Spatial Descriptions and Referential Continuity

Three experiments were carried out in which subjects drew diagrams to depict spatial layouts corresponding to such three-sentence verbal descriptions as: "The apple is on the left of the banana. The banana is in front of the carrot. The carrot is on the left of the doughnut." In Experiment 1, a referentially continuous description in which each adjacent pair of sentences had a referent in common required less listening time and elicited more correct diagrams than a referentially discontinuous description in which the first and second sentences had no referent in common. In Experiment 2, descriptions in which the second and third sentences had no referent in common were no more difficult to remember than continuous descriptions. Both, however, were considerably easier than the discontinuous descriptions. In Experiment 3, the subjects read the descriptions rather than heard them. Its results replicated the earlier findings, and the times taken to read the individual sentences in a description suggested that subjects try to integrate each incoming sentence into a single coherent mental model, and that those sentences which cannot be immediately integrated are represented in a propositional form.