← Briefs
Technology & DataOpen
Create a small, trustworthy dataset for a Nigerian language
Build and document a carefully checked sample dataset that could improve a language tool.
- Brief type
- Training
- Level
- Intermediate
- Time
- 2 weeks
- Format
- Solo or team
The brief
WHAT YOU ARE MAKING
A small, documented language dataset for Hausa, Yoruba, Igbo or another Nigerian language you know well. Choose one useful task, such as common greetings, public-service phrases, short translations or speech samples.
START HERE
Choose one language and one narrow use case. Work with at least two fluent speakers. Agree on spelling, dialect and meaning before collecting data.
YOUR SUBMISSION
Provide 100 to 300 checked entries in CSV or spreadsheet format, a one-page data card explaining how the entries were collected, the dialect or variety represented, consent, known limitations and how someone may reuse the dataset.
KEEP IT SAFE AND HONEST
Only record people who have freely agreed. Do not collect names, phone numbers or private conversations. Do not scrape copyrighted material or claim that a small sample represents every speaker or dialect.
GOOD WORK LOOKS LIKE
The files are tidy, another person can understand every column, fluent speakers have checked the entries, and the limitations are stated plainly.
Skills you will practise
Data collectionLanguage researchQuality assurance