SEMANTIC INFORMATION

Semantic Information

Semantic Information

During reading, people create rich mental representations corresponding to the events described in a text. These representations are constructed using the semantic content communicated in the text, together with a reader's knowledge about the way in which such events typically unfold in the world. Much research in the text-processing literature has been directed toward determining the type of semantic information that is used in the construction of these models. There is strong evidence that these models involve some degree of perceptual simulation (e.g., Barsalou, 1999; Glenberg & Kaschak, 2002; Glenberg & Robertson, 2000; Zwaan & Taylor, 2006), and that they represent key aspects of the situation at hand (e.g., Therriault, Rinck, & Zwaan, 2006; Zwaan, Langston, & Graesser, 1995; Zwaan & Radvansky, 1998). What is important to note about these frameworks is that they tend to focus on the information that is used to generate discourse-level representations, rather than on the time course of the mechanisms involved in the construction of those representations themselves.

EXPERIMENT 1

In Experiment 1, we examined readers' sensitivity to spatial information. O'Brien and Albrecht (1992) demonstrated a reading time penalty for sentences containing information that contradicted previously mentioned spatial information in a story. For example, in one of the experimental passages that we examined in Experiment 1, a singer (“Carly”) is introduced as standing either to the side or at the centre of the stage. After several filler sentences, the critical target sentence, “After the first verse she moved to the centre of the stage where she would stay until the song ended,” appears. According to O'Brien and Albrecht, a reading time penalty would be expected when this sentence is presented in the version of the story that introduced Carly as already standing at the centre of the stage. An early sensitivity account predicts that this penalty should be observable on reading times to the region of text containing the inconsistent information (i.e., the prepositional phrase, “to the centre of the stage”). Alternatively, under a delayed processing account, we would expect to observe a reading time penalty only toward the end of the sentence containing the critical material (i.e., on the post-critical region) and not on the region containing the prepositional phrase itself. This is what was reported by Guzman and Klin (2000). However, as noted earlier, it is possible that the measure of processing that they used was not sufficiently sensitive to detect an effect on the critical material. We based our experimental materials on those of O'Brien and Albrecht, but ensured that the critical region did not occur in sentence-final position. We constructed our materials so that the same target sentence could be compared across our two conditions, thus allowing us to rule out any possible effects driven by lexical differences between the words in our critical regions. In addition, we used word-by-word, self-paced reading to examine whether readers exhibit a reading time penalty for the region of text containing the anomalous information. An early sensitivity would be compatible with the ...