Cognitive and linguistic predictors of spelling in brazilian portuguese: a comparative study of students in the elementary school

This study investigated which cognitive–linguistic skills predict spelling in Brazilian Portuguese. The sample comprised 269 typically developing children (Grades 2–5) from one public and one private school. Analyses were conducted separately for Beginners (Grades 2–3; n = 119) and Advanced students (Grades 4–5; n = 150). The assessment protocol included word and pseudoword spelling, phonemic awareness, short-term phonological memory and working memory, lexical access, lexical orthographic knowledge (pseudohomophones and homophones), sublexical orthographic knowledge, morphological awareness, receptive and expressive vocabulary, visual discrimination, and visual memory. Multiple linear regression models with forced entry, adjusted for grade level and school type, were estimated separately for word and pseudoword outcomes. Among Beginners, the word-spelling model explained adjusted R2 ≈ .77; unique predictors were phonemic awareness, lexical orthographic knowledge (pseudohomophones and homophones), and morphological awareness. Receptive vocabulary showed a small negative coefficient, consistent with statistical suppression. For pseudowords, the model explained adjusted R2 ≈ .32, with phonemic awareness and morphological awareness as predictors. Among Advanced students, the word-spelling model explained adjusted R2 ≈ .68; unique predictors were lexical orthographic knowledge (pseudohomophones) and sublexical orthographic knowledge, with additional negative effects of RAN–Digits/Time and RAN–Objects/Errors, and a moderate positive effect of morphological awareness. For pseudowords, the model explained adjusted R2 ≈ .32, with only RAN–Digits/Time showing a marginal effect. Overall, the findings indicate a developmental shift in word spelling from early phonological predominance to greater reliance on stabilized orthographic representations and lexical access fluency, underscoring the multidimensional nature of orthographic acquisition.