Transposed-character effects during learning to read: When does letter and non-letter strings processing become different?

Efficient reading requires the association of different letter identities with their positions in the written word. This leads to the development of a specialized mechanism for encoding flexible location-invariant letter positions through learning to read. In this study, we investigated not only the emergence and development of this position coding mechanism but also whether this mechanism is a consequence of the orthographic code (i.e., letter specific) or inherent to generic visual object recognition. To do so, the same–different matching task was used with children from Grade 1 to Grade 5 (Experiment 1) and with adults (Experiment 2). In both experiments, reference and target stimuli were composed of four-character strings (consonants, digits, and geometrical forms) and could be identical or different by transposing or substituting two internal characters. Analyses of response times, error rates, and discriminability indices revealed a transposed-character effect regardless of the type of characters in Grades 1 and 2, whereas transposed-character effects were greater for letter strings than for familiar non-letter strings in Grade 3, lasting up to Grade 5 as well as in adults. These results provided evidence in favor of a flexible position coding mechanism that is specific to letter strings, which emerges with reading experience as a consequence of parallel processing of letters within words.