Self-Explanation vs. Think Aloud: What Natural Language Processing Can Tell Us

EasyChair Preprint 3624

Abstract

Self-explanation is designed to increase coherence by encouraging students to activate prior knowledge, generate inferences, and make casual connections (McNamara, 2004). The current study used natural language processing to examine how readers’ responses differ when instructed to self-explain or think aloud. Self-explanations were found to contain more cohesion, semantic overlap, and causal, active, and positive emotion words than think-alouds. The results provide evidence that instructional differences significantly predicted linguistic differences reader’s responses to texts.

Keyphrases: Natural Language Processing, self-explanation, think aloud

@booklet{EasyChair:3624,
  author    = {Sarah D. Creer and Kathryn S. McCarthy and Joseph P. Magliano and Danielle S. McNamara and Laura K. Allen},
  title     = {Self-Explanation vs. Think Aloud: What Natural Language Processing Can Tell Us},
  howpublished = {EasyChair Preprint 3624},
  year      = {EasyChair, 2020}}