Morphemic structure and errors in reading aloud nouns derived from noun and verb bases: a study with italian children

In previous studies (Traficante & Burani, 2003; Traficante et al., submitted) we found that inflected and derived words embedding a verb base are read more slowly and less accurately than words with adjective or noun bases. We hypothesized that, in reading aloud single words, the high number of base+suffix combinations, typical of Italian verbs, could lead to a low predictability of the suffix, and so to slower RTs and lower accuracy. Suffix predictability may have a strong role when the reader has difficulty in accessing whole-word representations, either because of the word low frequency or for a low reading ability. Eighteen 4th-5th grade Italian children with dyslexia (CwD) and 36 age-matched typically developing children (TDC) read aloud 42 noun-derived words (e.g., artista, artist) and 29 verb-derived words (e.g., punizione, punishment). Children’s reading errors were analyzed. The word constituent (root or suffix) and the error type (root substitution, suffix substitution, base word production, hesitation/suffix missing, other errors) were considered in relation to the grammatical class of the base word and to the reader’s ability. All children were more accurate with noun-derived than with verb-derived nouns. The morphemic boundaries were preserved in about 60% of the errors and about 36% of the errors involved the suffix. However, while CwD are likely to produce substitution errors, TDC showed a prevalence of hesitation/suffix missing errors more with verb-derived nouns (es. privazione, deprivation  priva…privazione) (73%) than in reading noun-derived nouns (27%). Logistic regression models show the influence of the suffix length on the error production of CwD, as 4-5 letter long suffixes are more preserved than the shorter ones, and of the base frequency on increasing the hesitation/suffix missing error produced by TDC. Data confirm the role of morphemic structure in reading aloud and show that the reading of CwD is mainly influenced by morpho-orthographic features (the letters chunks corresponding to the suffixes), while TDC’s reading is influenced by distributional properties of morpho-lexical representations.