Linguistic vs. Conceptual Sources of Implicit Agents in Sentence Comprehension

While Mauner, Tanenhaus, and Carlson (1995) argued that implicit participant information is derived from linguistic forms (e.g., verb argument structures), their results do not rule out a purely conceptual source for this information. Three experiments were conducted to resolve this question. Experiment 1 compared, in a timed contradiction task, the processing of passive and intransitive forms of verbs like sell that logically entail an agent, although there is no agent in the argument structures of intransitive verbs. Contradictory clauses elicited equivalently accurate contradiction judgments to short passive and intransitive agent contradictory sentences and nonagent contradictory controls. Experiments 2 and 3 used rationale clauses (e.g., … to raise money for charity) to probe readers' representations of short passives and intransitives equated for agent entailments. In Experiment 2, anomaly effects emerged in rationale clauses only when they followed agent-entailing intransitive sentences. In Experiment 3, when sentence-initial rationale clauses which engendered an agent expectancy were followed by short passive or intransitive sentences, anomaly effects emerged at the main verbs of only intransitive sentences. These results suggest that verb argument structures are accessed immediately and are not reducible to generic conceptual knowledge representations.