Gender-Inclusive Grammatical Error Correction through Augmentation
This addresses gender inclusivity issues in NLP tools for users of GEC systems, though it is incremental as it builds on existing augmentation methods.
The paper tackled gender bias in Grammatical Error Correction (GEC) systems related to masculine/feminine terms and singular 'they', showing that a novel data augmentation technique for singular 'they' and refined techniques for other terms can reduce bias while maintaining quality.
In this paper we show that GEC systems display gender bias related to the use of masculine and feminine terms and the gender-neutral singular "they". We develop parallel datasets of texts with masculine and feminine terms and singular "they" and use them to quantify gender bias in three competitive GEC systems. We contribute a novel data augmentation technique for singular "they" leveraging linguistic insights about its distribution relative to plural "they". We demonstrate that both this data augmentation technique and a refinement of a similar augmentation technique for masculine and feminine terms can generate training data that reduces bias in GEC systems, especially with respect to singular "they" while maintaining the same level of quality.