LM-Critic: Language Models for Unsupervised Grammatical Error Correction
It addresses the high cost of manual annotation for GEC by enabling unsupervised learning, though it is incremental as it builds on the Break-It-Fix-It framework.
The paper tackles the problem of grammatical error correction (GEC) without labeled data by using a pretrained language model as a critic to judge grammaticality and bootstrap training pairs, achieving improvements of +7.7 F0.5 in unsupervised and +0.5 F0.5 in supervised settings across multiple datasets.
Training a model for grammatical error correction (GEC) requires a set of labeled ungrammatical / grammatical sentence pairs, but manually annotating such pairs can be expensive. Recently, the Break-It-Fix-It (BIFI) framework has demonstrated strong results on learning to repair a broken program without any labeled examples, but this relies on a perfect critic (e.g., a compiler) that returns whether an example is valid or not, which does not exist for the GEC task. In this work, we show how to leverage a pretrained language model (LM) in defining an LM-Critic, which judges a sentence to be grammatical if the LM assigns it a higher probability than its local perturbations. We apply this LM-Critic and BIFI along with a large set of unlabeled sentences to bootstrap realistic ungrammatical / grammatical pairs for training a corrector. We evaluate our approach on GEC datasets across multiple domains (CoNLL-2014, BEA-2019, GMEG-wiki and GMEG-yahoo) and show that it outperforms existing methods in both the unsupervised setting (+7.7 F0.5) and the supervised setting (+0.5 F0.5).