Verbal Learning in Boys With P-Type Dyslexia, L-Type Dyslexia, and Boys Without Learning Disabilities: Differences in Learning Curves and in Serial Position Curves*

Twenty-five boys with P-dyslexia, 23 with L-dyslexia, and 26 boys without reading disabilities were administered the Digit Span (Forward and Backward) and the Dutch version of the Rey Auditory-Verbal Learning Test. Compared to normal boys, dyslexic boys exhibited reduced scores on Digits Backward and recalled fewer words during the five learning trials. Nonlinear modeling of the data for the five learning trials revealed that dyslexic boys showed smaller learning parameters than did normal boys and that Ldyslexic boys exhibited more loss of information during learning than did P-dyslexic boys. In dyslexic boys, the word-list primacy effect was strongly reduced. In normal boys, but not in dyslexic boys, Digits Backward correlated moderately with the primacy measure. The results suggest that reduced word-list learning in dyslexics is a consequence of a temporal ordering deficit rather than a rehearsal deficit. Children with dyslexia tend to show deficits across a variety of verbal memory measures. Many studies, including our own, have demonstrated impaired performances on memory span and auditory verbal learning tasks (e.g., Levin, 1990; McDougall, Hulme, Ellis, & Monk, 1994; Van Strien, Bakker, Bouma, & Koops, 1990). The deficit on span measures seems to be specific to tasks requiring phonological processes (Share, 1994), although Van Strien et al. (1990) reported poor performances of dyslexic boys on a visuospatial span measure (Corsi block-tapping). Hulme and Mackenzie (1992) interpreted phonological short-term memory problems reflecting a rehearsal deficit in the articulatory loop. Diminished free recall of supraspan word lists can be explained as a consequence of either shortor long-term memory deficits, depending on the shape of the serial position curve. Normal subjects tend to recall the early items (primacy effect) and late items (recency effect) in a word list better than the midlist words. The primacy effect is thought to be associated with long-term memory (LTM), while the recency effect is thought to be associated with short-term memory (STM). In many patients with defective learning abilities, the primacy effect is reduced, which could indicate a LTM problem. According to Capitani, Della Salla, Logie, and Spinnler (1992), the interpretation of the primacy effect as associated with LTM and the recency effect as associated with STM is controversial. Like in other studies, these authors found a poor correlation between a span measure of short-term memory and the word-list recency measure. Moreover, they found a moderate correlation between span and the word-list primacy mea146 JAN W. VAN STRIEN sure. Capitani et al. interpreted their results in terms of a model of verbal short-term memory comprising two components, one responsible for span and one responsible for recency (as a inherent feature of any memory system with gradual loss of information over time). In addition to the serial position curve, the learning curve can also be examined. With reference to supraspan word-list learning, several authors have distinguished between memory span and learning capacity (Blachstein, Vakil, & Hoofien, 1993; Lezak, 1995). Memory span is the capacity to learn information in a single trial, while learning capacity reflects the ability to add information from trial to trial. Because there are limits to the number of words that can be learned from a supraspan word list, the learning curve is not linear but asymptotic, with less new words added in the later trials. This curve can be expressed in a nonlinear model with two different components: the proportion of information added in each trial and the proportion of already learned information forgotten in each trial. The two components can be estimated by means of the least-squares criterion (see Results). The serial position and learning curves can provide additional information about the nature of the memory deficits in dyslexics. Therefore, the purpose of the present study was to compare dyslexic boys and normal boys on the various word-list learning parameters. A distinction was made between two subtypes of dyslexia: Pand L-type (Bakker, 1990). According to Bakker’s (1990) neuropsychological model for the normal and deviant learning-toread process, the relative participation of leftand right-hemisphere functions in the process of reading alters with development. In the initial stage of the learning-to-read process, there is a predominance of right-hemispheric involvement: the novice reader will use visuoperceptual analysis to discriminate and identify the perceptually complex graphemic information. In the course of normal development, the grapheme identification becomes an automatism. The child will then switch to semantic and syntactic strategies, predominantly generated by the left hemisphere. This shift in hemispheric involvement has been demonstrated in electrophysiological studies (Bakker & Licht, 1986; De Graaff, 1995; Licht, Bakker, Kok, & Bouma, 1988). Reading disabilities develop when these respective hemispheric balances are disturbed. Children with P-type dyslexia are slow but relatively accurate readers, with a fragmented style of reading. Children with L-type dyslexia are relatively fast but inaccurate readers. Children with P-dyslexia rely primarily on righthemispheric visuoperceptual processing, and are unable to switch to left hemispheric, semantic strategies in the later stages of the learning-toread process (perceptual reading type). Children with L-dyslexia employ left hemispheric, linguistic strategies prematurely from the very beginning of the learning-to-read process (linguistic reading type). Examination of cognitive functioning has provided good external validity for the Pand L-type classification (Bakker, 1994; Bakker, Licht, & Van Strien, 1991). Van Strien et al. (1990) found lowered memory performances in both Pand L-types when compared to normal readers, but no differences in memory functioning between the two dyslexia subtypes. The present study explored whether the additional memory measures yielded differences in learning strategies utilized by Pand Ltypes.