Arabic1800s

A curated corpus of nineteenth-century Arabic texts for historical language-model research.

The collection currently includes books and collected newspaper articles. It aims to expand across regions, authors, genres and perspectives.

Preprocessing

Document-specific Python rules remove publisher material, selected footnotes, page references and formatting artifacts while preserving historical wording, diacritics and paragraph boundaries.

Limitations

The corpus is a work in progress and is not representative of all nineteenth-century Arabic-speaking communities. Some works contain earlier quotations, historical prejudices and possible transcription errors.

Sources and licensing

Sources include Hindawi and the Arabic E-Book Corpus:

Hallberg, Andreas (2024). The Arabic E-Book Corpus (Version 1). https://doi.org/10.5878/7rbh-gy93

CC BY 4.0 applies to this project's contributions where applicable. Source-specific rights and attribution remain in effect; original public-domain texts remain public domain.