Downloaded By: [University of Minnesota] At: 19:04 28 May 2008

Reading Psychology, 29:137164, 2008

C Taylor & Francis Group, LLC
Copyright
ISSN: 0270-2711 print / 1521-0685 online
DOI: 10.1080/02702710801963951

DIFFERENTIAL COMPETENCIES CONTRIBUTING

TO CHILDRENS COMPREHENSION
OF NARRATIVE AND EXPOSITORY TEXTS
RACHEL M. BEST, RANDY G. FLOYD, and DANIELLE S. MCNAMARA
The University of Memphis, Memphis, Tennessee, USA
This study examined the influences of reading decoding skills and world knowledge on third graders comprehension of narrative and expository texts. Children
read a narrative text and an expository text. Comprehension of each text was assessed with a free recall prompt, three cued recall prompts, and 12 multiple-choice
questions. Tests from the WoodcockJohnson III Tests of Achievement (Woodcock,
McGrew, & Mather, 2001) were used to assess reading decoding skills and world
knowledge. Comprehension was better for the narrative text than the expository
text, and the effects of reader competencies depended on text genre. Comprehension
of the narrative text was most influenced by reading decoding skills. In contrast,
expository text comprehension was most influenced by world knowledge.

Introduction
Children use texts from the onset of their elementary schooling.
Therefore, understanding texts and learning from them are of
paramount importance to academic success. A critical period in
the development of such reading skills appears to occur during the
third, fourth, and fifth grades. In fact, reading difficulties emerging during this period have been referred to as the fourth-grade
slump (Meichenbaum & Biemiller, 1998; Sweet & Snow, 2003). Our
study explored third graders reading comprehension, competencies enabling it, and the influence of text genre to understand
better this critical period and to inform effective interventions for
third graders who struggle with reading.

Address correspondence to Danielle S. McNamara, The University of Memphis,

202 Psychology Building, Memphis, TN 38152-3230. E-mail: d.mcnamara@mail.psyc.
memphis.edu

137

138

R. M. Best et al.

Downloaded By: [University of Minnesota] At: 19:04 28 May 2008

Text Genre
It is likely that changes in reading requirements contribute to the
fourth-grade slump. Prior to the third grade, most reading activities focus on decoding accuracy, decoding automaticity, and oral
reading fluency, whereas afterward, children are increasingly required to read to understand. Changes in reading activities result
in children being exposed to increasingly difficult texts during the
third grade. During this period, children move beyond narrative
texts to expository texts, such as science or social studies texts,
to learn about subject domains (Bowen, 1999; Snow, 2002). We
know that elementary-school children encounter greater difficulty
comprehending expository text than narrative text (Olson, 1985;
Spiro & Taylor, 1980; Tun, 1989). However, we have a fairly limited
understanding of the competencies that support third graders
comprehension of expository texts in comparison to narrative
texts (e.g., Cote, Goldman, & Saul, 1998; Williams, Hall, & Lauer,
2004). Therefore, we compared competencies involved in third
graders comprehension of expository and narrative texts with a
view toward pinpointing the difficulty associated with expository
text comprehension.
Competencies Enabling Reading Comprehension
Reading comprehension can be defined as the ability to obtain
meaning from written text for some purpose (Vellutino, 2003,
p. 51). To comprehend successfully, readers must identify a series
of letters as a word, access the meaning of the word from their lexicon or mental dictionary, and integrate individual word meanings
into a coherent sentence-level representation. As such, reading
comprehension requires the efficient coordination and integration of a number of underlying processes, which include reading
decoding skills and the utilization of world knowledge (Kintsch,
1988; 1998; Perfetti, 1985; Snow, 2002; Vellutino, Scanlon, &
Tanzman, 1994).
READING DECODING SKILLS
Reading decoding represents the ability to apply lettersound
correspondence rules when reading words and non-words. Many
children in the third and fourth grades may still exhibit slow

Downloaded By: [University of Minnesota] At: 19:04 28 May 2008

Differential Competencies

139

or inaccurate decoding skills (Brand-Gruwel, Aarnoutse, & Van

den Bos, 1988; Lyon, 2002; Mommers, 1987; Perfetti, 1985; Roth,
Speece, Cooper, & De La Paz, 1996; Taschow, 1969; Vellutino,
2003). Many theorists assume that slow or inaccurate decoding
skills appear to consume working memory resources that might
otherwise be used for comprehension processes, such as inference
generation (Cain, Oakhill, & Bryant, 2004; Curtis, 1980; Hannon
& Daneman, 2001; Perfetti, 1985). There is no doubt that decoding skills form a foundation for reading comprehension, but even
if children have acquired accurate and automatic decoding skills,
they may still encounter comprehension difficulties (Lyon, 2002;
Sweet & Snow, 2003).
WORLD KNOWLEDGE
Several models of reading comprehension convey that the
readers ability to decode words accurately and automatically allows
the reader to integrate individual word meanings and sentence
meanings into a coherent representation. For example, Kintschs
(1988, 1998) Construction-Integration (CI) model distinguishes
between the surface code, the propositional textbase, and the situation model. The surface code preserves the exact wording and syntax
of clauses and rapidly fades from memory. The textbase contains
explicit propositions that preserve the meaning but not the exact
wording and syntax. The situation model goes beyond the explicit
text content. It represents the global meaning of the text that is formulated from the textbase and readers world knowledge. For narrative text, the situation model refers to the readers understanding of the characters, settings, actions, and events in the readers
mental representation of the story. For expository texts, the situation model refers to the integration of the textbase with the
readers knowledge about the texts subject matter. For both types
of text, the situation model is constructed inferentially through
interactions between the explicit text, world knowledge, and the
readers comprehension goals (Graesser, Singer, & Trabasso, 1994;
Johnson-Laird, 1983; Kintsch, 1998).
Although narrative texts may contain unfamiliar information,
most children have, from first-hand experience, well-developed
schemas about the settings, actions, and events described by them
(Cote et al., 1998; Nelson, 1996). Accordingly, most children
possess the knowledge necessary to understand narrative texts.

Downloaded By: [University of Minnesota] At: 19:04 28 May 2008

140

R. M. Best et al.

Many narrative texts also follow a simple structurea sequence of

causally related events. As a result, typical first and second graders
understand narrative text structure (e.g., Williams et al., 2004).
Thus, working memory requirements are reduced and comprehension is facilitated (Cain, 1996; Cote et al., 1998).
In contrast to narrative texts, expository texts tend to place
increased processing demands on the reader due to their greater
structural complexity, greater information density, and greater
knowledge demands. In terms of structure, expository texts often contain abstract and logical relations, which can be difficult
to interpret (Stein & Trabasso, 1981). Children between the third
and fifth grades have relatively little knowledge of expository text
structure (Kamberelis & Bovino, 1999; Langer, 1986), which may
impede their ability to organize and process text content. Perhaps
more importantly, because expository texts often introduce many
new concepts and ideas (e.g., osmosis, gravity), childrens existing
knowledge is vital in order to integrate and assimilate the new information from the text. If children lack related knowledge about
the concepts and ideas described by the text, their comprehension will be limited because they cannot generate accurate inferences (Beck & McKeown, 1992; Langer, 1986). This link between
knowledge and comprehension of expository texts is well supported by research with adults as well as children in middle school
and high school (e.g., Afflerbach, 1986; Chi, Feltovich, & Glaser,
1981; Chiesi, Spilich, & Voss, 1979; Lundeberg, 1987; McNamara,
Kintsch, Songer, & Kintsch, 1996; McNamara & Kintsch, 1996;
Means & Voss, 1985; Reutzel & Morgan, 1990). In fact, a few studies
with children or adults indicate that the influence of knowledge on
comprehension of expository texts is greater than that of reading
skills (e.g., Chiesi et al., 1979; OReilly & McNamara, 2002; Rupley
& Willson, 1996).
Present Research
The ability to apply lettersound correspondence rules when reading words is the foundation of reading for understanding, but this
ability is not sufficient to explain the complex process of reading comprehension across text genres. For example, it is probable that the fourth grade slump may, at least in part, be related to
deficits in knowledge enabling comprehension of expository texts.

Downloaded By: [University of Minnesota] At: 19:04 28 May 2008

Differential Competencies

141

To investigate this issue, we examined the effects of text genre,

decoding skills, and world knowledge on third graders text comprehension. The texts employed in this study reflect those typically
found in basal readers and science textbooks designed for children
in the third and fourth grades. Based on a large body of research,
we expected that childrens comprehension of the narrative text
would be superior to the expository text. Because reading decoding is the foundation of reading comprehension, we expected that
comprehension of both the narrative text and the expository text
would be highly related to decoding skills. Based on previous research and theory, we expected that the strength of the relation
between comprehension of the expository text and world knowledge would be similar in magnitude to the relation between comprehension of the expository text and decoding skills. In contrast,
we expected the strength of the relation between comprehension
of the narrative text and world knowledge would be notably lower
in magnitude than the relation between comprehension of the
narrative text and decoding skills.
Method
Participants
The participants were 61 children enrolled in the third grade at
two public schools in a large metropolitan school district. Children
ranged in age from 8 years, 4 months, to 10 years, 7 months (M =
109.7 months, SD = 6.0 months). Girls formed 52% of the sample
(n = 32) and boys 48% (n = 29). Approximately 57% of the children were Black or African American (n = 35), 28% were White
(n = 17), 7% were biracial (n = 4), and 3% were AsianPacific
Islanders (n = 2). Using parent education level as an index of socioeconomic status, 5% of mothers and 5% of fathers did not complete high school; 61% of mothers and 52% of fathers graduated
from high school, completed some college, or completed technical school; and 33% of mothers and 32% of fathers obtained at
least a college degree. Parents reported living in 20 different ZIP
code areas. According to the 1999 United States Census, the average median income level across these ZIP code areas was $35,874
(SD = $13,426, range = $19,998 to $65,874).

Downloaded By: [University of Minnesota] At: 19:04 28 May 2008

142

R. M. Best et al.

On two screening measures, children in this sample demonstrated reading comprehension skills and vocabulary knowledge
that were average for their age. Across children, values from the
WoodcockJohnson III (WJ III) Tests of Achievement (ACH) Passage Comprehension test (Woodcock et al., 2001) were M = 99.7
and SD = 9.2, and values from the WJ III ACH Picture Vocabulary
test were M = 102.3 and SD = 8.5. These results support the assertion that the participating children displayed an age-appropriate
range of reading abilities and competencies.
Materials
TEXTS
The texts were drawn from a pool of 60 narrative texts and
67 expository texts obtained from basal readers and science textbooks. For these 127 texts, conventional measures of text dimensions associated with comprehension, including number of words,
number of sentences, FleschKincaid grade level, and Flesch Reading ease, were derived from Coh-Metrix, Version 1.4 (Graesser,
McNamara, & Louwerse, & Cai, 2004). Preliminary selection criteria for inclusion were FleschKincaid grade level (to be between 2.97 and 4.75) and word length (within the range of 304
to 471 words). From the text pool, we selected the narrative text
Orlando, which was obtained from Addison-Wesleys Phonics TakeHome Reader, Grade 2 (1998). We selected the expository text Needs
of Plants, which was obtained from McGrawHills Science, Grade
2 (2000). These texts were representative of the text pool and
closely equated in terms of conventional measures of text dimensions (see Table 1). In reviewing Table 1, readers should note that
it is not feasible to match texts from different genres on all text
dimensions.
TABLE 1 Characteristics of the Narrative and Expository Texts
Measure
Number of words
Number of sentences
FleschKincaid grade level
Flesch Reading ease

Narrative Text
(Orlando)

Expository Text
(Needs of Plants)

445
61
3.6
85.81

462
45
4.4
90.40

Downloaded By: [University of Minnesota] At: 19:04 28 May 2008

Differential Competencies

143

COMPREHENSION MEASURES
Comprehension was assessed using a combination of recall
tasks and multiple-choice questions. Multiple measures have the
benefit of providing more thorough evaluation of the breadth and
depth of comprehension.
We collected free recall and cued recall data to assess childrens comprehension of the texts. The free recall task required
children to report what they remembered about the passage they
had just read. Each child was provided the following directive: Tell
me everything you can remember about what you have just read.
Give me as many details as possible, like you were trying to tell
a friend about you just read. All responses were recorded on an
audiotape and later transcribed.
Cued recall, which assessed major themes in the text, was
used to evoke richer content from the children (Zinar, 1990).
The cued recall task required children to respond to three directives. The directives were designed to assess comprehension
of three major sections of the texts, and they essentially covered
the entire text. For the Orlando text, children were directed to
(a) Tell me everything that Salvador did after his mama told
him they would have to sell Orlando, (b) Tell me everything
that Salvadors mama said at first about what they needed to do
with Orlando, and (c) Tell me everything that happened to
Salvador and what he did after the storm began. For the plant
text, children were directed to (a) Tell me everything about
the parts of a plant and how those parts give the plant what it
needs, (b) Tell me everything that makes leaves important to
plants and how those parts give the plant what it needs, and
(c) Tell me everything that plants do in response to changes
in their environment. All responses were audiotaped and later
transcribed.
Twelve multiple-choice questions for each text measured
textbase and situation model levels of understanding (Givon, 1993;
Kintsch, 1998; Louwerse, 2002; McNamara et al., 1996). Six questions were classified as text-based, and six were classified as bridging inference questions (see Appendix). For the narrative text,
the odd-numbered questions were text-based questions, and the
even-numbered questions were bridging inference questions. The
order was reversed for the expository text. Each question and three
possible answers were presented orally and visually, and children

Downloaded By: [University of Minnesota] At: 19:04 28 May 2008

144

R. M. Best et al.

were required to vocalize the correct answer. Correct scores were

totaled for each text.
Two questions of each type from the narrative text were answered correctly by more than 90% of children. In a conservative
attempt to eliminate question level ceiling effects and to facilitate
a normally distributed total score for the narrative text, these two
questions were omitted (see Appendix B). Therefore, 10 questions
were used to create the total score. From the expository text, the response patterns to all 12 questions did not indicate ceiling effects.
For each child and each text, the proportion of correct answers to
the total number of acceptable questions was calculated.
READING COMPETENCY MEASURES
As part of a larger battery of assessment instruments, children
completed two tests from the WJ III ACH (Woodcock et al., 2001).
The WJ III ACH LetterWord Identification test measures letter
and word recognition skills. Examinees must identify printed letters and words. The test has a median internal consistency reliability coefficient of .94 for ages 8 to 10 (McGrew & Woodcock, 2001).
The WJ III ACH Academic Knowledge test measures knowledge
about the biological, physical, and social sciences and the humanities. Examinees must provide information about the biological
and physical sciences; about history, geography, government, and
economics; and about art, music, and literature. The test has a
median internal consistency reliability coefficient of .84 for ages
8 to 10 (McGrew & Woodcock, 2001). For both WJ III ACH tests,
age-based standard scores (M = 100, SD = 15) were obtained. For
LetterWord Identification, the standard score represents decoding
skills. The standard score for Academic Knowledge represent world
knowledge.
Procedures
RECRUITMENT
Children were recruited by sending letters of invitation to
parents through the childrens school classrooms. The letters provided information about the study and requested that parents contact researchers to schedule a testing session. Testing sessions were
conducted on four Saturdays during February, March, and May
2003. After completion of the testing, children were provided a

Downloaded By: [University of Minnesota] At: 19:04 28 May 2008

Differential Competencies

145

gift card to a department store, coupons from merchants, and

school supplies.
TESTING
An assessment battery was completed in approximately a onehour testing session. Graduate students, who had (a) successfully
completed a semester-long graduate course covering the administration of standardized tests and (b) completed additional training with all assessment tasks, completed all testing. Children first
silently read each text within a 5-minute period. Half the children
read the narrative text first, and half read the expositive text first.
After reading the first text, the text was removed from view. Children then answered the respective free recall task, the cued recall
task, and the 12 multiple-choice questions. This process was repeated with the remaining text. The children then completed the
battery of reading competency tasks. All protocols were checked
for accuracy by the second author.
CODING RECALL DATA
The analysis focused on the amount of information children
recalled that was explicitly stated in the text, and it included two
steps. First, the narrative and expository texts were propositionalized using a conventional method in which the information contained in each sentence was broken into main propositions and
subpropositions (see Kintsch, 1998). Main propositions consisted
of the main idea, whereas the subpropositions contained details
pertaining to the main idea. For instance, the main proposition for
the sentence Plants need sunlight, air and water to live consisted
of the notion that plants need things to live. The subproposition
consisted of the notion that plants need sunlight and water. There
were 61 main propositions and 43 subpropositions in the narrative
text and 45 main propositions and 47 subpropositions in the expository text. Second, the childrens transcribed recall data were
broken into idea units. Idea units were classified as utterances that
contained a subject, verb, and direct object. Idea units were separated by connectives, such as so, and, but, andbecause. Every idea unit
was matched to the propositions. In cases where children repeated
information, each idea unit was counted only once.
Our initial analysis also focused on inferences children generated about information stated in the text (i.e., information that

Downloaded By: [University of Minnesota] At: 19:04 28 May 2008

146

R. M. Best et al.

was extrapolated from but not directly specified in the texts). Using previous inference research as a guide (e.g., Kintsch, 1993),
inferences were classified as text-based, elaborative, global, or irrelevant. However, we did not further analyze inference data because
too few inferences were produced (inferences comprised only 3%
of cued recall question answers and 1% of free recall question
answers).
All the free recall and cued recall was coded by a trained graduate student. Half the data was second coded by the first author.
Interrater reliability was evaluated for all dimensions of coding
(see below). Simple agreement and kappa analyses indicated that
agreement reached 90% or above on all dimensions, which indicates a high level of agreement. Disagreements were resolved by
discussion between raters.
PROPOSITION ANALYSIS
The proposition-based analysis was conducted from the recall
transcriptions. The recall analyses assessed the number of propositions recalled. To account for the completeness of information recalled, a value of 1.0 was assigned to recall that contained the main
proposition and more detailed information cited in the subproposition. A value of 0.5 was assigned to recall data that contained the
main proposition but that did contain the detailed information in
the subproposition. A value of 0 was assigned when the proposition
was not recalled. For example, a value of 0.5 was assigned for the
sentence Plants need sunlight, water, and air to live when a child
stated that plants need water to live. A value of 1 was assigned when
the child stated that plants need water, sunlight, and air. Because
childrens propositional recall sometimes contained erroneous information, recall that contained such information received a score
of 0.5. For example, a child may have erroneously stated, Plants
do not need sunlight. Once recall scores were totaled for a text,
they were transformed into a proportion because there were an
unequal number of propositions contained in the two texts.
Free recall and cued recall propositions were summed into
separate scores to focus on different dimensions of comprehension
across tasks. Thus, it was possible for children to recall the same
information in the free recall task and in the cued recall task.
For the free recall analysis, proportion scores were calculated by
dividing the free recall score by the number of propositions that

147

Downloaded By: [University of Minnesota] At: 19:04 28 May 2008

Differential Competencies

could potentially be recalled for each text. For the cued recall
analysis, the cued recall score was divided by the number of directly
relevant propositions that could be potentially recalled.
To determine which sentences contained directly relevant information, raters identified propositions that directly related to
the cued recall directives from three categories: directly related
sentences, indirectly related sentences, and irrelevant sentences.
Note that the same proposition was sometimes classified as directly
relevant for more than one cued recall question; in such cases, the
same proposition was counted for each of the relevant questions.
Kappa analyses showed that there was a high level of agreement
between raters across both texts (weighted kappa = .85 expository
text and .85 for the narrative text). Disagreements were resolved
by discussion. There were 32 main propositions directly relevant
to the expository directives and 35 propositions directly relevant
to the narrative directives.
Results
Table 2 presents the descriptive statistics from the free recall,
cued recall, and multiple-choice questions. For the multiple-choice
questions, data were analyzed for all questions for each text, regardless of type (e.g., text-based or bridging inference). Internal consistency reliability (coefficient alpha) was notably higher for the
scores stemming from summing across all questions, and preliminary analysis did not indicate differences in childrens responses
for the different question types. Preliminary analyses indicated
that the proportion scores from the recall tasks were low, which
indicates that children recalled a limited amount of information
relating to the texts.

TABLE 2 Childrens Proportion Correct Scores on the Narrative

and Expository Comprehension Tasks
Questions
Comprehension Measure

Narrative Text
M (SD)

Expository Text
M (SD)

Multiple-choice questions
Free recall propositions
Cued recall propositions

.72 (.17)
.10 (.07)
.15 (.10)

.49 (.15)
.04 (.03)
.07 (.03)

148

R. M. Best et al.

Downloaded By: [University of Minnesota] At: 19:04 28 May 2008

Genre
The first set of analyses investigated of the role of genre in childrens comprehension of the narrative and expository texts. A review of the descriptive statistics included in Table 3 revealed that
across the comprehension measures, childrens scores were notably higher for the narrative text than for the expository text.
Within-subjects ANOVAs confirmed that children comprehended
the narrative text better than expository text across the multiplechoice questions, F(1,60) = 54.87, MSE = 0.029, p < 0.001, the free
recall task, F(1,55) = 40.62, MSE = 0.002, p < 0.001, and the cued
recall task, F(1,54) = 30.48, MSE = 0.006, p < 0.001. These results
demonstrate that that childrens comprehension, across different
methods of assessment, was notably affected by text genre.
We hypothesized that the childrens performance on the comprehension tasks would be affected by their competencies as well
as genre. To investigate the importance of the observed effects of
TABLE 3 Correlations Between Comprehension of the Narrative and
Expository Texts, World Knowledge, and Decoding Skills
Reading Competency Measure

Comprehension Measure
by Genre
Narrative multiple-choice
questions
Narrative free recall
propositions recalled
Narrative cued recall
propositions recalled
Expository multiple-choice
questions
Expository free recall
propositions recalled
Expository cued recall
propositions recalled

World
Knowledge
(M = 97.4,
SD = 11.2)
U

Decoding
Skills
(M = 102.1,
SD = 9.7)
U

t Value for
Differences
Between
Corrected
Correlations

.36

.46 .47

.64

4.85

.28

.36

.46

.63

6.49

.41

.52 .48

.65

4.01

.37

.47

.14

.21

3.67

.47 .58

.33

.48

2.76

.44 .55

.14

.21

3.09

Note. U = uncorrected correlation. C = corrected correlation.

p < .05; p < .01, p < .001.

Downloaded By: [University of Minnesota] At: 19:04 28 May 2008

Differential Competencies

149

genre, a series of ANCOVAs were conducted in which genre was

the within-subjects variable and the measure of world knowledge
and the measure of decoding skills were covariates. Interestingly,
none of the statistically significant effects of genre observed in the
first analyses remained significant in the covariate analyses across
the multiple-choice questions, F < 1, propositions recalled during free recall, F = 2.37, p = .12, and propositions recalled during
cued recall, F = 3.42, p = 0.07. These findings suggest that childrens performance on the narrative and expository comprehension assessments are importantly related to reader competencies.
Correlations
Table 3 reports correlations between the reading competency
measuresworld knowledge and decoding skillsand scores from
the multiple-choice questions, propositions recalled during free
recall, and propositions recalled during cued recall. Correlations
are presented between these measures for both the narrative text
and the expository text. Because the standard deviations of the two
reading competency measures were smaller than those expected
of the population (i.e., 15; see Table 3), all correlations were corrected for attenuation due to restriction of range. Because the corrected correlations best represent the correlations between the
measures in the population, only corrected correlations will be
discussed.
For the narrative text, all comprehension measures except
Narrative Free Recall were significantly and moderately correlated
with both world knowledge and decoding skills.1 Across all three
measures of reading comprehension for the narrative text, the corrected correlations with decoding skills were significantly higher in
magnitude than their respective corrected correlations with world
knowledge (Cohen, Cohen, West, & Aiken, 2003). For the expository texts, all three comprehension measures were significantly
and moderately correlated with world knowledge. The magnitude
of the correlations between these comprehension measures and
world knowledge was significantly higher than their respective corrected correlations with decoding skills.
Importantly, there was a strong positive correlation between
world knowledge and decoding skills, r = .69, corrected correlation = .79), such that children with high world knowledge also

Downloaded By: [University of Minnesota] At: 19:04 28 May 2008

150

R. M. Best et al.

had high decoding skills and vice versa. Thus, there are shared
influences between world knowledge and decoding skills in text
comprehension, particularly for narrative texts.
Multiple Regression
A series of hierarchical (also called sequential) multiple regression analyses were used to identify the relative importance and
incremental value of world knowledge and decoding skills to comprehension of the narrative and expository texts. Building on the
correlation analyses, scores from the multiple-choice questions,
the free recall propositions recalled, and the cued recall propositions recalled for each text were used as dependent variables.
Independent variables were world knowledge and decoding skills.
Table 4 presents a summary of the regression analyses for the narrative text, and Table 5 presents a summary of the regression analyses
for the expository text.
For the narrative text, when the decoding skills variable was
entered first in the equation as an independent variable, it was
TABLE 4 Summary of Hierarchical Regression Analyses for Comprehension of
Narrative Text
Regression and
Order of Entry

SE B

Narrative MC questions (n = 61)

1. Decoding skills
.01
.00
2. World knowledge .00 .00
1. World knowledge .01 .00
2. Decoding skills
.01
.00
Narrative free recall (n = 60)
1. Decoding skills
.00
.00
2. World knowledge .00 .00
1. World knowledge .00 .00
2. Decoding skills
.00
.00
Narrative cued recall (n = 58)
1. Decoding skills
.01
.00
2. World knowledge .00 .00
1. World knowledge .00 .00
2. Decoding skills
.00
.00
Note. MC = Multiple-choice questions.

R2

"R 2

F(df 1,df2)

.47
.09
.37
.41

.22
.23
.14
.27

.00

.09

16.82 (1,59)
.28 (1,58)
9.33 (1,59)
6.67 (1,59)

<.001
.598
.003
.012

.46
.07
.28
.50

.21
.21
.08
.21

.00

.13

15.43 (1,58)
.16 (1,57)
5.03 (1,58)
9.59 (1,57)

<.001
.692
.029
.003

.48
.16
.41
.38

.23
.25
.17
.25

.01

.08

16.96 (1,56)
.95 (1,55)
11.53 (1,56)
5.46 (1,55)

<.001
.333
.001
.023

151

Downloaded By: [University of Minnesota] At: 19:04 28 May 2008

Differential Competencies

TABLE 5 Summary of Hierarchical Regression Analyses for Comprehension of

Expository Text
Regression and
Order of Entry

SE B

Expository MC questions (n = 61)

1. World knowledge .06
.02
2. Decoding skills
.04 .03
1. Decoding skills
.03
.03
2. World knowledge .09
.03
Expository free recall (n = 56)
1. World knowledge .00
.00
2. Decoding skills
.00
.00
1. Decoding skills
.00
.00
2. World knowledge .00
.00
Expository cued recall (n = 56)
1. World knowledge .00
.00
2. Decoding skills
.00
.00
1. Decoding skills
.00
.00
2. World knowledge .00
.00

R2

"R 2

F(df1,df2)

.37
.21
.15
.51

.14
.16
.02
.16

.02

.14

9.49 (1,59)
1.53 (1,58)
1.34 (1, 59)
9.57 (1, 58)

.003
.221
.251
.003

.41
.12
.34
.33

.17
.17
.11
.17

.01

.06

10.71 (1,54)
.50 (1,53)
6.94 (1, 54)
3.81 (1, 53)

.002
.483
.011
.056

.47
.26
.13
.61

.19
.23
.02
.23

.04

.21

12.65 (1,54)
2.74 (1,53)
.95 (1, 54)
14.61 (1, 53)

.001
.104
.335
<.001

Note. MC = Multiple choice.

a statistically significant predictor of comprehension across the

multiple-choice questions, the free recall propositions recalled,
and the cued recall propositions recalled ( p < .001). It alone accounted for more than 20% of the variance in these comprehension measures. In comparison, when the world knowledge variable
was entered first in the equation, it also proved to be a significant
predictor of comprehension across all three measures ( p < .05).
It alone accounted for less than 18% of the variance in the comprehension measures (R 2 range from .08 to .17). However, once
decoding skills was included in the equation, world knowledge did
not add significantly to the prediction of comprehension. In fact,
it explained less than 2% of additional variance in any comprehension variable once reading decoding skills was controlled for.
In contrast, with world knowledge entered first into the equation,
decoding skills still proved to be a significant predictor of comprehension. For example, it accounted for an additional 13% of the
variance in the free recall propositions recalled variable beyond
that of world knowledge.
For the expository text, world knowledge was a much stronger
and often singular predictor of comprehension. When world

Downloaded By: [University of Minnesota] At: 19:04 28 May 2008

152

R. M. Best et al.

knowledge was entered first in the equation as an independent

variable, it was a statistically significant predictor of comprehension across the multiple-choice questions, the free recall propositions recalled, and the cued recall propositions recalled ( p <
.01). It alone accounted for between approximately 14 and 19% of
the variance in the comprehension measures. Once world knowledge was considered, decoding skills did not add significantly to
the prediction of comprehension. Consistent with these results,
when decoding skills was entered first into the equation, it was a
significant predictor for only the free recall propositions recalled
( p = .011). However, once decoding skills was controlled for, world
knowledge added significantly to the prediction of comprehension as measured by the multiple-choice questions and cued recall
propositions recalled ( p < .01). World knowledge added approximately 14% to the prediction of comprehension measured by the
multiple-choice questions and more than 21% to the comprehension measured by the cued recall.
Discussion
The study was designed to investigate factors that drive the comprehension of texts appropriate for children in the third grade
because these children are entering the period known as the
fourth-grade slump. Specifically, we examined the importance of
two reading competencies in the comprehension of expository
and narrative texts in a diverse group of third-grade students. Our
findings support our hypotheses that (a) narrative texts are often
comprehended more successfully than expository texts and (b)
reader competencies (world knowledge and decoding skills) have
differential importance during the comprehension of texts from
different genres. Collectively, these findings are evidence that the
skill sets that are necessary for understanding narratives are different from that of expository texts (Bridge & Tierney, 1981).
Our results support previous research indicating that word decoding skills are driving forces in the comprehension of narrative
texts (e.g., Juel, Grittith, & Gough, 1986). In fact, regression analyses demonstrate that even when world knowledge is considered
as a predictor of reading comprehension, decoding skills add notably to the predictive model. Accurate and automatic word decoding may free up childrens working memory space, which can be

Downloaded By: [University of Minnesota] At: 19:04 28 May 2008

Differential Competencies

153

used for memories of the text base and for inference making (Perfetti, 1985; Seigneuric, Ehrlich, Oakhill, & Yuill, 2000). Although
our analyses indicated notable relations between world knowledge
and comprehension of narrative texts, most children have welldeveloped schemata about the settings, actions, and events described by narrative texts (Cote et al., 1998; Nelson, 1996), so individual differences in world knowledge may be less important for
accurate comprehension than those associated with the accurate
and automatic decoding of words.
Too little research has examined the comprehension of expository texts with children in the middle of elementary school (cf.
Wilson & Rupley, 1997). However, previous research has shown that
there are large effects of knowledge on reading comprehension
for adults and older school students (e.g., Afflerbach, 1986; McNamara & Kintsch, 1996). Indeed, the effects of knowledge often
override effects of reading abilities (Chiesi et al., 1979). This study
provides evidence that world knowledge is an important consideration when comprehension problems are observed during the
elementary school yearsparticularly for expository texts. For example, successful comprehension of the expository text, a science
text focusing on the needs of plants, was shown to be influenced
by the level of readers prior knowledge. The relations between
comprehension of this text and world knowledge, across a number of measures of comprehension, were significant and moderate
in magnitude. Decoding skills demonstrated much lower and inconsistent relations with comprehension of our expository text. In
fact, when allowed to compete in the prediction of comprehension of expository texts, the influence of world knowledge largely
dwarfed that of decoding skills. We can conclude that children
with less prior knowledge will struggle to form a coherent situation model when reading expository texts because they are not
able to generate the necessary inferences.
Our results are indicative of one problem that children may
face when they reach the third and fourth gradeshow to deal
with low cohesion science texts that demand knowledge that these
readers may not yet possess. This situation, we expect, could spiral
such that the child fails to understand the learning material and
thus goes deeper into a knowledge debt (Stanovich, 1986). By
the fourth grade, the situation would be aggravated and readily
apparent. Certainly, other factors will come into play during this

Downloaded By: [University of Minnesota] At: 19:04 28 May 2008

154

R. M. Best et al.

period, such as changes in the nature of instruction and changes in

reader motivation and strategy use (Artelt, Schiefele, & Schneider,
2001; Carroll, 2000; Mokhtari & Reichard, 2002).
Implications and Possible Remedies
CONSIDER EFFECTS OF TEXT GENRE
It is important that more research examine the comprehension of expository texts with young readers during the period
spanning third through fifth grade. In addition, researchers examining reading comprehension of children should clearly convey the genre of texts and passages used in their studies when
they disseminate research findings. Test developers and others involved assessment practices should consider the effects of genre
when designing and using reading comprehension assessment instruments and during content reviews of the instruments. Unless
general measures of reading comprehension are desired, perhaps
distinct measures of comprehension for select genres should be
sought and used, due to their differential relations with decoding skills and knowledge. Finally, those engaged in the design of
curricula or interventions to improve childrens reading comprehension should consider using strategies and techniques tailored
to each genre (see Baker, Gersten, & Grossen, 2002).
USE STRATEGIES ASSOCIATED WITH KNOWLEDGE
Because world knowledge is so closely tied to comprehension
of expository texts, instructional techniques and student-driven
strategies should be employed to match childrens knowledge
more appropriately to texts. Unfamiliar words in text present interference with the comprehension of expository texts, so children
may benefit from strategy instruction for interpreting word meanings in context (Kuhn & Stahl, 1998). Word familiarity may also be
aided by previewing texts to identify challenging words and to promote access to external resources, such as dictionaries, to obtain
rich descriptions of the word meanings (Baker et al., 2002).
Children may benefit from being taught strategies that help
them integrate information in a text with their prior knowledge.
Connections to prior knowledge result in a more coherent and
stable representation of the text content (e.g., Kintsch, 1998; McNamara et al., 1996). For example, after previewing a text, children

Downloaded By: [University of Minnesota] At: 19:04 28 May 2008

Differential Competencies

155

may be prompted to recall prior knowledge about the content of

a text or to organize their prior knowledge in structures similar
to graphic organizers (Ellis, 1993; Englert & Mariage, 1991). Children may also be prompted to integrate information from a text
with prior knowledge using imaging strategies (Baker, 1996).
Reading strategy instruction has been found to help young
adults and high-school students to overcome their knowledge
deficits (e.g., Baker, 1985; Best, Ozuru, & McNamara, 2004; McNamara, 2004; McNamara & Scott, 1999; OReilly, Best, & McNamara,
2004; Rosenshine & Meisler, 1994). For example, Self-Explanation
Reading Training (SERT) has been shown to help low-knowledge
and less strategic readers overcome knowledge deficits when reading expository texts. SERT teaches students to self-explain text by
using active reading strategies, such as making bridging and elaborative inferences using common sense and logic. An automated
system called the Interactive Strategy Trainer for Active Reading
and Thinking (iSTART) has been developed to deliver SERT using
pedagogical agents and interactive dialog (McNamara, Levinstein,
& Boonthum, 2003; OReilly, Sinclair, & McNamara, 2004). By using the latest technologies in intelligent tutoring systems, iSTART
guides young readers through text by suggesting different reading
strategies. Although the techniques and systems developed thus
far are targeted at young adults, a similar system may help younger
readers learn strategies that help them navigate difficult text.
ACKNOWLEDGE EFFECTS OF TEXT COHESION
The readers fund of available knowledge becomes particularly important when the reader is faced with low cohesion texts.
Text cohesion refers to the degree to which readers need to generate inferences to construct coherent mental representations from
information explicitly stated in the text. Signals in a text (such as
headers, relations between adjacent sentences, and explanations
about concepts) determine the extent to which readers need to
make inferences. Texts are considered to be low cohesion when
the reader is provided few signals in text. For accurate comprehension of these texts, readers are required to generate the appropriate inferences during the process of comprehension.
Analyses of texts used during the elementary-school period
indicate that text cohesion is a particular problem (McNamara,
Graesser, Louwerse, & Ozuru, 2005). Although many texts at

Downloaded By: [University of Minnesota] At: 19:04 28 May 2008

156

R. M. Best et al.

elementary grade levels use short sentences and include only words
that are relatively familiar, these texts often lack the cohesive devises that young readers need to comprehend them well (Graesser,
McNamara, & Louwerse, 2003). For example, text writers may
assume a prior knowledge on the part of the reader and consequently leave out information that scaffolds understanding (Beck,
McKeown, Sinatra, & Loxterman, 1991). This omission of information can be considered a cohesion gap that is often detrimental
to comprehension for low-knowledge readers. In addition, thirdand fourth-grade texts often contain very few connectives, such as
although, because, consequently, and so on. Connectives are essential
for readers with less knowledge to know what relationships exist
between successive ideas.
We assume that text quality can be improved and that this quality can be tailored to reader competencies. For instance, comprehension for low-knowledge readers may be facilitated by more cohesive texts, which contain fewer cohesion gaps (Britton, Gulgoz,
& Glynn, 1993; McNamara et al., 1996; McNamara & Kintsch, 1996;
Vidal-Abarca, Martinez, & Gilabert, 2000). The general approach
to increasing text cohesion is to add surface-level indicators of
relations between ideas in the text, such as explicit linguistic elements (e.g., words, features, cues, signals, and constituents), that
guide the reader to the meaning of the text. Thus, signaling causal
connections, with terms such as because and consequently, help the
reader interpret and remember relationships between concepts.
Accordingly, one way to help readers comprehend texts is to modify
cohesion. This modification can be done in numerous ways, such
as adding low-level information (e.g., identifying anaphoric referents, synonymous terms, connective ties, and headers) and supplying background information that was previously left unstated in
the text.
However, one roadblock to enhancing the cohesion of texts
has been the lack of automated tools measuring cohesion. Currently, readers, writers, editors, educators, researchers, and policymakers can only estimate the appropriateness of a text using common readability measures, such as Flesch Reading Ease and the
FleschKincaid Grade Level. These measures are based on superficial factors, such as the number of words in the sentences and
the number of letters or syllables per word (Beck et al., 1991; Britton et al., 1993). A new automated tool, Coh-Metrix (version 1.4;

Downloaded By: [University of Minnesota] At: 19:04 28 May 2008

Differential Competencies

157

Graesser et al., 2004; McNamara, Louwerse & Graesser, 2003),

produces 50 measures of cohesion relations and over 200 measures of language and discourse by applying modules that use lexicons, classifiers, syntactic parsers, shallow semantic interpreters,
conceptual templates, latent semantic analysis, and other components widely used in computational linguistics. These measures
from Coh-Metrix, which are sensitive to a range of cohesion relations, classes of inferences, as well as reader abilities, should extend or replace standard readability measures. With the use of
these measures, a better understanding of the effects of cohesion
on comprehension will provide valuable insight and explicit direction on how to improve comprehension of expository texts.
Limitations
This study is limited in a number of ways. First, although we believe
that we selected the two most consistent, powerful, and theoretically valid competencies for reading comprehension, other reading competencies deemed important in previous research were
omitted from this study. They include verbal working memory ability (Cain et al., 2004; Seigneuric et al., 2000), oral language-based
abilities (Carver & David, 2001; Storch & Whitehurst, 2002), general intelligence (Vellutino, 2003), metacognitive strategy knowledge (Baker, 1985; Wong, 1985), and knowledge of text structure
(Williams et al., 2004). Future research should examine these competencies to determine whether they demonstrate differential relations with comprehension of narrative texts and expository texts.
Because the results of our regression analyses indicated that the
combination of world knowledge and reading decoding skills accounted for less than 25% of the variance across our numerous
comprehension measures, it is probable that other competencies
would offer incremental value in predicting reading comprehension performance and in improving the understanding of the process of reading comprehension.
Although this study used multiple methods to assess reading
comprehension (e.g., multiple-choice questions and recall data),
this study is limited because it included data from only one
narrative text and one expository text. Although we took care to
choose typical texts, perhaps our findings would not generalize
to other texts in the same genre. Future research should include

Downloaded By: [University of Minnesota] At: 19:04 28 May 2008

158

R. M. Best et al.

aggregated comprehension measures that stem from reading and

understanding more than one text to overcome the possible
specific influences of text characteristics, such as content and
structure.
Finally, it is possible that our sampling techniques may have
influenced our results. For example, it is possible a sample of children with lower reading competencies and comprehension skills
from low socioeconomic status background may have yielded different patterns of results. We recommend future research be conducted with such subgroups.
Conclusion
This article has addressed factors that drive reading comprehension among third-grade children and that that may lead to fourthgrade slump. Its results indicate the effects of text genre and the
strong influence of world knowledge on comprehension of expository texts. Specifically, children with low levels of world knowledge
are likely to encounter difficulty understanding expository texts,
which are first introduced during the third and fourth grades.
These results point to the possibility that the decline in reading
performance observed during the third and fourth grades may be
explained, at least in part, by deficits in the readers world knowledge relevant to expository texts.
Acknowledgements
The research was supported by the Institute for Education Sciences
(IES R3056020018-02). Any opinions, findings, and conclusions
or recommendations expressed in this material are those of the
authors and do not necessarily reflect the views of the IES. Portions of this research were presented at the annual meeting of the
American Educational Research Association (2004) and the Sixth
International Conference of the Learning Sciences (2004). Portions of this study were published in the nonarchival proceedings
from the Sixth International Conference of the Learning Sciences.
We express thanks to a number of individuals who contributed
to this project. We are particularly thankful to Art Graesser and
Max Louwerse for their help at all levels of this project. We are
grateful to Tenaha OReilly, Yasuhiro Ozuru, David Dufty, and Kim

Downloaded By: [University of Minnesota] At: 19:04 28 May 2008

Differential Competencies

159

Sumara for their contributions to data collection and analysis and

to Jayme Sayroo McIntyre for propositionalizing and coding the
recall data. We thank Celeste Arps, Scott Baldwin, Jill Bose
Deakins, Jennifer Harmon, Jennifer Key, Cheryl Lykins, Terra Mayberry, Melissa McGhee, Bonnie Parnicky, Renee Bergeron, and
Kristen Yanchak for serving as test examiners. Finally, we are indebted to the parents, teachers, staff, and children at Idlewild Elementary and White Station Elementary as well as Memphis City
Schools administration for their support.
Note
1. We recognize that there is no standard rule-of-thumb for providing nominal
labels for r values. We drew from the following general labels: negligible to very
weak (0.0 to 0.2), weak (0.2 to 0.4), moderate (0.4 to 0.7), strong (0.7 to 0.9), and
very strong (0.9 to 1.0).

References
Addison-Wesley. (1998). Phonics take-home reader, grade 2. Upper Saddle River, NJ:
Author.
Afflerbach, P. (1986). The influence of prior knowledge on expert readers importance assignment process. In J. A. Niles & R. V. Lalik (Eds.), National reading
conference yearbook, 35. Solving problems in literacy: Learners, teachers and researchers
(pp.3040). Rochester, NY: National Reading Conference.
Artelt, C., Schiefele, U., & Schneider, W. (2001). Predictors of reading literacy.
European Journal of Psychology of Education, 16, 363383.
Baker, L. (1985). Differences in the standards used by college students to evaluate
their comprehension of expository prose. Reading Research Quarterly, 20, 297
313.
Baker, L. (1996). Social influences on metacognitive development in reading. In
C. Cornoldi & J. Oakhill (Eds.), Reading comprehension difficulties (pp. 331351).
Hillsdale, NJ: Erlbaum.
Baker, S., Gersten, R., & Grossen, B. (2002). Interventions for students with reading comprehension problems. In M. R. Shinn, H. M. Walker, & G. Stoner
(Eds.), Interventions for academic and behavior problems II: Prevention and remedial approaches (pp. 731754). Bethesda, MD: National Association of School
Psychologists.
Beck, I. L., & McKeown, M. G. (1992). Young students social studies learning:
Going for depth. In W. H. Slater (Eds.). Elementary school literacy: Critical issues
(pp. 133156). Norwood MA: Christopher Gordon.
Beck, I. L., McKeown, M. G., Sinatra, G. M., & Loxterman, J. A. (1991). Revising
social studies text from a text processing perspective: Evidence of improved
comprehensibility. Reading Research Quarterly, 26, 251276.

Downloaded By: [University of Minnesota] At: 19:04 28 May 2008

160

R. M. Best et al.

Best, R. M., Ozuru, Y., & McNamara, D. S. (2004, June). Self-explaining science
texts: Strategies, knowledge and reading skill. Paper presented at the International
Conference for the Learning Sciences, Los Angeles, CA.
Bowen, B. A. (1999). Four puzzles in adult literacy: Reflections on the national
adult literacy survey. Journal of Adolescent and Adult Literacy, 42, 314323.
Brand-Gruwel, S., Aarnoutse, C. A. J., & Van den Bos, K. P. (1998). Improving
text comprehension strategies in reading and listening settings. Learning and
Instruction, 8, 6381.
Bridge, C. A., & Tierney, R. J. (1981). The inferential operations of children across
texts with narrative and expository tendencies. Journal of Reading Behavior, 8,
201214.
Britton, B. K., Gulgoz, S., & Glynn, S. (1993). Impact of good and poor writing
on learners: Research and theory. In B. K. Britton, A. Woodward, & M. R.
Binkley (Eds.), Learning from textbooks: Theory and practice (pp. 146). Hillsdale,
NJ: Lawrence Erlbaum.
Cain, K. (1996). Story knowledge and comprehension skill. In C. Cornaldi &
J. Oakhill (Eds.), Reading comprehension difficulties: Processes and intervention
(pp. 167192).Mahwah, NJ: Erlbaum.
Cain, K., Oakhill, J., & Bryant, P. (2004). Childrens reading comprehension ability: Concurrent prediction by working memory, verbal ability, and component
skills. Journal of Educational Psychology, 96, 3142.
Carroll, J. B. (2000). The analysis of reading instruction: Perspectives from psychology and linguistics. Scientific Studies of Reading, 4, 317.
Carver, R. P., & David, A. H. (2001). Investigating reading achievement using a
causal model. Scientific Studies of Reading, 5, 107140.
Chi, M., Feltovich, P., & Glaser, R. (1981). Categorization and representation of
physics problems by experts and novices. Cognitive Science, 5, 121152.
Chiesi, H. I., Spilich, G. J., & Voss, J. F. (1979). Acquisition of domain-related
information in relation to high and low domain knowledge. Journal of Verbal
Learning and Verbal Behavior, 18, 275290.
Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2003). Applied multiple regression/correlation analysis for the behavioral sciences (3rd ed.). Mahwah, NJ:
Lawrence Erlbaum.
Cote, N., Goldman, S. R., & Saul, E. U. (1998). Students making sense of informational text: Relations between processing and representation. Discourse
Processes, 25, 153.
Curtis, M. E. (1980). Development of components of reading skill. Journal of
Educational Psychology, 90, 294311.
Ellis, E. S. (1993). Integrative strategy instruction: A potential model for teaching content-area subjects to learning disabled adolescents. Journal of Learning
Disabilities, 26, 358383.
Englert, C. S., & Mariage, T. V. (1991). Making students partners in the comprehension process: Organizing the reading POSSE. Learning Disability Quarterly,
14, 123138.
Get a Fact Sheet for your community. (2006, July). Retrieved November 8, 2006 from
http://factfinder.census.gov/home/saff/main.html? lang=en
Givon,
T. (1993). English grammar. A function-based approach (2 vols.). Amsterdam:
John Benjamins.

Downloaded By: [University of Minnesota] At: 19:04 28 May 2008

Differential Competencies

161

Graesser, A. C., McNamara, D. S., & Louwerse, M. M. (2003). What do readers

need to learn in order to process coherence relations in narrative and expository text. In A. P. Sweet & C. E. Snow (Eds.), Rethinking reading comprehension
(pp. 8298). New York: Guilford.
Graesser, A. C., McNamara, D. S., Louwerse, M. M., & Cai, Z. (2004). Coh-Metrix:
Analysis of text on cohesion and language. Behavioral Research Methods, Instruments and Computers, 36, 193202.
Graesser, A. C., Singer, M., & Trabasso, T. (1994). Constructing inferences during
narrative text comprehension. Psychological Review, 101, 371395.
Hannon, B., & Daneman, M. (2001). A new tool for measuring and understanding
individual differences in the component processes of reading comprehension.
Journal of Educational Psychology, 93, 103128.
Johnson-Laird, P. N. (1983). Mental models. Cambridge: Cambridge University
Press.
Juel, C., Grittith, P. L., & Gough, P. B. (1986). Acquisition of literacy: A longitudinal study of children in first and second grade. Journal of Educational Psychology,
78, 243255.
Kamberelis, G., & Bovino, T. D. (1999). Cultural artifacts as scaffolds for genre
development. Reading Research Quarterly, 34, 138170.
Kintsch, W. (1988). The role of knowledge in discourse comprehension: A
construction-integration model. Psychological Review, 95, 163182.
Kintsch, W. (1993). Information accretion and reduction in text processing: Inferences. Discourse Processes, 16, 193202.
Kintsch, W. (1998). Comprehension: A paradigm for cognition. Cambridge, England:
Cambridge University Press.
Kuhn, M. R., & Stahl, S. A. (1998). Teaching children to learn word meanings
from context: A synthesis and some questions. Journal of Literacy Research, 30,
119138.
Langer, J. A. (1986). Relations between levels on prior knowledge and the organization of recall. In. M. L. Kamil & A. J. Moe (Eds.). Perspectives in Reading
Research and Instruction. Washington, DC: National Reading Conference.
Louwerse, M. M. (2002). Computational retrieval of themes. In M. M. Louwerse
& W. Van Peer (Eds.), Thematics: Interdisciplinary studies (pp. 189212). Amsterdam: John Benjamins.
Lundeberg, M. (1987). Metacognitive aspects of reading comprehension: Studying understanding in legal case analysis. Reading Research Quarterly, 22, 407432.
Lyon, R. G. (2002). Reading development, reading difficulties, and reading instruction: Educational and public health issues. Journal of School Psychology, 40,
36.
McGrew, K. S., & Woodcock, R. W. (2001). Technical manual. Woodcock-Johnson III.
Itasca, IL: Riverside.
McNamara, D. S. (2004). SERT: Self explanation reading training. Discourse Processes, 38, 130.
McNamara, D. S., Graesser, A. C., Louwerse, M., & Ozuru, Y. (2005, June). Using
Coh-Metrix to measure cohesion. Poster session presented at the 15th Annual
Meeting of the Society for Text and Discourse, Amsterdam.
McNamara, D. S., Kintsch, E., Songer, N. B., & Kintsch, W. (1996). Are good texts
always better?: Interactions of text coherence, background knowledge, and

Downloaded By: [University of Minnesota] At: 19:04 28 May 2008

162

R. M. Best et al.

levels of understanding in learning from text. Cognition and Instruction, 14,

143.
McNamara, D. S., & Kintsch, W. (1996). Learning from texts: Effects of prior
knowledge and text coherence. Discourse Processes, 22, 247288.
McNamara, D. S., Levinstein, I. B., & Boonthum, C. (2004). iSTART: Interactive
strategy trainer for active reading and thinking. Behavioral Research Methods,
Instruments, & Computers, 36, 222233.
McNamara, D. S., Louwerse, M. M., & Graesser, A. C. (2003). Coh-Metrix: Automated
cohesion and coherence scores to predict text readability and facilitate comprehension
(Tech. Rep). Memphis, TN: The University of Memphis.
McNamara, D. S., & Scott, J. L. (1999). Training reading strategies. In M. Hahn &
S. C. Stoness (Eds.), Proceedings of the Twenty-first Annual Meeting of the Cognitive
Science Society (pp. 387392). Hillsdale, NJ: Erlbaum.
Means, M. L., & Voss, J. F. (1985). Star wars: A developmental study of expert
and novice knowledge structures. Journal of Memory and Language, 24, 746
757.
Meichenbaum, D., & Biemiller, A. (1998). Nurturing independent learners: Helping
students take charge of their learning. Cambridge, MA: Brookline.
Mokhtari, K., & Reichard, C. A. (2002). Assessing students metacognitive
awareness of reading strategies. Journal of Educational Psychology, 94, 249
259.
Mommers, M. J. (1987). An investigation into the relation between word recognition skills, reading comprehension and spelling skills in the first two years
of primary school. Journal of Research in Reading, 10, 122143.
Moyer, R., Daniel, L., Hackett, J., Baptiste, P., & Stryker, P. (2000). Science, grade 2.
Columbus, OH: McGraw-Hill.
Nelson, K. (1996). Language in cognitive development: The emergence of the mediated
mind. Cambridge, UK: Cambridge University Press.
Olson, M. W. (1985). Text type and reader ability: The effects on paraphrase and
text-based inference questions. Journal of Reading Behavior, 3, 199214.
OReilly, T., Best, R., & McNamara, D. S. (2004). Self-explanation reading training:
Effects for low-knowledge readers. In K. Forbus, D. Gentner, & T. Regier (Eds.),
Proceedings of the 26th Annual Meeting of the Cognitive Science Society (pp. 1053
1058). Mahwah, NJ: Erlbaum.
OReilly, T., & McNamara, D. S. (2002). Whats a science student to do? In W. D.
Gray & C. D. Schunn (Eds.), Proceedings of the Twenty-fourth Annual Meeting of
the Cognitive Science Society (pp. 726731). Mahwah, NJ: Erlbaum.
OReilly, T. P., Sinclair, G. P., & McNamara, D. S. (2004b). Reading strategy training: Automated versus live. In K. Forbus, D. Gentner, & T. Regier (Eds.), Proceedings of the 26th Annual Cognitive Science Society (pp. 10591064). Mahwah,
NJ: Erlbaum.
Perfetti, C. A. (1985). Reading ability. New York. Oxford University Press.
Reutzel, D. R., & Morgan, B. C. (1990). Effects of prior knowledge, explicitness,
and clause of order on childrens comprehension of causal relationships, Reading Psychology, 11, 93114.
Rosenshine, B., & Meister, C. (1994). Cognitive strategy instruction in reading.
In D. A. Hayes & S. A. Stahl (Eds.), Instructional models in reading (pp. 85107).
Hillsdale, NJ: Lawrence Erlbaum.

Downloaded By: [University of Minnesota] At: 19:04 28 May 2008

Differential Competencies

163

Roth, F. P., Speece, D. L., Cooper, D. H., & De La Paz, S. (1996). Unresolved mysteries: How do metalinguistic and narrative skills connect with early reading?
Journal of Special Education, 30, 257277.
Rupley, H. R., & Willson, V. L. (1996). Content, domain, and world knowledge:
Relationship to comprehension of narrative and expository text. Reading and
Writing, 8, 419432.
Seigneuric, A., Ehrlich, M.-F., Oakhill, J. V., & Yuill, N. M. (2000). Working memory resources and childrens reading comprehension. Reading and Writing, 13,
81103.
Snow, C. (2002). Reading for understanding: Toward an R&D program in reading
comprehension. Santa Monica, CA: RAND.
Spiro, R. J., & Taylor, B. M. (1980). On investigating childrens transition from narrative to expository discourse: the multi-dimensional nature of psychological classification
(Tech. Rep. No. 195). Urbana-Champaign, University of Illinois, Center for
the Study of Reading.
Stanovich, K. E. (1986). Matthew effects in reading: Some consequences of individual differences in the acquisition of literacy. Reading Research Quarterly, 21,
360407.
Stein, N. L., & Trabasso, T. (1981). Whats in a story: An approach to comprehension and instruction. In R. Glaser (Ed.), Advances in the psychology of instruction
(vol. 2, pp. 213267). Hillsdale, NJ: Erlbaum.
Storch, S. A., & Whitehurst, G. J. (2002). Oral language and code-related precursors to reading: Evidence from a longitudinal structural model. Developmental
Psychology, 38, 934947.
Sweet, A. P., & Snow, C. E. (Eds.). (2003). Rethinking reading comprehension. New
York: Guilford.
Taschow, H. G. (1969). Reading improvement in mathematics. Reading Improvement, 6, 6267.
Tun, P. A. (1989). Age differences in processing expository and narrative text.
Journal of Gerontology: Psychological Sciences, 44, 915.
Vellutino, F. R. (2003). Individual differences as sources of variability in reading comprehension in elementary school children. In A. P. Sweet & C.
E. Snow (Eds.), Rethinking reading comprehension (pp. 5181). New York:
Guilford.
Vellutino, F. R., Scanlon, D. M., & Tanzman, M. S. (1994). Components of reading
ability: Issues and problems in operationalizing word identification, phonological coding, and orthographic coding. In G. R. Lyon (Ed.), Frames of reference
for the assessment of learning disabilities: New views on measurement issues (pp. 279
329). Baltimore, MD: Paul H. Brookes.
Vidal-Abarca, E., Martinez, G., & Gilabert, R. (2000). Two procedures to improve
instructional text: Effects on memory and learning. Journal of Educational Psychology, 92, 107116.
Williams, J. P., Hall, K. M., & Lauer, K. D. (2004). Teaching expository text structure to young at-risk learners: Building the basics of comprehension instruction, Exceptionality, 12, 129144.
Wilson, V. L., & Rupley, W. H. (1997). A structural equation model for reading
comprehension based on background, phonemic, and strategy knowledge.
Scientific Studies of Reading, 1, 4563.

Downloaded By: [University of Minnesota] At: 19:04 28 May 2008

164

R. M. Best et al.

Wong, B. (1985). Metacognition and learning disabilities. In D. Forrest Pressley, G.

MacKinnon, & T. Waller (Eds.), Metacognition, cognition, and human performance
(vol. 2, pp. 137180). New York: Academic Press.
Woodcock, R. W., McGrew, K. S., & Mather, N. (2001). WoodcockJohnson III. Itasca,
IL: Riverside.
Zinar, S. (1990). Fifth-graders recall of propositional content and causal relationships from expository prose. Journal of Reading Behavior, 22, 181199.

Appendix
Multiple-choice questions
Questions
Orlando questions
Q1. What was Orlando?
Q2. Why did Mama tell Salvador that they had to sell
Orlando?
Q3. What was Salvador doing when his Mama told him that
they had to sell Orlando?
Q4. Why did Salvador run from his house?
Q5. How did Salvador follow Orlando home?
Q6. Why did Salvadors mother decide to keep Orlando?
Q7. What did Salvador use to make his fort?
Q8. Why was Salvadors Mama worried about Salvador?
Q9. Why did the tree catch on fire?
Q10. What area was between Salvadors house and the forest
where his fort was?
Q11. What happened when Salvador thought of his Mama in
the fort?
Q12. What caused Salvadors Mama to stand by the door on
the porch?
Plant questions
Q1. How do humans and animals help to meet the needs of
plants?
Q2. What gas do plants use to make food?
Q3. How does carbon dioxide enter plants?
Q4. What do roots do for plants?
Q5. Why would a plant grow toward a sunlit window?
Q6. Why do plants need sunlight?
Q7. What do both plants and animals need to live?
Q8. Why are plants important to humans?
Q9. Why do plants need chlorophyll to trap energy from
sunlight?
Q10. Which statement best describes simple plants?
Q11. Why would a big leaf make more food for a plant than
a small leaf?
Q12. What do our bodies use when we eat plants?

Question Type
Text-based
Inference
Text-based
Inference
Text-based
Inference
Text-based
Inference
Text-based
Inference
Text-based
Inference
Inference
Text-based
Inference
Text-based
Inference
Text-based
Inference
Text-based
Inference
Text-based
Inference
Text-based

Note. Questions eliminated from the analysis as children performed at ceiling level
(above 90% correct).