This investigation compared the effects of explicit letter-sound training to holistic word training on the development of word recognition in a novel orthography paradigm. In a between-subjects design, participants were trained to read spoken English words printed in the alphabet script of Korean Hangul. Training took place over four separate sessions with assessment measures conducted throughout. Compared to the holistic training, the component training condition resulted in significantly better transfer to novel word forms and retention of previously learned items. Furthermore, compared to component training, holistic training yielded greater sensitivity to frequency. Variability in the holistically trained condition revealed bimodal distribution of performance: a high and low performing subset. Functional MRI measured cortical responses to the training conditions. Imaging results revealed generally greater responses in the "reading network" overall for the explicit component-based training compared to holistic training, in particular, regions of the inferior and superior parietal gyri as well as the left precentral gyrus. In a comparison of readers within the holistic group, we found that readers who implicitly derived the sublexical patterns in the writing system activated more of the reading network than those who did not sufficiently acquire this knowledge. This latter group primarily activated ventral visual regions. We conclude that explicit training of sublexical components leads to optimal word recognition performance in alphabetic writing systems due to the redundant mechanisms of decoding and specific word form knowledge.
展开▼
