Eyelit: Eye-movement and reader response data during literary reading

This dataset is an extensive eye tracking dataset of 102 participants reading three Dutch literary short stories each (7790 words per participant). The preprocessed data set includes (1) Fixation report (fixation-level), (2) Saccade report, (3) Interest Area report (word-level), (4) Trial report (aggregated data for each page; stories were split up into 30 pages each), (5) Sample report (the data were sampled at 500 Hz, this report includes data on all individual samples), (6) Questionnaire data on reading experiences and other participant characteristics, and (7) word characteristics for all words in the stories (with the potential of calculating additional word characteristics).The study for which this data set was collected explored the effect of simulation on reading behavior by means of eye tracking. We hypothesized (A) that simulation would lead to longer fixation times for parts of the text high in simulation-eliciting content. Additionally, we hypothesized (B) that we would find personal preferences in the reaction to different types of simulation-eliciting content. We expected (C) that the findings from the eye tracking data would be related to self-report of simulation. Finally, we expected (D) that the amount of simulation would be predictive of self-report of appreciation. We found (A) longer reading times for perceptual and mental event simulation, but shorter reading times for motor simulation. The strength of the relationship between simulation and reading times varied between participants, but was positively correlated across the different types of simulation-eliciting content (B). Regarding (C) and (D), we found that this variation in the strength of the relationship between simulation and reading times was indeed related to aspects of self-reported simulation, absorption and appreciation.The findings from this study are described in detail in Mak & Willems (2019). https://doi.org/10.1080/23273798.2018.1552007

本数据集为一项规模庞大的眼动追踪（eye tracking）数据集，共纳入102名被试，每名被试需阅读三篇荷兰语文学短篇故事，单名被试的阅读文本总字数达7790词。
预处理后的数据集包含以下7类内容：(1) 注视报告（Fixation report，注视层级fixation-level数据）；(2) 眼跳报告（Saccade report）；(3) 兴趣区（Interest Area）报告（词级word-level数据）；(4) 试次报告（Trial report，含各页面汇总数据；每篇故事被拆分为30个页面）；(5) 采样报告（Sample report，数据采样率为500 Hz，该报告包含所有单条采样数据）；(6) 阅读体验问卷及其他被试特征数据；(7) 故事中所有词汇的词汇特征（可用于计算额外词汇特征）。
本数据集所属的研究采用眼动追踪技术，旨在探究模拟操作对阅读行为的影响。我们提出如下假设(A)：当文本中包含高诱发模拟内容的段落时，模拟操作会使读者的注视时长更长。此外，我们提出假设(B)：读者对不同类型诱发模拟内容的反应存在个体偏好差异。我们提出假设(C)：眼动追踪数据的分析结果将与被试自我报告（self-report）的模拟体验相关联。最后，我们提出假设(D)：模拟的程度可预测被试的自我报告阅读满意度。
本研究验证假设(A)后发现：感知模拟与心理事件模拟对应的阅读时长更长，而运动模拟对应的阅读时长更短。模拟与阅读时长之间的关联强度存在个体差异，但在不同类型的诱发模拟内容间呈正相关关系（对应假设B）。针对假设C与假设D，我们发现模拟与阅读时长关联强度的个体差异，确实与被试自我报告的模拟体验、沉浸感及阅读满意度相关。
本研究的详细研究结果已发表于Mak与Willems（2019）的论文中，DOI链接为：https://doi.org/10.1080/23273798.2018.1552007