Concurrent Statistical Learning of Adjacent and Nonadjacent Dependencies

Abstract

When children learn their native language, they have to deal with a confusing array of dependencies between various elements in an utterance. The dependent elements may be adjacent to one another or separated by intervening material. Prior studies suggest that nonadjacent dependencies are hard to learn when the intervening material has little variability, which may be due to a trade‐off between adjacent and nonadjacent learning. In this study, we investigate the statistical learning of adjacent and nonadjacent dependencies under low intervening variability using a modified serial reaction time (SRT) task. Young adults were trained on mixed sets of materials comprising equally probable adjacent and nonadjacent dependencies. Offline tests administered after training showed better performance for adjacent than nonadjacent dependencies. However, online SRT data indicated that the participants developed sensitivity to both types of dependencies during training, with no significant differences between dependency types. The results demonstrate the value of online measures of learning and suggest that adjacent and nonadjacent learning can occur together even when there is low variability in the intervening material. Open Practices This article has been awarded an Open Materials badge. All materials are publicly accessible in the IRIS digital repository at http://www.iris-database.org. Learn more about the Open Practices badges from the Center for Open Science: https://osf.io/tvyxz/wiki

Non‐adjacent Dependencies Processing in Human and Non‐human Primates

“…In previous SRT experiments, we found that the learning of adjacent dependencies in triplets (Minier et al, 2016), or pairs of shapes (Fagot, Malassis, & Medam, 2018) gave rise to similar progressive (albeit earlier) decrease in response times throughout the exposure to these regularities. These data, along with previous findings s (Frost & Monaghan, 2016;Romberg & Saffran, 2013;Vuong, Meyer, & Christiansen, 2016), suggest that adjacent and non-adjacent dependencies might be extracted by similar statistical learning mechanisms. However, they do not allow disentangling between the single-process and the dual-system proposals described above.…”

Towards a theory of individual differences in statistical learning

“…This makes it possible to track the detailed time-course of learning. RT differences between predicted and unpredicted stimuli have also been demonstrated in other tasks in auditory [86] and audio-visual SL [20,31,32,56]. Importantly, online measures of SL have been found to correlate with sentence processing in L1 [20,31], providing preliminary evidence regarding its predictive validity.…”

Children With Developmental Language Disorder Have an Auditory Verbal Statistical Learning Deficit: Evidence From an Online Measure

“…Following this trend, we decided to use response times as an online measure of NAD learning, in particular measuring the disruption peak that occurs in the response time pattern when items are presented that are discordant with NAD rules. Previous work has shown that disruption peaks reflect sensitivity to NADs in adults (López-Barroso, Cucurell, Rodrígez-Fornells, & de Diego-Balaguer, 2016;Misyak et al, 2010;Vuong, Meyer, & Christiansen, 2015) and in primary-school-aged children (Lammertink, van Witteloostuijn et al, 2018). The use of disruption peaks as an index of statistical learning has its roots in the serial reaction time task literature (Nissen & Bullemer, 1987), and the reason to work with disruption peaks rather than a decrease in response times over the first few training blocks is that such a response time decrease is not necessarily the result of statistical learning.…”