Linguistic Focus and Good-Enough Representations: An Application of the Change-Detection Paradigm
Sturt, Patrick, Sanford, Anthony J., Stewart, Andrew, Dawydiak, Eugene, Psychonomic Bulletin & Review
A number of lines of study suggest that word meanings are not always fully exploited in comprehension. In two experiments, we used a text-change paradigm to study depth of semantic processing during reading. Participants were instructed to detect words that changed across two consecutive presentations of short texts. The results suggest that the full details of word meanings are not always incorporated into the interpretation and that the degree of semantic detail in the representation is a function of linguistic focus. The results provide evidence for the idea that representations are only good enough for the purpose at hand (Ferreira, Bailey, & Ferraro, 2002).
Recent work on discourse processing reveals growing evidence that discourse representations may vary in the detail they encode and may sometimes be underspecified (Ferreira, Bailey, & Ferraro, 2002; Sanford, 2002; Sanford & Sturt, 2002). Ferreira et al. claim that we represent language input at a degree of specification that is sufficient only for the situation that we are in-what they call good enough representations. The present article aims to explore a factor that might influence the degree of specification with which the meaning of a word is represented. We test the hypothesis that the degree of specification is influenced by information structure: A word's meaning is represented in more detail when it is focused in a sentence's information structure than when it is not. This is analogous to claims made by researchers in visual cognition that visual memory is selective, and that information that is "important" in some sense is represented in more detail than information that is not (e.g., Simons & Levin, 1997). Our experimental method is related to the change-detection technique used in visual cognition research (e.g., Simons & Levin, 1997), except that in our case, the stimuli are short texts rather than visual scenes. Participants are required to read the short texts twice, and to detect whether or not a word has changed from the first presentation to the second.
One important demonstration that word meaning is sometimes underutilized comes from the study of semantic illusions. With the Moses illusion (Erickson & Matteson, 1981), many people consider (1) as true, even though Moses did not put animals on the Ark:
Moses put two of each sort of animal on the Ark. True or false? (1)
Failures to detect the anomaly in (1) are genuine and have been attributed to a failure to retrieve the basic information that it was Noah, not Moses, who did this. The Moses illusion has typically been treated within a memoryretrieval framework (e.g., Reder, 1982, 1987), but other failures to detect anomalies reflect more clearly a failure to utilize (dictionary style) word meaning rather than aspects of encyclopedic knowledge. Thus Barton and Sanford (1993) tested variants of the following:
After an air crash on the border of France and Spain, where should the survivors be buried? (2)
The proportion of people who detected the anomaly depended on a variety of factors, including the general scenario. So, when bicycle crash was substituted for air crash, the proportion of detections increased dramatically. Barton and Sanford ( 1993) argued that survivors is less relevant in the context of bicycle crashes than air crashes, and that the amount of detail with which a word's meaning is represented in the discourse representation is a function of its fit to a context: Semantic detail increases as fit decreases.
Such illusions show that word meaning is not always fully specified in the discourse representation. However, a major question is whether there are any general principles governing the extent to which a word's meaning will be specified. One potentially important factor, explored here, is linguistic focus.
Rooth (1992, 1995) defines linguistic focus in terms of sets of alternative interpretations. A simple way to illustrate this is to consider declarative sentences like (4) as answers to questions like (3A-3C):
Who introduced Bill to Sue? …