Adequately learning to read and write is indispensable in our technological societies; however, it is an effortful process that requires years of formal education. In transparent alphabetic orthographies, such as Dutch, the crucial building blocks for literacy skills are the correspondences between letters and speech sounds (LS pairs). Previously, we demonstrated that LS pairs reliably produce a congruency-sensitive fMRI response in the superior temporal cortex (STC) of Dutch fluent readers, but not in dyslexic readers. Here, we used fMRI to investigate whether a similar congruency-sensitivity exists in STC of readers of opaque orthographies, such as English, where the relation between letters and speech sounds is unreliable, and may therefore not provide a similarly important basis for literacy. Eighteen subjects passively perceived congruent and incongruent audiovisual pairs that have different transparencies in English: letters / speech sounds (LS; irregular), letters / letter names (LN; fairly transparent) and numerals / number names (NN; transparent). We detected a congruency by pair-type interaction in STC bilaterally, that was produced by a strong congruency effect for NN, a weaker one for LN, and an incongruency effect for LS. No congruency effects for LS pairs were found in the predicted direction (congruent > incongruent), in contrast to the previous results in Dutch readers. These results indicate that through education, the STC becomes tuned to the congruency of transparent audiovisual pairs, but suggest different neural processing of irregular pairs. This orthographic dependency of LS integration has important implications for remediation strategies for dyslexia in different languages.

reading; orthographic consistency; education; fMRI; audiovisual; cross-linguistic