Enhancing lecture accessibility for hearing-impaired and nonverbal primary school students through complex term detection

Abstract

Problem: Due to limited exposure to speech, and hearing impairment, a partial or whole inability to hear can cause delays in a child's language development. Children with hearing impairments encounter additional difficulties when learning to read and write in English since English texts frequently combine simple and complex vocabulary. When hearing-impaired students are exposed to a lot of complex terminology in texts like stories, essays, and academic materials, their comprehension slows down and their learning gaps expand compared to their hearing colleagues. The necessity for a mechanism to help hearing-impaired children get past these linguistic barriers particularly when it comes to comprehending complicated vocabulary is discussed in this research. Solution: To solve this, a machine learning-based system using a Random Forest Classifier was developed. Texts are preprocessed using NLP techniques (tokenization, POS tagging, etc.), and key features like word length, frequency, and syllable count are extracted. The model identifies difficult words and offers simplified synonyms and images to help students understand them better. Test Results: The system was tested on labeled educational content and achieved over 95% accuracy in classifying complex words. Evaluation metrics such as precision, recall, and F1- score confirmed its effectiveness. The tool outperformed basic heuristic methods and proved helpful in enhancing vocabulary support for hearing-impaired learners.