A Corpus Linguistic Perspective on Error Detection and Correction in English Academic Texts

Students and researchers from various fields of study and from different language backgrounds use English as a lingua franca when writing academic texts. However, even advanced and proficient users continue to make errors (Lennon 1991; Uzuner 2008). Therefore, the arrival of new AI writing tools might seem like the ideal solution, especially for non-native users of English struggling with writing academic texts. The goal of this study is to provide an overview of how useful AI tools can be for error correction purposes from a linguistic perspective. A representative selection of typical German learner errors from different linguistic problem areas was identified using a corpus linguistic approach. Nine AI tools were used to correct fictitious learner texts containing these errors and results indicate notable discrepancies in terms of the tools’ detection accuracy and correction suitability. Critical evaluation of AI applications is one aspect of AI literacy (D. T. K. Ng et al. 2021, OECD 2025), a future skill needed to use these technologies responsibly and competently. This paper provides a basis for further research in this rapidly developing field and offers practical insights for users of English who wish to leverage AI tools as artificial proofreaders for their academic texts.