Lukics, K. S. 1, 2 , Dobó, D. 1, 2 & Lukács, &. 1, 2
1 Department of Cognitive Science, Budapest University of Technology and Economics
2 MTA-BME Lendület Language Acquisition Research Group, Budapest, Hungary
Two-alternative forced choice (2AFC) tasks as measures of statistical learning received criticism due to their metacognitive nature and poor test-retest reliability (Franco et al., 2015; Isbilen et al., 2017). Attempts to seek more valid measures include measuring syllable detection RTs online in predictable versus unpredictable conditions (e.g., Batterink, 2017), and an implicit offline measure, the statistically-induced chunking recall (SICR) task (Isbilen et al., 2017). Testing young adults, we aimed to examine the sensitivity of three different measures of statistical learning in a word segmentation paradigm and to test relationships between them: online measures from a syllable detection task, offline implicit measures from a SICR and offline explicit measures from a 2AFC task. We found significant learning effects in all subtests. The size of the learning effect in the syllable detection task significantly correlated with subjects? performance in the 2AFC task. No significant relationships were found between SICR performance and either the 2AFC task or online measures. Syllable detection proved to be a sensitive non-metacognitive measure of SL which, unlike SICR is also related to metacognitive measures. Further explorations of different measures of SL should help us find better ways of testing different populations and variation in SL abilities.