Is spelling memory improved by reading aloud ?

Résumé En

Reading a word aloud allows for better memory performance than simply reading it silently (MacLeod et al., 2010). This production effect is robust but little is known about the kind of representations that undergird it. Does the production effect rely only on phonological representations or also on orthographic representations ? In three experiments, the MacLeod et al. (2010) procedure was used : In a first phase (learning phase), pseudowords were read silently or aloud according to their color. In a second phase (test phase), a recognition task was used with all the pseudowords presented in the learning phase (the « targets » for which French participants had to answer « yes ») and with an equal number of fillers (participants had to answer « no »). In the first experiment, the fillers in the test phase (e.g., ganud) were phonologically and orthographically different from the targets (e.g., pasto). The results showed that recognition accuracy was greater in the reading aloud condition than in the reading silently condition (30% difference). In the second experiment, the phonological information was reduced : the fillers (e.g., pavto) differed by only one phoneme from the targets (e.g., pasto). The results showed a weaker but still significant production effect (13% difference). In the third experiment, phonological information was not available to distinguish fillers from targets, and only orthographic information was helpful. For instance, if the pseudoword « pasto » was presented in the learning phase, the corresponding filler « pasteau » was presented in the test phase (both pseudowords were phonologically identical in French but orthographically different). Although much smaller, the production effect still remained (5% difference). Overall, the results showed that the production effect is mainly based on phonological information but it can be observed even when only orthographic information is available.