Writing and rewriting arabic numerals: dissociated processing pathways?

The present paper describes two patients, AB and FS, whose deficit in transcoding verbal to Arabic numerals was greatly affected by the format of the input. Despite intact comprehension of written verbal numerals and otherwise intact production of Arabic numerals they both transcode inefficiently the former into the latter. Yet, their ability to write the same Arabic numerals on dictation was fully preserved. Moreover, in both patients' performance, a systematic error pattern emerged reflecting the influence of the lexico-syntactic structure of the input numerals in the transcoding processes. Current models of number transcoding may not easily account for this pattern of dissociation without postulating different code-dependent pathways for verbal to Arabic transcoding. Within a more parsimonious approach, it is tentatively suggested that spoken and written verbal codes activate with different efficiency the same transcoding algorithm.