On the Robustness of Language Encoders against Grammatical Errors
This work addresses the robustness of language models for NLP applications, particularly in handling errors from non-native speakers, but it is incremental as it builds on existing models and tasks.
The study investigated how pre-trained language encoders like ELMo, BERT, and RoBERTa perform when exposed to grammatical errors, finding that their performance is affected to varying degrees, with fixed contextual encoders able to locate error positions and BERT capturing error-token interactions.
We conduct a thorough study to diagnose the behaviors of pre-trained language encoders (ELMo, BERT, and RoBERTa) when confronted with natural grammatical errors. Specifically, we collect real grammatical errors from non-native speakers and conduct adversarial attacks to simulate these errors on clean text data. We use this approach to facilitate debugging models on downstream applications. Results confirm that the performance of all tested models is affected but the degree of impact varies. To interpret model behaviors, we further design a linguistic acceptability task to reveal their abilities in identifying ungrammatical sentences and the position of errors. We find that fixed contextual encoders with a simple classifier trained on the prediction of sentence correctness are able to locate error positions. We also design a cloze test for BERT and discover that BERT captures the interaction between errors and specific tokens in context. Our results shed light on understanding the robustness and behaviors of language encoders against grammatical errors.