We report two studies that examine the role of semantic influences in
the assignment of thematic roles. Semantic factors were manipulated by
contrasting sentences in which one noun argument was a plausible fill
er of only one thematic role (e.g., the painting in The artist dislike
d the painting) with sentences in which both noun arguments were plaus
ible fillers of both thematic roles (e.g., The robin ate the insect).
Subjects were required to make plausibility judgments to sentences pre
sented auditorily. Experiment 1 examined RTs of normal subjects on the
plausibility judg ment task. In Experiment 2, the same sentences were
presented to aphasic patients identified as ''asyntactic'' comprehend
ers. In Experiment 1, RTs were speeded by semantic constraints on them
atic assignment, particularly when the role-constrained NP occurred ea
rly in the sentence (as in The painting was disliked by the artist). T
he aphasic performance patterns in Experiment 2 paralleled those of no
rmal subjects, but in greatly exaggerated fashion. The patients exhibi
ted high error rates on sentences where semantic constraints conflicte
d with the syntactically based assignments, even on sentences with can
onical (S-V-O) word order (e.g., #The deer shot the hunter). (C) 1998
Academic Press.