Julia Cerveira-Bianchi¹ (), LeAnne Spino-Seijas¹, Alisa Baron¹; ¹University of Rhode Island

Eye tracking with a violation paradigm is an online, real-time measure, used within the field of second language acquisition to measure implicit or automatic knowledge. Native speakers (L1) demonstrate implicit knowledge and second language learners (L2) exhibit various levels of developing implicit knowledge, which serves as a measure of proficiency. In a violation paradigm, L1 and L2 speakers read grammatical and ungrammatical stimuli. The eye-tracking data provides insight into participants’ sensitivity to grammatical violations. Previous studies have included comprehension questions, an offline measure, to ensure that participants read for meaning. However, eye tracking has rarely been applied to comprehension questions themselves, creating a methodological gap in the research. We fill this gap by investigating the relationships between eye-tracking measures, gaze duration and total duration, with accuracy and response time to comprehension questions. A total of 37 participants, L1 (N = 10) and L2 (N = 27) Spanish speakers, read grammatical and ungrammatical sentences in Spanish across two conditions, temporal reference and adverb placement. Participants read 69 sentences, each followed by a yes or no comprehension question which probed a non-critical region of the sentence, or the area without the grammatical error. The sentences were pseudorandomized to ensure that ungrammatical sentences of the same condition did not occur consecutively and that participants read only one version of all experimental stimuli. Our preliminary results show that L2 speakers exhibited significantly longer gaze duration and total duration than L1 speakers (p = 0.027 and p = 0.025 respectively). We also found a strong effect of sentence type, with adverb placement yielding a longer gaze duration than the temporal reference condition (p < 0.001). Finally, longer response times to comprehension questions were associated with a significant decrease in comprehension accuracy (p < 0.001), suggesting that slower responses are reliably linked to reduced comprehension accuracy.

Acknowledgements: Funding acknowledgment: This project was funded in part by two Project Completion Grants from the Division of Research & Economic Development at the University of Rhode Island (PI: Spino-Seijas; PI: Alisa Baron)