Question-based computational language approach outperforms rating scales in quantifying emotional states

This study demonstrates that descriptive word responses analyzed using natural language processing (NLP) show higher accuracy in categorizing emotional states compared to traditional rating scales. Participants generated narratives related to four emotional states (depression, anxiety, satisfaction, harmony), summarized them with five descriptive words, and rated them using scales. Another group evaluated these narratives. NLP quantified the words, and machine learning categorized responses. Word-based categorization (64%) significantly outperformed rating scale categorization (44%), challenging the presumed superiority of rating scales in measuring emotional states.

Sverker Sikström, Leva Valavičiūtė, Inari Kuusela, Nicole Evors