According to the dual-route model (DRM) of spelling, two processes operate in parallel: a lexical procedure, which relies on accessing word-specific orthographic memory, and a sublexical route, which relies on exploiting soundto-spelling regularities (e.g., phoneme-grapheme correspondences). Also according to the DRM, pseudoword spelling and consistency effects (words with consistent transcription spelled better than inconsistent ones) are considered markers of the sublexical procedure; inconsistent word spelling and the lexicality effect (words spelled better than pseudowords) together with word frequency effects (high-frequency stimuli spelled better than lowfrequency ones) are markers of the lexical procedure. The two-route view has been implemented in different spelling models (e.g., Barry, 1994;Hillis & Caramazza, 1991;Patterson, 1986;Perry, Ziegler, & Coltheart, 2002) and has received support from behavioural data derived from both adults with and those without spelling disturbances (for a review see Tainturier & Rapp, 2001) and children learning in various orthographic systems. As far as the ontogenetic acquisition of the two procedures is concerned, cross-linguistic studies have indicated that the more regular the writing system the more children rely on sublexical processing (for an English-Italian comparison see Marinelli, Romani, Burani, & Zoccolotti, 2015
AbstractWe examined how whole-word lexical information and knowledge of distributional properties of orthography interact in children's spelling. High-versus low-frequency words, which included inconsistently spelled segments occurring more or less frequently in the orthography, were used in two experiments: (a) word spelling; (b) lexical priming of pseudoword spelling. Participants were 1st-, 2nd-, and 4th-grade Italian children. Word spelling showed sensitivity to the distributional properties of orthography in all children: accuracy in spelling uncommon transcription segments emerged progressively as a function of word frequency and schooling. Lexical priming effects emerged as a function of age. When related primes contained an uncommon segment, 2nd-and 4th-graders preferred uncommon segments than common ones in spelling target pseudowords, thus inverting the response trend found in the control condition. A smaller but significant effect was present in 1st-graders, who, unlike 2nd-and 4th-graders, still preferred common segments, only slightly increasing the use of uncommon ones. A larger priming effect emerged for high-frequency primes than low-frequency ones. Results indicate that children learning to spell in a transparent orthography are sensitive to the distributional properties of the orthography. However, whole-word lexical representations are also used, with larger effects in more skilled pupils.