Regression Modeling of Reader's Emotions Induced by Font Based Text Signals

In this work we presented a mathematical model for the readers’ emotional state responses triggered by font style, type and color. It is based on multiple regression analysis of the repeated measures from 45 students and for 35 textual stimuli using the Self-Assessment Manikin test. Based on the dimensional theory of emotions, we propose a model on how emotional dimensions Pleasure, Arousal, and Dominance vary according to the typographic text signals: font style, font type and font/background color combinations. We observe that “Pleasure” dimension is affected negatively by font type (“Arial” and “Times New Roman”) and positively by color brightness difference of font/background color combinations. “Arousal” and “Dominance” are affected only by color brightness difference (negative correlation). According to the proposed model, font type “Arial” elicits more pleasant emotional state than “Times New Roman”. The results can be applied to augment user interface experience or to add expressivity in Text-to-Speech systems and provide accessibility of typography induced text signals.