Abstract
The goal of this study was to improve our understanding of the central cognitive functions of spelling: orthographic long-term memory, phoneme-to-grapheme conversion, and orthographic working memory. To do so, we present a methodological innovation to support vector regression lesion-symptom mapping, which uses a deficit spectrum index to identify brain areas associated with one of each pair of spelling deficits. Using this approach, we find evidence of distinct neural substrates that are selectively associated with each of the three central spelling components, supporting a componential functional architecture of spelling. The specific findings provide neural evidence relevant to various debates regarding the nature of these key cognitive processes and their relationships, namely, questions concerning the degree to which phoneme-to-grapheme conversion is distinct from general phonological processing, the existence of distinct lexical and sublexical routes for translating sound to print, and whether or not working memory is distinct from long-term memory or embedded within it. A better understanding of these issues has clinical implications in terms of our understanding of dysgraphic deficits and approaches to intervention.