Language Analyzer: AI-Based Marathi Grammar Correction Using Hybrid Deep Learning

This study highlights the inadequacies of the digital tools currently available for Marathi grammar correction and offers a fresh approach to improve their accuracy. A comparative analysis was conducted using a sample Marathi paragraph that contained intentional grammatical errors. Popular AI tools like ChatGPT, Gemini, Copilot, and Perplexity were used to process the paragraph. Accuracy and fluency tests were used to gauge their level of proficiency with Marathi text. Performance varied, according to the results, with ChatGPT scoring the highest at 95.72%, Copilot coming in second at 90.18%, and Gemini and Perplexity at 90.90%. Despite these positive results, the tools handling of complex grammatical structures, including verb conjugations, gender agreement, and Marathi-specific contextual nuances, was woefully inadequate.To address these problems, this study proposes a specialized Marathi Grammar Correction Tool that incorporates more intricate linguistic rules and context-aware mechanisms. By combining rule-based features with state-of-the-art deep learning architectures, the proposed system aims to increase accuracy, provide insightful feedback, and enable the effective use of Marathi in literature, education, and digital communication.