A Conventional Orthography for Algerian Arabic

Houda Saadane, Nizar Habash

Introduction

Algerian Arabic (ALG) is a vibrant dialect spoken by millions, yet it remains an "under-resourced" language in the field of Natural Language Processing (NLP). This paper introduces a standardized writing system known as CODA (Conventional Orthography for Dialectal Arabic) specifically tailored for the Algerian dialect. By establishing a consistent mapping between the spoken language and the Arabic script, the authors provide the linguistic foundation necessary to develop effective digital tools like machine translation, sentiment analysis, and speech recognition.

The Challenge of Polyglossia

The primary hurdle in processing Algerian Arabic is polyglossia. While Modern Standard Arabic (MSA) is the formal written standard, dialects like ALG are the primary mode of daily communication and social media interaction.

ALG differs significantly from MSA across phonology, morphology, and vocabulary. Furthermore, because ALG lacks a standardized spelling system, users write it inconsistently online, often using spontaneous phonetic spellings or "Arabizi" (Arabic written with Latin characters and numbers). This inconsistency makes it difficult for traditional NLP tools, designed for the rigid structure of MSA, to process dialectal text accurately. To build better tools, researchers first need a "conventional" way to write the dialect that is consistent, machine-readable, and preserves its unique features.

Methodology: The CODA Framework

To address this, the authors extended the CODA framework—previously applied to Egyptian and Tunisian Arabic—to the specific needs of the Algerian dialect. CODA is an ad hoc convention that utilizes the standard Arabic script but applies specific rules to handle dialectal variations while maintaining etymological links to MSA where possible.

Morphological Standardization

One of the most complex tasks is handling the agglutinative nature of the dialect. ALG features a variety of prefixes and suffixes that do not exist in MSA, such as specific passive voice markers (t- or tn-) and unique enclitics. CODA provides rules for segmenting these parts to ensure machine readability and structural consistency.

The morphological breakdown of an Algerian Arabic word into proclitics, stem, suffixes, and enclitics.

Lexical Guidelines

Algerian Arabic is a linguistic melting pot, incorporating a significant number of loanwords from Berber, Turkish, Italian, Spanish, and French. The CODA framework establishes guidelines for spelling these borrowed terms consistently, ensuring they are recognized as part of the standard dialectal lexicon rather than foreign intrusions.

Examples of Algerian Arabic words showing diverse etymological origins including Berber, Turkish, Italian, Spanish, and French.

Key Linguistic Insights

The development of the Algerian CODA Map revealed several unique features of the dialect that require specialized orthographic handling:

  • The "Qaf" Multiplicity: The letter qaf (ق) is particularly versatile in ALG, representing sounds like /q/, /g/, /k/, or even a glottal stop depending on regional variation. CODA retains the MSA qaf to preserve the word's etymological root while acknowledging this phonetic diversity.
  • Consonant Clusters: Unlike many Middle Eastern dialects, Algerian Arabic frequently drops short vowels at the start of words (elision), leading to complex consonant clusters (e.g., the MSA qatal becomes the ALG qtal). CODA rules are designed to represent these "fςal" verb schemes accurately.
  • Grammatical Simplification: The dialect often collapses complex MSA structures. For instance, the dual form and the feminine plural are frequently merged into the masculine plural, a shift that CODA must formalize for accurate parsing and translation.
  • The "N" of Number: A distinctive feature of ALG is the addition of an /n/ sound to numbers 11–19 when they precede a noun (e.g., sTaAšn TaAblaħ for "16 tables"). CODA formalizes this "N of Number Construct" to ensure syntactic consistency.

Conclusion and Future Work

The primary result of this research is the Algerian CODA Map, a comprehensive set of orthographic guidelines. This system allows researchers to standardize spontaneous, inconsistent input into a uniform format that preserves the dialect's linguistic identity while enabling computational processing.

While the current map focuses on the Algiers dialect as a default, future work will explore how these rules generalize across other regional variations like Oranais or Saharan Arabic. The ultimate goal is to use these guidelines to annotate large-scale corpora, paving the way for robust, dialect-specific NLP models that can finally serve the millions of Algerian Arabic speakers worldwide.

Summary updated

BibTeX
@inproceedings{saadane2015conventional,
  author = {Houda Saadane and Nizar Habash},
  title = {{A Conventional Orthography for Algerian Arabic}},
  booktitle = {ANLP @ ACL},
  year = {2015},
  doi = {10.18653/v1/w15-3208},
}
Figure