Journal of Reading Behavior

1981, Vol. XIII, No. 1

COMPREHENSION MONITORING, MEMORY, AND

STUDY STRATEGIES OF GOOD AND POOR READERS

Scott G. Paris
University of Michigan, Educational Psychology Department, 3210 School of Education,
Ann Arbor, MI 48109

Meyer Myers II
Purdue University

Abstract. Comprehension and memory skills of fourth grade good and poor readers
were compared in two studies. Their ability to monitor comprehension of difficult
and anomalous information was measured in three ways; by spontaneous self-
corrections during oral reading, by directed underlining of incomprehensible words
and phrases, and by study behaviors. Poor readers engaged in significantly less
monitoring on all three measures and this was correlated with poorer comprehension
and recall scores. An additional metacognitive measure of perceived reading strategy
effectiveness indicated that poor readers are often unaware of the negative in-
fluences of some strategies. The patterns of responses on the multiple measures sug-
gest that poor readers may adopt decoding rather than meaning comprehension goals
during reading and they are less accurate in applying monitoring skills towards
resolving comprehension failures.

Reading comprehension involves many perceptual and cognitive skills, but a ma-
jor component is the ability to monitor one's level of understanding while reading.
This kind of mental pulse-taking is important because it is a measure of progress
towards a reading goal and a signal for comprehension failures. Checking com-
prehension thus provides a link between the reader's purposes, progress, and
behavior. In our view of reading as a means-goals problem-solving task, there are
three distinct aspects of comprehension monitoring: evaluation, planning, and
regulation. The evaluation component involves checking one's current state of
knowledge while reading and provides answers to questions such as, "Does this
make sense? Do I understand this word? Do these ideas fit with previous informa-
tion?" If the answer to any of the previous questions is negative, then the reader
must generate a plan to rectify the comprehension problem (or, alternatively, change

Reprints may be obtained from Scott Paris, Combined Program in Education and Psychology, 3210 School of
Education, University of Michigan, Ann Arbor, Michigan, 48109. This research was supported by a grant
DownloadedNIE,
from the National Institute of Education, from jlr.sagepub.com
G-77-0014.by guest on February 29, 2016
6 Journal of Reading Behavior

the original goal). The planning component involves the recruitment and selection of
corrective strategies. Once a plan is selected from possible strategies, the reader
must implement the regulatory behavior, for example, rereading, using contextual in-
foimation, looking up words in a dictionary or requesting help. Thus, successful
monitoring involves reflection on one's own comprehension state, generation of
plans for improving comprehension, and implementation of the behavior. Monitoring
should also be flexible and adaptive so that one can generate alternative plans and
employ appropriate numbers and kinds of plans to solve the task.
Monitoring meaning during reading is hardly a new concept and has been em-
phasized by many others since Dewey (1910) and Thorndike (1917). Stauffer's (1969)
view of reading as a goal-directed activity, for example, stresses the relations
between strategies and goals. A similar point was made by Locke (1975) in a "how to
study" guide:
Studying actually requires a double or split mental focus. On the
one hand, you need to be focused on the material itself (that is, on
learning it). At the same time, however, you need to be constantly
checking to see that you are actually performing those mental
operations that produce learning. In short, you need to monitor
your mental processes while studying. (Locke, 1975, p.126).
Comprehension monitoring implies some awareness of goals and strategies for
meeting them and has been considered to be an important aspect of metacognition.
However, children seem to have difficulty in monitoring their own understanding
while watching magic tricks, listening to stories, or trying to study and remember in-
formation (Brown, 1978; Markman, 1979; Paris, 1978). While there has been little
research on comprehension monitoring while reading, some evidence suggests that
young and poor readers may not evaluate and regulate their understanding accurate-
ly. For example, fourth-graders identified as poor readers were not disrupted in their
oral reading by the substitution of inappropriate words within sentences (Isakson &
Miller, 1976). Good readers attend to meaning more than decoding and have tactics
for keeping track of the sense of the information. DiVesta, Hayward, and Orlando
(1979) found that the ability to use subsequent information in text to aid comprehen-
sion improved with age and reading skill among adolescents. In two excellent
reviews, Golinkoff (1976) and Ryan (1979) have noted that a crucial distinction
between good and poor readers is their ability to provide strategies for improving
comprehension. This difference may be partly due to the adoption of different
reading goals. Smith (1975) suggested poor readers often concentrate on decoding in-
dividual words and do not try to construct the meaning of sentences. Also, young
and poor readers seem to be less aware of the existence and value of techniques for
regulating comprehension (Forrest & Waller, 1979; Myers & Paris, 1978).
Comprehension monitoring is an important aspect of reading and cognitive
development and, based on preliminary evidence, may be a skill that differentiates
beginning and accomplished readers. It is certainly not the only differentiating factor
between good and poor readers but it is an hypothesis that deserves systematic in-
vestigation because of the importance of comprehension monitoring for educational
practices and theories of reading.
The purpose of the present research was to examine the comprehension skills of
good and poor readers in more detail with several different measures of comprehen-
Downloaded from jlr.sagepub.com by guest on February 29, 2016
 Comprehension Monitoring 7

sion monitoring. We also wanted to provide measures of perceived strategy utility, a

metacognitive assessment, in a within-subjects, within-task design so that understan-
ding and use of comprehension monitoring skills could be correlated. A final purpose
was to provide comparisons between good and poor readers on traditional com-
prehension and memory tests. The multiple measures and converging evidence were
intended to yield a more complete picture of children's use of comprehension skills
and the differences between good and poor readers.

EXPERIMENT 1

The purpose of the first study was to examine differences in comprehension

monitoring between good and poor readers during oral reading. The frequency of
monitoring was compared in two situations, spontaneous and directed monitoring.
This distinction is important because poor readers may have the ability to notice and
correct comprehension failures when required to do so but they fail to monitor on
their own. This kind of "production deficiency" for cognitive strategies is especially
evident on memory tasks (Paris & Lindauer, 1977) and may be a general characteristic
of learning disabled children (Torgeson, 1977). To test for spontaneous versus
directed differences in monitoring, we used two indices. First, we asked the children
to read orally two stories that contained nonsense words and phrases and we record-
ed their spontaneous hesitations, repetitions, and self-corrections. Secondly, we
presented new stories and told children that parts of the story might not make sense
and they should underline any words and phrases that they did not understand. In ad-
dition to providing two different measures of comprehension monitoring, we felt that
the directed underlining measure might provide a better index of monitoring ability
since spontaneous self-corrections could be contaminated by factors such as
decoding goals and pronunciation errors. Indeed, construction of appropriate, quan-
titative measures of comprehension monitoring is difficult since (a) many checking
behaviors may be subtle or covert and (b) hesitations and repetitions do not always
reflect deliberate attempts at monitoring and correcting the meaning. Thus, the direc-
tions to underline incomprehensible information may be a better measure of monitor-
ing than the admittedly weak observations of spontaneous hesitations, repetitions,
and self-corrections.

Method
Subjects. Thirty-two fourth graders from rural Indiana schools were subjects.
Two groups of 16 good and 16 poor readers, with equal sex representation, were
formed on the basis of test scores derived from the achievement series of the SRA
Assessment Survey. The mean grade equivalent reading scores for the total popula-
tion was 4.1 and the good and poor reader groups were defined according to devia-
tions from this mean. The mean of the poor reader group was 2.8 [SD = .68, range =
1.0-3.4). The mean grade equivalent reading score for the good reader group was 5.4
(SD = 3.6, range = 4.8-6.2). In order to match children on nonreading school
achievement, each poor reader was matched with a good reader of the same sex on
the basis of mathematics achievement scores from the SRA Survey. Although we
tried to match each pair of subjects within .4 grade equivalent math scores, two pairs
differed by .6. The mean grade equivalent math score for the poor reader group was
Downloaded from jlr.sagepub.com by guest on February 29, 2016
8 Journal of Reading Behavior

3.7 (SD = .37, range = 3.1-4.2) and the mean score for the good reader group was
3.8 (SD = .43, range = 3.1-4.3). The mean scores of both groups were slightly
above the 3.5 population mean on the mathematics achievement test. The mean age
of the poor readers was 10.4 years (SD = .64, range = 9.7-11.5) while the mean age
of the good readers was 10.1 years (SD = 2.6, range = 9.6-10.5). Thus, the two
groups differed significantly on reading achievement but were matched for
mathematics scores, ages, and sex.
Although the groups were equated as nearly as possible on nonreading dimen-
sions in this yoked design, it is probable that they varied in other ways. Certainly, the
poor readers were so classified for a variety of reasons. We have no other data from
their records to distinguish the etiologies of reading problems and thus, there was
considerable heterogeneity within groups. The heterogeneous etiologies actually bias
the study against finding differences and thus any observed differences in monitoring
attest to the pervasiveness of the behavior.

Materials
The stimulus passages consisted of two third- and two fifth-grade level stories
(and their corresponding sets of eight comprehension questions) selected from the
Spache (1972) Diagnostic Reading Scales. Thus, children in each group read two
stories that were appropriate for their reading level. Each story was modified by
replacing two nouns with phonologically acceptable nonsense words (e.g., klids,
kales) and by rearranging the words within two clauses to produce non-meaningful
phrases. Examples of each kind are:
1. He saw a black wolf and a red fox in their kales.
2. A big clown rode around one ring on a little pony on
aleet.
3. He must be very brave to train lions and wants do to
what he tigers.
4. Bob stopped to watch the other animals the of out
the park of the way.
The four nonsense words and phrases were scattered throughout the stories with the
stipulation that none of the changed information directly affected the answers to
eight comprehension questions.

Procedure
The reading tasks were administered to individual children in a quiet room of the
school. The sessions were informal and lasted approximately 25 minutes. The tasks
were described to children as reading and memory games. Each child read aloud one
third- and one fifth-grade level story in the spontaneous monitoring condition first
and the remaining third- and fifth-grade level stories in the directed underlining con-
dition. The two monitoring conditions were always presented in a fixed order so that
that the instructions to monitor did not contaminate the spontaneous self-corrections
of children. In the spontaneous condition, children were instructed to read the stories
aloud carefully and to try to remember them because they would be asked questions
about each story later. Following these stories, children were instructed that it helps
to pay attention to what the story means and to the parts of the story that do not
Downloaded from jlr.sagepub.com by guest on February 29, 2016
 Comprehension Monitoring 9

make sense. They were provided a pencil and told to underline any of the words or
sentences in the story that they did not understand. Children were told that this
underlining might help them answer the questions about each story. Following the
presentation of all four stories and their questions, children were asked to recall as
much of each story as they could. Recall was prompted by the title of each story and
followed the same order as presentation. The orders of stories were counterbalanced
within and across conditions and each child within a matched pair received the same
story order.

Measures
There were four dependent variables of interest in the comparisons between
good and poor readers. Spontaneous comprehension monitoring was assessed
through children's repetitions, hesitations, and self-corrections during oral reading in
the spontaneous monitoring condition. The experimenter recorded each of these
events as the children read aloud. The second measure of monitoring was the number
of words and phrases underlined by children in the directed condition. The third
dependent variable was the number of comprehension questions answered correctly
by each child for each story. The fourth measure was free recall. Children's recall
was transcribed verbatim from tape recorded sessions and scored according to the
number of clauses recalled from the story. For scoring purposes, each story was par-
titioned into clause units and recall of a clause was scored correctly if a child
reported the exact words or close synonyms for the subject, verb, and noun phrase
of each clause. Two judges scored all protocols and resolved the few questionable
cases mutually. In order to compare recall across stories, the number of clauses cor-
rectly recalled were converted to percentages of each story's number of clauses.

Results
Spontaneous monitoring. The percentages of anomalous words and phrases for
which children hesitated, repeated, or self-corrected was calculated for each group
and story. If a child did more than one type of monitoring on a given word or phrase,
it was counted as a single monitoring response in order to weight subjects equally.
There were no story differences within difficulty levels and so stories were collapsed
into third- and fifth-grade level stories.
The mean percentages of monitoring responses are shown in the top of Table 1.
These data were subjected to a Group (2) x Sex (2) x Unit (2) x Difficulty (2)
analysis of variance with repeated measures on the last two factors. Anomalous
phrases were noticed and corrected significantly more often than anomalous words
by both groups of readers, F(l, 28) = 42.60, p < .01. No other main effects were
significant and the only significant interaction was Group X Unit x Difficulty, F(l,
28) = 4.71, p < 0.5. The source of this interaction was traced to the different pat-
terns of monitoring within units of different difficulty. Poor readers noticed more
anomalous words in the third grade stories than good readers but were poorer in
detecting all other anomalous information (ps < .05). Spontaneous monitoring of in-
comprehensible information was considerably inferior for poor readers on fifth
grade stories. Even though these stories were beyond the poor readers' usual
abilities, they did not actively monitor or correct nonsense words and phrases. It
should be noted, however, that good and poor readers monitored only about 42 per-
Downloaded from jlr.sagepub.com by guest on February 29, 2016
1o Journal of Reading Behavior

cent and 32 percent of the nonsense words and phrases, respectively. Either
decoding the printed word was a major goal of the children's oral reading or they
were often oblivious to the meaning of sentence construction.

TABLE 1
Mean Percent Monitoring Responses

Words Phrases

Story Level Story Level

Third Fifth Third Fifth

Spontaneous Monitoring
Poor Readers 31.3 6.2 50.0 43.8
Good Readers 12.5 28.1 68.1 62.5

Directed Underlining
Poor Readers 18.8 18.8 46.9 21.9
Good Readers 25.0 25.0 78.1 62.5

Poor readers' inaccurate monitoring was not due to a lower frequency of

monitoring responses overall, though. The incidence of monitoring acceptable words
and phrases in the stories was also calculated. Poor readers noticed or self-corrected
an average of 2.3 good words in each third grade story and 3.2 acceptable words in
fifth grade stories. Good readers monitored 1.3 and 2.5 words respectively. In a
Group (2) x Sex (2) x Difficulty (2) analysis of variance, only the difficulty factor
was significant, F(l, 28) = 4.90, p < .05. Both groups of children were monitoring
the difficult stories more frequently but they did not differ on frequency of monitor-
ing other information. Poor readers' failure to notice anomalous information was not
due to lower absolute levels of monitoring, but to less accurate comprehension
checking. This indicates that poor readers did indeed hesitate, repeat, and self-
correct while reading but they did not evaluate the anomalous information to the
same degree as good readers. They failed to monitor exactly the information that
most required comprehension checking.

Directed underlining
There were no differences between stories within each level of difficulty and
the frequencies of underlining were summed over stories. The percentages of
anomalous words and phrases underlined are shown in the bottom of Table 1. These
data were subjected to a Group (2) x Sex (2) x Unit (2) x Difficulty (2) analysis of
variance. Significant main effects were obtained for Group, F(l, 28) = 12.99, p <
.01, and Unit, F(l, 28) = 17.31, p < .01, indicating that poor readers underlined
nonsense information less often than good readers and that both groups underlined
more anomalous phrases than words. There was also a significant Group x Unit in-
Downloaded from jlr.sagepub.com by guest on February 29, 2016
 Comprehension Monitoring 11

teraction, F(l, 28) = 4.11, p < .05 due to the large difference between groups on
underlining phrases. Good readers recognized 707o of the incomprehensible phrases
while poor readers only noticed 35 To. Again, poor readers failed to check the mean-
ingfulness of phrases (particularly on difficult stories) to the extent that good readers
did.
Three other facts should be noted about these data. First, both groups did not
underline many anomalous words and the good readers were surprisingly poor at
detecting them. Second, poor readers actually underlined more legitimate words'and
phrases than good readers (28 vs. 6) so that the less accurate monitoring of poor
readers was not due to a lower absolute frequency of underlining. Third, only three
poor readers and one good reader failed to underline some anomalous information in-
dicating that the effect was not due to a few subjects.

Comprehension questions
Eight questions from the Spache stories were asked of each subject following
each story. The percentages of errors for each group, condition, and level of difficul-
ty (summed over stories again) are shown in Table 2. These data were subjected to a
Group (2) x Sex (2) x Condition (2) x Difficulty (2) analysis of variance. Significant
main effects were obtained for Group, F(l, 28) = 22.93, p < .01, and Difficulty,
F(l,28) = 113.94, p < .01, indicating that poor readers made more errors than good
readers and fifth grade stories were more difficult to understand for all children. The
significant Group x Difficulty interaction, F(l, 28) = 21.54, p < .01, indicates that
good and poor readers were both highly accurate in their answers to third grade story
questions while the poor readers made many more errors on the difficult stories. The
significant Condition x Difficulty interaction, F(l, 28) = 3.98, p < .05, indicates
that comprehension improved in the underlining condition but mostly for the fifth
grade stories.

TABLE 2
Mean Percent Errors on Comprehension Questions

Spontaneous Directed
Monitoring Underlining

Third Fifth Third Fifth

Poor Readers 13.3 55.5 10.9 46.1
Good Readers 8.6 28.9 6.3 16.4

Free recali
The percentages of clauses correctly recalled from stories were subjected to a
Group (2) x Sex (2) x Condition (2) analysis of variance. Means are displayed in
Table 3 and significant main effects were obtained for Group, F{1, 28) = 18.40, p <
.01; Difficulty, F(l, 28) = 156.01, p < .01; and Condition, F(l, 28) = 4.44, p < .05;
indicating that good readers recalled more than poor readers, a higher percentage of
Downloaded from jlr.sagepub.com by guest on February 29, 2016
12 Journal of Reading Behavior

clauses were recalled from the third-grade level stories than were recalled from fifth
grade level stories, and children recalled more following the instructions to underline
than in the no instruction condition.

TABLE 3
Mean Percent of Clauses Recalled

Spontaneous Directed
Monitoring Underlining

Third Fifth Third Fifth

Poor Readers 38.8 17.3 48.8 18.6
Good Readers 58.4 30.8 61.3 34.1

The significant Sex x Story interaction, F(l, 28) = 5.34, p < .05, indicated that
males recalled more than females on the third grade level stories. The Group x Sex
x Condition interaction F(l, 28) = 15.71, p < .01, indicates that good reader
females and the poor reader males exhibited the most improvement in the underlining
condition. Despite these interactions, good readers generally recalled more informa-
tion than poor readers and recall was significantly higher in the underlining situation.

Discussion

This study is an initial demonstration that poor readers do not engage in accurate
monitoring as frequently as good readers. Poor readers did not evaluate anomalous
information as incomprehensible to the same degree as good readers even though
they did engage in monitoring other words and phrases within the stories they read.
Poor readers also demonstrated less accurate comprehension and recall of the stories
than good readers and the deficits in accurate comprehension monitoring observed in
this study and others (DiVesta, Hayward, & Orlando, 1979; Isakson & Miller, 1976)
are clearly correlated with poor story understanding.
However, the interpretation of the results of this study must be limited. First,
poor readers did hesitate, repeat, self-correct, and underline parts of stories they did
not understand. The intent to monitor was present yet poor readers did not evaluate
the comprehensibility of stories as accurately as good readers. Part of the problem
may be that poor readers adopted decoding goals and did not attend to the mean-
ingfulness of sentences. The finding that poor readers monitored the comprehen-
sibility of words at levels similar to good readers yet failed to evaluate scrambled
clauses well supports this interpretation.
Additionally, it should be noted that our poor readers were selected to match
good readers on math achievement and age but that etiology of reading problems was
not controlled. The study does not allow an examination of comprehension monitor-
ing skills according to a particular diagnosed reading disability and some problems
may be more correlated with comprehension checking than others. In general,
Downloaded from jlr.sagepub.com by guest on February 29, 2016
 Comprehension Monitoring 13

though, keeping track of the sense of the passage seems to be troublesome for
average children with reading disabilities.
Despite the limitations, the present study clearly demonstrates that comprehen-
sion monitoring, especially when directed, is less accurate in poor readers than good
readers and is related to the typically inferior comprehension and recall of these
children. However, the reasons for inaccurate comprehension monitoring remain
unspecified. It may be that the goal of decoding and pronouncing words takes
precedence over comprehension evaluation and regulation in poor readers. This
could be due to a processing limitation or the adoption of insufficient reading goals.
Several researchers have suggested that young children and poor readers are
unaware of the value of monitoring meaning and have few behavioral means
available for regulating comprehension (Ryan, 1979).

EXPERIMENT 2

The purpose of the second study was to provide additional information about
the differences between good and poor readers' comprehension skills. Measures of
comprehension monitoring are difficult since much of the evaluation may be covert.
One possibility for externalizing this process is to assess study behaviors such as
note-taking, question-asking, and dictionary use. Thus, in Experiment 2, the study
behaviors of good and poor readers were compared with particular attention to
children's strategies for deriving meaning for difficult vocabulary words. A second
aim of this study was to provide measures of children's understanding of useful and
harmful reading strategies. The metacognitive understanding of the utility of these
strategies may be correlated with reading abilitiy and performance. Although
metacognitive differences between good and poor readers have been suggested, no
study has measured this knowledge and related it to performance within subjects. A
final concern of this study was to assess long-term memory (i.e., one week) of good
and poor readers for sequentially organized story relationships. These three
measures were intended to supplement the results of Experiment 1 and to ascertain
more specific differences in meaning-monitoring behavior.

Method
Subjects. Two groups of 14 good and poor fourth grade readers, with equal sex
representation, were formed on the basis of test scores derived from the California
Achievement Test (CAT). The good and poor reading groups were defined according
to deviations from local norms on the CAT. The percentile mean of the poor reader
group was 16.1 (SD = 8.8, range = 2 to 27). The mean of the good reader group was
78.3 [SD = 7.2, range = 62 to 88). The mean age of the poor readers was 9 years, 9
months (SD = 3.7 months, range = 9-4 to 10-4) while the mean age of the good
readers was 9 years, 9 months (SD = 3.0 months, range = 9-5 to 10-2). Although
none of the poor readers were labeled retarded or enrolled in special education, it is
likely that they were inferior to the good readers on other academic and intellectual
skills. For example, the CAT mathematics percentile mean scores for good and poor
readers were 63 percent and 21 percent respectivly. No IQ data were available.

Downloaded from jlr.sagepub.com by guest on February 29, 2016

14 Journal of Reading Behavior

Materials
Story Materials. The stimulus passage was constructed by sequentially combin-
ing two stories selected from the fourth-grade levels of the Spache (1972) Diagnostic
Reading Scales. Story cohesiveness was obtained by including two introductory
sentences and a four sentence concluding paragraph, thus creating three episodes.
The story was further modified by replacing three nouns and one verb with
synonyms that were judged to be above the average fourth grader's reading
vocabulary (e.g., anther, papaya, meandered, menageria). The final passage con-
sisted of 20 sentences that conformed with the story grammar structure of Mandler
and Johnson (1977) and Stein and Glenn (1978) and included advanced vocabulary
words that were scattered throughout the passage in contextually appropriate posi-
tions. The structure of the passage involved an overall episode of two children telling
their father about school field trips. Each subepisode concerned each child's specific
account. The six categories of story grammar information within each episode were:
setting, initiating event, internal response, attempt, consequence, and reaction.
Although each episode included these categories, the number of categories and
sentences per episode was slightly variable. For example, the superordinate episode
had two examples of initiating events and the final subepisode has three setting
statements. The entire passage is provided in Table 4 with the categories of story
grammar information. The purpose of using a structured passage was to provide a
framework for analyzing free recall similar to other studies on children's recall of
prose.

Strategies. Twenty reading strategies that could affect memory for stories were
generated according to four equal categories. The grouping consisted of 10 positive
strategies that could facilitate comprehension and remembering and 10 negative
strategies that could be detrimental. The positive and negative groups were divided
further into internal reading strategies (e.g., positive: "ask yourself questions about
the ideas in the story"; negative: "think about something else while reading") and in-
to external strategies requiring additional materials or other people (e.g., positive:
"look up words you don't know in the dictionary"; negative: "watch TV while you
read"). Five neutral questions reflecting information irrelevant to memory and com-
prehension were also included (e.g., "does it help you remember the story if it's
typed in blue instead of black ink"). The complete list of strategies is shown in Table
5.

Hating Scale. A visual aid was constructed in order to facilitate children's

strategy evaluations and ratings. A graph resembling a histogram was drawn on an
814' x 11" sheet of paper. A .75 x .50 inch box with "No differences" written in it
was drawn in the center of the page. A horizontal axis was drawn on the paper. To
the right, four boxes successively increasing in height by .75 inches were drawn
above the axis and, to the left, four boxes inversely repeating the right side sequence
were drawn below the axis. The first and last boxes to the right of the neutral box
were respectively labeled "helps a little" and "helps a lot", while the first and last
boxes to the left were labeled "hurts a little" and "hurts a lot". This scale was used
by children to rate each of the reading strategy questions.

Downloaded from jlr.sagepub.com by guest on February 29, 2016

 Comprehension Monitoring 15

TABLE 4

Stimulus Passage and Story Grammar Episodes and Categories

Episode Unit Phrase/Sentence

1 S Bob and Mary were sitting at the breakfast table one morning talking to
their father.
1 IE They were telling him about the field trips they had taken at school.
3 S Mary's teacher took the class on a nature walk last week.
2 IR Mary remembered that it was a warm, bright Monday for the trip.
2 IE Every time the class came to a new plant in the forest, the teacher ex-
plained its parts.
2 IR Mary was amazed by all the different parts of a flower.
2 A The teacher pointed out the flower's stems, petals, and leaves.
2 C She learned how bees make honey from flowers and how insects eat
the anthers.
2 R Mary was so excited to learn about plants that she picked a flower to
show to her mother.
3 S Then Bob told his father that his class took a field trip to the city
market.
3 S It did not look like a grocery store.
3 S There were many big boxes of fruits and vegetables.
3 IE As Bob walked through the market, he saw huge crates of apples,
oranges, and all sorts of fruits, even a box of papayas.
3 IR He was hungry and wanted just one piece of fruit to eat.
3 A When Bob asked a man to sell him just one apple,
3 C the man laughed and gave him an extra large one but wouldn't take any
money.
3 R Bob meandered through the market eating his juicy apple and looking at
all the food.
1 IE As Bob and Mary told their stories,
1 IR their father smiled because he was happy that they had so much fun.
1 IR But he wanted to go somewhere too.
1 A He jumped up and said, "Let's all go to the zoo today."
1 R Bob and Mary rushed upstairs to tell their mother.
1 C They packed a picnic basket and went to the menagerie.
Note: S = Setting, IE = Initiating Event, IR = Internal Response, A = Attempt, R = Reaction,
C = Consequence.

Downloaded from jlr.sagepub.com by guest on February 29, 2016

16 Journal of Reading Behavior

TABLE 5

Reading Strategies by Type of Influence

Does it help to remember the story if you...

External positive
underline important parts
take notes summarizing the story
write it down in your own words
ask questions about the parts you don't understand
look up words you don't know in the dictionary

External negative
look up all of the story words in the dictionary
someone talks to you while you study
watch TV while you read
write down all the words in the story
cover up all the words in the story with you hand except the one word you're reading

Internal positive
check through the story to see if you remember all of it
say the main ideas over and over
imagine the story like a movie in your mind
reread it several times
ask yourself questions about the story (ideas in the story)

Internal negative
read the story backwards
read the story as fast as you can
skip the parts you don't understand
say every word over and over
think about something else while reading

Neutral questions
it's typed in blue instead of black ink
you are taller than other kids in your class
you have red hair
you can jump higher than other kids in your class
the story is in a book or on a separate sheet of paper

Procedure
The tasks were administered to individual children in a quiet room in the school.
The initial and follow-up sessions were informal and lasted approximately 25
minutes and 10 minutes respectively and were described to the children as reading
and memory games. A dictionary, pencil, and blank sheet of paper were positioned
Downloaded from jlr.sagepub.com by guest on February 29, 2016
 Comprehension Monitoring 17

in front of each child. Before the story was presented, children were instructed to
read and study the story so they could remember it later and told that they could
study in any way that they liked. The experimenter suggested that they could use the
pencil to write on the blank paper or on the story sheet, use the dictionary, or ask
questions. Children were then presented the passage to read, told to write their name
on it, and instructed to signal the experimenter when they knew the story well
enough to remember it. During the study period, the experimenter recorded the
number and type of observed strategies. Following the study period children were
given a two-minute interpolated task consisting of addition problems and subse-
quently asked to verbally recall the story. Following recall, children were asked to
report on their activities during the study period and were asked to define the four
advanced vocabulary words. Free recall, reported study activities, and definitions
were tape recorded.
The second phase of the initial session involved subject ratings of the utility of
strategies affecting reading. The nine point graphic rating scale indicating the degree
to which a factor could facilitate or hinder memory was described to the children.
Subjects were verbally presented with 25 strategies in random order and asked to
point to the location on the rating scale that best reflected the utility of that factor.
The experimenter numerically coded and recorded each rating response.
The follow-up session involved a second recall of the story one week later.
Children were asked to recall the story presented the previous week and given a cue
that the story was about "field trips". Their verbal recall was tape recorded.

Results
This study yielded information on three types of reading skills; study behavior,
strategy evaluations, and recall. Each type will be discussed separately.

Study Behavior. The average study time for all children was approximately four
minutes with no difference between the good and poor reader groups. However, the
use of comprehension strategies was more frequent for good readers. Eleven good
readers (79%) and four poor readers (29%) exhibited a study strategy for resolving
unknown words. Five of the good readers and only two poor readers used the pencils
and paper to write notes or summaries. Six of the good readers looked up words in
the dictionary while none of the poor readers did. The poor readers appeared more
concerned with pronunciation than meaning of words. Of the six children who asked
the experimenter about unknown words (four poor and two good readers), only the
good readers asked for the meanings and definitions.
Children were also asked to define the four advanced vocabulary words and to
report whether they had known the words before the session. Three poor and six
good readers knew "papaya" and one good reader knew the meaning of
"meandered". Excluding the words previously known, eight good readers but only
one poor reader learned the definitions of at least one word during the task. The
mean number of words learned by poor readers was .07 and .93 by good readers,
t(26) = 2.90, p < .01. The eight good readers all reported using strategies of asking
questions or referring to the dictionary to determine word meanings.
Thus, good readers exhibited more spontaneous strategies for monitoring and
resolving word meanings than poor readers. The few poor readers who noticed the
Downloaded from jlr.sagepub.com by guest on February 29, 2016
18 Journal of Reading Behavior

unfamiliar words only wanted help in pronouncing them. The use of comprehension
strategies was related to the learning of new words by good readers. The mean
number of overall strategy use (notes, questions, and dictionary usage) in the poor
reader group was only .5 while good readers employed an average of 1.4 study
techniques, t(26) = 2.06, p < .05.

Strategy Evaluations. All children were asked to rate the usefulness of 20

reading strategies on a nine point scale. The mean ratings are shown in Table 6. Poor
readers generally gave higher rankings to the strategies, particularly the negative in-
stances. However, differences in response bias seem unlikely since both good and
poor readers rated the neutral factors similarly, 4.96 and 4.97 respectively. The rating
scores were analyzed in Group (2) x Positive-Negative (2) x External-Internal (2)
ANOVA with repeated measures on the last two factors. The analysis revealed a
main effect of reading group, F(l, 26) = 11.7, p < .01, due to the higher ratings
given by poor readers. The analysis also revealed that all children could discriminate
positive from negative factors since positive factors were rated significantly higher,
(1, 78) = 51.3, p < .01. The main effect of locus was also significant, F(l, 78) =
8.80, p < .01, due to the higher ratings given to external factors. There was also a
significant interaction between Positive-Negative and External-Internal factors, F{1,
78) = 31.4, p < .01. As seen in Table 6, positive internal strategies were rated slight-
ly higher than positive external strategies while negative internal strategies were
rated much lower than negative external strategies. A separate ANOVA on the
ratings of negative strategies revealed significant main effects of group, F(l, 26) =
10.24, p < .01, and locus, F(l, 26) = 16.08, p < .01. Thus, poor readers were less
aware of the detrimental influences on comprehension of negative factors than good
readers while their ratings of positive and neutral factors were equal.

TABLE 6
Mean Ratings of 20 Reading Strategies*

Positive Strategies Negative Strategies

External Internal External Internal

7.2 8.1 4.8 3.5

Poor Readers
(1.8)° (1.2) (2.7) (2.8)

7.2 7.3 3.9 2.5

Good Readers
(1.5) (1.5) (2.4) (2.0)

"Scale ranged from 1-9 with higher numbers indicating a helpful strategy and 5 a neutral mid-
point
"Standard deviations in parentheses
Downloaded from jlr.sagepub.com by guest on February 29, 2016
 Comprehension Monitoring 19

Poor readers also showed more reversals of ratings, that is rating negative
strategies as positive and vice versa. The mean number of strategies receiving revers-
ed rankings was 3.1 for poor readers and 1.5 for good readers, t(26) = 3.18, p < .05.
For example, one internal negative strategy, "saying every word over and over",
was rated as very helpful by poor readers (7.4) and neutral by good readers (4.9)
which may reflect poor readers' concentration on word by word decoding strategies.
In general, poor readers were less aware than good readers of the negative strategies
and variables that influence story comprehension and memory.

Recall. Children's free recall from immediate and delayed tests was transcribed
from tape recordings and scored according to the story grammar categories. The per-
cent agreement of two judges independently scoring the data was 92 percent. The
recall scores were converted to percentages of story grammar units to control for the
variable numbers of propositions and are shown in Table 7. These data were analyz-
ed in a Group (2) x Test (2) x Grammar Unit (6) ANOVA with repeated measures on
the last two factors. The main effect of grammar unit was significant, F(5, 26) =
19.8, p < .01 and subsequent Newman-Keuls tests revealed that reactions and inter-
nal responses were recalled significantly less often than all other units. The percen-
tage recalled in the delayed condition was significantly less than the immediate test,
F(l, 26) = 8.46, p < .01, as might be expected. The main effect of the reading group
did not reach significance, F(l, 26) = 3.3, p < .08. However, an analysis of in-
dividual protocols revealed that poor readers forgot more information on the delayed
test (that they had successfully recalled on the immediate test) than good readers,
367o and 19% respectively. This difference was significant, t(26) = 2.02, p < .05.

TABLE 7

Percent Recall by Category, Recall Test, and Reading Ability

Immediate Initiating Internal Conse-

Recall Setting Event Attempt Response quence Reaction Total

Poor
Readers 49 46 48 24 45 38 41.7
Good
Readers 51 61 67 30 64 41 52.3

Delayed
Recall

Poor
Readers 41 45 36 13 43 26 34.0
Good
Readers 53 64 60 26 50 26 46.5

Downloaded from jlr.sagepub.com by guest on February 29, 2016

20 Journal of Reading Behavior

Another way to analyze recall performance is to examine correct recall of the se-
quential information within the three episodes of the passage. We grouped the setting
and initiating event categories into a group called "Beginnings", attempts and inter-
nal responses into "Actions", and consequences and reactions into "Outcomes" to
analyze the frequency with which children recalled information from each episode
according to the correct sequence of beginning-action-outcome. A sequence was
defined as recall of at least one unit from each group in proper order. The frequen-
cies of sequential recall were calculated for each episode, each recall test, and each
reading group and are shown in Table 8. A series of Cochran's Q tests on the
dichotomous frequency data revealed that episode 3 was recalled significantly better
than episode 2 in immediate recall and significantly better than both episodes 1 and 2
in delayed recall, ps < .05. The patterns of recall across episodes did not vary ac-
cording to group or test, though. For this reason the data were summed over episodes
for a Group (2) x Recall Test (2) ANOVA with repeated measures on the last factor.
This analysis of the frequencies of sequential recall of units indicated a significant ef-
fect of reading group, F(l, 26) = 5.60, p < .05, and a marginal effect for tests, F(l,
26) = 3.70, p < .06. Although the difference between good and poor readers for
percent recalled was not quite significant statistically, the organized, sequential
recall within episodes was significantly better for good readers.

TABLE 8
Frequencies of Sequence Recall by Episodes, Group, and Recall Test

Episodes
Immediate Recall 1 2 3

Poor readers 4 4 8
Good readers 10 4 11

Delayed Recall

Poor readers 3 2 6
Good readers 7 3 11

Correlations between reading strategies and recall

Following immediate recall, children were asked to report any strategies that
they had used to study the story. Responses included rereading, note-taking, im-
agery, concentrating, asking questions, and using the dictionary. As noted earlier,
good readers used and reported these strategies more often than poor readers. To
determine the relationship between study behavior and recall, a conditional pro-
bability analysis was conducted. The probability of recalling more than the mean
number of grammar units from the story given the reported use of a strategy was .71.
Clearly, strategies were associated with better than average recall and were more
frequent for good readers.
Downloaded from jlr.sagepub.com by guest on February 29, 2016
 Comprehension Monitoring 21

The relationship between recall and strategy ratings was assessed through multi-
ple correlations. A Pearson correlation was calculated between percent recall and
each of the 20 rated strategies, the number of reversed ratings, and reported strategy
use. The patterns for immediate and delayed tests were similar and revealed that
recall was negatively correlated with ratings for rereading the story, asking yourself
questions, looking up all the story words in the dictionary, reading the story fast, and
saying every word over and over. These are precisely the items that poor readers
rated highly. Also, the number of reversed ratings was negatively correlated with
recall while reported strategy use was positively correlated. All these correlations
were significant at the .05 level.

Discussion

In this experiment, poor readers engaged in few spontaneous study behaviors

and failed to ask questions, take notes, or use a dictionary as often as good readers.
Consequently, the poor readers did not notice and resolve the comprehension
failures of the unknown words. Poor readers were also less aware of the debilitating
effects on comprehension of negative strategies. They remembered somewhat less
of the stories than good readers and showed significant forgetting and less organized
recall. The lack of strategies, awareness, and organized memory were general
characteristics of the poor readers.

SUMMARY

Poor readers in these studies exhibited inferior comprehension and memory for
stories that they read orally or silently compared to good readers. This is the ex-
pected result and in correlation with the SRA and CAT criterion instruments used for
group identification. The present studies reveal additionally that the differences in
reading performance are correlated significantly with the use of accurate com-
prehension monitoring skills. One of the strengths of the research is the use of multi-
ple, observable measures of comprehension monitoring. These techniques may
prove fruitful for future researchers as well as classroom teachers because they pro-
vide ready indications of children's intent to monitor meaning, accuracy in checking,
and means available for improving comprehension. The use of interviews and
metacognitive measures also provide valuable information about children's percep-
tions of the reading process and can be added easily to research and classroom prac-
tices. A more complete picture of children's reading problems is provided by multi-
ple measures of reading goals, strategies, and performance.
In an earlier study, we observed that eight year olds do not understand the
variables that influence reading as well as twelve year olds and they have little
knowledge about strategies for achieving or resolving comprehension (Myers &
Paris, 1978). The failure of young and poor readers to understand the special
strategies required for reading has been noted by others (Gibson, 1974; Golinkoff,
1976) and has implications for teaching. The production deficiency for comprehen-
sion strategies may be eliminated in part by explicit instruction regarding strategies.
Such instruction must emphasize the child's awareness of the goal of meaning con-
struction during reading and the functional value of specific means for achieving
Downloaded from jlr.sagepub.com by guest on February 29, 2016
22 Journal of Reading Behavior

comprehension. Further research is needed to specify how good and poor readers
differ in their metacognitive understanding of reading skills and how training can
promote the utilization of monitoring strategies.

REFERENCE NOTE

1. FORREST, D. L., & WALLER, T. G. Cognitive and metacognitive aspects of reading. Paper
presented at the biennial meeting of the Society for Research in Child Development, San
Francisco, March 1979.
REFERENCES

BROWN, A. L. Metacognitive development and reading. In R. J. Spiro, B. C. Bruce, and G. W.

F. Brewer (Eds.), Theoretical issues in reading comprehension. Hillsdale, NJ: Lawrence
Erlbaum Associates, 1978.
BROWN, A. L. The development of memory: Knowing, knowing about knowing, and knowing
how to know. In H. W. Reese (Ed.), Advances in child development and behavior. New
York: Academic Press, 1975.
DEWEY, J. How we think. Boston: Heath, 1910.
DIVESTA, F. J., HAYWARD, K. G., & ORLANDO, V. P. Developmental trends in monitoring
text for comprehension. Child Development, 1979, 50, 97-105.
GIBSON, E. J. Trends in perceptual development: Implications for the reading process. In A.
Pick (Ed.), Minnesota symposia on child psychology, 1974, 8. 24-54.
GOLINKOFF, R. A comparison of reading comprehension processes in good and poor
comprehenders. Reading Research Quarterly, 1976, 11, 623-659.
ISAKSON, R. L., & MILLER, J. W. Sensitivity to syntactic and semantic cues in good and poor
comprehenders. Journal of Educational Psychology, 1976, 68, 787-792.
LOCKE, E. Q. A guide to effective study. New York: Springer, 1975.
MARKMAN, E. M. Realizing that you don't understand: Elementary school children's
awareness of inconsistencies. Child Development, 1979, 50, 643-655.
MYERS, M., & PARIS, S. G. Children's metacognitive knowledge about reading. Journal of
Educational Psychology, 1978, 70, 680-690.
PARIS, S. G. Coordination of means and goals in the development of mnemonic skills. In P.
Orstein (Ed.), Memory development in children. Hillsdale, NJ: Lawrence Erlbaum
Associates, 1978.
PARIS, S. G., & LINDAUER, B. B. Constructive aspects of children's comprehension and
memory. In R. V. Kail and J. W. Hagen (Eds.), Perspectives on the development of
memory and cognition. Hillsdale, NJ: Lawrence Erlbaum Associates, 1977.
RYAN, E. B. Identifying and remediating failures in reading comprehension: Toward an
instructional approach for poor comprehension: Toward an instructional approach for
poor comprehenders. In T. G. Waller and G. E. MacKinnon (Eds.), Advances in reading
research. New York: Academic Press, in press.
SMITH, F. Comprehension and learning. New York: Holt, Rinehart, & Winston, 1975.
STAUFFER, R. G. Directed reading maturity as a cognitive process. New York: Harper & Row,
1969.
THORNDIKE, E. L. Reading as reasoning: A study of mistakes in paragraph reading. Journal of
Educational Psychology, 1917, 8, 323-332.
TORGESEN, J. K. Memorization processes in reading-disabled children. Journal of Educational
Psychology, 1977, 69, 571-578.

Downloaded from jlr.sagepub.com by guest on February 29, 2016