language:
- ar
license: cc-by-sa-4.0
task_categories:
- text-classification
- token-classification
- feature-extraction
- translation
- question-answering
task_ids:
- language-modeling
- named-entity-recognition
- semantic-similarity
pretty_name: Hadrami Arabic Dialect Dataset
dataset_info:
version: 2.0.0
size_categories:
- 1K<n<10K
annotations_creators:
- expert-generated
language_creators:
- expert-generated
multilinguality:
- monolingual
source_datasets:
- original
tags:
- arabic
- dialect
- nlp
- linguistics
- low-resource
- yemen
- hadramawt
- arabic-nlp
- lexicon
- semantic-intent
- transliteration
- rag
- dialect-translation
- yemeni-arabic
- hadrami
- morphology
- lexicography
thumbnail: cover.png
Hadrami Arabic Dialect Dataset
A structured dataset of 1,000 entries from the Hadrami Arabic dialect (spoken primarily in the Hadramawt region of Yemen). Each entry includes the dialectal word alongside its Modern Standard Arabic (MSA/Fusha) equivalent, linguistic metadata, usage examples, proverbs, and semantic tags.
🔊 Roadmap: Audio pronunciations for each entry are planned for a future release.
Dataset Summary
| Field | Value |
|---|---|
| Entries | 1,000 |
| Language | Arabic (Hadrami dialect + MSA) |
| Source | قاموس اللهجة الحضرمية — فهد بن هلابي |
| Format | CSV / JSON |
| License | CC BY 4.0 |
| Task Tags | NLP, Dialectal Arabic, Lexicography, Low-resource Language |
Schema
| Column | Type | Description |
|---|---|---|
id |
int | Unique entry identifier |
word_vocalized |
string | Dialectal word with full diacritics (تشكيل) |
word_clean |
string | Dialectal word without diacritics |
root |
string | Arabic trilateral/quadrilateral root. May be null for loanwords or proper names |
pos |
string | Part of speech (see POS values below) |
fusha_equivalent |
string | Modern Standard Arabic equivalent word or phrase |
definition |
string | Definition in MSA |
region |
string | Geographic usage region within Hadramawt |
synonyms |
JSON array | List of dialectal synonyms |
phonetic_variants |
JSON array | Alternative pronunciations |
note |
string / null | Linguistic or usage note. null if not applicable |
source |
string | Source reference (book title, author, page) |
examples |
JSON array | Usage examples (see structure below) |
proverbs |
JSON array | Related proverbs in dialect |
tags |
JSON array | Semantic category tags |
examples field structure
Each example is a JSON object with two keys:
{
"f": "الجملة بالفصحى",
"h": "الجملة بالحضرمية"
}
f— sentence in Modern Standard Arabic (Fusha)h— equivalent sentence in Hadrami dialect
POS values
Noun, Verb, Adjective, Adverb, Expression, Preposition, Particle, Noun/Verb, Verb/Adjective, Verb/Phrase, Noun/Adjective
Notable Statistics
| Field | Coverage |
|---|---|
| Entries with examples | 1,043 / 1,000 (99.6%) |
| Entries with synonyms | 742 / 1,000 (70.9%) |
| Entries with proverbs | 89 / 1,000 (8.5%) |
| Entries with notes | 620 / 1,000 (59.2%) |
note nulls |
427 — intentionally null, not missing |
Top Semantic Tags
Daily Life · Nature · Agriculture · Social Customs · Architecture · Food · Clothing · Tools · Expressions · Home · Water · Weather · History
Usage (Python)
import pandas as pd
import json
df = pd.read_csv("hadrami_dictionary_v2.csv")
# Parse JSON array columns
for col in ["synonyms", "phonetic_variants", "examples", "proverbs", "tags"]:
df[col] = df[col].apply(json.loads)
# Example: get all nouns
nouns = df[df["pos"] == "Noun"]
# Example: get entries tagged with Food
food = df[df["tags"].apply(lambda t: "Food" in t)]
Loading from HuggingFace
from datasets import load_dataset
ds = load_dataset("saeedbark/hadrami-arabic-dialect-dataset")
Citation
If you use this dataset, please cite the original source:
فهد بن هلابي. قاموس اللهجة الحضرمية. (Fahd bin Hilabi, Dictionary of the Hadrami Dialect)
Contributing
Found an error or want to add entries? Open an issue or pull request on the dataset repository.
Changelog
v2.0.0 (2026-05-04)
- Added field (NLP semantic role classification)
- Added field (Latin script for each entry)
- Added field (pre-built Arabic full-text search blob)
- Normalized all and values
- Tags converted to lowercase for consistency
- 1,000 verified entries (quality-reviewed subset)
v1.0.0
- Initial release with 1,047 entries