Identification of written discourse topics by structure coherence and analogy strategies" General aspects and individual differences
N a c h s h o n M e i r a n b, P a z O r e f a
~'Department of Poetics and Comparative Literature, TeI-Aviv University, 1L-69978 TeI-Aviv, Israel h Department of BehavioralSciences, Ben-Gurion University of the Negev, 1L-84105 Beer-Sheva, Israel
Received September 1994; revised version April 1995 Abstract Categorically structured informative texts exhibit their discourse-topic in the beginning• When asked what the text is about, the skilled reader would deeply process the first proposition and skim through the rest for disconfirmation. S/he will, therefore, performpoorly on incoherent texts whose discourse-topic is displaced. Gifted and normal high-school students from a high socioeconomic neighborhood correctly identified more topics in coherent than in incoherent texts (Experiment 1). Low socioeconomic status subjects performed more poorly than the high socioeconomic status subjects on coherent texts, but better on incoherent texts (Experiment 2). Analogyimproved performance on coherent texts among low socioeconomic status subjects, who came from academic classes, but did not affect performance on incoherent texts. Experiment 3 studied discourse-topic identification of schematically organized texts by low socioeconomic status subjects, and found that analogies impaired it. The results are discussed in terms of the distinction between generalcomprehension and text-specific strategies.
1. Introduction An essential aspect of text comprehension is the identification o f the topic o f the discourse. The present study tackles questions concerning strategies of informative text comprehension• The major function of an informative text is to convey information in the most economical way. In this respect it differs, for example, from theliterary text, whose distinctive function is aesthetic. One type of informative text, called categorical, is organized like a taxonomy. Its principle of organization is similarity (Giora, 1985b). Such a text is highly redundant: Each of its propositions repeats information shared by the rest, while at the same time adding new informa* Corresponding author. Order of authorship is alphabetical.0378-2166/96/$15.00 © 1996 Elsevier Science B.V. All rights reserved SSDI 0378-2166(95)00040-2
R. Giora and al. / Journal of Pragmatics 26 (1996) 455~t74
tion. The text's shared information is its discourse-topic, on which the whole text is a comment (see detailed description later). The other type, called schematic, is organized in terms of spatial and temporal contiguity. Unlike thecategorical text, which requires abstraction in processing, the schematic organization relies on previous experience with spatial-temporal relations (see, e.g., Mandler and Johnson, 1977). Our major interest here concerns processing strategies of categorically organized informative texts. What does a reader do while reading an informative text? Two possible types of processing strategies come to mind: Onethat is text specific, and another that relies on general cognitive procedures. The distinction between task- (text-) specific and general cognitive processes is a crucial distinction in cognitive psychology. Several researchers suggest that exposure and practice lead subjects to form task-specific strategies. Consequently, skilled and less skilled processing are qualitatively different. Lessskilled processing is slow, error-prone, and effort consuming, while skilled processing is fast, relatively error free and requires only minimal effort (Ackerman, 1988; Anderson, 1983; Navon and Gopher, 1979; Shiffrin and Schneider, 1977; Schneider and Shiffrin, 1977). Task-specific strategies usually rely on the probabilistic structure of the domain: They are best at dealing with typical cases but...