Plural pronouns and the representation of their antecedents

Three experiments were carried out to investigate what kind of representation is accessed during the search for the antecedent of a plural pronoun. In the first two experiments, subjects read texts that contained two participants of different gender introduced by their proper names and, after finishing the texts, either a plural or a singular pronoun was presented. The subjects' task was to produce aloud a plausible continuation of the texts. The time between the presentation of the pronoun and the onset of the subjects' utterances was recorded. In the first experiment, the two participants were placed in the same or in different scenarios. In the second experiment, the two participants were presented in a conjoined noun phrase (John and Mary) or as subjects of different sentences (John ... and Mary ...). In the third experiment, the stimuli from Experiment 2 were used in a probe recognition task. The results showed that antecedents for pronouns are found by consulting a discourse model of the text, and that antecedents in the same scenario are likely to be represented as two different tokens linked by a plural address.