8/9/2019 Kintsch W & van Dijk

http://slidepdf.com/reader/full/kintsch-w-van-dijk 1/1

CC/NUMBER 34AUGUST 23, 1993

This Week's Citation Classic ®

Kintsch W & van Dijk T A. Toward a model of text comprehension and production.

Psychol. Rev. 85:363-94,1978.[University of Colorado, Boulder, CO; and University of Amsterdam, The Netherlands]

The semantic structure of texts can be describedboth at the local microlevel and at a more globalmacrolevel. A model for text comprehensionbased on this notion accounts for the formationof a coherent semantic text base in terms of acyclical process constrained by limitations of working memory. Furthermore, the model in-cludes macro-operators, whose purpose is to

reduce the Information in a text base to its gist,that is. the theoretical macrostructure. Theseoperations are under the control of a schema,which is a theoretical formulation of thecomprehender's goals. The macroprocesses arepredictable only when the control schema canbe made explicit. On the production side, themodel is concerned with the generation of recalland summarization protocols. This process ispartly reproductive and partly constructive, in-volving the inverse operation of the macro-operators. The model is applied to a paragraph

from a psychological research report, and meth-ods for the empirical testing of the model aredeveloped. [The SSCI ® and the SCI ® indicatethat this paper has been cited in more than 800publications.]

The Long and Crooked Way Toward a Model of

Text Comprehension

Walter Kintsch Department of Psychology

University of Colorado Boulder, CO 80309-0345

In the early 1970s psychologists were busy(re)discovering meaning and discourse. Thecognitive revolution was young, vigorous, andvictorious, and, having discarded nonsensesyllables, we were impatient with word listsand even isolated sentences, and wanted towork with real texts. We turned to linguists and

logicians to show us how. Several authorsdeveloped systems for the representation of meaning which proved to be quite useful Inthat they allowed a great deal of psychologicalexperimentation with textual materials. My owncontribution to this effort was The Representa- tion of Meaning In Memory. 1 These systemsprovided unite of analysis, such as the "propo-sition,'' not to be confused with its logicalancestor, that allowed us to scale and measure

the salient characteristics of the texts to beinvestigated as well as the reader's responsesto these texts, which were often free recallprotocols. With "Toward a Model of Text Com-prehension and Production," T.A. van OijkandI added a new element in 1978: no longer concerned with merely analyzing language, weproposed to model how subjects understand

and produce that language. That is, we shiftedthe level of discourse from descriptive sys-tems to process models. This proved to be amajor step, leading up to the current scenewhich is characterized by rich experimentaland theoretical work on discourse processingfrom a large group of researchers in psychol-ogy, education, linguistics, and artificial intel-ligence.

What is important for me about this paper isthat it was a beginning, not an end. Our 1983book 2 Is an elaboration of this article, showinghow the framework developed there could beapplied to a broad range of discourse compre-hension phenomena. The next step beyondthat was a 1988 paper, 3 where the previouslyneglected problem of knowledge use was ap-proached in a new way: A hybrid model com-bining the virtues of production systems andconnectionist constraint satisfaction mecha-nisms was introduced to model the role of knowledge in comprehension.

Van Dijk, a Dutch linguist, and I began tocollaborate in 1975.1 had read his dissertationand he had read some of my work, and we were

both interested. He came to Boulder, and wespent several exciting days in animated dis-cussion. When I woke up the day after he left,and he arrived in Amsterdam, we both foundthat we could not remember anything from our discussions. On future occasions we werecareful to take copious notes, so that we couldwork for several months independently on theideas we had generated together. This soundsquaint in the days of e-mail, but it worked wellfor quite a few years. After the 1983 book our ways separated, van Dijk turning in a socio-HnguisrJc direction, whereas I chose to staywith the initial cognitiveemphasis on languagecomprehension and memory.

Fifteen years after its publication "Toward aModel of Text Comprehension and Produc-tion" is still a paper graduate students read tolearn about text processing. There have beenmany new developments in this field, but theyhave built on the foundation provided in thispaper, rather than supplanting it

1. Kinte ch W. The representation of meaning in memory . Hillsdale, NJ: Erlbaum, 1974. 279 p. (Cited 890 times.)2. Tan Dijk T A & Kintsch W. Strategies of discourse comprehension. New Yorke Academic Press. 1983. 418 p. (Cited 205 times.)3. Hnuch W. The use of knowledge in discourse processing: a construction-integration model. Psychol. Rev. 95:163-82.1988.

(Cited 95 limes.) Received June. 1993

CURRENT CONTENTS® ©1993 by ISI®