# TTLab at AlexandriaX-2026: A Fine-Tuned Surface Tagger for Arabic Machine-Translation Error-Span Detection and Classification

arXiv: https://arxiv.org/abs/2609.29633 · HTML: https://mlsanity.com/paper/2609.29633/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 12 of its 18 references that are in the map.
- First posted on arXiv 24 Sep 2026.

## Builds on (most cited first, 12 of 12)

- [AraBERT: Transformer-based Model for Arabic Language Understanding](https://mlsanity.com/paper/2003.00104/) (2020)
- [Experts, Errors, and Context: A Large-Scale Study of Human Evaluation for Machine Translation](https://mlsanity.com/paper/2104.14478/) (2021)
- [xCOMET: Transparent Machine Translation Evaluation through Fine-grained Error Detection](https://mlsanity.com/paper/2310.10482/) (2023)
- [GECToR -- Grammatical Error Correction: Tag, Not Rewrite](https://mlsanity.com/paper/2005.12592/) (2020)
- [ARBERT & MARBERT: Deep Bidirectional Transformers for Arabic](https://mlsanity.com/paper/2101.01785/) (2020)
- [GEMBA-MQM: Detecting Translation Quality Error Spans with GPT-4](https://mlsanity.com/paper/2310.13988/) (2023)
- [The Interplay of Variant, Size, and Task Type in Arabic Pre-trained Language Models](https://mlsanity.com/paper/2103.06678/) (2021)
- [NileChat: Towards Linguistically Diverse and Culturally Aware LLMs for Local Communities](https://mlsanity.com/paper/2505.18383/) (2025)
- [SaudiBERT: A Large Language Model Pretrained on Saudi Dialect Corpora](https://mlsanity.com/paper/2405.06239/) (2024)
- [Advancing Dialectal Arabic to Modern Standard Arabic Machine Translation](https://mlsanity.com/paper/2507.20301/) (2025)
- [LQM: Linguistically Motivated Multidimensional Quality Metrics for Machine Translation](https://mlsanity.com/paper/2604.18490/) (2026)
- [AlexandriaX 2026: The First Shared Task on Dialectal Arabic Machine Translation](https://mlsanity.com/paper/2609.22796/) (2026)

## Abstract (arXiv)

We present TTLab's submission to the AlexandriaX-2026 Subtask~3 on Arabic MT error span detection and classification. Our system frames the task as token-level classification over surface forms, preserving character offsets to ensure exact alignment with the evaluation metric. To handle severe label imbalance, we employ a focal loss with class weighting and dialect-specific decoding thresholds. Among six Arabic pre-trained encoders, MARBERTv2 achieves the best overall performance of 40.8 and 40.91 on the development and test set, respectively, ranking nth3 out of all participating teams. While our system localizes error spans effectively, classification of rare error types remains challenging, highlighting the need for data augmentation for tail categories. The code is available at hrefhttps://github.com/ENTAILab/arabic-dialectal-mt-error-span-detectionfaGithub~ TTLab at AlexandriaX-2026