Using Human Interaction with Natural Language Processing Techniques to Reinforce Vocabulary Comprehension and Usage

This dissertation proposes SENCE: SENtence Curation and Evaluation - a Natural Language Processing (NLP) aid to be used in an educational setting for children. SENCE is designed as an AI-augmented tool for educators such as general and special education teachers and practicing school-based speech-language pathologists who work with children. While several commercially available products incorporate NLP techniques for teaching adults language skills, the field is still nascent for incorporating NLP into teaching aids for children with learning disorders. SENCE uses NLP techniques to reinforce vocabulary comprehension and usage in children. Additionally, it integrates human interaction with NLP techniques, allowing domain specialists to improve results before they are presented to students. SENCE uses off-the-shelf NLP libraries such as spaCy and Stanza in combination with NLP techniques such as lemmatization, part-of-speech tagging, and vocabulary similarity. These methods are integrated to identify key vocabulary words and sentences using those keywords. An evaluation is created based on these keywords and sentences. SENCE thereby creates an automated process to gauge students’ vocabulary comprehension over time. The evaluations can be shared between classes and instructors. Further, students can be quickly assessed for retention of words taught earlier in the school year. Through these methods, SENCE provides a novel and easy-to-use NLP-powered application for non-computer scientists to use NLP for everyday classroom tasks.