Oral vocabulary and learning to read morphologically complex words

Orthographic skeletons have been proposed as one cognitive mechanism used to link oral representations of words with their written forms. Orthographic skeletons are expectations about the spellings of orally exposed words that have not yet been seen. This mechanism has previously been tested with morphologically simple (e.g., “vish”) and inflected novel words (e.g., “vished”) but not derived novel words (e.g., “visher”). This study aimed to test whether adult readers form orthographic skeletons for the stems of novel words (e.g., “vish”) upon oral training of derived (e.g., “visher”) and inflected (e.g., “vished”) forms of these words using a training paradigm followed by (a) a lexical recognition task (Experiment 1: 81 L1 English speakers) and (b) eye-tracking within a sentence reading task (Experiment 2: 80 L1 English speakers). Novel words were allocated to either a trained or an untrained condition, with each set consisting of items with predictable (e.g., vish for /ˈvıʃ/) and unpredictable (e.g., jayf for /ˈdʒeıf/) spellings. In both experiments, shorter latencies were observed for predictable compared to unpredictable words. Moreover, a significant interaction was found between training and predictability both in lexical recognition latencies and key eye-tracking measures (gaze duration and total reading time), which is an indication of an orthographic skeleton effect. However, the effect of the morpheme type was absent. The results are consistent with previous findings for simple and inflected novel words and demonstrate that the orthographic skeleton effect extends to derived novel words.