A constituent comparison model of a picture-first verification task

In a recent critique of Carpenter and Just’s (1975) verification model, Tanenhaus, Carroll, and Bever (1976) pointed out that the estimates of the comparison parameter of the model varied considerably among tasks. In particular, estimates of this parameter were higher for picture-first than for sentence-first sentence-picture verification tasks. The picture-first estimates, however, were based upon an analysis of only a subset of the trials: namely, those in which the preposition “above” was embedded in the sentence. When a more comprehensive model is applied to the data of picture-first experiments, the estimates of the comparison operation are comparable to those based on sentence-first tasks.