MasakhaPOS: Part-of-Speech Tagging for Typologically Diverse African Languages
Abstract
In this paper, we present MasakhaPOS, the largest part-of-speech (POS) dataset for 20 typologically diverse African languages. We discuss the challenges in annotating POS for these languages using the UD (universal dependencies) guidelines. We conducted extensive POS baseline experiments using conditional random field and several multilingual pre-trained language models. We applied various cross-lingual transfer models trained with data available in UD. Evaluating on the MasakhaPOS dataset, we show that choosing the best transfer language(s) in both single-source and multi-source setups greatly improves the POS tagging performance of the target languages, in particular when combined with cross-lingual parameter-efficient fine-tuning methods. Crucially, transferring knowledge from a language that matches the language family and morphosyntactic properties seems more effective for POS tagging in unseen languages.
Keywords
Cite
@article{arxiv.2305.13989,
title = {MasakhaPOS: Part-of-Speech Tagging for Typologically Diverse African Languages},
author = {Cheikh M. Bamba Dione and David Adelani and Peter Nabende and Jesujoba Alabi and Thapelo Sindane and Happy Buzaaba and Shamsuddeen Hassan Muhammad and Chris Chinenye Emezue and Perez Ogayo and Anuoluwapo Aremu and Catherine Gitau and Derguene Mbaye and Jonathan Mukiibi and Blessing Sibanda and Bonaventure F. P. Dossou and Andiswa Bukula and Rooweither Mabuya and Allahsera Auguste Tapo and Edwin Munkoh-Buabeng and victoire Memdjokam Koagne and Fatoumata Ouoba Kabore and Amelia Taylor and Godson Kalipe and Tebogo Macucwa and Vukosi Marivate and Tajuddeen Gwadabe and Mboning Tchiaze Elvis and Ikechukwu Onyenwe and Gratien Atindogbe and Tolulope Adelani and Idris Akinade and Olanrewaju Samuel and Marien Nahimana and Théogène Musabeyezu and Emile Niyomutabazi and Ester Chimhenga and Kudzai Gotosa and Patrick Mizha and Apelete Agbolo and Seydou Traore and Chinedu Uchechukwu and Aliyu Yusuf and Muhammad Abdullahi and Dietrich Klakow},
journal= {arXiv preprint arXiv:2305.13989},
year = {2023}
}
Comments
Accepted to ACL 2023 (Main conference)