Igbo is spoken primarily in southeastern Nigeria and includes regional varieties. Language technology projects should define the variety and domain they need, especially when data will be used for speech recognition, machine translation, search, or conversational systems.
Plan the data before collection
Specify the intended modality, collection source, contributor criteria, consent, permitted use, target volume, metadata, and acceptance rules. For text, document spelling and normalization conventions. For speech, define audio quality, speaker labels, transcript format, and the treatment of repetitions, mixed-language speech, and unclear segments.
Make annotation reviewable
Annotation instructions should explain labels, edge cases, examples, and how to flag uncertainty. A pilot sample and second review can expose confusing instructions and systematic disagreements. Track changes to the guidelines so the final dataset has a clear audit trail.
For Igbo language data collection, transcription, or annotation, a useful request describes the exact use case and the regional, technical, and quality constraints. Langigs can discuss the requirements; specific availability and capacity must be confirmed.
