arabictyper
About the words

Acknowledgements.

The frequency dictionary on this site is built from two open sources. Both ask for credit and a link, and both are share-alike, so the word list we made from them carries the same licence: CC BY-SA 4.0.

Wiktionary, through kaikki.org

Headwords, vocalization, definitions, roots, plurals and feminine forms come from the English Wiktionary’s Arabic entries. So do the inflection tables we use to recognise a word’s conjugated and declined forms. We read them from the machine-readable extract that Tatu Ylonen’s Wiktextract project publishes at kaikki.org.

  • Text: Wiktionary contributors, licensed under CC BY-SA 4.0. Each word page links to its own entry.
  • Extract: Tatu Ylonen, Wiktextract: Wiktionary as Machine-Readable Structured Data, Proceedings of LREC 2022, pp. 1395–1403.

FrequencyWords, from OpenSubtitles

The order of the list, and the rank shown on each page, follow how often a word appears in Arabic film and TV subtitles. The counts are Hermit Dave’s FrequencyWords list for Arabic (2018 edition), derived from the OpenSubtitles2018 corpus that OPUS distributes.

  • Content: CC BY-SA 4.0.
  • Underlying corpus: OpenSubtitles.org, via P. Lison and J. Tiedemann, OpenSubtitles2016: Extracting Large Parallel Corpora from Movie and TV Subtitles, LREC 2016.

What we changed

Wiktionary’s fully vocalized spellings are trimmed to the lighter style used across this site: no fatḥa before a long alif, and tanwīn written after its alif. Long definitions are cut to two or three glosses, with the everyday sense first. Arabizi spellings are generated by the site’s own rules. Frequency ranks come from matching subtitle words to headwords and their inflection tables, which works well for most words and imperfectly for a few. Corrections are made by hand as we find them.

Reusing the list

You are welcome to reuse the dictionary under the same CC BY-SA 4.0 terms: credit Wiktionary, FrequencyWords and arabictyper.com, link to the licence, and share anything you build on it under the same licence. If you want the list as a file, get in touch.