Author ORCID Identifier
https://orcid.org/0009-0000-5464-8863
Date of Award
Spring 5-28-2024
Document Type
Thesis (Undergraduate)
Department
Cognitive Science
First Advisor
Mark Thornton
Second Advisor
Amisha Vyas
Abstract
Is there dimensionality in how people form impressions of speakers? The first portion of the study generated trait terms via text survey responses to recordings from the Emotional Speech Database (ESD). Following the analysis of this initial data, a second survey invited subjects to rate the speaker on a specific trait for each of many ESD recordings. The trait terms were selected from the text responses from the first survey that were most strongly correlated via a mixed-effects model with the speaker of each recording rather than the emotion or sentence. Following an exploratory factor analysis of the trait ratings responses in the second survey, a single dimension (“factor x”) was identified. Factor x can be roughly related to concepts of valence, arousal, and sociality. Findings from this study are compatible with general trait impression research findings in the field and support a generally dimensional framework of trait impressions across modalities.
Recommended Citation
Conway, Caroline W., "What We Hear in Voices: Identifying Dimensionality in Speech Stimuli Trait Impressions" (2024). Cognitive Science Senior Theses. 4.
https://digitalcommons.dartmouth.edu/cognitive-science_senior_theses/4