Development of linguistically annotated parallel language resources for four South African languages

For this project, we collected and annotated data to develop language resources for the four official South African Nguni languages written with a conjunctive orthography. The data for these four languages is parallel to allow for comparative (computational) linguistic studies. The corpora have been annotated for three types of linguistic information (morphology, part-of-speech and lemma). The article focuses on the annotation procedure, design choices that were made along the way as well as the quality control steps used. Hopefully this description will give some guidance for similar projects on under-resourced languages in the future.