 

v v .,
.3

 

 

 

 

1 Inn-1%..uﬂli...‘
3w, . 3a
. . , z a...
'l

This is to certify that the i
dissertation entitled '

EFFECTS OF TEXT LEVEL AND PRESENTATION FORMAT
ON INFERENCE-MAKING PROCESSES WITH NARRATIVE
TEXTS

presented by
CAROLYN AMANDA JAYNES

has been accepted towards fulﬁllment
of the requirements for the

degree In Educational Psychologx

598“ 77777 7);

Major Profess r’ 3 Signature

777M / ,, 7007

Date

 

 

MSU is an Aﬁinnative Action/Equal Opportunity Institution

 

'WLI .
Michigan State
University

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

 

 

DATE DUE DATE DUE DATE DUE
EQUG 3 62 00‘
4*; f: ,5? f;

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/CIRC/DaleDue.p65-p.15

EFFECTS OF TEXT LEVEL AND PRESENTATION FORMAT ON INFERENCE-
MAKING PROCESSES WITH NARRATIVE TEXTS

By

Carolyn Amanda Jaynes

A DISSERTATION
Submitted to
Michigan State University
in partial ﬁilﬁllment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Department of Counseling, Educational Psychology, and Special Education

2004

ABSTRACT

EFFECTS OF TEXT LEVEL AND PRESENTATION FORMAT ON INFERENCE-
MAKING PROCESSES WITH NARRATIVE TEXTS

By
Carolyn Amanda Jaynes
This study examined the relationship between text difﬁculty, mode of text
presentation, and inference-making, and was guided by the following questions: What
inference-making processes do students engage in as a narrative text varies according to
level of difficulty and presentation format? What types of inferences do students make
and how frequently are they generated? The project involved 27 fourth and ﬁfth graders
ﬁ'om the San Francisco Bay Area whose comprehension was at the third grade
instructional level. These students participated in two 3-part, one-on-one sessions. In the
ﬁrst session, students provided verbal reports of their thinking as they attempted to
comprehend three different narrative texts at their instructional reading level. One text
was presented in audio form, another text was presented in print form, and a third text
was presented in video form. The text presentation format as well as the sequence of
texts and presentation formats varied by student and were counterbalanced to assure that
there were comparable numbers of students who experienced the same text-format match,
sequence of texts, and sequence of presentation formats. In the second session, students
repeated the process with three texts at two grade levels above their instructional reading
level. Students’ verbal protocols were segmented into lists of propositions that were
categorized according to the types of cognitive processes they represented. Analyses
focused on the types of inferences students made, and how frequently they were

generated. While ﬁndings suggest that format effects on inference-making were

signiﬁcant and text level effects were not, a format by level interaction effect complicates
these ﬁndings. Most notably, it appears that individual differences play an important
role, especially since the inﬂuence of story, format, sequence, and story-format
combinations was not signiﬁcant. That is, determining which condition (print, audio, or
video) was more helpful or problematic, overall, was an impossible task. For example,
while some students appeared to have an easier time generating inferences within the
audio context, others clearly struggled. Similarly, ﬁndings associated with recall-level
data also resulted in an interaction effect that prompted a closer consideration of
individual differences since there was no signiﬁcant effect due to text level or
presentation format. It is interesting to note that for some students print texts elicited
more recall-level propositions with easier texts, while video texts prompted more recall

with more challenging texts.

To Mom and Dad

iv

ACKNOWLEDGMENTS

This dissertation is the ﬁuit of many hands. I thank my mother for
communicating her love and conﬁdence in me in ways that are clearly felt at my core. I
thank my father who honored my choice of profession. I thank Jeff, Cathy, Marvie, Tim,
Doug, Tinina, and all the kids, for their love and encouragement. I thank my ﬁ’iends—
those in California who surrounded me with support from afar, as well as my Michigan
pals—Cheryl, Gina, Lisa, Mark, Michael, Penelope, Shane, Stacey, Stephanie, Suzy, and
Tanna. On a bumper sticker 1 once read that “friends don’t let friends go to grad school,”
however, I smile and know that graduate school brought me treasured friendships. I
thank my committee members—Bob, Laura, Mark, and Nell for their insight, nudging,
and mentorship. I thank my chair, David, whose unfailing support and guidance gave me
the space and opportunity to do good work. I thank Taylor Bones, a girl’s best friend. I
thank the 27 students, along with their parents and teachers, who let me talk to them
about what they were understanding so I could learn something——may I use it well. And
I thank Grayson and Maria who inspired my questions. Finally, I acknowledge with all

conﬁdence that my Redeemer lives, and I thank Him above all.

TABLE OF CONTENTS

LIST OF TABLES .............................................................................. viii

LIST OF FIGURES ................................................................................ x
INTRODUCTION ........................................................................................................... 1
Limited Capacity Models ............................................................................................. 2
Reading as Thinking Models ....................................................................................... 3
Schema Theory ........................................................................................................ 4
Construction-Integration Model ............................................................................... 5
Versions of the Simple View of Reading ..................................................................... 6
Models of Reading in the Classroom ........................................................................... 8
Investigating Inference Generation ............................................................................ 12
Inference Generation across Presentation Formats: Audio, Print, and Video .............. l4
Inference Generation across Text Difficulty Levels ................................................... 17
METHODS ................................................................................................................... 20
Data Collection .......................................................................................................... 20
Research Sites and Participants .............................................................................. 20
Elementary School Sites .................................................................................... 20
Students ............................................................................................................. 22
Parents ............................................................................................................... 23
Materials ............................................................................................................... 24
Think Aloud Training Materials ......................................................................... 24
Research Session Texts ...................................................................................... 24
Procedures ............................................................................................................. 28
Think Aloud Training ........................................................................................ 28
Research Sessions .............................................................................................. 28
Data Preparation ........................................................................................................ 33
Segmenting Verbal Protocols into Propositions ...................................................... 33
Coding Propositions .............................................................................................. 35
Coding Propositions into General Categories ..................................................... 36
Coding Inference-level Propositions into Subcategories ..................................... 40
Data Analysis ........................................................................................................ 46
RESULTS ..................................................................................................................... 47
Effects of Text Level and Presentation Format on Inference-Making ......................... 47
Illustrating Predominant Patterns ........................................................................... 49
Predominant Patterns with Print and Video ........................................................ 51
Pattern 1: Print low, video high ..................................................................... 51
Pattern 2: Print high, video low. ..................................................................... 53
Predominant Patterns with Audio Text: Feast or Famine ................................... 54
Pattern 3: Audio as an inference suppressing condition. ................................. 57

vi

Pattern 4: Audio as an inference enhancing condition ..................................... 58

Examining Text Level More Careﬁilly ................................................................... 58
Effects of Text Level and Presentation Format on Various Sets of Inferences ........ 59
Effects of Text Level and Presentation Format on Six Inference Subcategories .. 63
Association-level inferences with the audio format ........................................ 64
Predictions with the audio format. .................................................................. 65

Analyses of Effects of Story, Order, & Story-Format Combinations ...................... 65
Effect of Story on Inference-Making .................................................................. 66

Effect of Story Order on Inference-Making. ....................................................... 66

Effect of Presentation Format Order on Inference-Making ................................. 67

Effect of Story-Format Combinations on Inference-Making ............................... 68
Analyses of Recall-Level Data ................................................................................... 7O
LIMITATIONS ............................................................................................................. 72
DISCUSSION ............................................................................................................... 74
The Inﬂuence of Presentation Format ........................................................................ 74
The Inﬂuence of Text Level ...................................................................................... 80
Implications ............................................................................................................... 82

vii

LIST OF TABLES

Table 2.]: School Information ..................................................................................... 2]
Table 2.2: Texts Used in Research Sessions ................................................................. 25
Table 2.3: Examples of Story Transitions .................................................................... 27
Table 2.4: Sample of Session Schedule ........................................................................ 29
Table 2.5: Inference Questions .................................................................................... 32
Table 2.6: Example of Protocol Segmentation ............................................................. 34
Table 2.7: Average number of student-generated propositions for each story ............... 35
Table 2.8: General Proposition Categories ................................................................... 37
Table 2.9: Examples of Inaccurate and Implausible Propositions ................................. 38
Table 2.10: Recall and Inference-Level Propositions, Accuracy and Plausibility ........... 40
Table 2.11: General Propositions Sums and Means (Excluding Inaccurate/Implausible

Propositions) ................................................................................................................. 40
Table 2.12: Inference Subcategories Deﬁnitions and Examples .................................... 41
Table 2.13: Inference Subcategories Sums and Means .................................................. 45
Table 3.1: Total Inferences Based on Text Level and Presentation Format ................... 47
Table 3.2: Patterns of Inference-Making ...................................................................... 50
Table 3.3: Pattern 1: Print Low, Video High ................................................................ 52
Table 3.4: Pattern 2: Print High, Video Low ................................................................ 54
Table 3:5: Pattern 3: Audio Low ............................................................... 55
Table 3.6: Six Categories of Inferences ........................................................................ 60
Table 3.7: Text-Connected and Self-Connected Inferences .......................................... 61
Table 3.8: Spontaneous and Prompted Inferences ........................................................ 62

viii

Table 3.9: Inferences by Story ..................................................................................... 66

Table 3.10: Inferences by Order of Story ..................................................................... 67
Table 3.11: Inferences by Order of Presentation Format ............................................... 68
Table 3.12: Inferences by Story-Format Combinations ................................................. 69

Table 3.13: Within-Subjects Analysis of Recall Based on Text Level and Presentation
Format ........................................................................................................................... 70

ix

LIST OF FIGURES

Figure 3.1: Discrepancies Based on the Overall Trend of Increases from Print to Audio
to Video ........................................................................................................................ 56

INTRODUCTION

Perhaps the entire range of plausible models of the reading process that have been
put forward by scholars of reading can be observed by noting the individual differences
encountered when working with children on a daily basis. Matching one’s observation of
a child’s reading process with a particular model may be relatively easy—so long as the
observation remains within a single slice of time and the range of activity does not vary
too much. For example, consider which models of reading may be supported by the
stories of Maria, age 11, and Grayson, age 10.

Maria ’s Story

Maria is a sixth grader who is riveted to the television each week
day at four o ’clock to watch her favorite animated series Pokemon.
Maria is equally enthralled with the Harry Potter novels that her
mother reads to her each night. It is not unusual for Maria to talk
at length about the various characters and storylines portrayed in
these programs and novels, and oftentimes, she is able to generate
rather astute inferences about characters ’ motives and causal
relationships. When I observed Maria reading to herself however,
she decoded at the third grade level, and demonstrated limited
understanding. Was the effort expended on decoding encroaching
on her ability to infer and construct meaning?

Grayson ’s Story

During a tutoring session earlier that week Grayson had clearly
struggled to make inferences about a text he had just read. So
later on, perhaps it should have been expected when he asked
several inference-based questions about the plot of the movie he
was watching. But didn ’t this viewing context ﬁee Grayson up to
focus on constructing meaning ﬁom the audio-visual ‘text, ’ a ﬁlm
entitled The Rookie? Since decoding was not an issue, and the
vocabulary was not terribly sophisticated would not his ability to
make inferences improve?

One cannot attempt to explain stories like Maria’s and Grayson’s without

revealing an inclination toward particular models of reading, or at least, certain elements

within those models. While some models present reading as a sequence of processes one
develops over time and experience, with higher level cognitive activities building upon a
foundation of lower level processes, other models de-emphasize sequence in favor of the
interaction of those processes at every stage of development. Although models in this
second general category may deem some processes more inﬂuential or integral, the
hierarchy is not sequential, but rather, based on what accounts for more of the variance in
reading ability. In the following section a range of reading models are discussed in

attempting to explain the observations of Grayson and Maria.

Limited Capacity Models

Some theories of reading development suggest that one’s attentional resources
have a limited capacity, and that there exists a hierarchy of processes from lower order
perceptual and decoding processes to higher order comprehension processes (LaBerge &
Samuels, 1974; Thurlow & van den Broek, 1997). Proponents of this view assert that
text comprehension involves two levels of information processing. The ﬁrst level
involves the technical process of decoding and word recognition, while the second level
entails the comprehension process of ‘creating a coherent representation of meaning of
the text’ (Beishuizen, Grand, & Schalk, 1999). Some researchers suggest that the
cognitive ﬂexibility required to balance the activities at these levels and manage more
than one aspect of the reading task at a time is something that develops over time
(Cartwright, 2002). This same line of theoretical work asserts that as readers begin to
perform lower order processes more automatically, they can allocate more attention to

comprehension processes (LaBerge & Samuels, 1974).

It appears that this limited capacity model of reading helps to explain Maria’s
story. If, indeed, it is the case that an individual is more capable of attending to higher
order processes when the lower order processes are more automatic, or even done for the
individual, it would follow that the individual would be more proﬁcient at engaging in
higher level processes like making inferences when decoding is not required because a
story is presented in either video form, as with a television program, or in audio form, as
when text is read aloud by a proﬁcient reader. In this example, it appears that Maria had
an easier time generating inferences within the viewing and audio contexts (watching the
Pokemon television program, and listening to her mother read aloud) where decoding
was not a factor, as opposed to the independent reading task where decoding was
required. Maria’s story, and others like it, illustrate the notion underlying Laberge and
Samuel’s limited capacity model, that readers possess a ﬁnite amount of cognitive capital
available for higher level processes. Meanwhile, this model does not help to explain the
story of Grayson who appeared to struggle with inference-making regardless of whether
or not any of his cognitive resources were occupied by lower level processes such as

decoding. An alternative model must be used to account for this observation.

Reading as Thinking Models
In contrast to these sequential, limited capacity models of reading, other theorists
view reading more as a general cognitive processes. As early as 1917, E. L. Thorndike
deﬁned reading as reasoning and prompted investigations of reading as a thinking process

guided by symbols (Powell, 1989; Thorndike, 1917). With its roots in the early 20th

century, this idea of reading as a reﬂective, purposeful, and constructive thinking process
is what Paris calls the ‘new-old’ interactive view of reading as thinking (Paris, 1991).

In general, it may be that a reading as thinking model would best explain
Grayson’s story. Ifreading and thinking processes are comparable, then it would be
reasonable to predict that an individual would experience comparable levels of ease or
challenge with higher level processes across a range of contexts. For Grayson,
generating inferences was equally challenging when he was reading a book and watching
a movie. In this instance, the view of reading as thinking makes sense. For Maria,
however, this overall perspective does not seem to ﬁt as well. Nevertheless, some views
of reading that fall within this general reading as thinking category might help to explain

elements of both Maria’s and Grayson’s stories.

Schema Theory
For example, schema theory proposes that comprehension involves the interaction

of new information with knowledge already stored in memory as the reader’s schemata
(Anderson & Pearson, 1984). This model suggests that in the act of comprehending, the
reader must either ﬁnd a mental home for information in the text or modify an existing
mental home in order to accommodate new information (p. 255). These mental homes,
or schemas, inﬂuence three basic processes in comprehension: 1) determining which
inferences the reader makes during comprehension as well as when these inferences are
made, 2) determining how the reader allocates attention to various elements in the text,
and 3) determining what the reader remembers about the text. (Anderson & Pearson,

1984)

In Maria’s case, then, it could be that the new information from the independent
reading text was more difﬁcult to accommodate, simply because it was less familiar than
the information from the Pokemon program and the Harry Potter novels that were very
familiar to Kayla who was exposed to these storylines on a regular basis. It may be that
the mode of text presentation did not account for the difference, but simply that, for
Maria, the Pokemon and Harry Potter content had an easier time ﬁnding a mental home.
And in Grayson’s case, it could be that the new information presented by both the novel
and the ﬁlm were equally difﬁcult to accommodate given Grayson’s available schemata.
It may be that what made inference-generation equally challenging across these media
contexts was more closely tied to the content within each media context than to the

possibility that inference-making is a challenge for him in general.

Construction-Integrgtion Model

Similarly, Kintsch’s construction-integration model can be used to explain parts
of both Maria’s and Grayson’s stories. This model combines a construction process in
which a text base is constructed ﬁom linguistic input as well as from the comprehender’s
knowledge base, with an integration phase in which this text base is integrated into a
coherent whole, which Kintsch calls the “construction-integration model” (Kintsch,
1994). In this model, word identiﬁcation processes are embedded within the process of
discourse understanding, and the lexical node serves as just one entry point into the
readers’ prior knowledge and experience (p. 973). Merging these construction and
integration phases, rather than assigning them a particular sequence or level, makes it

possible to use this model to explain both Maria’s and Grayson’s stories. As with schema

theory, apparent differences or consistencies in inference-generation across media may be
attributed to background knowledge and experience, and it is not clear that text
presentation format would either facilitate or hinder these processes.

While one version of the reading as thinking model is supported by Grayson’s
story, which implies that higher level processes like inference-making occur similarly
regardless of the need for cognitive activity at the decoding level, other reading as
thinking models are supported also by Maria’s story. In the section that follows a third

set of views can be applied to both readers as well.

Versions of the Simple View of Reading

Models related to Gough and Tunmer’s (1986) Simple View of Reading are
difficult to classify as either limited-capacity or reading as thinking models. The Simple
View of Reading asserts that an individual’s reading comprehension ability (R) is a
function of the product of two factors: decoding (D) and language or listening
comprehension (L), or R = D x L (Gough & Tunmer, 1986). This model suggests that if
an individual struggles with either decoding or general language comprehension, reading
comprehension will suffer. And according to Gough and Tunmer, individuals who
struggle with both components are considered to be “garden-variety” poor readers.

This simple view of reading can be used to explain elements of both Grayson’s
and Maria’s stories. For Grayson, who appeared to experience similar challenges with
inference-making whether he was required to decode a printed text or to comprehend the
audio and visual content presented in a film, this model might identify him as a “garden

variety” poor reader who struggles with both decoding and listening comprehension. Or,

it could be that he has difﬁculties with general language comprehension, so whether or
not decoding is required, comprehension is limited. For Maria, who appeared to generate
more inferences with content presented in audio and video forms than with content
presented in print, it might be said that the audio and video contexts enhanced her
language comprehension which contributed to her ability to engage in more higher-level
processes.

Finally, Chen and Vellutino (1997) offer a version of the original Simple View of
Reading that includes a linear combination, decoding plus listening comprehension (D +
L), which they say explains more of the substantive variance in comprehension ability.
This linear combination is added to Gough and Tunmer’s multiplicative combination,
decoding times listening comprehension, which Chen and Vellutino claim accounts for
more of the unique variance in comprehension ability. In other words, with the Chen and
Vellutino model (R = (D + L) + (D x L)) the linear term, decoding plus listening
comprehension, is able to identify the main effects, while the multiplicative term,
decoding times listening comprehension, is able to determine the interaction effects.
Therefore, this version of the model helps to explain those occasions when the interaction
term is signiﬁcant but none of the main effects is signiﬁcant (Chen & Vellutino, 1997).

Although the limited nature of both Grayson’s and Maria’s observations prevents
an analysis of the levels of effects addressed in Chen and Vellutino’s model, it remains
evident that both versions of the Simple View of Reading can be used to explain elements
of both students’ stories. And while these versions of the Simple View of Reading model
combine both decoding and language comprehension, they do not suggest that these

factors are limited in nature, or that the use of one is necessarily at the expense of the

other. Nor do they imply a particular sequence of development. Instead, these models
propose the notion that less skilled readers may be struggling due to a) nonautomated
lower level processing, b) deﬁcits in language comprehension, or c) a combination of the
two (Curtis, 1980). The apparent range of possible factors inﬂuencing one’s ability to
construct meaning underscores the importance of considering what models best explain
comprehension. It must be acknowledged that decisions related to comprehension
instruction, curriculum, assessment, and intervention are grounded in models of reading

to which teachers, administrators, and policy-makers subscribe—consciously or not.

Models of Reading in the Classroom

Attempts to map observations of students reading onto existing models of reading
highlight the distinctions among a range of well-respected theories reﬂected in current
policy and practice. While most models explain the inﬂuence of context, the role of
background knowledge and prior experience with text, as well as the effect of the
reader’s attention and self-monitoring, two apparent differences among these models are
related to sequence and cognitive capacity. In today’s educational climate, much of the
current reading policy reﬂects an inclination toward a more sequential, limited capacity
model. For example, many early reading interventions imply that the initial emphasis
should be on making sure that students can decode, presumably because once this is done
more automatically, they can move on to improving comprehension (AutoSkill, 2003;
Waterford, 2000). This implicit message is conveyed by measures used to determine the
success of such interventions in which oral reading accuracy and rate scores outweigh

comprehension scores. For example, in the San Francisco bay area, some low-

performing schools have implemented an assessment system called RESULTS that
determines students’ instructional reading level with three equally weighted measures:
oral reading accuracy; oral reading ﬂuency (i.e., rate only); and comprehension, as
determined by responses to three vague questions (CRLP, 1996). The temptation, then, is
to shift instructional focus away from the thinly measured meaning-level component to
the more skills-based components in order to get students up to ‘grade level.’

Concerns about how limited capacity models get instantiated in commercial
reading programs and then played out in schools, particularly with students who are
reading below grade level expectations, underlie the present study, which was designed to
add some insight into contrasting models of reading through an examination of struggling
readers’ comprehension processes within a range of media contexts.

First, by looking at how weaker readers engage in the comprehension process of
inference-making with narrative texts presented in three formats—text, audio, and
video—this study takes a closer look at this concept of “limited cognitive capacity.”
Admittedly, limited capacity models of reading are not designed to address
comprehension beyond the print context, however, this study questions the general notion
that individuals possess ﬁnite cognitive resources that are employed at one another’s
expense. That is, if it is true that individuals have a limited capacity for engaging in
various levels of cognitive processing, texts that vary according to difﬁculty level and
presentation format might help or hinder one’s ability to engage in higher level activities.
On the one hand, it might be expected that students would be able to engage in more
inference-making processes with texts at their instructional reading level, and with audio

and video text formats that free up some of the cognitive load by alleviating the student’s

need to focus on decoding. On the other hand, it might be that comprehending audio
texts, which “go in one ear and out the other,” may be more challenging, for some
students, than comprehending texts that appear in black and white. In other words, it
could be the case that even when decoding is not necessary, comprehending audio texts
may require more of the limited resources than comprehending texts in print. For
instance, the rate of textual input may differ with listening and reading. When an
individual is listening to a text, the rate of textual input is determined by the speaker, or
in this study, by the audiotape. Alternatively, when an individual is reading a text, the
rate is controlled by the reader, and the reader has the option of rereading portions of the
text (Carlisle & Felbinger, 1991). Therefore, it could be that within the audio context,
even when cognitive resources are not being spent on lower level processes like
decoding, the individual’s inability to monitor the speed of input or reread, or at least
rescan, portions of text, however small, may actually increase the cognitive load, leaving
the individual with fewer attentional resources to devote to higher level processes like
inference-making.

Finally, there is the possibility that the video context may actually require more
cognitive resources as an individual synthesizes both audio and visual information. One
could argue that as incoming information increases, the cognitive load intensiﬁes. For
example, audio and video versions of texts contain additional prosodic or visual
information that the reader must integrate. While this additional information could
enhance understanding and increase one’s ability to engage in higher level processes, it

also could be that integrating this supplementary content could be using up the

10

supposedly limited cognitive supply, leaving fewer attentional resources available for
inference generation.

In addition, this study attempts to investigate the usefulness of the reading as
thinking models. If it is the case that reading is thinking, then it would be expected that a
student’s ability to generate inferences would be similar across text levels and
presentation formats. While this study does not control for the potential inﬂuence of
background knowledge and prior experience, it does use six distinct texts with each
student making this less of a concern. That is, with a variety of unique texts, students’
knowledge is likely to be distributed across the set of narratives. This provision
diminishes the possible effect of prior knowledge, and highlights the effects of text level
and presentation format.

Finally, this study has the potential to examine the usefulness of the Simple View
of Reading. If it is true that comprehension is a function of an individual’s decoding
skills and language comprehension, then it would be likely that within the audio and
video contexts or with easier texts, where decoding is less challenging or not even
required, students whose weakness is related to decoding would have an easier time
engaging in higher level comprehension processes. Furthermore, students who can
decode but struggle with language comprehension would be expected to demonstrate
equal levels of challenge across all three formats and with both easy and difﬁcult texts.
Finally, the so-called garden-variety poor readers, those who struggle both with decoding
and language comprehension, would most likely do better within the audio and video
contexts or with easier texts where at least one of their problem skills, decoding, is less

challenging or not even required.

11

In an attempt to shed some light on the utility of these various models of reading,
this study analyzed verbal protocols from 4th and 5th grade students and was guided by
two questions: What inference-making processes do students engage in with narrative
tms varying according to level of dtﬂiculty and presentation format? What types of

inferences do students make and how frequently are they generated?

Investigating Inference Generation

As deﬁned in this study, inferences require an individual to integrate prior
knowledge with new information in an effort to construct coherent meaning (Rickheit,
1985; Beal, 1990; Thurlow, 1997) The importance of investigating the inference-making
processes of upper elementary students is grounded in research demonstrating that as
readers advance beyond the beginning stages, higher level processes account for a more
substantial proportion of the variance in reading comprehension ability than do lower
level word recognition processes (Hannon & Daneman, 2001). In addition, generating
and evaluating inferences is considered to be an essential comprehension-fostering
activity (Brown, Palincsar, & Arrnbruster, 1994). And according to Trabasso and
Magliano, inference-level processes work together with memory to help readers construct
a coherent representation of text (Trabasso & Magliano, 1996).

This study examines not only students’ overall inference-making, but a range of
speciﬁc inference-level processes through verbal protocol analysis. It was anticipated
that students would generate various types of inferences noted in protocols from previous
studies of comprehension. For example, Trabasso and Magliano identiﬁed three

categories of inferences including explanations, associations, and predictions. Although

12

they found that, for the most part, comprehension was most closely linked to explanation-
based inferences that provide the rationale and conditions for various events and
behaviors (Trabasso & Magliano, 1996), the purpose of the present study is to identify
the types of inferences that were generated most often by struggling readers across
different presentation formats and text difficulty levels. Understanding more about
inference generation in struggling readers may have instructional implications. For
example, we may learn that there were certain types of inferences that struggling readers
tend to avoid or gravitate toward within particular comprehension contexts, and these
might be the types of inferences that need to be either reinforced or supplemented.

In addition, this study takes into account the possibility that different types of
inferences call upon different sorts of cognitive activity. For example, Trabasso and
Magliano suggest that association-level inferences are activated by other inferences as
well as the content of the text (T rabasso & Magliano, 1996). Given this ﬁnding, one
reasonable prediction would be that associations would occur with a certain level of
ﬁ'equency due to the fact that that these types of inferences are connected to both content
and other inferential processing. While these sorts of predictions were not formally
tested as part of the research questions, the discussion of the data acknowledges that
categories of inference-making may require different types of cognitive resources and
that there may be occasions when inference-level activity might overlap.

Similarly, the within-subjects design of the study, which allowed for analyses of
individuals’ inference-making across a range of presentation formats and text levels,
permitted an examination of readers’ relative use of certain inference-making processes.

For instance, Trabasso and Magliano’s data suggest that readers may be consistent with

13

their use of one inference strategy over that of others (Trabasso & Magliano, 1996). The
design of this study made it possible to investigate the inference-making processes of an
individual within while encountering six “texts” that varied by presentation medium and
difﬁculty level. These categories of inferences are deﬁned in the Methods chapter.
Finally, in the sections that follow the rationale is presented for choosing to investigate
inference generation across a range of presentation formats and with different levels of

text difﬁculty.

Inference Generation across Presentation Formats: Audio, Print, and Video

While many studies have investigated inference-making as readers interact with
printed text, fewer have examined these processes across a range of media. Furthermore,
the research that has investigated inference-generation with various presentation formats
offers conﬂicting ﬁndings.

Some results indicate that higher level reading processes remain consistent
whether an individual is reading or viewing a text. For example, Neuman’s 1989 study
of 44 third grade students indicated no signiﬁcant differences in retellings of groups who
a) watched two video versions of stories, b) watched a video version and read a picture
book version, and c) read two picture book versions (Neuman, 1989). Likewise, in 1992,
Neuman found no signiﬁcant differences of inference strategy use among 83 ﬁﬁh grade
high and low achieving students who either read or watched brief episodes of stories
(Neuman, 1992).

Other scholars assert that while this consistency may occur with older readers, the

same does not hold true for younger readers. For example, Pezdek and her colleagues

14

(Pezdek et al., 1987) found that reading and television comprehension of non-illustrated
news stories was signiﬁcantly correlated for high school students. These ﬁndings were
contrary to those from their previous study with elementary students. These results
suggest that for older readers, who may well have strong decoding processes under
automatic control, comprehension may be a general cognitive construct that operates
similarly across various types of media. For younger readers format does have an effect.

A third line of work claims that while presentation format does not inﬂuence the
number of inferences, it does affect the bases upon which the inferences are generated.
Meringoff compared 48 children’s apprehensions of a story either read to them from an
illustrated book or presented as a televised ﬁlm. She found that while there were no
systematic differences between groups in the inferences themselves, the children who
watched the televised story relied more on visual content as a basis for inferences
(Meringoff, 1980). In addition, Beagles-Roos and Gat exposed 48 ﬁrst through fourth
grade students to a television and audio story and found that children’s ability to generate
inferences based on actions improved in the television context, while their ability to make
inferences based on verbal sources or knowledge unrelated to the story improved in the
audio context (Beagles-R003 & Gat, 1983).

Finally, a fourth line of work supports the notion that presentation format does
affect students’ ability to provide plausible answers to a higher level comprehension
questions that require inferential reasoning. Pezdek and her colleagues compared
comprehension scores of third and sixth grade students who read an illustrated storybook
and then were presented with either the televised or audio version. Based on students’

responses to sixteen comprehension questions including those requiring inference

15

generation, the researchers found that overall performance was better in the television
and reading conditions than in the audio-only condition (Pezdek, Lehrer, & Simon,
1984). Finally, Ricci and Beal compared comprehension scores of 66 ﬁrst graders who
were exposed to stories in an audio only, audiovisual, or interactive media formats. They
found that students in the audio only group performed more poorly on a set of inference-
level questions (Ricci & Beal, 2002).

While some ﬁndings suggest that presentation format does affect students’
inference-making processes, others suggest just the opposite. And although some results
indicate that inference generation is consistent across media contexts with older readers,
but varies with younger students, other ﬁndings indicate consistency or variation across
presentation formats is the same across age levels. These contradictory results prompted
the present study, which was designed to examine the effect of presentation format with
older readers who comprehend below grade level.

Additional reasons for investigating the inﬂuence of presentation format. were,
once again, related to observed trends within the current educational climate. For
example, within schools the notion that the print-only context is the most challenging is
evident in various methods used for accommodating the needs of weaker readers.
Teachers working with struggling readers, or with students who are learning to read in
English, oftentimes are encouraged to incorporate audio or visual components to reading
instruction. In addition, teachers may be encouraged to read aloud to students, to have
students use listening centers, to provide both oral and written forms of test questions,
and to include pictures or video representations of the content they are attempting to

teach. The present study was designed to examine the assumptions that underlie such

16

interventions by investigating higher level processes across various media contexts.
Based on the assumptions underlying these common intervention strategies, it might be
predicted that this study would ﬁnd more students struggling within the print-only
context, and having an easier time generating inferences when audio and video
information was added. The study was created, however, to leave the possibilities open

and actually call such predictions into question.

Inference Generation across Text Difﬁculty Levels

Along with considering the medium of text presentation, this study takes into
account students’ instructional reading levels. The potential inﬂuence of students’
instructional reading level and relative text difficulty is examined in a way that builds
upon existing research investigating how inference-making processes vary among
“skilled” and “unskilled” readers (Curtis, 1980; Hannon & Daneman, 1998) and how text
difficulty bears on inference-making (Pezdek et al., 1987). This study attempts to look at
readers at the same instructional reading level in all three medium contexts, using three
presentation formats.

While earlier work has considered the inference-making processes of both
“skilled” and “unskilled” readers, using the same text, this study examines students’
inference-making processes ﬁrst with texts that are accessible (matching students’
instructional level in terms of silent reading comprehension), and again with texts that
present a challenge (above students’ instructional level in terms of silent reading
comprehension). Students reading below grade level were selected in an effort to

examine individual differences in a group that often experiences standardized

17

interventions. This close consideration of students’ instructional reading level and its
match to text difﬁculty, builds upon the relatively small body of work that investigates
individual differences in inference-generation (Thurlow & van den Broek, 1997).

According to van den Broek and his colleagues, a reader’s ability to engage in
higher level process such as inference-making may depend on the relative ease of the
reading context. For example, van den Broek et al suggest that when individuals are
attempting to comprehend within a context that is not necessarily conducive to meaning
construction, they tend to engage in limited inferential activities. Within these sorts of
circumstances, readers engage in van den Broek et al’s interpretation of something called
Minimalist reading, when the context may be less than conducive to comprehension
based on a number of factors including unfamiliar or challenging text structure, content,
or vocabulary. On the other hand, Maximalist reading, which involves deep processing
and plenty of inferential activity, occurs within contexts that are more conducive to
meaning construction because they may involve well-structured texts, content familiarity,
and so on (van den Broek, Fletcher, & Risden, 1993).

This study was designed to examine students’ abilities to generate inferences in
two contexts that may have been more or less conducive to meaning construction based
on the difficulty levels of the texts. While one might predict that students would
participate in more Maximalist reading with the easier texts that contain more familiar
vocabulary and more concrete story lines, one might expect more Minimalist reading,
with limited inferential activity, while students engage with texts that are more abstract

and contain more challenging vocabulary and sentence structures.

18

Overall, this study was designed to investigate how text level and presentation
format inﬂuences struggling readers’ abilities to generate inferences. And ultimately, the
investigation was intended to shed light on factors that are identiﬁed in various models

that attempt to explain reading comprehension processes.

19

METHODS

This study was designed to examine the nature and frequency of readers’
inferences as they engage in texts that vary according to level of difficulty and
presentation format. Verbal protocols were collected during two 3-part, one-on-one
research sessions. Analyses of these protocols focused on the types and frequencies of
inferences generated with both accessible and more challenging texts in all three
presentation formats.

Data Collection
Resemh Sites and Particm
Elementary School Sites
The study was conducted at ﬁve public elementary schools in the San Francisco

Bay Area. Table 2.1 provides general information about each school site.

20

Table 2.1

School Information

 

%African American

% Free /Reduced Lunch
% American Indian

% Asian

% Paciﬁc Islander

% Filipino

% Hispanic

% White

% Multiple lNo Response

2002 Statewide Rank“
% ELL

'No. of Participants

School
2002 Base API*

 

561 l 83 37 0.4 1.7 0.6 7.6 41.0 41.2 7.2 0.4

Parker
3

 

M51) 67 22 0.0 5.0 2.9 18.1 22.6 47.3 4.2 0.0

658 4 (2001)

Wiley
0\

 

”9°" 92 68 0.0 8.1 1.0 2.1 74.0 9.3 3.2 2.3

496 l wrde

Donner
Ur

 

 

II/USP

(1999) 56 34 0.6 6.3 6.3 6.2 52.4 13.9 14.4 0.0

arper

H

 

Not
school- 18 20 0.5 33.0 0.2 0.9 10.8 8.0 37.5 8.9

wide

868

Oliver

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

‘Academic Performance Index (API) which measures the performance and growth of California schools
using a numeric scale that ranges from a low of 200 to a high of 1000. The interim statewide API
performance target for all schools is 800. A school’s growth is measured by how well it is moving toward
or past that goal. A school’s base year AOU is subtracted from its growth API to determine how muchthe
school grew in a year. For elementary schools, the 2002 Base API incorporates the results of l) the
Stanford 9 in all content areas, and 2) the California Standards Tests in English-Language Arts,
Mathematies and Social Science.

** API scores are divided into deciles numbered from 1 (the lowest) to 10 (the highest).

21

These sites were selected in an effort to study inference-making processes with
students in schools that are performing below the average on standardized reading
measures. In the state of California, many underperforming schools are in high poverty
communities with students that are ethnically and linguistically diverse. Previous studies
have shown that students in hi gh-poverty schools with populations that are ethnically and
linguistically diverse often experience reading instruction that is oriented more toward a
skills-based approach and less toward a meaning-making approach (Allington, 1991;
Fitzgerald, 1995). For this reason it is particularly important to represent such schools in
studies that examine meaning-making processes. Twenty-six of the twenty-seven
participants attended schools that rank below the state average on standardized measures,
have a high proportion of students who qualify for free and reduced lunch, and serve a

large number of English language learners.

Swim

Participants included 4th and 5th grade students who comprehend at the 3rd grade
level as measured by the Reading Inventory for the Classroom (F lynt & Cooter, 1998).
Teachers were asked to provide a list of students that were likely to fall within this
window. During the one-on-one assessment sessions for the RIC, students silently read
narrative passages and responded to a set of comprehension questions including literal,
inferential, and evaluative questions related to character, setting, story problem, and
theme (see Appendix A). Those who comprehended at the third grade instructional level
were asked to read the third grade level passage aloud for a measure of oral reading

ﬂuency. Students who fell into the third grade comprehension level were given consent

22

forms to take home and complete with their parents (see Appendix B). All students who
returned consent forms participated in the study. Participants included sixteen fourth
graders and eleven ﬁfth graders. There were twenty girls and seven boys, and ten
students spoke English as a second language. Participants included fourteen Aﬁican
American students, eight Latino students, two Asian students, two White/Caucasian
students, and one Filipino student, all comprehending below grade level expectations.
While earlier work has compared inference-making processes of ‘good’ and ‘poor’
readers (Neuman, 1990) and older and younger children (Meringoff, 1980) (Beal, 1990),
this study examines older students all comprehending at the third grade level in an effort
to highlight the possible effects of text format and text level on inference-making
strategies. In addition, struggling readers are underrepresented in studies using protocol
analysis, as researchers are inclined to use talented readers for such work because they
are often more verbal (Afﬂerbach & Johnston, 1984; Afflerbach, 2000). The present
study analyzes verbal reports of struggling readers in an attempt to add these voices to the

existing research.

Par—cuts

Students’ parents were asked to complete a survey about their child’s reading,
listening, and viewing habits, as well as how verbal they tend to be one-on-one, as well as
in small and large group settings. Surveys were sent home with each participant (see
Appendix C). Although the survey was brief, offered a phone survey option, and
included an addressed, stamped envelope, only ﬁve surveys were returned. As a result,

survey data was not included in analysis.

23

Materials

Think Aloud Training Materials

In whole-class or one-on-one sessions, students learned how to engage in the
think aloud process that was used in the two research sessions. For the one-on-one
sessions, the poem “If I Were In Charge of the World” by Judith Viorst was used to
introduce the process of ‘thinking aloud.’ This poem was selected because previous
experience has taught me that its content generates interest and provokes lively
discussion with upper elementary students (see Appendix D). For the whole class
sessions, the video version of Mwenye Hadithi and Adrienne Kennaway’s Hot Hippo was
used in addition to the poem. Additional explanation of the think aloud training is

included in the procedures section.

Research Session Texts

The six narrative texts used for the research sessions were selected for a number
of reasons. First, since the aim of the study was to investigate readers’ inference-making,
texts were chosen based on van den Broek’s suggestion that investigations of mental
models usually involve stories that revolve around the intentions of the protagonist (van
den Broek, Fletcher, & Risden, 1993). In addition, texts were selected because they also
appear in video formats (versions that last about ten minutes) that closely match the
original written text, include animation, and fall within the third or ﬁfth grade levels of

difficulty and interest.

24

While text leveling can be a rather subjective venture, involving a number of
factors including vocabulary, number of syllables and characters per word, number of
words per sentence, and sophistication of content and concepts, to name a few, these texts
were leveled based on one or more of the following leveling systems: F ountas and
Pinnell leveled texts lists, Fry Readability scale, F lesch-Kincaid Grade Level scale,
and/or Reading Recovery/Early Intervention levels. Finally, texts were evaluated using
the Literature Scale in Chall’s Qualitative Assessment of Text Diffrculty, and this
information was added to the synthesis resulting in the ﬁnal grade level determination
(Chall, 1996). Table 2.2 lists the texts used in the research sessions.

Table 2.2

Texts Used in Research Sessions

 

 

 

 

 

 

 

Final rating
. (incorporating
Reading Foutas & Fry Flesch- Clnll’s
Text Recovery Pinnell Readability Kincaid Qualitative
rating rating rating rating Assessment

of Text

Diﬂiculty) Rationale
The Amazing Bone
(Steig, l 97 6) 29 N 4.0 3.9 3.0

F '1'
Doctor De Soto settings
(Steig, I983) 26 L/M 4.0 3.9 3.0 and plot
. structures
Madelme and the
m? 1959) na" an 5.0 3.9 3.0
Why Mosquitoes More
BE: 1“ P°°p1° 5 na an 5.0 4.6 5.0 abstract,
(Aanderna, 1992) wk“
Strega Nona structures,
(dc Paola, 1975) 113 113 7.0 5.0 5.0 and
languages

A Story-A Story Other than

 

 

 

 

 

 

 

 

 

*na = not available

25

For the print version of the text used by the students during the research sessions,
each story was typed in 14-point Times New Roman font, using the published text. In
order to provide appropriate junctures for eliciting verbal protocols, each story was then
divided into three episodes at natural transition points in the story. The stopping points
for each episode were set at moments in the story when a new element of the plot was
introduced, when there was a change of scene, and/or when a shift in the story was about
to occur. In addition, transitions were placed at points when inferences would be likely
to occur. Table 2.3 displays examples of transition points and their rationale. The
printed texts used in the research sessions contained no illustrations because the print
context was designed to examine the student’s ability to generate inferences from verbal
input only, without the support of illustrations. A new page was started for each episode
so students held only one episode at a time and could not reread or rescan text read

previously (see Appendix E).

26

Table 2.3

mmples of Story Transitions

 

 

 

 

Transition Example Rationale for Transition Point Types of Inferences Likely to
Occur
Introduction of the main Predictions (e. g. about

...Thc night grew longer and problem in the story whether/why Mother Owl will
longer. The animals of the decide to once again wake the
forest knew it was lasting much Shift in time and place sun)

too long. They feared that the

sun would never come back. Evaluations (cg about whether

Mother Owl’s actions are

III-IIIIIIIIIIIIIIIIIIII jliStIﬁed)
gﬂﬂﬁmﬁ 5:116}, Personal connections (cg. about
eamc and sat down, pem’ pem’ fears of the dark)

pem, around a council ﬁre. ..

Introduction of the main Explanations (c. g about why

...And Strega Nona called Big problem of the story Big Anthony is tempted by the
Anthony in for supper. But too pot)

bad for Big Anthony, because Shift in time and place

he didn’t see Strega Nona blow Associations (e. g about Strega
three kisses to the magic pasta Nona’s relationship with

pet. And this is what happened. Anthony)
IUIIIIIIIIIIIIIIIIIIIIII InfemceSBasedon

The next day when Big Knowledge/Beliefs about the
Anthony went to the town World (63' about why MIC
square to fetch the water, he told show off)

everyone about the pasta pot. ..

 

 

 

Finally, audio versions of each text were recorded in three parts onto a compact

disc. These CD’s along with the DVD video versions of ﬁve of the texts, acquired

through Weston Woods, were played on a laptop computer. One text, Madeline and the

Gypsies, was not available in DVD, so a VHS version was used instead. While the video

versions omitted some descriptors and ﬁmctional words (eg. said Strega Nona) and

shifted some word order, they were very similar to the original print text, with minimal

textual omissions and substitutions. Participants used headphones to listen to the audio

and video versions.

27

 

Procedures

Think Aloud Training

Students were introduced to the think aloud process in either whole class or
individual settings. The 13 students from Parker Elementary School participated in
whole class training sessions for the think aloud procedure. The think aloud process was
modeled using the video version of Mwenye Hadithi and Adrienne Kennaway’s Hot
Hippo, a text not used in the research sessions Next, students had the opportunity to
participate in guided and independent practice of the think aloud procedure using the
poem “If I Were In Charge of the World” by Judith Viorst. At the other school sites,
there was no opportunity to conduct the think aloud training in a whole-class format.
Instead, students were introduced to the think aloud procedures during one-on-one

sessions. The same poem was used to model and practice the process of thinking aloud.

Resaarch Sessions

At the start of each session students were reminded that this study was being
conducted in order to learn more about how people ﬁgure out what is going on in a story,
and how individuals help themselves understand what they are reading, what they are
listening to, and what they are watching. Students participated in two 3-part research
sessions. In Session 1, students were exposed to three third grade level texts, each
presented in a different format. In Session 2, the process was repeated with three ﬁfth

grade level texts. The more accessible third grade level texts were used in the ﬁrst

28

session in an effort to develop trust and comfort among participants who already

struggled with reading. Table 2.4 presents a sample of how these sessions were

organized for three of the participants.

Table 2.4

Sample of Session Schedule

 

 

 

 

 

 

Session 1 Session 2
Third Grade Level Texts Fifth Grade Level Texts
Part1 Part2 Part3 Part1 Part2 Part3
Amazing Doctor De Madeline A Story Mosquitoes Strega Nona
Student 1 3°” S°‘°
VIDEO PRINT AUDIO PRINT VIDEO AUDIO
Madeline Amazing Doctor De Strega Nona A Story Mosquitoes
Student 2 Bone Soto
PRINT AUDIO VIDEO VIDEO AUDIO PRINT
Doctor Dc Madeline Amazing Mosquitoes Strega Nona A Story
Student 3 Soto Bone
AUDIO VIDEO PRINT AUDIO PRINT VIDEO

 

 

 

 

 

 

 

The design was counterbalanced for story-format combinations as well as

sequence of story and presentation format in order to control for the effects of particular

story-format, story-order, and format-order combinations (see Appendix F for story-

forrnat-order selection matrices).

At the beginning of each session students were provided with a brief introduction

reminding them of the purpose of the task (i.e., I want to learn more about how kids

 

ﬁgure out what’s going on in a story and how they help themselves understand what they
are reading, what they are listening to, and what they are watching). Then, as each story
was presented, students were asked to indicate whether it was familiar. They were asked

again at the end of the story. This information was noted and used as a covariate in

29

analysis because earlier work has shown a prior knowledge effect on inferable knowledge
(Pearson, Hansen, & Gordon, 1979).

Students started by reading, viewing, or listening to the ﬁrst part, or episode, of
the ﬁrst story. When they were done, they were asked to talk about what they were
thinking at this point in the story (i.e., Okay, (student name), what are you thinking at this
point in the story?”). Ifany responses required clariﬁcation or elaboration, students were
asked follow-up questions (cg, Why do you say that? Why do you think so? Whom do
you mean when you say ‘they’?) . This process was repeated after the students read,
viewed, or listened to the second and third parts of the story.

Probes were kept general and non-directive following Ericsson’s suggestion that
this design avoids the possibility of induced effects, affecting the think aloud process less
(Ericsson & Simon, 1980). Moreover, follow-up questions and probes were used in a
ﬂexible, but mindful manner in response to students’ individual differences. For
example, if a student’s initial report supplied the rationale or support for earlier
statements, the prompt ‘Why do you say that?’ was not used. If a student did not supply
such information, the follow-up prompts were used. Sometimes it was the case that
students did not have any support or rationale for earlier statements and did not respond
to the prompt. Other times, the prompt allowed more reticent or less verbal students the
opportunity to offer rationale or support not stated in the initial report. This limited
probe-when—necessary technique is similar to that used in other studies of reading-related
phenomena (Phillips, 1988; Neuman, 1992), and was designed to result in more complete
verbal reports from students. Afﬂcrbach suggests that the value of verbal report data for

investigating reading related processes is demonstrated through both methodological

30

rigor and ﬂexible use (Afflerbach, 2000). The process used in this study was designed to
be rigorous in its attempt to gather rationale or support whenever available, but also
ﬂexible in its attempt to respond to individual differences related to how verbal students
were in the interview situation. In addition, readers were stopped just two times during
the reading of the narrative in an attempt to maintain the ﬂow of reading required to
engage in inference-level processes.

When students ﬁnished thinking aloud after the third and ﬁnal part of the story,
they were asked three inference-level questions. The ﬁrst question for each story was an
explanation based question, requiring students to make inferences about what motivated a
main character’s actions, thoughts, or emotions. The second question was an association
based question, asking students to make inferences about a main character’s traits,
features, qualities, and/or characteristics. The third question for each story was an
evaluation question, requiring students to make infer any lessons learned by the

characters in the story. Table 2.5 contains the questions for each story.

31

Table 2.5

Inference Questions

 

 

 

 

 

The Amazing Bone Doctor De Soto Madeline and the Gypsies
Why does Pearl think that her Why does Doctor De Soto decide Why does Madeline suggest that
parents will say that she only to let the Fox into his office? they write Miss Clavcl a letter?
imagined that the bone was Why do you say that? Why do you say that?
talking to her? Why do you say
that?
How would you’describe Pearl? How would you describe Doctor How would you describe
What sorts of thmgs seem De Soto? What sorts of things Madeline? What sorts of things
“1190113111 ‘0 her? Why do you seem to be important to him? seem to be important to her?
say “131? Why do you say that? Why do you say that?
”WWW” “the Do youthinkofthe charactersin Doyouthinkanyofthc
characters [I] ““5 5’95] learned a this story learned a lesson or will characters in this story learned a
Legion or will do things? Wh ? do things differently in the lesson or will do things
“6‘1“” 1“ “‘3 mm“ 0 future? Who? Why do you say differently in the future? Who?
Why do you say that? that? Why do you say that?
Strega Nona A Story-A Story Why Mosquitoes Buzz in
Peoples’ Ears
Why does Big Anthony decide to Why does Ananse want to buy Why does the mosquito feel

use the pasta pot? Why do you
say that?

 

the Sky God’s stories? Why do
you say that?

 

guilty? Why do you say that?

How would you describe Strega How would you describe How would you describe King
Nona? What sorts of things seem Ananse? What sorts of things Lion? What sorts of things seem
to be important to her? Why do seem to be important to him? to be important to him? Why do
you say that? Why do you say that? you say that?
Doyouthinkanyofthe Doyouthinkanyofthe Doyouthinkanyofthe
characters in this story learned a characters in this story learned 3 characters in this story learned a
lesson or will do things lesson or will do things lesson or will do things
differently in the future? Who? differently in the future? Who? diﬁcrently in the future? Who?
Why do you say that? Why do you say that? Why do you say that?

 

After responding to the inference questions, students proceeded to the second part

of the interview, following the same procedures with a different text and presentation

format. Finally, students completed the ﬁnal part of the interview with a third

text/presentation format. Each part of the interview took about 20 minutes. Students’

think aloud segments and responses to inferences questions were audio-taped and

transcribed.

32

 

Data Preparation
Segmenting Verbal Protocols into Propositions
To begin, each verbal protocol was divided into a list of idea units, or propositions
using a procedure similar to those used in earlier studies of reading related processes
(Phillips, 1988; Rickheit & Strohner, 1985; Ericsson & Simon, 1993). Each proposition
consisted of one predicate and one or more arguments conveying its meaning. While
arguments were considered independent linguistic expressions that can stand alone,
predicates were viewed as inherently dependent expressions of the arguments, elements
such as qualities, relations, acts, properties, and states (F rawley, 1992). Table 2.6
presents an example of how a section of one protocol was segmented. The main

predicates have been underlined.

33

Table 2.6

Ex_ample of Protocol Segmentation

 

 

 

Original protocol Segmented protocol
(R = Researcher; S = Student) (R = Researcher; S = Student)
Amazing Bone Amazing Bone

R: So, (Student A), what I’d like you to do is tell me
what you’re thinking at this point in the story.

8: About, how is, how she’s gonna tell her parents
about the bone.

R: Okay, anything else?

S: (shakes head)

R: Okay, why are you wondering about that?

S: Because, her parents, her parents aren’t gonna
believe her because they’re gonna just say that she’s
R: Okay, (Student A), what are you thinking now?

8: Uh, the bone is gonna ﬁnd a way to get them
outta the room.

R: Why do you think that?

S: Because he stood up for hisself when the fox um
took him, but he got back to Pearl and now I think
that he’s gonna do it again and get outta the room.

R: Okay, we shall see...

 

R: So, (Student A), what I’d like you to do is tell me
what you’re thinking at this point in the story.

About, how is, how she’s gonna tell her parents
about the bone.

R: Okay, anything else?
(shakes head)
R: Okay, why are you wondering about that?

Because, her parents, her parents aren’t gonna
My; her

beeause they’re gonna just sg that she’s imaginary
stuﬂ‘.

R: Okay, (Student A), what are you thinln'ng now?

Uh, the bone is gonna ﬁnd a way to get them outta
the room.

It Whydoyouthinkthat?

Beeause hemmrhisself

when the fox um ml; him,

but he got back to Pearl
andnowlthinkthathe’s gonnadoitagain
andge_touttatheroom.

R: Okay, we shall see.

 

While there was variation among students and stories, the average number of

propositions per protocol was 34.6 per story. Table 2.7 summarizes the average number

of student-generated propositions for each story.

34

 

Table 2.7

Average number of student-generated propositions for ermh story

 

 

 

 

 

 

 

 

 

Story Mean N Standard Range
deviation
The Amazing Bone 43.3 27 22.6 91
Doctor De Soto 34.7 27 15.8 60
Madeline and the Gypsies 30.3 27 11.7 40
A Story-A Story 31.8 27 15.3 47
Why Mosquitoes Buzz in Peoples’ Ears 32.2 27 16.7 63
Strega Nona 35.4 27 13.7 44
TOTAL 34.6 162 16.6 93

 

 

 

 

 

In order to check for reliability of this segmenting process, external raters
segmented 10% of the protocols. Interrater agreement was 82.14%. Raters included
three students in the Developmental Teacher Education program at the University of
California, Berkeley. Each rater was provided with one example protocol in both original
and segmented form, along with directions for segmenting the protocols into a list of

propositions.

CodingProposjtions
While Ericsson suggests that the process of encoding of verbal protocols cannot
be made thoroughly objective (Ericsson & Simon, 1993), efforts were made to avoid
problems that can arise when coding is affected by the coder’s knowledge of the student,
the context, and memories of previous encodings. Each segmented 3-part interview
protocol was divided into three separate files, one for each story interview, and given a 6-

digit identification number. The student names, as well as presentation format and

35

 

sequence information were removed. Coding of each story protocol was done in a
random order, with no knowledge of the student, presentation format, or sequence. This
way the protocols contained an interview associated with an entire story, oﬁ‘ering enough
context to permit more of the semantic content of the protocol to be retained. At the
same time, coding decisions were made independent from knowledge about the student,

presentation format, and sequence.

Codingiropositions into General Categories

Next, these propositions were mapped onto a set of general categories. The ﬁrst
step was to encode each proposition in terms of the simplest process that could account
for it, keeping in mind Ericsson’s assertion that processes themselves cannot be encoded
directly from verbalizations, but must be inferred. This analysis reﬂected an iterative
process involving two passes through the data. While some propositions mapped onto
existing categories from earlier research, others formed novel or modiﬁed categories. For
the first pass through the data, each proposition was read alongside lists of general
verbalization categories from earlier work (Ericsson & Simon, 1993; Meringoff, 1980;
Neuman & Roskos, 1992; Pressley & Afﬂerbach, 1995; Trabasso & Magliano, 1996).

The categories, along with definitions and examples are displayed in Table 2.8.

36

Table 2.8

General Proposmon Categories

 

 

 

 

 

 

 

 

 

Category Deﬁnition Examples
Reeall Verbalizations that match or and she put him in her purse
paraphrase content found in the
text and she felt the breeze
Inference Verbalizations that require 'Cause she probably thought that they was
generating processes worried.
sohewastryingtoprove hispoint.
Reiteration Reiterating a previous (response to the question ‘Why does the
verbalization or question mosquito feel guilty?’)
It feels guilty because um...it feels guilty
Metacomment Verbalizations about ones I forgot the rest...
thinking
That, Ireaditbutldidn‘tunderstandit
Unrelated/contrived Verbalizations that are not Is there a queen animal in the jungle?
related to the task or purely
making up story content and then they gave the bone back to the witch.
Unintelligible Verbalizations that are and then after they scared people, they um, they
incomplete, inaudible, or could (inaudible)
syntactically confusing
but I‘m not so sure I think once the fox gets like,
like ﬁne, he gets okay,
Reﬁ'ain Not responding or stating that Um. ..
one cannot or will not respond

 

 

Becauseumheliked,um,ldon‘tknow.

 

While verbalizations were placed into these broad categories, those that fell into

the seventh category, Refrain, were noted, but not included as propositions. Likewise,

recall and inference level propositions that were either inaccurate or implausible were

noted, but not included in the overall analyses. Applying a process similar to that used in

McCormick’s study of disabled readers' erroneous responses to inferential

comprehension questions, propositions were considered inaccurate or implausible if they

were wild guesses, erroneous information, or illogical inferences unsupported by the

37

 

story content (McCormick, 1992). Examples of inaccurate and implausible propositions

are displayed in Table 2.9.

Table 2.9

 

@mples of Inaccurate and Implau_sible Propositions

 

 

Context (eg question, prompt) Proposition Rationale for Rating as Inaccurate
or Implausible
Student’s response to the question He (Ananse) was showing oﬂ” The Sky God was not Ananse’s
“How would you describe that his dad was a sky 80d, father. Ananse wanted to buy the
Ananse, what sorts of things seem Sky God’s stories.

to be important to him?” (A
Story-A Story)

 

Student’s open response to the

And he (Dr. De Soto), it was his

Dr. De Soto’s wife is his

 

 

 

 

 

 

 

ﬁrst segment of the story (Doctor patient that became his wife assistant, not his patient

De Soto).

Student’s open response to the Um, it's about this girl (Pearl) Pearl is on her way home from

ﬁrst segment of the story (The who's going to school school.

Amazing Bone).

Student’s open response to the That Miss (Clavcl), the teacher The teacher did not get lost.

last segment of the story got lost. Madeline and Pepito were left on

(Madeline and the Gypsies). the Ferris Wheel.

Student’s response to the question but she (Strega Nona) didn't like Strega Nona gave him food to eat

“Why did Big Anthony decide to give him (Big Anthony) that and a place to sleep. Big

use the pasta pot?” (Strega Nona) much—he didn't get that much Anthony’s curiosity and desire to
food. prove himself to the villagers

prompts him to use the pot.

Student’s response to the question Because she, Miss Clavel's sick Miss Clavcl was not sick.

“Why did Madeline suggest that Madeline wrote the letter beeause

they write Miss Clavel a letter?” she feared that Miss Clavcl was

(Madeline and the Gypsies) worried.

Student’s open response to the because he (The Fox) thought his The fox reconsiders eating the De

second segment of the story tooth wasn't gonna get better, Soto’s beeause they have helped

(Doctor De Soto). his tooth feel better.

Student’s response to the question No. The Fox learns a lesson about

“Do you think any of the
characters in this story learned a
lesson or will do things

differently in the future?” (Doctor

De Soto)

 

 

kindness and honesty.

 

38

 

Such inaccurate and implausible responses were not included in the analysis
because this study was designed to examine the effects of presentation format and text
level on struggling readers’ ability to generate inferences that would allow them to
construct coherent meaning of narrative texts. While other studies have analyzed
students’ errors in comprehension responses (McCormick, 1992; Williams, 1993), this
study examined effective inference-making across text levels and presentation formats.
Although other work has suggested that students who have a difficult time with
comprehension may be more inclined to make wild guesses, provide erroneous responses,
and generate illogical inferences (McCormick, 1992; Williams, 1993), the present study
made comparisons, across levels and formats, of students’ inference-making, not their
lack of inference-making.

Table 2.10 displays the sums and means of the following: accurate recall
propositions, inaccurate recall propositions, total recall pr0positions (both accurate and
inaccurate), plausible inference—level propositions, implausible inference-level
propositions, total inference-level propositions (both plausible and implausible). Again,
only accurate recall propositions and plausible inference-level propositions were included

in the data analyses.

39

Table 2.10

Recall and Inference-Level Propositions, Accuracy and Plausibility

 

 

 

 

 

 

 

 

 

 

 

N Sum Mean

Accurate Recall propositions 27 2651 98.19
Inaccurate Recall propositions 27 176 6.52
Total Recall propositions (accurate + inaccurate) 27 2827 104.70
Plausible Inference-Level propositions 27 2316 85.78
Implausible Inference-Level propositions 27 126 4.67
Total Inference-Level propositions 27 2442 90.44
(plausible + implausible)

 

Table 2.11 reports on the sums and means of the various proposition categories.

Table 2.11

GengalPropositions Sumsand Means (Excluding Inaccrgate/Impjausible Propositions)

 

 

 

 

 

 

 

 

N Sum Mean
Recall 27 2651 98.19
Inference 27 23 16 85.78
Reiteration 27 90 3.33
Metacomment 27 68 2.52
Unrelated/contrived 27 37 1.37
Unintelligible 27 15 .56

 

 

 

 

Coding Inference-level Propositions into Subcategories

Next, propositions labeled “inferences” were examined and compared with the

types of inferences delineated in earlier studies (Graesser & Kreuz, 1993; Meringoff,

1980; Neuman, 1990; Pressley & Aﬁlerbach, 1995; Thurlow & van den Broek, 1997;

Trabasso, 1981; Trabasso & Magliano, 1996). Again, this analysis involved an iterative

 

 

process with two passes through the data, with some propositions mapping onto existing

categories and others forming novel or modiﬁed categories. The resulting six

subcategories included inferences related to predictions, explanations, associations,

evaluations, world knowledge or beliefs, and personal connections. Table 2.12 displays

the subcategories along with deﬁnitions and examples.

Table 2.12

Inference Subcatggories Deﬁnitions and Examples

 

 

 

 

 

 

 

 

Inference Deﬁnition Example
Subcategory

Prediction Verbalizations that predict future events or causal Um, I'm thinking that like

consequences somethings gonna happen
when the fox is gonna try to eat
Pearl

Explanation Verbalizations that provide the basis, rationale, Because he wants to be like, he
motive, condition, or occasion of events, states, wants to be him self,
or actions

Association Verbalizations that provide information about but in the story, I thought the
who does what to whom with what, when, and mosquito was happy in a way
where, or provide information on features,
properties, relations, and ﬁlnctions of persons,
objects, or concepts

Evaluation Verbalizations that evaluate the style or content Because he, he, he 1ear-, he
of the text or evaluate characters’ actions or should learn not to mess with
behaviors other people

World knowledge or Verbalizations based on one’s knowledge or Beeause no regular bone could

beliefs beliefs about the world talk

Personal connection Verbalrzatr' 'ons based on a personal association or Because I wouldn't let nobody
connection to the characters or events take my stories

 

 

 

The ﬁrst three categories (prediction, explanation, and association) are borrowed

from Trabasso and Magliano’s investigation of eight readers’ conscious understanding of

41

 

eight narrative texts using think-aloud methodology (Trabasso & Magliano, 1996). The
ﬁrst two types of inferences (predictions and explanations), integrate sentences across the
text. For example, predictions are forward-oriented and are conveyed as expectations of
what is to come, based on what has been read. Predictions that are substantiated then
serve to link sentences or thoughts. Trabasso and Magliano hypothesized that if
comprehension is expectation-based, it seems likely that readers would generate a
relatively high proportion of prediction-level inferences (Trabasso & Magliano, 1996).
What they found, however, is that while readers did make a number of predictions, the
majority of the readers’ inferences were backward-oriented explanations that link what is
being read to prior knowledge or text information. These ﬁndings suggesting that
comprehension is more explanation-based are aligned with assumptions based on earlier
work that consider explanations to be the main form of understanding in various theories
about comprehension (e. g. Graesser, Singer, & Trabasso, 1994; Trabasso & Sub, 1993).
According to this earlier work, explanations may serve to guide comprehension as
readers pursue a need to know the reasons for story events.

Finally, Trabasso and Magliano noted readers’ use of associations which are
concurrent and usually elaborate the text, adding new information which may or may not
be used again. For example, a prior association may be retrieved and used to integrate
text through a prediction or explanation. It is important to note, then, that in the present
study, inference-level propositions were coded only once, with only one category per
proposition. So, associations that may have served explanatory or predictive functions in

the readers’ mind are recognized as associations only (Trabasso & Magliano, 1996).

42

The fourth category, evaluation, is found in Trabasso’s earlier work (Trabasso,
1981) and also classiﬁed by Pressley and Afﬂerbach as one of three types of activities
readers engage in during constructively responsive reading: constructing meaning,
monitoring, and evaluating (Pressley & Aﬂlerbach, 1995). In Pressley and Afflerbach’s
scheme, evaluating includes readers’ assessments of the text style (e. g., Is the writing
good or bad? Are the examples effective and compelling?), as well as the text content
(e. g., Is the content plausible? Is the storyline surprising? Is the main character doing the
‘right’ thing?) (Pressley & Afﬂerbach, 1995). And as with Trabasso and Magliano,
Pressley and Afflerbach recognize the existence of a dynamic interplay among the three
activities and admit that it borders on reductionistic to separate these three activities.
Nevertheless, the present study codes the propositions only once.

The ﬁfth and sixth categories, inferences based on beliefs or knowledge about the
world and personal connection are borrowed from Trabasso’s earlier work (Trabasso,
1981) as well as Meringoﬁ’ s investigations of the bases of readers’ inferences as they
attempted to comprehend stories presented in both video and storybook formats
(Meringoff, 1980). Meringoff found that some inferences were based on the readers’
general beliefs or knowledge about the world, while others were based on their personal
association with a character or a particular situation. For this study, the term personal
connection was used (rather than personal association) to avoid any confusion with the
general association category.

In addition, two aggregate variables were created: text-connected inferences and
self-connected inferences. The text-connected inferences variable combined the three

subcategories of prediction, explanation, and association, since such inferences required

43

the students make predictions, offer explanations, and designate associations directly
related to the story events and characters. The self-connected inferences variable
combined the three subcategories of evaluation, world knowledge or beliefs, and personal
connection, since these inferences required students to assert evaluations and make
connections based on their own experiences, background knowledge, opinions, and value
systems.

The ﬁ'equencies with which the various types of inferences occurred were noted.
As reported in Table 2.13, the types of inference that were most prevalent were
associations. The second most frequent inferences were evaluations, followed by
explanations, predictions, inferences based on knowledge or beliefs about the world, and
personal connections. And on average, students generated more text-connected than self-
connected inferences. This frequency trend is one that continues across additional
analyses reported in the Results section, and may be not surprising based on the nature of
the follow-up questions described above that may have prompted students to make more
associations, evaluations, and explanations. It is interesting to note, however, that
students’ overall inference-making was not signiﬁcantly inﬂuenced by the order of the
story. In other words, in discussing the sixth story, having heard similar questions
throughout the previous interviews, students’ inference-making did not necessarily
increase. This is surprising given that van den Broek and his colleagues found that
questioning may beneﬁt comprehension when the need for decoding is eliminated (van
den Broek et al., 2001). In this study inference-generation did not appear to increase with
each successive story, regardless of whether the format required decoding. Finally, I

noted the number of inferences that were made spontaneously as well as those prompted

44

by explicit questions or follow-up prompts such as ‘Why do you say that?’ This
information was used as a covariate in the analysis.

Table 2.13

Inference Subcategories Sums and Means

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

N Sum Mean Std. Deviation
Prediction 27 195 7.22 10.32
Explanation 27 436 16.15 5.26
Association 27 804 29.78 14.24
Evaluation 27 755 27.96 24.68
World knowledge/beliefs 27 105 3.89 5.71
Personal connection 27 21 0.78 2.17
Text-connected inferences 27 1435 53.15 20.77
Self-connected inferences 27 881 32.63 29.37
TOTAL inferences 27 2316 85.78 44.36
Prompted inferences 27 1582 58.59 21.83
Spontaneous inferences 27 734 27.19 25.16

 

In order to check for reliability of this coding process, external raters coded 10%
of the protocols. Interrater agreement was 80.88%. Raters included the same three
students in the Developmental Teacher Education program at the University of
California, Berkeley used to assess interrater agreement of protocol segmentation. Raters
coded segmented protocols that were different from the ones they had segmented, but
randomly selected from those that remained. Each rater was provided with a sample
segmented protocol that had been coded, along with deﬁnitions and examples of the

various types of propositions as deﬁned in Table 2.8 and 2.11.

45

Data Analysis

Analyses of data gathered in the form of audio-taped and transcribed verbal
reports, were examined in light of the research questions: What inference-making
processes do students engage in with narrative texts varying according to level of
difficulty and presentation format? What types of inferences do students make, and how
frequently are they generated? Means of the six types of inferences generated within the
three presentation formats at the two levels of text difficulty were analyzed using a
within-subjects 2x3 x6-way analysis of variance. Noted differences among means
(described in the next section) prompted a post hoc analysis using the Least Signiﬁcant
Difference test with a 6-leve1 variable (1=Grade 3 text in print, 2=Grade 3 text in audio,
3=Grade 3 text in video, 4=Grade 5 text in print, 5=Grade 5 text in audio, 6=Grade 5 text
in video). Three mean differences were signiﬁcant and are discussed in the section that
follows. Finally, follow-up analyses were conducted to determine whether the story, the
order of story or presentation format, and the counterbalanced story-format combinations

had signiﬁcant effects on students’ inference-making.

46

RESULTS

In this chapter, data gathered from students’ verbal protocols are presented in
three parts. The ﬁrst set of results reports the main effects of presentation format and text
level on students’ inference-making. A second set of analyses examines the potential
inﬂuence of story, the order of story or presentation format, and story-format
combinations on students’ inference-making. A third set of results reports students’
recall-level data. In addition, student examples are woven throughout the chapter to

illustrate the ﬁndings in a more descriptive manner.

Effects of Text Level and Presentation Format on Inference-Making
Preliminary analyses examined the main effects of text level and presentation
format on students’ inference-making. Table 3.1 reports the means of students’ total
inferences based on text level and presentation format.
Table 3.1

Total Inferences Based on Text Level and Presentation Format

 

 

 

 

 

 

 

 

 

 

Print Audio Video All Formats '
Level 3 Texts M: 12.52 M: 14.70 M: 16.78 M: 44.00
SD: 7.71 SD: 10.11 SD: 7.33 SD: 22.96
Level 5 Texts M: 14.19 M: 12.52 M: 15.07 M: 41.78
SD: 9.13 SD: 7.80 SD: 9.04 SD: 23.50
All Texts M: 26.70 M: 27.22 M: 31.85 M: 85.78
SD: 15.89 SD: 16.71 SD: 14.50 SD: 44.36

 

Means of the total number of inferences generated with the three presentation
formats at the two levels of text difﬁculty were analyzed using a within-subjects 2 (text

level) x 3 (presentation format) way analysis of variance. Variability in students’

47

inference-making attributed to presentation format was signiﬁcant, (F = 5.031, p =
.010)‘. As the data in Table 3.1 indicate, when numbers of inferences were combined
across both text levels, students generated the fewest number of inferences with texts
presented in print format (26.70), more with texts presented in audio format (27 .22), and
the greatest number of inferences with texts presented in video format (31.85). While
some students showed gains of just a small percentage of propositions, others showed
more notable increases (ﬁ'om 30% to over 100%) based on presentation format,
regardless of the order in which the format was presented. It is important point to bear in
mind is that the order in which students were exposed to print or audio or video formats
was randomly assigned, and while the trend indicates increases in inference-making from
print to audio to video conditions, this is not to imply that students were exposed ﬁrst to
print, second to audio and third to video.

While format effects were signiﬁcant, variability in students’ inference-making
attributed to text level was not signiﬁcant, (F = .699, p = .412). When inferences were
combined across all three presentation formats, students generated a mean of 44.00
inference-level propositions with the level 3 texts, and a mean of 41 .78 inference-level
propositions with the level 5 texts.

Although it appears that the ﬁndings indicate an overall trend, with print
prompting the fewest inferences and video prompting the most, it is important to note that
a straightforward interpretation of the main effects is complicated by a format by
difficulty level interaction (F = 3.097, p = .054). Speciﬁcally, while text difﬁculty
performed in one way in the audio and video formats—with easier passages eliciting

more inferences—this pattern was reversed in the print format, where the harder passages

 

l

elicited more inferences. This ﬁnding associated with the print format serves as the main
source of the interaction.

The sections that follow include closer examinations of the ﬁndings noted above
along with a post hoc analysis prompted by the interaction effect. Examples that
correspond to the most prevalent trends are presented to illustrate the main effects. In
addition, examples based on ﬁndings that deviate from the overall means are presented in

an effort to better understand the interaction effect.

Illustrating Predominant Patterns
In this section, patterns associated with the print and video contexts are followed
by analyses of data that call for a closer look at the audio context. Propositions for all
participants were included in the 2 x 3 way analysis of variance noted above. Table 3.2
reports patterns of inference-making for the twenty-seven students, noting the formats
that prompted the fewest to greatest numbers of inferences. These data highlight the

predominant patterns further illustrated by the tables and descriptive cases that follow.

49

Table 3.2

 

 

 

 

 

 

 

 

 

 

 

 

 

Patterns of Inference-Making
Lowest—Medium— No. No. students No. students No. students No.
Highest (number of students whose whose whose instances
inferences) whose inference- inference- inference- when this
inference- making making making pattern
making followed this followed followed this occurred in
followed pattern with this pattern pattern with a single
this pattern Level 5 texts when both Level 3 research
with Level inferences and Level 5 session (out
3 texts generated texts of 54
with Level sessions—
3 texts were 27 students,
combined 2 sessions
with per student)
inferences
generated
with Level
5 texts
PATTERN 1: Print Low,
Video High
Print—Audio—Video l3 5 8 3 18
PATTERN 2: Print High,
Video Low
Video—Audio—Print 1 5 l 0 6
PATTERN 3: Audio Low
Audio—Print—Vrdeo 2 3 6 1 5
Audio—Video—Print 1 8 2 0 0
PATTERN 4: Audio High
Print—Video—Audio 5 2 3 1 7
Print-AudiolVideo (equal) 1 1 0 o 2
Print/Audio (equa1)——Video 0 0 l 0 0
Audio—Plint/Vrdeo (equal) 2 o 1 o 2
Audio/Video (equa1)—Print 0 1 2 0 l
Video—Print—Audio 1 2 2 0 3
Video—Print/Audio (equal) 0 0 1 O 0
Video/Audio/Print (equal) 1 0 0 0 0

 

 

 

 

 

 

50

 

Predominant Pﬂems with PriMnd Video

ﬁttern 1: Print low. video high As noted in Table 3.2, when inferences were
totaled across text levels, eight of the twenty-seven participants made the fewest number
of inferences within the print context, more within the audio context, and the greatest
number within the video context, making this the most predominant trend. With just the
level 3 texts, data from thirteen students followed this general pattern, and with just the
level 5 texts, data from ﬁve participants matched the trend (making a total of eighteen
instances when this occurred out of the ﬁfty-four research sessions, two per student).
Finally, three participants followed this overall trend with both level 3 texts and level 5
texts. Table 3.3 reports the data from these three students, including information about
the story-format combinations and sequence of presentation. Such information is
provided in this ﬁrst table in order to illustrate how story-format combinations and
sequences varied by individual and will not be included in each subsequent table. The
table is followed by a case designed to illustrate the data from one of the students in a

more descriptive manner.

51

Table 3.3

Battern 1: Print Low, Video High

 

 

 

 

 

 

 

 

Student Level Print Audio Video
Madeline and the The Amazing Bone Doctor De Soto
Gypsies (presented (presented SECOND) (presented FIRST)
3 THIRD)
6 16
Len 4
A Story-A Story Strega Nona Why Mosquitoes Buzz
5 (presented SECOND) (presented THIRD) in Peoples’ Ears
(presented FIRST)
3 6 9
The Amazing Bone Madeline and the Doctor De Soto
(presented THIRD) Gypsies (presented (presented FIRST)
3 SECOND)
10 18
Clarissa 13
A Story-A Story Why Mosquitoes Buzz Strega Nona
5 (presented FIRST) in Peoples’ Ears (presented SECOND)
(presented THIRD)
4 14 17
The Amazing Bone Doctor De Soto Madeline and the
(presented THIRD) (presented FIRST) Gypsies
3 (presented SECOND)
20 24
Ravelle 25
Why Mosquitoes Buzz A Story-A Story Strega Nona
5 in Peoples’ Ears (presented THIRD) (presented FIRST)
(presented SECOND)
27 29 31

 

 

 

 

 

 

Before presenting the descriptive case of one student, Len, it is necessary to make
one disclaimer about these illustrative data. The within-subjects design employed in this
study meant that each student responded to a different text for each format; thus these
examples can only be regarded as generally illustrative of the patterns observed in the ﬁll]
analyses across all students—an analysis in which texts are balanced across formats. The
example of Len follows.

Lﬂ Len is one of three students who generated the fewest number of inferences
within the print context, more within the audio context, and the greatest number within

the video context with both level 3 and level 5 texts. For example, when the level 3 text,
52

Madeline and the Gypsies, was presented in print format (the third text presented), Len
was able to generate just a few inferences such as predicting that Madeline might become
a scientist because she is smart. When The Amazing Bone was presented in audio format
(the second text presented), Len was prompted to make more inferences such as
suggesting that the bone must be important to the main character, Pearl, because she went
out of her way to take care of it, and predicting that the bone might return to the witch
and turn into a boy. And when Doctor De Soto was presented in video format (the ﬁrst
text presented), Len generated the greatest number of inferences. He surmised that
Doctor De Soto was a good dentist who was motivated to help the fox because he felt
sorry for him. In addition, Len determined that Doctor De Soto and his wife were smart
because they devised a plan to ensure their safety. Finally, Len concluded that Doctor De
Soto valued doing his job well and was motivated to take good care of his family because
he was the husband.

Like Len, Clarissa and Ravelle also generated the fewest number of inferences
Within the print context and the greatest number within the video context, a pattern that
matches the trend when inferences are combined across text levels. Nevertheless, other

students’ inference-making patterns followed a slightly different course.

httem 2: Print high. video low. While the ﬁrst pattern of print low, video high
was more prevalent with level 3 texts for thirteen participants, this trend was reversed for
one student with level 3 texts, and for ﬁve students with level 5 texts. The numbers of
inferences these latter ﬁve students generated within each context are reported in Table

3.4, which is followed by a more descriptive illustration of one case.

53

Table 3.4

Pattern 2: Print High, Video Low

 

 

 

 

 

 

 

Student Level Print Audio Video
Katrina 5 35 30 25
Akisha 5 1 1 7 3
Lily 5 39 35 31
Roberta 5 12 10 6
Rosanna 5 8 6 5

 

 

 

 

 

 

KM Katrina is one of ﬁve students who generated the greatest number of
inferences within the print context, and the fewest within the video context with level 5
texts. For example, when Strega Nona, was presented in print format (the third text
presented), Katrina generated the greatest number of inferences. She was able to surmise
why Strega Nona lived alone, and why Anthony decided to disobey her instructions.
When A Story-A Story was presented in audio format (the ﬁrst text presented), Katrina
made fewer, more general inferences related to the moral of the story. And when Why
Mosquitoes Buzz in Peoples’ Ears was presented in video format (the second text
presented), Katrina generated even fewer inferences, simply evaluating the structure of
the text. It appears that Katrina, like four of her peers, had an easier time generating
inferences when more difﬁcult texts were presented in print, which is a shiﬁ ﬁ'om the

ﬁrst, most prevalent pattern (print low, video high).

Predominant Patterns with Audio Text: Feast or Famine
Additional ﬁndings deviate from the most predominant trend (print low, video

high), and illustrate the interaction effect noted above. While data from some students
54

correspond with the most prevalent pattern on an individual basis at either or both text
levels, data from other students reveal two alternative patterns that are hidden within the
overall analysis of the main effects. It is interesting to note that analyses of the numbers
of inferences generated by some students followed an alternative pattern that calls for a
closer consideration of the audio context. Since the overall trend suggests that, across
both text levels, there were more inferences generated within the video context than the
audio context, and more inferences generated within the audio context than the print
context, an analysis was run on the discrepancies between video over audio, and audio
over print. In other words, calculations were conducted to determine the differences
between the numbers of inference-level propositions generated within the video and
audio contexts. Likewise, differences were noted between the numbers of inferences
generated within the audio and print contexts. Figure 3.1 portrays these discrepancies

graphically.

55

Figure 3.1

Discrepancies Based on the Overall Trend of Increases from Print to Audio t9 Video

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Number ofInference-level Propositions
'5’ 8 o
Lily
Evan
A ub ry
Karina E
Veronica g
R o berta E
Katrina 9'
M issy ml m ‘3 5:
Alex :1 cg cg o
R 0 san n a § § '3-
. § § §
lu an m g cr 0
Ravelle """ g g g
g Aklsha g :5. g
o ' . ...,...;.;. a. n. 9‘
g C chat 5' 8 9.,
................. o o 5"
a Evette «3 «j» g
- ........... '1: an
Sarlna 3 §'— 8
L o u relle """""""""" '8’ 3, 8’
N a ri h a a a
.s a ........ ..... 5’ a E-
Clarissa ........... m 8
E rln ................ a
S h an ik a ............................... i.
............................... o
L °“ ........................... §
S e th ‘ ..... -...:.-.t.;.;.;.;.;.;.;.;.;._...;.W4...
C h a ['1 e s _.,.,._.;.;.;.;.;.;.;.,.;.;.;. ------ ' ......... ' ...........
Desiree .............................................................................
E v e re t t ................................................

 

 

56

It appears that for those students with the greatest variation in inference-
generation due to presentation format, the audio context was either particularly conducive
or problematic. For the students represented at the far ends of the graph, the audio
versions of the texts prompted notably greater or fewer numbers of inferences. That is,
students at the far right of the graph had a more difficult time generating inferences
within the audio context. While they were more successﬁil within the video context,
which corresponds to the general trend, they also were more successful within the print
context, which deviates from the overall trend. Alternatively, students at the far left of
the graph appeared to have an easier time making inferences within the audio context.
While they were less successful within the print context, which matches the general
trend, they also were less successﬁil within the video context, which deviates from the
overall findings. Students from both sides of the spectrum are represented in the sections
that follow.

Battern 3: Audio a_s__ an inference suppressing condition. Along with Figure 3.1,
Table 3.5 presents additional data on students for whom the audio context was noticeably
less fruitful in terms of inference-generation with both easy and more difﬁcult texts.
While there were three students for whom this was true with level 3 texts, and eleven
students for whom this was the case with level 5 texts (for a total of fourteen instances
across ﬁfty-four research sessions), there were three students for whom audio texts
prompted the fewest number of instances with both levels of text. For example, Charles

generated less than half the inferences with texts presented in audio format at both Level

57

3 and Level 5. The audio context appeared to present similar challenges for Desiree and
Everett.
Table 3.5

httern 3: Audio Low

 

 

 

 

 

 

 

 

Student Level Print Audio Video

3 l3 5 14
Charles

5 20 8 16

3 24 9 24
Desiree

5 16 9 15

3 9 8 12
Everett

5 14 12 3O

 

 

 

 

 

 

Battern 4: Audio as_ an inference enhancing condition. Finally, for some students
the audio context appeared to enhance their ability to generate inferences. While there
were ﬁve students for whom this was true with level 3 texts, and two students for whom
this was the case with level 5 texts (for a total of seven instances across ﬁfty-four
research sessions), there was one student for whom audio texts prompted the greatest
number of instances with both easy and challenging texts. For Evan, the audio context
prompted twelve more inferences than the print context, and ﬁve more inferences than
the video context. Along with these ﬁndings related to presentation format, other results

oﬁ‘er ﬁirther insight into the format by text level interaction effect noted above.

Examining Text Level More Careﬁilly
While the overall effect of text level proved not to be signiﬁcant, it was involved
in an interaction with format. Of particular interest is the ﬁnding within the print format

58

condition, where the average number of inference-level propositions generated with the
more challenging level 5 texts (14.19) was higher than the mean with the easier level 3
texts (12.52), however, the results of a simple effects ad hoc test indicate that the

variation due to text level was not signiﬁcant within the print condition ((F = 2.267, p =

.144).

Effects of Text Leveland Presentation Format on Various Sets of Inferences

Three additional overall tests were conducted to determine the direction for
further analysis. As described in the Methods section, inference-level propositions were
analyzed in a variety of ways. First, propositions were coded as one of six types of
inferences: predictions, explanations, associations, evaluations, inferences based on world
knowledge or beliefs, and personal connections. In addition, two aggregate variables
were created: text-connected inferences that combined the categories of prediction,
explanation, and association; and self-connected inferences that combined the categories
of evaluation, world knowledge or beliefs, and personal connection. Finally, inferences
were coded as either prompted or spontaneous. The results of overall tests examining the
effect of text level and presentation format on these three sets of inferences determined
which line of inquiry to pursue.

The ﬁrst test checked the effects of text level and presentation format on students’
generation of the six types of inferences. Means of the six types of inferences generated
with the three presentation formats at the two levels of text difﬁculty were analyzed using

a within-subjects 2 (text level) x 3 (presentation format) x 6 (inference type) way analysis

59

of variance. Table 3.6 reports the means of the six types of inferences across both text

levels and all three presentation formats.

Table 3.6

Six Qatggon'es of Inferences

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

73’ ._
S .8 g
c: g 8 g g g g, g
1% g 3 'g 3 a 6: =5
' a 3 2 E 8
Level Format 2 )3 £1 § 3: E3 E,
Print Mean 0.78 2.59 4.89 3.52 0.67 0.07 12.52
(SD) (1.37) (2.48) (2.93) (3.38) (1.62) (0.27) (7.71)
3 Audio Mean 1.48 2.11 5.78 4.33 0.89 0.11 14.70
(SD) (3.03) (2.19) (5.36) (4.59) (1.31) (0.42) (10.11)
Video Mean 1.30 3.19 5.89 5.11 0.96 0.33 16.78
(SD) (2.93) (2.20) (3.13) (5.10) (1.45) (1.21) (7.33)
Pﬁm Mean 1.11 2.63 5.26 4.56 0.59 0.04 14.19
5 (SD) (2.38) (1.86) (3.12) (5.47) (1.37) (0.19) (9.13)
And” Mean 0.59 2.59 3.85 5.00 0.26 0.22 12.52
(SD) (1.01) (1.93) (2.96) (5.49) (0.81) (1.15) (7.80)
Video Mean 1.96 3.04 4.11 5.44 0.52 0.00 15.07
(SD) (2.95) (1.91) (2.74) (6.03) (1.31) (0.00) (9.04)
Mean 3.56 7.89 16.56 12.96 2.52 0.52 44.00
3 (SD) (6.27) (3.25) (8.86) (11.71) (3.36) (1.28) (22.96)
ALLEN“ Mean 3.67 8.26 13.22 15.00 1.37 0.26 41.78
5 (SD) (5.04) (4.13) (6.85) (14.14) (2.69) (1.16) (23.50)
mm Mean 1.89 5.22 10.15 8.07 1.26 0.11 26.70
(SD) (3.14) (2.76) (5.15) (8.28) (2.73) (0.42) (15.89)
BOTH Audi Mean 2.07 4.70 9.63 9.33 1.15 0.33 27.22
levels ° (SD) (3.44) (2.76) (7.11) (9.12) (1.63) (1.21) (16.71)
W den Mean 3.26 6.22 10.00 10.56 1.48 0.33 31.85
(SD) (5.16) (2.83) (4.41 (9.76) (2.41) (1.21) (14.50)

 

The second test examined the inﬂuence of level and format on the aggregate

variables of text-connected and self-connected inferences. Means of the two aggregate

6O

 

variables generated with the three presentation formats at the two levels of text difficulty
were analyzed using a within-subjects 2 (text level) x 3 (presentation format) x 2
(aggregate variable) way analysis of variance. Table 3.7 reports the means of the two
aggregate of variables across both text levels and all three presentation formats.

Table 3.7

Text-Connectemd Self-Connected Inferences

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Total
inference-
Text-connected Self-connected level
Level Format inferences inferences propositions
Prim Mean 8.26 4.26 12.52
3 (SD) (4.36) (4.32) (7.71)
Audio Mean 9.37 5.33 14.70
(SD) (7.63) (5.48) (10.11)
Video Mean 10.37 6.41 16.78
(SD) (4.29) (5.84) (7 .33)
Print Mean 9.00 5.19 14.19
5 (SD) (5.22) (6.44) (9.13)
Audio Mean 7.04 5.49 12.52
(SD) (3.76) (5.94) (7 .80)
Video Mean 9.11 5.96 15.07
(SD) (5.26) (6.28) (9.04)
Mean 28.00 16.00 44.00
3 ALL Formats (SD) (11.94) (14.20) (22.96)
Mean 25.15 16.63 41.78
5 (SD) (11.05) (16.20) (23.50)
Prim Mean 17.26 9.44 26.70
(SD) (8.29) (10.40) (15.89)
BOTH A di Mean 16.41 10.81 27.22
levels “ ° (SD) (9.42) (10.42) (16.71)
Video Mean 19.48 12.37 31.85
(SD) (6.68) (l 1.02) (14.50)

 

 

Finally, a third overall test checked the variation in students’ spontaneous and
prompted inference-making due to text difficulty and presentation format. Means of
spontaneous and prompted inferences generated with the three presentation formats at the
two levels of text difﬁculty were analyzed using a within-subjects 2 (text level) x 3

(presentation format) x 2 (response type) way analysis of variance. Table 3.8 reports the

61

means of both spontaneous and prompted inferences across text levels and presentation
formats.

Table 3.8

Spontaneous and Prompted Inferences

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Total

Spontaneous Prompted inference-level

Level Format inferences inferences propositions
3 Print Mean 3.77 8.74 12.52
(SD) (4.06) (4.46) (7.71)
Audio Mean 4.93 9.78 14.70
(SD) (5.77) (5.00) (10.11)
Video Mean 6.22 10.56 16.78
(SD) (5.39) (3.83) (7.33)
Print Mean 4.63 9.52 14.19
5 (SD) (5.83) (4.74) (9. 13)
Audio Mean 3.19 9.33 12.52
(SD) (3.74) (5.61) (7.80)
Video Mean 4.44 10.67 15.07
(SD) (4.68) (5.91) (9.04)
3 Mean 14.93 29.07 44.00
ALL Formats (SD) (13.49) ( 10.94) (22.96)
5 Mean 12.26 29.52 41.78
(SD) (12.68) (12.64) (23.50)
Print Mean 8.41 18.26 26.70
(SD) (9.03) (8.12) (15.89)
BOTH Audi Mean 8.11 19.11 27.22
levels 0 (SD) (8.66) (9.40) (16.71)
Video Mean 10.67 21.22 31.85
(SD) (9.22) (7.93) (14.50)

 

 

 

 

 

 

 

As expected, in all three tests, variability in students’ inference—making attributed
to presentation format was signiﬁcant, (F = 5.03 1, p = .010), (F = 5.031, p = .010), and
(F = 5.186, p = .010), respectively. Likewise, as with the previous 2 x 3 way ANOVA,
variability in students’ inference-making attributed to text level was not signiﬁcant in any
ofthe three tests, (F = .696,p = .412), (F = .696,p = .412), and (F = .695,p = .412),
respectively.

The ﬁrst test, the 2 (level) x 3 (format) x 6 (inference type) way ANOVA, was

identiﬁed as having the potential to explain more of the variation since it was the only

62

test that showed signiﬁcant effects associated with text level or presentation format (a
text level by inference-type interaction). Therefore, the section that follows takes a closer

look at the six subcategories of inferences identiﬁed in the Methods chapter.

Effects of Text Level and Presen4ation F ormat on Six Inference Subcategories

The six types of inferences were part of the within-subjects 2x3x6-way analysis of
variance which identiﬁed a format effect, a type of inference effect, and a text level by
inference type interaction. Noted differences among means presented above in Table 3.6
prompted a post hoc analysis using the Least Signiﬁcant Difference test with a six-level
variable (1=Grade 3 text in print, 2=Grade 3 text in audio, 3=Grade 3 text in video,
4=Grade 5 text in print, 5=Grade 5 text in audio, 6=Grade 5 text in video). This test was
used to determine which means were signiﬁcantly different. The six-level variable was
created because post hoc analyses on general linear models with repeated measures
require a designated between-subjects factor. It was not possible to run this sort of
analysis with just format and text difﬁculty because there are only two text levels, thus
the six-level variable was created. Three mean differences were statistically signiﬁcant,
and all three differences were associated with the audio versions of the more challenging
level 5 texts.

First, students generated fewer association inferences with the audio versions of
the level 5 texts than with the video versions of the level 3 texts. On average, students
generated 3.85 association inferences with the level 5 audio versions, but 5.89 with the

video versions of the easier level 3 texts, and these differences were signiﬁcant (p = .033)

63

Second, students generated fewer association inferences with the audio versions
of the level 5 texts than with the audio versions of the level 3 texts. Again, on average
students generated 3.85 association inferences with the level 5 audio versions, but 5.78
with the audio versions of the easier level 3 texts, and these differences were signiﬁcant
(p = .044).

Finally, students generated fewer prediction inferences with the audio versions of
the level 3 texts than they did with the video versions of the level 5 texts. On average,
students generated 0.59 prediction inferences with the level 5 audio versions, but 1.96
with the video versions of the more challenging level 5 texts, and these differences were

signiﬁcant (p = .039).

Associ_ation-level inferences with the amdio format; Two student cases illustrate
instances when the level 5 audio texts prompted signiﬁcantly fewer association-level
inferences than level 3 audio or video texts. For example, with the level 5 audio text, A
Story-A Story, Evette was unable to make a single association. On the other hand, with
the level 3 video text, The Amazing Bone, and the level 3 audio text, Doctor De Soto, she
was able to generate a total of 16 associations related to the roles that characters played in
the respective plots. Similarly, when Roberta listened to the level 5 text, A Story-A Story,
she was unable to generate more than 3 association-type inferences. However, with the
level 3 video text, Madeline and the Gwsies and the level 3 audio text, Doctor De Soto,
Roberta was able to make 16 associations related to character’s personalities. For Evette
and Roberta, it seems that the audio form of the more challenging level text presented a

particular challenge with making association-related inferences, and the overall ﬁndings

64

represented by these two examples help to explain the text level by inference type

interaction.

Predictions with the audio forma; Two additional cases illustrate occasions when
notably fewer predictions were made with the audio versions of level 5 texts than with
the video versions of level 5 texts. For example, when Juan was working with the level 5
audio text, A Story-A Story, he made just 2 predictions. On the other hand, with the level
5 video text, Strega Nona, he was able to generate 9 predictions about ﬁrture story events.
Likewise, when Narisha was engaged with the level 5 audio text, Mry Mosquitoes Buzz
in Peoples ’ Ears, she was unable to make a single prediction. However, with the level 5
video text, A Story-A Story, she was able to make 7 predictions. For Juan and Narisha, it
appears that with difﬁcult texts, the audio format presented a particular challenge when
making predictions.

Given earlier analyses of how the audio condition inﬂuenced overall inference-
making, these ﬁndings on how this condition appeared to inﬂuence these speciﬁc
categories of inferences is perhaps not surprising, nevertheless the results help to identify

the type of inferences that were more affected by this medium.

Analyses of Effects of Story. Order. & Story-Format Combinations
A second set of analyses was conducted to rule out any spurious effects of story,
and particular story-by-format, order-by-story, or order-by-format combinations that
might compromise a straightforward interpretation of the main effects. These secondary

analyses indicated that the inﬂuence of story, story order, format order, and story-format

65

combination on students’ overall inference-making was not signiﬁcant. The results of

these analyses are presented in Tables 3.9 through 3.12.

Effect of Story on Inference-MaLing

Table 3.9 reports the means of students’ total inference-level propositions by
story, however, a test of between-subjects effects indicates that the variability in students’
overall inference-making based on story was not signiﬁcant (F = .413, p = .839).
Table 3.9

Inferences by Story

 

 

 

 

 

 

 

 

 

 

 

 

 

Level Story N Mean Std. Deviation
The Amazing Bone 27 14.37 8.43
3 Doctor De Soto 27 14.93 9.51
Madeline and the Gypsies 27 14.70 7.95
A Story-A Story 27 12.89 10.64
5 Why Mosquitoes Buzz in Peoples’ Bars 27 13.33 6.83
Strega Nona 27 15.56 8.08

 

Effect of Story Order on Inference-Making.

Table 3.10 reports the means of students’ total inference-level propositions by the
order in which the story was presented. For example, The Amazing Bone was the ﬁrst
story for nine of the participants. For those nine students, the mean number of total
inference-level propositions was 16.33. A test of between-subjects effects indicates that
the variability based on the order in which a story was presented was not Signiﬁcant (F =

1.287, p = .247).

66

Table 3.10

Inferences by Order of Story

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Order Story Mean N Defitgtion
First The Amazing Bone 16.33 9 7.70
Doctor De Soto 19.38 8 14.96
Madeline and the Gypsies 11.90 10 5.11
A Story-A Story 7.56 9 5.61
Why Mosquitoes Buzz in Peopleﬁ’ Ears 15.90 10 8.28
Strega Nona 15.13 8 8.58
Second The Amazing Bone 13,44 9 10.49
Doctor De Soto 11.89 9 5.30
Madeline and the Gypsies 16.78 9 8.93
A Story-A Story 13.67 9 11.19
Why Mosquitoes Buzz in Peoples’ Ears 12.88 3 7.6g
Strega Nona 16.80 10 7.18
Third The Amazing Bone 13.33 9 7.43
Doctor De Soto 14.10 10 5.88
Madeline and the Gypsies 15.88 8 9.58
A Story-A Story 17.44 9 12.45
Why Mosquitoes Buzz in Peoples’ Ears 10.89 9 2.30
Strega Nona 14.56 9 9.32

 

 

 

Effect of Presentation Format Order on Inference-Mak_ing

Table 3.11 reports the means of students’ total inference-level propositions by
order of presentation format. Participants were exposed to each presentation format two
times—once with a Level 3 text, and again with a Level 5 text. For example, there were

seventeen instances when the ﬁrst text was presented in print format, and as stated in the

67

Methods section, students were never assigned the same order of presentation format for
both research sessions. A test of between—subjects effects indicates that the variability of
students’ inference-making based on the order of presentation format was not signiﬁcant
(F = 2.327,p = .061).

Table 3.11

Inferences by Order of Presentation F orrnat

 

 

 

 

 

 

 

 

 

 

Order Presentation format Mean N Std. Deviation
Fir“ Pm“ 11.41 17 5.73
““110 17.56 18 12.69
Vid°° 13.63 19 6.64
sewn“ Pﬁm 15.32 19 10.46
Audi° 11.89 18 5.84
Vid°° 15.76 17 8.32
Third Prim 13.11 18 8.10
Audi° 11.39 18 5.77
““60 18.50 18 9.20

 

 

 

 

 

 

 

Effect of Story-Format Combinations on Inference-Making

Finally, Table 3.12 reports the means of students’ total inference-level
propositions by story-presentation format combinations. For example, ten participants
were exposed to the print version of The Amazing Bone, nine participants were assigned
to the audio version of this same text, and so on. A test of between-subjects effects
indicates that the differences based on the particular combinations were not signiﬁcant (F

= 1.453, p = .167).

68

 

Table 3.12

Inferences by Story-Format Combinations

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Story tPresentation Sid--
ormat Mean N Devratron
Print 14.40 10 10.05
The Amazing Bone Audio 13.56 9 9.29
Video 15.25 8 5.82
Print 12.50 8 7.07
Dr. De Soto Audio 17.90 10 13.60
Video 13.78 9 4.84
Print 10.44 9 5.15
Madeline and the Gypsies Audio 1200 8 4.34
Video 20.70 10 8.93
Print 13.89 9 13.84
A Story-A Story Audio 11.56 9 8.13
Video 13.22 9 10.34
Print 15.75 8 5.90
Why Mosquitoes Buzz in PoopleS’ 5313 Audio 13.10 10 8.16
Video 11.44 9 6.04
Print 13.20 10 6.34
Strega Nona Audio 12.88 8 7.95
Video 20.56 9 8.34

 

 

 

 

 

Given that the effects of story, and particular story-by-format, order-by-story, or

order-by-format combinations were not signiﬁcant, a straightforward interpretation of the
main effects and interactions of interest was possible.

Finally, while this study was designed to investigate the main effects of text level

and presentation format on students’ higher level processing, the last section of this

69

 

chapter examines the inﬂuence of difﬁculty and medium on students’ recall-level

processing.

Analyses of Recall-Level Data
This last section examines the main effects of text level and presentation format
on students’ recall using a 2 (text level) x 3 (format)ANOVA. As deﬁned in Table 2.8 of
the Methods section, recall-level propositions refer to verbalizations that either match or
paraphrase content found in the text. Table 3.13 reports the means of recall—level
propositions for both text levels and each presentation format.
Table 3.13

Within-Subjects Analysis of Recall Based on Text Level and Presentation Format

 

 

 

 

 

 

 

 

 

 

Print Audio Video All Formats
M: 20.37 M: 18.04 M: 13.74 M: 52.15
Level 3 Tex“ SD: 23.07 SD: 16.16 SD: 13.50 SD: 43.98
M: 14.67 M: 13.19 M: 18.19 M: 46.04
Level 5 Tem SD: 14.27 SD: 14.56 SD: 18.99 SD: 42.24
M: 35.04 M: 31.22 M: 31.93 M: 98.19
A“ Tam SD: 32.42 SD: 28.74 SD: 32.00 SD: 82.69

 

While neither text level (F = 1.684, p = .206) nor format (F = .358, p = .701)
elicited a statistically reliable effect on recall, they did produce a signiﬁcant interaction
effect (F = 4.254, p = .019). The source of the interaction is illustrated in Table 3.13; for
level 3 texts, video proved to be the least effective format for promoting the recall-level
propositions, while it was the most effective for level 5 texts. It appears that for easy
texts, the print condition prompted more recall, whereas for harder texts, the video was

more conducive to recall.

70

This chapter has presented three sets of ﬁndings related to a) the effects of
presentation format and text difficulty, b) the potential inﬂuences of story, the order of
story or presentation format, and story-format combinations, and c) recall-level data.
While format effects were signiﬁcant and text level effects were not, it is important to
bear in mind that the interaction effect complicates these ﬁndings, and it appears that
individual differences play an important role, especially since the inﬂuence of story,
format, sequence, and story-format combinations was not signiﬁcant. That is,
determining which condition (print, audio, or video) was more helpful or problematic,
overall, was an impossible task. While some students appeared to have an easier time
generating inferences within the audio context, others struggled. Similarly, ﬁndings
associated with recall-level data, also resulted in an interaction effect that prompted a
closer consideration of individual differences since there was no signiﬁcant effect due to
text level or presentation format. It is interesting to note that for some students print texts
elicited more recall-level propositions with easier texts, while video texts prompted more
recall with more challenging texts. In the following chapters, possible explanations for

such ﬁndings are presented, along with potential limitations of the study.

71

LIMITATIONS

The limitations of this study reﬂect many of the general limitations associated
with the use of verbal reports as data. First, one must consider the potential effects due to
the nature of the one-on-one interview context. Afﬂerbach states that there is “no more
intimate reading research methodology than protocol analysis,” and suggests that
students’ verbalizations may be inﬂuenced by the student-interviewer relationship,
cultural differences in using language, and differences in how students View their roles as
reporters of reading processes (Afﬂerbach, 2000, p. 172). In addition, the interview
context may have created a sense of pressure to elaborate on stories which, as Beishuizen
suggests, may cause readers to infer more, or even fewer, relationships than they would
in a natural reading situation (Beishuizen et al., 1999).

In addition, the transcribed verbal reports did not include data related to the
differences inherent in spoken discourse. During the interviews students made use of
intonation, inﬂection, and pauses that were not accounted for in the transcribed protocols
(Afﬂerbach & Johnston, 1984).

Finally, although there was an attempt to gather an additional source of data in the
form of student and parent surveys, the low percentage of return resulted in a failure to
compare these data with those gathered ﬁom the interviews. Moreover, there was no
opportunity to demonstrate correlations between more objective measures of inference-
making and the students’ verbal reports (Afﬂerbach, 2000).

An additional limitation is related to the difﬁculty in ﬁnding students willing to
participate that ﬁt within the small window of comprehending at the 3rd grade level

according to a single measure, Flynt & Cooter’s Reading Inventory for the Classroom.

72

Although seven school sites were approached, only ﬁve sites had one or more teachers
who were willing to have their students pre-assessed to see if they qualiﬁed for the study.
While a total of 145 students were pre-assessed, only 63 students qualiﬁed and were
invited to participate. The 27 students that returned signed consent forms participated in
the two research sessions. AS a result, there were not the 36 students required to ﬁll
every cell within the story-format-order matrices displayed in the Methods section in
Table 2.4 and Table 2.5. Therefore, while the random selection process was
counterbalanced to ensure maximally comparable groups, the matrix, or block, remained

incomplete.

73

DISCUSSION

Despite the limitations acknowledged in the previous chapter, ﬁndings from this
study do advance our understanding of the role of text difﬁculty and presentation format
on inference-level processes, especially for populations of students experiencing
difficulty in mastering reading. They also provide direction for ﬁrrther investigations and
may contribute to our knowledge of how such higher level processes might be fostered in
elementary classrooms, although the study was not intended to provide any direct test of
instructional interventions. This study set out to build upon the existing research on
students’ comprehension, focusing on the higher order process of inference-making. The
within-subjects design of this investigation allowed for the efﬁcient analyses of a range
of factors: presentation format, text level, story, format order, story order, and format-
story combinations. In addition, this study examined the inﬂuence of these factors on
students’ overall inference-making, as well as students’ generation of various types of
inferences and recall-level propositions. A discussion of the results is presented in three
parts, beginning with a closer look at the ﬁndings related to presentation format, followed
by results associated with text level, and concluding with a set of implications for both
research and practice. The discussion examines the ﬁndings in light of concepts

associated with the various models of reading outlined in the introduction.

The Inﬂuence of Presentation Format
The overall test, a 2x3x6-way analysis of variance, concluded that the mode of

presentation had a statistically signiﬁcant effect on students’ generation of inferences,

74

mm

 

albeit complicated by an interaction between format and text difﬁculty level. The
following section examines this ﬁnding with a discussion of the most prevalent patterns
of inference-making by presentation format as well as an examination of the real numbers
at both text levels.

On one hand, the most predominant pattern ——students generating the fewest
inferences within the print context—lends support to the notion that individuals possess a
limited capacity for different levels of cognitive processing. That is, with the struggling
readers in this study, as the need for lower order perceptual processes like decoding
decreased, engagement in higher order processes like inference-making increased. In this
way, it appears as though these results help explain the story of Maria, 8 student from the
introduction. Observations of Maria’s inference-making resemble the data on students’
overall inference-making. Like several students in this study, Maria appeared to generate
more inferences within the audio and video conditions than within the print condition.
Nevertheless, while the variation for many students followed this general pattern, other
students followed somewhat different paths that are masked by the main effects and

visible only through an analysis of the real numbers.

While the analysis of variance determined that the effect of presentation format on
overall inference-making was statistically signiﬁcant when the numbers of inferences
with both text levels were combined, the actual numbers suggest alternative patterns.

This section takes a closer look at the actual numbers of inferences generated by the

twenty-seven participants at each text level across the various presentation formats.

75

In the Results chapter, Table 3.2 reported the numbers of students whose
inferences increased according to various format-related progressions. For example, with
the easier texts, several students (thirteen total) generated the fewest number of
inferences within the print condition and the greatest number within the video condition,
when the numbers for both text levels were combined (Level 3 + Level 5). On the other
hand, for eight students, the progression shifted with the more difﬁcult level 5 texts. That
is, with the level 5 texts, these eight students generated the fewest number of inferences
within the audio context, and greater numbers within the video and print contexts.

This shift in the progression warrants a closer look at the concept of “limited
capacity.” While this general notion offers an explanation for the ﬁndings associated
with the easier level 3 texts, it is unable to account for the results associated with the
more challenging level 5 texts in the same way. When students were interacting with the
more difﬁcult texts, it appears that the print context—which places a burden on lower
level decoding processes and presumably leaves fewer cognitive resources available for
higher level processes—prompted the greatest number of inferences for thirteen students.
On the one hand, these ﬁndings challenge the limited capacity idea which implies that
eliminating the need for lower level perceptual processes leaves more attentional
resources available for higher level processes.

On the other hand, these ﬁndings can lend support to a general limited capacity
notion which acknowledges the possibility that although audio and video may alleviate
the need for decoding, these media may require other perceptual and integrative
processes that leave fewer resources available for inference-making. As Magliano and

his colleagues suggest, understanding a story depicted in video form requires an

76

individual to integrate visual, auditory, and discourse information (Magliano, Dijkstra, &
Zwaan, 1996), and this integration process may make the video context all the more
challenging.

And ﬁnally, it also could be the case that inference-making is more of a
compensatory move that readers employ when their understanding is more, rather than
less, tenuous. In this way, ﬁndings related to the predominant trends are difﬁcult to apply
to particular comprehension models, or even to general concepts upon which they are
based. Instead, these ﬁndings appear to highlight the ways in which presentation format

had a range of effects for different students.

Just as the print context rendered variable results for different students, the audio
context was a source of notable variation as well. When texts were presented in audio
format, inference generation was inﬂuenced in a range of ways. As noted in the Results
chapter, some students followed the most prevalent pattern and generated more
inferences with audio than print, which required decoding, and fewer inferences with
audio than video, which offered visual and auditory content. For other students,
however, the audio context appeared to be particularly conducive or problematic.

Some students were able to generate notably greater or fewer numbers of
inferences when texts were presented in audio format. For some students who found the
print or video context to be particularly conducive to inference-level processes, they
found the audio context to be much less so. Conversely, for other students, the print and
video contexts were more problematic, while the audio context was much more

favorable. Once again, these ﬁndings can be interpreted in more than one way. While

77

 

the ﬁrst ﬁnding can support the assertion that the audio context may be more demanding
than the print and video contexts (for example because of the fact that there is nothing to
attract one’s visual attention), it could be the case that more inferences were generated
within the print and video conditions because readers were attempting to assist their own
construction of meaning by engaging in more inference-level thinking aloud.
Similarly, while the second ﬁnding may suggest that the video and print contexts present
more of a challenge than the audio context, it is possible that the audio context prompted
more inferences because it was more, not less, challenging. For example, as mentioned
in the introduction, when listening to a text, the individual is bound to the rate of textual
input determined by the speaker, or in this study, by the audiotape. However, when
reading a text, the individual controls the rate of textual input and has the option of
rereading portions of the text (Carlisle & Felbinger, 1991). The cognitive processes
required to comprehend within the constraints associated with the rate as well as the
transitory nature of audio-taped text, may leave fewer resources available for inference-
making and render the audio context more challenging. Moreover, the fact that some
students were able to generate more inferences within the print context than within the
audio context suggests that perhaps decoding processes do not occur at the expense of
inference level processes, and our attentional resources are not so limited after all.

In addition, this second ﬁnding lends support to elements of Kintsch’s
commotion-integration model outlined in the introduction. This model suggests that a
text base is constructed from linguistic input as well as from an individual’s knowledge

base, and then integrated into a coherent whole—what Kintsch calls the Situation model

78

(Kintsch, 1994). Within the video context, this text base can be supplemented with

additional visual input that is unavailable within the print or audio contexts.

While results related to students’ overall inference-making suggest that the audio
context may prompt either greater or fewer numbers of inferences with texts at different
levels, it is interesting to note that the only statistically signiﬁcant ﬁndings related to
specific types of inferences were associated with the level 5 audio texts, which appeared
to present a particular challenge. As noted in the Results chapter, students generated
signiﬁcantly fewer predictions and association-based inferences with the more
challenging level 5 texts only when presented in audio form. The fact that the notable
differences are related to these two types of inferences is not surprising when one
considers earlier research which suggests that various types of inferences interact. For
example, Trabasso and his colleagues claim that association-level inferences may be
activated by other inferences (Trabasso & Magliano, 1996). This assertion provides not
only a possible explanation as to why association-level inferences were most prevalent,
but also a possible rationale for the ﬁnding that this type of inference was generated
signiﬁcantly less often within a particular media context. If association-level inferences
are activated by other inferences, perhaps differences in this most prevalent inference-
type within a particular context is even more noteworthy. That is, if association-type
inferences are activated by other inferences, the fact that students generated signiﬁcantly
fewer associations within a particular context (audio), suggests that students are

generating fewer activating inferences within that context as well.

79

Similarly, this notion of inference interaction can be used to explain the ﬁnding
that students made signiﬁcantly fewer prediction-level inferences within the level 5 audio
context. Van den Broek and his colleagues suggest that inferences are not independent
and that forward inferences, such as predictions, may affect other inferential processes
(van den Brock, Fletcher, & Risden, 1993). Once again, if it is the case that predictions
interact with other inference-level processes, the fact that students generated notably
fewer predictions within the level 5 audio context is perhaps even more signiﬁcant, and

perhaps the audio context with more challenging texts presents a particular challenge.

The Inﬂuence of Text Level

While text difﬁculty appeared to inﬂuence the effect of presentation format in
some respects, it is surprising that across all three media contexts, text level appeared to
play a less signiﬁcant, and sometimes even unpredictable, role. As noted in the Results
chapter, the effect of text level on students’ overall inference-making was not statistically
signiﬁcant. And a closer look at the means for each presentation format reveals even
more unexpected ﬁndings that warrant ﬁirther discussion.

The ﬁrst surprise is that within the print context, students generated more
inferences with the more challenging level 5 texts than they did with the easier level 3
texts. This ﬁnding challenges the limited capacity model, which would predict that
students would engage in less inference-making with challenging texts, especially those
in print format, that require them to decode unfamiliar words and comprehend longer,

more complex sentences.

80

In addition, this ﬁnding raises additional questions about the role of inference-
making in narrative comprehension. As mentioned above, while it may be true that
increases in inference-making indicate a deeper level of comprehension, it may be the
case that inference-making increases when students experience greater difficulty building
a straightforward text base from the transparent recall of explicitly stated propositions;
that is, as suggested by a compensatory view of inference-making, students have to ﬁll in
more gaps when the text information does not stay with them long enough to build that a
stable model of meaning.

In essence, two possible motives for increased inferencing in this situation present
themselves. On the one hand, readers may be generating inferences that are allowing
them to maintain coherence and better understand the text (Thurlow & van den Broek,
1997). On the other hand, readers may be generating inferences because the knowledge
required to assimilate new information is not available, and the only recourse is to make
deductions that are hopefully useful (Trabasso & Sub, 1993). So while it might be
expected that readers would engage in Minimalist reading with less inferential activity
when they are interacting with texts that are less conducive to comprehension because the
text content, structure, or vocabulary presents a particular challenge (as with the level 5
texts), and in Maximalist reading with texts that contain more familiar content and
vocabulary (as with the level 3 texts), these ﬁndings call these notions into question and

prompts a closer consideration of inference-making as a measure of comprehension.

Equally interesting is the ﬁnding that students generated more recall-level

propositions with the level 5 video texts than the level 3 video texts. While one might

81

assume that the less challenging video text would have allowed for more recall, it may be
the case that because there was more inference-making occurring within this context,
students were less likely to include additional recall information as well. In addition, it
may be the case that since video is perceived as the ‘easy’ medium (Beagles-R008 & Gat,
1983), less effort was invested in making recall-level responses with the easier texts and
more emphasis was placed on raising the level of the task by engaging in more inference-
level activity. While no data were gathered about these possibilities, it is interesting to
consider what factors may have inﬂuenced students’ tendencies to either recall or make
inferences. Likewise, it is interesting to consider how researchers, practitioners, and
students might deﬁne comprehension. That is, what proportion is recall-level activity and
what proportion is inferential-level activity? The section that follows describes

implications for each of these groups.

Implications

The results of this study both challenge and support elements of the various
models of reading outlined in the introduction, however, the ﬁndings leave the implicit
question in this study clearly unanswered—which model of reading is more accurate,
complete, or worthy? While this investigation surely answers the original research
questions related to the types and frequencies of inferences generated across a range of
presentation formats and text levels, between the lines was the pursuit of a reasonably
complete model of comprehension and one was not found. Perhaps not surprisingly, the

ﬁndings remained closely tethered to the original research questions, and have simply

82

'I‘H -. 1m

 

 

 

 

 

111'

for

initiated a journey toward a fuller understanding of the differential effects of presentation
format for twenty-seven struggling readers.

Implications for further research center around pursuing this line of inquiry in an
effort to broaden the deﬁnition of what it means to comprehend by examining
individuals’ meaning-construction processes within a range of media conditions.
Ultimately, such investigations could lead to an expanded taxonomy of reading
difficulties, one that includes more than just word recognition, reading comprehension,
and listening comprehension—factors measured in many assessments of reading
difﬁculty and addressed in various models of reading. For instance, as outlined above,
the Simple View of Reading considers two of these three factors and suggests that weaker
readers struggle with word recognition, or listening comprehension, or both (Chen &
Vellutino, 1997; Gough & Tunmer, 1986). Similarly, Carlisle points out that while
reading disability is identiﬁed when a student’s listening comprehension is age
appropriate and reading comprehension is weaker, a general comprehension problem is
identiﬁed when a student’s reading and listening comprehension are below grade level
(Carlisle & Felbinger, 1991).

The ﬁndings from this study suggest that there may be more to consider. The data
highlighting such a broad range of individual differences call for a closer look at how
comprehension, or at least inference-making, is inﬂuenced by various media, and how
this variability might offer insights into a broader range of comprehension abilities and
disabilities. For example, it would be worthwhile to further investigate comprehension
processes of those students who found the audio and video contexts to be either

particularly conducive or problematic. Results could be plotted on a graph with the audio

83

 

 

data on the x-axis and the video data on the y—axis, and follow-up studies could be
conducted with students falling into four broad categories (high-audio/high-video; high-
audio/low-video; low-audio/high-video; and low-audio/low-video). Such investigations
might lead to a broader deﬁnition of comprehension that allows for a wider range of
explanations for students’ struggles with comprehension

Finally, although this was not a study of classroom instruction, ﬁndings from this
investigation may suggest possibilities for practice that build upon this notion of an
expanded taxonomy of reading difﬁculties. Just as the explanations for comprehension
ability expand, so must the assessments used to measure these abilities. That is, if it is
the case that—as these ﬁndings suggest—students’ abilities to engage in higher level
processes like inference-making are signiﬁcantly inﬂuenced by presentation format in
rather individual ways as noted by the format by level interaction effects, it becomes
important to assess these abilities in a more comprehensive manner. In other words, one
could argue that a more comprehensive understanding of students’ comprehension
abilities with print, audio, and video texts would result in more effective instructional
decisions. However, it is not yet determined that simply knowing more about a child’s
abilities in terms of word recognition, silent reading comprehension, listening
comprehension, and viewing comprehension, and then “ﬁlling the gaps” with
corresponding strategy instruction is the most effective route.

Nevertheless, the results of this study do highlight the existence of individual
differences through an investigation of the nature and prevalence of various inferences
generated across a range of presentation formats and levels of text. While the ﬁndings

were unable to suggest a single model of reading to explain students’ comprehension

84

 

 

 

processes, they do forge paths for future studies that may broaden our understanding of
inference and recall level activities that contribute to this phenomenon called

comprehension.

85

REFERENCES

86

REFERENCES

Aardema, V. (1992). Why mosquitoes buzz in peoples' eel; New York: Pufﬁn.

Afflerbach, P. (2000). Verbal reports and protocol analysis. In M. L. Kamil, P. B.
Mosenthal, P. D. Pearson, & R. Barr (Eds), Handbook of Reading Research (Vol. III, pp.
163-179). Mahwah, NJ: Lawrence Erlbaum Associates.

Afﬂerbach, P., & Johnston, P. (1984). Research methodology on the use of verbal
reports in reading research. Journal of Reading Behaviorpl6(4), 307-322.

 

Allington, R L. (1991). Children who ﬁnd learning to read difﬁcult: School
responses to diversity. In E. H. Hiebert (Ed), Literacy for a diverse society: Perspectives.
practices, and policies (pp. 23 7-252). New York: Teachers College Press.

Anderson, R. C., & Pearson, P. D. (1984). A schema-theoretic view of basic
processes in reading comprehension. In P. D. Pearson (Ed), HmdbooLof reading
research (pp. 25 5-291). New York: Longman.

 

AutoSkill. (2003). Academy of Reading: AutoSkill International, Inc.

Beagles—R005, J ., & Gat, I. (1983). Speciﬁc impact of radio and television on
children's story comprehension. Journal of Educationallsychology, 75(1), 128-137.

Beal, C. R. (1990). Development of knowledge about the role of inference in text
comprehension. Child Development. 61. 1101-1023.

Beishuizen, J ., Grand, J. L., & Schalk, J. (1999). No correlation between
inferencing causal relations and text comprehension? L__earningand Instruction, 9. 37—56.

 

Bemelmans, L. (1959). Madeline and the gypsies. New York: Viking Press.

Brown, A. L., Palincsar, A. S., & Armbruster, B. B. (1994). Instructing
comprehension-fostering activities in interactive learning situations. In R. B. Ruddell, M.
R. Ruddell, & H. Singer (Eds), Theoretical models and proceases of reapipg (4th ed., pp.
757-787). Newark, DE: International Reading Association.

Brunet, J. (1985). Models of the Learner. Educational Researcher. 14(6), 5-8.

Carlisle, J. F., & Felbinger, L. (1991). Proﬁles of listening and reading
comprehension. Journ_al of Educational Research. 34(6), 345-354.

87

Cartwright, K. B. (2002). Cognitive development and reading: The relation of
reading-speciﬁc multiple classiﬁcation skill to reading comprehension in elementary
school children. Journal of Educational Psychology, 94(1), 56-63.

Chall, J. S., Bissex, G. L., Conrad, S. S., & Harris-Sharples, S. (1996). Qpalitative
assessment of text difﬁculty: A praakalguide for teachers and writers. Cambridge, MA:
Brookline Books.

Chen, R. S., & Vellutino, F. R. (1997). Prediction of reading ability: A cross-
validation study of hte simple view of reading. Journal of Literacy Research, 29(1), 1-24.

CRLP. (1996). Pre-K-12 RESULTS program: California Reading & Literature
Project.

Curtis, M. E. (1980). Development of components of reading skill. qurnal of
Educational Psychology, 72(5), 656-669.

de Paola, T. (1975). Strega Nona. New York: Simon & Schuster.

Ericsson, K. A., & Simon, H. A. (1980). Verbal reports as data. Pachological
Review 87(3), 215-251.

 

Ericsson, K. A., & Simon, H. A. (1993). Protocol analysis: Verbal reports as data.
Cambridge: The MIT Press.

 

Fitzgerald, J. (1995). English-as-a-second-language reading instruction in the
United States: A research review. Joumi of Reading Behavior. 27. 115-152.

F lynt, E. S., & Cooter, R. B. (1998). Reading Inventory for the Classroom. (Third ed.).
Upper Saddle River, NJ: Merrill.

Frawley, W. (1992). Lingaistic Semantics. Hillsdale, NJ: Lawrence Erlbaum
Associates.

Gough, P. B., & Tunmer, W. (1986). Decoding, reading, and reading disability.
Remedial and Special Education. 7, 6-10.

 

Graesser, A. C., & Kreuz, R. J. (1993). A theory of inference generation during
text comprehension. Discourse Processas. 16. 145-160.

Graesser, A. C., Singer, M., & Trabasso, T. (1994). Constructing inferences
during narrative text comprehension. Psychological Review, 101(1), 371-395.

Hannon, B., & Daneman, M. (1998). Facilitating knowledge-based inferences in
less-skilled readers. Contemporary Educational Psychology, 23, 149-172.

88

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(1), 103-128.

Hayley, G. (1988). A story, a story. Glenview, IL: Scott Foresman.

Kintsch, W. (1994). The role of knowledge in discourse comprehension: A
construction-integration model. In R. B. Ruddell, M. R. Ruddell, & H. Singer (Eds),
Theoreticaamodels and processes of remg (4th ed., pp. 951-995). Newark, DE:
International Reading Association.

LaBerge, D., & Samuels, S. J. (1974). Toward a theory of automatic information
processing in reading. @gnitive Psychology, 6, 293-323.

Magliano, J. P., Dijkstra, K., & Zwaan, R. A. (1996). Generating predictive
inferences while viewing a movie. Discourse Processes 22, 199-224.

McCormick, S. (1992). Disabled readers' erroneous responses to inferential

comprehension questions: Descripton and analysis. Reading Research Quarterly, 27(1),
55-77.

Meringoff, L. K. (1980). Inﬂuence of the medium on children's story
apprehension. Journal of Educational Psychology, 72(2), 240-249.

Neuman, S. B. (1989). The impact of different media on children's story
comrpehension. Reading Research and Instructiona28(4), 38-47.

Neuman, S. B. (1990). Assessing children's inferencing strategies. In J. Zutell &

S. McCormick (Eds), Literacy theory and research: Analyses from multiple paradigms
(pp. 267-274). Chicago, IL: National Reading Conference.

Neuman, S. B. (1992). Is learning from media distinctive? Examining children's
inferencing strategies. American Educational Research Jourml, 29(1), 119-140.

Neuman, S. B., & Roskos, K. (1992). Literacy objects as cultural tools: Effects
on children's literacy behaviors in play. Reading Research Quarterly. 27(3), 203-225.

Paris, S. G. (1991). Assessment and remediation of metacogntive aspects of
children's reading comprehension. Topics in Langpage Disordera; 2(1), 32-50.

 

Pearson, P. D., Hansen, J ., & Gordon, C. (1979). The effect of background
knowledge on young children's comprehension of explicit and implicit information.
Journal of Reading Behavior, 11(3), 201-209.

Pezdek, K., Lehrer, A., & Simon, S. (1984). The relationship between reading and
cognitive processing of television and radio. Child Development. 55, 2072-2082.

89

 

Pezdek, K., Simon, S., Stoeckert, J ., & Kiely, J. (1987). Individual differences in
television comprehension. Memory & Cognition. 15(5), 428-435.

Phillips, L. M. (1988). Young readers' inference strategies in reading
comprehension. C_ognition and Instruction. 5(3), 193-222.

Powell, J. L. (1989). How well do we measure real reading? ERIC Digest.

Pressley, M., & Afflerbach, P. (1995). Verbal protocols of reading: The nature of
constructively responsive reading. Hillsdale, NJ: Lawrence Erlbaum Associates.

 

Ricci, C. M., & Beal, C. R. (2002). The effect of interactive media on children's
story memory. Journal of Educational Psychology, 94(1), 138-144.

Rickheit, G., & Strohner, H. (Eds). (1985). Inferences in text processing.
Bielefeld, F.R.G.: Elsevier Science Publishers B.V.

Steig, W. (1976). The amazing bone. New York: Farrar, Straus, Giroux.
Steig, W. (1983). Doctor De Soto. New York: Farrar, Straus, and Giroux.

Thorndike, E. L. (1917). Reading as reasoning: A study of mistakes in paragraph
reading. The Joumza of Educational Psychology, 7(June), 323-332.

Thurlow, R., & van den Broek, P. (1997). Automaticity and inference generation

during reading comprehension. Reading and Writing marterly: Overcoming Learning
Difﬁculties. 13. 165-181.

Trabasso, T. (1981). On the making of inferences during reading and their
assessment. In J. T. Guthrie (Ed), Comprehension and teaching: Research reviews (pp.
56-76). Newark, DE: International Reading Association.

Trabasso, T., & Sub, S. (1993). Understanding text: Achieving explanatory
coherence through on-line inferences and mental operations in working memory.
Discourse Processes 16, 3-34.

Trabasso, T., & Magliano, J. P. (1996). Conscious understanding during
comprehension. Discourse Processes. 21. 255-287.

van den Broek, P., Fletcher, C. R., & Risden, K. (1993). Investigations of

inferential processes in reading: A theoretical and methodological integration. Discourse
Processea, 16, 169-180

90

van den Broek, P., Risden, K., Tzeng, Y., Trabasso, T., & Basche, P. (2001).
Inferential questioning: Effects on comprehension of narrative texts as a function of grade
and timing. Journal of Educational Psychology, 93(3), 521-529.

Waterford. (2000). Waterford Early Reading Progpam: Waterford Institute, Inc.

Whitney, P., & Budd, D. (1996). Think-aloud protocols and the study of
comprehension. Discourse Processes. 21. 341-351.

Williams, J. P. (1993). Comprehension of students with and without learning

disabilities: Identiﬁcation of narrative themes and idiosyncratic text representations.
Journal of Educational Psychology, 85(4), 631-641.

91

APPENDICES

92

 

APPENDIX A

Consent Form

To the family of

 

I am a graduate student at the University of California at Berkeley conducting a research study to
ﬁnd out how students help themselves understand what is going on in a story when they are
reading to themselves, listening to someone else read, or watching a video version of a story.
Ultimately, I hope this work will help educators learn more about how to foster students’ reading
comprehension, and I would like to invite your child to participate in the study.

If you agree to have your child participate, she will take part in two one-hour sessions at school.
During the ﬁrst session I will ask your child to talk about how she is helping herself understand
the stories I present to her in three formats: written, audio, and video. In the second session,
scheduled on another day, your child will repeat this process with three diﬂ’erent stories. With
your permission, the interviews will be tape-recorded. In addition, you and your child will
complete a brief survey about her reading, listening, and viewing habits. This survey may be
completed by hand or over the telephone. Parts of these interviews and surveys may appear in
published reports, but your child will never be identiﬁed by name when I write down or discuss
the information that I collect. Her privacy will be protected to the maximum extent allowable by
law. The records of this study will be kept in a locked ﬁle.

If you will allow your child to participate, please Sign the permission below. Please also ask your
child to Sign the form and have her return the signed form to her teacher. Keep in mind that you
can withdraw your child from the study at any time. Whether or not you choose to allow her to
take part in the research will have no bearing on her standing or grades at school.

Ifyou have questions, you may contact me at (510) 407-1390 or jaynesca@uclink.berkeley.edu.
In addition, you may contact my supervisor, P. David Pearson, at (510) 643-6644 or
ppearson@socrates.berkeley.edu , or Richard Prawat at Michigan State University at (517) 353-
6417 or rsprawat@msu.edu. Finally, you may contact Ashir Kumar, MD. chair of the University
Committee on Research Involving Human Subjects (UCRIHS) if you have questions about your
rights as a human subject of research. He is at 246 Administration Building, Michigan State
University, East Lansing, MI 48824-1046, by phone (517) 355-2180, by FAX (517) 353-2976, or
by email at ucrihs@msu.edu.

Sincerely,

Carolyn Jaynes

You will be given a copy of this form to keep for your records.

I have read the above information. I give permission for my child to participate in the

reading study.

Parent/guardian’s signature: Date

 

 

Child’s signature: Date

 

93

 

 

 

APPENDIX B

Assessment Protocol

Flynt, E. S., & Cooter, R. B. (1998). Reading Inventory for {he Classroom. (Third ed). Upper Sadrﬂe
River, NJ: Merrill.

The Big Bad Wolf

One day Mr. Wolf was walking through the forest. He was enjoying an afternoon
walk and not bothering anyone. All of a sudden it started to rain and he became wet and
cold.

Just when Mr. Wolf was about to freeze to death, he saw a small house in the
woods. Smoke was coming from the chimney, so he knocked on the door. No one was
home, but a note on the door said:

Come in and make yourself warm. I’ll be back about 2:00 p.m.
Love,
Granny

The poor wet wolf came in and began to warm himself by the ﬁre. He saw one of
the Granny’s nightgowns on the bed, so he decided to put it on instead of his wet clothes.
Since he was still very, very cold he decided to get into Granny’s bed. Soon he was fast
asleep.

Mr. Wolf fell into a deep Sleep. When he awoke, Mr. Wolf found an old woman,
a little girl wearing a red coat, and a woodcutter standing around the bed. The
woodcutter was yelling at Mr. Wolf and saying something about how he was going to kill
him with his axe. Mr. Wolf jumped out of the bed and ran for his life.

Later that day, Mr. Wolf was ﬁnally safe at home. His wife said, “Just you wait,
those humans will make up a story about how big and bad you were.”

Questions:

Who was the story about? (character/literal)

Where was Mr. Wolf when he saw the house? (setting/literal)

Why did Mr. Wolf need to get into the house? (story problem/literal)

What made Mr. Wolf think it was okay to go into the house? (problem resolution
attempts/inferential)

5. What did Mr. Wolf do after entering the house? (problem resolution
attempts/literal)

Why did Mr. Wolf have to run for his life? @roblem resolution attempts/literal)
What lesson did Mr. Wolf learn? (theme/evaluative)

8. What did Mrs. Wolf say that would make you think She didn’t trust humans?
(character/literal)

99°19?”

N?

94

APPENDIX C

Think Aloud Training Poem

If I Were In Charge of the World
by Judith Viorst

[H were in charge of the world

I'd cancel oatmeal,

Monday mornings,

Allergy shots, and also Sara Steinberg.

IfI were in charge of the world

There'd be brighter night lights,

Healthier hamsters, and

Basketball baskets forty eight inches lower.

If I were in charge of the world
You wouldn't have lonely.

You wouldn't have clean.

You wouldn't have bedtimes.
Or "Don't punch your sister."
You wouldn't even have sisters.

IfI were in charge of the world

A chocolate sundae with whipped cream and nuts would be a vegetable
All 007 movies would be G,

And a person who sometimes forgot to brush,

And sometimes forgot to ﬂush,

Would still be allowed to be

In charge of the world.

95

APPENDIX D

Texts Used in Research Sessions

Doctor De Soto
By William Steig

Doctor De Soto, the dentist, did very good work, so he had no end of patients. Those close to his own
size—moles, chipmunks, et cetera—sat in the regular dentist’s chair. Larger animals sat on the ﬂoor,
while Doctor De Soto stood on a ladder. For extra-large animals, he had a special room. There Doctor De
Soto was hoisted up to the patient’s mouth by his assistant, who also happened to be his wife. Doctor De
Soto was especially popular with the big animals. He was able to work inside their mouths, wearing rubber
boots to keep his feet dry; his ﬁngers were so delicate, and his drill so dainty, they could hardly feel any
pain

Being a mouse, he refused to treat animals dangerous to mice, and it said so on his Sign When the doorbell
rang, he and his wife would look out the window. They wouldn’t admit even the most timid-looking eat.
One day, when they looked out, they saw a well-dressed fox with a ﬂannel bandage around his jaw.

“I cannot treat you, sir!” Doctor De Soto shouted. “Sir! Haven’t you read my Sign?”

“Please!” the fox wailed. “Have mercy, I’m suffering!” And he wept so bitterly it was pitiful to see.

“Just a moment,” said Doctor De Soto. “That poor fox,” he whispered to his wife. “What shall we do?”
“Let’s risk it,” said Mrs. De Soto. She pressed the buzzer and let the fox in.

He was up the stairs in a ﬂash “Bless your little hearts,” he cried, falling to his knees. “I beg you, do
something! My tooth is killing me.”

“Sit on the ﬂoor, sir,” said Doctor De Soto, “and remove the bandage, please.”

Doctor De Soto climbed up the ladder and bravely entered the fox’s mouth “Oooo-wowl” he gasped The
fox had a rotten bicuspid and unusually bad breath.

“This tooth will have to come out,” Doctor De Soto announced. “But we can make you a new one.”

“Just stop the pain,” whimpered the fox, wiping some tears away.

Despite his misery, he realized he had a tasty little morsel in his mouth, and his jaw began to quiver. “Keep
open!” yelled Doctor De Soto.

“Wide open!” yelled his wife.

“I’m giving you gas now,” said Doctor De Soto. “You won’t feel a thing when I yank that tooth.

Soon the fox was in dreamland. “M-m-m, yummy,” he mumbled. “How I love them raw...with just a
pinch of salt, and a...dry...white wine.”

They could guess what he was dreaming about. Mrs. De Soto handed her husband a pole to keep the fox’s
mouth open. Doctor De Soto fastened his extractor to the bad tooth Then he and his wife began turning
the winch. Finally, with a sucking sound, the tooth popped out and hung swaying in the air.

“I’m bleeding!” the fox yelped when he eame to.
96

Doctor De Soto ran up the ladder and Stuffed some gauze in the hole. “The worst is over,” he said. “I’ll
have your new tooth ready tomorrow. Be here at eleven Sharp.”

The fox. still woozy, said goodbye and left. On his way home, he wondered if it would be shabby of him to
eat the De Sotos when the job was done.

After oﬂice hours, Mrs. De Soto molded a tooth of pure gold and polished it “Raw with salt, indeed,”
muttered Doctor De Soto. “How foolish to trust a fox!”

“He didn’t know what he was saying,” said Mrs. De Soto. “Why should he harm us? We’re helping him.”
“Because he’s a fox!” said Doctor De Soto. “They’re wicked, wicked creatures.”

That night the De Sotos lay awake worrying. “Should we let him in tomorrow?” Mrs. De Soto wondered.
“Once I start a job,” said the dentist ﬁrmly, “I ﬁnish it. My father was the same way.”

“But we must do something to protect ourselves,” said his wife.

They talked and talked until they formed a plan “I think it will work,” said Doctor De Soto. A minute
later he was snoring.

The next morning, promptly at eleven, 3 very cheerful fox turned up. He was feeling not a particle of pain.
When Doctor De Soto got into his month, be snapped it shut for a moment, then opened wide and laughed
“Just a joke!” he chortled.

“Be serious,” said the dentist sharply. “We have work to do.” His wife was lugging the heavy tooth up the
ladder.

“Oh, I love it!” exclaimed the fox. “It’s just beautiful.”
Doctor De Soto set the gold tooth in its socket and hooked it up to the teeth on both Sides.

The fox caressed the new tooth with his tongue. “My, it feels good,” he thought “I really shouldn’t eat
them. On the other hand, how can I resist”

“We’re not ﬁnished,” said Doctor De Soto, holding up a large jug. “I have here a remarkable preparation
developed only recently by my wife and me. With just one application, you can be rid of toothaches
forever. How would you like to be the ﬁrst one to receive this unique treatment?”

“I certainly would!” the fox declared. “I’d be honored.” He hated any kind of personal pain
“You will never have to see us again,” said Doctor De Soto.

“No one will see you again,” said the fox to himself. He had deﬁnitely made up his mind to eat them—
with the help of his brand-new tooth.

Doctor De Soto stepped into the fox’s mouth with a bucket of secret formula and proceeded to paint each
tooth. He bummed as he worked. Mrs. De Soto stood by on the ladder, pointing out spots he had missed.
The fox looked very happy.

When the dentist was done, he stepped out. “Now close your jaws tight,” he said, “and keep them closed
for a full minute.” The fox did as he was told. Then he tried to open his mouth—but his teeth were stuck
together!

97

i

“Ah, excuse me, I should have mentioned,” said Doctor De Soto, “you won’t be able to open your mouth
for a day or two. The secret formula must ﬁrst permeate the dentine. But don’t worry. No pain ever
again!”

The fox was stunned He stared at Doctor De Soto, then at this wife. They smiled, and waited All he
could do was say, “Frank 00 berry mush” through his clenched teeth, and get up and leave. He tried to do
so with dignity.

Then he stumbled down the stairs in a daze.

Doctor De Soto and his assistant had outfoxed the fox. They kissed each other and took the rest of the day
OR.

98

The Amazing Bone
By William Steig

It was a brilliant day, and instead of going straight home from school, Pearl dawdled. She watched the
grownups in town at their grownup work, things she might someday be doing.

She saw the street cleaners sweeping the streets and She looked in at the bakery on Parsnip Lane and saw
the bakers taking hot loaves of pumpernickel out of the oven and powdering cmllers with sugar dust.

On Cobble Road She stopped at Maltby’s barn and stood gawking as the old gaffers pitched their ringing
horseshoes and spat tobacco juice.

Later She sat on the ground in the forest between school and home, and Spring was so bright and beautiful,
the warm air touched her so tenderly, she could almost feel herself changing into a ﬂower. Her light dress
felt like petals. “I love everything,” she heard herself say.

“So do I,” a voice answered
Pearl straightened up and looked around. No one was there. “Where are you?” she asked.

“Look down,” came the answer. Pearl looked down. “I’m the bone in the violets near the tree by the rock
on your right” Pearl stared at the small bone.

“You talk?” she murmured

“In any language,” said the bone. “And 1 can imitate any sound there is.” The bone made the sounds of a
trumpet calling soldiers to arms. Then it sounded like wind blowing, then like paneling rain. Then it
snored, then it sneezed.

Pearl couldn’t believe what She was hearing. “You’re a bone,” she said. “How come you can sneeze?”
“I don’t know, the bone replied “I didn’t make the world”
“May I take you home with me, wonderful bone?” Pearl asked.

“You certainly may,” said the bone. “I’ve been alone a long time. A year ago, come August, I fell out of a
witch’s basket. I could have yelled after her as she walked on, but I didn’t want to be her bone any longer.
She ate snails cooked in garlic at every meal and was always complaining about her rheumatism and asking
nosy questions. I’d be happier with someone young and lively like you.”

Pearl picked the bone up and gently put it in her purse. She left the purse open, so they could contimre their
talk, and started home, forgetting her schoolbooks on the grass. She was eager to Show this bone to her
parents, and she could guess what would happen when She did. She would tell about the talking bone, and
her mother would say “You’re only imagining it.” Her father would agree, and then the bone would
ﬂabbergast them both by talking.

The spring green sparkled in the spring light. Tree toads were trilling. “It’s the kind of wonderful day,”
said Pearl, “when wonderful things happen—like ﬁnding you.”

“Like ﬁnding you!” the bone answered. And it began to whistle a walking tune that made the going very
pleasant. But not for long. Who should rush out from in back of a boulder and spoil everything but three
highway robbers with pistols and daggers. Pearl couldn’t tell what bmd of animal they were, because they
wore cloaks and Halloween masks, but they were ﬁerce and spoke in chilling voices.

99

 

 

 

“Hand over the purse!” one commanded. Pearl would have gladly surrendered the purse, just to be rid of
them, but not with the bone in it.

“You can’t have my purse,” She said, surprised at her own boldness.
“What’s in it?” said another robber, pointing his gun at Peari’s head.
“I’m in it!” the bone growled. And it began to hiss like a snake and roar like a lion.

The robbers didn’t wait around to hear the rest, in case there was any more. They ﬂed so fast you couldn’t
tell which way they’d gone. It made Pearl laugh The bone, too.

They continued on their way, joking about what had just happened and chatting about this and that. But it
wasn’t long before a fox stepped forth from behind a tree and barred their path He wore a sprig of lilac in
his lapel, he carried a eane, and he was grinning so the whole world could see his Sharp white teeth.

“Hold everything,” he said Pearl froze. “You’re exactly what I’ve been longing for,” he went on
“Young, plump, and tender. You will be my main course at dinner tonight” And he seized Pearl in a tight
embrace.

“Unhand her, you villain,” the bone screamed, “or I’ll bite your ears off!”

“Who is that speaking?” asked the surprised fox.

“A ravenous crocodile who dates on fresh fox chops, that’s who!” answered the bone.

The wily fox was not as easily duped as the robbers. He saw no dangerous crocodile. He peered into
Pearl’s purse, where the sounds seemed to be coming from, and pulled out the bone. “AS I live and
ﬂourish!” he exclaimed. “A talking bone. I’ve always wanted to own something of this sort.” And he put
the bone in his pocket, where it roared and ranted to no avail.

Pushing Pearl along, the fox set out for his hideaway. Pearl’s sobs were so pitiful the fox couldn’t help
feeling a little sorry for her, but he was determined she would be his dinner.

“Please, Mr. Fox,” Pearl whinrpered, “may I have my bone back, at least until I have to die?”

“011, all right,” said the fox, disgusted with himself for being so softhearted, and he handed her the bone,
which she put back in her purse.

“You must let this beautiful young creature go on living,” the bone yelled. “Have you no Shame, Sir!”
The fox laughed “Why should I be ashamed? 1 can’t help being the way I am. I didn’t make the world”
The bone commenced to revile the fox. “You coward!” it sneered. “You worm, you odoriferous wretch!”

These expletives were annoying “Shut up, or I’ll eat you,” the fox snarled. “It would be amusing to gnaw
on abone that talks...and screams with pain”

The bone kept quiet the rest of the way, and so did Pearl.

When they arrived at the fox’s hideaway, he shoved Pearl, with her bone, into an empty room and locked
thedoor. Pearlsatontheﬂoorandstaredatthewalls.

“I know how you feel,” the bone whispered.

“I’m only just beginning to live,” Pearl whispered back. “I don’t want it to end.”
1 00

\ .

“I know,” said the bone.
“Isn’t there something we can do?” Pearl asked.

“I wish I could think of something,” said the bone, “but 1 can’t. I feel miserable.”

“What’s that?” Pearl asked. She’d heard some sounds from the kitchen
“He’s sharpening a knife,” the bone whispered.

“Oh, my goodness!” Pearl sobbed. “And what’s that?”

“Sounds like wood being put into a stove,” answered the bone.

“1 hope it won’t all take too long,” said Pearl. She could smell vinegar and oil. The fox was preparing a
salad to go with his meal. Pearl hugged the bone to her breast. “Bone, say something to comfort me.”

“You are very dear to me,” said the bone.

“Oh, how dear you are to me!” Pearl replied. She could hear a key in the lock and was unable to get a
word out of her throat or turn her eyes toward the door.

“Be brave,” the bone whispered Pearl could only tremble.
She was dragged into the kitchen, where she could see ﬂames in the open Stove.

“I regret having to do this to you,” sighed the fox. “It’s nothing personal.”
“Yibbam!” said the bone suddenly, without knowing why he said it.
“What was that?” said the fox, standing Stock-still.

“Yibbam sibibble!” the bone intoned “Jibrakken sibibble digray!” And something quite unexpected took
place. The fox grew several inches shorter.

“Alabam Chinook beboppit gebozzle!” the bone continued, and miraculously the fox was the Size of a
rabbit. No one could believe what was happening, not Pearl, not the fox, not even the bone, whose words
were making it happen.

“A doonis ishgoolak keebolotin yibapp!” it went on The fox, clothes and all, was now the size of a mouse.

“Scrabboom‘t!” the bone ordered, and the mouse—that is, the minuscule fox—scurried away and into a
hole.

“I didn’t know you could do magic!” Pearl breathlessly exclaimed.
“Neither did I,” said the bone.
“Well, what made you say those words?”

“I wish I knew,” the bone said. “They just came to me, I had to say them. I must have picked them 11p
somehow, hanging around with that witch.”

“You’re an amazing bone,” said Pearl, “and this is a day I won’t ever forget!”

101

 

It was dark when they reached Pearl’s house. The moment the door swung open she was in her mother’s
arms, and right after that in her father’s.

“Where on earth have you been?” they both wamw to know. “We were frazzled with worry.”
Pearl didn’t know what to say ﬁrst. She held up the bone. “This bone,” She said, “can talk!”

And just as She had expected, her mother said, “A talking bone? Why, Pearl, it’s only your imagination.”
And her father said something similar.

And also as Pearl had expected, the bone astonished them both by remarking, “You have an exceptional
daughter.”

Before her parents had a chance to get over their shock, Pearl began telling the story of her day’s adventure,
and the bone helped out. It was all too much for Pearl’s parents. Until they got used to it.

The bone stayed on and became part of the family. It was given an honored place in a silver tray on the
mantelpiece. Pearl always took it to bed when she retired, and the two chatterboxes whispered together
until late in the night. Sometimes the bone put Pearl to Sleep by singing, or by imitating soft harp music.
Anyone who happened to be alone in the house always had the bone to converse with. And they all had
music whenever they wanted it, and sometimes even when they didn’t.

102

g
i“

 

 

Madeline and the Gypsies
By Ludwig Bemelmans

In an old house in Paris that was covered with vines lived twelve little girls in two straight lines. In two
straight lines they broke their bread and bnlshed their teeth and went to bed. They left the house at half-
past nine—the smallest one was Madeline. In another old house that Stood next door lived the son of the
Spanish Ambassador. He was all alone; his parents were away; he had no one with whom to play. He
asked, “Please come, I invite you all, to a wonderful Gypsy Carnival.” And so, dear reader, here we go!

Up and down and down and up—they hoped the Fenis wheel would never stop. Round and round; the
children cried, “Dear Miss Clavel, just one more ride!”

A sudden gust of wind, a bolt of lightning, even the Rooster found it ﬁightening. The big wheel stops; the
passengers land How fortunate there is a taxi stand!

“Hurry, children, off with these things! You’ll eat in bed.”
Mrs. Murphy brings the soup of the evening; it is half-past nine. “Good heavens, where is MADELINE?”

Poor Miss Clavel, how would she feel if she knew that on top of the Fenis wheel, in weather that turned
from bad to rotten, Pepito and Madeline had been forgotten?

Pepito said, “Don’t be afraid I will climb down and get some aid”

It was downpouring more and more as he knocked on the Gypsies’ caravan door.

The Gypsy Mama with her umbrella went and got some help in the circus tent With the aid of the strong
man and the clown, Madeline was safely taken down.

The Gypsy Mama tucked them in and gave them potent medicine. The big wheel was folded, and the tent.
They packed their wagons and away they went. For Gypsies do not like to stay—they only come to go
away.

A bright new day—the Sky is blue; the storm is gone; the world is new. This is the Castle of
Fountainblue—“All this, dear children, belongs to you.”

How wonderful to ﬂoat in a pool, watch other children go to school, never to have to brush your teeth, and
never—never—to go to sleep. The Gypsies taught them grace and speed, and how to ride the circus steed.
Then Madeline said, “It’s about time we sent dear Miss Clavel a line.”

Poor Miss Clavcl—a shadow of her former self ﬁom worrying, because, instead of twelve, there were only
eleven little girls—stopped brushing their curls and suddenly revived when the postal card arrived “Thank
heaven,” She said, “the children are well! But dear, oh dear, they’ve forgotten how to Spell.”

She studied the postmark, and then fast and faster they rushed to the scene ofthe disaster. The Gypsy
Mama didn’t like at all what she saw in her magic crystal ball The Gypsy Mama said, “How would you
like to try on this lovely costume of a lion?”

With a curved needle and some string She sewed both the children in, and nobody knew what was inside
the tough old lion’s leathery hide. A circus lion earns his bread by scaring people half to death And after
that, he’s put to bed. A lovely dawn and all was well; the lion roamed through wood and dell. He smelled
sweet ﬂowers; he eame to a farm; he frightened the bamyard—intending no harm. They saw a man and
said, “Please help us to get out of this old pelt.”

103

 

 

 

The man was a hunter. He took his gun. He got to his feet and started to mn.

Said the lion, “We’d better go back, for if we’re not in a 200 or circus, we’ll surely be shot.”
They got to the tent in time for the show. “Look,” said Madeline, “There in the ﬁrst row. .
“Oh yes,” said Pepito, “There are people we know!”

“Dear Miss Clavcl! At last we found you! Please let us put our arms around you.”

The Gypsy Mama sobbed her grief into her only handkerchief. The strong man suddenly felt weak, and
tears were nmning down his cheek. Even the poor clown had to cry as the time came to say good-bye.

The best part of the voyage—by plane, by ship, or train—is when the trip is over and you are home again

Here is a freshly laundered shirty—It’s better to be clean than dirty. In two straight lines they broke their

bread and brushed their teeth and went to bed. “Good night, little girls, thank the Lord you are well! And
now PLEASE go to sleep,” said Miss Clavel. And she turned out the light and closed the door—and then

she came back; just to count them once more!

104

Why Mosquitoes Buzz in People’s Ears
By Verna Aardema

One morning a mosquito saw an iguana drinking at a waterhole. The mosquito said, “Iguana, you will
never believe what I saw yesterday.”

“Try me,” said the iguana.
The mosquito said, “I saw a farmer digging yams that were almost as big as I am.”

“What’s a mosquito compared to a yam?” snapped the iguana gnunpily. “I would rather be deaf than
listen to such nonsense!” Then he stuck two sticks in his ears and went off, mek, mek, mek, mek, through
the reeds.

The iguana was still grumbling to himself when he happened to pass by a python. The big snake raised his
head and said, “Good morning, Iguana.” The iguana did not answer but lumbered on, bobbing his head,
badamin, badamin. “Now, why won’t you speak to me?” said the python to himself. “Iguana must be
angry about something I’m afraid he is plotting some mischief against me!”

He began looking for somewhere to hide. The ﬁrst likely place he found was a rabbit hole, and in it he
went, wasawusu, wasawusu, wasawusu. When the rabbit saw the big snake coming into her burrow, she
was terriﬁed. She scurried out through her back way and bounded, krik, krik, lo'ik, across a clearing.

A crow saw the rabbit nmning for her life. He ﬂew into the forest crying kaa, kaa, kaa! It was his duty to
spread the alarm in case of danger.

A monkey heard the crow. He was sure that some dangerous beast was prowling near. He began
screeching and leaping kili wili through the trees to help warn the other animals As the monkey was
crashing through the treetops, he happened to land on a dead limb. It broke and fell on an owl’s nest,
killing one of the owlets.

Mother Owl was not at home. For though she usually hunted only in the night, this morning she was still
out searching for one more tidbit to satisfy her hungry babies. When she returned to the nest, she found
one of them dead Her other children told her that the monkey had killed it. All that day and all that night,
she sat in her tree—so sad, so sad, so sad!

Now it was Mother Owl who woke the sun each day so that the dawn could come. But this time, when she
should have booted for the sun, she did not do it. The night grew longer and longer. The animals of the
forest knew it was lasting much too long. They feared that the sun would never come back.

At last King Lion called a meeting of the animals. They came and sat down, pem, pem, pem, around a
council ﬁre. Mother Owl did not come, so the antelope was sent to fetch her.

When she arrived, King Lion asked, “Mother Owl, why have you not called the sun? The night has lasted
long, long, long, and everyone is worried.”

Mother Owl said, “Monkey killed one of my owlets. Because of that, I cannot bear to wake the sun.”

The king said to the gathered animals. “Did you hear? It was the monkey who killed the owlet—and now
Mother Owl won’t wake the sun so that the day can come.”

Then King Lion called the monkey. He came before him nervously glancing from side to side, rim, rim,
rim, rim. “Monkey,” said the king, “why did you kill one of the Mother Owl’s babies?

105

“Oh, King,” said the monkey, “it was the crow’s fault He was calling and calling to warn us of danger.
And I went leaping through the trees to help. A limb broke under me, and it fell on the owl’s nest.”

The king said to the council. “So, it was the crow who alarmed the monkey, who killed the owlet—and
now Mother Owl won’t wake the sun so that the day can come.”

Then the king ealled for the crow. That big bird came ﬂapping up. He said, “King Lion, it was the rabbit’s
fault! I saw her running for her life in the daytime. Wasn’t that reason enough to spread an alarm?”

The king nodded his head and said to the council. “So, it was the rabbit who startled the crow, who
alarmed the monkey, who killed the owlet—and now Mother Owl won’t wake the sun so that the day can
come.”

Then King Lion called the rabbit. The timid little creature stood before him, one trembling paw drawn up
uncertainly. “Rabbit,” cried the king, “why did you break the law of nature and go running, running,
running, in the daytime?”

“Oh, King,” said the rabbit, “it was the python’s fault I was in my house minding my own business when
that big snake came in and chased me out”

The king said to the council. “So, it was the python who scared the rabbit, who startled the crow, who
alarmed the monkey, who killed the owlet—and now Mother Owl won’t wake the sun so that the day can
come.”

King Lion called the python, who came slithering, wasawusu, wasawusu, past the other animals. “But,
King,” he cried, “it was the iguana’s fault! He wouldn’t speak to me. And I thought he was plotting some
mischief against me. When I crawled into the rabbit’s hole, I was only trying to hide.”

The king said to the council. “,So it was the iguana who frightened the python, who scared the rabbit, who
startled the crow, who alarmed the monkey, who killed the owlet—and now Mother Owl won’t wake the
sun so that the day can come.”

Nowthe iguana wasnotattherneeting, forhehadnotheardthesummons Theantelopewassenttofetch
him All the animals laughed when they saw the iguana coming, badamin, badamin, with the sticks still
stuck in his ears!”

King Lion pulled out the sticks, pump, pump. Then he asked, “Iguana, what evil have you been plotting
against the python?”

“None! None at all!” cried the iguana “Python is my friend!”
“Then why wouldn’t you say good morning to me?” demanded the snake.

“I didn’t hear you, or even see you!” said the iguana “Mosquito told me such a big lie, I couldn’t bear to
listen to it So I put sticks in my ears.”

“Nge, nge, nge,” laughed the lion. “So that’s why you had sticks in your ears!”

“Yes,” said the iguana. “It was the mosquito’s fault”

King Lion said to the council. “So, it was the mosquito who annoyed the iguana, who frightened the
python, who scared the rabbit, who startled the crow, who alarmed the monkey, who killed the owlet—and
now Mother Owl won’t wake the sun so that the day can come.”

“Punish the mosquito! Punish the mosquito!” cried all the animals.
1 O6

When Mother Owl heard that, she was satisﬁed. She turned her head toward the east and hooted: “Hoo!
Hoooo! Hooooooo!”

Andthesrmcameup.

Meanwhile the mosquito had listened to it all from a nearby bush. She crept under a curly leaf, semm, and
was never found and brought before the council. But because of this the mosquito has a guilty conscience.
To this day she goes about whining in people’s ears: “Zeee! Is everyone still angry at me?” When she
does that, she gets an honest answer.

 

107

A Story, A Story
By Gail Haley

Once, oh small children round my knee, there were no stories on earth to hear. All the stories belonged to
Nayame, the Sky God. He kept them in a golden box next to this royal stool.

Ananse, the Spider rmn, wanted to buy the Sky God’s stories. So he spun a web up to the sky. When the
Sky God heard what Ananse wanted, he laughed: “Twe, twe, twe. The price of my stories is that you bring
me Osebo, the leopard of-the-terrible—teeth, Mmboro the hornet who-stings-like-ﬁre, and Mmoatia the fairy
whom-men-never-see.”

Ananse bowed and answered: “I shall gladly pay the price.”

“Twe, twe, rwe,” chuckled the Sky God “How can a weak old man like you, so small, so small, so small,
pay my price?”

But Ananse merely climbeddown toearthtoﬁndthethingsthat the Sky Goddemanded Ananseranalong
the jungle path—yiridi, yiridi, yiridi—till he came to Osebo the leopard-of-the-temhle-teeth

“Oho, Ananse,” said the leopard, “you are just in time to be my lunch.”

Ananse replied: “As for that, what will happen will happen. But ﬁrst let us play the binding binding
game.”

The leopard, who was fond of games, asked: “How is it played?”

“With vine creepers,” explained Ananse. “I will bind you by your foot and foot Then I will untie you, and
you can tie me.”

“Verywell,” growledtheleopardwhoplannedtoeatAnanseassoonasitwashisturntobindhim

So Ananse tied the leopard by his foot, by his foot, by his foot, with the vine creeper. Then he said: “Now,
Osebo, you are ready to meet the Sky God.” And he hung the tied leopard in a tree in the jungle.

Next, Anansecutafrondfromabananatreeandﬁlledacalabashwithwater. Hecreptthroughthetall
grasses, sora, sora, sora, till he came to the nest of Mmboro, the hornets-who-sting-like-ﬁre. Amuse held
the banana leafover his head as an umbrella Then be poured some ofthe water in the calabash over his
head

The rest he emptied over the homet’s nest and cried: “It is raining, raining, raining. Should you not ﬂy
into my calabash, so that the rain will not latter your wings?”

“Thank you. Thank you,” hummed the hornets, and they ﬂew into the ealabash—fom! Ananse quickly
stopped the mouth of the gourd.

“Now, Mmbora, you are ready to meet the Sky God,” said Ananse. And he hung the calabash full of
hornets onto the tree next to the leopard

Ananse now carved a little wooden doll holding a bowl. He covered the doll from top to bottom with
sticky latex gum Then he ﬁlled the doll’s bowl with pounkd yams. He set the little doll at the foot of a
ﬂamboyant tree where fairies like to dance. Ananse tied one end of a vine round the doll’s head and,
holding the other end in his hand, he hid behind a bush.

108

 

In a little while, Mmoatia the fairy-whom-no-man-sees came dancing, dancing, dancing, to the foot of the
ﬂamboyant tree. There she saw the doll holding the bowl of yams. Mmoatia said, “Gum baby, I am
hungry. May I eat some of your yarns?”

Ananse pulled at the vine in his hiding place, so that the doll seemed to nod its head. So the fairy took the
bowl from the doll and ate all the yams. “Thank you, Gum baby,” said the fairy. But the doll did not
answer.

“Don’t you reply when I thank you?” cried the angered fairy. The doll did not stir.

“Gum baby, I’ll slap your crying place unless you answer me,” shouted the fairy. But the wooden doll
remained still and silent. So the fairy slapped her crying place—pa! Her hand stuck fast to the gum baby’s
sticky cheek.

“Let go of my hand, or I’ll slap you again” —Pa! She slapped the doll’s crying place with her other hand
Now the fairy was stuck to the gum baby with both hands, and she was furious. She pushed against the doll
with her feet, and they also stuck fast

Now Ananse came out of hiding. “You are ready to meet the Sky God, Mmoatia” And he carried her to
the tree where the leopard and the hornets were waiting. Ananse spun a web round Osebo, Mmboro, and
Mmoatia. Then he spun a web to the sky. He pulled up his captives behind him, and set them down at the
feet of the Sky God.

“0, Nyame,” said Ananse, bowing low, “here is the price you ask for your stories: Osebo the leopard-of-
the-terrible-teeth, Mmboro the hornets-who-sting-like-ﬁre, and Mmoatia the fairy-whom-men-never—see.”

Nyame the Sky God called together all the nobles of his court and addressed them in aloud voice: “Little
Ananse, the spider man, had paid me the price I ask for my stories. Sing his praise. I command you.”

“From this day and going on forever,” proclaimed the Sky God, “my stories belong to Ananse and shall be
called ‘Spider Stories. ’”

“Eeeee, Eeeee, Eeeee,” shouted all the assembled nobles

So Ananse took the golden box of stories back to earth, to the people of his village. And when he opened
the box all the stories seattered to the corners of the world, including this one.

This is my story which I have related. If it be sweet, or if it be not sweet, take some elsewhere, and let
some come back to me.

109

 

 

 

Strega Nona
By Tomie dePaola

In a town in Calabria, a long time ago, there lived an old lady everyone called Strega Nona, which meant
“Grandnm Witch”

Although all the people in town talked about her in whispers, they all went to see her if they had troubles.
Even the priest and the sisters of the convent went, because Strega Nona did have a magic touch

She could cure a headache, with oil and water and a harpin. She made special potions for the girls who
wanted husbands. And she was very good at getting rid of warts.

But Strega Nona was getting old, and she needed someone to help her keep her little house and garden, so
she put up a sign in the town square.

And Big Anthony, who didn’t pay attention, went to see her. “Anthony,” said Strega Nona, “you must
sweep the house and wash the dishes. You rmrst weed the garden and pick the vegetables. You must feed
the goat and milk her. And you must fetch the water. For this, I will give you three coins and a place to
sleep and food to eat.”

“Oh, grazia,” said Big Anthony.

“The one thing you must never do,” said Strega Nona, “is touch the pasta pot. It is very valuable and I
don’t let anyone touch it!”

‘Oh, si, yes,” said Big Anthony.

And so the days went by. Big Anthony did his work and Strega Nona met with the people who came to see
herforheadachesandhusbandsandwarts. Big Anthorryhadanicebedtosleepinnexttothegoatshed,
and he had food to eat.

One evening when Big Anthony was milking the goat, he heard Strega Nona singing. Peeking in the
window, he saw Strega Nona standing over the pasta pot.

She sang,
Bubble, bubble, pasta pot.

Boil me up some pasta, nice and hot.
I’m hungry and it’s time to sup.
Boil enough pasta to ﬁll me up.

And the pasta pot bubbled and boiled and was suddenly ﬁlled with steaming hot pasta

Then Strega Nona sang,

Enough, enough, pasta pot,
I have my pasta, nice and hot

So simmer down my pot of clay
Until I’m hrmgry another day.
“How wonderful!” said Big Anthony. “That’s a magic pot for sure!”

And Strega Nona called Big Anthony in for supper. But too bad for Big Anthony, because he didn’t see
Strega Nona blow three kisses to the magic pasta pot.

And this is what happened

110

2' (1"! 1"“! “ﬂ

The next day when Big Anthony went to the town square to fetch the water, he told everyone about the
pasta pot. And naturally everyone laughed at him, because it sounded so silly—a pot that cooked all by
itself. “You’d better go and confess to the priest, Big Anthony,” they said. “Such a lie!”

And Big Anthony was angry and that wasn’t a very good thing to be. “I’ll show them!” he said to himself.
“Someday I will get the pasta pot and make it cook! And then they '11 be sorry.”

That day came sooner than even Big Anthony would have thought, because two days later Strega Nona said
to Big Anthony, “Anthony, I must go over the mountain to the next town to see my friend, Strega Amelia.
Sweep the house and wwd the garden Feed the goat and milk her and for your lunch, there are some bread
and cheese in the cupboard. And remember, don’t touch the pasta pot.”

“Oh, yes—yes—Strega Nona,” said Big Anthony. But inside he was thinking, My chance has come! As
soon as Strega Nona was out of sight, Big Anthony went inside, pulled the pasta pot off the shelfand put it
on the ﬂoor.

“Now, let’s see if I can remember the words,” said Big Anthony. And Big Anthony sang,
Bubble, bubble, pasta pot,
Boil me some pasta, nice and hot
I’m hungry and it’s time to sup.
Boil enough pasta to ﬁll me up.

And sure enough, the pot bubbled and boiled and began to ﬁll up with pasta “Aim!” said Big Anthony,
and he ran to the town square, jumped on the fountain and shouted, “Everyone get forks and plates and
platters and bowls. Pasta for all at Strega Nona’s house. Big Anthony has made the magic pasta pot
work.”

Of course, everyone laughed, but ran home to get forks and plates and platters and bowls and sure enough,
when they got to Strega Nona’s the pasta pot was so full it was beginning to overﬂow. Big Anthony was a
hero! He scooped out pasta and ﬁlled the plates and platters and bowls. There was more than enough for
all the townspeople, including the priest and the sisters from the convent. And some people came back for
two and three helpings, but the pot was never empty.

When all had had their ﬁll, Big Anthony sang,
Enough, enough, my pasta pot,
Ihave mypastaniceandhot.
So sinuner down, my pot of clay
Until I’m hungry another day.

But, alas, he did not blow the three kisses!

He went outside and to the applause of the crowd, Big Anthony took a bow. He was so busy listening to
compliments from everyone that he didn’t notice the pasta pot was still bubbling and boiling, until a sister
from the convent said, “Oh, Big Anthony, look!”

AndpastawaspouringoutofthepotallovertheﬂoorofStregaNona’shouseandwascomingoutthe
door! Big Anthony rushed in and shouted the magic words again, but he pot kept bubbling He took the
potofftheﬂoor,butpastakeptonpouringfromit BigAmhonygrabbedacoverandputitonthepotand
sat on it But the pasta raised the cover, and Big Anthony as well, and spilled on the ﬂoor of Strega Nona’s
house.

“Stap!” yelled Big Amhony. But the pasta did not stop and if someone hadn’t grabbed poor Big Anthony,

thepastawouldhavecoveredhimup. Thepastalmdallbutﬁlledthelittlehouse. Outofthewindowsand
through the doors camethe pastaandthe pot kept right onbubbling.

111

a. O”
. , . . ' o .
s I

The townspeople began to worry. “Do something, Big Anthony,” they shouted. Big Anthony sang the
magic song again but without the three kisses it did no good! By this time the pasta was on its way down
the road and all the people were nmning to keep ahead of it.

“We must protect our town from the pasta,” shouted the mayor. “Get mattresses, tables, doors—anything
to make a barricade.” But even that didn’t work. The pot kept bubbling and the pasta kept coming!

“We are lost,” said the people, and the priest and the sisters of the convent began praying. “The pasta will
cover our town,” they cried.

And it certainly would have, had Strega Nona not come down the road, home from her visit. She didn’t
have to look twice to know what had happened.

She sang the magic song and blew the three kisses, and with a sputter the pot stopped boiling and the pasta
came to a halt.

“Oh, grazia—thank you, thank you, Strega Nona,” the people cried. But then they turned on poor
Anthony. “String him up,” the men of the town shouted.

“Now, wait,” said Strega Nona. “The punishment must ﬁt the crime.” And she took a fork from a lady
standing nearby and held it out to Big Anthony. “All right, Anthony, you wanted pasta from my magic
pasta pot,” Strega Nona said, “and 1 want to sleep in my little bed tonight So start eating.”

And he did—poor Big Anthony.

112

 

 

APPENDIX E

Survey of Reading, Listening, Viewing, and Speaking Habits

Student:

1.

On a typical school day, about how much time do you spend reading to yourself or to
someone else? (Please, do NOT include the time you spend reading while you are at
school).

CIRCLE ONE:

0 min. 15 min. 30 min. 45 min. 1 hour More thanl hour

When it is NOT a school day (e.g. weekends, holidays, summer), about how much time
do you spend reading to yourself or to someone else?
CIRCLE ONE:

0min 15 min. 30min. 45min. lhour Morethanlhour

On a typical school day, about how much time do you spend listening to someone read to
you or listening to books on tape? (Again, please do NOT include time you spend
listening to someone read or listening to books on tape while you are at school).
CIRCLE ONE:

0min. 15min 30min 45min. lhour Morethanl hour

When it is NOT a school day, about how much time do you spend listening to someone
read to you or listening to books on tape?
CIRCLE ONE:

0min. 15min 30min. 45min. lhour Morethanl hour

On a typical school day, about how much time do you spend watching television?
CIRCLE ONE:

0 min 15 min. 30 min. 45 min 1 hour More than] hour

When it is not a school day, about how much time do you spend watching television?
CIRCLE ONE:

0 min. 15 min 30 min. 45 min. 1 hour More thanl hour

On a scale from 1 to 5, with 1 being low and 5 being high, how would you rate
yourself on the following:

Participation in whole class discussions or conversations:

CIRCLE ONE:
1 2 3 4 5
(I never participate) (I participate a lot)

Participation in small group discussions or conversations in class
CIRCLE ONE:

113

L"

 

Parent:

1 2 3 4 5
(I never participate) (I participate a lot)

Participation in discussions or conversations with a group of friends or relatives

CIRCLE ONE:
1 2 3 4 5
(I never participate) (I participate a lot)

Participation in discussions or conversations with a friend or relative
CIRCLE ONE:

1 2 3 4 5
(I never participate) (I participate a lot)

On a typical school day, about how much time does your child spend reading to
him/herself or to someone else? (Please, do NOT include the time s/he spends reading
while s/he is at school).

CIRCLE ONE:

0 min. 15 min 30 min. 45 min. 1 hour More thanl hour

When it is NOT a school day, about how much time does your child spend reading to
him/herself or to someone else?
CIRCLE ONE:

0 min. 15 min. 30 min 45 min 1 hour More than! hour

On a typical school day, about how much time does your child spend listening to
someone read or listening to books on tape? (Again, please do NOT include time s/he
spends listening to someone read or listening to books on tape while s/he is at school).
CIRCLE ONE:

0min. 15 min 30min. 45min. lhour Morethanl hour

When it is NOT a school day, about how much time does your child spend listening to
someone read or listening to books on tape?
CIRCLE ONE:

{fa-— -.-‘--q'

0 min. 15 min. 30 min 45 min. 1 hour More than] hour

On a typical school day, about how much time does your child spend watching
television?
CIRCLE ONE:

0min. l5min 30min 45min. lhour Morethanl hour
When it is not a school day, about how much time does your child spend watching
television?

CIRCLE ONE:

114

 

0 min. 15 min. 30 min. 45 min. 1 hour More than! hour

On a scale from 1 to 5, with 1 being low, how would you rate your child on the
following:

Participation in discussions or conversations with a group of friends or relatives
CIRCLE ONE:

1 2 3 4 5
(My child never participates) (My child participates a lot)

Participation in discussions or conversations with a ﬁiend or relative
CIRCLE ONE:

1 2 3 4 5
(My child never participates) (My child participates a lot)

115

APPENDIX F

Story-Format-Order Selection Matrices

 

LEVEL 3

Audio

Video

Audio
Video
Print

Print
Audio
Video

Print
Video
Audio

Video
Audio
Print

Video
Print
Audio

 

Amazing Bone
Dr. De Soto
Madeline

tit

Student 10

Student 2

***

Student 4

Student 13

 

Amazing Bone
Madeline
Dr. De Soto

Student 14

Student 24

##*

Student 22

Student 20

Student 12

 

Dr. De Soto
Amazing Bone
Madeline

Student 9

at”

Student 16

##18

Student 17

Student 6

 

Dr. De Soto
Madeline
Amazing Bone

***

Student 21

Student 19

Student 26

Student 8

***

 

Madeline
Amazing Bone
Dr. De Soto

Student 15

***

Student 27

Student 11

Student 25

Student 1

 

 

Madeline
Dr. De Soto
Amazing Bone

 

Student 5

 

Student 7

 

Student 18

 

Student 3

 

***

 

Student 23

 

 

 

LEVEL 5

Audio

Video

Audio
Video
Print

Audio
Video

Video
Audio

Video
Audio

Video
Print
Audio

 

A Story
Mosquitoes
Strega Nona

***

***

Student 10

Student 15

Student 23

Student 25

 

A Story
Strega Nona
Mosquitoes

Student 11

Student 19

Student 6

Student 8

ttt

Student 27

 

Mosquitoes
A Story
Strega Nona

Student 16

Student 20

Student 9

#*#

Student 13

Student 17

 

Mosquitoes
Strega Norm
A Story

Student 2

Student 4

iii

Student 14

Student 5

Student 7

 

Strega Nona
A Story
Mosquitoes

**¥

Student 1

Student 26

Student 18

Student 22

*#*

 

 

Strega Nona
Moquitoes
A Story

 

Student 24

 

Student3

 

Student 12

 

*tt

 

***

 

Student 21

 

 

116

i
!

 

 

   

llllllllllllllllHUNllllllllHllllHllllllHlHlllHl
3 1293 02551 6802