NeuTral Rewriter: A Rule-Based and Neural Approach to Automatic Rewriting into Gender Neutral Alternatives

Eva Vanmassenhove, Chris Emmery, Dimitar Shterionov


Abstract
Recent years have seen an increasing need for gender-neutral and inclusive language. Within the field of NLP, there are various mono- and bilingual use cases where gender inclusive language is appropriate, if not preferred due to ambiguity or uncertainty in terms of the gender of referents. In this work, we present a rule-based and a neural approach to gender-neutral rewriting for English along with manually curated synthetic data (WinoBias+) and natural data (OpenSubtitles and Reddit) benchmarks. A detailed manual and automatic evaluation highlights how our NeuTral Rewriter, trained on data generated by the rule-based approach, obtains word error rates (WER) below 0.18% on synthetic, in-domain and out-domain test sets.
Anthology ID:
2021.emnlp-main.704
Volume:
Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing
Month:
November
Year:
2021
Address:
Online and Punta Cana, Dominican Republic
Editors:
Marie-Francine Moens, Xuanjing Huang, Lucia Specia, Scott Wen-tau Yih
Venue:
EMNLP
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
8940–8948
Language:
URL:
https://aclanthology.org/2021.emnlp-main.704
DOI:
10.18653/v1/2021.emnlp-main.704
Bibkey:
Cite (ACL):
Eva Vanmassenhove, Chris Emmery, and Dimitar Shterionov. 2021. NeuTral Rewriter: A Rule-Based and Neural Approach to Automatic Rewriting into Gender Neutral Alternatives. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 8940–8948, Online and Punta Cana, Dominican Republic. Association for Computational Linguistics.
Cite (Informal):
NeuTral Rewriter: A Rule-Based and Neural Approach to Automatic Rewriting into Gender Neutral Alternatives (Vanmassenhove et al., EMNLP 2021)
Copy Citation:
PDF:
https://aclanthology.org/2021.emnlp-main.704.pdf
Video:
 https://aclanthology.org/2021.emnlp-main.704.mp4
Data
OpenSubtitlesWinoBias