Struggling Readers Smiling on the Inside and Getting Correct Answers

• In Event: Perspectives on Emotion and Engagement Among Struggling Adolescent Readers: Findings From an Online Literacy Intervention

Abstract

Purpose: Middle-school struggling readers have built up many years of frustrating and negative emotions associated with reading in school. In investigating an intervention designed to help these students, we simultaneously face the challenge of investigating the interplay of emotion and cognition (Okon-Singer, Hendler, Pessoa, & Shackman, 2015), and have the opportunity to use new emotional measures (Calvo & D’Mello, 2010) to unpack how we can improve our online environments for learners with different reading abilities. These new emotion measures are just starting to gain momentum in terms of responding to students’ emotions (Herold, 2016). In emotion research many studies rely on self-report and frequently use an ad-hoc list of words (Weidman, Steckler, & Tracy, 2016). We use a novel method of analysis by applying sentiment analysis methods to both the fixed self-report words and to open-ended student discussion comments in an online platform called PRODUCT NAME. The result is a comparable set of valence measures that facilitates an investigation of a more nuanced understanding of emotion in learning.

Theoretical Framework: Fiedler and Beier (2014) describe a broaden and build theory of positive emotion (Fredrickson, 1998, 2004), saying that both positive and negative emotions influence learning. Negative emotions are thought to be more effective for considering a complex learning task in isolation (Clore & Huntsinger, 2007), while positive emotions are thought to be associated with assimilative learning tasks (Fidler & Beier, 2014). A competing perspective is that emotions in general are a deficit to learning: cognitive load theory advocates that neutrality is optimal for student performance (Fraser et al., 2012). These perspectives provide a theoretical frame to investigate student emotion during learning.

Method: This study investigates middle school remedial reading classrooms in 9 US states, students (N=246) could choose short fiction and non-fiction readings, and then self-report their emotional reaction, make open-ended discussion comments, and answer a comprehension check. Sentiment analyses were used on both the self-report and comments to generate two emotional predictions about the student experience. Logistic regression was used to predict the relationship between the comprehension check and the emotional experiences of students.

Results: Students who provided a more positive self-report of emotional reaction to the reading, and demonstrated a more neutral presence in discussion, were more likely to provide a correct answer to the comprehension check (see Table 2). This supports both the relationship of positive emotion supporting assimilation and the relationship of neutral emotion supporting performance. This combination of feeling positive but behaving neutrally has been described as “smiling inside” (Schall, Martiny, Goetz, & Hall, 2016). Our model of “smiling inside” predicts success 43 out of 97 times (see Table 3).

Educational Significance: Applying sentiment analysis to both self-report mechanisms and online discussion comments provides methodological guidance on generating two comparable traces of students’ emotions for researchers. The specific results could inform teacher practice to recognize the role of emotions in reading, especially differences in emotions felt vs. those expressed, and how to support learners in a technology-supported classroom.

Authors

• Garron Hillaire, Massachusetts Institute of Technology

• Boris Goldowsky, CAST, Inc.

• Bart Rienties