International Journal of Computational Linguistics Research Fouzi Harrag 17 3 2026 https://doi.org/10.6025/ijclr/2026/17/3/151-171 https://www.dline.info/jcl/fulltext/v17n3/jclv17n3_2.pdf Vocabulary acquisition remains a cornerstone of second language (L2) proficiency; however, traditional instructional methods frequently fail to optimize long term lexical retention. Recent advancements in Natural Language Processing (NLP) and artificial intelligence offer novel avenues for personalized learning, dynamic text difficulty assessment, and affective intervention. This comprehensive review investigates the comparative efficacy of semantic versus thematic lexical clustering on ESL learner's vocabulary acquisition, alongside the application of AI-driven lexical metrics. Drawing upon contemporary research regarding morphological families, bilingual lexical diversity, and word feature influences, we analyze how specific clustering strategies impact both immediate recall and deep contextual integration. Furthermore, this paper explores the integration of NLP-based interventions designed to identify and disrupt habitual procrastination patterns, thereby reducing stress among secondary and post secondary ESL students. Findings synthesized from recent empirical studies indicate that while semantic clustering effectively supports initial word mapping, thematic approaches foster richer, more adaptable word learning experiences. Additionally, AI-mediated personalized text assessment significantly reduces cognitive load, allowing learners to engage with appropriately challenging materials without experiencing overwhelming frustration. Ultimately, this research underscores the critical need to align advanced lexical metrics with targeted clustering strategies and AI-driven affective interventions. By bridging psycholinguistic theory with modern educational technology, this study provides a robust framework for developing adaptive, evidence-based vocabulary assessments and instructional tools in contemporary language education.