jaylee8864/korean-vocabulary-5000

Koko Korean 5K — Multilingual Vocabulary Dataset

5,000 carefully curated Korean vocabulary entries with English translations, romanization, contextual usage notes, and example sentences. Each entry is also translated into 9 additional languages, giving researchers and developers a high-quality parallel corpus of 50,000 aligned vocabulary records anchored to Korean.

The dataset reflects real conversation patterns from K-dramas, K-pop, and everyday Korean — not textbook-only material — making it especially useful for training language models and apps that target modern, casual Korean.

Languages

Code	Language
en	English (source)
es	Spanish
pt-BR	Portuguese (Brazilian)
ja	Japanese
zh-TW	Chinese (Traditional)
id	Indonesian
vi	Vietnamese
de	German
fr	French
th	Thai

Korean (ko) is present in every record as korean_term, romanization, and example_sentence_korean.

Use cases

• Train Korean language models (vocabulary acquisition, masked LM, translation)
• Build vocabulary apps and flashcard tools
• Linguistic research on Korean honorifics, romanization, and word categorization
• Translation quality evaluation across 10 target languages
• Cross-lingual retrieval and embedding benchmarks

Data fields

Every record across every language file shares the same schema:

Field	Type	Description
slug	string	URL-safe identifier (stable across all languages)
english_term	string	English meaning (or translated meaning in non-EN files)
korean_term	string	Korean (Hangul) text
romanization	string	Revised Romanization
context_description	string	Cultural and usage context
example_sentence_korean	string	Korean example sentence
example_sentence_english	string	English (or translated) gloss for the example
difficulty	string	Beginner / Intermediate / Advanced
category	string	Topic (Emotions, Greetings, Food, K-pop, etc.)

The slug is consistent across all 10 language files, which lets you join records by slug to assemble parallel multilingual rows.

Quick start

from datasets import load_dataset

# Load just the English source
ds = load_dataset("jaylee8864/korean-vocabulary-5000", "en")
print(ds["train"][0])

# Load Japanese version
ds_ja = load_dataset("jaylee8864/korean-vocabulary-5000", "ja")

Or load raw JSONL directly:

import json
with open("words.en.jsonl") as f:
    rows = [json.loads(line) for line in f]

Categories

Records cover ~80 categories including: Greetings, Emotions, Food, K-pop, K-drama Expressions, Travel, Shopping, Work, Family, Slang, Idioms, Numbers, Body Parts, Weather, Honorifics, and more.

Source & live demo

Curated and audio-paired by Koko AI — an AI-powered Korean conversation tutor. Each vocabulary entry has native pronunciation audio (TTS) and a dedicated learning page on the site.

Browse the full dataset with audio at kokoai.im/word.

License

CC BY 4.0 — free to use commercially or non-commercially with attribution to kokoai.im.

Citation

@dataset{koko_korean_5k_2026,
  author    = {Koko AI},
  title     = {Koko Korean 5K — Multilingual Vocabulary Dataset},
  year      = {2026},
  url       = {https://huggingface.co/datasets/jaylee8864/korean-vocabulary-5000},
  homepage  = {https://kokoai.im}
}