| |
Abstract:
Williams Syndrome (WS) is a genetic disorder resulting in
moderate retardation, but surprisingly good language abilities, a
dissociation widely cited in arguments for language modularity.
Karmiloff-Smith et al. (1997) have questioned the validity of WS as
evidence for modularity, based in part on gender agreement errors
in French WS adolescents. This contrasts with excellent WS
performance with other morphosyntactic rules (Bromberg et al.,
1994; Clahsen et al., 1999). We believe the French results may have
reflected well-documented lexical retrieval problems in WS. We
report an elicited production study (based on Gordon, 1985)
examining one aspect of morphosyntactic knowledge which does not
depend on retrieval. 12 WS children (age 8-16) and 12
mental-age-matched control children (age 4-7) were prompted to
produce nouns in three forms: singular (rat, foot), plural (rats,
feet), and inside a compound (rat-eater, foot-eater). The WS
children almost never produced illicit compounds like "rats-eater",
but correctly allowed compounds like "mice-eater" (5% vs. 59%).
Controls performed equivalently. This contrasts sharply with the
performance of Specific Language Impaired (SLI) children (van der
Lely and Christian, 2000), who produced "rats-eater" violations 35%
of the time. This pattern belies the results from a standardized
comprehension test often cited as evidence for WS language
impairment-our WS children scored 2 years younger in age
equivalence than the SLI children. The collective results support
the view that the computational system of language is selectively
spared in WS.
|
