Do "English" Named Entity Recognizers Work Well on Global Englishes?
This highlights a critical limitation in NER tools for analyzing global English use, which is incremental as it identifies a gap but does not propose a new method.
The study tackled the problem of whether English named entity recognition (NER) models generalize to global English varieties by testing them on a new Worldwide English NER Dataset, finding that models trained on standard datasets like CoNLL 2003 or OntoNotes experienced significant performance drops of over 10 F1 points on global data.
The vast majority of the popular English named entity recognition (NER) datasets contain American or British English data, despite the existence of many global varieties of English. As such, it is unclear whether they generalize for analyzing use of English globally. To test this, we build a newswire dataset, the Worldwide English NER Dataset, to analyze NER model performance on low-resource English variants from around the world. We test widely used NER toolkits and transformer models, including models using the pre-trained contextual models RoBERTa and ELECTRA, on three datasets: a commonly used British English newswire dataset, CoNLL 2003, a more American focused dataset OntoNotes, and our global dataset. All models trained on the CoNLL or OntoNotes datasets experienced significant performance drops-over 10 F1 in some cases-when tested on the Worldwide English dataset. Upon examination of region-specific errors, we observe the greatest performance drops for Oceania and Africa, while Asia and the Middle East had comparatively strong performance. Lastly, we find that a combined model trained on the Worldwide dataset and either CoNLL or OntoNotes lost only 1-2 F1 on both test sets.