MudawanSn: A Gold-Standard Wolof-Arabic Parallel Corpus for Machine Translation

MudawanSn is a publicly available parallel corpus containing 1,271 sentence-aligned pairs for the Wolof–Arabic language pair. The corpus consists of manual translations from Wolof into Modern Standard Arabic, with source texts drawn from the MasakhaNER corpus. It was created by author mbaye930 and last updated on the platform in June 2026.

Use Cases

• Train machine translation models based on the 1,271 gold-standard sentence pairs.
• Benchmark translation quality for the Wolof-Arabic language pair based on the manually translated corpus.
• Analyze news discourse patterns across languages based on texts covering politics, society, religion, and sports.
• Study linguistic features and translation challenges for Modern Standard Arabic based on translations from Wolof.

Strengths

• Contains 1,271 manually translated, sentence-aligned pairs, constituting a gold-standard resource.
• Source texts are drawn from the established MasakhaNER corpus, covering multiple domains of Senegalese news.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for large-scale model training.
• Last updated 2026-06-11 16:52:35; freshness should be verified.

Provenance

Source

Source texts are drawn from the MasakhaNER corpus.

Collection Method

Consists of manual translations from Wolof into Modern Standard Arabic.

Freshness

Last updated 2026-06-11 16:52:35.

Geography

The source texts cover Senegalese news discourse.