AbjadNLP 2027 · Shared Task 04

AbjadTranslit 2027

Arabic–Latin cross-script transliteration

Sentence-level conversion between Arabizi and Arabic script using parallel text data.

One language. Two writing systems.

Arabizi · Latin script mar7aba bikom
⇅
Arabic script مرحبا بكم

Illustrative example of the two writing systems.

About the shared task

Connecting Arabizi and Arabic script

AbjadTranslit 2027 focuses on sentence-level transliteration between Arabic written in Latin characters, commonly known as Arabizi, and Arabic script. The task uses parallel text to study conversion in both directions.

Latin → Arabic

Arabizi to Arabic script

Given a sentence written in Arabizi, produce the corresponding sentence in Arabic script.

Arabic → Latin

Arabic script to Arabizi

Given a sentence in Arabic script, produce the corresponding sentence in Arabizi.

Parallel text, sentence by sentence

The shared task centres on paired Arabizi and Arabic-script sentences. Dataset access, release details and licensing information will be announced here.

Taking part

Participation details

The task description is available now. Please check this page for data releases, evaluation guidance and participation instructions as they are confirmed.

Visit the AbjadNLP 2027 workshop page →
Data access and licensing
To be announced
Evaluation protocol and metrics
To be announced
Registration and submissions
To be announced
Shared-task dates
To be announced

The team

Shared-task organisers

  • Dr Pranav Gupta
  • Niranjan Kumar M
  • Balaji Nagarajan
  • Dr Imed Zitouni
  • Dr Mo El-Haj