Grammatical Error Correction in Slovak Using Neural Machine Translation and No Language Left Behind

Grammatical error correction aims to identify and rectify linguistic inaccuracies in text, thereby enhancing communication and comprehension. This study investigates the ap-plication of neural machine translation techniques, particularly the No Language Left Behind technique for GEC in Slovak, a morphologically rich language. We constructed datasets containing deliberate grammatical errors and used them to train models capable of detecting and correcting these errors. By introducing random character and word-level transformations, we simulated realistic grammatical errors, creating a robust testing ground for model evaluation. Our experimental results demonstrate significant improvements in performance metrics such as BLEU score, Word Error Rate, and Sentence Error Rate. The findings highlight the potential of NMT-based approaches in developing efficient and accurate GEC systems for underrepresented languages.