Evaluating click detection as a measure of statistical learning: a pilot study of children with poor auditory attention

Briscoe, J. . & Heeks, G.

University of Bristol

Capturing statistical learning (SL) of transitional probabilities from children?s exposure to unfamiliar speech offers important theory and clinical insights. Although children show levels of SL similar to adults (Saffran et al., 1997), the characterisation as adevelopmental is problematic. Using 2AFC tests to assess individual children is difficult to interpret when individual scores are at, or significantly below chance. Our study used the click detection paradigm with children aged 5-7 years to probe SL of transitional probabilities during speech exposure. Since attentional regulation is a potential source of individual differences, two groups of children were compared on a click detection paradigm; children with atypical scores on an attentional subtest of the TEACH (Walk Don?t Walk) (n=19; 14 female, WDW scaled score < 7) and matched controls (n=19, 14 female, age; p = .51) as sampled from mainstream schools. Both groups were further matched on response rate implying good be
havioural alignment to the click task during encoding, and matched on 2AFC scores at test; WDW Mean = 63%; TPD Mean = 57%; p = .43, n=16) although group scores were near chance . Click detection latencies will be further discussed to probe the validity of online measures of SL during language exposure.