[PS-3.10]Online measure of nonadjacent dependency learning in typically developing children

Lammertink, I. 1, 2 , van Witteloostuijn, M. 1, 2 , Boersma, P. 1, 2 , Wijnen, F. 3, 4 & Rispens, J. 1, 2

1 University of Amsterdam, The Netherlands
2 Amsterdam Center for Language and Communication ACLC, The Netherlands
3 Utrecht University, The Netherlands
4 Utrecht Institute of Linguistics OTS, The Netherlands

Sensitive methods tracking on-line learning of nonadjacent dependencies (NADL) in children are scarce. We investigated the suitability of recording reaction times (RT) during NADL to measure learning success.

Forty-six native Dutch speaking children (M = 7.3 years) participated in a child-friendly adaptation of a NADL experiment (López-Barroso et al. 2016). Forty percent of the utterances followed the NAD-rule tep_X_lut and 40% followed the NAD-rule sot_X_mip (Gómez, 2002; the first element predicts the third element). Twenty percent were fillers in which the first element never predicted the third element. Children were asked to press a green button in response to hearing the target (lut or mip). They had to press a red button when the third element was not this target. After three regular blocks (180 trials), a block was administerd in which lut or mip did not follow the NAD-rules (random block; 30 trials), followed by a regular block (recovery block; 60 trials).

The RT-trajectory of NAD-rules differed from the fillers across the final training, random and recovery block (F(4,42) = 6.79,p< .001; = .39). The random and recovery RTs correlated (r =.68, p< .001). These results support the reliability of RT as online measure of NADL in children.