- [WMT24++: Expanding the Language Coverage of WMT24 to 55 Languages & Dialects](https://mlsanity.com/paper/2502.12404/) (2025)
- [Selective Attention for Context-aware Neural Machine Translation](https://mlsanity.com/paper/1903.08788/) (2019)
- [Microsoft Translator at WMT 2019: Towards Large-Scale Document-Level Neural Machine Translation](https://mlsanity.com/paper/1907.06170/) (2019)
- [Embarrassingly Easy Document-Level MT Metrics: How to Convert Any Pretrained Metric Into a Document-Level Metric](https://mlsanity.com/paper/2209.13654/) (2022)
- [Leveraging GPT-4 for Automatic Translation Post-Editing](https://mlsanity.com/paper/2305.14878/) (2023)
- [Measuring and Increasing Context Usage in Context-Aware Machine Translation](https://mlsanity.com/paper/2105.03482/) (2021)
- [When Does Translation Require Context? A Data-driven, Multilingual Exploration](https://mlsanity.com/paper/2109.07446/) (2021)
- [How to Select Datapoints for Efficient Human Evaluation of NLG Models?](https://mlsanity.com/paper/2501.18251/) (2025)
- [Identifying Context-Dependent Translations for Evaluation Set Production](https://mlsanity.com/paper/2311.02321/) (2023)
- [On Measuring Context Utilization in Document-Level MT Systems](https://mlsanity.com/paper/2402.01404/) (2024)
- [Estimating Machine Translation Difficulty](https://mlsanity.com/paper/2508.10175/) (2025)
- [Multi-Dimensional Machine Translation Evaluation: Model Evaluation and Resource for Korean](https://mlsanity.com/paper/2403.12666/) (2024)

## Abstract (arXiv)

Recent calls for harder machine translation benchmarks have not clarified what difficulty should mean. We argue that one meaningful and currently unmeasured axis is referential reach, the distance a segment must look back into its document to resolve the entities and pronouns it contains. We formalize this as discourse dependency (DDP), a metric-free, source-side measure computed from named entity re-mentions and pronominal coreference. Validated against gold coreference, DDP errs one-sidedly in 99.2% of segments, so a high-DDP segment is certified to require long-range context. Applying DDP to WMT24++ and WMT25 shows that both are heavily skewed toward low-DDP segments, which domain labels do not distinguish. Building on DDP, we compare five context injection strategies in an English-Korean post-editing setup, varying context size and selection. As DDP grows, no strategy keeps pace with human post-editing. On segments with DDP >= 15 raters prefer human translations, while automatic metrics register no difference. As frontier systems saturate aggregate scores, DDP shifts evaluation from how well models score to how far they can reach.