SCHEMA ACTIVATION IN MEMORY FOR PROSE 1. Michael A. R. Townsend State University of New York at Albany

Transcription

1 Journal of Reading Behavior 1980, Vol. II, No. 1 SCHEMA ACTIVATION IN MEMORY FOR PROSE 1 Michael A. R. Townsend State University of New York at Albany Abstract. Forty-eight college students listened to two ambiguous passages about familiar topics under one of four context conditions. Half of the students were given appropriate contextual information about the passages either just before or just after hearing the passages. The remaining students either received no contextual information or received inappropriate contextual information about the passages. Following an interpolated activity, recall of the passage information was significantly lower for those subjects who were not able to activate an appropriate schema at the time of acquisition of the material. Students given inappropriate contextual information recalled less material than students given no contextual information. The results extend previous research with sentences to larger units of connected discourse and are discussed within the framework of schema shifting, a cognitive process essential in the comprehension of continuous prose. Communication depends on a reader or listener receiving linguistic information that refers to counterparts in his or her knowledge system. Comprehension occurs when the person has found "a set of schemata which appear to give a sufficient account of the information" (Rumelhart & Ortony, 1977, p. 112). Comprehension may fail either because the knowledge structures (schemata) do not exist, or because the person is unable to generate the appropriate structures in a particular situation. The generation of appropriate structures is aided by the semantic context surrounding the message. Bransford and his colleagues (e.g., Bransford & Franks, 1971; Bransford & Johnson, 1972; 1973) have dramatically shown that appropriate contextual information can facilitate comprehension of ambiguous passages, but only if the information is provided prior to receiving the passage. The converse expectation is that provision of an inappropriate semantic context should retard comprehension and recall since a meaningful interpretation cannot be constructed if the person is unable to utilize those schemata that best account for the incoming information. Since the utilization of an inappropriate context will prevent the reader or listener from independently searching for an appropriate contextual base for the information, as may happen

2 50 Jo urnal of Rea ding Beha vior when no contextual information is provided, prose text should be less comprehensible under an "inappropriate context" condition than under a "no context" condition. This issue has been investigated in twj) studies involving memory for sentences. Johnson, Doll, Bransford and Lapinski (1974} had adults listen to a list of unrelated ambiguous sentences such as "He kicked twice but got no change" under conditions in which they were first provided with an appropriate referential context ("vending machine" in this case), an inappropriate referential context, or no context. Under a presentation rate approximating that of ordinary conversation (i.e., one second pauses between sentences), there were significant differences between all three conditions in the mean number of sentences recalled, with recall being lowest in the "inappropriate context" condition and highest in the "appropriate context" condition. However, when seven second pauses between sentences were introduced the difference in recall between the "inappropriate context" and "no context" conditions did not reach significance. Similarly, a study by Doll and Lapinski (1974) using eight second pauses between unrelated sentences revealed no significant differences in recall or processing time between sentences cued with an inappropriate semantic context and those given no contextual information. Since it cannot be assumed that unrelated sentences and connected discourse are processed similarly, this study extends the research to prose text. Subjects METHOD The subjects were 48 students (37 females) enrolled in an undergraduate educational psychology course. Materials Two passages concerning topics for which high level schemata are presumed to exist were constructed. One paragraph was called "Making Toast" while the other was called "Making a Kite." The passages were constructed such that they did not explicitly mention the topic, yet the sentences remained consistent with the grammatical and semantic constraints of ordinary English. Three judges independently parsed the passages according to Bransford and Johnson's (1973) definition of idea units as "individual sentences, basic semantic propositions or phrases" (p. 393). Any disagreements were resolved by consensus. Using this procedure, the 98 word "Toast" passage contained 16 idea units and the 104 word "Kite" passage contained 20 idea units. Inappropriate contexts for the two passages were "Catching a Fish" (for "Making Toast") and "Changing a Light Bulb" (for "Making a Kite"). Audio tape recordings were made of an adult female reading the passages at approximately two words per second with normal pauses between sentences. Procedure Testing was carried out in groups of about ten students, each randomly assigned to one of four context conditions. Students in the Appropriate Context Before condition were given booklets containing appropriate titles for the passages and the titles

3 Schema Activation 51 were read before hearing the passages. Students in the Inappropriate Context condition were given booklets containing inappropriate'titles for the passages and the titles were read before hearing the passages. Students in the No Context condition received booklets which contained no contextual information concerning the passages. Finally, students in the Appropriate Context After condition were given booklets containing appropriate titles for the passages but each title was not seen until after the passage had been heard and written recall requested. Recall was required at the end of each passage and the instructions stressed that students be as accurate as possible but that they use their own words to express any ideas in the passage that could not be remembered exactly. A 60 second interpolated activity (nonverbal arithmetic problems) was inserted between the end of the passage and free recall. The order of the two passages was counterbalanced within each context condition. RESULTS Protocols were scored for any meaning-preserving approximation of a text idea unit. A sample of protocols was rescored, with reliability of scoring being r =.96. The following analyses are based on an arcsin transformation of the proportion of idea units recalled for each passage. There were no significant effects for sex or passage order. Thus, the main analysis was a 4 (context condition) x 2 (passage) repeated measures analysis of the transformed proportions. The means and standard deviations relating to this analysis are shown in Table 1. TABLE 1 Proportion of Recall in Each Context Condition for Each Passage Passage Context Condition» Toast Kite Appropriate Context Before No Context Inappropriate Context Appropriate Context After n = 12 for each condition

4 52 Journal of Reading Behavior There was a significant main effect for context condition, F(3, 44) = 13.70, p <.001. Three planned orthogonal comparisons indicated greater recall in the two appropriate context conditions (combined) than in the JVo Context and Inappropriate Context conditions (combined), i(44) = 3.07, p <.01. Furthermore, recall was greater in the Appropriate Context condition than in the Appropriate Context After condition, f(44) = 2.88, p <.01. Finally, recall was lower in the Inappropriate Context condition than the No Context condition, f(44] = 1.69, p <.05 (one tail). The main effect for passage was also significant, F(l, 44) = 9.31, p <_.01, the "Toast" passage [ =.341) being better recalled than the "Kite" passage ( =.237). The interaction effect (context condition passage) was not significant, F(3, 44) = 0.76, p >.50. DISCUSSION The results of this study indicate that potentially meaningful prose text can remain relatively incomprehensible when a context does not activate appropriate schemata at the time of acquisition of the material. High levels of comprehension depend on relevant cognitive structures or schemata being engaged with the linguistic material as the result of contextual cueing. Furthermore, extending the work of Johnson et al. (1974) with sentences, these results demonstrate that activation of inappropriate schemata to comprehend prose passages is likely to be more debilitating to comprehension than providing no contextual information at all. It seems likely that the activation of inappropriate cognitive structures prevents a person from independently attempting to find an interpretation, perhaps idiosyncratic, that allows for greater instantiation. This finding is important to the concept of schema shifting. During the reading process the reader must engage an appropriate cognitive structure in order to comprehend the material. As the context of the material changes, so must the cognitive structures if comprehension is to be maintained. If a person fails to shift schemata as the context changes, then the previously activated schema, now inappropriate because of a change in context, is being used in an effort, albeit unsuccessful, to comprehend the new material. Current research indicates that there are both individual differences (Spiro & Tirre, 1980) and developmental differences (Townsend, 1978) in flexibility of schema engagement that result in differences in reading comprehension. The finding that activation of an appropriate schema after hearing the passage does not facilitate comprehension and recall over being given no contextual cueing replicates previous studies using ambiguous materials. To clarify the role of high level schemata in this study, a further 25 students were asked to write for approximately five minutes on each of the topics "Making Toast" and "Making a Kite." Although the free stories were approximately 25 percent longer (by line count) than the recall protocols, the proportion of idea units coinciding with the experimental passages was only.038 and.048 for the "Toast" and "Kite" passages respectively. Clearly, although the retroactive contextual cueing failed to provide a framework to facilitate retrieval of the passage information, all four contextual groups recalled more text information than could be expected on the basis of prior knowledge alone. Future research could profitably explore the interaction between contextual effects and text specificity.

5 Schema Activation 53 The importance of the semantic context that is active during acquisition of information cannot be overstated. Activation of an appropriate schema facilitates comprehension and recall while activation of an inappropriate schema is harmful to comprehension; this is important to the issue of schema shifting, a cognitive process that underlies the comprehension of continuous prose. FOOTNOTE 1 The author wishes to express his gratitude to Cindy Parry for her assistance with this study. Requests for reprints should be addressed to the author at Education B12, The University at Albany, 1400 Washington Ave., Albany, NY REFERENCES BRANSFORD, J.D., & FRANKS, J.J. The abstraction of linguistic ideas: A review. Cognitive Psychology, 1971, 2, BRANSFORD, J.D., & JOHNSON, M.K. Contextual prerequisites for understanding: Some investigations of comprehension and recall. Journal of Verbal Learning and Verbal Behavior, 1972, 11, BRANSFORD, J.D., & JOHNSON, M.K. Considerations of some problems of comprehension. In W.G. Chase (Ed.), Visual information processing. New York: Academic Press, DOLL, T.J., & LAPINSKI, R.H. Context effects in speeded comprehension and recall of sentences. Bulletin of the Psychonomic Society, 1974, 3, JOHNSON, M.K., DOLL, T.J., BRANSFORD, J.D., & LAPINSKI, R.H. Context effects in sentence memory. Journal of Experimental Psychology, 1974, 103, RUMELHART, D.E., & ORTONY, A. The representation of knowledge in memory. In R.C. Anderson, R.J. Spiro, and W.E. Montague (Eds.), Schooling and the acquisition of knowledge. Hillsdale, NJ: Lawrence Erlbaum Associates, SPIRO, R.J., & TIRRE, W.C. Individual differences in schema utilization during discourse processing. Journal of EducationalPsychology, 1980, 72, TOWNSEND, M.A.R. Development of flexibility of schema shifting in prose comprehension. (Doctoral Dissertation, University of Illinois at Urbana-Champaign, 1977). Dissertation Abstracts International, 1978, 38 (10), 6029-A. (University Microfilms No ).