We assessed writing abilities in a cohort of 31 patients with a diagno
sis of DAT (in two subgroups, with minimal [MMSE 24-28] and mild [MMSE
16-23] levels of dementia), and 10 matched controls. Central aspects
of writing were assessed by both written and oral spelling to dictatio
n of 72 single words varying in frequency (high or low) and predictabi
lity of sound-to-spelling correspondences (predictable, unpredictable
and irregular). All subjects achieved better scores on high, as compar
ed to low, frequency words. The performance of both patient groups was
significantly affected by degree of predictability, and was equivalen
t in the written and oral spelling conditions. Phonologically acceptab
le alternative spellings (e.g.'wade' --> WAID) constituted the majorit
y of errors. More peripheral processes in writing were assessed by cop
ying and cross-case transcription of single letters. Subjects were mor
e successful at copying within case than transcribing across case. Per
formance was also better-substantially so for the mild DAT group-when
the target response in either task was an upper- rather than a lower-c
ase letter. There was considerable heterogeneity in performance on the
spelling and the letter tasks. Some patients (even in the more affect
ed DAT group) were unimpaired on both tasks, suggesting that dysgraphi
a is not a constant feature in early DAT. When writing deficits do bec
ome apparent, in the earliest stages of the disease the pattern is mos
t likely to be one of mild surface dysgraphia, a form of central dysgr
aphia; impairments in more peripheral aspects of writing tend to emerg
e once the disease has progressed beyond the minimal stage. (C) 1997 E
lsevier Science Ltd.