Abstract
Multi-modal techniques offer significant untapped potential to unlock improved NLP technology for local languages. However, many advances in language model pre-training are focused on text, a fact that only increases systematic inequalities in the performance of NLP tasks across the world’s languages. In this work, we propose a multi-modal approach to train language models using whatever text and/or audio data might be available in a language. Initial experiments using Swahili and Kinyarwanda data suggest the viability of the approach for downstream Named Entity Recognition (NER) tasks, with models pre-trained on phone data showing an improvement of up to 6% F1-score above models that are trained from scratch. Preprocessing and training code will be uploaded to https://github.com/sil-ai/phone-it-in.- Anthology ID:
- 2022.acl-long.364
- Volume:
- Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
- Month:
- May
- Year:
- 2022
- Address:
- Dublin, Ireland
- Editors:
- Smaranda Muresan, Preslav Nakov, Aline Villavicencio
- Venue:
- ACL
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 5306–5315
- Language:
- URL:
- https://aclanthology.org/2022.acl-long.364
- DOI:
- 10.18653/v1/2022.acl-long.364
- Cite (ACL):
- Colin Leong and Daniel Whitenack. 2022. Phone-ing it in: Towards Flexible Multi-Modal Language Model Training by Phonetic Representations of Data. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 5306–5315, Dublin, Ireland. Association for Computational Linguistics.
- Cite (Informal):
- Phone-ing it in: Towards Flexible Multi-Modal Language Model Training by Phonetic Representations of Data (Leong & Whitenack, ACL 2022)
- PDF:
- https://preview.aclanthology.org/naacl24-info/2022.acl-long.364.pdf
- Code
- sil-ai/phone-it-in
- Data
- MasakhaNER