Yoruba is a major language of West Africa, and written Yoruba can use tone marks and underdots that distinguish sounds and meanings. A language-data project should define how these marks are expected to appear in its target text and how reviewers should handle inconsistent source material.
Build conventions into the specification
For transcription and text collection, document orthography, diacritics, punctuation, borrowed words, names, and acceptable normalization. If the downstream application depends on tone information, removing diacritics may discard important distinctions. If the application uses a normalized form, explain the transformation and preserve source material where required.
Annotation and evaluation
Intent, sentiment, entity, or translation tasks need examples that reflect the intended domain and language use. Reviewers should receive guidance on ambiguity, code-switching, spelling variants, and disagreements. A small pilot can reveal unclear instructions before a large collection begins.
Organizations seeking Yoruba language data, Yoruba transcription, or Yoruba annotation should state the task, target variety, script conventions, domain, volume, and quality threshold. Langigs can review the brief and confirm whether the requested scope is feasible.
