Negative Language Transfer Identification in the English Writing of Chinese and Farsi Native Speakers

Effective communication in English can facilitate educational and employment opportunities for learners of English as an additional language (EAL) who tend to employ rules from their native language while communicating in English. This results in negative language transfer (NLT) when the rules from the mother tongue do not match those of English. One way of assisting EAL learners is to identify NLT errors in their English writing as a first step in the feedback process. However, manually identifying and providing feedback on learner NLT is a difficult task that requires time and expertise. A model that automatically identifies NLT in learner writing could facilitate this process. In this study, four classification algorithms were implemented to identify NLT in EAL learner writing automatically. Two of the language modelling approaches employed to classify learner errors (n-gram and recurrent neural network) were grounded in the linguistic nature of NLT, whereas the other two classifiers were general-purpose classifiers (random forest and logistic regression). The results show that the models could identify NLT in the English writing of Chinese and Farsi native speakers. Random forest outperformed all other models, yielding average weighted F1-scores of 78.1% on the Chinese FCE dataset and 94.8% on the Farsi Lang-8 dataset. This work shows that the implemented models could be used to automatically identify NLT errors in the English writing of Chinese and Farsi native speakers. The correct identification of such errors and subsequent provisioning of appropriate feedback could facilitate language learning and improve educational and employment opportunities for EAL leaners.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.