# Researchers release madartb, a parallel Multi-Dialect Arabic dependency treebank

Digest AI · Research · published 2026-10-22T04:00:00Z

Canonical: https://digestai.news/story/researchers-release-madartb-a-parallel-multi-dialect-arabic-dependency

## Summary

A new resource called MADARTB has been introduced for Arabic dependency parsing. It covers five city dialects—Beirut, Cairo, Doha, Rabat and Tunis—plus an existing Modern Standard Arabic version, totaling 5,000 sentences annotated with the CATiB scheme. The authors compare syntactic similarity across the varieties, finding overall high similarity but notable local divergences.

The paper evaluates parsing performance using a supervised parser and four large‑language‑model (LLM) setups. The supervised parser achieves 87.5% labeled attachment score (LAS), while the best LLM configuration reaches 85.7% LAS across all six varieties on the test set. All annotations and the trained models are released publicly for the research community.

## Key points

- MADARTB contains 5,000 sentences annotated in five Arabic city dialects plus MSA using the CATiB scheme.
- Supervised parser reaches 87.5% LAS; the top LLM setup reaches 85.7% LAS across six varieties.
- The treebank and associated models are released publicly for research use.

## Why it matters

Providing a large, parallel treebank for Arabic dialects enables better parsing models, closing a resource gap and improving downstream NLP applications for regional Arabic.

## Sources

1. [A Parallel Multi-Dialect Arabic Dependency Treebank](https://arxiv.org/abs/2610.28823) (arXiv cs.CL, 2026-10-22, primary source)

## Cite

Digest AI, "Researchers release madartb, a parallel Multi-Dialect Arabic dependency treebank", 22 October 2026, https://digestai.news/story/researchers-release-madartb-a-parallel-multi-dialect-arabic-dependency

---

Written by Digest AI's editorial model from the linked sources; the sources are the record. Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse
JSON: https://digestai.news/story/researchers-release-madartb-a-parallel-multi-dialect-arabic-dependency.json