Automated analysis of dream sentiment -- the royal road to dream dynamics?

Abstract

Valence Aware Dictionary for sEntiment Reasoning (VADER) is an automated software program for analyzing textual data based on an established lexicon and annotated lexical features. Support-vector machine (SVM) is a popular machine-learning model for solving classification problems. VADER and SVM can serve as potential alternatives to the conventional content analysis and Linguistic Inventory and Word Count analysis of dream emotions. The study presented here aimed to evaluate the overall affective valence of dreams using both the VADER and SVM methods. A total of 2,600 dreams primarily obtained from an open source—including dreams reported by American, German, Hong Kong, Peruvian, and Taiwanese people—were subjected to the 2 automated algorithms for sentiment analysis. The mean VADER and SVM sentiment scores indicate overall balanced sentiment in dream reports. Accordingly, an average dream report contains positive and negative emotions of similar intensity. Notwithstanding their different algorithms and methodological strategies, the marked consistency between the VADER and SVM scoring suggests that VADER and SVM can provide reliable, effective, yet distinct tools for dream sentiment analysis. In addition, the analysis of Chinese people’s dreams suggests that the discrepancy between dream sentiment scored by automated algorithms and subjective feelings experienced by dreamers may reveal some dynamic processes during dreaming, such as working through concerns and desensitizing feelings. (PsycInfo Database Record (c) 2022 APA, all rights reserved)