Swoboda, K. , Witton, C. , Talcott, J. B. & Shapiro, L.
School of Life and Health Sciences, Aston University, UK
The debate concerning the association of developmental dyslexia with deficits in basic auditory processing is ongoing. Research studies have yielded variable results which are difficult to interpret. The first objective of our study was to quantitatively assess the variability of effect sizes across the studies measuring one type of basic auditory processing, frequency discrimination (FD). Our second aim was to explore whether any of the cognitive or task design variables could moderate the variability of the FD effect sizes. To meet our first objective we performed a meta-analysis of n=29 experiments from 27 published articles which psychophysically assessed FD in dyslexic and control participants. Statistical significance of both the mean effect size (M=0.735; SEM=0.076) and the effect size variability demonstrated that although individual effect sizes were on average significant, they were inconsistent across studies. We also found considerable inconsistency among the cognitive and language measures used (real and non-word reading, phonemic deletion, non-verbal IQ, verbal working memory, and oral language), specifically 1) common use of non-standardised tests, and 2) lack of inclusion of key cognitive measures relevant to reading. Therefore, our finding that the effect size heterogeneity was not moderated by any of the participant variables needs to be treated with caution. Importantly, however, we found a relationship between the type of FD task and the variability of the FD effect sizes, such that larger effect sizes tend to co-occur with the FD tasks which may be more working memory-dependent, and smaller effect sizes with the tasks relying more on pattern recognition processes. The results of our study suggest that it would greatly benefit the literature if authors reported the key cognitive measures, including standardised tests, to facilitate across-study comparisons. Furthermore, the choice of FD task is important because performance on different FD tasks may depend on different underlying mechanisms.