Nizar Habash · page 2
Papers
- Unified Guidelines and Resources for Arabic Dialect Orthography (2018 · 118 citations)
Modern Standard Arabic has a fixed spelling standard, but spoken Arabic dialects have none. As a result, one dialectal word can appear online in dozens of forms, which fragments data and hurts NLP systems. Earlier Conventional Orthography for Dialectal Arabic (CODA) work covered one dialect at a time. Its rules were… - ArabicMMLU: Assessing Massive Multitask Language Understanding in Arabic (2024 · 117 citations)
Arabic-capable LLMs are usually tested for knowledge and reasoning on MMLU translated from English. Translation can introduce errors and leaves out content rooted in the Arab world. To fill this gap, the authors built ArabicMMLU, the first MMLU-style benchmark written natively in Modern Standard Arabic (MSA). They… - Automatic Transliteration of Romanized Dialectal Arabic (2014 · 114 citations)
Egyptian Arabic (EGY) is often typed in Latin letters and digits, a practice called Arabizi. Existing EGY tools expect Arabic script, so they cannot read this text. The task is also hard to define, because the dialect has no standard spelling even in Arabic script. The authors fix the target as CODA, a conventional… - Dependency Parsing of Modern Standard Arabic with Lexical and Inflectional Features (2013 · 109 citations)
The paper asks which morphological features help dependency parsing of MSA, using CATiB trees and two transition-based parsers (MaltParser and Easy-First). - NADI 2022: The Third Nuanced Arabic Dialect Identification Shared Task (2022 · 106 citations)
NADI 2022 is the third edition of a shared task on Arabic dialects. It targets fine-grained variation at the country level in short social media posts. The organisers set two subtasks. Subtask 1 asks systems to recognise which of 18 national dialects a tweet comes from. Subtask 2 asks for sentiment (positive… - NADI 2020: The First Nuanced Arabic Dialect Identification Shared Task (2020 · 102 citations)
NADI 2020 is a shared task on recognizing where an Arabic tweet's author comes from based on dialect. It has two subtasks: assigning one of 21 Arab countries (Subtask 1) and one of 100 provinces (Subtask 2). Earlier finer-grained work, such as MADAR, relied on commissioned parallel sentences. NADI instead uses… - Conventional Orthography for Dialectal Arabic (2012 · 100 citations)
Dialectal Arabic (DA) is the everyday spoken Arabic of the Arab world, and it is increasingly written online in blogs, forums, SMS and email. - Don't Throw Those Morphological Analyzers Away Just Yet: Neural Morphological Disambiguation for Arabic (2017 · 97 citations)
A morphological analyzer gives several readings for each Arabic word when taken out of context, and the task is to choose the right one in context. - The Second QALB Shared Task on Automatic Text Correction for Arabic (2015 · 96 citations)
QALB-2015 was the second shared task on automatically correcting errors in Modern Standard Arabic text. Its predecessor, QALB-2014, covered only native-speaker comments on Aljazeera articles. This edition kept that native track and added a second track for essays by learners of Arabic (L2), drawn from the ALWC and… - Building a Corpus for Palestinian Arabic: a Preliminary Study (2014 · 96 citations)
Most Arabic NLP tools target Modern Standard Arabic (MSA). Spoken dialects differ from it in sound, morphology and vocabulary, have no agreed spelling, and appear in writing mostly as informal online chatter. This paper reports early work on Curras, an annotated corpus of Palestinian Arabic (PAL) of roughly 43K words… - Morphological Analysis and Generation for Arabic Dialects (2005 · 96 citations)
The paper presents MAGEAD, a finite-state morphological analyzer and generator for Modern Standard Arabic and its dialects. - The Illinois-Columbia System in the CoNLL-2014 Shared Task (2014 · 94 citations)
This paper presents the Illinois-Columbia entry to the CoNLL-2014 shared task on correcting grammatical mistakes in essays by non-native English writers. - A Conventional Orthography for Algerian Arabic (2015 · 91 citations)
Algerian Arabic (ALG) is a widely spoken Western Arabic dialect with no standard spelling. Speakers write it inconsistently, and NLP tools built for Modern Standard Arabic (MSA) perform poorly on it. The authors propose ALG CODA, an Algerian version of the Conventional Orthography for Dialectal Arabic (CODA), which… - A Conventional Orthography for Tunisian Arabic (2014 · 91 citations)
Tunisian Arabic (TUN) is the everyday spoken variety of Tunisia. It is increasingly written online, but it has no standard spelling, few texts or dictionaries, and it blends into Modern Standard Arabic (MSA) through many mixed forms. Inconsistent spelling makes it hard to build and share NLP resources. The authors… - Developing and Using a Pilot Dialectal Arabic Treebank (2006 · 84 citations)
Arabic dialects are spoken rather than written, so parsers lack dialect training data. For the 2005 JHU workshop on Parsing Arabic Dialects (PAD), the LDC quickly built a pilot Levantine Arabic Treebank (LATB). It covers about 26,000 words of Jordanian conversational telephone speech. The goal was a development and… - A Morphological Analyzer for Gulf Arabic Verbs (2017 · 83 citations)
Most Arabic NLP tools target Modern Standard Arabic (MSA) and perform poorly on dialects. Before this work, no full analyzer for Gulf Arabic (GLF) produced segmentation, POS analysis and lemmas together. The authors build CALIMA GLF, a morphological analyzer and generator for GLF verbs with a focus on Emirati Arabic.… - NADI 2023: The Fourth Nuanced Arabic Dialect Identification Shared Task (2023 · 82 citations)
NADI 2023 is the fourth round of a shared task on Arabic dialects. Arabic is really a family of varieties, and some of them are not mutually intelligible. The organizers set up three subtasks under one common evaluation: - LDC Arabic Treebanks and Associated Corpora: Data Divisions Manual (2013 · 82 citations)
LDC publishes its Arabic treebanks as single undivided collections, while NLP experiments need separate training, development and test portions. - A Corpus for Modeling Morpho-Syntactic Agreement in Arabic: Gender, Number and Rationality (2011 · 82 citations)
The authors add manually annotated functional features to the nominals of PATB part 3: gender, number and rationality (human or not), as opposed to the form-based features the affixes show. - AraBART: a Pretrained Arabic Sequence-to-Sequence Model for Abstractive Summarization (2022 · 80 citations)
Arabic summarization research has mostly produced extractive systems. The few abstractive ones were either trained without pretraining or reused BERT, where only the encoder carries pretrained weights and the decoder and cross-attention start from random values. The authors present AraBART, a BART Base–style seq2seq… - Transliteration of Arabizi into Arabic Orthography: Developing a Parallel Annotated Arabizi-Arabic Script SMS/Chat Corpus (2014 · 80 citations)
Arabic social media text is often typed in Arabizi, an informal Latin-letter romanization that uses digits for sounds Latin lacks (e.g., 3 for ع). - Arabic diacritization in the context of statistical machine translation (2007 · 77 citations)
Arabic script usually omits diacritics, the optional marks for short vowels, gemination and nunation. This leaves words lexically and morphologically ambiguous. The authors ask whether restoring some or all of these marks helps phrase-based SMT from Arabic. Earlier preprocessing work mostly tried to shrink sparsity.… - Automatic Gender Identification and Reinflection in Arabic (2019 · 75 citations)
English often leaves the speaker's gender unexpressed in first-person sentences. Arabic usually has to mark it, so a single-output translator ends up guessing, typically along stereotyped lines. The authors do not try to debias the translation model itself. Instead they add a monolingual post-processing layer that… - A Morphologically Annotated Corpus of Emirati Arabic (2018 · 73 citations)
Dialectal Arabic is the variety people actually speak and increasingly write online, yet its NLP resources lag well behind those for Modern Standard Arabic (MSA). - Arabic Natural Language Processing (2022 · 72 citations)
This is not an experimental paper. It is a short abstract for an introductory three-hour tutorial on processing Arabic, given at EMNLP 2022 by Nizar Habash of the CAMeL Lab at NYU Abu Dhabi. It targets developers and linguists who want to build Arabic tools but lack background on the language, which exists in…
This page in Markdown → https://md.research.elamur.ai/free/en/authors/nizar-habash/page/2/