The Relationship between Keystroke Actions, Self-Reported Emotion Regulation, and Facial Expressions during Self-Reflective Writing

• In Event: A Comprehensive Look at Emotion Formation and Regulation Across Different Contexts With Various Methodologies

Abstract

Objectives and Theoretical Frameworks
Current models of self-regulation and writing (Graham, 2018; Zimmerman & Risemberg, 1997) incorporate socio-cognitive theories about metacognition, which include the role of emotion during writing. Graham (2018) suggests that emotions modulate writing, meaning that all aspects of the writing process may affect/be affected by emotion. According to Gross (1998), emotions can be regulated using different strategies during stages of emotion generation, such as reappraising emotions generated while writing. Therefore, understanding how emotions modulate writing is key to understanding emotion generation and regulation.
Following Graham’s (2018) and Gross’ (1998) models, we triangulated keystroke-logging data to examine writing input actions, facial expressions using facial detection software to examine emotion, and self-reports to examine emotion regulation and posed the following research question: What is the relationship between writing actions, facial expression of emotion, and self-reported emotion regulation during writing?

Methods, Data Sources, and Results
Undergraduate participants (N = 60) enrolled in first-year writing in the Southeastern U.S. composed written reflections about their own writing and completed the Emotion Regulation Questionnaire (Gross & John, 2003). During the writing sessions we collected keystroke log-file data via InputLog (Leijten & Van Waes, 2013) and recorded videos of participants’ facial expressions to analyze using Affdex (iMotions, 2018). InputLog records pause duration between input actions and classifies each input as production, deletion, or insertion. For this study, we focused on input actions following pauses over 1,000ms as these actions are more likely to reflect participants’ metacognitive monitoring during writing (Galbraith & Baaijen, 2019; Rosenqvist, 2015). Affdex provides evidence scores (from 0-100) for each emotion, and we focused on anger, contempt, disgust, joy, and surprise (see Figure 1).
Mixed-effects models reveal that production (in comparison to insertion, i.e., dummy coded) is a significant predictor of all emotions (see Tables 1 and 2). When adding pause time and pause frequency as predictors (see Table 3), pause time and pause frequency are significant positive and negative predictors of anger, respectively. Pause frequency is also a positive predictor of disgust and surprise. When adding self-reported cognitive reappraisal and expressive suppression as predictors (Table 4), cognitive reappraisal is a negative predictor of disgust. Therefore, fewer instances of pausing and an increased cognitive reappraisal score are each associated with a lower facial expression evidence score of disgust.

Scholarly Significance
Our study offers evidence confirming Graham’s (2018) argument that emotions modulate writing. This was evident in the significant relationship between pause behaviors and facial expression of emotion, in that longer pause duration predicted greater expression of anger and more frequent pausing during writing predicted greater expression of disgust and surprise, and lower expression of anger. This suggests that as writers pause to engage in monitoring, their emotional expression may modulate the writing actions they take following the pause. Further, emotion regulation may result in subdued expression of emotion, as was the case for disgust with an increase in cognitive reappraisal associated with a decrease in disgust. Further research may examine emotion generation and regulation during long pauses between writing actions.

Authors

• Michelle Taub, University of Central Florida

• Joel Schneier, University of Central Florida

• Caroline Myrick, Independent Scholar

• LaVonda R. Walker, University of Central Florida

• Sierra Outerbridge, University of Central Florida