Text this: Performance (F1 score (%)) of the fine-tuned PLMs and ensemble models on closely related language combinations. The average weighted F1 with 95% confidence intervals (CIs).