I L L I NO I S

UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN

PRODUCTION NOTE
University of Illinois at
Urbana-Champaign Library
Large-scale Digitization Project, 2007.
 T
E
C
H R
N E Technical Report No, 50

I p SCHEMA-DIRECTED PROCESSES IN
C o°O LANGUAGE COMPREHENSION

A R Richard C. Anderson
L T University of Illinois at Urbana-Champaign

S July 1977

Center for the Study of Reading

TA-iE
LilRARY V
OCT 7 198
AT !UP" - 'AIGN

UNIVERSITY OF ILLINOIS
AT URBANA-CHAMPAIGN BOLT BERANEK AND NEWMAN INC.
1005 West Nevada Street 50 Moulton Street
Urbana, Illinois 61801 Cambridge, Massachusetts 02138
 CENTER FOR THE STUDY OF READING

Technical Report No. 50

SCHEMA-DIRECTED PROCESSES IN
LANGUAGE COMPREHENSION

Richard C. Anderson

University of Illinois at Urbana-Champaign

July 1977

University of Illinois
at Urbana-Champaign Bolt Beranek and Newman Inc.
1005 West Nevada Street 50 Moulton Street
Urbana, Illinois 61801 Cambridge, Massachusetts 02138

A version of this paper was presented at the NATO International

Conference on Cognitive Psychology and Instruction, Amsterdam, June 1977.
The research reported herein was supported in part by the National
Institute of Education under Contract No. MS-NIE-C-400-76-0116 and
Grant No. HEW-NIE-G-74-0007. To appear in A. Lesgold, J. Pelligreno,
S. Fokkema, and R. Glaser (Eds.), Cognitive psychology and instruction,
New York: Plenum.
 Schema-Directed Processes

Schema-Directed Processes in Language Comprehension

In this paper I will develop the thesis that the knowledge a person

already possesses has a potent influence on what he or she will learn and

remember from exposure to discourse. I will begin by outlining some assump-

tions about the characteristics of the structures in which existing know-

ledge is packaged. Next, based on these assumptions, I will present a

speculative theoretical treatment of the processes involved in assimilating

the information and ideas in discourse. This is the topic that will be

given most attention in this paper. Data consistent with the theory will

be summarized. It should be emphasized in advance, however, that our

experiments to date show at most that the theoretical notions are inter-

esting and plausible. The research has not advanced to the point where

we have a firm basis for choosing between competing accounts. Finally,

I will make some observations about the implications of this research for

education.

Schematic Knowledge Structures

Like many others (Ausuble, 1963; Minsky, 1975; Schank & Abelson,

1975; Bower, 1976; Rumelhart & Ortony, 1977), I find it useful to postulate

that knowledge is incorporated in abstract structures that have certain

properties. These structures will be called schemata in deference to

Piaget (1926) and Bartlett (1932), who introduced the term to psychology.

What follows is an amalgam of my own thinking and that of other theorists.

A schema represents generic knowledge; that is, it represents what

is believed to be generally true of a class of things, events, or situations.

 Schema-Directed Processes

A schema is conceived to contain a slot or placeholder for each component.

For instance, a Face schema (Palmer, 1975) includes slots for a mouth,

nose, eyes, and ears. Encoding a particular object is conceived to be a

matter of filling the slots in the schema with the features of the object.

Part of schematic knowledge is the specification of the constraints on

what normally can fill the slots. An object will be recognized as a face

only if it has features that qualify as eyes, a mouth, a nose, and so on.

To be sure, the constraints on the slots in a Face schema are flexible

enough that we can tolerate considerable variation, as in a sketchy drawing

in a comic strip, the stylized and transformed representation in a cubist

painting or the exaggerated portrayal in a political cartoon (Gombrich,

1972). Nonetheless, there are limits beyond which an object is no longer

a face.

The encoded representation of a particular thing or event consists

of a copy of the schemata which were brought to bear in interpretation

plus the information inserted in the schemata's slots. Such particularized

representations are called instantiated schemata (cf. Anderson, Pichert,

Goetz, Schallert, Stevens, & Trollip, 1976). The slots in a schema may

be instantiated with information that could be said to be "given" in the

situation, or message, but often slots are filled by inference.

A schema is a knowledge "structure" because it indicates the typical

relations among its components. A Face schema will represent the relative

spatial positioning of the eyes and nose, for instance. Another attribute

of schemata with structural significance is that they exist at various

levels of abstraction and embed one within another (Rumelhart & Ortony,
 Schema-Directed Processes

1977). Contrast the knowledge that (1) a face has eyes, (2) an eye has a

pupil, (3) a pupil dialates in the dark. It is apparent that these propo-

sitions are arranged in decreasing order of importance to faces. This

variation in importance can be captured by assuming that Eye is a subschema

embedded in the Face schema; and that Pupil, in turn, is a subschema of

Eye. It is assumed that a person can employ a dominant schema without

necessarily accessing the knowledge available in embedded subschemata.

On the other hand, should the occasion demand it, the full meaning of a

subschema can be unpacked and a deeper interpretation given.

To comprehend a message is to place a construction upon it which gives

a coherent formulation of its contents. In schema terms, a "coherent

formulation" means a one-to-one correspondence between the slots in a

schema and the "givens" in the message. It is instructive to examine the

comprehension of a sentence devised by Bransford and McCarrell (1974) for

which a subsuming schema is not readily apparent: The notes were sour

because the seams split. The syntax is simple and the individual words

are easy, yet the sentence as a whole does not immediately make sense to

most people. However, the sentence becomes meaningful as soon as one hears

the clue bagpipe. Why is this clue effective? An answer is that it

enables the conception of a framework which maps onto a possible world.

Within the framework each word in the sentence can be construed to have a

referent with a sensible role to play in the possible world. That is to

say, the clue allows one to invoke a schema containing slots for the

objects, actions, and qualities mentioned in the sentence. The schema

gives a good account of the sentence and, therefore, there is the subjective

sense that it has been comprehended.

 Schema-Di rected Processes

Conceptions of the Reading Process

According to one view, reading is a "bottom-up" or "data-driven"

process (Bobrow & Norman, 1975). There is a series of discrete processing

stages each corresponding to a level of linguistic analysis. Analysis

proceeds from the most primitive low-order level to the most complex high-

order level. As a first step, feature analyzers are brought to bear to

discriminate horizontal, vertical, and oblique line segments; open and

closed loops; intersection with a horizontal plane; and so on. From these,

letters are identified. Strings of letters are analyzed into clusters

with morphophonemic significance. Words are recognized. Strings of words

are parsed into phrase constituents. Word meanings are retrieved from the

subjective lexicon. Eventually a semantic interpretation of a sentence

is produced. Sentence meaning is conceived to be the deterministic product

of the lower-order levels of analysis and, presumably, the meaning of a

text is a concantenation of the meanings of its component sentences.

Another view holds that reading is essentially a "top-down" or

"conceptually driven" process. Rather than analyzing a text squiggle by

squiggle, the reader samples it to confirm or reject hypotheses about its

content. In other words, reading is conceived to be a psycholinguistic

guessing game (Goodman, 1967). The reader's expectations represent a form

of preprocessing which should expedite and speed up subsequent analysis.

Occasionally expectation would be predicted to override the print, as

appears to happen when children make miscues in oral reading, substituting

semantically related words in place of those given.

 Schema-Directed Processes

There is an interesting difference between the bottom-up and top-down

theories about reading in their treatment of ambiguity. According to the

former view a high-order process does not affect low-order processes. Each

stage takes as its input the output from the preceding stage. If an

ambiguity arises at any stage, the alternative interpretations are sent

forward for resolution at a later point. For instance, it would be supposed

that all of the meanings of a homonym are accessed. Eventually, if the

message as a whole is not ambiguous, a process operating on syntax, seman-

tics, or pragmatics at the phrase, sentence, or text level, will permit a

choice among the homonym's senses.

From the perspective of a bottom-up model, reading is a matter of

growing a tree of possible interpretations. Any stage may add new branches,

or prune some of those already there. From the perspective of a model

that admits of possible top-down influences, on the other hand, not all

of the branches need be grafted on to the tree in the first place. Emerging

high-order expectations may forstall some interpretations before they occur.

With respect to the meaning of a homonym, it might be expected that normally

only the contextually most appropriate meaning would be accessed. this is

the implication of research by Schvaneveldt, Meyer, and Becker (1974)

using a lexical decision task. For instance, money was identified as a

word faster in the sequence save, bank, money than in either river, bank,

money, or the control sequences save, date, money or fig, date, money.

If all senses of a word were activated, bank should have primed money to

some extent even when preceded by river, but this did not happen. Con-

verging evidence has been obtained by Swinney and Hakes (1976) who found,
 Schema-Directed Processes

using a phoneme monitoring task, that a disambiguating context of a sentence

or two can constrain the interpretation of a subsequently encountered

homonym.

Of course, it is surely simplistic to imagine that reading is either

a bottom-up or a top-down process. Rumelhart (1976) has presented a

persuasive case that reading must involve continuous interactions among

many levels of analysis. I am dealing in this paper with how concepts

brought to a text influence comprehension, learning, and recall but, to

assert the obvious, the processes involved in analyzing the print itself

are also crucial.

Schemata and Text Interpretation

We have used several tricks to get people to bring different schemata

into play when reading text. Several studies have employed whole passages

which were ambiguous. For instance, Schallert (1976) constructed passages

that could be given two distinct interpretations. One of the passages

told of a character who was afraid that his best pitchers would crack in

the heat. The passage was entitled "Worries of a baseball manager" or

"Worries of a glassware factory manager." Scores on a multiple-choice

test--constructed so that the interpretation of pitcher and other similarly

ambiguous elements could be distinguished--indicated that the interpretation

of this and other passages was strongly related to the title.

In the absence of strong contextual cues, such as titles and intro-

ductions, the schemata by which people assimilate ambiguous passages can

be expected to depend upon their background and life situation. Anderson,

Reynolds, Schallert, and Goetz (1977) wrote the following passage:

 Schema-Directed Processes

Every Saturday night, four good friends get together. When

Jerry, Mike, and Pat arrived, Karen was sitting in her living
room writing some notes. She quickly gathered the cards and
stood up to greet her friends at the door. They followed her
into the living room but as usual they couldn't agree on
exactly what to play. Jerry eventually took a stand and set
things up. Finally, they began to play. Karen's recorder
filled the room with soft and pleasant music. Early in the
evening, Mike noticed Pat's hand and the many diamonds. As
the night progressed the tempo of play increased. Finally,
a lull in the activities occurred. Taking advantage of this,
Jerry pondered the arrangement in front of him. Mike inter-
rupted Jerry's reverie and said, "Let's hear the score." They
listened carefully and commented on their performance. When
the comments were all heard, exhausted but happy, Karen's
friends went home.

Most people interpret this passage in terms of an evening of cards but it

can be interpreted as about a rehearsal of a woodwind ensemble. Another

passage is usually seen as about a convict planning his escape from prison,

however it is possible to see it in terms of a wrestler hoping to break

the hold of an opponent. These passages were read by a group of physical

education students and a group of music students. Scores on a multiple-

choice test and theme-revealing disambiguations and intrusions in free

recall indicated that the interpretation given to passages bore the

expected strong relationship to the subject's background. An example

of an intrusion showing a card theme was, "Mike sees that Pat's hand has

a lot of hearts." One showing a music theme was, "As usual they couldn't

decide on the piece of music to play."

 Schema-Directed Processes

Of special significance to the discussion in this section were res-

ponses on a debriefing questionnaire. Subjects were asked whether they

became aware of another possible interpretation of either passage. The

interesting fact is that 62% reported that another interpretation never

occurred to them, while an additional 20% said they became aware of an

alternative interpretation during the multiple-choice test or when res-

ponding to the debriefing questionnaire. Less than 20% said they were

aware of a second interpretation while reading a passage. Many people would

not wish to place too much stock in retrospective reports. Still, these

are the results that would be expected on the basis of top-down, schema-

based processing.

Gordon Bower (cf. 1977) and his coworkers at Stanford have completed

several studies which parallel those done in my laboratory. One study

involved stories about characters who visit the doctor. An examination is

completed and the doctor smiles and says, "Well, it seems my expectations

have been confirmed." The base story was, in Bower's words, "a sort of

neutral Rorshach card onto which subjects could project their own meanings"

(1977, p. 8). The introduction to one version of the story describes the

character as worried about whether she is pregnant. Here subjects tended

to recall the doctor's remark as, "Your fears have been confirmed" or

simply, "You're pregnant." An alternate introduction described the main

character as a wrestler worried about being underweight. Subjects who

read this version remembered that the doctor told the character he was

gaining weight.
 Schema-Directed Processes

10

In another study, Bower and his associates used a story about a series

of mishaps that happen when a TV commercial involving water skiing is filmed.

Alternate introductions were written to cause the reader to identify with

either Harry, the boatdriver, or Rich, the water skier. On a recognition

test subjects tended to rate as explicitly part of the text statements

formulated from the perspective of the character with whom they were led

to identify. For instance, more subjects given the water skier than the

boatdriver introduction identified, "The handle was torn from Rich's grasp

as the boat unexpectedly jumped ahead," as a proposition from the text.

The reverse was true of the parallel formulation of the same episode written

from the boatdriver's perspective: "Rich slipped and lost control and the

handle went skipping across the water."

The general point illustrated by these experiments is that the meaning

of a text arises in an interaction between the characteristics of the

message and the reader's existing knowledge and analysis of context.

Ambiguous passages are useful for making transparent the role of world

knowledge and context. However, there is every reason to suppose that

they are equally important when comprehending material which would be said

to be "unambiguous." A message has an unambiguous meaning just in case

there is consensus in a linguistic community about the schemata that

normally will subsume it. The role of knowledge of the world is merely

less obvious to the psychologist doing prose memory research in these cases,

for the author, reader, and the judges who score the protocols employ

complementary schemata and thus give essentially the same interpretation

to the material.
 Schema-Directed Processes

11

Schemata and the Significance of Text Elements

Since Binet and Henri (1894; Thieman & Brewer, in press) worked with

French school children at the end of the nineteenth century, it has been

known that people are more likely to learn and remember the important

than the unimportant elements of a prose passage. No doubt authors pro-

vide linguistic cues to the important points in a text; however, I shall

argue that importance is largely a derivative of the schemata the reader

imposes on the text.

The schema brought to bear on a text will contain embedded subschemata

which generally can be conceived to form a hierarchy. The position of a

subschema in the hierarchy is one index of its importance. Significant

text information instantiates high-order slots in the structure. The schema

could be said to "give" such information its importance. It follows that

the importance of a text element would vary if readers were caused to

invoke schemata in which the text element played a greater or lesser role.

This hypothesis has been investigated in two lines of research in my

laboratory.

Anderson, Spiro, and Anderson (1977) wrote two passages--one about

dining at a fancy restaurant, the other a closely comparable story about

shopping at a supermarket. The same eighteen items of food, attributed

to the same characters, were mentioned in the same order in the two stories.

Subjects read one of the stories and then, after an interval, attempted

reca 11.

The first prediction was that the food items would be better learned

and recalled when presented in the restaurant narrative. The reasoning

 Schema-Directed Processes

12

was that a dining-at-a-fancy restaurant schema contains a more finely

articulated structure. That is, certain categories of food will be

ordered and served. And, there are constraints on the items that can fit

into these categories; hot dogs will not be the main course nor Koolaid

the beverage. Just about any food or beverage fits a supermarket schema.

This prediction was confirmed in two experiments.

The second experiment involved food categories determined on the basis

of a norming study to have a high or a low probability of being in an

individual's restaurant schema. An entree and a drink during the meal

are examples, respectively, of the low and high categories. An entree

is an essential element. No fine meal would be complete without one. A

drink during dinner is a less central, perhaps optional element. Subjects

who read the restaurant story recalled substantially more of the foods

and beverages from three high probability categories than subjects who

read the supermarket story. In contrast there was no difference between

the two passages on items from three low probability categories. This shows

that the restaurant narrative did not indiscriminately facilitate performance

as would be expected if it were overall more interesting, coherent or

memorable. Instead, as predicted, there was selective enhancement of

items from just those categories that have special importance in a res-

taurant schema.

The next prediction was that subjects would more accurately ascribe

foods to characters when given the restaurant story. Who gets what food

has significance within a restaurant schema whereas it matters not in a

supermarket who throws the brussel sprouts into the shopping cart. In
 Schema-Directed Processes

13

both experiments the conditional probability of attributing a food item

to the correct character given that the item had been recalled was higher

among subjects who received the restaurant than the supermarket story.

Finally, it was predicted that order of recall of food items would

correspond more closely to order of mention for subjects who read the

restaurant passage. There is not, or need not be, a prescribed sequence

for selecting foods in a grocery store, but at a fine restaurant it would

be peculiar to have a strawberry parfait before the escargot. In the first

experiment, the average correlation between recall order and order of

mention was significantly higher for the group that received the res-

taurant than the supermarket narrative. The trend was in the same direc-

tion but not significant in the second experiment, perhaps because recall

was attempted shortly after reading. There had been an hour and a half

interval before recall in the first study. Maybe surface order information

is available shortly after reading and this makes the generic order informa-

tion inherent in a schema superfluous.

The experiments just described used the trick of weaving the same

information into two different narratives in order to get readers to

assimilate that information to two different schemata. The device in a

second, parallel line of research was to ask subjects to read a narrative

from alternative points of view which, presumably, caused them to invoke

different schemata. Pichert and Anderson (1977) asked subjects to read

stories from one of two perspectives or no directed perspective. One of

the stories ostensibly was about what two boys do when skipping school.

They go to one of the boy's homes since his mother is never home on
 Schema-Directed Processes

14

Thursdays. It is a well-to-do family with a luxuriously appointed home.

It has a number of attractive features such as spacious grounds, a tall

hedge that hides the house from the road, and a new stone fireplace.

However, it also has some defects including a musty basement and a leaky

roof. The family has many valuable possessions--silverware, a coin collec-

tion, a color TV set. Readers were asked to approach the story from the

viewpoint of a burglar or a prospective homebuyer. Obviously a coin

collection is important to a burglar but unimportant to a homebuyer. The

opposite is true of a musty basement or a leaking roof. In a preliminary

experiment the average intercorrelation of rated idea unit importance

across three perspectives on each of two stories was determined to be

quite low, which is in itself evidence that schemata determine the sig-

nificance of text elements.

The next experiment manipulated perspective to investigate the effects

of schemata on text learning and recall. The previously obtained ratings

of idea unit importance were strongly related to immediate recall and,

independently, to delayed recall. This was true just of ratings obtained

under the perspective the subject was directed to take, not other possible

but nonoperative perspectives. Rating of importance under the operative

perspective was a significant predictor of recall in five of six step-

wise multiple regression analyses (one for each of three perspectives on

each of two stories). It was the only significant predictor in four of

these analyses.

The past few years have seen increasing refinement of the notion of

importance in terms of theories of text structure (cf. Kintsch, 1974;

 Schema-Directed Processes

15

Meyer, 1975; Rumelhart, 1975; Mandler & Johnson, 1977). These are more

properly regarded as theories of the structure of the schemata by which a

linguistic community normally will subsume a message, as some theorists

expressly acknowledge. But a text need not be read "normally." Depending

on the reader's goal, task, or perspective he or she may override the

conventions a linguistic community ordinarily uses to structure a text.

When the schema changes, then, so will the importance of text elements.

Possible Effects of Schemata on Encoding and Retrieving Text Information

In this section I shall give a more detailed account of some of the

mechanisms by which schemata may affect the processing of text information.

The phenomenon that I will concentrate on explaining is the primacy of

important text in recall illustrated in the preceding section.

Significant text elements might be better recalled because they are

better learned. In other words, the effect might be attributable to a

process at work when a passage is read. An attractive possibility is that

the schema provide the device by which a reader allocates attention. Extra

attention might be devoted to important text elements whereas insignificant

elements might be skimmed or processed less deeply. A second possibility

on the encoding side is that a schema provides "ideational scaffolding,"

to use Ausubel's (1963) apt term, for selected categories of text informa-

tion. A schema will contain slots for important information, but may

contain no slots, or only optional slots, for unimportant information.

According to this view information gets encoded precisely because there

is a niche for it in the structure. This is an interesting idea, but as

 Schema-Directed Processes

16

yet I have been unable to think of any implication of the ideational

scaffolding hypothesis that might permit it to be distinguished from the

regulation-of-attention notion.

The fact that people recall more important than unimportant text

elements might be due to processes at work when information is retrieved

and used, instead of, or in addition to, processes acting when the informa-

tion was initially encoded. There are several possible retrieval mechanisms

that fall out of a schema-theoretic orientation which might account for

the primacy of important text information in recall.

The first can be called the "retrieval plan" hypothesis. The idea is

that the schema provides the structure for searching memory. Consider for

illustration the burglar perspective on the story about two boys playing

hooky from school. The rememberer will possess the generic knowledge that

burglars need to have a way of entering a premise; that they are interested

in finding valuable, portable objects that can be fenced easily; that they

are concerned to avoid detection; and that they aim to make clean getaways.

Memory search is presumed to start with the generic concerns of a burglar.

Generic concerns implicate selected categories of text information. For

instance, the fact that all burglars need to enter the place to be robbed

is assumed to provide a mental pathway or implicit cue for the specific

proposition that the side door was kept unlocked. On the other hand,

information in a text which may have been encoded but does not connect

with the schema guiding memory search should be relatively inaccessible.

For example, the passage about the boys playing hooky from school asserts

that the house has new stone siding. Presumably there are no pointers in
 Schema-Directed Processes

17

a burglary schema to information of this type and, thus, this information

is unlikely to be retrieved even if it were stored.

We have termed another possible retrieval explanation the "output

editing" hypothesis. The assumption is that the schema contains within

itself an index of importance. The rememberer establishes a response

criterion based jointly on this index, motivation, and demand characteris-

tics. There are several variants on how output editing might work. In

crudest form, the subject simply might not write down information that

occurred to him or her because it falls below the response criterion.

I will consider, finally, the possibility that people may remember

more important than unimportant information because of a process of "infer-

ential reconstruction" (Spiro, 1977). There may be information missing

from memory either because the information was not stored, or because it

has been forgotten. The conceptual machinery of the schema and the informa-

tion that can be recalled may permit the rememberer to fill gaps by

inference. Anderson, Spiro, and Anderson (1977) have illustrated how the

process might work as follows. Suppose that a person is trying to recall

a story about a meal at a fine restaurant (see the preceding section).

The beverage served with the meal cannot be recalled, but since there is

a slot in a restaurant schema for such an item, the rememberer is led to

try to reconstruct one. If the information that beef was served for the

main course can be recalled, then red wine may be generated as a candidate

beverage. There are a couple of possible scenarious at this point. Red

wine might be produced simply as a plausible guess. A good guess and an

element actually remembered often will be indistinguishable to a judge,

 Schema-Di rected Processes

18

particularly one applying lenient, gist scoring criteria. Or, it might

be that once a candidate element, such as red wine, has been produced it

is checked against an otherwise inaccessible memory trace. To say this

another way, the process might be one of generation followed by recogni-

tion and verification (Kintsch, 1974). In any event, the foregoing gives

an account of the primacy of important text information, for the schema

is more likely to contain the concepts for reconstructing important than

unimportant elements.

Evidence for Encoding and Retrieval Benefits

We have completed several experiments to determine whether schemata

have independent effects on the encoding and retrieval of text information

and, if so, to begin to pin down the specific mechanisms that are responsible

My student, Jim Pichert, and I (Anderson & Pichert, 1977) asked under-

graduates to read the story about two boys playing hooky from school from

the perspective of either a burglar or a homebuyer. The story was recalled

once from the same perspective from which it had been read. Then everyone

recalled the story for a second time. Half the subjects did so again from

the same perspective. The other half changed perspectives. Based on

previously obtained ratings, a cluster of information important to a

burglar but unimportant to a homebuyer (e.g., a collection of rare coins),

and another cluster important to a homebuyer but unimportant to a burglar

(e.g., a fireplace), were identified. As expected, subjects produced on

the second recall a significant amount of new information--that is, informa-

tion that had not been recalled the first time--which was important in
 Schema-Directed Processes

19

the light of the new perspective, but which was unimportant in terms of

the perspective operative when the passage was read and recalled the

first time. There does not appear to be any way to explain this finding

solely in terms of encoding mechanisms. Thus, it seems to be rather

strong evidence for a retrieval mechanism independent of encoding.

In the preceding section three explanations within schema theory for

an influence on retrieval were discussed. To review briefly, the first

is the retrieval plan hypothesis: a new schema will furnish implicit cues

for different types of text information. The second is the output editing

hypothesis: when the schema changes different types of information are

above a response criterion. The third is the inferential reconstruction

hypothesis: a new schema will provide the concepts for infering different

categories of important but unavailable information.

In a follow-up study, Pichert and I replicated the retrieval benefit

identified in the experiment described above. We also collected subjects'

introspective descriptions of the processes of learning and remembering.

Most subjects discussed strategies and tactics for remembering in a manner

consistent with the retrieval plan hypothesis. A number said in so many

words that reviewing the concerns of a burglar or homebuyer caused them

to think of previously unrecalled information related to these concerns.

For example, one subject said, "I was thinking . . . was there anything

wrong with the house? And then I remembered the basement was damp."

Another said, "I remembered [the color TV] in the second one, but not

the first one. I was thinking about things to steal, things to take and

steal . . ."
 Schema-Directed Processes

20

The self-report protocols generally gave little support to the output

editing hypothesis. Most subjects insisted that they wrote down everything

they could remember. John Surber, another student of mine, manipulated

the incentive for recall. He reasoned that if the increment in recall in

the perspective-shift group were due to output editing, then the increment

would disappear under conditions of high incentive. What he actually found

was a difference in favor of subjects who shifted perspective regardless

of whether a 25¢ bonus was paid for each new idea. Thus, two strands of

evidence weigh against an output editing interpretation of the results of

this series of experiments. I do not wish to argue that people never

suppress information available to them, only that this probably was not a

major factor under the conditions that have prevailed in our research.

Spiro (1977) has obtained convincing evidence for reconstructive

processes in memory for discourse. Subjects read a story about a couple

engaged to be married. The man is strongly against having children. In

one version of the story the woman is elated to find this out because she

doesn't want children either. In the other version, she is horrified

because a large family is important to her. Several minutes after reading

the story subjects are told either that the couple did get married or that

they broke up. Based on the assumption that people's common-sense psychology

of interpersonal relations could be represented in terms of Heider's

principle of structural balance, Spiro predicted the particular types of

"reconciling errors"' subjects would introduce into their recall protocols

when the situation described to them was imbalanced. For instance, when

the couple got married despite the serious disagreement about having
 Schema-Directed Processes

21

children, it was argued that subjects would modify the story to reconcile

the incongruity by claiming, for instance, that "the problem was resolved

when they found out that Margie couldn't have children anyway." The

expected types of reconciling inferences appeared with increasing fre-

quency over a retention interval of six weeks. Subjects were more confi-

dent their inferences had been part of the story than they were that

propositions that had an explicit basis in the text had been present.

The perspective shift studies described earlier in this section all

showed a retrieval benefit but, for a couple of reasons, none clearly

established that schemata have an encoding influence as well. This was

the purpose of another experiment completed by Jim Pichert and me. A

story was recalled just once, from either the same perspective from which

it was read or a different one. Both the perspective from which the

story was read and the perspective from which it was recalled, which were

orthogonal factors in the design employed, had a substantial effect on

performance. Thus, both encoding and retrieval influences were demonstrated.

When asked how the assigned perspective affected the manner in which

the story was read, most subjects described a process of directing atten-

tion to important elements. For example, one subject told to take the

burglar perspective said, "I kept in mind all of the critical things a

burglar would be looking for such as getting in and out, the items that

it would be easy to move and take from the house itself." One assigned

the homebuyer perspective reported, "I spent most of the time looking for

items to be interested in when buying a house." A straightforward way to

get converging evidence on the regulation-of-attention hypothesis would

 Schema-Directed Processes

22

be to time subjects on chunks of text material whose importance has been

manipulated in some way. We haven't done experiments of this type yet.

In summary, in this section I have reviewed evidence that a schema

operative when a passage is read affects encoding, possibly by directing

attention to text elements that are significant in the light of the schema.

Evidence was presented which shows that later the schema affects remembering,

probably in part by providing the plan for searching memory. Schemata

probably also provide the basis for inferential elaboration when a passage

is read and inferential reconstruction when there are gaps or inconsis-

tencies in memory.

Implications of Schema Theory for Education

Text information is interpreted, organized, and retrieved in terms of

high-level schemata. It follows that the student who doesn't possess

relevant schemata is going to have trouble learning and remembering the

information encountered in stories and textbooks. Consider for illustra-

tion the description of an unfamiliar nation in a geography text (cf.

Anderson, Spiro, & Anderson, 1977). The mature student will bring to

bear an elaborate Nation schema which incorporates well-formed subschemata

for assimilating information about the topography, climate, economy,

culture, and political system. It is only a slight oversimplification

to say that the task for the advanced student is simply to fill the slots

in an already formed schema with the particular information in the text

about the unfamiliar nation. The information will be readily acquired

and, once acquired, easily retrieved when needed.

 Schema-Directed Processes

23

How about the young reader who, for the sake of the argument, will

be assumed not to possess a refined Nation schema? In the worst case a

description of an unfamiliar nation would be unintelligible to such a

reader, like the Bransford and Johnson (1973) passages for mature readers

when a schema-evoking context was not provided. More likely, the young

reader will have a partly formed Nation schema sufficient for some level

of understanding of the material, but which will not enable a representa-

tion of great depth or breadth.

Whether people possess the schemata appropriate for assimilating a

text should be an important source of individual differences in reading

comprehension. Smiley, Oakley, Worthen, Campione and Brown (1977) have

obtained some evidence suggesting that this may be the case. Good and

poor readers drawn from seventh-grade classes read one folktale and

listened to another. Following each story, they were tested for compre-

hension and recall. Under both reading and listening conditions, good

readers recalled a greater proportion of the stories and the likelihood

of their recalling a particular element was an increasing function of the

element's structural importance. Poor readers not only recalled less of

the stories, but their recall was not as clearly related to variations

in importance. Smiley et al. went on to show that it was necessary to

test children as young as first grade before finding another group which

showed as little sensitivity to gradations of importance as poor reading

seventh graders (see also Brown & Smiley, 1977). On the other hand,

Perfetti and Lesgold (in press) have summarized several studies which,

by and large, have not revealed substantial differences among good and
 Schema-Directed Processes

24

poor readers in sensitivity to sentence structure or text structure.

Thus, based on evidence already available, it is too early to say

whether variations in high level schemata, or facility in using these

schemata, will turn out to be a consistent difference between good and

poor readers. I hope only to have shown that this is a very reasonable

place to look for differences. If differences are consistently found

there will be implications for diagnosis, design of lesson materials, and

approaches to teaching.
 Schema-Directed Processes

25

References

Anderson, R. C., & Pichert, J. W. Recall of previously unrecallable informa-

tion following a shift of perspective (Tech. Rep, 41). Urbana, Ill.:

Center for the Study of Reading, University of Illinois, April 1977.

Anderson, R. C., Pichert, J. W., Goetz, E. T., Schallert, D. L., Stevens,

K. V., & Trollip, S. R. Instantiation of general terms. Journal of

Verbal Learning and Verbal Behavior, 1976, 15, 667-679.

Anderson, R. C., Reynolds, R. E., Schallert, D. L., & Goetz, E. T. Frame-

works for comprehending discourse. American Educational Research

Journal, 1977, in press.

Anderson, R. C., Spiro, R. J., & Anderson, M. C. Schemata as scaffolding

for the representation of information in connected discourse (Tech.

Rep. 24). Urbana, Ill.: Center for the Study of Reading, University

of Illinois, March 1977.

Ausubel, D. P. The psychology of meaningful verbal learning. New York:

Grune and Stratton, 1963.

Bartlett, F. C. Remembering. Cambridge, England: The Cambridge University

Press, 1932.

Binet, A., & Henri, V. La memoire des phrases. L'annee psychologique,

1894, 1, 24-59.

Bobrow, D. G., & Norman, D. A. Some principles of memory schemata. In

D. G. Bobrow and A. M. Collins (Eds.), Representation and under-

standing: Studies in cognitive science. New York: Academic Press,

1975.
 Schema-Directed Processes

26

Bower, G. H. Experiments on story understanding and recall. Quarterly

Journal of Experimental Psychology, 1976, 28, 511-534.

Bower, G. H. On injecting life into deadly prose: Studies in explanation

seeking. Invited address at meeting of Western Psychological Associa-

tion, Seattle Washington, April 1977.

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 and D. Palermo (Eds.), Cognition and

the symbolic processes. Hillsdale, N.J.: Erlbaum, 1974.

Brown, A. L., & Smiley, S. S. Rating the importance of structural units

of prose passages: A problem of metacognitive development. Child

Development, in press.

Gombrich, E. H. J., Hochberg, J., & Black, M. Art, perception, and reality.

Baltimore: Johns Hopkins University Press, 1972.

Goodman, K. S. Reading: A psycholinguistic guessing game. Journal of the

Reading Specialist, 1976, 4, 126-135.

Kintsch, W. The representation of meaning in memory. New York: John

Wiley & Sons, 1974.

Mandler, J. M., & Johnson, N. S. Remembrance of things passed: Story

structure and recall. Cognitive Psychology, 1977, 9, 111-151.

Meyer, B. J. F. The organization of prose and its effects on memory.

Amsterdam: North-Holland Publishing Company, 1975.

 Schema-Directed Processes

27

Minsky, M. A framework for representing knowledge. In P. H. Winston (Ed.),

The psychology of computer vision. New York: McGraw-Hill, 1975.

Palmer, S. E. Visual perception and world knowledge: Notes on a model

of sensory-cognitive interaction. In D. A. Norman, D. E. Rumelhart,

and the LNR Research Group, Explorations in cognition. San Francisco:

Freeman, 1975.

Perfetti, C. A., & Lesgold, A. M. Discourse comprehension and sources

of individual differences. In M. Just and P. Carpenter (Eds.),

Cognitive processes in comprehension. Hillsdale, N.J.: Erlbaum,

in press.

Piaget, J. The language and thought of the child. New York: Harcourt,

Brace, 1926.

Pichert, J. W., & Anderson, R. C. Taking different perspectives on a story.

Journal of Educational Psychology, 1977, in press.

Rumelhart, D. E. Notes on a schema for stories. In D. G. Bobrow and A. M.

Collins (Eds.), Representation and understanding: Studies in cognitive

science. New York: Academic Press, 1975.

Rumelhart, D. E. Toward an interactive model of reading. In S. Dornic

(Ed.), Attention and performance VI. London: Academic Press, 1976.

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, N.J.: Erlbaum, 1977, in press.

Schallert, D. L. Improving memory for prose: The relationship between

depth of processing and context. Journal of Verbal Learning and Verbal

Behavior, 1976, 15, 621-632.

 Schema-Directed Processes

28

Schvaneveldt, R. W., Meyer, D. E., & Becker, C. A. Contextual constraints

on ambiguous word recognition. Paper presented at the meeting of the

Psychonomic Society, Boston, November 1974.

Schank, R., & Abelson, R. P. Scripts, plans, and knowledge. Proceedings

of the Fourth International Joint Conference on Artificial Intelligence.

Tbilisi, Georgia: U.S.S.R., 1975.

Smiley, S. S., Oakley, D. D., Worthen, D., Campione, J. C., & Brown, A. L.

Recall of thematically relevant material by adolescent good and poor

readers as a function of written versus oral presentation (Tech. Rep.

23). Urbana, Ill.: Center for the Study of Reading, University of

Illinois, March 1977.

Spiro, R. J. Remembering information from text: Theoretical and empirical

issues concerning the 'State of Schema' reconstruction hypothesis.

In R. C. Anderson, R. J. Skiro, and W. E. Montague (Eds.), Schooling

and the acquisition of knowledge. Hillsdale, N.J.: Erlbaum, in press.

Swinney, D. A., & Hakes, D. T. Effects of prior context upon lexical access

during sentence comprehension. Journal of Verbal Learning and Verbal

Behavior, 1976, 15, 681-689.

Thieman, T. J., & Brewer, W. J. Alfred Binet on memory for ideas. Genetic

Psychology Monographs, in press.

 CENTER FOR THE STUDY OF READING

TECHNICAL REPORTS

No. 1: Halff, H. M. Graphical Evaluation of Hierarchical Clustering Schemes,

October 1975.

No. 2: Spiro, R. J. Inferential Reconstruction in Memory for Connected

Discourse, October 1975.

No. 3: Goetz, E. T. Sentences in Lists and in Connected Discourse, November 1975.

No. 4: Alessi, S. M., Anderson, T. H., & Biddle, W. B. Hardware and Software
Considerations in Computer Based Course Management, November 1975.

No. 5: Schallert, D. L. Improving Memory for Prose: The Relationship Between

Depth of Processing and Context, November 1975.

No. 6: Anderson, R. C., Goetz, E. T., Pichert, J. W., & Halff, H. M. Two
Faces of the Conceptual Peg Hypothesis, January 1976.

No. 7: Ortony, A. Names, Descriptions, and Pragmatics, February 1976.

No. 8: Mason, J. M. Questioning the Notion of Independent Processing Stages

in Reading, February 1976.

No. 9: Siegel, M. A. Teacher Behaviors and Curriculum Packages: Implications

for Research and Teacher Education, April 1976.

No. 10: Anderson, R. C., Pichert, J. W., Goetz, E. T., Schallert, D. L., Stevens,
K. V., & Trollip, S. R. Instantiation of General Terms, March 1976.

No. 11: Armbruster, B. B. Learning Principles from Prose: A Cognitive Approach

Based on Schema Theory, July 1976.

No. 12: Anderson, R. C., Reynolds, R. E., Schallert, D. L., & Goetz, E. T.
Frameworks for Comprehending Discourse, July 1976.

No. 13: Rubin, A. D., Bruce, B. C., & Brown, J. S. A Process-oriented Language
for Describing Aspects of Reading Comprehension, November 1976.

No. 14: Pichert, J. W., & Anderson, R. C. Taking Different Perspectives on

a Story, November 1976.

No. 15: Schwartz, R. M. Strategic Processes in Beginning Reading, November 1976.

No. 16: Jenkins, J. R., S Pany, D. Curriculum Biases in Reading Achievement

Tests, November 1976.

No. 17: Asher, S. R., Hymel, S., & Wigfield, A. Children's Comprehension of
High- and Low-Interest Material and a Comparison of Two Cloze
Scoring Methods, November 1976.

No. 18: Brown, A. L., Smiley, S. S., Day, J. D., Townsend, M. A. R., & Lawton,
S. C. Intrusion of a Thematic Idea in Children's Comprehension
and Retention of Stories, December 1976.
No. 19: Kleiman, G. M. The Prelinguistic Cognitive Basis of Children's Communi-
cative Intentions, February 1977.

No. 20: Kleiman, G. M. The Effect of Previous Context on Reading Individual

Words, February 1977.

No. 21: Kane, J. H., & Anderson, R. C. Depth of Processing and Interference
Effects in the Learning and Remembering of Sentences, February 1977.

No. 22: Brown, A. L., & Campione, J. C. Memory Strategies in Learning:

Training Children to Study Strategically, March 1977.

No. 23: Smiley, S. S., Oakley, D. D., Worthen, D., Campione, J. C., & Brown,
A. L. Recall of Thematically Relevant Material by Adolescent
Good and Poor Readers as a Function of Written Versus Oral Pre-
sentation, March 1977.

No. 24: Anderson, R. C., Spiro, R. J., & Anderson, M. Schemata as Scaffolding
for the Representation of Information in Connected Discourse,
March 1977.

No. 25: Pany, D., & Jenkins, J. R. Learning Word Meanings: A Comparison of
Instructional Procedures and Effects on Measures of Reading
Comprehension with Learning Disabled Students, March 1977.

No. 26: Armbruster, B. B., Stevens, R. J., & Rosenshine, B. Analyzing Content
Coverage and Emphasis: A Study of Three Curricula and Two Tests,
March 1977.

No. 27: Ortony, A., Reynolds, R. E., & Arter, J. A. Metaphor: Theoretical
and Empirical Research, March 1977.

No. 28: Ortony, A. Remembering and Understanding Jabberwocky and Small-Talk,

March 1977.

No. 29: Schallert, D. L., Kleiman, G. M., & Rubin, A. D. Analysis of Differences
Between Oral and Written Language, April 1977.

No. 30: Goetz, E. T., & Osborn, J. Procedures for Sampling Texts and Tasks
in Kindergarten through Eighth Grade, April 1977.

No. 31: Nash-Webber, B. Anaphora: A Cross-Disciplinary Survey, April 1977.

No. 32: Adams, M. J., & Collins, A. A Schema-Theoretic View of Reading

Comprehension, April 1977.

No. 33: Huggins, A. W. F. Syntactic Aspects of Reading Comprehension, April 1977.

No. 34: Bruce, B. C. Plans and Social Actions, April 1977.

No. 35: Rubin, A. D. Comprehension Processes in Oral and Written Language,

April 1977.

No. 36: Nash-Webber, B., & Reiter, R. Anaphora and Logical Form: On Formal
Meaning Representations for Natural Language, April 1977.
No. 37: Adams, M. J. Failures to Comprehend and Levels of Processing in
Reading, April 1977.

No. 38: Woods, W. A. Multiple Theory Formation in High-Level Perception,

April 1977.

No. 39: Nickerson, R. S., & Adams, M. J. Uses of Context in Speech Under-
standing and Reading, April 1977.

No. 40: Brown, J. S., & Collins, A. Model-Based Versus Text-Based Reasoning,
April 1977.

No. 41: Anderson, R. C., & Pichert, J. W. Recall of Previously Unrecallable

Information Following a Shift in Perspective, April 1977.

No. 42: Mason, J., Osborn, J., & Rosenshine, B. A Consideration of Skill
Hierarchy Approaches to the Teaching of Reading, April 1977.

No. 43: Collins, A., Brown, A. L., Morgan, J. L., & Brewer, W. F. The
Analysis of Reading Tasks and Texts, April 1977.

No. 44: McClure, E. Aspects of Code-Switching in the Discourse of Bilingual

Mexican-American Children, April 1977.

No. 45: Schwartz, R. M. Relation of Context Utilization and Orthographic

Automaticity in Word Identification, May 1977.

No. 46: Anderson, R. C., Stevens, K. C., Shifrin, Z., & Osborn, J. Instantia-
tion of Word Meanings in Children, May 1977.

No. 47: Brown, A. L. Knowing When, Where, and How to Remember: A Problem
of Metacognition, June 1977.

No. 48: Brown, A. L., & DeLoache, J. S. Skills, Plans, and Self-Regulation,
July 1977.

No. 49: Goetz, E. T. Inferences in the Comprehension of and Memory for Text,
July 1977.

No. 50: Anderson, R. C. Schema-Directed Processes in Language Comprehension,

July 1977.

No. 51: Brown, A. L. Theories of Memory and the Problems of Development:

Activity, Growth, and Knowledge, July 1977.

No. 52: Morgan, J. L. Two Types of Convention in Indirect Speech Acts,

July 1977.

No. 53: Brown, A. L., Smiley, S. S., & Lawton, S. C. The Effects of Experience
on the Selection of Suitable Retrieval Cues for Studying from
Prose Passages, July 1977.

No. 54: Fleisher, L. S., & Jenkins, J. R. Effects of Contextualized and

Decontextualized Practice Conditions on Word Recognition,
July 1977.
No. 55: Jenkins, J. R., & Larson, K. Evaluating Error Correction Procedures
for Oral Reading, August 1977.

No. 56: Anderson, T. H., Standiford, S. N., & Alessi, S. M. Computer Assisted
Problem Solving in an Introductory Statistics Course, August 1977.

No. 57: Barnitz, J. Interrelationship of Orthography and Phonological Structure

in Learning to Read, August 1977.

No. 58: Mason, J. M. The Role of Strategy in Reading in the Mentally Retarded,
September 1977.

No. 59: Mason, J. M. Reading Readiness: A Definition and Skills Hierarchy

from Preschoolers' Developing Conceptions of Print, September 1977.

No. 60: Spiro, R. J., & Esposito, J. J. Superficial Processing of Inferences

Presented in Text, October 1977.

No. 61: Spiro, R. J., & Smith, D. Distinguishing Sub-Types of Poor Comprehenders:
Overreliance on Conceptual vs. Data-Driven Processes, October 1977.

No. 62: Spiro, R. J., & Rittenhouse, R. K. Attribute Clarity in Conservation

Instructions and Horizontal D~calage in Normal Hearing and Deaf
Children, October 1977.

No. 63: Spiro, R. J., & Anderson, M. C. Theme and Subordination in Sentence
Recall, October 1977.

No. 64: Spiro, R. J., & Martin, J. E. Contextual Factors in the Recall of
Alternative Surface Structures, October 1977.

No. 65: Brewer, W. F. Memory for the Pragmatic Implications of Sentences,

October 1977.