ComprehensionWorkshop

Reading Time Regularity as Predictor for Text Comprehension

Authors:
Tschense, Monika, monika.tschense@leuphana.de, Leuphana University Lüneburg
Wallot, Sebastian, sebastian.wallot@leuphana.de, Leuphana University Lüneburg

Keywords: reading time regularity (RTR), recurrence quantification analysis (RQA), text reading, eye tracking, self-paced reading

Abstract:

Reading a text is a complex and dynamic task that involves a precise interplay of motoric, perceptual, and higher-order cognitive processes. Moreover, the reading process adapts to text properties (e.g, difficulty; Kaan & Swaab, 2003), task demands (e.g., proofreading vs. skimming; Schotter et al., 2014), and reader characteristics (e.g., prior knowledge; Kendeou & Van den Broek, 2007). It is well-established that eye movements and reading times reflect the coupling between a text and a reader (Engbert et al., 2005; Rayner et al., 2016). In a previous study, Tschense and Wallot (2022) demonstrated that this relationship can be quantified by the degree of regularity in time series of the reading process. The authors further proposed that regularity might also be informative about outcome measures such as text comprehension. To test this assumption, participants read three fictional texts (~3,000 words, ~15 minutes) in three conditions: at a comfortable pace, as fast as possible, and as accurately as possible. After each text, comprehension was assessed using open-ended wh-questions and yes/no statements. In Study 1, participants’ eye movements were recorded (EyeLink 1000, binocular recording, 500 Hz). Two additional studies measured self-paced reading times: in Study 2, words were presented in isolation, while Study 3 employed a word-by-word text build-up. Time series from all studies were analyzed using Recurrence Quantification Analysis (RQA), and the resulting regularity measures (RR, DET, MDL, ADL, LAM, and TT) were passed on to mixed-effects models. In both studies, regularity measures significantly predicted participants’ text comprehension scores. However, differences emerged between the two types of comprehension items. Moreover, eye movements and word reading times yielded distinct patterns of results. This highlights the need for a more systematic investigation of different types of both comprehension measures and process measures, as well as their suitability for RQA. Additionally, it underscores potential advantages of RQA over traditional measures of the reading process, such as average reading times, the number of fixations, and average fixation durations (Mézière et al., 2023; Southwell et al., 2020).