ghananlpcommunity/pristine-twi

This dataset is shared under CC BY-NC 4.0, which means you are free to use, share, and adapt it for non-commercial research and educational purposes with attribution. You can read the full license at https://creativecommons.org/licenses/by-nc/4.0/.

Pristine Twi Dataset

A large-scale Twi language dataset containing clean and naturally sounding Twi text across four distinct styles — monologue, narrative, dialogue, and storytelling — generated from real Ghanaian news topics as inspiration to keep it grounded on Ghanaian named entities and vocabulary.

This dataset was built to support the development of Twi language models, tokenizers, and other NLP tools for the Akan language family.

Dataset Details

Field	Detail
Language	Twi (tw) — Akan, Ghana
License	CC BY-NC 4.0
Author	Mich-Seth Owusu
Organization	Ghana NLP Community
Format	Parquet (columnar)

Columns

Column	Type	Description
twi	string	Twi text snippet
style	string	Monologue, Narrative, Dialogue, Storyful
char_count	int	Character count of the Twi text

Text Styles

The dataset contains Twi text categorized into four distinct styles:

• Monologue (ɔkasafo biako) — passionate, personal perspective
• Narrative (abakɔsɛm) — journalistic, structured account
• Dialogue (nkɔmmɔdie) — conversational exchange
• Storyful (anansesɛm) — dramatic, metaphorical storytelling

Generation

Text was generated using the Gemini API with Ghanaian news paragraphs as prompts, separated into stylistic chunks, then cleaned to remove speaker labels, markdown artifacts, and any responses that failed Twi language validation.

The code for generating the dataset can be found here: https://github.com/GhanaNLP/NLP-scripts/blob/main/text/generate_twi-gemini.py

---

Made with ❤️ by Mich-Seth Owusu for the Ghana NLP Community