Online processing fluency and explicit knowledge in data-driven learning: Rethinking grammatical difficulty

This study tracks changes in high school students’ online processing fluency and explicit knowledge over time following data-driven learning (DDL) instruction and explores what their performance on two structures differing in difficulty may suggest about grammatical difficulty. A total of 37 participants received DDL instruction targeting the two grammatical structures. Online processing fluency was assessed through a self-paced reading task and explicit knowledge through an untimed grammaticality judgment test, with qualitative data from learning journals used to contextualize the quantitative findings. Analyses using mixed-effects models revealed two main patterns. First, gains in online processing fluency were sustained through the delayed test for both structures, whereas gains in explicit knowledge were not retained—a divergence that held regardless of grammatical difficulty. Second, the structure classified as less difficult based on acquisition order, linguistic complexity, and frequency nonetheless showed greater processing demands and lower comprehension accuracy, suggesting that semantic-logical properties such as embedded negation may also contribute to grammatical difficulty. These findings underscore the value of examining online processing fluency and explicit knowledge separately following DDL instruction and are consistent with a multidimensional view of grammatical difficulty. Pedagogical implications and directions for future research are discussed.