[bambozya]

This adds Lahga to the catalogue: an open, collaborative dictionary of spoken Arabic. Each Modern Standard Arabic headword lists the forms the dialects use for it, each attributed to one variety, with usage notes and example sentences.

• Dataset: https://huggingface.co/datasets/lahga/arabic-dialects
• Site and current files: https://lahga.fyi/data
• 25,072 dialect forms under 2,012 headwords, across 24 varieties; CC BY-SA 4.0
• Metadata JSON: datasets/lahga.json

Two things a reviewer should know:

• There is no paper. The paper and venue fields are empty.
• More than half of the forms were first drafted by a language model and read by the maintainer before import. Source lists LLM and Annotation Style lists LLM annotation for that reason. The 3,606 forms no speaker has confirmed yet carry needs_review: true in the data, and the dataset card explains what the flag does and does not promise. Palestinian forms come from Maknuune and part of the Egyptian, Levantine, Hejazi, Gulf, Moroccan and Tunisian forms from Wiktionary, hence Derived From.

The volumes in Dialect Subsets are counts of forms per country as of 2026-10-02. I made the file by hand from the fields of recent entries, not through the Masader form; tell me if a field should be different and I will fix it.

🤖 Generated with Claude Code

[bambozya]

Hi @zaidalyafeai , a gentle ping on this one when you have time. I made the file by hand rather than through the form. If the form is easier for you to review, I'm happy to resubmit it that way, or to change any field. Thanks for maintaining Masader :)