How discussion questions influence children's story understanding

This research examined the influence of discussion questions with a dual focus on important and implied text ideas in children's story understanding. In a three-week instructional study, questions with such a focus formed the basis of the treatment for the four experimental groups, while control group questions came from the basal series teachers' manuals containing the experimental stories. Questions for each experimental group differed in focus (prediction or review) and in whether they required answer justification. Comprehension of 106 third-grade children randomly assigned to the five groups was measured with free and probed story recalls. Results of mixed hierarchical analyses indicated that justifying answers significantly improved the performance of groups asked prediction questions but not groups asked review questions. There was also a nonsignificant trend toward superior performance of the experimental groups over the control group, suggesting that importance and inexplicitness of ideas are appropriate foci in designing discussion questions. Fielding, Anderson, & Pearson HOW DISCUSSION QUESTIONS INFLUENCE CHILDREN'S STORY UNDERSTANDING When Frank Smith (1975, p. 34) called comprehension "the condition of having one's cognitive questions answered," he was using questions as a generic term for a whole set of activities and cognitions that skilled readers seem to engage in--predicting, reviewing, making inferences, and modifying interpretations, to name a few. Many researchers and practitioners, in fact, seem to view teachers' questions as a means of modeling and engaging children in various active processes that a skilled reader carries out independently to make meaning while reading. The key to making questions useful in that way may be asking systematically derived and strategically placed questions whose purpose and pattern eventually might be internalized by children--because, after all, the ultimate goal of comprehension instruction is student independence in reading, not giving answers to teacher-posed questions. The present study is an attempt to integrate two lines of research about comprehension instruction to devise a method of generating and asking discussion questions about complex stories. The purpose is to study the role of discussion questions with a dual focus on important and implied text ideas in children's comprehension of discussed stories and stories that they read independently. Related Literature The present study fits into a body of research about the nature of text and teacher questions and their effects on comprehension that has grown considerably, especially in the last decade. Classroom observations of comprehension instruction (Durkin, 1978-79; Guszak, 1967; Mason & Osborn, 1982; Morrison, 1986) and analyses of the instructional suggestions in basal reader teachers' manuals (Beck, McKeown, McCaslin, & Burkes, 1979; Durkin, 1981) have documented the preponderance of story questions; the emphasis they place on low-level, factual information while failing to develop plot or event sequences through lines of questions about stories; and the tendency to ask questions that may distract readers from a story's central content. Such descriptions of the state of classroom comprehension instruction have spurred a number of investigations into the changes in instruction that might lead to more desirable learning outcomes. Consistent with current views of reading as a schema-theoretic process (e.g., Adams & Collins, 1979; Anderson & Pearson, 1984; Rumelhart, 1977, 1980), these investigations highlighted the role of background knowledge in comprehension--both in its role of providing a prototypical organizing framework, a story schema for example, that enables readers to anticipate and organize text ideas (e.g., Fitzgerald, 1984; Fitzgerald/Whaley, 1981; Stein & Glenn, 1979), and in its contribution of the raw material out of which crucial story inferences can be made (e.g., Paris & Lindaeur, 1976; Pearson, Hansen, & Gordon, 1979). With respect to the role of questions, most notable are two general findings: (a) both training and practice in answering inference questions can improve not only children's ability to answer inference questions but also their general understanding of what they read (Gordon & Pearson, 1983; Hansen, 1981; Hansen & Pearson, 1983; Raphael & McKinney, 1983; Raphael & Pearson, 1985; Raphael & Wonnacott, 1985; Redfield & Rousseau, 1981; Sundbye, 1987); and (b) attention (through questions or otherwise) to the typical structure of stories (e.g., Fitzgerald & Spiegel, 1983; Gordon & Pearson, 1983; Idol, 1987; Singer & Donlan, 1982) or to the information otherwise identified as central or important in stories (Beck, Omanson, & McKeown, 1982; Omanson, Beck, Voss, & McKeown, 1984) can improve comprehension. The present study addresses several gaps in our knowledge about the role of story discussion questions in comprehension instruction: (a) the lack of integration between our knowledge about the structure of stories and our knowledge about inferential reading comprehension; (b) our incomplete knowledge about the effectiveness of a special class of inference questions, those that require predictions; and (c) our failure to test systematically the common-sense notion that it is beneficial for children to justify their answers to discussion questions with evidence from the text or their background knowledge. Discussion Questions 2 Fielding, Anderson, & Pearson These issues are important in planning effective text-based instruction that centers on discussion questions. First, although the value of both central or important story questions and inferential questions has been documented, the relationship between the two has not been investigated. Even in stories for young children, what is stated in the surface structure usually gives only a partial account of the complex relationships one must recognize to understand and appreciate the story. For example, characters' goals, the relationship between their goals and their actions, and the conflict between two or more characters' goals often are left to inference, although they provide the locus of causality in stories (Bruce, 1980a, 1980b, 1984; Bruce & Newman, 1978). Yet it is not clear how any of the most prevalent story structure theories represent inferences or account for their role in the comprehension of complex narratives. Furthermore, a close look at the questions in the redesigned story lessons of Beck and her associates (Beck et al., 1982; Omanson et al., 1984) shows that while all questions tapped central story content, less than one-third required inferring even the most straightforward connections between different sentences, and only half of those required integration of information in noncontiguous sentences or integration of text information with background knowledge. The general approach to potentially difficult text ideas in the work of Beck et al. (1982) seems to be to supply the difficult ideas (e.g., that raccoons, because of their masks, might look a little like bandits) instead of leading children to figure them out through questions and discussion. On the other hand, it is not clear how theories of inferential reading comprehension explain which inferences, out of the almost limitless number that could be generated about a complex story, are the most important ones for story comprehension. Trabasso and his colleagues (Nicholas & Trabasso, 1980; Omanson, 1982; Trabasso & Nicholas, 1980; Trabasso, Secco, & Van Den Broek, 1984; Warren, Nicholas, & Trabasso, 1979) suggest that out of all the inferences that could be made while reading a story, the ones most critical are those necessary for the comprehension of an event chain in a narrative. The degree to which inference instructional studies have focused on such inferences is unclear. In Hansen's (1981; Hansen & Pearson, 1983) studies, for example, important text ideas were the basis for inferences in the group taught an inference-making strategy, but only a strategy for what Warren et al. call slot-filling inferences was taught. The training strategy for making inferences was based on selecting important text ideas, then having students predict what would happen in the story from their prior experience in similar situations. For Hansen's inference question-only group (the group that practiced answering a lot of inference questions), it is not clear whether any inferences required what Warren et al. (1979) call text connecting nor whether the inference questions tapped important text ideas. Gordon and Pearson (1983) and Raphael and her colleagues (Raphael & McKinney, 1983; Raphael & Pearson, 1985; Raphael & Wonnacott, 1985) taught strategies for making both textconnecting and slot-filling inferences, but it is unclear what role importance played in the selection of examples for instruction or in the construction of inference questions. Besides this key issue of no research with a combined focus on importance and explicitness of the story ideas that are targeted with questions, the present research also addresses the issue that no studies have investigated the relative effects on comprehension of review and discussion questions. Although predicting is a characteristic of good readers (e.g., Collins, Brown, & Larkin, 1980; Fitzgerald, 1984; Fitzgerald/Whaley, 1981), a special case of inference-making (Pearson & Johnson, 1978), and an especially active and independent form of engagement with the text (e.g., Stauffer, 1975), few investigators have involved children in prediction through questions. The successful strategies of Hansen (1981; Hansen & Pearson, 1983) and Palincsar and Brown (1984) incorporated prediction. Furthermore, recent work by Anderson and his associates (Anderson, Wilkinson, Mason, & Shirey, 1987; Wilkinson, Anderson, & Wilson, 1987) showed that prediction questions produced better memory for stories than did word-level questions when each