A Comprehensive Analysis of Arabic Natural Language Processing Research: Trends, Topic Evolution, and Research Gaps -- A Bibliometric and Topic-Based Study

Mullosharaf K. Arabov

cs.CLAug 24, 2026

A Comprehensive Analysis of Arabic Natural Language Processing Research: Trends, Topic Evolution, and Research Gaps -- A Bibliometric and Topic-Based Study

Authors: Mullosharaf K. Arabov

Organizations: Kazan Federal University, Institute of Computational Mathematics and Information Technologies, Kazan, Russia

Abstract

Arabic Natural Language Processing (NLP) has grown rapidly over the past decade, driven by digital transformation in the Arab world, social media, and large language models (LLMs). Despite this growth, a comprehensive quantitative meta-analysis remains absent. This study presents a bibliometric and topic-based analysis of 7,120 Arabic NLP papers published between 1960 and 2026, sourced from five platforms (arXiv, ACL Anthology, Semantic Scholar, Crossref, OpenAlex) plus an additional targeted OpenAlex subset. We employ BERTopic for topic modeling, regression analysis, social network analysis, and geographic mapping. Our findings show a significant publication surge after 2020, driven by transformer models and LLMs. Topic modeling identifies 19 themes, the largest centered on text, speech, translation, and recognition (2,942 papers). Citation analysis reveals a positive correlation between paper age and citations (r = 0.245, p < 0.001); regression (R^2 = 0.105) shows that indexing in OpenAlex or Semantic Scholar and institutional affiliation are associated with higher citations. Saudi Arabia, the United States, and Egypt lead in research output. A task-dialect gap matrix identifies understudied areas, including summarization for Maghrebi, Iraqi, and Sudanese dialects. The largest topic has the highest H-index (90), followed by sentiment analysis (57). Our quantitative approach complements existing qualitative surveys and offers recommendations to prioritize under-resourced dialects and develop culturally aligned benchmarks.

Figures & tables

Figure 1: Overview of the methodological pipeline.

Source	Relevant records loaded ( $\geq 2$ terms)
arXiv	613
ACL Anthology	657
Semantic Scholar	6,865
OpenAlex	4,054
OpenAlex (extra)	206
Crossref	3,985

Table 1: Data sources and integration statistics.

Statistic	Value
Total papers	7,120
Unique authors (normalized)	14,805
Unique DOIs	6,114
Year range	1960–2026
Papers with citation data	4,877 (68.5%)
Papers with country information	2,568 (36.1%)

Table 2: Descriptive statistics of the final corpus.

Figure 2: Publications by year with key events (AraBERT in 2018, ChatGPT/LLM boom in 2023).

Author	Papers
Nizar Habash	151
Muhammad Abdul-Mageed	95
Mona Diab	86
Wajdi Zaghouani	57
Ahmed Ali	55
AbdelRahim Elmadany	54

Table 3: Top-20 most productive authors (normalized names).

Figure 3: Publication activity of top-5 authors over years.

Topic ID	Topic Name	Count
0	text, speech, translation, recognition	2,942
1	sentiment, sentiment analysis, detection, social	614
2	patients, version, validity, questionnaire	305
3	algerian, algeria, french, dialect	137
4	ner, entity, named, named entity	107
5	similarity, identification, relatedness, dialect identification	65

Table 4: BERTopic topics with paper counts (after reduction to 20 topics; topic -1 omitted). The outlier topic (-1) contains 2,535 papers (35.6% of the corpus); the sum of the listed topics is 4,585.

Figure 4: Trends of major topics over years.

Metric	BERTopic	LDA
Topic Diversity	0.868	0.720
Topic Coherence	0.772	0.028

Table 5: Topic quality metrics: BERTopic vs. LDA.

Rank	Paper	Citations
1	AraBERT: Transformer-based Model for Arabic Language Understanding	1,480
2	Translation Techniques Revisited	1,023
3	Modern Arabic: Structures, Functions and Varieties (Holes)	771
4	ARBERT & MARBERT: Deep Bidirectional Transformers for Arabic	714
5	The Phonology and Morphology of Arabic	688
6	Arabic Natural Language Processing (survey)	551

Table 6: Top-10 most cited papers.

Figure 6: Average citations by topic (Top-10).

Predictor	Coefficient	p-value
year_c	-1.234	<0.001
source_Semantic_Scholar	+5.455	<0.001
source_OpenAlex	+11.086	<0.001
has_institutions	+8.730	<0.001
source_OpenAlex_(extra)	-9.042	0.004
matched_terms_count	+1.208	0.044

Table 7: Significant predictors of citation count (p < 0.05).

Figure 7: Geographic distribution of Arabic NLP publications by country of affiliation.

Country	Code	Affiliations
Saudi Arabia	SA	519
United States	US	463
Egypt	EG	266
Jordan	JO	253
United Kingdom	GB	212
United Arab Emirates	AE	182

Table 8: Top-10 countries by affiliation count.

Figure 8: Top-20 institutions by affiliation count.

Dialect	Mentions
MSA (Modern Standard)	1,553
Egyptian	462
Maghrebi (Darija)	341
Levantine (Shami)	187
Gulf (Khaliji)	105
Iraqi	79

Table 9: Dialect mentions in abstracts (total counts).

Figure 10: Dialects and major topics: trends over years.

Task	MSA	Egyptian	Levantine	Gulf	Maghrebi	Iraqi	Sudanese	Yemeni	Najdi
sentiment	53.9	16.6	2.4	4.7	15.9	2.7	2.7	0.3	0.7
translation	55.3	19.3	5.5	4.6	10.7	3.8	0.2	0.0	0.6
ner	58.0	15.7	6.1	4.0	10.4	2.6	0.4	1.2	1.5
asr	51.3	24.2	5.1	4.0	11.2	2.5	0.4	0.4	1.1
dialect_id	56.0	26.6	4.3	2.7	8.7	0.5	0.0	0.0	1.1

Table 10: Task–dialect coverage matrix for core NLP tasks (row-normalized, %).

Task	MSA	Egyptian	Levantine	Gulf	Maghrebi	Iraqi	Sudanese	Yemeni	Najdi	Hejazi	Hassaniya
ocr	37.5	25.0	4.2	0.0	33.3	0.0	0.0	0.0	0.0	0.0	0.0
summarisation	76.9	15.4	0.0	7.7	0.0	0.0	0.0	0.0	0.0	0.0	0.0

Table 11: Task–dialect coverage for specialised tasks (row-normalized, %).

Topic	CAGR (%)
text, speech, translation, recognition	11.54
patients, version, validity, questionnaire	5.63
sentiment, sentiment analysis, detection, social	3.69
ner, entity, named, named entity	-5.26
algerian, algeria, french, dialect	-9.80

Table 12: Compound Annual Growth Rate (CAGR) for top topics (last 5 years).

Concept	Frequency
Linguistics	2,998
Philosophy	2,899
Arabic	2,823
Computer science	2,728
Artificial intelligence	2,262
Natural language processing	2,241

Table 13: Top-20 concepts by frequency.

Matched Term	Frequency
arabic	6,981
arabic language	2,996
arabic dialect	1,455
modern standard arabic	1,329
dialectal arabic	457
arabic dataset	447

Table 14: Top-20 matched terms by frequency.

Degree Centrality		Betweenness Centrality
Author	Degree	Author	Betweenness
Mustafa Jarrar	0.166	Nizar Habash	0.025
Nizar Habash	0.165	Marwan Torki	0.020
Muhammad Abdul-Mageed	0.163	Allan Ramsay	0.019
Omer Nacar	0.160	Khaled Shaalan	0.019
AbdelRahim Elmadany	0.159	Bashar Talafha	0.018

Table 15: Top-7 authors by degree and betweenness centrality.

Survey	Key Claim	Empirical Support
Shi and Agrawal (2025)	Sentiment analysis is a dominant theme	Topic 1 is second-largest (8.6%) with H-index 57
Iwidat and Abu Helou (2026)	Shift from ML to transformers/LLMs	Approximately 82% of papers postdate 2020
Dahou et al. (2025)	Dialect identification is foundational	Present as a specialized topic (Topic 5) in our model
Hamed et al. (2025)	Code-switching is under-studied	No dedicated topic; subsumed under Maghrebi studies
Alzubaidi et al. (2025)	Arabic LLM benchmarks are emerging	Early signs of LLM-related research (Topic 17) with 13 papers
Mashaabi et al. (2026)	Arabic LLMs focus on MSA	MSA dominates; Hejazi and Hassaniya nearly absent

Table 16: Alignment between our findings and existing surveys.

Arabic NLP

Arabic natural language processing develops methods and resources for understanding and generating Arabic across dialects, domains, and media.

See all 104 papers

Oct 9, 2026

Towards the Roof of the World: NLP Progress in Tibetan Languages

Tibetan, spoken by some seven million people on and around the Qinghai-Xizang Plateau, remains underrepresented in natural language processing (NLP). We present, to our knowledge, the first peer-reviewed systematic bibliometric survey of Tibetan NLP: 140 publications from 2004 to 2026 in the ACL Anthology and DBLP, annotated for venue type, research area, first-affiliation country, and resource release. Output has grown rapidly, with 84% of the publications appearing since 2018, and has expanded bottom-up along the NLP pipeline, from corpora and word segmentation through script, speech, and applications to large language models (LLMs). Researchers based in China first-authored 87% of the publications, which span all 25 represented research areas, whereas work from other countries concentrates on Classical Tibetan. We argue that this full-pipeline paradigm, realized under the compounded constraints of a complex script, dialectal diversity, and geographic remoteness, offers a reference model for other low-resource languages of the Global South. We also identify gaps in dialect coverage, responsible NLP, native and comparable evaluation, cross-community visibility, and resource release (only 29% of the publications release data, code, or models), and recommend open science, shared standards, and community participation for the LLM era.

Xi Cao, Yijie Li, Yuan Sun +3

School of Philosophy and Religious Studies, Minzu University of China · Institute of National Security, Minzu University of China · National Language Resource Monitoring & Research Center | Minority Languages Branch +3

Low-Resource Language Processing Bibliometric Analysis

Ask PDF

Feb 18, 2026

Aladdin-FTI @ AMIYA Three Wishes for Arabic NLP: Fidelity, Diglossia, and Multidialectal Generation

Arabic dialects have long been under-represented in Natural Language Processing (NLP) research due to their non-standardization and high variability, which pose challenges for computational modeling. Recent advances in the field, such as Large Language Models (LLMs), offer promising avenues to address this gap by enabling Arabic to be modeled as a pluricentric language rather than a monolithic system. This paper presents Aladdin-FTI, our submission to the AMIYA shared task. The proposed system is designed to both generate and translate dialectal Arabic (DA). Specifically, the model supports text generation in Moroccan, Egyptian, Palestinian, Syrian, and Saudi dialects, as well as bidirectional translation between these dialects, Modern Standard Arabic (MSA), and English. The code and trained model are publicly available.

Jonathan Mutal, Perla Al Almaoui, Simon Hengchen +1

Faculté de traduction et d’interprétation, Université de Genève

Arabic NLP Arabic Dialects Arabic Language Models Machine Translation Multilingual Language Models Low-Resource Language Processing

Ask PDF

Oct 5, 2026

DialectSentEval 2026: Arabic Dialect Sentiment Analysis and Swapping Shared Task

Sentiment analysis is a fundamental problem in Natural Language Processing (NLP). Standard sentiment classification for the Arabic language remains challenging due to the high volume of dialectal Arabic. To advance research in this area, this paper proposes the Shared Task on Sentiment Analysis and Swapping in Arabic Dialects (DialectSentEval), hosted with the Arabic Natural Language Processing Conference (ArabicNLP 2026). This shared task consists of two subtasks: Subtask 1 focuses on multi-class and multi-dialect sentiment analysis, requiring models to identify sentiment polarity across various Arabic dialects. Subtask 2 introduces a generative task for Arabic sentiment swap, challenging models to invert sentiment polarity while preserving core semantics. In this overview paper, we present the motivation, dataset creation, and summarize the main findings from participating models.

Saad Ezzini, Shadi Abudalfa, Maram Alharbi +6

King Fahd University of Petroleum and Minerals, KSA · Onaizah Colleges, KSA · University College of Applied Sciences, Palestine +5

Arabic NLP Sentiment Analysis Arabic Dialects

Ask PDF

Abstract

Arabic NLP

Explore similar work

Towards the Roof of the World: NLP Progress in Tibetan Languages

Aladdin-FTI @ AMIYA Three Wishes for Arabic NLP: Fidelity, Diglossia, and Multidialectal Generation

DialectSentEval 2026: Arabic Dialect Sentiment Analysis and Swapping Shared Task