From Selection to Organization: Effects of Structure-Guided Digital Highlighting on Learning from Expository Text

This dataset accompanies an experimental study examining whether digital learner-generated highlighting can better support learning when aligned with the compare-and-contrast structure of an expository text. A total of 186 seventh-grade students from three schools in Santiago, Chile, were randomly assigned to four conditions: study-only, traditional highlighting, color-coded highlighting, or matrix-guided highlighting. The main hypothesis was that structure-guided highlighting would improve learning relative to study-only and traditional highlighting. Color-coded highlighting required students to classify selected information according to its structural role, whereas matrix-guided highlighting additionally required students to organize selected information within a spatial matrix. A second hypothesis predicted that both structure-guided conditions would produce more eye-movement transitions between the two text paragraphs. The dataset includes memory and comprehension outcomes, time on task, total gaze duration on the text, between-paragraph transitions, baseline reading-comprehension scores, and gender. Memory was assessed with a 14-item cloze test and comprehension with a written summary scored using a four-criterion rubric. Eye movements were recorded with a Tobii Pro Nano eye tracker. No significant group differences were found for memory. For comprehension, matrix-guided highlighting significantly outperformed study-only (d = 0.77) and traditional highlighting (d = 0.61), although it did not differ significantly from color-coded highlighting. Both structure-guided conditions produced more between-paragraph transitions than study-only and traditional highlighting. All highlighting conditions also required more time and produced longer gaze duration than study-only. The findings should not be interpreted as showing that greater time, visual attention, or transition frequency necessarily improves learning. Rather, the clearest comprehension benefit occurred when selected information was organized within an external relational representation. Between-paragraph transitions should be interpreted as indicators of comparative visual search rather than direct evidence of successful integration.本数据集为一项实验研究的配套数据，该研究旨在探究当数字式学习者自主标注（digital learner-generated highlighting）与说明性文本的对比结构相适配时，能否更有效地支持学习。本研究招募了来自智利圣地亚哥三所学校的186名七年级学生，将其随机分配至四种实验条件：仅阅读组、传统标注组、色标标注组及矩阵引导标注组。本研究的核心假设为，相较于仅阅读组与传统标注组，结构引导标注（structure-guided highlighting）可提升学习效果。其中，色标标注组要求参与者根据信息的结构功能对所选内容进行分类，而矩阵引导标注组则额外要求参与者将所选内容整合至空间矩阵中。第二项假设预测，两类结构引导标注组均会在两个文本段落间产生更多眼动转移。 本数据集包含记忆与理解能力测评结果、任务时长、文本总注视时长、段落间眼动转移次数、基线阅读理解得分及参与者性别信息。记忆能力通过14题完形填空测验（cloze test）进行评估，理解能力则通过书面总结进行测评，总结得分采用四维度评分量表（four-criterion rubric）计分。眼动数据通过Tobii Pro Nano眼动仪采集。记忆能力测评未发现组间存在显著差异。理解能力方面，矩阵引导标注组的表现显著优于仅阅读组（d=0.77）与传统标注组（d=0.61），但其与色标标注组的表现无显著差异。两类结构引导标注组的段落间眼动转移次数均多于仅阅读组与传统标注组。所有标注组的任务时长均长于仅阅读组，且文本注视总时长也更长。 本研究结果不应被解读为更长的任务时长、更高的视觉注意力投入或更高的眼动转移频率必然能提升学习效果。相反，最显著的理解能力提升出现在所选内容被整合至外部关系表征的情境中。段落间眼动转移次数应被视为比较性视觉搜索的指标，而非信息成功整合的直接证据。