Randi C. Martin & A. L. Inglis
A recent study by Saffran, Schwartz and Linebarger (1998) examined aphasic patients' ability to judge the plausibility of sentences varying in syntactic complexity and type of mismatch between the verb and its arguments. There were two types of mismatch: (a) proposition-based sentences --- both nouns could plausibly be either agents or themes of the verb, but the overall proposition was implausible, e.g., "It was the bird that the worm ate", (b) verb-constrained sentences --- one noun was implausible as an agent or theme of the verb, e.g., "It was the hunter that the deer shot". Their counter-intuitive finding for a mixed group of patients with syntactic comprehension deficits was significantly more errors for the verb-constrained implausibility compared to the proposition-based, especially for their most syntactically complex sentences (i.e., object clefts). Age-matched controls failed to echo this pattern in reaction times, being faster to decide the verb-constrained type compared to proposition-based. Saffran et al. argued that the results for the aphasic patients supported the interactive theory of semantic and syntactic processing. When semantic constraints on role assignment were weak, as in the proposition-based sentences, the patients relied on the output of syntactic processes for role assignment, which were sufficiently preserved to allow for above-chance performance. When, however, semantic constraints strongly biased assigning a noun to an incorrect role (e.g., assigning "deer" as theme rather than agent of "shot"), then these constraints overrode the patients' weakened ability to use the output of the syntactic system. (See Cupples & Inglis, 1993, for somewhat related findings).
We present a case study of a patient, ML, who, like the patients reported by Saffran et al., showed syntactic comprehension deficits on sentence-picture matching tasks (in terms of error rates for auditory presentation but only latencies for visual presentation). On auditory plausibility judgments, however, he was significantly worse at the proposition-based items, compared to verb-constrained, on errors and times, especially on more complex items. One factor possibly distinguishing this patient from those of Saffran et al. and Cupples and Inglis is that ML has a semantic STM deficit. His comprehension deficit may result from a consequent difficulty in maintaining or re-activating lexical-semantic representations in object-relative forms rather than from a difficulty exploiting a syntactic parse. Thus, output from his syntactic system is not weakened, preventing override by semantic constraints. The results are interpreted in terms of the role of semantic STM in sentence processing.
Saffran, E., Schwartz, M. F., & Linebarger, M. C. (1998). Semantic influence on thematic role assignment: Evidence from normals and aphasics. Brain and Language, 62(2), 255-297.
Cupples, L. & Inglis, A. L. (1993). When task demands induce "asyntactic" comprehension: A study of sentence interpretation in aphasia. Cognitive Neuropsychology, 10(3), 201-234.