Writing arabic numerals in an agraphic patient.

We report on the writing of Arabic numerals in a patient whose alphabetical script was restricted to graphemic jargon (Schonauer & Denes, 1994). The analysis of writing errors in Arabic script over three testing sessions (4, 10, and 13 months after stroke) confirmed the separate processing of syntactic and lexical information in number production proposed by current models. The changing error pattern over time reflected some difficulties observed in developmental studies on the acquisition of Arabic numeral writing. Errors were mostly of the syntactic type and (at a certain stage) were based on the verbal form of the numerals. As reported in neuropsychological (Noel & Seron, 1995) and developmental (Power & Dal Martello, 1990; Seron & Fayol, 1994) studies, sum relations were more difficult to transcode than product relations within complex numerals.