First-Attempt Articulation Error Counts and Digit Span Scores in Japanese Preschool Children: An Exploratory...

Digit span tasks are commonly used as indicators of verbal short-term memory, but their relation to speech production accuracy is complex and remains insufficiently examined in Japanese preschool children. This exploratory study examined whether a task-specific count of first-attempt articulation errors was associated with Number Recall scaled scores from the Kaufman Assessment Battery for Children, Second Edition (KABC-II).

In this exploratory cross-sectional study, 31 community-based Japanese children aged 56-82 months completed the Japanese Articulation Test 50-word version and the Number Recall subtest of the Kaufman Assessment Battery for Children, Second Edition. The primary speech outcome was the first-attempt articulation error count, defined as the number of elicited words containing at least one substitution, omission, or distortion on the first response. Spearman correlation was used for the primary analysis, with bias-corrected and accelerated bootstrap 95% confidence intervals. Age-adjusted sensitivity analyses included partial Spearman correlation and negative binomial regression.

First-attempt articulation error counts were right-skewed (mean, 3.0; standard deviation, 3.4; range, 0-16). The mean KABC-II Number Recall scaled score was 10.3 (standard deviation, 2.5). Higher error counts had a negative point estimate for association with Number Recall scaled scores, but the estimate was imprecise and compatible with no association (rho = -0.341; p = 0.061; BCa 95% CI, -0.638 to 0.065). The age-adjusted partial Spearman correlation was similar to the unadjusted estimate (rho = -0.341; p = 0.066). In negative binomial regression adjusted for age in months, the digit-span term yielded an incidence rate ratio of 0.900 (95% CI, 0.78-1.04; p = 0.147).

In this small exploratory sample, first-attempt articulation error counts showed an imprecise inverse association with KABC-II Number Recall scaled scores that was compatible with no association. The findings should be interpreted as preliminary and hypothesis-generating and should not be used to infer clinical severity, mechanism, or diagnostic implications. Larger studies with age-balanced samples, formal hearing screening, and phonologically motivated task designs are needed.