Text Analysis in Sentiment Detection: A Cross-Disciplinary Approach Combining Linguistics and Psychology

Abstract

Sentiment detection, a key task in natural language processing (NLP), involves identifying and classifying emotions expressed in text. While traditional models often focus on linguistic features such as word choice and syntax, they frequently fail to capture the complexities of human emotion. This article explores the integration of psychology with linguistic analysis to improve sentiment detection accuracy. Drawing on psychological models such as Ekman’s universal emotions and Plutchik’s Wheel of Emotions, we propose a hybrid approach that combines linguistic features with emotional insights from psychology. Through experiments on a standard sentiment analysis dataset, we demonstrate that the hybrid model outperforms traditional linguistic models in accuracy, precision, recall, and F1-score. Our findings suggest that a cross-disciplinary approach, combining insights from linguistics and psychology, leads to more accurate and nuanced sentiment detection. This work has implications for a range of applications, including social media analysis, customer feedback, and mental health monitoring, where understanding the emotional nuances of language is crucial.