Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/260411127

Making Connections: Linking Cognitive Psychology and Intervention Research


to Improve Comprehension of Struggling Readers

Article in Learning Disabilities Research and Practice · February 2014


DOI: 10.1111/ldrp.12026

CITATIONS READS

59 3,411

3 authors, including:

Christine Espin Paul van den Broek


Leiden University Leiden University
82 PUBLICATIONS 3,843 CITATIONS 197 PUBLICATIONS 13,927 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Paul van den Broek on 12 February 2018.

The user has requested enhancement of the downloaded file.


Learning Disabilities Research & Practice, 29(1), 17–24

C 2014 The Division for Learning Disabilities of the Council for Exceptional Children

Making Connections: Linking Cognitive Psychology and Intervention


Research to Improve Comprehension of Struggling Readers
Kristen L. McMaster
University of Minnesota
Christine A. Espin and Paul van den Broek
University of Leiden

Many studies have demonstrated the efficacy of reading comprehension interventions for
struggling readers, including students with learning disabilities. Yet, some readers continue to
struggle with comprehension despite receiving these interventions. In this article, we argue that
an explicit link between cognitive psychology and intervention research contributes to knowl-
edge regarding for whom and under what conditions reading comprehension interventions are
most likely to be beneficial. First, we provide a brief overview of a cognitive perspective on
reading comprehension. Next, we illustrate an application of this perspective by describing
a collaborative project in which we examined interactions among reader characteristics, text
properties, and instructional contexts. Last, we highlight directions for future research and
implications for practice.

Many students, including those with learning disabilities ple, in their review of reading comprehension intervention
(LD), struggle to learn to read—a skill that is critical for suc- research, Fagella-Luby and Deshler (2008) noted that few
cess in school and society. Whereas many struggling readers interventions were grounded explicitly in theory, and argued
have difficulties primarily with basic skills such as decoding, that greater progress would be made if research and practice
others have difficulties with both decoding and comprehen- were built upon theory. Thus, the purpose of this article is
sion or with comprehension alone. Readers who struggle to describe a collaborative program of research that connects
with comprehension, and are at risk for or have been identi- research, practice, and cognitive theory to address compre-
fied with LD, are the focus of this article. hension difficulties of struggling readers. We describe results
Several research syntheses have described interventions of this collaborative work, and illustrate how such connec-
that are effective for improving comprehension skills of tions can lead to advances in both cognitive theory and inter-
struggling readers (Berkeley, Scruggs, & Mastropieri, 2010; vention research. First, we describe reading comprehension
Edmonds et al., 2009; Gajria, Jitendra, Sood, & Sacks, 2007; from a cognitive psychology perspective (also see Kendeou,
Gersten, Fuchs, Williams, & Baker, 2001; Swanson, 1999). van den Broek, Helder, & Karlsson, this issue). Next, we
Convergent findings from these syntheses are that: (1) com- describe factors related to reading comprehension difficul-
prehension of struggling readers, including those with LD, ties, and discuss how we have examined interactions among
can be improved; (2) effective interventions include identifi- these factors in our research. Last, we highlight directions
cation of text structure, vocabulary instruction, use of prior for further research and implications for practice.
knowledge, and cognitive strategy use; and (3) instruction
is most effective if it is overt and explicit. Despite these
positive findings, research also reveals that not all strug- WHAT IS READING COMPREHENSION?
gling readers benefit from generally effective interventions
(Fagella-Luby & Deshler, 2008), and that positive inter- Within a cognitive psychology perspective, reading compre-
vention effects are not always maintained and generalized hension is often defined as: (1) the translation of written
(Gersten et al., 2001). Thus, research focused on improving code into meaningful language units that are combined to (2)
struggling readers’ comprehension remains a priority in the build a coherent mental representation of the text (Kintsch
field. & van Dijk, 1978; Trabasso & van den Broek, 1985; van
Part of the call for continued research emphasizes the im- den Broek, White, Kendeou, & Carlson, 2009). Although the
portance of a stronger connection between theory and prac- breakdown and organization of these components vary some-
tice (e.g., Fagella-Luby & Deshler, 2008; Pressley, Graham, what among different theories of reading comprehension,
& Harris, 2006; van den Broek & Espin, 2009). For exam- most theories include these two components. For example,
the Simple View of Reading (Gough & Tunmer, 1986) por-
Requests for reprints should be sent to Kristen McMaster, University of trays reading comprehension as the product of decoding and
Minnesota. Electronic inquiries should be sent to mcmas004@umn.edu. linguistic comprehension. Decoding includes phonological
18 MCMASTER, ESPIN, AND VAN DEN BROEK: MAKING CONNECTIONS

1 2 3

Sammy and the New Bike

1. Once there was a boy named Sammy. 4

2. Sammy had dark hair

3. and brown eyes. 5

4. Sammy wanted a bike.

5. He asked his mother for a bike, 6 7

6. but she refused.

7. Sammy’s mother loved to cook. 8

8. Sammy decided to earn some money,

9. and asked for a job at a nearby grocery store. 9 10 11

10. The grocery store had lots of shelves,

11. and the shelves were different colors. 12

12. Sammy made deliveries for the grocery store,

13. and earned lots of money. 13

14. Finally, Sammy bought a new bike.

14

FIGURE 1 Illustration of a causal network for a story about Sammy and his bike.

awareness, phonics, and word recognition—skills needed to Most likely, your summary included the following story
translate written code into meaningful language units. Lin- events: There was a boy named Sammy (1) who wanted a
guistic comprehension includes vocabulary, syntactic and bike (4). He asked his mother for a bike (5) but she refused
semantic awareness, inference making, and construction of to buy it for him (6). Sammy decided to earn some money
mental schemas—skills needed to build coherent mental rep- (8) by working in a grocery story (9) to buy the bike himself
resentations of text. Much research has been conducted to (not in the text; requires background knowledge about earn-
test and refine theories for how these components interact to ing money to be able to buy something). He earned enough
support (or prevent) reading comprehension. See, for exam- money (13) to buy the bike (14). Your summary likely did
ple, work on the lexical quality hypothesis (e.g., Perfetti, not include the following events: Sammy had dark hair (2)
2007), interactive–compensatory model (e.g., Stanovich, and brown eyes (3), his mother loved to cook (7), the grocery
1984), and semantic–deficit model (e.g., Nation & Snowling, story had lots of shelves (10) and the shelves were different
1998). colors (11).
In our research, we have been primarily concerned with The events that readers are most likely to include in sum-
the construction of a coherent mental representation of text. maries of a story are those that are causally and logically
Specifically, we have focused on the connections that readers related to each other, and thus contribute to the coherence
must identify to build coherence. To build a coherent mental of the text (Trabasso, Secco, & van den Broek, 1984; van
representation of a text, readers must connect and integrate den Broek, Fletcher, & Risden, 1993). For two events (A and
current information in the text with information that appeared B) to be causally and logically related, the following criteria
earlier in the text and/or with background knowledge (van must be met: (1) A must occur prior to B; (2) A must be ac-
den Broek et al., 2009). To illustrate what it means to build tive at the time of B; (3) A must be necessary for B to occur
a coherent representation, we invite you to read the story within the context of the story; and (4) A must be sufficient
in Figure 1, and then create a summary of the story before for B to occur within the context of the story. In “Sammy
reading on. and the New Bike,” the events Sammy wanted a bike and he
LEARNING DISABILITIES RESEARCH 19

asked his mother for a bike are causally and logically related; erties of the text, and the instructional context (McNamara
the relation between the two events fits the above criteria. & Kendeou, 2011; RAND Reading Study Group, 2002; van
Sammy wanting a bike occurs prior to and is active when he den Broek et al., 1993, 2013). Reader characteristics include
asks his mother for a bike, and it is a necessary and sufficient decoding and comprehension skill, as well as motivation,
reason for him to ask his mother for a bike. interest, and goals for reading. Text properties include the
In building a coherent representation of a text, readers genre, content, format, and difficulty of a text. Instructional
create a causal network representing the causal and logical context includes the purpose for learning, the instructional
relations among events in the story (Trabasso et al., 1984). A tasks, the mode of delivery, student group size, duration and
causal network for the Sammy story is depicted in Figure 1. frequency, feedback and reinforcement, and so on. Any one
Each circle—or node—represents an event in the story, and of these factors—and the interactions among them—can con-
the arrows represent the causal connections between events. tribute to reading comprehension success or failure (van den
As shown in Figure 1, some events have many connections; Broek & Kramer, 1999). Below, we discuss how we have
others have few or none. Events with few or no causal con- addressed each of these factors, and the interactions among
nections do not contribute significantly to a coherent rep- them, in our research.
resentation of the story. For example, Event 7, “Sammy’s
mother loved to cook,” is not related to any other event in
the story. In fact, you may have stopped reading for a mo- A COLLABORATIVE PROGRAM OF RESEARCH
ment when you reached that event, and wondered why it was ON READERS, TEXTS, AND INSTRUCTIONAL
in the story. You may even have reread part of the story to CONDITIONS
determine whether you missed something, or scanned ahead
to see if his mother’s love for cooking would become impor- In the past decade, our research team has sought to better
tant. Other events, such as “Sammy had dark hair and brown understand when and how comprehension processes break
eyes,” have few connections and therefore are not central to down, with the ultimate goal of designing interventions that
the meaning of the text. address struggling comprehenders’ specific problems. We
When asked to recall or summarize texts, good readers have focused on fourth-grade readers, because late elemen-
tend to include the most highly connected events, and leave tary school marks a critical transition point when academic
out events with few connections. Van den Broek, Helder, and requirements include comprehension of increasingly com-
van Leijenhorst (2013) refer to this characteristic as sensitiv- plex texts. At this point, some students who appeared to be
ity to structural centrality, or “the extent to which a reader on track in reading in the early grades begin to struggle—a
identifies and processes information central to the semantic phenomenon often referred to as the “fourth-grade slump”
structure of the text” (p. 132). Younger readers and readers (e.g., Chall, Jacobs, & Baldwin, 1990; Sweet & Snow, 2003).
who struggle with comprehension are less sensitive to struc- Throughout this research, we have identified students
tural centrality, whereas more proficient readers are more whose primary difficulties are with comprehension rather
sensitive to structural centrality. Good readers are able to than decoding. Our general approach has involved two steps.
generate connections that span multiple episodes and events First, we have identified students who perform at the bot-
in a story, and to connect abstract parts of the text such as tom 25th percentile in overall reading proficiency, using
intentions and goals of the characters (van den Broek et al., Curriculum-Based Measurement (CBM) maze tasks (e.g.,
2009). In addition, good readers tend to add information from Fuchs & Fuchs, 1992). Students’ reading skill level is then
their background knowledge to fill in gaps in the coherence of confirmed by other reading assessments (e.g., standardized
a text. For example, in the Sammy story, you may have used measures of reading comprehension or district measures),
background information about earning money to be able to and teacher judgment. Second, we have screened out students
purchase something to infer that Sammy used his own money whose primary difficulties relate to decoding and fluency
to buy the bike. In contrast, struggling readers often fail to (e.g., those performing below grade-level on oral reading
add relevant background knowledge (even if they have it) or fluency). Below, we describe this research, focusing on how
they add irrelevant knowledge (e.g., Williams, 1993). characteristics of struggling comprehenders interact with text
Sensitivity to structural centrality and the building of properties and instructional contexts.
semantic networks depend in part on the reader’s “stan-
dards of coherence” (van den Broek, Bohn-Gettler, Kendeou,
Carlson, & White, 2011; van den Broek & Espin, 2012). Understanding Characteristics of Struggling
“Standards of coherence” refers to the “type and degree of Readers
semantic coherence that a reader aims to maintain while
reading” (van den Broek & Espin, 2012: p. 41). Poor read- Not all struggling comprehenders fit one specific profile, but
ers often have lower standards of coherence. That is, dur- rather form a heterogeneous group (e.g., Cain & Oakhill,
ing reading, they are more easily satisfied with a less-than- 2006; Nation, Clarke, & Snowling, 2002). An initial goal of
coherent depiction of the text, so they devote less attention our research was to better understand individual differences
and energy to the processes required to build a coherent among struggling readers, as well as how the comprehen-
representation. sion processes of struggling readers differ from nonstrug-
Many factors contribute to the standards of coherence gling readers (Rapp, van den Broek, McMaster, Kendeou, &
adopted by a reader, and the subsequent representation of Espin, 2007; van den Broek et al., 2006). Thus, we iden-
text, including the characteristics of the reader, the prop- tified struggling, average, and good readers using maze
20 MCMASTER, ESPIN, AND VAN DEN BROEK: MAKING CONNECTIONS

(Fuchs & Fuchs, 1992), confirming their status with the Think-aloud results indicated that struggling readers gen-
Gates-MacGinitie Reading Comprehension test (MacGinitie, erated more invalid elaborative inferences than did average
MacGinitie, Maria, & Dreyer, 2002). Participants completed and good readers. Further, a cluster analysis of struggling
measures of word identification, word attack, fluency, vocab- readers’ think-aloud data revealed two subgroups of strug-
ulary, listening comprehension, IQ, working memory, and gling readers. One group, called Paraphrasers, made few in-
motivation, along with eye tracking and think-aloud tasks. ferences but produced significantly more text repetitions and
Typical comprehension measures (such as recall or ques- paraphrases than did average and good readers as well as the
tioning) assess the products of reading—what the reader un- other group of struggling readers. The other group, called
derstands about a text after reading it. Yet, it is during the Elaborators, made many elaborative inferences—similar to
process of reading that comprehension occurs (Rapp et al., the number of elaborative inferences made by average and
2007). To better understand characteristics of readers who good readers and reliably more than Paraphrasers—but their
struggle with comprehension, we sought ways to assess the inferences often were inaccurate or invalid. These subgroups
reading process itself—what the reader is doing while reading replicated across several samples in several different years in
a text. We used two approaches commonly used in cognitive studies that have used the think-aloud approach with fourth-
psychology to assess the process of reading: eye-tracking and grade struggling readers (e.g., Carlson, Seipel, & McMaster,
think-aloud tasks. unpublished data; McMaster et al., 2012; van den Broek et al.,
Eye-tracking is a behavioral measure that provides infor- 2006).
mation about the moment-by-moment processes in which The above findings suggest that subgroups of struggling
readers engage during reading (Rayner, Chace, Slattery, & comprehenders engage in different processes while they read,
Ashby, 2006). In our research, participants wore a head- and thus may struggle with comprehension for different rea-
mounted eye-tracker while they read short passages dis- sons. These findings led to questions such as: (1) Does a
played on a computer screen. The eye-tracker recorded read- student’s subgroup classification as Paraphraser or Elabo-
ers’ eye movements from one point in the text to another rator depend on the type of text? and (2) Does a student’s
(saccades), where and for how long readers paused in the subgroup classification as Paraphraser or Elaborator interact
text (fixations), and where readers looked back in the text with instructional contexts?
(regressions).
Think-aloud tasks (see Ericsson & Simon, 1993) are de-
signed to reveal what people are thinking during completion Do Subgroups of Struggling Readers Vary
of a task. With regard to reading, think alouds are used to as a Function of Text Properties?
reveal the processes readers employ while reading a text. In
a think aloud task, participants read passages one sentence at Whether a student’s subgroup classification as Paraphraser
a time and say whatever comes to mind after each sentence. or Elaborator depends on the type of text has potential in-
In our research, think-aloud protocols were audio recorded, structional implications. For example, a struggling reader
transcribed, parsed into idea units, and coded to identify spe- might paraphrase extensively while reading stories, but rely
cific cognitive processes reflected in each thought unit. These more on irrelevant or inaccurate elaborative inferences while
processes include elaborative and predictive inferences (use reading informational texts. Alternatively, a struggling reader
of background knowledge to explain or predict text, coded might paraphrase exclusively across all types of texts, which
as either valid or invalid depending on relevance to the text), might support comprehension of some texts but not oth-
connecting and reinstatement inferences (inferences related ers. Such information would be useful in designing instruc-
to either the immediately preceding sentence or to earlier tion to help struggling readers comprehend a wide range of
parts of the text), associations (comments relating the text to texts.
a personal experience), paraphrases/text repetitions (restate- To address this question, we replicated the subgroup anal-
ments of the current sentence in the readers’ own words or yses with both narrative and informational texts across two
verbatim), metacognitive comments (comments on the read- fourth-grade samples in 2 years. We chose informational
ers’ own understandings or opinions about the text), and text, given the focus on this type of text in the late elemen-
questions (questions about the text that could not be coded tary grades (Best, Floyd, & McNamara, 2008; Sweet & Snow,
in the other categories). 2003). Struggling readers were identified using the two-step
Among the fourth graders in the study (n = 241), strug- process described previously (Year 1, n = 60; Year 2, n =
gling readers performed lower than average and good readers, 61). A think-aloud task was administered individually to each
and average readers performed lower than good readers, on student, using both narrative and informational texts.
word attack, fluency, vocabulary, listening comprehension, For both years, think-aloud data were submitted to sepa-
and IQ. There were no reliable group differences with re- rate cluster analyses for the narrative and informational texts.
spect to motivation or working memory. Eye-tracking results Similar subgroups emerged as in previous research (McMas-
showed that struggling readers had similar numbers of fixa- ter et al., 2012; van den Broek et al., 2006): Paraphrasers
tions as did average and good readers, but spent more time (Year 1, n = 16 from stories and 18 from informational
on each fixation. Further, struggling readers looked back at texts; Year 2, n = 22 from stories and 21 from informa-
previous text as often as did average and good readers but tional text) and Elaborators (Year 1, n = 44 from stories and
spent more time rereading text, suggesting that struggling 42 from informational texts; Year 2, n = 39 from stories
readers were encountering greater difficulty comprehending and 40 from informational text). The subgroups emerged
the text (cf. Rayner et al., 2006). in similar proportions as in previous research, with more
LEARNING DISABILITIES RESEARCH 21

Elaborators than Paraphrasers. A majority of students (93 In an initial test of this approach (van den Broek et al.,
percent in Year 1, 85 percent in Year 2) were identified as 2006), we conducted a controlled experiment in which strug-
being in the same subgroup using both narrative and infor- gling, average, and good fourth-grade readers (n = 78) an-
mational texts. This finding suggests that the Paraphraser and swered Causal or General questions while reading narrative
Elaborator subgroups are relatively stable across text types, texts in separate, counterbalanced, one-to-one tutoring ses-
but that a small number of students may vary in their sub- sions. The primary dependent variable involved verbal recall
group status depending on text type, at least across the narra- of the texts read during the Causal and General conditions.
tive and informational texts included in these studies. Future Recalls were coded for the proportion of all story events
research should broaden the range of text types (including dif- recalled, as well as for the proportion of highly connected
ferent types of informational texts, such as sequence, descrip- events recalled. Results indicated that Causal questioning led
tion, compare–contrast, problem–solution, cause–effect) to to significantly more complete recalls of text than did general
determine whether subgroups remain stable. Questioning (d = .53). Findings did not vary by struggling,
average, or good reader status.
Next, we applied the questioning approach to a classroom-
Do Subgroups of Struggling Readers Respond based intervention in a small, randomized field trial (McMas-
Differently to Instruction? ter et al., 2012). Struggling-, average-, and good fourth-grade
readers (n = 246) were randomly assigned to Causal, Gen-
In addition to understanding the characteristics of struggling eral, or “W” (who, what, where, when) Questioning condi-
readers and how they interact with different types of text, tions. Again, questioning was embedded in narrative texts,
we have sought ways to improve struggling readers’ com- but this time the intervention was delivered in a classwide
prehension of text. To do so, we developed and compared peer-mediated format, three times per week for 20–30 min-
different questioning interventions that build on cognitive utes over 9 weeks. Overall, readers made statistically sig-
models of reading comprehension. We examined whether nificant improvements from pre- to posttest on story recall
the type of questioning matters, and whether the effects of (ps < .001 to .04), but there was no reliable main effect of
different questioning approaches vary for readers in different question type, and no interaction between question type and
subgroups. struggling, average, or good comprehender status.
As discussed earlier, building a coherent mental repre- Further analyses revealed a more nuanced picture of
sentation of text depends on the extent to which the reader the possible effects of the different questioning approaches.
identifies important causal or logical relations among events We conducted post hoc analyses to determine whether sub-
and ideas in the text (Trabasso et al., 1984; van den Broek group status moderated struggling readers’ response to the
et al., 1993). One way to prompt readers to identify such different questioning approaches. These analyses revealed
relations is to ask specific “causal” questions at points in the an interaction between question type and subgroup: Elab-
text where it is important to make connections. For example, orators outperformed Paraphrasers in the Causal condition
in “Sammy and the New Bike” (Figure 1), a good causal (d = .86), Paraphrasers outperformed Elaborators in the Gen-
question might be, “Why did Sammy decide to earn some eral condition (d = 1.46), and Paraphrasers in the General
money?” This question is designed to prompt the reader to condition outperformed Paraphrasers in the Causal condition
attend to Sammy’s desire to get a bike, which is an event with (d = 1.52). In other words, struggling readers’ response to
many causal relations to other events in the story. To gen- the different questioning approaches appeared to vary de-
erate a coherent representation of the text, the reader must pending on subgroup status (see McMaster et al., 2012, for
connect Sammy’s decision to earn money to his desire to get a complete description of these results).
a bike. A less useful question might be, “What color hair did The differential response by the two subgroups provides
Sammy have?” because Sammy’s hair color is not relevant to an example of an interaction between reader characteristics
his desire to get a bike (or to get a job and earn money to get and instructional contexts. Specifically, the effects of dif-
the bike). Recalling the unimportant details is not necessary ferent questioning approaches may depend on the way strug-
for the reader to generate a coherent mental representation gling readers process text. Recall that, during the think-aloud
of this story. tasks, Elaborators tended to rely on background knowledge
In three separate studies, we contrasted the causal ques- to explain what they were reading (but not always in ways that
tioning approach with a “general” questioning approach, in supported comprehension), whereas Paraphrasers tended to
which readers were periodically prompted to make any con- simply reiterate the last sentence they read. Perhaps causal
nection between what they were currently reading and what questioning, which requires the reader to make specific text-
happened earlier in the text (by asking, “How does this sen- based connections, helps Elaborators focus on important in-
tence connect to an earlier part of the text?”). We hypoth- formation within the text. In contrast, general questioning,
esized that making causal connections is particularly im- which prompts the reader to make any text-based connection,
portant to constructing a coherent representation of text (as may help Paraphrasers think about the text beyond the current
opposed to making any kind of text-based connection). We sentence. This interpretation suggests that cognitive charac-
were also interested in practical implications: causal ques- teristics of struggling readers have instructional relevance.
tioning has a strong theoretical foundation, but it is rather However, the findings must be interpreted cautiously, given
labor intensive to develop good causal questions. General that analyses were conducted post hoc and subgroups were
questioning, in contrast, is easily applied to any text, because not randomly assigned to questioning condition. Replication
it simply repeats the same question. of these findings is necessary.
22 MCMASTER, ESPIN, AND VAN DEN BROEK: MAKING CONNECTIONS

Most recently, we have examined ways to provide a further research is needed to better understand the instruc-
more intensive intervention, using the Causal and General tional conditions under which questioning approaches are
questioning approaches in a small-group format. In a recent most beneficial, and whether the effects of instruction are
investigation (McMaster et al., unpublished data), we iden- moderated by reader characteristics, including subgroup sta-
tified fourth-graders struggling with comprehension, placed tus. Research is also needed to determine the developmental
them in groups of three to five students, and assigned groups nature of the subgroups (e.g., do younger and older strug-
randomly to Causal or General questioning. Tutors provided gling readers fit similar profiles and respond to intervention
30 minutes of instruction, three times per week for 18 ses- in similar ways?). Preliminary research has revealed similar
sions, consisting of (1) previewing a story and preteaching subgroups in 7th and 9th grades (van den Broek et al., 2005),
key vocabulary, and (2) reading the story with Causal or but requires replication.
General questions embedded. These questions remain central to our current work, given
In contrast to the previous findings, there were no reli- our interest in finding ways to meet the individual needs
able main effects of question type on students’ text recalls or of struggling comprehenders. Complicating this issue is a
oral reading fluency, although an effect size of d = .88 for practical matter: Teachers have emphasized that small-group
the proportion of gist story units recalled favored the causal intervention is more feasible than one-to-one intervention in
questioning approach (McMaster et al., unpublished data). schools (McMaster et al., unpublished data). Yet, as indi-
Further, there were no reliable interactions between ques- cated above, small groups will likely comprise readers who
tioning approach and reader subgroups. Failure to replicate struggle with comprehension in different ways with different
the earlier findings may have been due, at least in part, to types of texts. Thus, it is important to find ways to be respon-
changes in the structure of the intervention. In the previous sive both to the group as a whole and to individual students.
study, the intervention was structured as a classwide, peer- Using the causal questioning framework, our understanding
mediated approach, in which stronger readers were paired of the way Paraphrasers and Elaborators process text, and
with weaker readers to conduct the reading and questioning extensive observations of how readers respond to question-
activities, whereas the current intervention was conducted in ing during intervention, we are exploring ways to provide
small groups with adult tutors. This change in structure was systematic corrective feedback that is both responsive to in-
part of an effort to intensify the intervention for struggling dividual readers and engages the entire group.
readers (by providing an adult tutor), while maintaining the Another remaining question is whether causal question-
feasibility of the intervention (by delivering instruction in ing, which has shown promise when used with stories, can be
small groups rather than one-to-one). applied to other types of text. In fourth grade, current stan-
The use of small groups inevitably led to combinations dards indicate that students should be able to read and under-
of Paraphrasers and Elaborators within each group, which stand a wide variety of challenging informational texts (e.g.,
likely changed the dynamic of the intervention. For example, history/social studies, science, and technical texts). Texts that
students likely heard a range of paraphrasing and elaborative have a narrative structure, such as historical or biographical
responses to questions from their peers, as well as the cor- texts, are most likely to align well with a causal question-
responding feedback from the tutor, which could have had ing approach. Thus, we are currently examining whether
an impact on the way they processed the text. These changes causal questioning shows promise with biographical texts as
in structure ultimately prevented a true replication of the well as with stories, and whether readers’ responsiveness to
previous study. In addition, it was impossible to randomly intervention using stories versus biographical texts varies by
assign Paraphrasers and Elaborators to condition, given lo- subgroup status.
gistical constraints in schools. Thus, a true experimental test If subgroups identified through think-aloud methods have
of the instructional relevance of the subgroups remains to be instructional relevance, another important goal will be to
conducted. find a more feasible way to identify them. Think-aloud pro-
tocols require intensive training and time to code reliably, and
are not practical for educators to score in a timely manner
SUMMARY AND FUTURE DIRECTIONS in schools. Researchers could examine whether a less fine-
grained coding approach would yield similar results. Further
Findings from this work support the following conclu- research may shed light on other characteristics that distin-
sions: First, the nature of struggling comprehenders’ diffi- guish the two subgroups, which may lead to more practical
culties may be revealed through the processes in which they identification approaches.
engage during reading (Rapp et al., 2007). Specifically, think-
aloud methods consistently reveal two subgroups of strug-
gling comprehenders: one characterized by frequent para- IMPLICATIONS FOR PRACTICE
phrasing and text repetition, and the other by heavy reliance
on elaborative inferences. Second, subgroup membership ap- Whereas many questions remain as to how to best promote
pears to be relatively stable when either narrative or informa- struggling readers’ generation of coherent representations of
tional texts are used to identify them; yet, for a small number different types of texts, the following implications of this
of students, subgroup classification appears to vary based on work are relevant to practice. First, we have observed that
text type. Third, subgroups of struggling comprehenders may many students who struggle with comprehension are able
have instructional relevance, in that questioning approaches to make connections needed to generate coherent representa-
may be differentially effective for the two groups. However, tions of text, but often require support to do so. Second, not all
LEARNING DISABILITIES RESEARCH 23

struggling readers experience comprehension difficulties in grants R305G040021 and R324A110046 to the University
the same ways, and these differences may have instructional of Minnesota. The opinions expressed are those of the au-
implications. Teachers should be aware of possible differ- thors and do not represent views of the Institute or the U.S.
ences and be responsive to the needs of individual students. Department of Education.
To address these two issues, teachers can provide support
by explicitly directing students to attend to important, highly
connected parts of the text they are reading, and by providing
explicit feedback that is responsive to individual students’ REFERENCES
ways of processing text. We have conceptualized such sup-
port as a causal questioning approach that includes: Berkeley, S., Scruggs, T. E., & Mastropieri, M. (2010). Reading compre-
hension instruction for students with learning disabilities, 1995–2006:
1 Brief preteaching of vocabulary and concepts that are A meta-analysis. Remedial and Special Education, 31, 423–436.
critical to understanding the text. Best, R. M., Floyd, R. G., & McNamara, D. S. (2008). Differential com-
petencies contributing to children’s comprehension of narrative and
2 Questions embedded in text that prompt the reader to expository texts. Reading Psychology, 29, 137–164.
make connections to the most important parts of what Cain, K., & Oakhill, J. (2006). Profiles of children with specific reading
they have read so far, and explicit modeling of how comprehension difficulties. British Journal of Educational Psychology,
to look back at the text to make good connections. 76, 683–696.
We use very specific “Why” or “How” questions; Chall, J. S., Jacobs, V. A., & Baldwin, L. E. (1990). The reading crisis: Why
poor children fall behind. Cambridge, MA: Harvard University Press.
for example, “Why did Sammy decide to earn some Edmonds, M., Vaughn, S., Hjelm, J., Reutebuch, C., Cable, A., & Tackett,
money?” K. (2009). A synthesis of reading interventions and effects on reading
3 Explicit feedback that is responsive to individual stu- outcomes for struggling readers. Review of Educational Research, 79,
dents. For example, a student who tends to para- 266–300.
Ericsson, K. A., & Simon, H. A. (1993). Protocol analysis: Verbal reports
phrase might say, “Because he needed some money,” as data. Cambridge, MA: MIT Press.
to which the teacher might respond, “You’re on the Faggella-Luby, M. N., & Deshler, D. D. (2008). Reading comprehension in
right track, but why did Sammy need some money?” adolescents with LD: What we know; what we need to learn. Learning
Or, a student who tends to elaborate might say, “Be- Disabilities Research & Practice, 23, 70–78.
cause he wanted to be really, really rich and then he’d Fuchs, L. S., & Fuchs, D. (1992). Identifying a measure for monitoring
student reading progress. School Psychology Review, 21(1), 45–58.
be really, really happy,” to which the teacher might Gajria, M., Jitendra, A. K., Sood, S., & Sacks, G. (2007). Improving compre-
respond, “That may be true, but what does the text say hension of expository text in students with LD: A research synthesis.
about why Sammy decided to earn some money?” Journal of Learning Disabilities, 40, 210–225.
Gersten, R., Fuchs, L. S., Williams, J. P., & Baker, S. (2001). Teaching
Our goal is to continue development and evaluation of reading comprehension strategies to students with learning disabilities:
A review of research. Review of Educational Research, 71, 279–320.
this intervention approach in order to provide guidance to Gough, P. B., & Tunmer, W. E. (1986). Decoding, reading, and reading
educators who seek to support struggling comprehenders. disability. Remedial and Special Education, 7, 6–10.
Kendeou, P., van den Broek, P., Helder, A., & Karlsson, J. (2014). A cognitive
model of reading comprehension: Implications for reading difficulties.
CONCLUSION Learning Disabilities Research and Practice, 29, 10–16.
Kintsch W., & van Dijk, T. A. (1978). Toward a model of text comprehension
and production. Psychological Review, 85, 363–394.
The work described in this article illustrates the importance MacGinitie, W. H., MacGinitie, R. K., Maria, K. & Dreyer, L. G. (2002).
of connecting cognitive theory to reading intervention. Cog- Gates-MacGinitie Reading Tests Fourth Edition: Technical Report.
nitive theory provides insights and methods to examine the Itasca, IL: Riverside Publishing.
ways that struggling readers process text and the extent to McMaster, K. L., van den Broek, P., Espin, C. A., White, M. J., Kendeou,
which they are able to create a coherent mental representation P., Rapp, D. N., et al. (2012). Making the right connections: Differ-
ential effects of reading intervention for subgroups of comprehenders.
of the text. Drawing on these insights, we have undertaken Learning and Individual Differences, 22, 100–111.
a program of research to systematically identify these pro- McNamara, D. S., & Kendeou, P. (2011). Translating advances in reading
cesses and determine how reader characteristics interact with comprehension research to educational practice. International Elec-
text properties and instructional contexts, with the ultimate tronic Journal of Elementary Education, 4, 33–46.
Nation, K., Clarke, P., & Snowling, M. J. (2002). General cognitive ability
goal of identifying conditions under which struggling com- in children with reading comprehension difficulties. British Journal of
prehenders are likely to benefit from intervention. It is our Educational Psychology, 72(4), 549–560.
hope that findings of this work may contribute to knowledge Nation, K., & Snowling, M. J. (1998). Semantic processing and the de-
of how to improve outcomes for struggling readers, includ- velopment of word recognition skills: Evidence from children with
ing those with LD. Further, because instructional application reading comprehension difficulties. Journal of Memory and Language,
39, 85–101.
provides a strong test of cognitive models of reading compre- Perfetti, C. A. (2007). Reading ability: Lexical quality to comprehension.
hension, our hope is that findings from intervention research Scientific Studies of Reading, 11, 357–383.
will serve to inform and refine cognitive theory. Pressley, M., Graham, S., & Harris, K. (2006). The state of educational in-
tervention research as viewed through the lens of literacy intervention.
British Journal of Educational Psychology, 76(1), 1–19.
Acknowledgments RAND Reading Study Group. (2002). Reading for understanding; To-
ward an R&D program in reading comprehension. Santa Monica, CA:
RAND.
The research reported here was supported by the Institute of Rapp, D. N., van den Broek, P., McMaster, K. L., Kendeou, P., & Espin, C. A.
Education Sciences, U.S. Department of Education, through (2007). Higher-order comprehension processes in struggling readers: A
24 MCMASTER, ESPIN, AND VAN DEN BROEK: MAKING CONNECTIONS

perspective for research and intervention. Scientific Studies of Reading, van den Broek, P., & Espin, C. A. (2012). Connecting cognitive theory and
11, 289–312. assessment: Measuring individual differences in reading comprehen-
Rayner, K., Chace, K. H., Slattery, T. J., & Ashby, J. (2006). Eye movements sion. School Psychology Review, 41, 315–325.
as reflections of comprehension processes in reading. Scientific Studies van den Broek, P., Fletcher, C. R., & Risden, K. (1993). Investigations
of Reading, 10, 241–255. of inferential processes in reading: A theoretical and methodological
Stanovich, K. E. (1984). The interactive-compensatory model of reading: A integration. Discourse Processes, 16, 169–180.
confluence of developmental, experimental, and educational psychol- van den Broek, P. W., Helder, A., & Van Leijenhorst, L. (2013). Sensitivity
ogy. Remedial and Special Education, 5, 11–19. to structural centrality: Developmental and individual differences in
Swanson, H. L. (1999). Reading research for students with learning disabil- reading comprehension skills. In M. A. Britt, S. R. Goldman, & J. F.
ities: A meta-analysis of intervention outcomes. Journal of Learning Rouet (Eds.), Reading: From words to multiple texts (pp. 132–146).
Disabilities, 32, 50–532. New York: Routledge.
Sweet, A. P., & Snow, C. E. (Eds.). (2003). Rethinking reading comprehen- van den Broek, P., & Kremer, K. E. (1999). The mind in action: What it
sion. New York: Guilford. means to comprehend during reading. In B. Taylor, M. Graves, & P.
Trabasso, T., Secco, T., & van den Broek, P. (1984). Causal cohesion and van den Broek, (Eds.), Reading for meaning (pp. 1–31). New York:
story coherence. In H. Mandl, N. L. Stein, & T. Trabasso (Eds.), Teacher’s College Press.
Learning and comprehension of text (pp. 83–111). Hillsdale, NJ: van den Broek, P., McMaster, K., Rapp, D. N., Kendeou, P., Espin, C., &
Erlbaum. Deno, S. (2006, June). Connecting cognitive science and educational
Trabasso, T., & van den Broek, P. (1985). Causal thinking and the repre- practice to improve reading comprehension. Washington, DC: Paper
sentation of narrative events. Journal of Memory and Language, 24, presented at the Institute of Education Sciences Research Conference.
595–611. van den Broek, P., White, M. J., Kendeou, P., & Carlson, S. (2009). Read-
van den Broek, P., Bohn-Gettler, C., Kendeou, P., Carlson, S., & White, M. J. ing between the lines: Developmental and individual differences in
(2011). When a reader meets a text: The role of standards of coherence cognitive processes in reading comprehension. In R. K. Wagner, C.
in reading comprehension. In M. T. McCrudden, J. P. Magliano, & G. Schatschneider, & C. Phythian-Sence (Eds.), Beyond decoding: The
Schraw (Eds.), Text relevance and learning from text (pp. 123–140). behavioral and biological foundations of reading comprehension (pp.
Greenwich, CT: Information Age Publishing. 107–123). New York: Guilford.
van den Broek, P., & Espin, C. A. (2009). Improving reading comprehension: Williams, J. P. (1993). Comprehension of students with and without learning
Connecting cognitive science and education. Cognitive Critique, 2, disabilities: Identification of narrative themes and idiosyncratic text
1–25. representations. Journal of Educational Psychology, 85, 631–641.

About the Authors

Kristen L. McMaster, Ph.D., is an Associate Professor of Special Education in the Department of Educational Psychology,
University of Minnesota. She received her Ph.D. in special education from Vanderbilt University. Her research interests include
(1) promoting teachers’ use of data-based instruction and (2) developing individualized interventions for students for whom
generally effective instruction is not sufficient.

Christine Espin, Ph.D. (University of Minnesota), is a Professor at Leiden University in the department of Education and
Child Studies, Special Education. Her research focuses on the development of progress-monitoring systems in reading, written
expression, language learning, and content-area learning for secondary students with learning disabilities, and on teachers’ use
of data for educational decision-making.

Paul van den Broek, Ph.D. (University of Chicago), is Director of the Brain and Education Laboratory at Leiden University
and holds faculty appointments in the Department of Education and Child Studies at Leiden University and the Center for
Cognitive Sciences at the University of Minnesota. His research focuses on the cognitive and neurological processes involved
in reading comprehension in children and adults, and on the development of instructional methods for proficient and struggling
readers.

View publication stats

You might also like