Phonological spelling in a DAT patient: The role of the segmentation subsystem in the phoneme-to-grapheme conversion

We are presenting a single-case study of a DAT patient whose writing output is severely impaired while performance in reading aloud and repetition is almost flawless. The large corpus of errors collected from written and oral spelling tasks shows two important characteristics: (1) in both tasks, OE relies on the non-lexical route for spelling and produces 'phonologically plausible errors" (PPEs) and non-phonologically plausible errors" (NPPEs), and (2) the proportion of NPPEs affecting four phonological features [+/- voiced], [+/- nasal], [a continuant], and [+/- rounded] is higher in written than in oral spelling. Analysis of PPEs and NPPEs reveals that the proportion of PPEs varies in inverse relation to the phonological complexity of the stimuli, i.e. fewer PPEs are produced in syllabically complex stimuli. According to our proposal, OE's functional lesion is localised in the segmentation subsystem of the phoneme-to-grapheme conversion mechanism. More specifically, OE suffers from a phonological impairment, that is, a lowered tolerance to syllabic complexity, which is exacerbated in any task, including phonological spelling, that requires an explicit segmentation of the auditory input form. A second deficit affecting the phonological working memory system is responsible for the production of the single feature errors. We suggest that the single feature errors are more abundant in written than in oral spelling because OE suffers from a deficit affecting the transfer from abstract graphemic representations to letter forms without affecting the transfer to letter names