Journal of Literacy Research

0 downloads 0 Views 535KB Size Report
The online version of this article can be found at: ... about spiders read a passage about spiders and answered wh-questions ..... Zip has stronger legs than Webby because he must catch his ... running across the yard when she is hungry.
Journal ofhttp://jlr.sagepub.com/ Literacy Research

The Effect of Background Knowledge on Young Children's Comprehension of Explicit and Implicit Information P. David Pearson, Jane Hansen and Christine Gordon Journal of Literacy Research 1979 11: 201 DOI: 10.1080/10862967909547324 The online version of this article can be found at: http://jlr.sagepub.com/content/11/3/201

Published by: http://www.sagepublications.com

On behalf of: Literary Research Association

Additional services and information for Journal of Literacy Research can be found at: Email Alerts: http://jlr.sagepub.com/cgi/alerts Subscriptions: http://jlr.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav Citations: http://jlr.sagepub.com/content/11/3/201.refs.html

>> Version of Record - Sep 1, 1979 What is This?

Downloaded from jlr.sagepub.com by guest on January 15, 2012

Journal of Reading Behavior 1979, Vol. XI, No. 3

THE EFFECT OF BACKGROUND KNOWLEDGE ON YOUNG CHILDREN'S COMPREHENSION OF EXPLICIT AND IMPLICIT INFORMATION1

P. David Pearson University of Illinois at Urbana-Champaign Jane Hansen University of New Hampshire Christine Gordon University of Calgary

Abstract. To investigate the applicability of schema-theoretic notions to young children's comprehension of textually explicit and inferrable information, slightly above-average second grade readers with strong and weak schemata for knowledge about spiders read a passage about spiders and answered wh-questions tapping both explicitly stated information and knowledge that necessarily had to be inferred from the text. Main effects were found for strength of prior knowledge [p < .01), and question type (p < .01). Simple effects tests indicated a significant prior knowledge effect on the inferrable knowledge (p < .025) but not on explicitly stated information. A follow-up study was conducted to verify the fact that the question type effect was not due to the chance allocation of inherently easier questions to one of the two question types. We found a reliable decrease in question difficulty attributable to cueing propositional relations explicitly in the text (p < .01). These data were interpreted as supporting and extending the arguments emerging from various "schema theories".

The present study was designed to assess the role that background knowledge plays in determining young children's ability to process relationships that are explicitly and fully specified in a text in comparison to those that are only partially specified by the same text. The study stems from the theoretical tradition of schema theory as explicated by Rumelhart and Ortony (1977), Rumelhart (in press) and Anderson (1977), as well as the research tradition established by Bransford and McCarrell (1973), Bransford and Johnson (1973), R. Anderson, Spiro, and M. Anderson (1977), and Anderson, Reynolds, Schauert, and Goetz (1977). Schema theory suggests that schemata serve at least two important functions during comprehension. First, they provide a framework for classifying concepts

Downloaded from jlr.sagepub.com by guest on January 15, 2012

202

Journal of Reading Behavior

presented in a text. Hence, the stronger the framework, the more likely concepts are to be classified and available for subsequent retrieval from long term memory. R. Anderson, Spiro. and M. Anderson (1977) found evidence for this generalization: Subjects who encountered certain items of food within a restaurant schema recalled those items better than subjects who encountered the same items within a grocery shopping schema. Bransford and Johnson (1973) found that subjects given a passage theme in the form of a title or a picture before reading an obscure passage recalled significantly more textually presented concepts than those not given a theme or given a theme after reading. Another corollary of this function is that concepts presented in text will be remembered as a function of the schema in which they are initially encoded into memory. Anderson et al. (1977) found that recall and comprehension of passages which invited two schematic interpretations (wrestling vs. a prison break or cardplaying vs. music rehearsal) was highly related to the background knowledge of the readers and/or environment in which the testing occurred. Physical education students in a physical education class setting chose the wrestling interpretation of the first passage but the card playing interpretation of the second; music students in a music class chose the alternative interpretation of each passage. Bransford and McCarrell (1973), using similar ambiguous passages, found that subjects tended to recall propositions that were consistent with the particular theme they were given. A second function that schemata serve in comprehension is to allow readers to fill in gaps not completely specified in the text—in the language of schema theory, assigning a default value for a variable whose value is unspecified in the text. Hence a reader who reads that John pounded in a nail is likely to tell you that he used a hammer to perform the deed, even though no instrument is mentioned in the text. While no studies have directly tested the effect of prior knowledge on slot-filling inferences, a study by Chi (in press) has suggested that young children with strongly developed knowledge structures perform a variety of cognitive tasks (recall, metamemorial monitoring, prediction) more like older children than like their knowledge deficient peers. In other words, apparent developmental changes in cognitive strategy may in fact turn out to be knowledge-based differences. Indeed, a study by Omanson, Warren, and Trabasso (1978), with its finding that inferencing performance was independent of story recall, invites such an interpretation. The present study, while continuing in the same tradition as those previously cited, differs in several specific features. First, unambiguous text (a second grade selection about spiders) was used. Second, young subjects (average ability second grade students) served as the population of readers. Third, comprehension was assessed by asking wi-question probes rather than eliciting free recall. Fourth, prior knowledge was manipulated by assessing how much each subject knew about the topic to be read rather than by implanting some schematic information in the text or the reader's mind prior to reading. Unambiguous text was used for reasons of ecological validity; it was reasoned that while ambiguous text can be used to establish the power of a variable, validation in natural text is necessary prior to wide-scale acceptance of a conclusion. Younger subjects were chosen to investigate the applicability of schema-theoretic operations to a novice rather than a mature population of readers. tV/i-comprehension probes served a twofold function: (1) to examine the schema-theoretic hypotheses with dif-

Downloaded from jlr.sagepub.com by guest on January 15, 2012

Background Knowledge and Comprehension

203

f erent dependent measures, particularly those commonly used in school settings¡ and (2) to look at the differential effects of prior comprehension of ideas that are fully versus only partially specified in the text. The strength of previously available schemata was assessed by asking students direct questions about the topic because it was felt that such a technique might ultimately be useful to classroom teachers as a diagnostic tool. The predictions from schema theory for the present experiment are straightfoward: (1) because well-developed schemata increase the likelihood that an item will be bound to a schema at all, students with high prior knowledge scores, in comparison to student with low prior knowledge scores, should exhibit superior comprehension of ideas explicitly stated in the text; however, (2) because of the added operation of filling slots not specified in the text, high prior-knowledge students should demonstrate an even greater advantage over low prior-knowledge students on probes requiring the comprehension of ideas only partially specified in the text. EXPERIMENT 1 Method Subjects. The subjects were second grade students who had. attained grade equivalent scores within a range of 2.5 - 3.7 on the Metropolitan Achievement Test, Form A, in September. Assuming eight months growth between September and the time of testing in May, the grade equivalent range of the students was estimated to be 3.3-4.5. The students were selected from four classrooms, two classrooms in each of the two schools in a middle class suburb of St. Paul, Minnesota. Twenty-five students were given a test on knowledge about spiders. The ten with the highest and the ten with the lowest scores were selected to participate in the experiment. The ten lowest (the weak schema group) received scores of 2 or 3 on the eight pretest questions. The ten highest (the strong schema group) received scores of either 5, 6 or 7. The mean number of correct responses given by the group with the weak schemata was 2.7 [SD = .81); the mean number correct for the strong schema group was 5.8 [SD = .63). This difference was significant [t =9.09, df = 18, p < .001). The difference between the two groups on the reading subtest of the Metropolitan Achievement Test was not significant. The mean for the weak schema group was 3.13 and for the strong schema group, 3.32 (f = .909, df = 18, p > .05). The difference between the two groups on the I.Q. was also not significant. The mean I.Q. for the weak schema group was 114.80 and for the strong schema group, 120.40 (f = 1.36, df = 18, p > .05). The two groups, though different in amount of background information on spiders, were similar in reading ability and measured I. Q. Materials. A list of eight pretest questions was prepared to assess background knowledge of spiders. A basal reader selection on spiders was rewritten to include additional information on spiders and a narrative line. The readability level of the revised selection was computed to be 2.8 by applying the Spache Readability Formula. The selection was typed on a primary typewriter. A list of twelve posttest questions was prepared using criteria from Pearson and Johnson (1978). Six of the questions fell into a category that Pearson and Johnson labelled textually explicit. Such questions are derived by performing a wA-transformation on some immediate

Downloaded from jlr.sagepub.com by guest on January 15, 2012

204

Journal of Reading Behavior

constituent of a sentence in the text, as in (2). They are identical to Bormuth's (1969) category of rote questions. Six questions fell into Pearson and Johnson's scriptally implicit category. Such questions, while derived from and related to the text, necessarily require the reader to refer to prior knowledge to generate an answer, as in (3). (1) (2) (3)

Webby [the spider] has two jaws in front of her mouth. These jaws have fangs at the end. Webby bites the insect with these fangs. What does Webby bite insects with? What part of Webby's body is nearly the same as part of a snake's body ?

Procedure The pretests were administered individually over a one-week period. Prior to administering the pretest questions the following directions were given: I have eight questions to ask you. I'll ask you each question and you tell me the answer so I can write it down. Some of the questions are hard so just tell me what you think is correct. Some of them you may not know, so then tell me you don't know. The first question is: The questions were then administered orally. One follow-up query was allowed per answer if the appropriateness of the answer was not clear. The prior knowledge probes were developed where possible to relate to background knowledge that was at least tangentially related to the questions that were asked in thé posttest. For example, one of the posttest questions related to the fact that spiders don't stick to their own webs. In the prior knowledge assessment, the general principle of lubrication to prevent sticking was assessed by asking why people put oil in a pan before frying an egg. Related to questions (2) and (3) above, the children were asked what fangs were. All of the oral responses were recorded verbatim and scored later. Responses were classified independently by each experimenter. There were no disagreements. After a one week interval, the students read the actual selection. A small vacant room in each school was used to test students individually. The following directions were given: Read this story to yourself. Read it just once. Read it carefully and don't hurry. If you meet some words you don't know pronounce them to yourself as best you can and then go on. When you have finished reading, return the story to me. Then I'll ask you some questions about the story. The twelve posttest questions were presented orally in an order that followed the story sequence; the six implicit and six explicit questions were interspersed. Again, all responses were recorded and scored independently by each experimenter; there were no disagreements. Results The posttest results for the two groups and for both question types are reported in Table 1.

Downloaded from jlr.sagepub.com by guest on January 15, 2012

Background Knowledge and Comprehension

205

TABLE.l Mean Number of Correct Responses on Post Test (Experiment 1) Question Types

Prior Knowledge X Groups Strong Schema Weak Schema Total

Explicit SD

4.70 3.80 4.25

(1.16) (1.69) (1.48)

X

Implicit SD

2.80 1.00 1.90

(1.62) (1.05) (1.02)

Total X

SD

7.50 4.80

(1.80) (2.30)

The strong schema group (M = 7.50) performed significantly better than the weak schema group (M = 4.80) overall, F(l,18) = 8.40, p < .01. There was a significant within subjects main effect for question type, F(l,18) = 30.32, p < .01 indicating that explicit questions (M = 4.25) were easier than implicit questions (M = 1.90). The prior knowledge by question type interaction was not significant, F(l,18) = 1.13,p> .05. Discussion The findings are partially consistent with the prediction made on the basis of earlier schema theoretic studies: There was an overall effect for prior knowledge on comprehension but no interaction effect. Taken by itself such a result would imply that students with well developed schemata were superior to those with weakly developed schemata on both types of questions. Such a result would lead to the conclusion that both functions of schemata—helping to instantiate textual concepts within a framework and filling in textual gaps though the assignment of default values were operative in the study. However, a visuaHnspection of mean differences between the two groups on textually explicit items [D = .90) versa textually implicit items [D = 1.80) prompted post-hoc Scheffe tests for the simple effects of prior knowledge within each question type. The Scheffe tests indicated that the effect of prior knowledge is more pronounced for implicit, F(l,18) = 7.46, p < .025, than for explicit questions, F = 1.13, p > .10.2 Significant simple effects for prior knowledge on both question types, coupled with a significant prior knowledge by question-type interaction would have provided stronger support for those predictions. Yet the results are in the right direction, and the prior knowledge effect for implicit questions appears quite reliable. The experiment has some limitations. First, it would be useful to replicate the effects with a real "population" of passages rather than a simple passage. Second, the question type effect is somewhat suspect. That is, it may be that the greater difficulty of implicit questions may have been an artifact of the particular set used in this

Downloaded from jlr.sagepub.com by guest on January 15, 2012

206

Journal of Reading Behavior

study. If this were the case then both the question type effect and the simple effect of prior knowledge on implicit questions could be questioned. To investigate this possibility a second study was carried out. EXPERIMENT 2 Method Subjects. Twenty second-grade students from a middle class suburban school who were reading at or within one year above grade level in the Metropolitan Achievement Test participated in the. study. Materials. The same passage used in Experiment 1 was rewritten in two forms. Ten questions were developed such that the five that were textually explicit in Form 1 would be scriptally implicit in Form 2 and vice-versa. This was accomplished by differentially adding and deleting information between Forms 1 and 2. For example, passage (4) might have been rewritten as (5) so that questions (6) and (7) would change categories from one form to the next. In other words, the questions remained constant from one form to another; however, the information available in the passage was varied between forms. (4)

(5)

(6) (7)

Right now Zip is very hungry. So he runs across the yard looking for insects. Zip has stronger legs than Webby because he must catch his food by running and jumping on it. Webby doesn't need to go running across the yard when she is hungry.... When an insect touches the web, it gets caught in the sticky threads. But Webby's feet don't stick to the web. Right now Zip is very hungry. So he runs across the yard looking for insects. Zip has stronger legs than Webby. Webby doesn't need to go running across the yard when she is hungry.... When an insect touches the web, it gets caught in the sticky threads. But Webby's feet don't stick to the web because she has a special oil on her feet that keeps her from getting caught. Why does Zip have stronger legs than Webby? Why don't Webby's feet stick to the web?

Procedures. With the omission of a prior knowledge test, the data were collected and scored exactly as in Experiment 1. Results. Posttest results (Table 2) indicated a significant main effect for question type, F(l,18) = 17.64, p < .001, but not for form. F{1,18) = .34, p > .05. Hence, the overall question type effect was replicated. The most interesting effect was the significant form by question type interaction, F(l,18) = 11.56, p < .01. The interaction results from the addition of a constant amount of difficulty for question type (a mean of about 1,00) to two sets of questions which differ inherently in average difficulty (2.8" versus 3.65). Discussion. These results suggest that while questions may differ from one another in their basic difficulty, there is a relatively constant amount of difficulty attributable to removing textually explicit information useful in answering the ques-

Downloaded from jlr.sagepub.com by guest on January 15, 2012

Background Knowledge and Comprehension

207

TABLE 2 Mean Number of Correct Responses on Post Test (Experiment 2) Question Types X Form I Form II Total

Explicit SD

3.3 4.2 3.75

(.68) (.92) (.91)

X

Implicit SD

3.1 2.3 2.7

(.88) (.81) (.72)

Total X

SD

6.4 6.5

(1.26) (1.18) —

tion. Hence the possible limitation noted in the discussion of Experiment 1 seems unwarranted. Comprehension of textually explicit information is easier than comprehension requiring integration of textual information and prior knowledge. And the previous conclusion that comprehension requiring such integration is especially facilitated by strong schemata remains plausible. GENERAL DISCUSSION In general these results confirm and extend the conclusions drawn by those who have previously demonstrated the effect of schemata on the comprehension of text. Students with well developed schemata on a topic are able to answer more questions about a passage than those with weakly developed schemata. This effect is particularly prominent when the questions require prior knowledge to be accessed. By way of extension, schema-theoretic functions have been shown to operate (1) with younger populations, (2) in typical environments with a typical text and (3) across different dependent measures. The results provide some hints about other studies that might be undertaken to detennine the practical implications of the prior knowledge effect that has once again been substantiated in this study. That prior knowledge facilitates comprehension, particularly inferential comprehension, seems well established. However, a teacher is likely to raise the question, so what! What do I do when students embark on a topic for which they have weakly developed schema? In fact cognitive psychologists (e.g., Bransford, Nitsch, & Franks, 1977) have raised the same point in discussing issues of learning or what they call "changing states of schema." What needs to be demonstrated is that students with weak schemata given direct instruction for concepts about the topic to be read subsequently comprehend more like students who come to the task with strong schemata than untreated students with weak schemata. In this regard, the salutary effect of semantic network pre-tegching found by Swaby (Note 1) and Schachter (Note 2) for specific populations of children with specific types of text should be mentioned. Sloan and Pearson (1978) found that almost any type of teacher intervention helps poor readers' comprehension of difficult technical

Downloaded from jlr.sagepub.com by guest on January 15, 2012

208

Journal of Reading Behavior

material. Armbruster (Note 3) found that adolescents' recall of expository text was facilitated by an independent study strategy which forced them to create semantic maps for the text. Nonetheless, much more instructional research is needed before we can specify the types of students and texts for whom and which such intervention is effective. The consistently greater difficulty of inference questions in relation to questions tapping textually explicit relations suggests that instructional studies which provide a direct strategy for making inferences may prove helpful. What needs to be determined is whether the difference is due to inherent differences in the tasks or to an artifact of instruction: children get much more practice at "literal" comprehension (Guzak, 1968). FOOTNOTES 1

Requests for reprints should be addressed to P. David Pearson, Center for the Study of Reading, 174 Children's Research Center, 51 Gerty Drive, Champaign, IL 61820. 2 Recall that in the original f test to evaluate pre-experimental differences between the two groups on I.Q. the obtained t = 1.36, p > .05, with the I.Q. advantage going to the strong schema group. Just to make doubly sure that I.Q. was not a factor in the posttest differences, separate analyses of co-variance were conducted for each question type using I.Q. as a covariate. The results for the analyses of co-variance replicated the results of the post hoc Scheffe tests: For explicit probes there were no significant differences between the two groups, F(1,17) = 1.56, p > .05; for implicit probes, the difference favored the strong schema group over the weak schema group, F(1.17) = 5.96, p < .026. 3 These means result from averaging the diagonally adjacent cell means in Table 2. REFERENCE NOTES 1. SWABY, B. The Effects of Advance Organizers and Vocabulary introduction on the Reading Comprehension of Sixth Grade Students. Unpublished doctoral dissertation, University of Minnesota, 1977. 2. SCHACHTER, S. An Investigation of the Effects of Vocabulary Instruction and Schemata Orientation Upon Reading Comprehension. Unpublished doctoral dissertation, University of Minnesota, 1978. 3. ARMBRUSTER, B. An Investigation of the effectiveness of "Mapping" text as a studying strategy of middle schools. Unpublished doctoral dissertation, University of Illinois, 1979. REFERENCES ANDERSON, R. C. The notion of schemata and the educational enterprise: General discussion of the conference. In R. C. Anderson, R. J. Spiro, & W. E. Montague (Eds.), Schooling and the acquisition of knowledge. Hillsdale, NJ: Lawrence Erlbaum Associates, 1977. ANDERSON, R. C., REYNOLDS, R. E., SCHALLERT, D. L., & GOETZ, E. T. Frameworks for comprehending discourse. American Educational Research Journal, 1977, 14, 367-381. ANDERSON, R. C., SPIRO, R. J., & ANDERSON, M. C. Schemata as scaffolding for the representation of information in connected discourse. American Educational Research Journal, 1978, 15, 433-440. BORMUTH, J. R. An operational definition of comprehension instruction. In K. Goodman & J. T. Fleming (Eds.), Psycholinguistics and the teaching of reading. Newark, DE: International Reading Association, 1969.

Downloaded from jlr.sagepub.com by guest on January 15, 2012

Background Knowledge and Comprehension

209

BRANSFORD, J. D., & JOHNSON, M. K. Considerations of some problems of comprehension. In W. G. Chase (Ed.), Visual information processing. New York: Academic Press, 1973. BRANSFORD, J. D., & MCCARRELL, N. S. A sketch of a cognitive approach to comprehension. In W. Weimer & D. Palermo (Eds.), Cognition and the symbolic processes. Hillside, NJ: Lawrence Eribaum Associates, 1974. BRANSFORD, J. D., NITSCH, K. E., & FRANKS, J. J. Schooling and the facilitation of knowing. In R. C. Anderson, R. J. Spiro, & W. E. Montague (Eds.), Schooling and the acquisition of knowledge. Hillsdale, NJ: Lawrence Erlbaum Associates, 1977. CHI, M. T. Knowledge development and memory performance. In M. Friedman, J. P. Das, & N. O'Connor (Eds.), Intelligence and ¡earning. New York: Plenum Press, in press. GUZAK, F. J. Diagnostic reading instruction in the elementary school. New York: Harper & Row, 1972 OMANSON, R. C , WARREN, W. H., TRABASSO, T. Goals, inferential comprehension, and recall of stories by children. Discourse Processes, 1978, 1, 323-336. PEARSON, P. D., & JOHNSON, D. D. Teaching reading comprehension. New York: Holt, Rinehart, & Winston, 1978. RUMELHART, D. E. Schemata: The building blocks of cognition. In R. J. Spiro, B. C. Bruce, W. F. Brewer [Eds.], Theoretical issues in reading comprehension. Hillsdale, NJ: Lawrence Erlbaum Associates, in press. RUMELHART, D. E., & ORTONY, A. The representation of knowledge in memory. In R. C. Anderson, R. J. Spiro, & W. E. Montague (Eds.), Schooling and the acquisition of knowledge. Hillsdale, NJ: Lawrence Erlbaum Associates, 1977. SLOAN, M., & PEARSON, P. D. The effect of teacher intervention strategies on poor readers' comprehension of technical matter in a technical vocational school. The Journal of Vocational Education Research, 1978, 3, 9-20.

Downloaded from jlr.sagepub.com by guest on January 15, 2012