ABSTRACT
It is well-established that poor readers exhibit deficits in paired associate learning (PAL), and there is increasing evidence for a phonological locus of these deficits. However, it remains unclear whether poor performance stems from difficulties specific to the phonological output system or difficulties that affect both phonological input and output processes. Understanding these deficits is important not only in the context of PAL but also for informing broader theories of typical and atypical reading development. We developed a novel paradigm that allowed us to assess PAL in the presence and absence of phonological output demands. In total, 14 poor readers and 14 age-matched controls were first trained to criterion in verbal-visual PAL before being tested in the visual-verbal direction. The results showed that poor readers learned at the same rate as controls in verbal-visual PAL, even when the nonword stimuli were phonologically confusable. Yet, despite having reached the same criterion as controls in verbal-visual PAL, poor readers exhibited robust impairments for those same paired associates in visual-verbal PAL. The overall pattern of results is most consistent with the conclusion that PAL deficits reflect impairments to the phonological output system; however, results that may challenge this interpretation are also discussed.
Subject(s)
Dyslexia/physiopathology , Paired-Associate Learning/physiology , Pattern Recognition, Visual/physiology , Psycholinguistics , Reading , Speech Perception/physiology , Adolescent , Child , HumansABSTRACT
Recent research suggests that visual-verbal paired associate learning (PAL) may tap a crossmodal associative learning mechanism that plays a distinct role in reading development. However, evidence from children with dyslexia indicates that deficits in visual-verbal PAL are strongly linked to the verbal demands of the task. The primary aim of this study was to disassociate the role of modality and verbal demand in driving the PAL-reading relationship. To do so, we compared performance across four PAL mapping conditions: visual-verbal, verbal-verbal, visual-visual and verbal-visual. We reasoned that if crossmodal mapping demand accounts for the PAL-reading relationship, both visual-verbal PAL and verbal-visual PAL should exhibit significant relationships with reading ability. The results were incompatible with the crossmodal hypothesis. Only tasks requiring verbal output (visual-verbal PAL and verbal-verbal PAL) significantly correlated with reading ability. In addition, visual-verbal PAL and verbal-verbal PAL were well represented by a latent "verbal output PAL" factor. Structural equation modeling showed that this factor fully accounted for the PAL-reading relationship; visual-verbal PAL did not add anything to the prediction of reading above and beyond this latent factor. The results are interpreted according to an alternative verbal account of the PAL-reading relationship.