Oculomotor Evidence of Visual and Auditory Statistical Learning in Children

Deficits in statistical learning have been associated with various language disorders in adolescents and adults, but evidence is lacking concerning statistical learning in preschool-age children. This dissertation investigated the use of two new serial reaction time (SRT) task adaptations designed to measure young children’s statistical learning ability in response to visual and auditory input. In Study 1, typically developing 5- and 6-year-old monolingual Englishspeaking children were randomly assigned to either a visual or an auditory oculomotor serial reaction time (SRT) task. Twelve participants, ages 5;4 to 6;8 (years; months; mean = 5;8) completed the visual oculomotor SRT task and 12 children, ages 5;3 to 6;10 (mean = 5;10) completed the auditory oculomotor SRT task. In Study 2, which concerned the auditory oculomotor SRT task only, 17 typically developing monolingual four-year-olds (ages 4;0 to 4;11, mean = 4;7) completed the task. The tasks used in both studies involved oculomotor responses to 200 trials distributed over three phases: (a) a patterned phase, in which an 8-element sequence repeats 15 times (120 trials); (b) an unpatterned phase (40 trials) in which the 8 elements occur in a pseudorandom sequence; and (c) a second patterned phase, in which the original 8-element sequence repeats 5 times (40 trials). A significant increase in oculomotor reaction time (RT) at the pseudorandom phase was defined as the index of statistical learning. The hypothesis for both studies was that the mean RT z score during the 40 pseudorandom trials would be significantly slower relative to the final 40 patterned trials of the immediately preceding set. This hypothesis was supported, with statistically significant differences and large effect sizes for the visual task (Hypothesis 1a) in Study 1, (t(11) = -4.51; p < .001, Cohen’s dz = 1.57 [95% CI 0.65, 2.49]); for the auditory task (Hypothesis 1b) in Study 1, (t(11) = -5.76; p < .001, (Cohen’s dz = 1.79 [95% CI 0.81, 2.77]) and for the auditory task (Hypothesis 2) in Study 2, (t(16) = -4.76; p < .001, Cohen’s dz = 1.30 [95% CI 0.56, 2.03]). Findings from these studies suggest that these novel oculomotor SRT tasks successfully detect statistical sequence learning in typically developing monolingual children as young as four years of age. Accordingly, these and similar tasks may enable studies of the relationship between statistical learning and language ability in children much younger than those investigated previously with SRT tasks.