Reading behaviors, psychological characteristics, and

Reading behaviors, psychological characteristics, and experiential background of one poor comprehender: A Grounded Theory case study

A Dissertation Presented to the Graduate Faculty of the University of Louisiana at Lafayette In Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy

Tobias A. Kroll Fall 2014

© Tobias A. Kroll 2014 All Rights Reserved

Reading behaviors, psychological characteristics, and experiential background of one poor comprehender: A Grounded Theory case study Tobias A. Kroll

APPROVED:

___________________________ Jack S. Damico, Chair Professor of Communicative Disorders

___________________________ Ryan L. Nelson Associate Professor of Communicative Disorders

___________________________ Nancye Roussel Professor of Communicative Disorders

___________________________ Holly L. Damico Assistant Professor of Communicative Disorders

___________________________ Mary Farmer-Kaiser Interim Dean of the Graduate School

This work is dedicated to all for whom the mechanistic outlook on human life has become an obstacle to the actualization of their full humanity.

No scientific explanation of human behavior could ever be complete. In fact, no unpoetic description of the human condition can ever be complete. Berliner (2002, p. 20)

ACKNOWLEDGMENTS No achievement in human life is brought forth alone: All that we accomplish happens thanks to the example and with the help of others. In finalizing this work, I am particularly indebted to the following people. Dr. Jack S. Damico: for his inspiration, for a thorough and scholarly education, and for his guidance and support throughout the thesis process. Dr. Ryan L. Nelson: for hours of invaluable help with the technical aspects of this dissertation. Drs. Nicole Müller and Martin J. Ball: for their patient advice and constant availability for any questions. Drs. Nancye Roussel and Holly L. Damico: for taking over committee duties as Drs. Müller and Ball followed their calling overseas. To the entire Department of Communicative Disorders: for almost five years of being a place of learning and fellowship, and in some ways a home in a new country. To my parents, Alf and Ursula Kroll: for instilling curiosity in me, and a lifelong love of learning. To my brother, Daniel J. Kroll, and my extended family: for quiet sustenance extended across the globe. And to my wife, Mrs. Anuradha Chandrasekaran: for never failing to cheer me up when I thought this work could not be done, and for always reminding me of my abilities and the love in which they thrive.

Table of Contents Acknowledgments ........................................................................................................... vi List of Tables ................................................................................................................... xi List of Figures .................................................................................................................. xii Chapter 1: Introduction ...................................................................................................1 Chapter 2: Review of the Literature ...............................................................................9 Reading Comprehension ......................................................................................9 A brief history of views on reading comprehension .......................... 10 Definitions of reading comprehension ................................................ 12 Poor Reading Comprehension ........................................................................... 22 History and definitions of the term ..................................................... 23 Text coherence and integration ........................................................... 25 Use of contextual information and background knowledge ............... 30 Inferencing .......................................................................................... 33 Metacognition and psychological factors in text processing .............. 36 Working memory ................................................................................ 44 Studies on linguistic processing .......................................................... 50 Language impairment and cognitive abilities ..................................... 53 Intervention studies ............................................................................. 56 Explanations for Poor Comprehension .............................................................. 60 General Discussion and Rationale for the Present Study................................... 63 General discussion .............................................................................. 63 Intentionality and contextual complexity: The qualitative alternative ............................................................................................ 68 Qualitative Research as Contextualized Inquiry ................................................ 76 Participant meanings vs. expert knowledge ........................................ 78 A pragmatic approach to qualitative inquiry ...................................... 79 Framing the Question ......................................................................................... 82 Chapter 3: Methods ........................................................................................................ 85 Methodological Framework: Grounded Theory ................................................ 85 Grounded theory: An overview .......................................................... 85 Substantive versus formal theory ........................................................ 89 Choosing an epistemology .................................................................. 90 The verificationist pitfall..................................................................... 94 Ethnographic Methods: Modified Participant Observation, Video Analysis, Interviews........................................................................................... 98 Basic ethnography: Modified participant observation and video analysis ...................................................................................... 98 Interviews ...........................................................................................100

Miscue Analysis/Eye Movement Miscue Analysis ..........................................103 Miscue analysis in the study of reading .............................................103 Studying eye movements in reading: History and technology ..........105 Studying eye movements in reading: The findings............................109 Combining approaches: Eye Movement Miscue Analysis ................111 Artifact Analysis ...............................................................................................112 Participant .........................................................................................................112 Number of participants/data sets ........................................................112 Recruitment ........................................................................................114 Assessment and qualifying criteria ....................................................115 Procedures: Data collection ..............................................................................117 Home portion: Interviews and audio/video recorded reading samples; field notes ............................................................................118 Laboratory portion: Eye tracking and audio/video recordings ..........120 Data Analysis ....................................................................................................122 Step 1: Line-by-line coding of interview data ...................................123 Step 2: Focused coding of interview data ..........................................124 Step 3: Miscue analysis ......................................................................124 Step 4: Incident-by-incident coding of miscue data ..........................125 Step 5: Eye Tracking Miscue Analysis (EMMA) ..............................126 Step 6: Axial coding and comparison to sensitizing case ..................127 Step 7: Formulation of a grounded theory on poor reading comprehension ...................................................................................128 Chapter 4: Results..........................................................................................................129 Demographic and Clinical Information ............................................................129 Interviews ..........................................................................................................132 Theme One: READING VS. COMPREHENDING ......................................136 Theme Two: (LACK OF) SUPPORT ......................................................137 Theme Three: HALF THE STORY .........................................................139 Theme Four: ISSUES EXPRESSING HIMSELF.........................................143 Theme Five: (LACK OF) MOTIVATION & NEED FOR SCAFFOLDING/STRUCTURE ..................................................................146 Theme Six: OTHER-ORIENTATION ......................................................150 Theme Seven: TANGIBILITY ...............................................................153 Theme Eight: EMERGING AWARENESS ...............................................156 Connotative theme: UNCERTAINTY ....................................................157 Home Reading: Behavioral and Miscue Analysis ............................................159 Home readings one & two: Behavioral analysis of pre-reading/between-reading behaviors .............................................159 Miscue analysis home reading one ....................................................166 Miscue analysis home reading two ....................................................186

viii

Eye Movement Miscue Analysis ......................................................................198 EMMA text one: Silent story book reading .......................................199 EMMA text two: Story book read aloud............................................210 EMMA text three: Expository text read aloud ...................................239 Overview and Preliminary Axial Coding of Results ........................................249 Chapter 5: Discussion and Interpretation of Results .................................................254 The Developmental Dynamics of P‘s Poor Comprehension ............................254 Addressing the Puzzle: Comparison to a Sensitizing Case ...............................258 Clinical evaluation of the sensitizing case .........................................259 Interview with private SLP: New themes and a breakdown ..............261 Resolution of the breakdown .............................................................269 Poor Comprehension as a Systemic Phenomenon ............................................279 Systems theory: A brief overview......................................................280 Systems ontology and epistemology in poor reading comprehension ...................................................................................281 Systems values: Qualitative vs. quantitative stances .........................286 Comprehension: Structure or Quale? ................................................................291 Towards a Grounded Theory i): Poor Comprehension as an Emergent Phenomenon ......................................................................................................297 Towards a Grounded Theory ii): An Ecological Metaphor for Poor Comprehension .................................................................................................303 ‗Struggle‘ vs. ‗drift‘: A Darwinian heterodoxy .................................304 Poor reading comprehension as a quasi-ecological ‗niche‘ ...............307 Chapter 6: Conclusion ...................................................................................................314 Theoretical Implications ...................................................................................315 The concept of poor comprehension ..................................................315 Assessment of poor comprehension...................................................318 Models of reading ..............................................................................321 Clinical Implications .........................................................................................330 P‘s case from a clinical angle: The pitfalls of isolated intervention ........................................................................................330 General clinical implications .............................................................332 Directions for Future Research .........................................................................334 Future research: Social-psychological aspects...................................334 Future research: Linguistic-behavioral aspects ..................................337 Future research: Lack of response to intervention .............................341 Limitations ........................................................................................................342 Concluding Remarks: Ontological Non-Commitments and a Poetic Pragmatist Stance ..............................................................................................346

ix

Bibliography ...................................................................................................................349 Abstract ...........................................................................................................................382 Biographical sketch........................................................................................................384 Appendices .............................................................................................................. see insert

x

List of Tables Table 4.1: Demographics, Assessment, and Clinical Information for P ..........................132 Table 4.2: List of Themes Derived from Interviews with P and his Mother ...................135 Table 4.3: Stimuli for P Home Reading...........................................................................160 Table 4.4: Pattern Analysis of Miscues for P Home Reading One ..................................168 Table 4.5: Home Reading One Miscues Grouped by Combined SemanticSyntactic Alterations ........................................................................................................172 Table 4.6: Themes from Home Reading One ..................................................................186 Table 4.7: Pattern Analysis for P Home Reading Two ....................................................188 Table 4.8: Home Reading Two Miscues Grouped by Combined SemanticSyntactic Alterations ........................................................................................................192 Table 4.9: Joint Themes from Home Readings One and Two .........................................198 Table 4.10: Stimuli for EMMA .......................................................................................198 Table 4.11: Eye Movement Measurements for Text One ................................................203 Table 4.12: Developmental Trends in Eye Movement Measures ....................................204 Table 4.13: Pattern Analysis for EMMA Text Two ........................................................212 Table 4.14: EMMA Text Two Miscues Grouped by Combined SemanticSyntactic Alterations ........................................................................................................213 Table 4.15: Eye Movement Measurements for Text Two ...............................................219 Table 4.16: Joint Themes from Home Readings and EMMA Texts One and Two.........239 Table 4.17: Pattern Analysis for EMMA Text Three ......................................................240 Table 4.18: EMMA Text Three Miscues Grouped by Combined SemanticSyntactic Alterations ........................................................................................................241 Table 4.19: Eye Movement Measurements for Text Three .............................................246 Table 4.20: Overview of All Themes Derived from the Data .........................................250

List of Figures Figure 3.1: The ASL MobileEye Head Worn Eye Tracker .............................................106 Figure 3.2: Example of a Fixation Plot Generated Automatically from Data from a Static Device.........................................................................................................108 Figure 3.3: Example of a Fixation Plot Generated Manually from the Head Worn Eye Tracker Used in this Study .............................................................................109 Figure 3.4: Example Stimulus Material for Eye Tracking ...............................................121 Figure 4.1: A Prosody-based Miscue Indicating Syntactic Weakness ............................169 Figure 4.2: Another Prosody-based Miscue Indicating Syntactic Weakness ..................170 Figure 4.3: An Omission Indicating Syntactic Weakness ...............................................170 Figure 4.4: A Partially Acceptable Miscue (other miscue markings removed) ...............171 Figure 4.5: Grade Level Results for Oral Reading vs. Comprehension Between Assessments ......................................................................................................176 Figure 4.6: Instances of ‗Choppy‘ Reading .....................................................................178 Figure 4.7: Examples of End-of-Line Hesitations ...........................................................178 Figure 4.8: Examples of Between- and Within-Word Hesitations ..................................179 Figure 4.9: Examples of Hesitations Between Phrases ....................................................180 Figure 4.10: Examples of Hesitations between Syllables ................................................181 Figure 4.11: The Only Completely Acceptable Miscue in the Sample (Corrected) .......................................................................................................................184 Figure 4.12: Examples of Good Integrative Strategies ....................................................185 Figure 4.13: Miscues Indicative of Constructive Prediction ...........................................189 Figure 4.14: Interaction of LACK OF INTEGRATION and STRUGGLE FOR ACCURACY .............193 Figure 4.15: Example for PIECEMEAL PROCESSING resulting in Meaning Loss ................193 Figure 4.16: Examples for STRUGGLE WITH PREDICTION ...................................................194

Figure 4.17: Paratextual Comments .................................................................................195 Figure 4.18: Fixation Progressions on Slide One before and after Adjustment ..............201 Figure 4.19: ‗Cluster‘ Pattern of Regressions ..................................................................205 Figure 4.20: Long-range Regression................................................................................206 Figure 4.21: Variability in Fixation Durations ................................................................207 Figure 4.22: Comparison of Variability in Fixation Durations........................................208 Figure 4.23: Comparison of Variability in Fixation Durations per Paragraph ................209 Figure 4.24: Fixations during Lookback ..........................................................................210 Figure 4.25: Miscues indicative of STRUGGLE WITH PREDICTIONS ....................................215 Figure 4.26: Instances of PIECEMEAL PROCESSING (line edited for presentation) .............216 Figure 4.27: Example of Possible Inaccuracies in Recording of Text Two ....................217 Figure 4.28: ‗Fixation‘ vs. ‗Foveal‘ Count of Fixated Words .........................................218 Figure 4.29: Examples of Miscued Words not Directly Fixated on First Pass ................221 Figure 4.30: Example of a Miscued Word Directly and Repeatedly Fixated ..................222 Figure 4.31-1: Nonlinear Relations of Fixations .............................................................222 Figure 4.31-2: …and Concurrent Productions .................................................................223 Figure 4.32: Eye Movement Patterns from Paulson (2005) ............................................224 Figure 4.33: Juxtaposition of Eye Movement Patterns and OR-ER Deviations ..............225 Figure 4.34: Eye Movements and Corrected Miscue on halfpipe ..................................227 Figure 4.35: Patterns Indicative of Struggle in Absence of Struggle ..............................228 Figure 4.36: Fixation Durations on Dozens (left) and have had their turn (right) ...........229 Figure 4.37: Fixation Progression and Durations on momentarily ..................................230

xiii

Figure 4.38: Fixation Progressions on First Reading vs. Search for Errors ....................231 Figure 4.39: Examples for ‗Settling in Behavior‘ ............................................................233 Figure 4.40: Comparison of Variability in Fixation durations per Paragraph .................236 Figure 4.41: Text Fixated During but not Included in Retell ..........................................238 Figure 4.42: Examples for STRUGGLE FOR ACCURACY, PIECEMEAL PROCESSING ................242 Figure 4.43: Another example for STRUGGLE FOR ACCURACY, PIECEMEAL PROCESSING .......................................................................................................................242 Figure 4.44: Examples for STRUGGLE WITH PREDICTION ...................................................243 Figure 4.45: Evidence for Systematic Recording Error ...................................................245 Figure 4.46: Efficient Sampling of Picture Information ..................................................248 Figure 4.47: A Hypothetical Outline of P‘s Path to Poor Comprehension ......................251 Figure 5.1: A Thematic Outline of P‘s Trajectory ...........................................................255 Figure 5.2: Amended Thematic Outline of P‘s Trajectory ..............................................272 Figure 5.3: A Thematic Outline of S‘ Trajectory ............................................................274 Figure 5.4: P‘s Trajectory, Conceptualized as a Loop .....................................................275 Figure 5.5: The Causal-Componential View of Poor Comprehension ............................298 Figure 5.6: The Systemic View: Poor Comprehension as an Emergent Phenomenon.....................................................................................................................299 Figure 5.7: Niche Construction ........................................................................................309 Figure 5.8: The Niche of Poor Comprehension (for P‘s case) .........................................311 Figure 6.1: P‘s Places in a Simple Taxonomy of Reading Difficulties ...........................320

xiv

Chapter One: Introduction There is no shortage of evidence pointing to the fact that some individuals experience reading comprehension difficulties. The nature and origins of reading comprehension difficulties, however, are not so clear. Nation (2005, p. 248) In ‗cutting up‘ a system, the analytical method destroys what it seeks to understand. Cilliers (1998, p. 2) Poor comprehenders are defined as children who ―read [out loud] accurately but have specific difficulty understanding what they read‖ (Nation, 2005, p. 249). Similarly, Cain and Oakhill (2006) speak of ―children with fluent and accurate word reading in the presence of poor text comprehension‖ (p. 683). According to some estimates, 16 to 31 percent of struggling readers between second and eighth grades exhibit such specific comprehension deficits (Catts, Hogan, & Adlof, 2005). The population is highly heterogeneous: some poor comprehenders have been diagnosed with language impairment, others with cognitive impairment, but neither of these impairment types is found in the population of poor comprehenders as a whole (Nation, 2005; Nation, Clarke, Marshall, & Durand, 2004). The first author to study the phenomenon of poor comprehenders was Cromer (1970); research has been conducted ever since. Yet despite the prevalence of the phenomenon, and despite the fact that it has been investigated for more than four decades, no sufficient explanation for its occurrence and etiology has been found (Nation, 2005); Nelson comments that among populations with reading difficulties, poor comprehenders are possibly the least understood (personal communication in Gruen & Nelson, 2011). It was this puzzling state of affairs that sparked my interest and prompted me to undertake the present study. While reviewing the pertinent literature, I became convinced that it was not lack of effort or interest on part of the academic community that was at the heart of the issue. Rather,

it seemed that the methods used for investigation, and the theoretical background on which research is based, are ill-suited to the task. This is especially true for studies conducted in the wake of Yuill and Oakhill‘s (1991) landmark tome on the subject; their impressive effort not only re-ignited interest in the phenomenon, but set the tone for a very specific line of inquiry. Most studies since their publication nearly 25 years ago have been operating from a ‗reductionist‘ viewpoint (Brigandt & Love, 2008), using experimental methods to detect assumed causal relations between different aspects of poor comprehension. This endeavor has been unsuccessful so far in that no single causal factor for poor comprehension has been identified (Cornoldi, De Beni, & Pazzaglia, 1996; Nation, 2005), and no satisfactory theoretical explanation for the phenomenon has been offered. Inquiry into poor reading comprehension is embedded in the framework for the study of reading in general, and it could therefore be assumed that it is subject to the same methodological and theoretical quandaries. Much of mainstream reading research has concentrated on lower-level literacy competencies such as phonemic awareness and ‗decoding‘ of print, as well as pertinent literacy difficulties such as dyslexia. Less attention has been paid to higher-level competencies such as linguistic processing, inferencing, use of context and background knowledge, and comprehension (Damico & Nelson, 2010; Wagner, Schatschneider, & Pythian-Sence, 2009a). The same is not true for research on poor comprehension, which has extensively investigated linguistic aspects of literacy. However, this work has done so from a very specific perspective, at least for the past two decades. The general theoretical background for research post-Yuill and Oakhill (1991) is the Simple View of Reading (Hoover & Gough, 1990), which posits a disconnect between decoding and comprehending. In this orientation, decoding is viewed as a skill specific to reading and

2

comprehending is thought to involve the same competencies as needed to understand oral language. Consequently, reading is conceptualized as the product of these correlated but theoretically unrelated skills. Implicit in this model is the rarely articulated assumption that reading is largely a passive process of accurately transforming incoming stimuli from the visual to the auditory modality so that linguistic processing can occur rather than an active effort at making sense of visual linguistic input (cf. Damico & Nelson, 2010; Pearson, 2009; Smith, 2004). Informed by this model and its underlying assumptions, most research over the past two decades has studied a variety of specific linguistic behaviors as potential causes for poor comprehension, such as resolution of pronouns, or making inferences during processing. Based upon many of these studies, it has been shown that poor comprehenders have deficiencies in basically every aspect of linguistic processing of print. In this sense, research on the population has been highly successful, but as the underlying theoretical assumptions require the detection of one single causal factor, authors conclude, in a somewhat puzzled manner, that the phenomenon is poorly understood. The irony here is that comprehension is an unobservable mental process (Pearson, 2009), which begs the question as to why its behavioral concomitants should be indicative of ontologically discrete, causally influential mental entities (cf. Smith, 2004). These assumptions are axioms on which most current studies rest, and which are rarely questioned (for one such incidence of doubt see Nation, 2005). It is interesting to juxtapose this state of affairs with the research prior to Yuill and Oakhill (1991). Investigators in the 1970s and 1980s had a more ‗active‘ view of readers, and viewed reading as an integrated linguistic process that comprises all language skills from the

3

graphic/phonemic to the discourse level. Hence, they adopted a slightly different perspective: instead of exploring poor comprehenders‘ deficits (the absence of expected skills), they were primarily interested in the extent to which these children are different from typical readers in what they do when reading. Also, researchers used more variegated methods, adding qualitative analyses of reading behaviors and interviews with poor comprehenders to experimental designs. Their results are concurrent with later findings, but there is a major difference in their conclusions – earlier authors had no compunctions about interpreting their results. Based on their view of reading as an active, integrated process, they concluded that poor comprehenders are characterized by an undue focus on lower-level aspects of print such as phonics or phrases, and by a concomitant failure to use higher-level linguistic skills such as inferencing or comprehension monitoring. Garner (1981) coined the term ‗piecemeal processing‘ to describe the prevalent reading behavior in the population. In addition, poor comprehenders were shown to have little understanding of the reading process. They did not seem to grasp that comprehension is the goal of reading, and tended to lack motivation to engage in it. In sum, the comparison of the earlier with the later period of research left me puzzled; while findings are completely coherent, conclusions drawn from them are quite different. The best explanation I had was that the divergent conclusions were the result of the different theoretical models that informed methods and interpretations. I therefore decided to design the present study as an explicit alternative to current research, harking back to but going beyond methods used in the earlier period of investigation. Poor comprehension, I resolved,

4

needed to be investigated using qualitative methodologies, and a different set of theoretical assumptions for guidance. The rationale for this decision was that qualitative methods of inquiry provide an alternative to the reductionist quest for experimentally verifiable causal relations. In the qualitative tradition, the goal is to generate rich, comprehensive descriptions of naturally occurring, complex phenomena (Geertz, 1973; Lincoln & Guba, 1985), including observable behaviors as well as ‗participant meanings‘ or lived experiences (Damico & SimmonsMackie, 2003; Janesick, 1994; Simmons-Mackie & Damico, 2003). A qualitative account of a phenomenon of interest is successful if it enables understanding (verstehen) of its subject matter, as opposed to merely explaining it in terms of causal relations (Bransen, 2001). So far, only a few studies have investigated poor comprehenders from a qualitative perspective, and some evidence suggests this might be a fruitful avenue. As an example, it was found that some children in this population may have less print exposure than typical peers (Cain, 1996; Cain & Oakhill, 2011), and that both their understanding of reading and their experience of it are different from good comprehenders (Canney & Winograd, 1979; Persson, 1994). However, subsequent studies have not followed up on these findings. Likewise, there has been virtually no investigation of poor comprehenders‘ reading behaviors such as eye movements or miscues during reading of authentic texts (i.e. texts that have not been devised specifically for experimental purposes). In other words, while there is a wealth of knowledge on poor comprehenders‘ performance on a range of experimental tasks, their actual behaviors while reading have gone undocumented. To provide guidelines for the inquiry, Goodman‘s (1994) transactionalsociopsycholinguistic model of reading was chosen as a theoretical framework. Accordingly,

5

miscue analysis (Goodman & Goodman, 1994) and eye movement miscue analysis (Paulson, 2000) were adopted as main methodological tools, and supplemented by classic methods of qualitative inquiry such as interviews, participant observation, and artifact analysis (Damico & Simmons-Mackie, 2003). The initial goal of the study was to investigate, as comprehensively as possible, poor comprehenders‘ reading behaviors (specifically their miscues and eye movements); their lived experience of reading; and the history of their exposure to print. The goals of the study were a) to provide a rich description of poor comprehenders, and b) to devise a theoretical explanation for the observations obtained through rich description. Given the latter goal, it was decided that the study would operate within the methodological framework of Grounded Theory (Glaser & Strauss, 1967). However, things did not progress as I had initially anticipated (a common feature of qualitative research, cf. Janesick, 1994). Two obstacles changed the course of the inquiry. First, finding participants turned out to be more difficult than expected. This was partly due to the unwillingness of local stakeholders (teachers and parents) to disclose information on and grant access to their young readers. In addition, there were issues with the assessment tools employed. My initial goal was to study a minimum of three cases; I ended up with a comprehensive dataset on one poor comprehender, a less than comprehensive set on a second one, and an extensive amount of data on a young reader who turned out to be an overall poor reader when I reassessed my participants. This situation turned out to be less problematic than it could have been, for the second obstacle I encountered was the sheer scope of the inquiry. Comprehensive, in-depth analysis of a large amount of variegated data such as those garnered in this study is surprisingly labor-

6

intensive. As will be seen, analysis of the data on the first case alone made for such richness of interpretation that it was ultimately decided to limit the investigation to this participant. What emerged from the one dataset examined was a true alternative to prior studies, not only in terms of findings, but also as regards their interpretation and the ensuing theoretical outlook on the issue of poor comprehension. As will be seen, all findings appeared to converge to lead me to the conclusion that the causal-componential view is a misguided way of thinking about the issue. Instead, I was pointed to the distinct possibility that poor comprehension may best be viewed as a systemic phenomenon. That is, an emergent property of the interactions of a reader with his/her environment over time wherein reader-intrinsic factors collude with aspects of the readers social and academic experience to create what we view as poor comprehension, but what from the reader‘s view may simply be a good-enough fit of his/her abilities with environmental expectations. From this perspective, it would seem unsurprising that the quest for single underlying causes should have been futile. As in the quote by Cilliers (1998) above, this quest has dismantled the whole that it sought to understand. I am aware that these are bold assertions, in particular as they are based on one case alone. It therefore needs to be mentioned that this study is the attempt to walk a tightrope between boldness of interpretation, and humility of purported scope. It is bold in that it uses its data to demonstrate a new way of studying and understanding poor comprehension. At the same time, it aims for humility in cautioning that its methods, findings, and conclusions are merely suggestions; they may be a better way of thinking about and studying the issue. Its claim is to introduce a promising alternative to existing research and theorizing – nothing more, nothing less.

7

The discussion will proceed as follows. In Chapter Two, I will review the literature, not only on poor comprehension, but also on models of comprehension in general, and I will tie it back to the theoretical views on reading that have informed research. In Chapter Three, I will introduce the methodology and tools I used, and the epistemological background in which I place this inquiry. Chapter Four will provide an in-depth discussion of findings; Chapter Five will be devoted to bringing these findings together by formulating a preliminary grounded theory on poor comprehension. Chapter Six will conclude the discussion with an outlook on theoretical and clinical implications, and suggest lines for further research.

8

Chapter Two: Review of the Literature And so to completely analyze what we do when we read would almost be the acme of a psychologist‘s achievements, for it would be to describe very many of the most intricate workings of the human mind. Huey (1968/1908, p. 6) The wonder that the ‗intricate workings of the mind‘ during reading and reading‘s ‗tangled story‘ evoked in Huey can still be felt a century later. It can also be argued that the acme of complete analysis of the reading process has not and may never be reached. The workings of mind ultimately elude description, and the observer has to resort to what is observable in human behavior in order to get a glimpse of deeper phenomena. This review will proceed as follows. First, there will be a brief discussion of the notion of reading comprehension and the overall history of reading research and instruction in which it is embedded. Then there will be a review of the literature on poor comprehension in the current meaning of the term, which developed around 1970. Reading Comprehension Comprehension is now agreed upon as the goal of literacy instruction. However, this has not been the case throughout history. Until recently, comprehension has gone rather unacknowledged. As we will see, the reason for this is twofold. First, definitions of literacy have changed over time. Second, and possibly more importantly, comprehension is by its very nature elusive and difficult to pinpoint, as ―the boundaries of the topic are so broad and so poorly marked. Reading comprehension is only a subset of an ill-defined larger set of knowledge [about communication]‖ (Paris & Hamilton, 2009, p. 32). This review will first focus on the historical development of how we understand comprehension, and then proceed to discuss descriptive accounts and definitions that have been offered.

A brief history of views on reading comprehension. At any given point in history, what is viewed as ‗reading‘ is a social construction: Being literate, over the course of the centuries, could mean anything from the ability to sign one‘s name to reading at a late elementary level (Damico, Nelson, & Bryan, 2005). This partly explains why, for much of recorded literacy history, reading comprehension was not the primary object of educational effort and theoretical attention: ―The default indicator of reading prowess in the 17th to 19th centuries was definitely oral capacity‖ (Pearson, 2009, p. 4). The advent of modern schooling and the changing needs of an industrializing society changed this outlook. In the first half of the 20th century, behaviorist psychology introduced multiple-choice comprehension questions and retells as comprehension measures. Starting with the ‗cognitive revolution‘ in the second half of the century, more elaborate models of comprehension were devised, drawing, for example, on schema theory and social-interactional views of learning. Still, it was not until the 1980s that the comprehension of reading material became truly established as a matter of import in educational research and practice. After a brief period of bloom, this outlook was again superseded by a return to the ‗basics‘ of accuracy and fluency during the mid-1990s. Comprehension assumed its old place in the background and has only recently begun to reemerge (Pearson, 2009). The history of reading comprehension as an object of study and practice is embedded in the similarly lively overall history of reading research and instruction. If we follow Huey‘s (1968/1908) account, reading teachers‘ method of choice, from classical antiquity up until the late 1800s, was a bottom-up approach he terms the ‗alphabet method‘: ―the child learned first (…) the alphabet. (…) Then the combinations like ab, eb, ib, were spelled out and pronounced (…) Then (…) gradually longer and longer words were used‖ (Huey, 1968/1908,

10

p. 254). Despite the prevalence of this method, approaches that resemble modern practices were developed and implemented from as early as the 16th century onwards. Huey describes materials similar to today‘s alphabet and illustrated text books; early educators advocated for sight-word reading, but also for contextualized approaches where words are first encountered in text and analyzed later. Others devised phonetic alphabets for unambiguous decoding. In Huey‘s time, the ‗alphabet‘ method appears to have been retreating, and replaced by combinations of ‗phonic/phonetic‘ methods and ‗word‘ or ‗sentence‘ instruction. ‗Phonic‘ refers to the same method used today, where words are sounded out letter by letter. ‗Phonetic‘ instruction used specially devised alphabets to this end in which each character represents a sound to facilitate decoding. ‗Word‘ approaches were based on sight-word recognition. The ‗sentence‘ method is intriguing in that it appears to represent an early childcentered approach: A sentence uttered spontaneously by a student was written down by a teacher for the student to practice recognition of printed materials on a text of interest. Finally, the author mentions a method in which children imitate in chorus what the teacher reads while attending to the pertinent print (Huey, 1968/1908, pp. 265-275). Starting with the introduction of widespread educational testing in the first few decades of the 1900s, the notion of ‗reading skill‘ was developed, that is, the assumption that reading can be dissected into discrete, teachable components (Pearson, 2009). In consonance with the importance placed on oral reading capacity, much of early 20th-century educational practice centered on the extensive drilling of phonics (Beck, 1981). In the 1920s, the ‗looksay‘ or ‗whole-word‘ approach was advocated as an alternative to and eventually gained the upper hand over phonics; like the latter, however, it did not offer guidance for remediation in students who failed to achieve expected competencies. This prompted the forceful return of

11

phonics in the 1950s; the controversies over best practices in reading instruction have been persisting ever since. In 1967, Jean Chall published a landmark volume on the topic, in which she concluded that code-based instruction (phonics and related bottom-up approaches) are superior to more holistic approaches; she also introduced the notion of ‗stages‘ to be mastered in reading development, and claimed that ‗decoding‘ was one of these, usually occurring around 1st and 2nd grade (Beck, 1981; Calfee, 2009; Chall, 1983; Itzkoff, 1996; Pearson, 2009). The 1980s saw the reemergence of holistic outlooks, with Whole Language as the best-known position amongst a variety of integrated models of reading. The simmering controversy flared up again, and by the early 2000s the skills-based, sequential, bottom-up view had gained the upper hand, notably through introduction of the No Child Left Behind Act in 2001 (Calfee, 2009; Gavelek & Bresnahan, 2009; Gerstl-Pepin & Woodside-Jiron, 2005; Itzkoff, 1996; Pearson, 2009). Comprehension once more assumed a place on the back burner, but not for long: the late 2000s saw a reemergence of concern with meaning (Pearson, 2009), this time embraced even by proponents of skills-based instruction, as seen in book titles such as Beyond decoding (Wagner, Schatschneider, & Pythian-Sence, 2009b). Definitions of reading comprehension. Regardless of their orientation, theorists now agree that comprehension is the ultimate goal of reading instruction (e.g. Goodman & Goodman, 2009, p. 91; Paris & Hamilton, 2009, p. 32; Snowling & Hulme, 2005, p. 207; Wagner, Schatschneider, & Pythian-Sence, 2009a, p. xi). The main reason for its past lack of prominence is, of course, that comprehension cannot be directly observed. As Pearson explains, readers may tell the researcher whether or not they have understood the text, how they feel about it, answer comprehension questions, critique or summarize the text, and so on but ―all of these tasks (…) are little more than the residue of the comprehension process

12

itself‖ (Pearson, 2009, p. 4). Huey (1908) posits that ―the consciousness of meaning itself belongs (…) to that group of mental states, the feelings, which I regard with Wundt as unanalyzables‖ (p. 163, emphasis mine). If we assume with Pearson (2009, p. 6) that ‗consciousness of meaning‘ is the same as ‗comprehension‘, then this statement suggests that Huey viewed comprehension as one of the qualia, that is, as belonging to the category of ―introspectively accessible, phenomenal aspects of our mental lives‖ (Tye, 2009). This would entail that reading comprehension is not only part of a very poorly understood subject area, viz. the study of consciousness (van Gulick, 2011), but also that it is ultimately inaccessible to objective measurement (Tye, 2009). Philosophical reflections on the nature of understanding and meaning regarding comprehension echo these concerns. Wittgenstein (1984/1953), musing on the impossibility of verifying others‘ mental states, concludes that linguistic meaning depends on the way a word (or, we might add, any other unit of language) is conventionally used for interactional purposes. Understanding can only be inferred – but never directly described and accessed – by observing the other person‘s behavior. If the behavior accords with conventional expectations, we assume comprehension of the linguistic unit. Quine (1960) proposes that linguistic meaning is ultimately indeterminate and inscrutable; by extension, then, the same would apply to comprehension of meaning. Given this fundamental ineffability, it is not surprising that multiple and at times contradictory models of comprehension have been devised but no single one has been agreed upon. Paris and Hamilton (2009) list four different types of such models. ‗Information processing‘ types view the reader as a computer-like entity that applies circumscribed sets of mental operations to textual input, yielding output in the form of comprehension. ‗Construction-integration‘ models balance bottom-up and top-

13

down processes. Constructivist models conceptualize reading as an active and situated process that involves a reader guided by intention. Finally, ‗multi-component‘ models attempt to integrate all known facets of comprehension, tacitly assuming that those can be viewed as separate modules. They are ―open-ended because any linguistic, cognitive, or social skill might potentially influence comprehension‖ (p. 39). One of the puzzling aspects of this subject area is that despite the elusive nature of comprehension, educators and researchers have had no qualms to suggest that it can be adequately captured and described through varying sets of circumscribed behaviors. An early list of comprehension subskills was compiled by W. S. Gray. It includes ‗simple‘ and ‗complex interpretation habits‘; the sheer detail of the compilation is illustrative in that it makes clear the scope of behaviors and competencies that can be thought of as part of reading comprehension. Among the ‗simple habits‘ identified by Gray are: Concentrating attention on the content; associating meanings with symbols; anticipating the sequence of ideas; associating ideas together accurately; recalling related experiences; recognizing the important elements of meaning; [and] deriving meanings from the context and from pictures. (Gray, 1925, cited in Pearson, 2009, p. 8-9; bullet points changed to enumeration) The ‗complex habits‘ include: Analyzing or selecting meanings (to select important points and supporting details, to find answers to questions…); associating and organizing meanings, for example, to grasp the author‘s organization, to associate the what is read with previous experience, to prepare an organization of what has been read; evaluating meanings, for example, to appraise the value of significance of

14

statements, to compare facts read with items of information from other sources, to weigh evidence presented, to interpret critically; retaining meanings, for example, to reproduce for others, to use in specific ways. (Gray, 1925, cited in Pearson, 2009, p. 9; bullet points changed to enumeration) Pearson identifies Gray‘s list as one of the first in a long line of attempts to define reading comprehension as a set of teachable subskills. Interestingly, subsequent lists became considerably less exhaustive. In 1949, McKee listed vocabulary, use of context for inference of word meanings, figurative language, syntax, linkage of ideas across sentences, distinguishing emotions from facts, and text discussion as conducive for comprehension. In 1968, Clymer offered a taxonomy of text properties (main ideas, supporting details, sequence, comparison, causes and effects, and character traits) and how they fit into two types of comprehension, literal (recognition and recall) and inferential (Pearson, 2009). With the emergence of psycholinguistics in the late 1950s, and the ‗cognitive turn‘ in psychology during the 1970s, researchers were empowered, after a long period of domination by behaviorism, to ―re-embrace and extend constructs such as human purpose, intention, and motivation to a greater range of psychological phenomena‖ (Pearson, 2009, p. 12). Schema theory came to be one of the ―most popular and influential movements‖ (p. 12) of this era. According to this account, human knowledge is organized in structured mental representations that guide our experience through a system of ‗slots‘ we expect to be ‗filled‘ with information of a certain kind. As an example, the ‗cocktail party‘ schema would include slots for alcohol drinks and stand-up conversations but not for sitting down and having elaborate meals, which would belong in the ‗restaurant‘ schema. Schemata can operate both in terms of background/world knowledge, and in terms of print and genre knowledge,

15

facilitating grasp of subject matter as well as organization of a text (Pearson, 2009; Schmidt & Weischenberg, 1994). They are formed from prior knowledge; schema theory therefore provided a theoretical rationale for investigating such knowledge in readers, which was found to be an excellent predictor of comprehension. Since comprehension was viewed as an active process, this theory favored a constructivist-interactional view of comprehension. Meaning and comprehension were thought to reside neither in texts nor in readers, but in the interaction in which a reader makes sense of print (Pearson, 2009). Another important facet of the schema view was the introduction of the notion of metacognition into the field. The reader was not only a ‗builder‘, constructing meaning from text, but also a ‗fixer‘ with the ability to detect faulty design of his own meanings, and to call up remedial strategies where needed. Metacognitive knowledge, (i.e. knowing when and why a specific piece of knowledge should be put to use), came to be seen as a third type of knowledge, added to the agreed-upon distinction of knowing-that (declarative) vs. knowinghow (procedural). ―The real contribution was helping us understand that we cannot characterize comprehension (…) without including all of these‖ (Pearson, 2009, p. 15). Schema theory had just become the most influential approach by the mid-1980s when it was challenged for being too simplistic. Schemata seemed too static and abstract to account for the dynamic, episodic, image-rich mental activities during reading; hence, an enhanced version was devised under the name of ‗mental models‘ (e.g. Johnson-Laird, 2005). This work ―reached its zenith (…) in the work of Kintsch‖ (Pearson, 2009, p. 18). In Kintsch‘s account, text is seen as providing insufficient information for more than cursory understanding. For full comprehension, the reader needs to construct a mental representation of what is described in the text (Kintsch & Rawson, 2005). As the authors note, this ―requires

16

the integration of information provided by the text with relevant prior knowledge and the goals of the comprehender‖ (p. 211). Comprehension is viewed as a function not only of knowledge and processing but also of reading purpose. Spiro and colleagues (1987), start from a similar premise critical of schema theory, and arrive at a similar yet broader conclusion suggesting that understanding complex and illstructured domains requires cognitive flexibility and nonlinear learning; comprehension may require assimilation of seemingly contradictory aspects of the same domain, rather than simple application of a ready-made schema or model. Like Kintsch, the authors highlight the role of reading purpose, explaining that ―instead of prepackaged schemata, purpose-sensitive situational schemata are constructed, thus allowing knowledge to be used in different ways on different occasions for different purposes‖ (p. 4). Another main strand of thought, based on the Vygotskian tradition, shifted the focus from presumed mental structures to the context in which reading occurs. No activity, the argument went, happens in abstract space; any reading done by a student is embedded in a specific sociocultural and interpersonal situation, as well as in a web of interactions with text (Pearson, 2009). It may be added here that reading comprehension is also grounded in bodily experience: As Lakoff and Johnson (2003/1980) show, much of everyday language is based on metaphorical translations of physical perception into more abstract domains. Kaschak et al. (2009) sum up pertinent research and conclude that even though results so far have been promising, they have not yet yielded a full-fledged theory of embodied text comprehension. A final alternative was offered by the notion of reader-text transactions, with a shift in emphasis towards personal reactions of readers (Pearson, 2009). As the rationale for this study is partly based on these models, they will be discussed in greater detail here.

17

The classic transactional model comes out of Rosenblatt‘s work. In a summary of her outlook, Rosenblatt (1994) likens the transition towards transactional accounts to the paradigm shift in natural sciences that led from a Cartesian/Newtonian world-view to a ‗postEinsteinian‘ view informed by quantum mechanics. Both reject the notion of an observer (or reader) separate from the object of observation (or text); any act of observation is a whole that encompasses both: ―The knower, the knowing, and the known are seen as aspects of ‗one process.‘ Each element conditions and is conditioned by the other in a mutually constituted situation‖ (p. 1058). The term ‗transaction‘ was chosen by Dewey and Bentley to emphasize the holistic aspect of the concept; instead of separate entities ‗interacting‘, ‗transaction‘ evokes a more ‗ecological‘ dynamic in which reader and text take part, and through which they are organically changed. Meaning (or comprehension) arises out of this process (one is tempted to say that meaning is the change that occurs in it) and is seen as dynamic, rather than static. Rosenblatt does not reject the notion of mental patterns such as schemata, but insists with Bartlett that such concepts be seen as processual, momentary and situated, rather than simple entities sitting on a mental shelf to be pulled off for usage. Goodman (1994) adopts a very similar stance but gives it a distinctly interdisciplinary flavor. Like Rosenblatt (1994), he locates text meaning in the process of reading and writing: The text has a potential to evoke meaning but has no meaning in itself. (…) Meaning (…) does not pass between writer and reader. It is represented by a writer in a text and constructed from a text by a reader. Characteristics of writer, text, and reader will all influence the resultant meaning. (p. 1103)

18

The author then sets out to make explicit the major factors that shape what happens in the reading transaction. These can broadly be identified as social, psychological, and linguistic. The social provides context for reading in two ways. First, any use of written language is an act of communication, and as such constrained by factors present in any communicative situation, for example, reliance on unspoken ‗contracts‘ or ‗maxims‘, as noted by Grice (1975). In reading, the main contract between reader and writer is that ―the reader wants to understand and will try to do so [since] no text can be so successfully constructed (…) that it can be comprehended without the reader actively trying to make sense of it‖ (Goodman, p. 1104). Similarly, in order to serve its communicative functions, written language has to adhere to accepted conventions. Obviously, much of reading is socially induced, as is the case, for example, while reading at work or at school. The second social aspect pertains to the process of learning linguistic conventions in that all language development is the expression of (personal) meaning by conventional (social) means. To do so, the learner must ‗invent‘ the pertinent convention for him- or herself, that is, construct personal meaning by making tentative use of conventional means. This leads into the second, psychological aspect of the reading transaction. Goodman (1994) presents this as a wide field, broadly informed by schema theory and systemicfunctional linguistics but going beyond both. Both purpose and intention for engaging in a reading transaction play a vital role, as ―different intentions may strongly influence comprehension. [E.g.] reading to cram for an exam is different from reading for a term paper or to fulfill a burning desire to know more about something‖ (p. 1118). Goodman also lists a set of cognitive strategies, such as initiation, sampling, inferencing, predicting, selfmonitoring, and correction, that are deployed during the transaction while the reader goes

19

through ‗cycles‘ of processing visual input, organizing it into perceptual ‗chunks‘ to construct a structure from which words derive their meanings. In the process, Piagetian processes of assimilation (of incoming information to existing schemata) and accommodation (of existing schemata to incoming information) take place continuously. Given the dynamic nature of the transactional process, Goodman suggests a distinction between comprehending (making sense of text) and comprehension (the end product of making sense of text). His approach places strong focus on studying the former, chiefly by investigating the linguistic aspect of reading, described in terms of cueing systems, i.e. subsystems of language that offer cues to the appropriate deployment of the strategies described above. The taxonomy of cueing systems is taken directly from established linguistic description, commonly divided into graphophonics (the written-mode equivalent to phonetics/phonology), syntax, semantics, and pragmatics (Goodman & Goodman, 1994; Goodman, Watson, & Burke, 2005). At this point, brief mention needs to be made of Smith‘s (2004) account. It bears strong similarity to transactional views but adds an information-theoretical spin that elucidates the relation between visual input and linguistic processing. According to Smith, visual information is sampled as needed to enable prediction of linguistic content; prediction is made possible by using both this information and prior knowledge to reduce uncertainty of what might come next. The prediction is then gauged against new input and either confirmed or corrected. Given the well-established limitations of human working memory (Miller, 1956), input must be integrated (or ‗chunked‘) into a larger context (of text or background knowledge) in order to be recognized ‗as‘ something. If input exceeds working memory limits before it is chunked (e.g. if a reader attempts to hold in mind more than ‗seven, plus or

20

minus two‘ letters at a time) working memory is easily overwhelmed and comprehension is lost. Therefore, visual sampling must be as efficient as possible, and reading cannot take place without organization of visual input into meaningful chunks. This model is quite different from a final account that merits consideration, the Simple View of reading (Gough, Hoover, & Peterson, 1996; Hoover & Gough, 1990), in which reading is operationalized as the product of language comprehension (both in reading and ‗auding‘, that is, listening for comprehension) and decoding, or R = C x D. If either skill is null, then reading is also not present. The Simple View is distinct from constructivist and transactional views discussed above in that it a) presupposes the dissociation of decoding and comprehension; b) assumes broad similarity of linguistic comprehension in reading and listening (conceptualized, for both cases, as ―the ability to take lexical information (…) and derive sentence and discourse interpretations‖ [Hoover & Gough, p. 131]); and hence c) does not take into account that reading requires active use of strategies, and that these strategies may be quite different from those used for ‗auding‘ (Hoffman, 2009). Consequently, the authors characterize reading as a bottom-up process not in the strict sense of the term (i.e. in the sense that all reading always occurs in a linear sequence from decoding to comprehension) but in the general sense that ―the more proficient the reader, the less the reliance on context (…) [Therefore,] fluent reading may best be characterized as a bottom-up process‖ (p. 131). It is not the author‘s intent to criticize the Simple View at this point; suffice to note that it is ignored by much of the scholarship in the field due to its failure to incorporate the idea of an active reader (Hoffman, 2009). Also, the notion that proficient reading correlates negatively with reliance on context is questionable, as we will see further below in this

21

chapter. The Simple View‘s importance for this study stems from the fact that most of the research of poor reading comprehension has explicitly or implicitly built on it. This can be explained by the acclaim it has received in the broader field of literacy research and policy (Hoffman, 2009), but also—and importantly—by the fact that it provides the conceptual framework commonly used to categorize reading problems. With its assumption of a disconnect between decoding and comprehending, it informs our ideas of what constitutes good reading (good comprehension and good decoding), dyslexia (poor decoding but no comprehension problems), poor comprehension (absence of decoding problems) and general poor reading (with problems in both comprehension and decoding) (Nelson, 2010, p. 147). In sum, extant models of reading comprehension range from the mechanistic and modular to the dynamic, organic, and transactional. Put otherwise, there is a continuum from an emphasis on the interplay of separate components of cognition to a view of reading as a communication situation that involves active individuals and their purposes. Much of the research on poor comprehension has been informed by the rather mechanistic, componential assumptions of the Simple View; it will be discussed, during the review, to what extent this informs the current state of scholarship in the area. Poor Reading Comprehension While reading comprehension has been under scrutiny for more than a hundred years, poor comprehension in the current sense of the term (i.e. absent decoding problems) has been discussed for approximately four decades. A brief discussion of the history and definitions of the term will therefore suffice; after this, a review of pertinent research will be presented. There is a distinct turning point in the literature that occurs after Yuill and Oakhill‘s (1991) landmark publication; virtually all subsequent research was done along the threads laid out in

22

their tome. The organization of the review will therefore roughly follow those threads and attention will be devoted to how this work may have influenced current views. History and definitions of the term. The earliest mention of poor reading comprehension in the absence of decoding problems seems to come from Cromer‘s (1970) study of poor readers who exhibited ―a mismatch between the individual‘s typical mode of responding and the pattern of responding assumed necessary for adequate reading‖ (p. 471). Cromer claimed that the reason for his participants‘ comprehension difficulties, their ‗typical mode of responding‘, was a lack of active construction of meaning from text: although they have the skill to read aloud and say individual words correctly, they do not organize their reading input in a meaningful way, that is, they organize input in a word-by-word fashion rather than into meaningful units such as phrases. (p. 472) Yuill and Oakhill (1991) remark that Cromer‘s findings were ―vigorously attacked‖ (p. 4) for they were at odds with the prevailing view that comprehension occurs automatically when decoding is established. Cromer‘s take on the issue is echoed by Levin (1971, p. 19), who focuses his review on ―the child who can identify the words, knows the meaning of individual words, but has difficulty in integrating the separate meanings into an organized whole‖. Golinkoff (1975) appears to be the first publication that uses the term ‗poor comprehenders‘. The author concurs with previous definitions: ―the poor comprehender seems to read text in a word-by-word manner, with a minimum of text organization. She is also generally inflexible to variations in task demands‖ (p. 36). Similarly, Garner (1981) proposes a ‗piecemeal processing‘ explanation, according to which

23

―proficiently decoding poor comprehenders‖ (p. 159) have trouble understanding print because they ―manage written language as bits and pieces, not as textual wholes‖ (p. 161). Oakhill and Garnham‘s (1988) treatment foreshadows the threads laid out in the classic tome by Yuill and Oakhill (1991). They define the problem as one of children having ―adequate word recognition skills but (…) not [understanding] what they read as well as they might be expected to‖ (p. 119), and differentiate it from hyperlexia, a similar phenomenon that is accompanied by language delays and usually occurs with developmental disorders such as autism. With one exception (Kintsch & Rawson, 2005), this distinction is held up throughout the literature. Yuill and Oakhill (1991) define the population as ―children who (…) are fluent readers, but who show little evidence of understanding what they read‖ (p. 2). Note that the definitions of the issue undergo a subtle but possibly decisive change. Prior to Yuill and Oakhill (1991), authors had little compunction attributing poor comprehenders‘ difficulties to what they do when they read, as seen in formulations such as ‗word-by-word reading‘ or ‗piecemeal processing‘. By contrast, later studies building on Yuill and Oakhill focused on necessary but absent behaviors (such as metacognition) or on deficits (e.g. in working memory). In Cromer‘s (1970) terms, a ‗difference‘ perspective was replaced by a ‗deficit‘ perspective, with potentially far-ranging consequences. The shift seems to have occurred within the classic works. Oakhill and Garnham (1988, p. 118-119) do remark that poor comprehenders‘ problems may be caused by lack of understanding that the goal of reading is meaning construction, and the consequent failure to adjust reading strategies. Yuill and Oakhill (1991, p. 10-11) discuss the role of motivation and attitudes in comprehension failure, but dismiss motivation as a factor since their experimental designs

24

were deemed to control for it, and do not pursue attitude further. As a result of this cursory treatment, these qualitative factors have scarcely been studied in subsequent research. In publications after Yuill and Oakhill (1991), the definition of poor reading comprehension has remained the same. Sometimes the term ‗specific comprehension difficulties‘ is used; also, ‗word recognition‘ is frequently used to mean ‗decoding‘. In this study, the terms ‗poor comprehenders‘ and ‗poor comprehension‘ will be taken up. However, I will distinguish between ‗decoding‘ and ‗word recognition‘ because in transactional views of reading, ‗recognition‘ of individual words is intimately connected with linguistic processing of the larger textual wholes in which they are embedded (Goodman, 1994; Smith, 2004), which is a problem area in poor reading comprehension. Consequently, poor comprehenders will be defined as being proficient at ‗decoding‘, not ‗word recognition‘. Text coherence and integration. The earliest study in the area (Cromer, 1970) investigated the abilities of college-aged poor comprehenders to integrate print to meaningful wholes and construct coherent meanings from it. Cromer classified half of his poor readers as ‗difference‘ and the other half as ‗deficit‘ as per their ‗vocabulary skills‘ (the ―ability to elaborate meanings of words‖, p. 472). Reading material was organized in four different ways: a) sentences, b) single words, c) meaningful phrases, and d) non-meaningful groups of words. While both groups of poor readers comprehended significantly less well overall than the good readers, the ‗difference‘ poor readers comprehended as well as controls when the material was ‗chunked‘ into meaningful phrases. Also, their performance did not differ between conditions a), b), and d), suggesting that they approached all three modes with the same behavior, viz. word-by-word reading. Cromer concludes that ‗difference‘ poor readers

25

do not adequately organize reading material mentally; they process it word by word and hence lose meaning despite absence of other difficulties. Yuill and Oakhill (1988a) investigated comprehension of anaphoric textual relations in 7 to 8 year olds to examine their ability to construct cohesive text representations. They had children follow along, on a text with marked anaphors, as an experimenter read a brief story. Afterwards children were asked to resolve the anaphors. Poor comprehenders showed significant difficulties on this task, with the exception of elliptic relations with a nearby antecedent (as in the sentence sequence, ―I hope you have brought your lunch!‖—―Of course I have‖, where the antecedent brought my lunch, found in the previous sentence, is omitted from the response). Despite being allowed to use the text, they performed so poorly that, in the words of the authors, ―it is not at all surprising that they have comprehension problems‖ (p. 183). Their answers suggested that they used the nearest syntactically appropriate item when resolving anaphors, or misapplied world knowledge to the text. Informal probing suggested that they seemed not to attend to text meanings and proffered idiosyncratic meanings instead. In their later work, Yuill and Oakhill (1991) report several experiments reiterating the difficulties that poor comprehenders have with anaphoric devices. First, they presented children with sentences containing two antecedents for target pronouns, varying ambiguity of pronouns as well as memory load, and asked forced-choice questions about the antecedents. As an example, the sentence Peter lent his coat to Martin because he was cold has two possible antecedents for he (Peter and Martin). The likelier antecedent is Martin, but the pronoun itself does not allow for differentiation unless sentence meaning is taken into account. By contrast, in Peter lent his coat to Sue because she was cold, the pronoun

26

unequivocally refers to the female proper noun. In addition to varying pronoun ambiguity, they also allowed children either to view the entire sentence while it was read to them, or to view one clause only, varying memory load. Reading was followed up with a forced-choice question such as ―Who was cold, Peter or Sue?‖ In all conditions, poor comprehenders were significantly less able to resolve pronouns than the good comprehending controls. In a follow-up study, the authors used a similar format but gave children a cloze task to fill in the appropriate pronoun, and asked them true-or-false questions about the sentence proposition. They also varied sentence complexity. Results showed the same pattern of lower performance among poor comprehenders, except that there was a significant difference across sentence complexity within this group. Taken together with findings from Yuill and Oakhill (1988a), these results suggest that this population has marked difficulty understanding pronouns, and that difficulties become more pronounced with complexity level. Their ability to mentally establish text cohesion is thus hampered. In line with these results, Tarlow (1990) found that her poor comprehending participants were less able to generate main ideas: Instead of constructing a generalized meaning from the entire text, they generalized smaller units while disregarding overall meaning, or chose topic sentences that were salient because of the way they were written but that were not important to the text as a whole. Yuill and Oakhill (1991, pp. 112-124) presented children with four propositions related to stories presented orally, one of which represented the main idea. Poor comprehenders were significantly less likely to choose the main idea when asked to do so than their good comprehending counterparts. The finding that poor comprehenders have difficulties integrating larger pieces of text into meaningful wholes was reiterated in Bilheimer (1992) and Long and Chong (2001). In

27

the former study, structures of seventh-grade poor and good comprehenders‘ text recalls were compared for narrative and expository text. Recall propositions were classified as being on a micro-, macro-, or superstructural level of text organization; both groups were equally proficient at recalling the narrative text on all levels, but the poor comprehenders were less able to recall expository information beyond the micro-level. Long and Chong found that poor comprehenders did not detect textual inconsistencies if the inconsistent items were separated in the text. While they did activate the information needed to detect the inconsistency, they did not establish a connection to what they were presently reading. Cain, Patson, and Andrews (2005) investigated children‘s ability to select the correct conjunction from a range of choices when presented with a cloze/blank task. Poor comprehenders had significantly greater difficulty with this task than control children, even though they did not differ from them on a test of grammatical knowledge. The authors characterize their performance as ―a general delay rather than deviance‖ (p. 890) and conclude that both age and comprehension ability are related to task performance, but do not provide an explanation of how they are related, or why poor comprehenders‘ performance might differ between a grammatical knowledge and a conjunction cloze task. Yuill and Oakhill (1991, pp. 153-176) report a number of studies that investigate poor comprehenders‘ abilities to deal with cohesion in narrative texts. In a retell of simple stories, aided by picture stimuli, they were able to reproduce cohesive devices given in the original story. In contrast to good comprehenders, however, they did not add any such devices, resulting in list-like stories as opposed to the livelier ones produced by the comparison group. When telling a story from picture stimuli, poor comprehenders used either a lot of cohesive devices or very few; they also made greater use of improper pronominal reference, and they

28

did not benefit from complete (as opposed to sequential) presentation of stimuli whereas good comprehenders did benefit in this condition. In a final study, the authors presented children with stories in three conditions: normal; with jumbled sentences; and with jumbled sentences revised to have appropriate referential continuity. Only good comprehenders‘ recall benefited from revised sentences; poor comprehenders‘ recall performance did not differ between the jumbled and the revised conditions. Cain (1996) elicited narratives from three groups of children, two of which (good and poor comprehenders) were of the same age, and of a group of good comprehenders matched to the poor group by comprehension age. Presenting participants with a topic line in one condition, and a set of pictures depicting an event sequence in another, she found that in the topic line condition, both good comprehender groups were more proficient than the poor comprehenders in producing narratives with a well-integrated structure of related events. The latter group benefited most of all groups from the picture sequence condition, eliminating the differences to the comprehension-age match group, and closing in on same-age good comprehenders. Cain takes this as an indication that lack of story structuring skills is causally related to poor comprehension, since it did not seem to be a consequence of comprehension age. Similar results were obtained by Cragg and Nation (2006), who elicited written narratives using a picture sequence, and found that their poor comprehending participants produced stories that were less well-structured, used sentences of lower complexity, and captured fewer main ideas than their counterparts. Carretti, Re, and Arfè (2012) reiterated these results, adding the finding that good and poor comprehenders performed similarly on a task involving description of pictures with no requirement of producing a narrative structure.

29

In sum, it is well established that poor comprehenders fail to integrate textual information to meaningful wholes and to mentally establish coherent connections between parts of texts, for example, as expressed by anaphors or conjunctions. They experience particular trouble interpreting anaphors with a distal antecedent, and do not heed text meaning when trying to resolve anaphoric relations. Similarly, they do not pay attention to textual inconsistencies between distal items despite adequate background knowledge. They also have trouble with the active use of conjunctions, and a reduced ability to produce coherent narratives. As a consequence, they are less apt than good comprehenders at recalling information beyond the micro-level, and when they generate main ideas, they do not take them from the whole text but from smaller units, choosing topic sentences based on surface criteria but not meaning. Expectably, they benefit from chunking of complex information into units of meaning; conversely, they have more problems with complex than with simple syntax. Cromer (1970) attributes his findings to habitual word-by-word processing of text; other authors of pertinent studies have not offered an explanation for coherence problems. Use of contextual information and background knowledge. As indicated in Cromer (1970) and Long and Chong (2001), prior knowledge may play a role in poor reading comprehension; not only its presence or absence but also whether or not it is put to use. A variety of studies have investigated this idea as well as poor comprehenders‘ use of contextual information. Kendall and Hood (1979) compared two groups of fifth graders, one with good comprehension but poor decoding and one group of poor comprehenders, for errors while reading out loud. The former group made more errors overall, but the poor comprehenders had a higher proportion of contextually inappropriate errors than the poor

30

decoder group, suggesting they made less proficient use of contextual information, and also that they may not be proficient at using ―meaningful relationships […] within sentences‖ (p. 47), which is reminiscent of the problems with cohesion seen above. Garner and Reis (1981) investigated ‗lookback‘ behaviors, (i.e. whether readers would use the strategy of looking back at previously read text when presented with an interspersed question). Results indicated that good comprehenders did so far more frequently than poor comprehenders; the authors also found a marked difference between older and younger good comprehenders, indicating that failure to look back to text to get information may be common in earlier typical reading development. In an important and elaborate study, Nation and Snowling (1998) investigated elementary school-age children‘s ability to benefit from context, and correlated it to a variety of measures of reading and language proficiency. To determine the extent to which children used context to recognize words, they compared word reading latency and accuracy for single words and for sentences. Results showed that words in context were read more accurately, and that good readers benefited more from context than less-skilled readers both in accuracy and rate. Of the manifold skills the authors had measured beforehand, listening comprehension specifically was positively correlated with context usage. The authors then followed up with a study that used the same paradigm to investigate context use in good readers as compared to children with dyslexia and to poor comprehenders. Results showed that all children read words better in context than in isolation. Importantly, when compared for improvement over baseline, context facilitation was most pronounced in children with dyslexia, and poor comprehenders benefited least from the availability of context. The authors suggest that ―incomplete information from a partial decoding attempt interacts with

31

contextual information and, in combination, these two sources of information may result in the correct pronunciation of the target word‖ (pp. 1007-1008). Hence dyslexics (who would have been least able to rely on decoding only) showed strongest context benefits, whereas poor comprehenders (who would have been proficient in decoding but least proficient in linguistic processing) showed least context effects. Of note, the authors‘ explanation bears strong resemblance to Smith‘s (2004) information-theoretic account. However, in contrast to Smith, they view decoding as more important than higher-level skills, and draw no further connection between poor comprehenders‘ difficulties and their finding that context use is best predicted by (receptive) linguistic proficiency. Similar findings are reported by Cain, Oakhill, and Lemmon (2005), who studied 9year old children‘s ability to explain the meaning of idiomatic expressions. Poor comprehenders were as proficient as good readers to explain transparent idioms presented in context, but had significantly more problems using context to explain opaque idioms. Van der Schoot, Vasbinder, Horsley, Reijntjes, and van Lieshout (2009) presented 10 to 12 year olds with ambiguous sentences in disambiguating context and used eye tracking and reading time measurements to determine how children used context for disambiguation. Poor comprehenders were as proficient as good readers to use context when it preceded the trouble spot, but not when disambiguating information followed an ambiguous sentence. This suggests that poor comprehenders did not monitor their comprehension as effectively as the good reading group, a finding made in a variety of studies, as will be seen below. In sum, it seems unequivocal that poor comprehenders‘ troubles are characterized by inadequate use of linguistic context when interacting with print. They produce more contextually inappropriate miscues than good comprehenders, make less use of context to

32

answer comprehension questions, benefit less when being presented with words in context, and are less apt at using context to resolve opaque idioms and disambiguate information. While some earlier studies attribute this to the type of instruction and the consequent habits of word-by-word processing, later authors do not attempt to explain this behavior. Inferencing. As Yuill and Oakhill (1991, pp. 65-80) point out, creating an integrated representation of a text requires inferences: to connect text portions, and to ‗fill in‘ any informational gaps (or slots) where the text doesn‘t provide specific information. A number of early studies on inferencing skills are reported in Oakhill and Garnham (1988, pp. 136139). The poor comprehending youngsters were as able as their better comprehending peers to answer literal questions about a text, but less apt at answering inferential ones, even when the text was made available to them. Similar problems were found on tasks requiring verbatim versus gist recall of sentences. The authors conclude that inferencing problems are a genuine ability issue for this group, and not due to differences in knowledge base. This also shows in their lesser proficiency at instantiation, a type of inference in which context is used to give greater specificity to a word meaning (e.g. people instantiated as Eskimo when embedded in a sentence such as The people build their houses from ice). Yuill and Oakhill report an additional experiment that substantiates earlier findings. In a sentence recall task for which children were presented with verbatim original sentences as well as sentences representing valid or invalid inferences, poor comprehenders were less likely to choose the valid inference than good comprehenders, but more likely to choose either a verbatim original or an invalid inference. This suggests, as per the authors, that good comprehenders construct integrated, meaningful wholes to represent text, whereas poor comprehenders rely on verbatim recall and have problems constructing integrated, coherent representations.

33

In subsequent studies, the link between inferencing difficulties and poor comprehension was corroborated. Gladstone (1991) compared seventh-grade poor comprehenders to two groups of good comprehenders attending fifth and seventh grade. She had them read two types of passages that differed in length and information density. All passage meanings were ambiguous to invite inferencing. Participants were asked to stop reading at pre-specified points in the text and answer inferencing questions. On the longer passage, both the poor comprehenders‘ and the fifth graders‘ answers were insufficiently comprehensive or erroneous. They produced ―implausible, internally inconsistent and poorly elaborated interpretations‖ (p. iii), which Gladstone attributes to insufficient ability to abstract and synthesize ideas (p. 127). By contrast, responses on the shorter passages were as appropriate as the older groups‘, suggesting that both poor comprehenders and younger readers benefit from the absence of irrelevant information, and from dense informational units that do not require the reader to detect and connect distal important ideas. Cain and Oakhill (1998) investigated young readers‘ aptitude at answering questions that either tapped information given in a text, or required cohesive as well as gap-filling inferences (the latter add information not given in but deductible from the text). They compared good and poor comprehenders to a group of younger children matched to the poor group by comprehension level, and found that poor comprehenders performed worse than the matched group on both types of inferencing questions, but similar on questions tapping literal information. The differences between groups diminished when children were able to look at the text, but poor comprehenders still fared worse on questions requiring gap-filling inferences. This was not due to lack of knowledge or ability to make such inferences; when children were asked directly to provide these inferences, all were able to do so.

34

Cain, Oakhill, Barnes, and Bryant (2001) found that poor comprehenders were less skilled at making both coherence and elaborative inferences. That is, inferences necessary for text comprehension versus inferences that enrich representation but are not crucial for understanding. Poor comprehenders also proved less competent at learning and recalling information, but the group differences in inferencing persisted even after differential text knowledge was controlled. Analyzing types of inference, the authors conclude that the main reason for inferencing failure in poor comprehenders was that this group selected the wrong information from their knowledge base. As the stories in question were read to the children, rather than by them, the authors also conclude that this problem is not limited to reading alone. Bowyer-Crane and Snowling (2005) compared good and poor comprehenders‘ performance on two different reading assessments, the Neale Analysis of Reading Ability— Revised (Neale, 1989) and the Wechsler Objective Reading Dimensions (Wechsler, 1990). The NARA II requires drawing of inferences based on real-world knowledge; poor comprehenders performed significantly worse than the comparison group on this test. By contrast, the WORD requires mostly recall of literal information; on this assessment, poor comprehenders scored within normal range. The authors suggest that the inferencing difficulties of this group are not caused by a lack of world knowledge but by a lack of awareness of the need to use this knowledge to generate necessary inferences. Geva and Massey-Garrison (2012) note that inferencing difficulties hold even across languages; both monolingual and ELL poor comprehenders display this particular characteristic. In sum, poor comprehenders‘ problems with inferencing are well documented. They are less able than good comprehenders to answer any kind of inferential questions, even

35

though they show the same abilities to answer literal questions; less apt at instantiating more specific word meanings from context use; more likely to recall verbatim information or to make invalid inferences; and less proficient at answering inferencing questions from texts when required to differentiate important and unimportant information. These difficulties persist regardless of the knowledge base available to them; selection of unrelated information from knowledge base and a lack of awareness of the need to use the knowledge base seem to be implied as well. Most authors appear to agree that these problems are not due to any fundamental deficit or inability; when poor comprehenders‘ knowledge base is controlled for and they are explicitly prompted to make inferences, differences when compared with good comprehenders disappear. Cain and Oakhill‘s (1998) study was specifically designed to address issues of causality; their poor comprehending participants proved less apt at inferencing tasks than comprehension-matched younger children, which led the authors to conclude that inferencing problems are likely a cause of, not caused by comprehension problems. Metacognition and psychological factors in text processing. In line with the importance placed on metacognitive competence in the last few decades, a rather large body of research exists in this area. Most of it is predicated on the concept of comprehension monitoring and/or error detection. Other studies have addressed psychological factors such as motivation or beliefs in one‘s own competency. In addition, many studies investigate metacognition and text processing simultaneously or draw conclusions regarding both aspects of the reading process. Hence, all three areas are discussed jointly in this section. The initial spark for studies on this topic seems to have come from Miller and Isakson (1976), who compared good and ‗difference‘ readers (in Cromer‘s [1970] sense) for

36

sensitivity to violations of sentence syntax and semantics. Presenting their participants with triads of one well-formed sentence, one with a semantic malformation and one with both a semantic and a syntactic deviation, they found that good comprehenders made significantly more reading errors on the former two types than the poor comprehending group. The authors take this to confirm their hypothesis that good comprehenders would be more sensitive to such violations. Garner (1981) modified short text passages to include either inconsistent information, or infrequent polysyllabic words that did not alter passage meaning but were expected to be unknown to her 5th to 6th grade poor comprehending participants. She then had children read the passages and rate them for comprehensibility. As expected, unaltered and inconsistent versions were not rated differently, but versions containing difficult words were rated as significantly harder to understand. When interviewed, participants who indicated having problems with the unaltered or the inconsistent passages were unable to provide details on their problems. By contrast, when asked about problems with the ‗difficult words‘ versions, all children gave specific answers such as ―long words‖ or ―tough words‖. Garner concludes that her findings corroborate the ‗piecemeal processing‘ explanation for poor comprehension, i.e. the idea that poor comprehenders process print in a word-by-word fashion, inattentive to meaning. She also suspects a lack of expectancy of meaningfulness to play a role. As mentioned above, Garner and Reis (1981) investigated ‗lookback‘ behavior displayed by good and poor comprehending students when presented with interspersed questions. In addition, they used qualitative observational data to find out to what extent their participants were monitoring their own comprehension as evidenced by displays of ―subject dissatisfaction‖ (p. 578) in the case of incorrect responses. Such displays were taken to

37

indicate awareness of comprehension problems. Poor comprehenders showed the least pertinent behaviors; by contrast, both younger and older good comprehenders displayed behaviors that suggested they were aware of comprehension failures, even though only the older group was able to resolve those through ‗lookbacks‘. Hahn and Smith (1983) had sixthgraders tutor younger students in answering comprehension questions; they found that the poor comprehenders amongst their tutors rarely used strategic behaviors such as looking back and skimming through the text, rephrasing of questions, and differentiating between different types of questions. Garner, Wagoner, and Smith (1983) used a very similar design to investigate use of text ‗lookbacks‘ and text sampling, and found that poorly comprehending tutors fared worse than their good comprehending peers on all of these. Hare and Borchardt (1985) modified the error detection design described above by asking their 7th and 8th grade participants to ‗act as editors‘ for expository text passages, some of which had been altered to incorporate words inconsistent with passage meaning. Students were asked to read each passage and rate it as ‗easy‘, ‗okay‘, or ‗difficult‘ to understand, and were then informed that the altered passages ―contained words which did not fit the sense of the larger passage‖ (p. 239) and asked to locate these. In contrast to earlier findings, good and poor comprehenders did not differ in their ratings regarding difficulty level, suggesting that good comprehenders in this study did not attend to textual inconsistencies the same way as in similar studies. In the specific information condition, however, both groups differed significantly from each other, with the former group (good comprehenders) locating more than 90% of the errors while the latter was only able to locate 15%. Also in this condition, when asked about causes for difficulties, poor comprehenders tended to point out ‗hard‘

38

words that were perfectly consistent with passage meaning. The authors view these results as corroboration that poor comprehension is characterized by word-by-word processing of text. Fleisher (1990) undertook a study of spontaneous self-corrections while reading narrative texts. Texts were organized so that they increased in difficulty; oral reading errors were subjected to a miscue-type analysis similar to Goodman‘s framework (e.g. Goodman & Goodman, 1994). That is, according to whether the observed response bore graphophonemic and/or semantic similarity to the expected response or not. Across difficulty levels, poor comprehenders corrected errors most frequently that were graphically similar to the text, and they corrected errors least frequently that were graphically dissimilar. Corrections were made regardless of whether errors disrupted text meaning or not. As Fleisher notes, this may suggest that poor comprehenders ―are more focused on graphic (…) than on semantic dissonance; their primary reading goal may be to pronounce the words rather than to obtain meaning‖ (p. 112). She also suggests that emphasis on decoding instruction in classrooms may be causally implied in this finding. Another strand of research in this area investigated children‘s understanding of what they are doing when reading, an approach broadly based on schema theory and similar ‗reader-as-builder‘ concepts. Canney and Winograd (1979) used a questionnaire to tap children‘s schema of reading; they also presented them with texts that were altered in several ways (semantic, syntactic, graphic) and had them comment on their readability. As they do not report procedures of participant recruitment, it is not clear whether their groups represent the profiles presupposed for this study; their results are still illustrative, which is why they are discussed at length here. In the interview, better comprehenders gave more information about how they understood reading than poorer comprehenders. Also, both younger children

39

and poorer comprehenders of all ages referred more often to mechanical skills, whereas older better comprehenders placed emphasis on meaning-related processes. Specifically, a majority of good comprehenders referred to thinking about what is read, and talked about connected texts as carriers of meaning; both schemata were only spuriously present among older, and not at all among younger poor comprehenders. In addition, older good readers were aware of subtle aspects of reading such as relaxing to improve concentration or personal involvement. Poor comprehenders were aware that they experienced problems but saw the solution in increased attention to lower-level mechanics and in thinking harder about reading (not about what is read). As the authors put it, ―lower comprehenders did not appear cognizant of the need for top-down processing behavior‖ (p. 38). Reacting to the altered passages, a majority of better comprehenders rejected semantically or syntactically changed texts as unreadable whereas most poor comprehenders stated they could read them. All children found graphically altered texts unreadable. The authors conclude that poor comprehenders‘ difficulties are due to inappropriate schemata for reading acquired in instruction: many poor comprehenders are passive readers, responding mechanically to graphic stimuli, just as they have been taught to do in grades 1, 2, and 3. No accommodation (…) of their schemata for reading occurs upon entrance to grade 4 because no such drastic change is perceived necessary. (p. 8) Wade (1990) used a Think-Out-Loud approach to devise a descriptive taxonomy of poor readers. She did not use standardized assessments for participant recruitment, so her results need to be viewed with caution, but two of her categories appear illustrative of the population of interest in this study. According to her results, the ‗non-risk taker‘ processes

40

text in a passive, bottom-up fashion, and fails to think beyond the information given; comprehension questions are answered by repeating verbatim and occasionally incoherent information. Wade attributes this group‘s difficulties at least partly to excessive attention to print, and to reluctance to take the risk of being wrong when thinking beyond the text surface. Similarly, the ‗non-integrator‘ fails to develop coherent understanding of a text beyond the immediate portion under scrutiny, possibly because this reader does not understand cohesive relations between ideas. Yuill and Oakhill (1991, pp. 124-141) examined comprehension monitoring and found that only good comprehenders slowed down when encountering textual inconsistencies while reading out loud. When asked to resolve such inconsistencies, both good and poor comprehenders were able to do so, but the latter groups‘ performance was impacted when processing load was increased by varying distance and position of resolving information. Kletzien (1991) had 10th to 11th grade good and poor comprehenders read three expository passages tailored to match independent, instructional, and frustration reading levels of each group. The texts were presented with blanks where content words had been taken out. Students were asked to complete the cloze and to report on the strategies used to do so. Cloze accuracy was also computed. There was no difference in accuracy between groups but the use of strategies differed; as the material got harder, poor comprehenders used a smaller variety of strategies, and used strategies less frequently than good comprehenders. In Kletzien‘s words, ―the difference between the groups was in regulation, rather than knowledge, of comprehension strategies‖ (p. 79). She discusses personal factors, such as affective reactions to the reading situation or belief in one‘s own ability to influence success while reading, as possible reasons for this discrepancy between groups.

41

De Sousa and Oakhill (1996) is one of the few studies that investigate broader psychological factors related to poor comprehenders‘ difficulties. The authors assessed children‘s abilities to identify problems embedded in brief reading passages, using two formats (game-like vs. not game-like), of which the game-like one was reported by the children to be more interesting to them. Good comprehenders‘ performance was not affected by format, whereas the poor comprehender group showed significant improvements in the high-interest condition. The authors suggest that deficits in comprehension monitoring may be linked to lack of motivation and interest in completing a given task. van der Schoot, Vasbinder, Horsley, and van Lieshout (2008) had good and poor comprehending children read two texts on science and social topics, assigning them the role of a journalist required to write about what they read. The rationale was to induce purpose and hence active engagement with the text, specifically with words important to the topic. The authors tracked participants‘ eye movement in relation to important words as well as to anaphors. Results showed that the good comprehenders spent significantly more time fixating important than unimportant words than their poor comprehending counterparts, but less time fixating anaphoric items. The authors conclude that while good comprehenders actively engage to meet a given purpose, poor comprehenders process text in a more passive way. They are also less proficient at constructing coherent mental representations from a text, as indicated by their increased time needs while processing anaphors. Geva and MasseyGarrison (2012) investigated oral language abilities in monolingual and ELL fifth-graders classified as good readers, poor decoders, or poor comprehenders. Regardless of language status, the latter group displayed general language weaknesses in higher-level language, plus difficulties in inferencing and listening comprehension.

42

Yuill (1996), building on the premise that appreciation of jokes and riddles requires metalinguistic competence, studied correlations between children‘s reading comprehension and accuracy on the one hand, and recall of joking riddles on the other. She found that riddle recall did not significantly correlate with reading accuracy, but did correlate significantly with reading comprehension for two of her four types of riddles. The two types are termed ‗word compound‘ and ‗metalinguistic‘ riddles, respectively; the former type relies on the possibility of assigning multiple meanings to lexical compounds (e.g. Would you like to buy a pocket calculator?—No thanks, I know how many pockets I have), whereas the latter involves metalinguistic thinking in the technical sense of the term (e.g. Which word is always loud, even when said softly?—Loud). She concludes that poor comprehenders have difficulties with at least some of the metalinguistic skills required for riddle appreciation and retention. In a follow-up study, Yuill added three more types of riddles: morphophonological (What kind of keys are furry?—Monkeys), syntactic (What animal can jump higher than a house?—All animals, houses can‘t jump), and absurd (How do you fit 6 elephants into a Mini?—Three in the front and three in the back). Children were presented with two answers to choose from, a joking one and a non-jocular, realistic one, and asked to decide which one ‗worked better‘. As in the previous study, children‘s performance on word compound and metalinguistic riddles correlated with comprehension; unlike the previous study, single-word based and pragmatic riddles were correlated with comprehension in the same way, likely due to the differences in experimental tasks. Interestingly, performance on morphophonological riddles correlated positively with reading accuracy, unlike any other riddle type. In sum, the data suggest that poor comprehenders do not make adequate use of metacognitive abilities when interacting with print. As a result, they are less sensitive to

43

semantic or syntactic malformations than their good comprehending counterparts, show less behavior indicative of awareness of comprehension failures, and are more likely to spontaneously correct reading miscues based on visual rather than on semantic information. They are also more prone to experience breakdowns in comprehension monitoring based on text surface criteria such as distance of needed information. When asked about causes for their troubles, they are more likely to refer to ‗difficult‘ words rather than to issues of meaning. Acting as tutors for younger students they use less strategic behaviors than good comprehenders. They also decrease their use of reading strategies the more difficult a text gets, and fail to appropriately allocate attention, treating important and unimportant information indiscriminately. Finally, they are less likely to accurately recall or complete joking riddles that require metacognitive skills. On the upside, their comprehension monitoring can be improved by engaging them in activities that are of interest to them. As for explanations for these findings, there seems to be some agreement that poor comprehenders differ from good comprehenders in the way they process print. Earlier authors describe their behavior as ‗piecemeal‘, word-by-word processing, with undue attention to graphic surface and too little attention to meaning, possibly because poor comprehenders do not expect texts to be meaningful. Amongst later studies, van der Schoot et al. (2008) point out that poor comprehenders appear to be passive rather than active in their engagement, and link their metacognitive troubles to failure in establishing coherent mental representations. Working memory. The role of working memory in poor comprehension is controversial; hence, pertinent studies will be described in greater detail than done in previous sections. Research on the issue began with Yuill, Oakhill, and Parkin (1989), who

44

had their participants perform two experimental tasks gauging their ability to retain shortterm information. In the first task, they asked children to read out lists of groups of threedigit numbers, and to recount the last digit for each group. Poor comprehenders fared significantly worse on this task than their good comprehending peers. The authors suggested that this result may be indicative of a general deficit in working memory not restricted to the linguistic domain. In the second experiment, they had their participants resolve semantic anomalies in brief stories, with resolving information placed either directly next to or two sentences away from the anomaly. When asked to resolve the anomaly, poor comprehenders performed as well as their counterparts on proximal information but significantly below them with distal information, a result reminiscent of the textual coherence issues reported above. Yuill and Oakhill (1991, pp. 96-112) presented children with a list of spoken words as well as a list of pictures, and asked the children to recall the words. No difference was found between good and poor comprehenders. The authors concluded that poor comprehenders do not appear to have general memory problems, as they are quite proficient in verbatim recall, but did have issues specific to working memory (i.e. holding information in mind while processing other information). Yuill and Oakhill (1991, pp. 96-112) devised two further experiments based on the assumption that drawing inferences while processing text reflects working memory capacity. In the first, children silently read sentences and then answered comprehension questions that required pronoun resolution. Good comprehenders performed slightly better than poor comprehenders on this task, even though the difference was not significant. Of interest, good comprehenders took somewhat longer than their counterparts to read the sentences and answer the questions, and even though only sentence reading time exhibited a significant

45

difference between the groups, there was a significant positive correlation of reading time and correct answers for the good comprehenders, whereas there was no such correlation for poor comprehending children. This suggests, the authors concluded, that the former (good comprehenders) put effort in resolving anaphors while reading the sentences, whereas the latter group (poor comprehenders) did not. To rule out interference of verbatim recall, the authors developed a similar task with longer texts. Children read brief stories that presented antecedent conditions; consequences needed to be inferred. Memory load was varied by inserting filler sentences. Results were ambiguous; while poor comprehenders did perform worse than their counterparts, reading times and accuracy were not affected by memory load in this group, but these variables were affected for good comprehenders. The authors suggest that this may reflect a higher sensitivity, in the good comprehenders, to textual cues that an inference is required, a result consistent with research reported above. They also suggest that working memory difficulties may be described in terms of inefficient use of strategies while reading. Despite the inconsistency of these results, much effort was subsequently devoted to research in this area. De Beni, Palladino, Pazzaglia, and Cornoldi (1998) presented collegeaged good and poor comprehenders with a verbal and numerical recall test, with final items of sentences and number lists to be recalled. Participants also had to answer true/false questions for the sentences. The authors found no differences in accuracy in numerical lists but did find such differences in sentences, where poor comprehenders recalled fewer words than their counterparts, and incorrectly recalled more words that were present but not the targeted last words in sentences (intrusion errors). These results were taken to indicate that working memory is hampered by an inability to inhibit irrelevant information – as poor

46

comprehenders had to first process the meaning of the sentence to answer the true/false questions, their ability to recall the final word may have been impacted. To test this hypothesis, a second experiment was devised. Participants were presented with strings of unrelated words, asked to tap when they heard an animal word, and to recall the last word of each string. Poor comprehenders recalled fewer words than good comprehenders, and fared worse as strings got longer. They also had more intrusions, particularly of animal nouns. To rule out the possibility that these results were simply due to task demands, another group of participants was asked to recall every word in the string. Groups did not differ significantly, suggesting that dual task demands were not the reason for poor comprehenders‘ previous performance. The authors concluded that this population suffers from a ―deficient inhibitory mechanism‖ (p. 319), which would explain their comprehension difficulties. Of course, it can be argued that the real cause of poor comprehenders‘ issues is not a deficiency in some presumed mechanism but the simple fact that wherever results were significant, tasks were linguistic rather than purely memory-based. Given the variety of problems with linguistic processing that this population encounters (see below), it may simply have been the nature of the tasks that overburdened their working memories. This argument is proffered by Nation, Adams, Bowyer-Crane, and Snowling (1999), who presented their participants with similar recall tasks as above but varied stimuli along linguistic parameters. Poor comprehenders proved no different than their peers to phonological variation but showed less proficiency on semantic tasks. Also, they performed within normal range on a non-verbal, spatial task requiring working memory resources. The authors conclude that working memory issues in this population are specific to the verbal domain. De Beni and Palladino (2000) responded by investigating recall of irrelevant

47

information in reading, and found that poor comprehenders recalled more such information than their counterparts. They took this to indicate inhibition difficulties; it needs to be noted, however, that these results can be just as well explained by incoherent or unorganized mental representation. Palladino, Cornoldi, De Beni, and Pazzaglia (2001) added a study on list recall of exclusively verbal items, and found the same patterns of higher intrusion errors in poor comprehenders, leading them to the same conclusions as previously. Cornoldi, Carretti, and De Beni (2001) present a review on pertinent studies and come to a conciliatory conclusion, according to which active engagement in distinguishing relevant from irrelevant information, and in activating inhibitory processes, is implied in the populations‘ problems. Marshall and Nation (2003) investigated children‘s recall of sentences and pointed out that poor comprehenders, unlike typical readers, were prone to errors which altered sentence meanings. Oakhill, Hartt, and Samols (2005) used the semantic anomaly paradigm to place their participants under differing working memory demands; poor comprehenders fared worse with higher demands, and also performed worse on a non-linguistic (numerical) working memory test. However, comprehension ability was as good a predictor of performance in anomaly detection as was working memory; the authors discuss the possibility that multiple cognitive abilities may be implied in comprehending. Cain (2006) followed up investigating poor comprehenders‘ recall of both linguistic and non-linguistic items, and found that in line with previous research, the target population did make more intrusion errors than their peers. However, they did not differ in short-term memory as measured by recall. Adding verbal and nonverbal working memory requirements, Cain then explored her participants‘ ability to recall last words of sentences they themselves provided in a cloze procedure. Poor comprehenders fared worse than typical children in both

48

tasks, but more pronouncedly on the verbal task. Few intrusion errors were made, leaving the question of inhibitory differences still open, so Cain devised a complex cloze task where children were presented with a low-probability ending to the sentence after providing a guess, and later with another cloze test aimed at recalling the low-probability items and disconfirming guesses. Poor comprehenders showed higher priming and lower inhibition of disconfirmed items (i.e. they were more prone to recalling items that had become irrelevant). Cain concluded that inhibitory differences are linked to comprehension issues but also cautioned that intrusion errors, the main measure of such differences, were not very frequent in her study, suggesting that other factors may be at work as well, and that inhibitory deficits may simply be a corollary of poor reading comprehension. Pimperton and Nation (2012) correlated measures of working memory with teacher ratings of poor comprehenders‘ problem behaviors in the classroom. As a group, poor comprehenders had working memory issues in the verbal but not in the nonverbal domain, which is in line with previous findings. However, the authors detected a subgroup of poor comprehenders who showed working memory-related problem behaviors in the classroom; this subgroup had trouble in this area across verbal and nonverbal domains, suggesting that a portion of the population may indeed be characterized by working memory difficulties. In sum, the role of working memory in poor comprehension is far from established. Swanson, Howard, and Sáez (2007) review a number of pertinent studies across reading disabilities and come to the conclusion that domain-general working memory deficits are causally implied in all of them. However, they advance their conclusion cautiously, conceding that ―there is a problem in arguing that comprehension deficits are related to a domain-general system‖ (p. 181). For example, there are no intervention studies in working

49

memory; by contrast, intervention that targets comprehension strategies has been shown to be successful with the poor comprehending population (p. 179). This would suggest that working memory overload occurs due to insufficient mental organization of reading material. By contrast, Borrella, Carretti, and Pelegrina (2010) reaffirm previous findings regarding deficient inhibition of no longer relevant information, and conclude that this very specific memory skill is causally implied in poor reading comprehension. But again, it can be argued that failure to inhibit irrelevant information is itself a result of failure to distinguish relevant from irrelevant input. Finally, at least one study (Geva & Massey-Garrison, 2012) has not found any working memory differences between poor comprehenders and good readers. Given these issues, Nation (2005) concludes that ―the relationship between language comprehension and verbal memory in poor comprehenders is an intimate one‖ (p. 258) but also that ―issues of causality are far from clear‖ (p. 258). Studies on linguistic processing. A flurry of studies has investigated various lowerlevel aspects of linguistic functioning in poor comprehenders, mostly under the label of ‗processing‘. The first publication of relevance in this area is Hess‘ (1982) investigation of an array of reading-related skills and processing speeds. In this study, poor comprehenders took longer to read the presented texts; were less apt at recognizing both verbatim and textimplicit information when presented with statements related to their readings; recalled less idea units in a free recall task; and produced more inappropriate thematic intrusions on the same task. However, poor comprehenders did prove apt at recalling important idea units; good comprehenders, in this task, recalled both important and less important idea units, suggesting that when faced with difficulties, poor comprehenders focused on the important aspects of the texts. In a prompted retrieval task in which children were asked to remember

50

instruments from agentive sentences, poor comprehenders took longer for responding, indicating increased needs for processing time. Interestingly, the author also found increased processing time on a task of decoding single common words; later studies did not confirm this finding. In sum, poor comprehenders performed worse or took longer than their good comprehending counterparts on virtually every aspect of the reading process assessed. Yuill and Oakhill (1991, pp.42-52) used card sorting tasks to investigate semantic processing. The children were asked to sort words into categories, and they had to match pictures to words. No differences were found between comprehender groups, suggesting that semantic access was not a problem for the poorer group. This task did not assess automaticity of such access; another experiment was therefore conducted that tapped children‘s abilities to rapidly name pictures and words, both directly and with interference from incongruent picture-word pairings as well as picture-nonword pairings. Results indicated significant differences between groups, but not between conditions for the poor comprehender group, suggesting that poor comprehenders did not experience disruptions of meaning access beyond their general difficulties with linguistic material. The authors point out that poor comprehenders were less apt at picture naming than their peers, which they take as a possible indication of poorer semantic representations. Yuill and Oakhill (1991, pp. 52-64) then investigated syntactic processing, first replicating the study on anomalous sentences by Miller and Isakson (1976) reported above. No significant differences between comprehender groups were found, suggesting that decoding may not have been sufficiently controlled for in the original study. In a more elaborate experiment, the authors tested the interaction between syntactic complexity and sentence processing, presenting children with an array of embedded and non-embedded

51

constructions that varied in clause numbers. No differences between groups were found for verbatim recall of sentences, but poor comprehenders performed worse both in gist recall and in answering comprehension questions than their counterparts. Sentence complexity did not affect these results, suggesting that groups did not differ in syntactic processing. To corroborate these findings, the authors administered a standardized measure of receptive syntactic skills (the Test for Reception of Grammar, Bishop, 1983), where children are read sentences and asked to select a picture that ‗goes with‘ the sentence. Groups did not differ in this task. The authors conclude that poor reading comprehension is not characterized by differences in processing at the word or sentence level. Subsequent research largely confirmed this finding. Stothard and Hulme (1995) assessed poor comprehenders‘ phonological skills and found them age-appropriate. Nation and Snowling (1998) showed that semantic deficiencies can interfere with phonological processing, again confirming that poor comprehenders‘ difficulties lie in meaning-related aspects of linguistic functioning. Nation and Snowling (1999) found poor comprehending children to experience difficulties in abstract, categorical semantic relations. In a task that required matching of words either by function (e.g. broom-floor) or by category (e.g. broombrush), and that were or were not lexically associated, poor comprehenders showed the same priming effects as their counterparts for functionally but not for categorically related words (except when those were also associated). The authors point out that this pattern resembles responses by developmentally younger children. Nation, Marshall, and Snowling (2001) compared picture-naming skills of typical, dyslexic, and poor comprehending youngsters. The latter group did not exhibit phonological problems regardless of length or frequency of target words, but was slower and less accurate in naming than typical controls, particularly as

52

regards low-frequency words, suggesting deficits in semantic processing or representation. Nation, Snowling and Clarke (2007) corroborated this finding in a vocabulary-learning task, in which poor comprehenders fared better on learning phonological than semantic aspects of novel linguistic concepts. The same results held on association/recall tasks (Weekes, Hamilton, Oakhill, & Holliday, 2008) and in learning of written novel words from context (Ricketts, Bishop, & Nation, 2008). Tong, Deacon, Kirby, Cain, and Parrila (2011) gave morphological derivation tasks to children in 3rd (t1) and again in 5th grade (t2); poor comprehenders performed worse than average and good comprehenders at t2 but not t1, suggesting that slower acquisition of morphological awareness plays a role in their problems. In sum, it is well established that poor comprehenders experience difficulties at lower-level aspects of linguistic processing wherever meaning and/or metalinguistic thinking are involved, but not at the level of isolated word meanings, or in purely syntactic tasks. There is also some evidence that some but not all of them have smaller vocabularies than typical young readers (Nation, 2009). None of the authors in this area venture to specify the role such skills may play in poor comprehenders‘ overall problem constellation. As Nation (2005) makes clear, none of the findings reported here lend themselves to causal conclusions. Language impairment and cognitive abilities. As poor comprehenders exhibit problems on virtually all levels of linguistic performance when meaning is involved, and in the spoken as well as the written mode, it is plausible to suspect that the population may be characterized by language impairment. A few studies have looked into this possibility. Nation, Clarke, Marshall, and Durand (2004) administered a battery of language assessments to poor comprehenders, tapping abilities from the phonological to the pragmatic level. As expected, the target group performed lower than typical controls on all measures except the

53

phonological ones. The authors then compared their scores to standardized measures of language impairment. Depending on the index used, between 5 and 19 poor comprehending children out of a sample of 25 were qualified as language impaired. The authors proffer a thoughtful discussion, pointing out that while poor comprehenders exhibit rather pervasive oral language problems, they do not show a clear profile of difficulties: ―the poor comprehender population is likely to be heterogeneous‖ (p. 210). One obvious difference when compared to the Specific Language Impairment (SLI) population is the relative strength of the poor comprehenders on the phonological plane; specific language-impaired children‘s proficiency in this area is below normal. Another difference concerns real-world difficulties implied in both profiles. One criterion for SLI is its impact on social or academic functioning; poor comprehenders, by contrast, often go unnoticed in classrooms. Finally, there are qualitative differences between the groups‘ performance. For example, poor comprehenders tend to overgeneralize regular past tense forms while SLI children fail to mark tense in the first place. The authors conclude that there is an overlap between the two profiles, but no general concomitance. In line with these findings, Kelso, Fletcher, and Lee (2007) identified 15 children out of a sample of 52 children with SLI as poor comprehenders. Adlof (2010) compared children‘s performance on morphosyntactic measures to established SLI norms; her poor comprehending participants showed difficulties with obligatory finiteness, irregular past tense, and subject-verb agreement, the first two of which are also known to be problematic for SLI children. Taken together, results indicate that while poor reading comprehension is concomitant with general language problems, it is not simply a consequence of language impairment. The two conditions may or may not co-occur; in either case, as Nation (2005)

54

points out, the fact that oral language problems are present in these children does not negate the fact that they have problems specific to reading. Also, these problems are stable over time. In a longitudinal study by Nation, Cocksey, Taylor, and Bishop (2010), poor comprehenders maintained their specific profiles from ages 5 through 8. Another factor to be considered is cognitive ability. As seen above, poor comprehending children do not perform differently from typical peers on a variety of nonverbal tasks. Nation, Clarke, and Snowling (2002) studied general cognitive abilities in the population using a broad-based assessment, the British Ability Scales II (Elliott, Smith, & McCulloch, 1996). The target group scored lower than controls on most measures, with highest discrepancies in the verbal domain. Nonverbal differences were significant but less pronounced; spatial cognition did not differ. General cognitive ability was within normal range for most of both groups, even though poor comprehenders fell on the lower end of that range, and a few scored below average. Two points of interest need to be noted. First, there was considerable variation between individual children in both groups as regards cognitive abilities; however, in the poor comprehender group reading comprehension was equally poor across cognitive skills. The authors argue that this may suggest different underlying causes for their comprehension issues, yet it would be more plausible to infer that reading comprehension is not dependent on general cognitive skills but requires specialized strategies that children in this group fail to use, regardless of their cognitive functioning. Second, while arithmetic skills were typical in poor comprehenders, quantitative reasoning (i.e. the detection and application of regular relations between numbers) was below normal. The authors do not discuss this finding but it could be argued that there is a parallel between the differential demands of decoding versus comprehending in reading, and number tasks vs.

55

reasoning task in mathematics. Both decoding and number tasks require comparatively lower-level skills whereas comprehending skills such as inferencing or establishment of textual coherence would seem to be more akin to mathematical rule detection and application. In this light, poor comprehenders‘ mathematical profiles would be expected to turn out as they did. In sum, there is evidence that language impairment and cognitive problems can be concomitant with poor reading comprehension, but also that they cannot be equated, nor that the former are causally implied in the latter. There are poor comprehenders who are not language impaired and who fall within normal range of cognitive functioning. Also, some reading theorists caution that written and oral language are so different in their characteristics that different skills are needed to process them (e.g. Smith, 2004). For example, written texts require mental organization of massive amounts of information in a way normal oral language does not. As Coots and Snow (1980) formulate, the existence of poor comprehenders ―provides considerable justification for a (…) view in which oral language comprehension ability plus decoding ability are regarded as not sufficient for the attainment of reading comprehension‖ (p. 21). Intervention studies. A large body of data has substantiated that poor comprehenders benefit from intervention (e.g., Keene & Zimmerman, 2007; Paris, Cross & Lipson, 1984;Vance, 1990). Rickards and Hatcher (1977-1978) had children read passages with interspersed questions: ‗meaningful learning‘ questions that required cognitive organization of the material, and ‗rote learning‘ questions that required literal recall. In a third condition, no questions were asked. Good comprehenders performed better than their poor comprehending counterparts on most of a variety of measures of text recall, and their

56

performance did not differ between conditions. By contrast, poor comprehenders performed significantly better in the ‗meaningful learning‘ condition than in the other two; also in this condition, they caught up to the comparison group on one of the measures. The authors conclude that poor comprehenders ―do not spontaneously organize paragraph details‖ (p. 551) whereas good comprehenders ―spontaneously generate a context while reading‖ (p. 552). Subsequent studies confirmed these findings. Gagné (1981) conducted a complex project involving poor comprehending middle school students. She found that ‗elaborative processing‘ of prose material (i.e. the addition of thoughts), improves retrievability of the information in the texts. Direct teaching of pertinent strategies proved to be of particular benefit, as did selection of information on the basis of readers‘ prior knowledge. Brown and Palincsar (1982) involved a small group of poor comprehenders in two different conditions of intensive, long-term comprehension instruction. In one, students answered comprehension questions on texts, received corrective feedback, and were then interactively taught comprehension strategies (main idea generation, grouping of parts of information, prediction of content and related questions, and resolution of confusion). In the other, the order of strategy training and corrective feedback was reversed. Students‘ abilities improved considerably in the first condition, but not much in the second. The authors do not elaborate on this finding but it can be speculated that their participants may not have been aware of their comprehension problems, and that it took corrective feedback to trigger such awareness and make them receptive to strategy training. Garner (1982) followed up on her studies of lookback behavior, teaching poor comprehenders to use looking back at the text to answer comprehension questions. It took

57

only one unit of explicit instruction to induce such behavior, and even though poor comprehenders‘ performance did not reach the same level as good comprehenders‘ performance who had received the same instruction, it did reach good comprehenders‘ preinstruction levels. Results were maintained in a follow-up session a week later. Palincsar and Brown (1984) taught poor comprehenders to summarize, question, clarify, and predict while reading, both in reciprocal, dialogic teaching and in a classroom-style format. Poor comprehenders benefited in both conditions but significantly more so in the reciprocal format, and did not only maintain this improvement over time but generalized it to classroom and standardized tests of comprehension. These findings were replicated by Lysynchuk, Pressley, and Vye (1989). Idol-Maestas (1985) had children use a guided comprehension probe involving a set of six rather specific instructions for interaction with print, to answer comprehension questions from a text. Performance improved significantly over baseline, but only older children were able to maintain this level. In a similar design, poor comprehenders benefited from strategy instruction implemented in a classroom-like setting and with a heterogeneous group of readers (Ritter & Idol-Maestas, 1986). Idol (1987) used a story mapping intervention in a likewise heterogeneous group; poor comprehenders benefited from the intervention, as did control children. As previously discussed, text organization and inferencing were found to be problem areas in poor reading comprehension. Yuill and Joscelyne (1988) as well as Yuill and Oakhill (1988b) addressed these in their studies. In the former, children were provided with titles and pictures that integrated story information, or they were trained to look for clues to infer consequences. Comprehension improved in both conditions. The latter study compared the effects of inferencing training to those of rapid decoding and standard comprehension

58

exercises (comprehension questions and corrective feedback including group discussion amongst children). Comprehension improved in all three conditions, but strongest gains were made in the comprehension exercise and notably the inference training conditions. The authors conclude that both kinds of training ―promoted comprehension processes normally lacking‖ (p. 43) in the population, and attributed the unexpected success of the comprehension exercises to children‘s spontaneous discussion activities. Both conditions seemed to provide children with greater awareness of their own comprehension. Oakhill and Patel (1991) trained their participants in imagery, using pictures that illustrated sequence and important details of stories read. Children were instructed to imagine the pictures in their mind. Over the course of three sessions, poor comprehenders‘ ability to answer comprehension questions about the texts improved significantly. Later studies substantiated that poor comprehenders appear to benefit from virtually any kind of study aimed at promoting active engagement with print and/or use of higher-level strategies. Johnson-Glenberg (2005) used interactive Internet tools that prompted students to generate questions and devise visual models from the texts they were reading. Her participants made gains in comprehension and engaged more frequently in rereading of text; the poorest comprehenders benefited the most from the intervention. Yuill (2007) improved children‘s performance on standardized comprehension tests by means of an interactive group intervention aimed at metalinguistic skills, using software that engaged children in discussing ambiguities joking riddles. The same results were achieved using a peer-to-peer discussion format (Yuill, 2009). In both studies, children also showed significant increases in metalinguistic comments. van der Schoot, Horsley, and van Lieshout (2010) instructed their participants to read a text either with the goal of understanding it, or to imagine the events

59

depicted as vividly as they could. Participants‘ comprehension benefited from the ‗imagination‘ condition; they also changed their patterns in allocation of attention as measured through eye tracking, devoting more time to aspects of text related to the situation model than in the ‗understanding‘ condition. In sum, it is unequivocal that poor comprehenders‘ difficulties are not irrevocable. Given appropriate instruction, they show rapid and clear improvement in their ability to comprehend print. Successful interventions include interspersed questions that require cognitive organization of texts; explicit instruction in a large variety of reading strategies, taught one or several at a time; induction of lookback behavior; imagery training; and interactive formats using peer-to-peer or learner-computer interaction. All of these have in common that they promote interaction with print at levels higher than visual information and surface text structure. These findings would seem to provide important insights into the etiology of poor reading comprehension, as will be discussed in the next section. Explanations for Poor Comprehension As mentioned above, the phenomenon of poor reading comprehension in the absence of decoding problems continues to puzzle researchers, and no agreed-upon cause or explanation has been established (Nation, 2005). This is somewhat surprising, given that the phenomenon has been so intensively studied, and a vast array of contributing factors has been determined. Poor comprehenders have been found to have difficulties in virtually all areas of linguistic processing of text except for the phonological/graphophonemic plane, and some surface aspects of syntax. Moreover, a variety of interventions have proven successful with the population, suggesting that we have a good grasp on what to do to help poor comprehenders. If this is the case, why do we not adequately understand their problem?

60

Reviewing the explanations that have been proffered over four decades of research, it becomes clear that they reflect different theoretical views of comprehension that underlie the respective studies, implicitly or explicitly. Some authors place more emphasis on the reader as an active constructor of meaning, whereas others conceptualize comprehension in terms of cognitive components that interact regardless of readers‘ intentions. Research into the phenomenon began from the former view, starting with Cromer‘s (1970) ‗difference‘ perspective, in which poor reading comprehension is characterized as a habitual way of interacting with print that differs from such interaction as would result in comprehension. Hence, this and related explanations focus on what poor comprehenders do when they read. Examples include Cromer himself, as well as Golinkoff (1975), who attributes poor comprehenders‘ difficulties to a habit of inflexible word-by-word processing of text. Similarly, Garner (1981) speaks of ‗piecemeal processing‘ of print, and suspects a lack of expectancy that texts can be meaningful. Hare and Borchardt (1985) point out that poor comprehenders‘ attention is mainly focused on surface aspects of text such as ‗hard‘ words. Canney and Winograd (1979) as well as Fleisher (1990) emphasize that the population views reading mostly as the mechanical process of pronouncing words. Wade (1990) as well as van der Schoot et al. (2008) describe poor comprehenders as passive processors who pay excessive attention to superficial aspects of print, a view in consonance with Yuill and Oakhill (1988a, 1991) who describe that their participants rarely take the effort to process spatially distal information. Use of the ‗active reader‘ view is concentrated in the earlier literature – at least up to Yuill and Oakhill‘s (1991) landmark tome. Their rather cursory treatment of motivation and attitude, mentioned above, is illustrative of the change in perspective that seems to have

61

occurred with their publication. This is not to say that the notion of the active reader is completely absent from later works. More emphasis, however, is placed on the absence of competencies and behaviors (a ‗deficit‘ perspective in Cromer‘s [1970] terms), and on impersonal skills and presumed mental entities. Borne out of the Simple View of reading, a latent ‗multi-component‘ model of comprehension has informed much of the last two decades‘ work either implicitly as a guiding principle for research, or explicitly at the core of explanatory accounts. As a result, much research has focused on pinpointing specific deficits as causes of poor comprehension. Unfortunately, this enterprise seems to have failed. Among cautious reviewers, a consensus exists that no single underlying causal factor has been established. Nation (2005) concludes that the population is characterized by heterogeneous problems in all areas of language (except phonology and graphophonemics). Heterogeneity is also the main finding in Cornoldi, de Beni, and Pazzaglia (1996), who compared individual profiles of poor comprehenders and found that they shared ―few common traits [and] no unique pattern of difficulties‖ (p. 132). Likewise, Cain and Oakhill (2006) in their study of individual profiles, ―did not identify any concurrent skill deficits that were consistently associated with poor comprehension‖ (p. 694) and conclude that ―a single underlying source of poor comprehension is unlikely‖ (p. 683). In the conclusion to Cain and Oakhill (2007), they echo Nation, citing inferencing, integration of text, metacognition, and story knowledge as candidate competencies (but excluding word and sentence level skills), and formulate that while ―a large number of skills are correlates of poor comprehension (…) our understanding of the skills that are causally implicated in comprehension is still developing‖ (p. 69).

62

In sum, current leaders in the field suggest that despite a wealth of existing data, we have not yet understood poor reading comprehension. By contrast, earlier authors made confident assertions about their understanding of the issue. Whether or not these assertions were premature, it can be suspected that the difference in conclusions stems not from insufficient data but from differences in theoretical perspectives on the reading process. General Discussion and Rationale for the Present Study In this section, the findings of the literature review will be summarized and an outline of the purpose of the present study, concluding with the research questions, will be presented. As this study presents an alternative to prior approaches, a discussion of the theoretical rationale for this novel way of inquiry has also been included. General discussion. As seen above, poor comprehenders have been shown to have difficulties with virtually all aspects of reading, except graphic and syntactic surface structure of written language. Problem areas include the integration of textual information into coherent mental models; use of context and background knowledge; inferencing; metacognition and comprehension monitoring; and various aspects of linguistic competence. Working memory issues seem to play a role as well, but may simply be concomitants of the problem. Some poor comprehenders have language and/or learning impairments but others do not; importantly, teaching of reading strategies has proven infallibly successful with the population, suggesting that intrinsic impairments are not the prime cause of their troubles. In the previous section, the puzzling divergence in explanations for the issue between earlier and later accounts has been discussed. To reiterate, in publications where the focus is on readers as active, intentional constructors of meaning, the poor comprehender profile appears quite clear; their difficulties are invariably characterized by failure to use higher-

63

level, meaning-making strategies when reading. This view is particularly prevalent in the earlier literature. Much of the later research, by contrast, has operated from componential notions of reading. Informed by the Simple View, it has concentrated on determining absent or deficient linguistic and cognitive skills and the causal effects these are presumed to exert on comprehension. As an example, Nation, Snowling, and Clarke (2007) present their results, stating, ―the source of poor comprehenders‘ difficulties with lexical learning may be localized to the semantic rather than phonological component of vocabulary learning‖ (p. 131). In formulations such as this, the reading process is treated as an aggregate of separate if interacting parts that operate smoothly in good readers but are somehow (and somewhat mysteriously) absent or malfunctioning in poor comprehenders. The failure to attribute causal status to any particular skill has led to a paucity of explanatory accounts in recent research. However, this paucity can hardly be attributed to insufficient data. As demonstrated in this review, there is a wealth of convergent findings. Rather, the problem may be that the notion of causal relations between isolable components of comprehension is not conducive to integrating these findings into a coherent theoretical account. Two distinct yet interrelated alternatives present themselves here. The first is the concept of the reader as active ‗builder‘ and ‗fixer‘, as intention-guided constructor of meaning; the second, a view of comprehension as a quale, an unobservable mental experience that can be inferred from behaviors but not equated to reified ‗cognitive components‘. Two examples from the literature may illustrate how these concepts could inform understanding of existing data. As previously discussed, Yuill and Oakhill (1991) report

64

several successful interventions targeting a variety of higher-level strategies. However, they appear somewhat puzzled by their own results: We still have only a limited understanding of exactly how the training we used affects children. We cannot assume that they adopt wholesale the strategies we had intended them to learn, but that through doing the tasks we set them, they evolved some processes and strategies of their own which later proved effective. (pp. 212-213) Had the authors conceptualized readers as active learners, they might have been able to account for the possibility that their participants may indeed have ‗evolved some processes and strategies of their own‘. Clearly, once poor comprehenders are shown better ways of engaging with text, they use them—sometimes to the surprise of the interventionists. Nation (2005) discusses the status of inferencing and comprehension monitoring as causal factors. While she finds the evidence for their general importance convincing, she notes that the causal weight attributed to them by some authors (see above) may be based on circular reasoning: ―children (…) were selected as poor comprehenders precisely because they performed poorly on a standardized test of discourse comprehension containing a high proportion of inference-based questions‖ (p. 260). According to some authors, she adds, higher-level difficulties, e.g. with inferencing, ―are not examples of structural or specific deficits that cause a comprehension problem. Rather, they are the comprehension problem‖ (p. 261) and are based on lower-level deficits. It is not surprising that the insistence on describing comprehension as a mechanistic interplay of a finite set of components should yield such contradictory accounts. In the history of pertinent scholarship, diverse skill lists of varying length and scope have been

65

devised, as reported above. The differences between detailed accounts such as Gray‘s and the limited skill set investigated in the last twenty years suggest that not only does there seem to be a certain degree of arbitrariness in such lists, but also that our understanding of what readers do may not necessarily have improved. Notably, there seems to be confusion between comprehending as an observable process, and comprehension as its result, a distinction made in transactional models of reading (Goodman & Goodman, 1994; Goodman, Watson, & Burke, 2005). If comprehension is viewed as a quale, confusions between process and the result evaporate, and inferences, as in Nation‘s (2005) example, can be seen as a behavioral indication that comprehending is happening. Likewise, the absence of inference-making would simply indicate that a particular part of the comprehension process is absent; there is no need to assume lower-level difficulties. The fact that recent research is characterized by a relative absence of the active reader and a focus on reified components of cognition cannot only be attributed to underlying concepts of reading but is also connected with experimental methodology. Damico and Ball (2010) offer a thorough critique of experimentalism in the human sciences. As they make clear, the observation that a regular conjunction of events occurs does not mean they are causally related; causation can be inferred but not proven. This is especially true when there is no theoretical explanation for it. Experimentalist research, with its background in the natural sciences, is geared towards determining ―whether one variable causes another rather than how it did so‖ (p. 9), and hence it has difficulty providing such explanations: Unlike phenomena in the natural sciences, human behavior and the resultant social actions are guided by the meanings negotiated by the human participants (…) Beliefs,

66

intentions, values, and interpretations held by the actors in the social realm greatly influence their observed behaviors and practices. (p. 7) As various ‗reader-as-builder‘ theories argue, beliefs, intentions, and values are crucial to understand how a reader engages with text. However, these aspects of the reading process have rarely been investigated in poor comprehenders, among other reasons because they cannot adequately be studied outside of the social contexts in which they are acquired and negotiated, and which are difficulty to access in tightly controlled experiments. Consequently, Damico and Ball (2010) call for qualitative studies to understand complex, situated phenomena in human communication. It is highly informative, at this point, to mention two relatively recent studies that did use a more contextualized approach. In 1994, Cain (mentioned in Oakhill and Yuill, 1996) investigated home reading habits in families of good and poor comprehenders. She found that the latter had less exposure to print than their peers: They were read to significantly less frequently, and had also less access to libraries. Cain and Oakhill (2011) correlated measures of reading habits, comprehension, and vocabulary growth in 8- to 16-year old poor comprehenders over the course of several years, and compared these to good comprehenders. While no evidence for ‗Matthew effects‘ was found—i.e. the gap between the groups did not widen over time—all three measures were positively correlated at every point in time, suggesting that they were interrelated. These findings suggest that environmental and experiential factors do play a role in poor reading comprehension. However, this insight is not reflected in other studies, or integrated into explanatory accounts.

67

In sum, while much is known about isolated behaviors in poor comprehenders, they have been studied less extensively as readers. This is true for contextualized descriptions of the reading process as a whole, as well as for factors such as attitudes, motivation, prior experience, and the like. Only few authors have attempted a theoretical account of the phenomenon; as seen, recent scholarship in particular has concluded that despite the vast amounts of data gathered, the phenomenon is poorly understood. However, this seems to be a result of theoretical and methodological orientation rather than of a dearth of knowledge; poor comprehension may simply be too complex and too elusive to be conceptualized in terms of causal relations of isolated components. For all these reasons, this study was specifically devised to provide a methodological alternative to most existing research by exploring the scientific potential of generating contextualized, qualitative descriptions of poor reading comprehension (as opposed to decontextualized, experimental ones). This approach is novel in four ways. First, it takes seriously the possibility that comprehension may not be reducible to component parts. Second, it takes into account human factors that have not been addressed in much recent research, such as intentionality, motivation, and lived experience. Third, it includes context, both historical and current, as a relevant factor in the development and functioning of poor comprehenders. Fourth, its goal is therefore to understand the phenomenon, rather than explain it as a result of linear-causal relationships. I will discuss these four points in turn to provide a rationale for my research question. Intentionality and contextual complexity: The qualitative alternative. We have seen that there is a distinct possibility that the unsatisfactory state of affairs reported above may be the outcome of a simplistic notion of reading comprehension. Reducing it to a

68

mechanistic interplay between component skills, or to observable behaviors, amounts to reifying an ineffable mental state. When purpose and intentionality are left out of the description of human behavior, and qualia like comprehension reified, confusion ensues. There are issues and findings within the experimental literature itself that substantiate the potential of an alternative approach. These include methodological reductionism, failure to account for qualitative and experiential factors, and failure to take context into account. Each will be discussed in turn to provide a rationale for a qualitative approach. Reduction of complexity as a problem of parsimony. A common criticism of experimental research is that methodological rigor comes at the price of reducing multifaceted phenomena to simple factors, with the peril of oversimplification (Damico & Ball, 2010; Smith, 1998). A precise account of this problem comes from Cronbach (1975), himself a towering figure in experimental psychometrics. Aware of the limitations of the paradigm he helped to advance, Cronbach formulates: The canon of parsimony, misinterpreted, has led us into the habit of accepting Type II errors at every turn, for the sake of holding Type I errors in check. There are more things in heaven and earth than are dreamt of in our hypotheses, and our observations should be open to them (…) From Occam to Lloyd Morgan, the canon has referred to parsimony in theorizing, not in observing. (p. 124) In other words, Cronbach warns social scientists against assuming that hypotheses about causal relations will infallibly cover all there is to know about a given subject matter; an assumption based, he argues, on a misconstrual of the principle of theoretical simplicity. Parsimony in theory construction does not mean researchers have to limit their observations

69

to single, controllable entities, especially since such control precludes observation of interactions with other factors that may occur outside of the experimental situation. There is reason to suggest that pertinent research has not only misconstrued the principle of parsimony but actually violated it. As Baker (2010) explains, one major way of reading Ockham‘s ‗razor‘ and related accounts is ‗ontological parsimony‘, or ―the number of kinds of entities postulated by [a] theory‖. We have seen that multiple ‗kinds of entities‘ have been postulated as component factors of reading comprehension: mental models (discussed under ‗coherence‘), reading strategies (e.g. use of context or inferencing), metacognition, linguistic skills, and working memory. Comprehending as a process, or comprehension as a mental state, are assumedly constituted by such autonomous if interrelated entities (even though no theoretical consensus on their interplay exists). The conceptual problems of this view have been discussed above. Now it becomes clear that there is also an epistemological problem: Viewing comprehension as constituted by various autonomous component skills actually requires more ontological assumptions about the autonomous existence of said components than viewing comprehension as a quale, of which any observable behaviors or inferred mental entities are but aspects. Hence, on the level of theorizing a holistic view of reading is, in fact, more parsimonious than a causal-componential view. ‗Aspects of a unitary process‘, even such distal ones as prior experience or general motivation, do not constitute a proliferation of ‗kinds of entities‘, whereas ‗components skills of reading‘ do. Qualitative and experiential factors overlooked. Given the state of affairs, alternative approaches could be justified simply by referring to the lack of explanatory accounts. However, three major arguments why a qualitative approach may be warranted can be drawn from the existing literature itself.

70

First, consider that qualitative factors in reading—purpose or motivation, experience, schemata for reading, or the notion of the active reader—have been amply used for explanation in the earlier literature. Among others, Canney and Winograd (1979), Golinkoff (1975), Garner (1981), Hare and Borchardt (1985), and Fleisher (1990) emphasize that poor comprehenders appear to process print passively, without expectancy of meaningfulness, and possibly employing a schema of reading-as-decoding. By contrast, later studies have only sporadically proffered such accounts. As mentioned, the shift seems to have occurred in the landmark publications by Oakhill and colleagues. Oakhill and Garnham (1988, p. 118-119) note that poor comprehenders may fail to grasp that to read is to construct meaning from a text, and hence to adjust their strategies. Yuill and Oakhill (1991, p. 10-11) briefly discuss motivation, claiming that it could not have played a role in their findings as their designs controlled for it. They do suggest that attitude or understanding of reading may have been a factor: ―many of the poor comprehenders seemed to revel in their ability, often precocious, to decode long and complicated words‖ (p. 10). They go on to musing on the reasons for this: It may seem surprising […] that anybody could enjoy reading something they could not understand, but for children who have just mastered some of the regular spellingto-sound correspondences, it can be an interesting and challenging task, which also gains ready praise from adults. (p. 11) It would seem that the authors‘ dismissal of motivation as a factor in poor comprehension was premature: if children revel in decoding because it gets them praise from adults, motivation does seem to play a role in how they understand reading, and what they focus on while engaging with print. However, this line of thought has not been pursued in

71

subsequent research. Of note, many findings, for example, poor comprehenders‘ failure to establish coherent mental models of text, use inferences, or monitor their own comprehension, could be fit within an explanatory complex of motivational, attitudinal, and schema-related factors. If this population does not view comprehension as the purpose of reading, possibly because they have not been taught to do so, then the finding that they fail to use comprehending strategies would not be surprising. Second, whenever later studies did adopt a broader view on the problem, results suggested that contextual and qualitative factors do warrant attention. Cain (1994, quoted in Oakhill and Yuill, 1996) found that poor comprehenders had less home exposure to print than their peers. Cain and Oakhill (2011) established that measures of reading habits, comprehension, and vocabulary growth were positively correlated over the course of several years, suggesting that they were interrelated. De Sousa and Oakhill (1996) reported that their poor comprehenders‘ performance in comprehension monitoring improved significantly when tasks were more interesting to them. In brief, every time experience and motivation were investigated, they proved important. Third, recall that basically any kind of intervention aimed at improving higher-level reading strategies proved successful with the population. Whether they were taught to mentally organize text to answer interspersed questions, look back to previous text portions to garner relevant information, or use imagery while reading, poor comprehenders inevitably benefited from remedial instruction. This would appear to suggest that a) part of their difficulties consisted in not employing such strategies as were taught, and that b) it cannot be excluded that the reason they did not employ such strategies was that they had not been

72

taught to do so. Experience, once more, shows up as a potentially crucial factor (Berliner & Biddle, 1995; Smith, 2006). Contextualization as a rationale for qualitative approaches. In addition to these philosophical and empirical rationales, there is the major issue of descriptive adequacy or, in experimentalist terms, external validity (Howe, 2004). Referring to Geertz‘s (1973) concept of ‗thick‘ description, Cronbach (1975) points out that accounting for controlled factors alone is insufficient in studying human behavior. Rather, the role of the social scientist is to describe, as carefully as possible, anything that may have a potential influence on the observed effect. Such factors include uncontrolled conditions, (…) personal characteristics, and (…) events that occurred during (…) measurement. As [the social scientist] goes from situation to situation, his first task is to describe and interpret the effect anew in each locale, perhaps taking into account factors unique to that locale of series of events. (pp. 124-125) In other words, instead of assuming that a phenomenon of interest is caused by one or few factors that can be isolated from the environment in which it occurs, the goal of the social scientist must be to understand individual instances of the phenomenon in terms of the unique factors that give rise to these instances. McDermott and Roth (1978), coming from a qualitative angle, make a similar point: instances of behavior, usually thought of as constrained either by participant-internal (psychological) factors, or factors pertaining to the social macro-structure they occur in, may better be conceptualized as emerging from a concrete interactional context. Of course, if these instances are to be classified as variations on a particular phenomenon, then the factors that give rise to them cannot be completely

73

different across instances. However, ―generalization comes late, and the exception is taken as seriously as the rule‖ (Cronbach, 1975, p. 125). What Cronbach advocates for here is a contextualized understanding of phenomena as they occur in a concrete environment, as opposed to a quest for context-free appraisal. In literacy-related research, the importance of context factors such as socioeconomic status is well known but all too often ignored (Nelson & Damico, 2006). Berliner (2002) makes the forceful point that given the myriads of potentially important factors, educational research is the ‗hardest science of all‘ (the title of his essay): ―we face particular problems and must deal with local conditions that limit generalizations and theory building‖ (p. 18). He gives the example of a study that found a general negative correlation between test anxiety and test performance—but in some of the 93 sites studied, the negative correlation was very strong, whereas in a few, the correlation was actually positive. Differences in contextual factors, and interactions between these factors, lead to wildly different correlations—a reality that was not captured by the general finding. And he goes on to describe the complex nature of educational realities: Any teaching behavior interacts with a number of student characteristics, including IQ, socioeconomic status, motivation (…), and a host of other factors. (…) Student behavior is interacting with teacher characteristics, such as (…) training (…), conceptions of learning, beliefs about assessment, and even (…) personal happiness (…). Other variables interact with those just mentioned (…). We are not even sure in which directions the influences work, and many surely are reciprocal. (p. 19)

74

In sum, then, it would appear that the shortcomings of the extant literature are threefold. First, a componential notion of reading may not do justice to the ineffable nature of the process. Second, reader intentionality and experience do not seem to have received due attention. Intentionality includes psychological factors such as motivation, interest, and understood purpose of reading in any given reading-related situation. Experience refers to prior exposure, instruction, print-related interaction etc., which can be assumed to shape intentionality. (General schemata for reading can be conceptualized as spanning both areas: They are formed through and become part of a reader‘s experience, and they are instantiated situationally for any given task, shaping the reader‘s intentions in the moment.) Finally, social-experiential context has not been appropriately addressed. We know little about the past and present environment in which poor comprehension develops. Qualitative research, the alternative offered in this study, was devised to address these issues. Its epistemological foundation rests precisely on the distinction between nonintentional, natural events and intentional, human actions; its methods of data collection include contextual factors as a matter of course. Understanding vs. explaining: the qualitative approach to intentionality. Scientific knowledge is frequently equated with the formulation of general laws that explain the causal effect of one variable on another (Bransen, 2001; Damico & Ball, 2010). But there is an alternative way of knowing that does not rely on such relationships. Traditionally termed verstehen (German for understanding), this approach is based on making sense of phenomena in empathetic, human terms, rather than through the formulation of general, causal laws (erklären, that is, explaining, in the original formulation). The justification for this approach is that human actions, unlike natural events, are based on intentionality: ―[they]

75

take place in order to bring certain future states of affairs about, not merely because certain past states of affairs happened‖ (Bransen, p. 6). Howe (1992) terms this type of cause-effect relationship therefore intentionalist causation. This type of causation is unlike linear-causal relations between naturally occurring phenomena, where outcomes are simply contingent on a prior event (Bransen). As an example pertinent to this study, Yuill and Oakhill (1991) seem to have been looking for contingent interactions between interventions and outcomes but were unable to find any. Had they deployed the notion of intentionalist causation, conceptualizing children as active learners with individual differences in schemata, attitude, and experience, their intervention successes could have been accounted for by understanding behavioral change as changes in intentionality while reading. This, in turn, could have been attributed to changes in schemata and attitudes, prompted by changes in children‘s experience due to the intervention they received. In sum, then, to understand a human action is to be able to connect it to mental states of an agent. Verstehen is the empathetic re-description of actions in terms of such mental states (Bransen, 2001). It is this philosophical stance that provides the foundation for the methodological choice of qualitative methods. Qualitative Research as Contextualized Inquiry Just as qualitative inquiry is suited to understand intentionality, it lends itself to studying interactions of factors of interest in a concrete environment (Berliner, 2002, p. 19). In fact, contextualization is part its very definition as offered by some authors: Qualitative research is a situated activity that locates the observer in the world. It consists of a set of interpretive, material practices that make the world visible. (…)

76

Qualitative researchers study things in their natural settings, attempting to make sense of, or interpret, phenomena in terms of the meanings people bring to them. (Denzin & Lincoln, 2005, p. 3) Conceptualized this way, qualitative research is the opposite of the creation of carefully controlled situations, the basis of experimental inquiry: Its gold standard is to observe and participate in ‗life as it is‘ in order to interpret human life-worlds in their own terms. There is no space here to discuss the possibility or impossibility of such a standard (see the discussion of Howe, 1992, below); suffice to mention that not all qualitative methods rely on naturalistic observation. Traditionally, this type of inquiry has also been carried out by way of interviews and collection of artifacts from the context in question (Denzin & Lincoln, 2005); other time-honored tools include analyses of linguistic and behavioral data (Damico & Simmons-Mackie, 2003). All these types of data collection were used in the present study (albeit with a qualification as regards observation, to be discussed below). They have in common a focus on interpretive understanding. A classic example for a contextualized verstehen-type analysis is found in Geertz (1973). In his explanation of the role of interpretation in ethnographic work, the author uses an example by the philosopher Ryle. Suppose somebody is winking to signal secret understanding, Geertz lays out. A strictly descriptive account of this observed behavior—i.e. an account that disregards agent intentionality—would be unable to tell it from an involuntary twitch of the eye. Suppose further that someone else may wink to mock the first person‘s wink. To the non-interpretive, ‗objective‘ observer, this intricate signaling amounts to nothing more than an inexplicable co-occurrence of twitches; if this observer were looking for a linear-causal explanation, it even could be taken to indicate causality (which of course

77

would be refuted with a subsequent observation of a twitch not followed by another). It takes an interpretive observer with a mind attuned to human meanings to tease out the reasons for this sequence of events. To such an observer, the movement of eye muscles can be interpreted as a complex layer of communicative meanings if he or she is sufficiently familiar with the ―universe of symbolic action‖ (p. 24) of which the movement is a part—in fact, it must be if it is to be more than a mere factual recount. This ‗universe‘ is constituted through past interactions, and it is present to participants as the symbolic reality that provides them with a blueprint for the interpretation of actions—in other words, a context. In sum, then, qualitative research, with its basis in verstehen epistemology and its natural incorporation of context, appears to have the potential to elucidate the puzzling phenomenon at hand. Its characteristics will be discussed in more detail below and in Chapter 3; first, however, it needs to be noted that there are issues that arise from the particular challenges of literacy research within a qualitative paradigm. These issues will inform the discussion leading up to the formulating of the research question. Participant meanings vs. expert knowledge. As Bransen (2001) lays out, if we take seriously that all human action has its basis in an intention which makes sense to the agent, then ―we cannot sidestep the attempts of agents to ascribe meaning to their actions. (…) We have to take seriously the commonsensical and folk-psychological interpretations of those involved‖ (p. 14). However, the relation between intention and action, between behavior and ‗commonsensical‘ intepretation is not as straightforward as in the case of winking discussed above. In the latter situation, the observer can ask the agents to explain their actions to corroborate his interpretation, and it can be assumed that they will give him a plausible answer. In studying poor reading comprehension, this is not possible in the same way. First,

78

children may not be aware of their own mental states while reading simply as a function of their cognitive development; for the same reason, they may not be able to adequately talk about them. Second, and more problematically, children with reading issues may describe their actions in terms that do not match the interpretation of an expert observer. As an example, children in Canney and Winograd (1979) reported paying more attention to lowerlevel and mechanical aspects of reading when encountering difficulties. In their understanding, this was an appropriate reaction; in the understanding of the observer, by contrast, this reaction actually got in the way of their reading. This discrepancy is common in studies of human action and interaction. As McDermott and Roth (1978) point out, internal inconsistencies in participant information and behavior, including lying and contradiction, have been observed in studies of various cultures around the world. Bransen (2001) continues to explain that neither the agent‘s nor the observer‘s understanding of an event is a priori privileged, and recommends ―careful sensitive confrontation‖ (p. 15) of divergent views in order to bring about verstehen. In Canney and Winograd (1979), however, the viewpoint of the authors obviously prevailed. The same is true, as will be seen, for an analytical tool central to this study, viz. miscue analysis, in which reading ‗errors‘ are taken to provide insights into readers‘ cognitive strategies. In this method, the viewpoint of the reader is not even consulted; it the expert‘s interpretation, based on careful observation, which provides understanding of behaviors. How, then, can verstehen be brought about? A pragmatic approach to qualitative inquiry. Howe (1992) provides a rationale for going beyond participant meanings for analysis. Discussing the contention between quantitative and qualitative approaches to educational research, he argues that these

79

approaches seem incompatible only when understood according to the most rigid and restricted interpretations of the underlying epistemologies—positivism and interpretivism, respectively—and that these interpretations are neither helpful nor tenable. Positivism, with its mechanistic outlook borrowed from the natural sciences, assumes that phenomena can be observed objectively, without recourse to and independent from theoretical notions researchers might bring to the observation. The criticisms of this position proffered by Quine (1951) and Kuhn (1962) have dealt a fatal blow to this assumption: ―positivism is dead as an epistemological doctrine‖ (Howe, p. 240). In informing research practice, though, it seems alive and well (Damico & Ball, 2010; Howe, 2004). Our own discussion of the literature certainly lends itself to the suspicion that much of the research on poor comprehension is shaped by positivist undercurrents. Take, for example, the implicit assumption that comprehension can be conceptualized as a mechanistic interplay between isolable components, or that motivation is not a factor that warrants scrutiny. The interpretivist stance that underlies qualitative inquiry allows the researcher to go beyond mechanistic outlooks. As seen, in this paradigm actions and behaviors can be interpreted as results of intentions by human actors in a concrete environment. However, Howe (1992) makes clear that a purely interpretivist stance is fraught with problems as well. First of all, it has been recognized that ‗going completely native‘ is as impossible as ‗being completely objective‘. Just as the experimentalist researcher uses theory to study ‗objective‘ facts, the interpretivist researcher cannot shed his or her own cultural and theoretical background while trying to make sense of the ‗other‘s‘ behavior (see also McDermott, 1987, for a sharp critique of qualitative studies in education which fail to move beyond majoritycultural preconceptions). The best he/she can hope to achieve is to find out how participants

80

think about their own actions, what they ―think they are up to‖ (p. 242). Which engenders a second problem: Informants may not always have a complete picture of their own situation; they may even be completely unaware of some systematic aspect of it. We can assume that this would be the case for poor comprehenders: It can hardly be expected that elementaryschool age children are able to formulate, in explicit terms, why they fail to understand what they read (in particular as they do not always seem to grasp the idea that comprehension is part of the definition of reading). Finally, in the realm of human behavior, actions are usually intended but outcomes may not be. For example, early research has suggested that poor comprehension may be a consequence of excessive focus on small-scale aspects of print, which in turn might be a behavior learned in reading instruction. If this were the case, poor comprehension would be an example of an unintended outcome of a succession of intentional actions. Howe (1992) concludes that ―positivism is untenable and interpretivism is incomplete‖ (p. 243) and calls for an approach that would acknowledge both participants‘ perspectives and their blind spots, both intentions for actions and unintended outcomes. He calls this approach ‗critical educational research‘ and lists the following as its characteristics: a) grounding the question in existing theory, and fitting the methodological approach to the problem at hand; b) using methodologically appropriate ways of collecting and analyzing data; c) referring back to existing knowledge in the process; and d) critically engaging with findings drawn from different approaches. This pragmatic stance seems to fit well with the situation in the present study. Previous knowledge exists but the phenomenon in question warrants further investigation, and the methodology needs to be fitted to address the problems of extant studies, notably to include an interpretivist orientation, while not dismissing unintended or

81

unacknowledged dynamics. Hence, Howe‘s suggestion provides an excellent rationale for an eclectic approach in which data is collected in various ways, and interpreted with recourse to the wealth of existing knowledge, making use both of insider perspective and ‗outsider‘ theoretical background. Framing the Question In devising a research question, I was faced with the problem of formulating it in a way that would account for the wealth of extant knowledge, as well as the need for contextualization and inclusion of intentionality. To be true to the tradition of qualitative inquiry, the general research question would have to be kept as open as possible; at the same time, it was deemed necessary to provide a skeletal framework to organize my inquiry. Surveying the entirety of the preceding discussion, I decided that Goodman‘s (1994) transactional-sociopsycholinguistic model of reading provided a good starting point for my inquiry, for two reasons. First, it has the broadest scope of all models of reading discussed above, providing the openness needed; also, this scope is ordered by context, from the micro (linguistic-behavioral) via the meso (psychological) to the macro (social-experiential). Second, it comes with analytical tools to process data of reading behaviors, which will be discussed below. Based on the conceptual areas of the model, then, the overarching question could be formulated as follows: General research question: What are the qualitative characteristics of poor reading comprehension and its context? ‗Qualitative‘, in this formulation, is an adjectival ascription to be read as synonymous with verstehen on the part of the researcher. Thus, another way of phrasing this question would be ‗How can poor reading comprehension be understood in its context?‘ The wording

82

above was necessary to provide a basis to frame inquiry in a more detailed manner. As per the socio-psycholinguistic model, the general question implies three sub-questions that correlate with each type of characteristic suggested by the model: 1) What are the linguistic and behavioral characteristics of poor reading comprehension? 2) What are the psychological characteristics of poor reading comprehension (lived experience, attitudes, and schemata)? 3) What are the contextual (social and developmental) characteristics of poor reading comprehension? The first sub-question was addressed using the analytical tools provided by the transactional model, viz. Miscue Analysis (Goodman & Goodman, 1994) to address the linguistic aspects of reading, as well as Eye Movement Miscue Analysis (EMMA) (Paulson, 2000; 2002; Paulson & Freeman, 2003), which adds eye tracking data to the linguistic approach. Other reading-related behaviors such as bodily movements or facial expressions were also incorporated in this set of data. The second and third sub-questions were addressed through interviews with poor comprehenders and their parents; where possible, professionals working with them, e.g. teachers or speech-language pathologists, were interviewed as well. Participant observations of reading behaviors, parent-child interactions, and environment of the children were also added to this end, as were analyses of literacy-related artifacts such as writings, homework, grade reports, IEPs, and the like. As implied in the question, the goal of the study was to formulate a (however tentative) empathetic ‗re-description‘ (Bransen, 2001)—a verstehen-based, interpretive account—of poor reading comprehension. To this end, it was placed within the framework of

83

Grounded Theory, a broad-based, qualitative methodology devised to investigate complex phenomena in the human realm with the specific aim of generating middle-range theoretical accounts of them (Bryant & Charmaz, 2007a, b; Charmaz, 2003, 2005, 2006; Glaser & Strauss, 1967; Strauss & Corbin, 1994). Analytical tools and methodology will be discussed in Chapter Three below. It should be noted that due to significant difficulties in participant recruitment, as well as the laborintensive nature of the analyses, only one child was studied. Any findings and conclusions must therefore be viewed as preliminary. More than anything else, this study should be seen as an exploration into the feasibility and usefulness of contextualized, qualitative investigations of poor reading comprehension.

84

Chapter Three: Methods It may well be asked, and until recently the question has scarcely been raised, (…) just what, indeed, do we do, with eye and mind and brain and nerves, when we read? (…) We have surely come to the place where we need to know just what the child normally does when he reads (…). Huey (1968/1908, pp. 8-9) As noted over 100 years ago, Huey‘s complaint that what readers actually do has rarely been investigated is still valid today as regards poor comprehenders. While it was not possible to study the role of ‗brains and nerves‘ in this dissertation, it was hoped that some light would be shed on what poor comprehenders ‗do with eye and mind‘ and how these processes are shaped by experience and social interaction. In this chapter, I will outline the methods and procedures of recruitment, data collection, and analysis that were used in this study. I will begin by outlining my rationale for adopting the methodological framework of Grounded Theory within a pragmatist paradigm. Methodological Framework: Grounded Theory In order to provide a justification for adopting Grounded Theory (GT) as the guiding methodology of this dissertation, I will first refer back to the discussion in the previous chapter to outline which characteristics of GT make it an appropriate choice for this study. I will then give an overview over its history and basic tenets. Grounded theory: An overview. Grounded Theory (GT) was first expounded by Glaser and Strauss (1967) to offer an alternative to the predominant verificationist sociological research of the day. The authors argued that there was an abundance of ‗greatman‘ theories (e.g. the accounts brought forth by Durkheim, Simmel, Marx, or Weber) that

attempted to explain the entirety of the social world, and left less-than-great researchers with the task of verifying hypotheses derived from them. And yet, they asserted, sociology was in dire need of theory – theory derived from (‗grounded in‘) data rather than from abstractive, logical speculation, that could provide satisfactory explanatory accounts of empirical or conceptual phenomena of a particular aspect of social life, rather than mere facts squeezed into a ‗great-man‘ paradigm for explication. Later, this type of theory was dubbed ―middlerange theories‖ (Charmaz, 2006, p. 7). As asserted by Glaser and Strauss (1967), the reasons for this lack of grounded, middle-range theorizing may be found in the history of sociology. Much of sociological fieldwork, prior to the World War II era, was done by meticulously amassing qualitative data. Subsequent theorizing—if any—tended to be related to the data in unsystematic ways, giving qualitative research a reputation for being impressionistic or anecdotal. After World War II, quantitative methods made great strides in the social sciences, resulting in hypothesis-verifying work as mentioned above; researchers were trained to find facts but not to generate explanations from them. Qualitative research, therefore, was relegated to fulfill purely exploratory purposes, to be a stimulant for more rigorous, quantitative pursuits. Glaser and Strauss (1967) set out to restore qualitative research to its rightful place in the social sciences by devising a sufficiently rigorous methodology so as to systematically generate credible theory (but not so rigorous as to be insensitive to incoming data). While they argue that both quantitative and qualitative methods have a place in social research, they focus on the latter for several reasons, of which the most important for the present study is that ―the crucial elements of a (…) theory are often found best with a qualitative method, that is, from data on structural conditions, consequences, deviances, norms, processes, patterns,

86

and systems‖ (p. 18). This resonates well with the need, formulated in the previous chapter, for a contextualized and verstehen-based inquiry of poor comprehension. At the same time, the researcher who wishes to use GT is not limited to using of data collected through fieldwork or interviews; rather, multiple types and sources of data can be mined for use including quantitative data (e.g. from surveys). This flexible approach provides the justification needed for this study. Grounded theory can accommodate naturalistic as well as numerical, elicited data since it does not tie the latter to a causal-statistical analysis, but subjects it to the same process of coding and comparison as qualitative data (Glaser, 2008). Since its inception, GT has become the most widely used qualitative research approach across the human sciences (Bryant & Charmaz, 2007a). It exists now in multiple varieties to fit the divergent needs of researchers (e.g., Bryant & Charmaz, 2007b; Charmaz, 2006; Skeat & Perry, 2008), and has been described as a valuable if somewhat underused tool for research in speech-language pathology (Skeat & Perry, 2008). The main innovation of GT was to introduce a systematic way of doing qualitative analysis that was open to scrutiny by colleagues. In particular, the following procedures are laid out by Bryant and Charmaz (2007a, b), Charmaz (2003, 2005, 2006), Corbin and Strauss (2008), Glaser (2008), Glaser and Strauss (1967), and Strauss and Corbin (1994). Collecting rich data. In principle, anything that is suitable to add explanatory power to the analysis of a phenomenon can be employed as data in GT. This includes classic data sources like field notes and observations, recordings, and interviews, but also textual data from libraries and other sources, and even quantitative data. With regard to many investigations, one important data source is the researcher him or herself. Throughout data collection and analysis, part of his/her task is to generate and refine (or reject) preliminary

87

ideas for interpretation, including such ideas as stem from sudden insights, anecdotal observations, or personal experience, and to note them down in memo form to make them amenable to further analysis. Explicit coding. To counter accusations that qualitative research is inherently ‗impressionistic‘ or ‗unsystematic‘, GT involves intricate and detailed coding procedures. Verbal data are coded line-by-line, and numerical data scrutinized for indices of directional relationships. These initial codes are then thematically chunked together into larger categories that are compared against incoming data. The relationship between these larger categories form the elements of the grounded theory that is the goal of the inquiry. Theoretical sampling. In contrast to both random sampling and purely naturalistic observation (which are often cited as the ‗gold standards‘ in the experimentalist and classical qualitative paradigm, respectively) GT involves the deliberate sampling of data that the researcher deems suitable to answer his questions. After completing a first round of ‗initial data collection‘ and seeing categories emerging from the analysis of that data, subsequent data is sampled specifically in order to probe and extend the fledgling categories. This can, but need not, include data that contradicts them, or comparison cases that differ from those under scrutiny. Charmaz (2006) makes the case that contradictory data should not be deliberately imported into the analysis but that if and when it arises, it suggests that the researcher‘s question or theory needs refinement. Glaser and Strauss (1967) hold that ‗differing‘ cases need not be studied if the goal of the GT inquiry is a ‗substantive‘ theory but that they do warrant attention if the goal is a ‗formal‘ theory (this distinction will be discussed in the next subsection).

88

Constant comparison and revision. As in most qualitative research (e.g. Janesick, 1994), GT requires the researcher to constantly compare existing data to incoming data, to compare incoming data to any preliminary categories that may have emerged, and to continuously revise any theoretical formulations in line with new insights from data and coding. This process continues until the emerging categories are ‗saturated‘, that is, until the point when ―gathering fresh data no longer sparks new theoretical insights, nor reveals new properties of your core theoretical categories‖ (Charmaz, 2006, p. 113). The exact methods of data collection and analytical procedures used in this study will be discussed in a subsequent section. But first, problems regarding the goal of this study and the conventions and epistemological background of GT need to be discussed. Substantive versus formal theory. GT typically results in two methodological outcomes, dubbed ‗substantive‘ and ‗formal‘, depending on whether the goal of the grounded theory is to explain an empirical area, or to formulate a conceptual phenomenon that holds across empirical areas (Glaser, 2007; Glaser & Strauss, 1967). As an example, one of the classical GT works was a substantive theory of death as a non-scheduled status passage (Glaser & Strauss). This theory could have been elaborated into a formal grounded theory by investigating other non-scheduled status passages, such as unintended pregnancy or loss of social status due to scandal (examples mine). However, this was not accomplished for the referenced study, which is illustrative of the fact that formal grounded theory is not often employed (Glaser). The main reasons for this seem to be that researchers either believe they can generate formal theory simply by rendering substantive theory in more abstract language, skipping cross-area comparison (Glaser & Strauss), or that they are confused by the various definitions of formal theory that have been suggested over the years (Glaser). Either way, a

89

decision had to be made prior to the beginning of this study: would the goal be to write a substantive grounded theory, or a formal one? The decision was not initially obvious, as most ‗substantive‘ areas of GT inquiry involve contextualized phenomena in concrete social environments (e.g. ―work, juvenile delinquency, medical education, mental health‖, Glaser & Strauss, 1967, p. 79). This reflects GT‘s origin in sociology, of course. Poor reading comprehension is a more elusive, cognitive phenomenon that this study wished to explore in the light of context but it is a phenomenon that is not bound to social groups the way classical substantive topics are. A topic area such as ‗Poor reading comprehension in rural African Americans‘ would have lent itself more obviously to being labeled ‗substantive‘, whereas the present study design does not situate the phenomenon in such a context. On the other hand, the fact that only one participant could be studied precluded any attempt at devising more than a substantive theory. It was therefore decided that the goal of this study would be the generation of a substantive grounded theory of poor reading comprehension. Choosing an epistemology. An unexpected problem that emerged during the preparation phase of this study was the need to choose an epistemological background. It turned out that GT, while providing an explicit methodology, has not been linked to one specific epistemology. That is, while much contemporary GT is couched either in constructivism (Bryant & Charmaz, 2007b; Charmaz, 2006) or pragmatism (Corbin & Strauss, 2008), the classical account by Glaser and Strauss has been credited with substantial positivist leanings (Bryant & Charmaz). As discussed in Chapter Two, positivism assumes that reality that can be adequately studied by isolating causal relations between entities and capturing these in logical propositions (Damico & Ball, 2010; Denzin & Lincoln, 2005;

90

Maxwell, 2004). By contrast, constructivism holds that all the propositions we can formulate in regard to reality are constructions within the realm of knowledge. Hence, we are in no position to speak of reality ‗as such‘ but only of our constructions as more or less ‗viable‘ (Charmaz, 2006; von Glasersfeld, 1985). Along these lines, shared and communicated experiences of participants are of major concern in constructivist approaches to GT (Charmaz) and to qualitative research in general (Denzin & Lincoln), as they provide valuable insight in the realities constructed by human agents. Positivism, by contrast, is antithetical to an interpretive stance, and was therefore ruled out as a guiding framework for the present study. However, constructivism was not an obvious choice either. As discussed above, while the lived experience of poor comprehenders was a crucial factor of interest for this study, it was also necessary to go beyond the participant perspective and to accommodate for the possibility that processes invisible to participants might be equally important. This seemed to call for an assumption of ―expert knowledge‖ (Bryant & Charmaz, 2007b, p. 40) on part of the researcher, which would have been in line with early positivist leanings in GT but at odds with constructivist takes on it. To complicate matters further, the transactional model that informed theoretical sampling as well as some of the tools used for data collection (notably miscue analysis and EMMA) has previously been couched in a realist epistemology (e.g. Goodman, 2008; Paulson & Goodman, 2008). This outlook eschews both positivist certitude of a reality equal to knowledge, and constructivist multiplicity of viable realities, in favor of a layered reality that is accessible through but not equal to knowledge representation. Scientific realists assume that observations are but the surface layer of a complex, dynamic reality. That is,

91

observations are our experiences of events, which are independent from (and hence not the same) as our observations. Events, in turn, are not random occurrences but instantiations of structures and processes constituting the ‗real‘, the deepest layer of a threefold reality. This stance allows the researcher to go beyond mere observation of contingencies. In this realist mode, explanations for observed patterns include structures that cannot be directly observed but can be inferred from observation. It is usually assumed that realism allows for the inclusion of intentional human agency or experiential contexts into explanations of phenomena of interest (Damico & Ball, 2010; House, 1991). In addition, a realist stance allows for the use of ‗expert knowledge‘; existing interpretive tools and prior knowledge of processes and structures can be used to make sense of behavioral data, and indeed have been in studies based on the transactional model (Goodman, 2008; Paulson & Goodman, 2008). However, realism has not been embraced as a guiding philosophy for GT, and serious concerns have been raised as to whether it can be reconciled to the constructivist epistemology widely used in GT. Denzin and Lincoln (2005) as well as Chakravartty (2013) state unequivocally that it cannot. From a constructivist viewpoint, it is untenable to presume an external reality outside of ―issues of value, ideology, power, desire, sexism, racism, domination, repression, and control‖ (Denzin & Lincoln, p. 13). With its emphasis on sociocultural meanings as determinants of truth and falsehood, constructivism ―stands opposed to the realist contention that theories can be understood as furnishing knowledge of a mind-independent world‖ (Chakravartty). To avoid embroilment in these philosophical conundrums, a pragmatist stance (Hookway, 2010; Peirce, 1878) was adopted for the purposes of the present study. As mentioned, at least one version of GT comes couched in this paradigm. Corbin and Strauss

92

(2008) explicitly orient themselves within pragmatism, espousing a dynamic epistemology in which ―acts of knowing‖ (p. 4) lead to practical consequences. We can infer a reality from the consequences, but pragmatism is not interested in making metaphysical assumptions about this reality, particularly since it is always reality for an epistemic actor: ―what is discovered about ‗reality‘ cannot be divorced from the operative perspective of the knower‖ (p. 4). This sounds constructivist at first glance, but there are two important differences between the two outlooks. The first is methodological: pragmatism is not exclusively concerned with insider perspective, and hence more flexible in choice of methods (see e.g. Feilzer, 2010; Morgan, 2007 for pragmatism in mixed-methods research). The second difference is a slight but crucial shift in epistemology. Constructivism espouses what might be called a ‗negative‘ view of reality: All we can say about it is that it permits or inhibits certain actions within our lived experience. Von Glasersfeld (1985) illustrates this using the example of a blind person walking repeatedly through a forest. All that she knows about the forest is a pattern of obstacles (trees) in her way, and of pathways between those. The difference between the constructivist and the realist stance is obvious: the former equates knowledge of the forest with a constructed mental map of wooden, leafy obstacles, while the latter deems it safe to infer a reality of wooden, leafy objects that exist independently from the mental construction. The pragmatist might counter that it was not actually the forest but the way through it that constituted the object of inquiry. Since the blind person has successfully and repeatedly traversed the forest, her knowledge about the way through it was ‗good enough‘ for all practical purposes. In interaction with reality, it is the results that count, not the relation

93

between knowledge representation and ‗reality‘, or the nature of the latter, as in the skirmishes between constructivism and realism. The same outlook can be applied to the present study. It is particularly valuable in the light of the novelty of the underlying approach, and the relative paucity of data. This study is the first to undertake a description of poor reading comprehension from a contextualized viewpoint. Hence, processes and structures underlying the phenomenon cannot be presumed, and given the time constraints placed on a dissertation as well as the lack of participants, there is no reason to suppose that the results of this study will place us in a position to make ontological assumptions about their existence. We can, however, be guided by a perspective of usefulness. For the pragmatist, this is what counts, not the purported ontological status of his findings. When this status is not taken for granted, philosophical conundrums evaporate into the open horizon of an inquiry whose outcome is undetermined. Thus, in the pragmatist epistemological stance taken here, no a priori ontological status was assigned to objects or methods of inquiry. Instead, a simple question was asked: What happens when we describe poor comprehension from a contextualized, interpretive angle, using both insider perspective and expert knowledge? The verificationist pitfall. After methodology and epistemology were chosen, one last issue remained to be addressed, which related back to the original impetus behind the development of Grounded Theory. Ideally, GT involves an outlook that is as fresh as possible, unencumbered by prior theorizing. The researcher starts out by simply asking, ―What‘s happening here?‖ (Charmaz, 2006, p. 20). Classically, it is even advised to minimize literature review before delving into the research in order to avoid being caught up in preconceived notions (Bryant & Charmaz, 2007a, b; Charmaz, 2003, 2005, 2006; Glaser &

94

Strauss, 1967; Strauss & Corbin, 1994). And yet, in the real world of the conventional dissertation process, ―professors, ethical reviewers, and funding agencies all (…) demand a thorough literature review prior to starting data gathering‖ (Noerager Stern, 2007, p. 123). As well they might, she adds, after all, a researcher, especially a fledgling one, is always a ‗dwarf standing on the shoulder of giants‘ whose work is informed and made possible by the work of those who came before. This raised a threefold problem for the study at hand. First, it was heavily influenced by prior research. Second, its very questions were shaped by a pre-existing paradigm, viz. the socio-psycholinguistic model of reading, and ‗active reader‘ views in general. Third, it was inexorably geared towards a specific kind of finding; it would either yield a useful or a less than useful contextualized account of poor comprehension, but no alternatives to such an account. Was it appropriate to use GT under these circumstances? Or was GT used to cloak an attempt at verifying the ‗great-man theory‘ of Goodman (1994)? To answer this, it is helpful to take a closer look at the ―tensions between data collection strategies and what constitutes ‗forcing‘[,] unresolved in grounded theory‖ (Charmaz, 2006, p. 18). This tension goes back to the original work by Glaser and Strauss (1967). Even though GT was developed with an emphasis on ‗fresh‘ outlooks on data, Glaser and Strauss did mention, albeit in passing, that the researcher ―must have a perspective that will help him see the data‖ (p. 3) and that insights sparked from existing models outside the data can be helpful (p. 6). At a later time, Strauss and Corbin (1994) bemoan that due to overemphasis in Glaser‘s and Strauss‘ original formulations, rigid inductivism has in some cases yielded ―sterile or boring studies‖ (p. 277), and point out that they even deem it appropriate to use GT to elaborate and refine existing (grounded) theories. These authors

95

then go on to invoke the potential of extant theories to guide GT research, either explicitly or implicitly in the form of ‗theoretical sensitization‘ to relevant questions, a view echoed by Charmaz (2006), who describes how symbolic interactionism guides her in approaching areas of inquiry. In sum, the use of preexisting theories and findings is well established in GT, and perfectly appropriate according to both the classical and contemporary accounts. Indeed, Corbin and Strauss (2008, pp. 32-35), elaborating on the concept of ‗sensitivity‘, explain that the researcher must be able to interact with the data, rather than being bewildered by it and that prior insights are indispensable for this. The question is not whether or not background knowledge should be used; it will be anyway. ―Insights (…) do not just occur haphazardly, they happen to prepared minds during interplay with the data‖ (p. 32). In fact, it is preferable to make conscious use of prior knowledge rather than to pretend being a blank slate. As Corbin and Strauss further stated, ―Forcing (…) the researcher‘s ideas on data is more likely to happen when the researcher ignores the relevance of self (…) and thinks that it is only the data talking‖ (p. 33). The authors then go on to outline various uses that can be made of existing literature (pp. 35-39). The following have been found to be relevant for the present study: it does give license that enables comparisons; enhances sensitivity; enables the use of prior knowledge and preexisting theory and findings as additional data; assists in interpretive stimulation via questions and enables comparison against findings. Given the limited data set of this study, these are valuable considerations. Regarding the present study, some of the findings from previous research are unanimous (poor comprehenders have weaknesses in higher-level cognitive processes), some preliminary (there may be social and psychological factors associated with poor

96

comprehension), and some contested (the role of excessive attention to lower-level processes). Comparing methodological approaches taken, it was found that reductionist, experimental research has not yielded explanatory accounts of the phenomenon, while interpretive studies have arrived at preliminary explanations, if not a coherent theory. Thus sensitized to a potential methodological issue, I concluded that according to the theoretical preconceptions that guided this study, this state of affairs is inevitable, as reading is too complex to be captured in simple experimental tasks or causal models. It seemed therefore appropriate to adopt an alternative conceptualization of reading as an integrated, dynamic, contextualized process in which a human agent uses whichever knowledge and experience he/she has at hand to interact with text. This stimulated the questions drawn from Goodman‘s (1994) transactional model. In other words, the transactional-sociopsycholinguistic view informed the present study in the same way symbolic interactionism informs Charmaz‘ (2006) work; it may operate as a guiding framework for data collection and interpretation. In addition, it can be argued that extant findings played the role of the first set of data in the present study. They acted as sensitizing concepts pointing to potentially important phenomena. That is, enabling the incorporation of higher- and lower-level cognitive processes as well as contextual and motivational factors. These were conceptualized, under the guidance of the sociopsycholinguistic-transactional view, as behavioral, experiential, and background aspects of poor comprehension, and data collection was geared towards samples illustrative of these three aspects. From a GT point of view, then, the ‗initial data collection‘ for this study could be regarded as completed by means of the literature review, and the actual data collection, described below in this chapter, was really the second step of the GT inquiry, viz. theoretical

97

sampling based on extant findings. The tools used, and the methods of analysis associated with them, were specifically geared towards elucidating the phenomena that had been identified as relevant. These tools include ethnographic as well as linguistic- behavioral methods; each of these will be discussed in turn. Ethnographic Methods: Modified Participant Observation, Video Analysis, Interviews Since one of the goals of the present study was to obtain as rich as possible a description of the context of poor reading comprehension, the encounter with the participant was done with an ‗ethnographic attitude‘. That is, I immersed myself in the experience and used field notes and memos as data, as well as video recordings of the participant during reading tasks. All these are time-honored methods in ethnography and GT; they will be discussed in turn. Basic ethnography: Modified participant observation and video analysis. Ethnography as an academic discipline has been variously traced to the exploits of Boas, Malinowski, or Radcliffe-Brown, but its roots trace back into classical antiquity accounts of ‗barbaric‘ cultures (Atkinson & Hammersley, 1994). The general concern of ethnography is the description and interpretation of ―complex and contextually sensitive aspects of social life and culture‖ (Damico & Simmons-Mackie, 2003, p. 136). Ethnography has been advocated for, as well as used, as a strategy of inquiry in communication sciences and disorders (Crago, 1990; Damico & Augustine, 1995; Damico & Simmons-Mackie, 2003; Kovarsky & Crago, 1990-1991); its tools, such as participant observation resulting in field notes, and open-ended interviews, are part and parcel of GT (Charmaz, 2003; Corbin & Strauss, 2008). Likewise, video recordings are commonly used both in ethnography and GT (Damico & SimmonsMackie, 2003).

98

Basic ethnographic data was collected in the form of field notes from encounters with the participant, both at home and in UL‘s eye-tracking laboratory, and as video recordings of his reading behaviors in two reading situations (see below). It should be noted, however, that the observational situation was somewhat different from what is typically considered ‗participant observation‘ (cf. Atkinson & Hammersley, 1994; Damico & Simmons-Mackie, 2003; Denzin & Lincoln, 2005; Kovarsky & Crago, 1990-1991). Rather than immersing myself in a given social situation, I created the situations. Thus, my observations cannot be taken as fully representative of my participant‘s life. However, they can still be interpreted since they were involved in authentic and contextual reading activities that require actual reading comprehension. Like other qualitative methodologies, ethnography comes in various philosophical and methodological approaches (Denzin & Lincoln, 2005). For this study, Geertz‘s (1973) concept of ‗thick description‘ was adopted as part of an overall framework of ‗rich‘ description. (‗Rich‘ refers to a varied array of data, whereas ‗thick‘ means the in-depth interpretation of individual data points based on the relevant contextual factors; J. S. Damico, 2014; personal communication). As discussed in Chapter 2, ‗thick‘ description is the interpretive probing of an action to uncover layers of meaning that make up participants‘ symbolic universe (Geertz, 1973). In the same way that successions of winks can be interpreted if the researcher is familiar with the potential meanings associated with winks, the physical environment and the visible behaviors (e.g., bodily movements, facial expressions) of a child placed in an unfamiliar reading situation can be interpreted if the researcher is sufficiently familiar with the ‗universe‘ of childhood communication and literacy socialization in contemporary American culture.

99

Interviews. Informal interviews using (mostly) open-ended questions are a mainstay of ethnographic research (Agar, 1996; Simmons-Mackie & Damico, 2003; Westby, 1990) and are commonly employed for GT studies (Charmaz, 2006; Corbin & Strauss, 2008). Neither Charmaz nor Corbin and Strauss give specific explanations on the interview process, other than that it should resemble a controlled conversation (i.e. a conversation that is centered on the broad topic area of interest to the interviewer). However, for the purposes of this study it was decided that a rudimentary structure was needed, for three reasons: a) the relative inexperience of the interviewer; b) the assumption that interviewees would not necessarily be aware of all the potential factors that might impact reading comprehension (cf. Howe, 1992, as well as the discussion on insider vs. expert knowledge above); and c) the relative specificity of the questions to be answered. The last of these reasons refers back to the notion that the literature review provided me with sensitizing concepts, making data collection more focused (‗theoretical‘ in GT terms) than it would have been if I had started from scratch. Consequently, I knew that I was looking for information regarding the participant‘s attitude towards reading and towards himself as a reader, as well as for his ‗schema‘ for reading. That is, what he thinks he is doing when reading (cf. Chapter Two). Likewise, since I was working from a transactional perspective (Goodman, 1994; Rosenblatt, 1994; Smith, 2004), I knew that in interviews with parents and (as available) teachers, SLPs, or other professionals, I would be looking for information about the experiential background of the participant. It was therefore decided that a structured interview outline would be used in order to garner the information needed. The outline was informed by the guidelines for semi-structured interviews given in Smith and Osborn (2003). Such interviews differ both from the informal approaches

100

classically used in GT and ethnography, and from structured interviews used in quantitatively oriented research (e.g. questionnaires). Structured interviews rely on questions asked in a standardized manner and sequence to allow for quantitative coding of answers. Semistructured interviews, by contrast, ―will be guided by the schedule [of pre-established questions/areas of interest] rather than be dictated by it‖ (Smith & Osborn, p. 56). Rapport with the interviewee, of minimal concern in structured interviews, is as important in semistructured approaches as it is in informal ones. Similarly, the interviewer is free to probe topics of interest emerging during the interaction, or to follow the interviewee‘s lead. The main difference between semi-structured and informal interviews is therefore the use of an outline (or ‗schedule‘, as the authors call it). For the present study, the notion of ‗funneling‘ questions in semi-structured formats proved invaluable. Smith and Osborne (2003) recommend starting with the most general questions and then ‗funneling‘ topics with questions that cast smaller nets. This gives interviewees the freedom to elaborate on a general topic from their viewpoint, and it gives interviewers a pathway into garnering more specific information after the initial overview. Semi-structured interviews thus appeared to be an ideal tool to achieve the purposes of this study, viz. garnering data based both on participant perspective and on expert knowledge. Interview with the poor comprehender. In the light of previous findings and the guiding model for this study, particular importance was placed on the participant‘s ‗schema‘ for reading, his attitude towards reading, and his attitude towards himself as a reader and learner. Thus, these aspects informed interview outlines. The following questions guided the process (‗funneling‘ is indicated by indention).

101

ATTITUDE QUESTIONS—READING: How do you feel about reading? What do you read? (Where? When? How often?) What do you like about reading? What don‘t you like about reading? Tell me about reading at school. What do you read at school? What do/don‘t you like about reading at school? Tell me about reading at home. What do you read at home? What do/don‘t you like about reading at home? ATTITUDE QUESTIONS—SELF-AS-READER/LEARNER: Tell me something about your reading. Are you a good reader? Why do you think so? Do you use strategies when you read? Tell me about school. What do/don‘t you like about school? SCHEMA QUESTIONS: Why do you read? Why do you read in school? Why do you read at home? Do you normally understand what you read? Is it important to understand what we read? Why do people read? Why do teachers/parents read?

GENERAL QUESTION FUNNELING QUESTIONS

Interviews with parents and professionals. No findings are reported in the literature that could inform parent or stakeholder interviews. Therefore, interviews with adult participants had to be more general, and were tied back to children‘s reading behaviors and interview data where possible. The main purpose was to shed light on poor comprehenders‘ print environment and literacy exposure at home or in educational/interventional environments; this purpose informed the outline for interviews with adult participants. PARENT/PROFESSIONAL QUESTIONS: Tell me about reading at home. Tell me about your reading. Tell me about your child‘s reading. Tell me about reading to your children when they were little. Tell me about reading for school. Tell me about the beginning of your child‘s reading problems. Tell me about the development of your child‘s reading problems. Tell me about what you think is the reason your child has reading problems. 102

Again, the ‗schedules‘ were used as an outline, not as a rigid sequence to be followed. As can be seen in Appendix E, the actual interviews were only loosely based on the actual sequence of the questions. However, the topics covered were informed by them in a substantial way. Miscue Analysis/Eye Movement Miscue Analysis Miscue analysis is a time-honored method of analyzing reading strategies and text comprehension that was borne out of naturalistic observation and applied psycholinguistics (Goodman, 1967; Goodman & Goodman, 1994; Goodman, Watson, & Burke, 2005). Eye Movement Miscue Analysis (EMMA) is a relatively recent tool for studying reading behaviors developed by Paulson (2000). It brings together miscue analysis and the equally time-honored study of eye movements in reading (Huey, 1908). In the present study, both methods were used. The reasoning behind this harkens back to the earlier discussion of naturalistic versus elicited data. Miscue analysis can be accomplished in any environment, with relatively simple, unobtrusive equipment (audio recorder and/or video camera), and with natural reading materials (such as books, as in the case of this study). EMMA requires eye-tracking technology and frequently takes place in a laboratory situation (as in this study). Given the potential influences of context on the data collection process, it was decided that it would be necessary to collect both eye tracking data and reading data that approximated, as closely as possible, a natural reading situation. Therefore, both a classical miscue analysis and an EMMA were conducted. Both will be discussed in this section. Miscue analysis in the study of reading. Miscue analysis is a qualitative approach to analyzing how readers interact with text. It rests on a transactional model of reading as outlined above, according to which readers engage in transactions with text that are

103

simultaneously social, psychological, cognitive, and linguistic. Hence, the exact nature of the transaction will be a function of the reader‘s current proficiency as well as of past experiences that make up his/her background information. Miscue analysis was developed by Kenneth Goodman in the late 1960s and has since been applied to a host of reading situations involving different populations of any age (Goodman, 1967, 1994, 2008 a, b; Goodman & Burke, 1968; Goodman & Goodman, 1994). In accordance with their transactional background, miscue analysts hold that readers need to actively engage with print to construct a mental representation from visual input that does not carry information by itself (Goodman, 1994; Smith, 2004). In doing so, four major ‗cueing systems‘ (or types of knowledge) are available: graphophonemics (letter-sound relationships), syntax (grammar), semantics (word meanings), and pragmatics (context and background knowledge). The oral production and comprehension of meaning and its subtleties that is not available entirely on the page that is being read is viewed as an outcome of the way the reader deploys these systems; the term ‗miscue‘ is used to convey that such occurrences are insights into the reader‘s strategies of meaning construction, rather than simply ‗errors‘. Miscues may show readers‘ deficiencies in the use of cueing systems; notably, ‗low-quality‘ miscues (i.e. unexpected responses which change the meaning of the text or result in meaningless renderings), suggest that the reader is not attending to meaning. Miscue analyses also include retells to provide insight into the reader‘s understanding of the text once the reading task is completed (Goodman & Goodman, 1994; Goodman, Watson, & Burke, 2005). In the common procedure, participants are presented with a complete text consisting of a minimum of 500 words. The text must be unfamiliar to them and ―difficult enough to challenge readers, but not so difficult that they can‘t continue independently‖ (Goodman,

104

Watson, & Burke, 2005, p. 46). Participants are asked to read the text out loud and to retell it afterwards. Miscues can be recorded on-line if the miscue analysis is used as an informal assessment tool, but for research purposes, the reading and retell are audio recorded and analyzed after the fact on a transcript of the text in question, numbered by lines. Goodman, Watson, and Burke outline three procedures of using miscue analysis; the ‗in-depth‘ procedure (pp. 131-162), in which minute attention is given to every miscue, was selected as appropriate for present purposes (for an outline see the Procedures subsection below). One of the strengths of miscue analysis, regardless of degree of formality, is its combination of descriptive and interpretive procedures, typical of qualitative research. As an example, occurrence of high-quality miscues (which do not change text meaning) in conjunction with an appropriate retell are commonly interpreted as fluent construction of meaning, whereas low-quality miscues (which do change text meaning) are seen as indicators of insufficient attention to meaning construction (e.g. Goodman, Watson, & Burke, 2005, p. 87). In the tradition of qualitative inquiry, miscue analysis thus allows for rich description and interpretation of reading behaviors. Studying eye movements in reading: History and technology. The study of eye movements in reading began in 1879, when Emile Javal observed that the eyes do not move smoothly over the text but rather in jerky patterns of leaps and stops, which he called ‗saccades‘ and ‗fixations‘ (Huey, 1908; Paulson & Freeman, 2003; Rayner, 1998). In order to obtain more precise information about these movements, researchers began to devise ways of recording them; Huey (1908; pp. 25-29) reports that his own device consisted of a plaster of Paris cup sandpapered until thin put directly on the cornea anesthetized with Holocaine or cocaine. Participants‘ heads were fixed to prevent head movements; an intricate mechanical

105

apparatus transmitted the movements of the eyeball to a metal pointer, which recorded them onto a paper covered with soot. The author assures us that despite this elaborate and intrusive setup, ―the reading proceeded as glibly and easily as could be desired‖ (p. 26). Eye trackers have become more sophisticated since those days, in particular after the advent of computerized technology. Current models commonly use an infrared beam to track the eye and no longer require headrests, offsetting the effect of head movements by counterbalancing eye tracking information with information from a head tracking camera. Large amounts of data can be collected and processed using specialized software programs (Applied Science Laboratories, 2008, 2009; Paulson & Freeman; Rayner, 1998). For this study, an ASL MobileEye head worn eye tracker (Applied Science Laboratories, 2008) was used after it proved difficult to capture the participant‘s pupil reflections with a desktop eye tracking unit (Applied Science Laboratories, 2009). The MobileEye consists of a pair of spectacles, similar to safety goggles used in industrial environments, with a camera unit and a mostly transparent mirror monocle mounted around the right spectacle (Fig. 3.1).

Figure 3.1: The ASL MobileEye Head Worn Eye Tracker (Applied Science Laboratories, 2009, p. 15).

106

The camera unit houses an eye camera and a scene camera, which together provide the device software with the information needed to compute eye movements relative to what is viewed. Via the mirror monocle, LEDs in the camera unit project three beams of near infrared light onto the eye. These beams are invisible to the participant but their reflection on the cornea, the ―thin film-like tissue that covers the eye‖ (Applied Science Laboratories, 2008, p. 16), yields three dots which are representations of the ‗corneal reflection‘ that are captured by the eye camera (via the monocle) together with the dark dot which is a reflection of the pupil. The relative position of these two visual marks provides the system with the information necessary to calculate gaze angle. Conveniently, the corneal reflection remains more or less stable while the pupil moves with the head; this allows for compensatory calculation of head movements. At the same time, the scene camera provides information of what is looked at, using the same angles to track eye movements on the scene under observation. This method of measuring eye movements is known as ‗Dark Pupil Tracking‘ (Applied Science Laboratories, 2008). Data is captured both numerically as a numbered sequence of fixations including beginning, end, and duration of fixations, and visually, showing eye movements in real time as crosshairs moving over the viewed scene. When using static eye trackers, i.e. units which are not head worn but mounted on a computer screen, visual data can be converted into fixation plots, which show fixations as numbered dots and saccades as lines against the background of the viewed scene (Applied Science Laboratories, 2010, a, b). This is the typical way that eye movement patterns during reading are reported in the literature. Figure 3.2 shows an example for such a plot from Nelson, Damico, and Smith (2008). Note that the dots are differentially sized, indicating fixation duration.

107

Figure 3.2: Example of a Fixation Plot Generated Automatically from Data from a Static Device (Nelson, Damico, & Smith, 2008, p. 300) Due to technical difficulties in capturing the participant‘s pupil with a static, mounted eye tracker, the head worn device as described above was used for this study. This required a different approach to visualizing the gaze data, as fixation plots cannot be generated automatically from a head worn eye tracker. Fixation locations are always relative to the position of the device; hence, any attempt at plotting a fixation progression would be thwarted by the fact that even the slightest head movement changes the apparent position of prior fixations. In other words, when using a head worn device, the displayed location of a fixation can be deemed accurate only while the fixation is being made. This technological limitation made it necessary to generate fixation plots manually in order to ensure comparability to the extant literature. For each fixation, location on the scene monitor was carefully transmitted to the original stimulus (the PowerPoint slides), marked with a dot, and numbered. Dots were then manually connected by a line in order to display fixation progressions. In the process, visual information of fixation durations was lost: it was not possible to accurately transmit dot diameters. Numerical information on durations was still available, however. Figure 3.3 shows an example of a plot generated manually for this study.

108

Figure 3.3: Example of a Fixation Plot Generated Manually from the Head Worn Eye Tracker Used in this Study Studying eye movements in reading: The findings. Over the course of the decades, a wealth of knowledge about eye movements in reading has been accumulated using technologies as described above. It is well established that visual information can be processed only while the eye is fixated. During this time, information in a 1 to 2 degree angle from the center of the fixation is in focus. This ‗foveal region‘ amounts to about 3 to 6 letter spaces, depending on font size and distance from text. Due to physiological limitations, information outside the foveal region will appear blurred to about a 5-degree angle (the ‗parafoveal region‘). Beyond that, no information will be taken in by the eye for processing; hence the need for saccades (Paulson & Freeman, 2003; Rayner, 2009). Information from both the foveal and the parafoveal regions is used during reading, but the region of usable information is asymmetric with regard to the center of the foveal region: it extends around three to four character spaces to its left, and about 14 to 15 character spaces to its right for readers of English. Information about a word‘s meaning is usually obtained from a region of about five to eight spaces to the right of the center (Rayner & Slattery, 2009). Results of eye tracking studies indicate that the eyes do not follow text in a linear, uniform fashion. Readers make regressions (i.e. leftward or other movements in the direction of text already passed); about 10 to 15 percent of all eye movements are of this regressive kind, and their frequency increases with text difficulty (Rayner & Slattery, 2009). Readers also skip about 20 to 40 percent of the words in the text they are reading depending on text

109

difficulty and reader characteristics. Skipping is associated with linguistic characteristics: short and predictable words are skipped more often than others, function words more often than content words (Paulson & Freeman, 2003; Rayner, 2009; Rayner & Slattery, 2009). Skipping and regression together make for a stark discrepancy between oral production and what is looked at during reading, as evidenced by comparisons of eye movements and oral reading performance (Paulson, 2002). Readers exhibit great variability on all measures of eye movement. This includes fixation duration, saccade length, and words skipped (Paulson & Freeman, 2003; Rayner, 2009; Rayner & Slattery. 2009). The extreme variability of eye movements while reading has been observed from the onset of the eye movement studies (Huey, 1908); researchers with quite divergent theoretical outlooks agree that this variability often is related to text meaning, even though the nature of the relationship is conceptualized in differing ways (Paulson, 2002; Paulson & Freeman; Rayner & Slattery). Paulson (2005) argues that, much like the weather, eye movements might best be described in terms of chaos theory, and he predicts that the more engaged a reader is in constructing meaning from text, the less linear and predictable their eye movements. The relation of text meaning, engagement with text, and eye movements was of paramount interest for the present study. Based on extant literature, it seemed reasonable to assume that participants‘ eye movements could be interpreted with regard to their interaction with print. Put otherwise, it was assumed that eye movements could be used, in conjunction with other reading behaviors and with interviews, to get an insight into what poor comprehenders actually do while reading, and to compare this with their schema as per their own reports, that is, with what they think they are doing.

110

Combining approaches: Eye Movement Miscue Analysis. Both eye tracking technology and miscue analysis offer ways to document reader behaviors, and interpret them in terms of underlying cognitive processes (Goodman & Goodman, 1994; Paulson, 2000, 2002). It makes sense, therefore, to combine them to gain a better understanding of what readers do while processing text. Eye movement miscue analysis (EMMA), introduced in Paulson (2000), is a hybrid approach to the description of reading behaviors that combines the recording of eye movements via eye tracking technology with the recording of oral reading performance, allowing for the comparative analysis of physiological and linguistic data (Nelson, Damico, & Smith, 2008). As an example, one of the findings in the original study was that miscues in adults, contrary to popular belief, are not caused by failure to fixate on a word (Paulson, 2000). In fact, miscued words tend to be fixated even longer than nonmiscued words. This finding was reiterated in a study examining a child with Downs Syndrome (Nelson, Damico, & Smith), showing that there is no a priori reason to assume that reading behavior including eye movements are fundamentally different in individuals with disabilities as compared to typical individuals. Other EMMA studies have shown how beginning readers use picture information to help meaning construction (Duckett, 2001, 2001); that bilingual fourth-graders do not substantially differ between languages in the reading strategies they employ (Freeman, 2001); that similar strategies are used even across semiotic systems such alphabetic text and musical notation (Kim, Knox, & Brown, 2007); or how excessive attention to the letter-by-letter sequence of print can hamper readers‘ efforts to construct meaning from the text (Brown, Kim, & O‘Brien Ramirez, 2012). Like miscue analysis, EMMA is a qualitative approach, which distinguishes it clearly from the experimental paradigm in which eye movement research is commonly used in the

111

study of reading (cf. the review of the field in Rayner, 1998). For this dissertation, EMMA was used as a means of generating rich descriptions of authentic reading behaviors, which had not been done previously for poor comprehenders. Artifact Analysis The description of cultural artifacts is a common aspect of ethnographic studies (Geertz, 1973; Simmons-Mackie & Damico, 2003). For present purposes, artifacts considered to be of potential interest included any written products related to the participant‘s overall literacy development. This included school documentation regarding the participant‘s literacy and literacy problems, academic performance, and other memos and notations that might lend depth to the participant description, as well as the participants‘ own writings and drawings, homework assignments, essays, and the like. No precedent for artifact analysis could be found in the GT literature; the actual analysis proceeded therefore as a type of incident-by-incident coding, as described below. Participant Participant recruitment for this study proved somewhat difficult. The problems began as early as the decision on how many participants to incorporate, persisted through recruitment, and resulted in the present case study. I will discuss these points in turn. Number of participants/data sets. The GT literature is equivocal as regards sample size. As mentioned above, initial data collection was seen as completed by means of literature review in the present study, and the actual data collection was regarded as theoretical sampling for the saturation of categories. In discussing data sets for theoretical sampling, Charmaz (2006) states that ―25 interviews may suffice for certain small projects‖ (p. 114) but no definite numbers are required. Noerager Stern (2007), by contrast, deems ―20

112

to 30 interviews and/or hours of observation adequate‖ (p. 117); like Morse (2007), she cautions against the pitfalls of collecting vast amounts of data. Corbin and Strauss (2008) are ―doubtful that five or six one-hour interviews can lead to saturation [of categories]‖ (p. 149), but do not make any statement as regards an acceptable number of interviews, leaving the researcher with an even wider range of what might be adequate. Thus, we have to infer that any number from 20 interviews upwards (and possibly less than that) may be sufficient for the development of a grounded theory. However, the present study differs from the more typical GT studies referred to in the quotations above. First, there was a need to take into account the temporal constraints of dissertations, and the time-consuming nature of the analyses suggested. It was clear that it would not be possible to incorporate more than five participants. Second, and more importantly, in the type of study referenced above, interviews are typically the only type of data, or a mix of interviews and observations. Read in this light, the literature recommends the use of (very roughly) 20 data sets consisting of one-hour interviews or of observations. In the present study, by contrast, an array of variegated data was collected; analytic procedures had to be adjusted accordingly. Consequently, the investigation was designed originally to conduct two to three interviews per poor comprehender (one with the participant and one with the parent, plus another one with other stakeholders or professionals such as SLPs), and to collect two sets of reading data from each participant (one in a vaguely naturalistic situation and one in the eye-tracking laboratory). These, in turn, were subjected to two types of analysis: miscue analysis/EMMA and behavioral analysis of video data. In addition, field notes and artifacts were obtained and analyzed. All in all, the following data sets were collected:

113



3 interviews (one each with the poor comprehender and his parent, and one with a speech-language pathologists (SLPs) working with the poor comprehenders, which was used as secondary data in the discussion section)



2 miscue analyses



2 EMMAs



1 eye movement recording while reading silently



1 set of observational data (field notes and video data)



1 set of artifacts which were consulted for interpretation as necessary

In sum, 10 data sets were obtained for this study. Obviously, this represents a less than minimal amount of data to work with for the purposes of GT, hence the cautionary statements made above have to be repeated – this dissertation, more than anything else, is an exploration of the usefulness of contextualized GT research into poor reading comprehension. Recruitment. Participant recruitment happened on an opportunistic basis, when and wherever the occasion arose. The recruitment process was started by initiating contact with local SLPs and area school boards, using both personal and professional connections. Despite variegated efforts, the original target quota of five participants was not reached. This was due to several factors. First, many persons who were contacted proved rather reluctant to work with the researcher; this applies to administrators, professionals, and parents alike. Second, where help was granted (and very generously so by some), suggested potential participants frequently did not meet definitional criteria. That is, a number of children labeled as poor comprehenders turned out to be poor overall readers. In the end, data were collected on two clear cases of poor comprehension, and one overall poor reader who

114

struggled just enough with decoding to fail to qualify as a poor comprehender (see below for assessment procedures). Due to various difficulties (technical and otherwise), the data on the second poor comprehender was impoverished as compared to the data on the first; also, analysis of the first participant‘s data proved so time-consuming (albeit fruitful) that it was decided to turn this dissertation into study of a single case focusing more on theoretical implications of the extensive data set. While unusual both for a dissertation and a GT study, the decision seemed justified on three accounts: the sheer richness of the data, the scope and depth of the analysis, and the fact that the study was never intended to be more than an exploratory account. In accordance with the literature on poor comprehension (See Chapter Two), the target age for potential participants was between 8 and 11 years; the participant for this study was 9;1 to 9;3 years old at the time of data collection. (Although this is not explicitly discussed in the literature, it would seem that poor reading comprehension becomes first tangible in this age range, when children have developed enough graphophonemic strategies to appear to be proficient readers, cf. Yuill & Oakhill, 1991). The participant was recruited through the University of Louisiana at Lafayette Summer Literacy Project, in which he had been enrolled for two consecutive summers at the time of this study. Assessment and qualifying criteria. To ensure comparability to extant literature, assessment and qualifying criteria followed the studies discussed in Chapter Two. After a potential participant was indicated, the child was administered a test of reading comprehension and a test of nonword decoding to determine whether he/she was part of the population in question (scoring low on reading comprehension), or dyslexic (scoring low on decoding), or an overall poor reader (scoring low on both instruments). The measures most

115

commonly used in the literature are the Neale Analysis of Reading Ability-Revised (Neale, 1997) for comprehension (this instrument also includes a word-reading accuracy test) and the Graded Nonword Reading Test (Snowling, Stothard, & McLean, 1996) for decoding. However, these tests could not be used for the present study, as they were not available for the North American market. It was therefore decided to use the Gray Silent Reading Test (Wiederholt & Blalock, 2000) for assessing children‘s comprehension, and the Word Attack subtest of the Woodcock Language Proficiency Battery (Woodcock, 1991) for assessing their decoding abilities. Both measures have been shown to be reliable indicators of the respective skill (e.g., Farnill & Hayes, 1995; Keller, 2003). Following Cain (2006), children were classified based on age equivalent scores. If they obtained reading comprehension scores equivalent to more than 12 months below both their chronological age and the age equivalent obtained on the decoding test, and if decoding age scores were not lower than their chronological age, they were classified as poor comprehenders, and qualified for this study. As previously mentioned, participant recruitment proved difficult; this was true not only in terms of making contact with cooperative referral sources and potential participants but also in terms of the instruments used. For reasons that cannot be adequately discussed in the present chapter, the decoding portion of the assessment had to be readministered (details of this aspect of the study will be examined in subsequent chapters). One of the participants, who had qualified as a poor comprehender on the original measure, did not qualify on the subsequent one; his difficulties with decoding on the second measure fit with an overall poor reader but not a poor comprehender profile. The participant who was eventually admitted to the study had divergent profiles on the different assessment used; this finding will be discussed in Chapter Four.

116

At this point, a note on other aspects of participants‘ functioning is in order. Several studies from the relevant literature controlled for children‘s nonverbal cognitive abilities, labeling them as ‗poor comprehenders‘ only if those abilities were similar to typical readers. However, at least in one instance (Cragg & Nation, 2006), poor comprehenders scored significantly below typical readers on nonverbal cognitive measures. This is in line with findings by Nation, Clarke, and Snowling (2002), according to which poor comprehension and cognitive difficulties may often be concomitant. It was therefore decided not to assess nonverbal cognitive abilities for the present study. Also, children with a diagnosis of language impairment were eligible participants, since language impairments are an agreedupon characteristic of many poor comprehenders; the same applies to mild learning impairments. Potential participants were excluded if they had had any of the following diagnoses: moderate to severe cognitive impairment; developmental disability; autism or Asperger‘s syndrome; or pervasive developmental disability. This exclusionary criterion was employed since comprehension issues in these populations would appear to be part of a problem constellation that is different from the general language or mild cognitive problems exhibited by poor comprehenders. None of the children that were assessed presented such problems. After assessment and admission to the study an appointment for data collection was made. Data collection was not effected on the same day as assessment to prevent fatigue. Procedures: Data collection After contact was established and consent/assent obtained (see Appendix A for examples of the respective forms), children‘s comprehending and decoding abilities were assessed as described in the previous section. If they qualified, data collection was effected. This subsection serves to describe the exact procedures for this part of the study.

117

Home portion: Interviews and audio/video recorded reading samples; field notes. As mentioned, the home reading sample was taken to gain contrastive data with respect to the laboratory situation (for the sake of convenience, interviews were done on the same occasion as home readings). In the eye tracking laboratory, participants are seated on a large adjustable chair in a tiny room full of computers and other equipment; outfitted with a head worn electronic device; and asked to read from a screen while being closely monitored by two (mostly) unfamiliar adults. It was assumed that this situation would not be very representative of a typical reading situation in the life of the participant, calling in question the validity of the data as regards the purpose of gaining a naturalistic reading sample (i.e. of data that shows what readers ‗normally do‘). The home reading situation was therefore created to provide a counterbalance to the laboratory context. Here, the participant was free to choose a comfortable location and familiar texts for this situation. (It was decided that it was more important to gain a naturalistic sample than to assess comprehension of previously unread texts for this portion of the data.) The participant and his mother were asked, prior to the data collection session, to choose a range of texts the participant would like to share with the researcher. To ensure presence of reading materials, age-appropriate materials were also provided by the researcher in case they were needed. These materials were selected according to presumed interest to the participant as based on the researcher‘s experience with literacy intervention and children‘s literature. They were roughly matched to the participant‘s chronological age using the Accelerated Reader Book Finder website (Renaissance Learning, Inc., 2012). However, these researcher-supplied materials were not needed and were not used. During the actual session, the participant picked two texts he had previously read from a pile of books provided

118

by his mother. The participant was free to take breaks during the data collection process but did not take advantage of this opportunity. He was asked to read three texts but wished to read no more than two, and the researcher complied. The first set of data collected during the home session was the parent interview with the participant‘s mother. Next, the participant was invited to sit with the investigator for an interview and the reading portion of the data collection. It should be noted that miscue analysis guidelines were adapted for the present study. Recall that the common procedure is to use at least one complete story with a length of minimally 500 words. For present purposes, however, it was decided to sample variegated reading performances including different genres, notably story books and expository (schooltype) texts in order to appraise differences in performance related to textual and genre differences. Due to difficulty finding suitable materials, it was decided that instead of reading complete texts, the participant would only read the beginnings of the respective texts (except for the second home reading text, where the participant insisted on reading a central portion instead of the beginning). As the texts were provided by the participant, the endpoint for reading could not be established beforehand. As will be further discussed in Chapter Four, the participant was asked to halt reading the first text by the researcher as he seemed to lose focus. For the second text, he had to be coaxed into further collaboration and hence was promised he would only have to read two pages of the book he had chosen. The home readings were both audio and video recorded to allow for behavior analyses. An Olympus VN-5200PC digital voice recorder was used for the audio portion, together with a Philips Voice Tracer 600 digital voice recorder for backup. The video

119

recording was done using a Sony DCR-SX45 handycam. Also, field notes were taken to capture the researcher‘s observations. Laboratory portion: Eye tracking and audio/video recordings. For the second round of data collection, the participant was invited to the University of Louisiana at Lafayette‘s eye tracking research laboratory. Due to scheduling issues, this part of data collection occurred 7 weeks after home reading. It was assumed that this would have no bearing on the data. As the EMMA was to be the centerpiece for the analysis of behavioral data, materials were carefully prepared. Texts were chosen using the Accelerated Reader Book Finder website (Renaissance Learning, Inc., 2012) to roughly match for age, and based on the researcher‘s knowledge of and experience with engaging children‘s literature. Texts were also chosen to represent different levels of difficulty. As with the home reading, it was appropriate to collect reading data across genres. Hence, two of the three stimuli were taken from children‘s storybooks; the third one was expository in nature. To ensure sufficiency of stimuli, materials for this portion of data collection were prepared to include a minimum of 400 words each. Also, during this data collection session the participant was not given choices and the expectation that all three texts were to be read was clearly communicated. The materials for the eye tracking samples were prepared for reading from a computer screen. Books were scanned on an HP Photosmart D110 series scanner with a resolution of 600 dpi and saved as *.bmp files. To ensure readability, measurability, and consistency of visual input, scanned pages were loaded into PowerPoint presentation software where they were divided vertically and scaled proportionally so that font size was roughly equal to 30pt on a 1024×768 desktop screen. Consistency of font size was assessed

120

by measuring the height of rounded, mid-sized letters (e and o) on the screen using a common measuring tape. To preserve consistency of size, each complete PowerPoint presentation was then saved as a ‗device independent bitmap‘. Figure 3.4 shows an example for an eye tracking stimulus.

Figure 3.4: Example Stimulus Material for Eye Tracking (from Avi, 1988). In order to maximize naturalness of reading situation, initial plans were to use an ASL eye tracking desktop unit (Applied Science Laboratories, 2009). However, it proved impossible to capture the participant‘s pupils with this type of eye tracker. In a second attempt using the head worn MobileEye device (Applied Science Laboratories, 2008), the data collection was successfully completed. The tasks for this part of data collection differed from the home reading portion. One of the readings was completed silently. The purpose of this was to compare eye movements between the reading out loud and the silent conditions, since some authors have made the point that both represent different types of reading (e.g. Goodman, 1967). All tasks included a retell as per common miscue analysis/EMMA procedures. Task sequence was as follows:

121

storybook silently read and retell; storybook out loud and retell; expository text out loud and retell. All readings were recorded in multiple ways. The central method of recording was the eye tracker, which by itself yielded two sets of data: video recordings of eye movements on the screen (including audio information), and numerical recordings of fixations and saccades. A camera (the same as above) recorded body language and facial expressions from a slightly further distance. Finally, two audio recorders (as previously mentioned) were used to provide backup audio data. Data Analysis As is common in GT (Charmaz, 2006; Corbin & Strauss, 2008), the data analysis process began before completion of data collection. At odds with GT traditions, though, collection was not informed by preliminary analysis results (‗theoretical sampling‘), since the data collection itself had been classified as theoretical sampling from the start. Also, the sequence in which data analysis was carried out could not conform to GT precedents, as the data was more variegated than is common in GT studies. To comply with the dissertation process, which requires an established procedure, while still following GT tradition, it was therefore decided to analyze data on a set-by-set basis using researcher ‗insights‘ (see above) that applied across sets, and to constantly compare the data under scrutiny to previously analyzed data; these constant comparisons precipitated in memos. At the same time, analysis was designed to avoid strict linearity in order to keep the ‗dance‘ aspect of qualitative inquiry (Janesick, 1994). Consequently, the first interviews (with the participant and his mother) were transcribed and subjected to line-by-line coding right after they had been collected in August 2012, and the first memos were drafted throughout the data collection process, which

122

lasted from August to December 2012. The analysis process was then paused and taken up again in June 2013. Barring these temporal disruptions, and against a background of continuous memo writing, analysis proceeded as described below. The processes and descriptors used are taken from Charmaz (2006) and Corbin and Strauss (2008). Note that since in qualitative research, analysis and results are not as strictly separated as in experimental studies (see Janesick, 1994, for an excellent description of the reiterative ‗dance‘ of qualitative analysis), the actual procedures will only be outlined in this chapter, and presented in more detail in the next. Step 1: Line-by-line coding of interview data. As mentioned, the first step in data analysis was line-by-line coding of the interviews with the participant and his mother. This type of coding translates the raw linguistic data into concepts generated by the researcher, making them amenable to further coding and analysis. The purpose of line-by-line coding is to stay as close to the data as possible while at the same time rendering it in a form that allows subsuming similar topics under one concept (Charmaz, 2006). As an example, the participant‘s mother spoke about ‗reading aloud‘ at various points in the interview; line-byline coding allowed for comparing the context of these mentions, thereby enabling the teasing out the significance of ‗reading aloud‘ in these various contexts. Another purpose of line-by-line coding was to discern the use of ‗in-vivo codes‘, that is, of ―participants‘ special terms‖ (Charmaz, 2006, p. 55) whose meanings illuminate some aspect of participant experience. For example, the participant‘s mother referred to ‗reading‘ multiple times during the interview; however, it became clear from the context that she was actually referring to oral reading. ‗Reading‘, for her, served as an in-vivo code for a conception of literacy that does not necessarily include comprehension.

123

It should also be noted that the moniker ‗line-by-line coding‘ is not to be taken literal. GT authors do not specifically mention this, but their examples make it clear that while each line of the interview (or any other type of data) is scrutinized for concepts, coding is only done when such concepts are actually found. In typical interview data, this is the majority of lines but by no means all of them. Also at this point of data analysis, field notes were scrutinized for observations of interest, and the first instances of the ‗constant comparisons‘ that make up the backbone of GT inquiry were between the interview data and the field notes. Emerging concepts precipitated as memos. Step 2: Focused coding of interview data. To establish higher-order categories from the coded concepts in the interviews, ‗focused coding‘ (Charmaz, 2006) was used, a type of coding that builds on line-by-line or incident-by-incident concepts, as well as in-vivo codes, to ―synthesize and explain larger segments of data‖ (p. 57). That is, the focused coding attempts were employed to establish important ‗themes‘ or ‗categories‘ from conceptual data; this is a common approach in qualitative research (e.g. Janesick, 1994; Simmons-Mackie & Damico, 2003). Step 3: Miscue analysis. After line-by-line coding of interviews was completed and in-vivo codes established, miscue analysis was effected on the home reading data. The recorded text was transcribed, marked, and coded in line with the guidelines from Goodman and Goodman (1994) and Goodman, Watson, and Burke (2005). As mentioned, this study used the ‗in-depth‘ procedure as described by Goodman, Watson, and Burke, which calls for the examination of each miscue in terms of the four cueing systems, as well as in terms of

124

meaning change and correcting behavior, using a set of six questions to be answered by the analyst: 1)

Does the miscue occur in a structure that is syntactically acceptable in the reader‘s dialect?

2)

Does the miscue occur in a structure that is semantically acceptable in the reader‘s dialect?

3)

Does the miscue change the meaning of the entire text?

4)

Is the miscue corrected?

5)

How much does the miscue (…) look like the text word (…)?

6)

How much does the miscue (…) sound like the expected response? (2005, pp. 135-136)

Each one of questions 1) through 4), can be answered Yes, Partial, and No by the analyst; questions 5) and 6) can be answered High, Some, or No (referring to ‗similarity‘). The coded transcript was then subjected to statistical and qualitative analysis, and the retell was evaluated for completeness and appropriateness. It needs to be borne in mind that miscue analysis is fundamentally a qualitative tool which is adaptable to the needs of the analysis process; as will be seen, the standard in-depth procedure was slightly altered for the purposes of the present study. Step 4: Incident-by-incident coding of miscue data. To embed the miscue analysis into the GT framework, the transcript and the statistical data were subjected to incident-byincident coding, the ―close cousin‖ (Charmaz, 2006, p. 53) of the line-by-line approach. This type of coding is used on data that is not propositional the way interviews are, such as behavioral phenomena. Thus, any observations made from miscue analysis data that

125

appeared to pertain to the research questions were coded this way. Also at this stage, home reading video recordings were evaluated for behavioral observations, which were coded using the same incident-by-incident method. This led to an intermediate step of behavioral analysis (see below). Constant comparison and memo writing continued as a preparatory activity for higher-level coding. Intermediate step: behavioral analysis. During the process of coding miscue data, attention was drawn to a brief interchange between the participant and myself, which occurred between the two home readings. Subjected to a sequential analysis of the interaction, it illustrated the themes gained from the data and will be reported in Chapter Four. Step 5: Eye Tracking Miscue Analysis (EMMA). The next step in the analysis process used the eye tracking data, and the video data from the eye tracking sessions. This step involved a fourfold analysis. First, eye movements were scrutinized to get an initial overview of patterns. The researcher watched the eye movement video and took notes on phenomena of interest, translating behavioral into conceptual data. Second, the actual EMMA was effected, for which reading materials were coded according to the in-depth procedure (Goodman, Watson, & Burke, 2005) while simultaneously comparing miscues (linguistic data) to eye movements and to concepts established previously. Third, the numerical data captured by the eye tracker was subjected to statistical analyses and used for comparison with established reading data. Fourth, the retell was evaluated as previously and compared to the miscue and eye movement data. Constant comparison and memo writing continued throughout.

126

Step 6: Axial coding and comparison to sensitizing case. With analysis progressed to this point, the process moved on to organizing the data along an overarching conceptual framework; hence, this part of the analysis will be presented as ‗discussion‘ (Chapter Five) of the ‗results‘ (Chapter Four). To some extent, the insertion of this step makes this investigation unique when compared to other studies using grounded theory and when compared to many previous studies of poor comprehenders. Note that the distinction here between ―discussion‖ and ―results‖ as applied here is only vaguely useful for a qualitative study in which interpretation is iterative and no clear-cut distinction between presentation and interpretation of data is made. Put more accurately, this step amounted to what is called ‗axial coding‘ in the GT literature. Axial coding was deemed a useful tool for this study even though it is not clearly defined between authors and seems to be falling out of favor with GT researchers. Charmaz (2006, pp. 60-63) defines axial coding as a way of relating categories to subcategories by organizing the latter around a central ‗axis‘, and she cautions that this may impose a rigid interpretive framework and actual hinder analysis. By contrast, Corbin and Strauss (2008, pp. 195-228) refer to axial coding simply as ―the act of relating concepts/categories to each other‖ (p. 198). Despite these divergent opinions on the process, it was deemed a useful step for this theoretically-oriented case study. For present purposes, a central axis was used in Step 6. The axis was provided by the account of the emergence of the participant‘s poor comprehension as given in the interviews. Given that the characteristics of the participant‘s development could be ‗naturally‘ arranged along this central timeline, it was deemed warranted to use it as an axis despite Charmaz‘ cautionary note. Also during this step, two descriptive metaphors emerged from the analysis and were used to formulate a grounded theory in the next step: a ‗systemic‘ and an

127

‗ecological‘ account of the case. Comparison to a sensitizing case I had encountered prior to the study provided further illustration of the emerging theory. Step 7: Formulation of a grounded theory on poor reading comprehension. The final step and culmination of the analysis process was the formulation of the grounded theory on poor reading comprehension, which was the goal of this study. As mentioned previously, this theory can only be viewed as tentative, given the paucity of data available in the framework of this dissertation. Despite its tentative status, however, the emerging theory proved sufficiently substantial to warrant further investigation using the same methodological paradigm, as well as suggestions for further research based on different methodological and conceptual approaches. The results of this investigation, including primary and secondary data and the interpretation of these data sets, are contained in Chapter Four. Chapter Five will discuss the results in light of the conceptual framework and the final chapter (Six) will provide appropriate summations and conclusions for this investigation.

128

Chapter Four: Results To know man as he is no mean aspiration. Cronbach (1975, p. 126) In the spirit of Cronbach‘s remark, this and the following chapter are dedicated to understanding the participant in terms of his characteristics—behaviors, attitudes and motivations, and experience. As outlined in the previous chapter, the discussion of the results will adhere roughly to the sequence of data collection, beginning with demographic and clinical information. This information will be treated as part of the results, as it constitutes an important part of the data. Field notes and artifacts will not be discussed separately but rather in an interspersed fashion throughout to provide background information. Demographic and Clinical Information The participant (henceforth P) is a Caucasian male who was 9 years and 1 month old when assessed for eligibility, and 9 years 3 months by the time data collection was finished. The period of data collection coincided with his entry to fourth grade. No formal assessment of socioeconomic status was effected but field notes describe his home environment as follows: ―a nice middle class neighborhood, their [P‘s family‘s] house is one of the bigger and nicer ones. 2 stories, wooden paneling, lots of space‖ (Appendix B, p. B1). Observation thus suggests that his household belongs to the middle-class stratum, albeit not its upper echelons, as he is attending a public school. P earned a raw score of 0 on the Gray Silent Reading Test, which places him in the < 7;0 age range and at the 1.0 grade level, the lowest age and grade equivalents specified on this instrument (Appendix C, pp. C1-C2). By contrast, his decoding score on the Word Attack subtest of the Woodcock Language Proficiency Battery was 25, placing him in the age

range for 29-54 years, or at the 14.4 grade equivalent (Appendix C, p. C3). In other words, P has a clear and rather extreme poor comprehender profile. P‘s mother mentioned in the interview that reading aloud helps him comprehend; it could thus be suspected that these results were an artifact of the assessment condition. However, they were validated by an additional assessment using the Gray Oral Reading Test 4th edition (Wiederholt & Bryant, 2001), on which P scored at an age equivalent of 7;3, and a grade equivalent of 2.2 for comprehension (Appendix C, p. C4). Thus, while not as extreme as on the silent assessment, his comprehension score was more than a year below his actual age and grade level, confirming the poor comprehender profile. P was recruited through the University of Louisiana at Lafayette Literacy Project, a summer intervention program that he had attended several times. No formal evaluation had been effected from the University of Louisiana at Lafayette‘s side, but school documentation provided by P‘s mother indicated that he had been receiving speech-language therapy through Lafayette Parish Public school board since he was 3;3 years of age, and was in ‗educational need‘ in the area of communication at the time of data collection (Appendix D, pp. D1-D2). The interview with his mother revealed further that P had been receiving speechlanguage therapy for significant language delay from 15 months of age. As she put it, ―he‘s always had about an 18-month speech delay. So, when he was 2, he still wasn‘t talking‖ (Appendix E, p. E4). In addition, P had also been treated for articulation difficulties. His mother attributed these issues to recurrent fluid build-up (otitis media) and resultant hearing problems in early childhood. First-grade school documentation indicates some persistent problems with articulation at age 5;10, and this documentation adds additional concerns in the following language areas,

130

―identifying antonyms and synonyms, answering comprehension questions, formulating questions, and using appropriate grammar‖ (Appendix D, p. D1). According to his mother, P‘s articulation was no longer a concern at the time of this study; intervention in this area had ceased in second grade (Appendix E, p.E7). The documentation corroborates this (Appendix D, p. D3). The most recent Individual Education Plan (IEP) progress report made available for this study is dated March 26, 2012. In this report, goals are described as follows: By March 2013, [P] will be able to read/respond or listen/respond to various text dependednt [sic] questions presented which are of the following type and/or those which require him to: understand and relate vocabulary, main idea, infer, evaluate, explain, describe, ect. [sic] (Appendix D, p. D2) In addition to school-based services, P received language intervention from an SLP in private practice at the time of this study (Appendix E, p. E12). P‘s mother also reported that her son had been diagnosed with attention deficit disorder (ADD) and was receiving medication which helped with school participation and homework, but not with his comprehension problems (Appendix E, pp. E15-E17). In sum, P presented with a clear poor comprehender profile and a history of speech and language difficulties. His articulation problems no longer seemed an issue at the time of this study, but as will be seen below, remnants may still be influencing his reading behaviors. His language problems persisted, notably in comprehension-related areas. Despite this, he had passed the iLEAP test (Louisiana Department of Education, 2007) and had been admitted to fourth grade (Appendix D, pp. D4-D6). Possibly in relation to this, the school had

131

announced that effective at the beginning of fourth grade, he would no longer be provided with extended school year services (Appendix D, p. D7). Table 4.1 sums up demographics, assessment results, and clinical information for P.

Characteristic

P

Age

9;1-9;3

Gender

Male

Race/ethnicity

Caucasian

SES

Middle class

Grade

4

Result GSRT

Age equivalent < 7;0, Grade equivalent 1.0

Result WLPB word attack

Age equivalent 29-54 years, Grade equivalent 14.4

Clinical history

18-months speech and language delay from infancy until elementary school age; articulation problems until 2nd grade, persistent language problems; attends language therapy

Table 4.1: Demographics, Assessment, and Clinical Information for P Interviews The interviews with P and his mother were effected on the same date as the home readings. I went to their house for the occasion, interviewed P‘s mother first, then P himself, and then set up equipment for the home reading. His mother was rather forthcoming and her interview lasted around 31 minutes; it turned out to be very rich in information. P himself was less talkative; my field notes state that he ―appeared a little reluctant and possibly overwhelmed (or confused?)‖ (Appendix B, p. B2). After completing this part of data collection, I transcribed the interviews and coded them line-by-line; simultaneously, I wrote various memos, which reflect the richness of these initial data. About a year later, I re-examined the data and subjected it to focused coding, 132

chunking line-by-line codes into themes. For this part of the analytic process, I treated both interviews as one set of data. This was done for two reasons. First, the interview with P yielded only limited usable information. For example, it was not possible to determine his attitude towards books with certainty. He reported both getting bored with books and not getting bored with books; when asked about specific books he liked, he mentioned storybooks such as Diary of a Wimpy Kid (Kinney, 2007-present) or Flat Stanley—a genre he usually dislikes according to his mother (Appendix E, p. E25). While it was generally difficult to elicit coherent answers to abstract matters from P, it is possible that these responses were prompted by the way I worded my questions, for example, introducing the topic of boredom into his discourse (see below.) The second reason I coded the themes together for both interviews is that where useful information was obtained from P, it matched squarely with the themes gleaned from P‘s mother‘s reports. I therefore decided that treating both interviews as one would be truer to the data than formulating themes separately for each interview and linking them through axial coding. Through this process, eight themes or thematic categories were generated and labeled as follows: READING VS. COMPREHENDING; (LACK OF) SUPPORT; HALF THE STORY; ISSUES EXPRESSING HIMSELF; (LACK OF) MOTIVATION & NEED FOR SCAFFOLDING/STRUCTURE; OTHERORIENTATION; TANGIBILITY; EMERGING AWARENESS.

Most of these themes were supported by

subordinate themes during the initial data analysis that were best explained by formulating the inclusive themes. Table 4.2 lists each subordinate theme and the theme under which each was organized for purposes of explication.

133

In addition, a ninth theme was generated that differed from the others in two ways. First, it was grounded in data running through all of them, rather than in a particular set of codes. At the same time, however, it did not seem appropriate to make it an overarching macro-theme since the second difference was that it had an even more ‗interpretive‘ aspect to it than the other themes; reflecting my ‗gut feelings‘ while reviewing the data and trying to verstehen the situation. In the literature I reviewed, there is no mention of such a qualitative (as opposed to taxonomical) distinction between types of themes. I will therefore introduce such a distinction for the purpose of this study. Corbin and Strauss (2008, pp. 45-53) explain how all descriptive labels we use for analysis are concepts; they ―stand for groups or classes of objects, events, and actions that share some major common property(ies)‖ (p. 45). In this sense, the eight themes mentioned are conceptual; based on common properties in the underlying data. By contrast, the last theme is more connotative, pointing to my reaction rather than to the code properties; a connotative theme as opposed to the conceptual themes listed above. This connotative theme is labeled UNCERTAINTY. The themes will be discussed in turn.

134

Themes

Subordinate Themes

READING VS. COMPREHENDING

(none)

(LACK OF) SUPPORT

(none)

HALF THE STORY

(Uncertainty of) Medical origin Language delay Early comprehension problems Focus and demand, Homework trouble Things of the past ADD as medical aspect ADD vs. comprehension Literacy environment

ISSUES EXPRESSING HIMSELF

Boredom Aggravation

(LACK OF) MOTIVATION & NEED FOR SCAFFOLDING/STRUCTURE

Motivation Need for scaffolding/structure Interests

OTHER-ORIENTATION

Coping mechanisms Role of teacher Group sports issues Lack of peers Passivity People pleaser Other-orientation Spelling = copying words

TANGIBILITY

Tangibility Comprehension intervention Positive affect vs. tangibility

EMERGING AWARENESS

Growing awareness of comprehension Increasing confidence Awareness issues

UNCERTAINTY

(none)

Table 4.2: List of Themes Derived from Interviews with P and his Mother

135

Theme One: READING VS. COMPREHENDING. There was a conspicuous distinction woven into P‘s mother‘s discourse; she consistently treated ‗reading‘ as something separate and distinguishable from ‗comprehending‘ what is read (a dichotomy reminiscent of, though not equal to, the Simple View of reading as discussed in Chapter Two). The distinction was introduced in her very first turn at talk in response to my introductory question, ―tell me anything you can think about‖; she stated, ―he… can read, he just has a (hard time) comprehending what he reads‖ (Appendix E, p. E1; see Appendix E, p. E32 for transcription conventions). She explained further that P ―has an easier time reading things when he reads it aloud‖ (Appendix E, p. E1) but that ―he has a hard time reading something silently and then processing it‖ (Appendix E, p. E2). She continued making this distinction while elaborating on P‘s skills and issues, explaining, for example, that ―he can read fine. He can read War and Peace. He just can‘t tell you what it‘s about‖ (Appendix E, p. E4). She also used the term ‗illiterate‘ in the same way (cf. the themes labeled ―Literacy‖ in Appendix E), explaining that, ―there‘s kids who‘re illiterate. He‘s not illiterate,‖ and that, ―[the school] they‘re gonna focus their attention on kids who are illiterate, not on kids who have a comprehension issue‖ (both p. E7). As seen in Chapter Two, there is no model of reading which views comprehension as something separate from reading, not even the dualistic Simple View. P‘s mother‘s use of the term represents an ‗in-vivo code‘, a term laden with specific participant meanings that differ from common usage (Charmaz, 2006, pp. 55-57). It also turned out that this in-vivo code was somewhat ill-defined, which is not surprising, as such codes are representative of lay lore, not of scholarly concepts. There was one notable exception to P‘s mother‘s consistent use of ‗reading‘ as ‗reading aloud‘ (and ‗illiterate‘ as ‗being unable to read aloud), namely, her

136

statement on silent reading quoted above. This inconsistency is puzzling. If she equates reading with verbal behavior, then how does she conceptualize what happens during silent reading? Unfortunately, I did not have the opportunity to probe into her understanding of silent reading. For now, therefore, all that we can conclude is that silent reading behaviors are mostly absent from P‘s mother‘s discourse. As we will see in other parts of the interview, P‘s school makes the same conceptual distinction between reading and comprehending, and some of the intervention for P‘s literacy issues seems to have been informed by it. Notably, his accommodations, as well as decisions to not provide accommodations, were informed by this concept. Finally, it is noteworthy that P himself seems to equate reading with its oral-verbal aspects. When I asked him if he ever goes back in a text in order to reexamine something he didn‘t understand, he responded: ―when I um… get the words wrong um… (my teacher) writes something (like) that then I have to start over. Wait not start over. (…) back where you were reading‖ (Appendix E, p. E45, emphasis mine). He also classified himself as a good reader (Appendix E, p. E47), which is only true when reading is defined as oral-verbal proficiency. Theme Two: (LACK OF) SUPPORT. The second theme emerged with P‘s mother‘s second turn at talk, and much of her discourse was laced with pertinent comments. She reported that she devotes considerable concern to the services provided to her son, both within and outside the school system. What struck me about this theme was an apparent disconnect in her appraisal of school efforts (cf. Memo Four, Appendix B, p. B6). On the one hand, it was clear that P had been receiving considerable accommodations, and continued to be provided with support. As an example, he was allowed to take his tests in the hallway,

137

which enabled him to read the questions aloud for better comprehension, and he was assisted by an aide marking his responses on the sheets. In addition, he was receiving speechlanguage therapy at school by the time of this study, and his mother appeared to be appreciative of the SLP‘s efforts (cf. Appendix E, p. E14). On the other hand, she was in an ongoing struggle with the school about recognition of P‘s ADD. Her goal was to get him an in-class aide to scaffold his comprehension during ongoing classroom tasks (more about this in subsequent themes). There was an undercurrent of frustration in her discourse while she was talking about this particular fight; the following quote conveys the flavor of her explanations: So I had to pay to have him tested. I wanted I wanted the school to do it. I wanted the school board to test him, they wouldn‘t. I wanted him reevaluated. I (…) had a request to have him reevaluated… for… years and I never—they never got it. So I‘m still fighting to s—today still fighting to get him reevaluated. So he can get um Special Ed minutes. (Appendix E, p. E16) Her frustration was even more palpable when she was talking about a past intervention decision made by the school. Due to P‘s early speech and language problems, he had been placed in a reading tutorial program as early as kindergarten. According to his mother, this was until about second grade, at which point, she said, ―they pulled him out. Because he was able to read‖ (Appendix E, p. E7). And she elaborated (I quote at length, omitting my own turns at talk, to illustrate how she expressed her feelings): The school doesn‘t care that he doesn‘t understand, the school cares whether he can read or not. And he can read. So they pulled him out of reading. Because there‘s kids

138

who‘re illiterate. He‘s not illiterate. So they‘re gonna focus their attention on kids who are illiterate, not on kids who have a comprehension issue. So this kid can understand what he reads, he just can‘t read. My kid can read, just can‘t understand what he reads. They‘re focusing on this because it‘s something tangible and not on him because it‘s something that—they‘re just not concerned about. So they—they kicked him out of the reading program. (Appendix E, pp. E7-E8) As mentioned above, this passage makes it clear that the school based its decisions regarding P‘s reading abilities on his oral-verbal behavior, not on his ability to comprehend texts. His mother accepts that distinction in general but is frustrated with the decision: she sees her son struggle with comprehension and wishes for the school to support him in this issue. But it appears that this is not on the school‘s agenda; it ‗doesn‘t care‘ about P‘s struggles as much as it cares about meeting ‗tangible‘ benchmarks (a theme we will explore below). Theme Three: HALF THE STORY. This theme centers on the etiology of P‘s comprehension problem as recounted by his mother; it subsumes several subordinate themes that lend credence to the idea that much is not known or understood regarding P‘s difficulties: (Uncertainty of) Medical origin, Language delay, Early comprehension problems, Focus and demands, Homework trouble, Things of the past, ADD as medical aspect, ADD vs. comprehension, and Literacy environment. It began to emerge with the account that his mother gave early in the interview. I quote the first two longer turns at talk (omitting a hearer feedback signal by myself):

139

…one of the theories is that when he was (…) about 12 months old, he was having a hard time walking, he was having a hard time talking. And it‘s because (…) he had so much fluid in his ears, he couldn‘t hear. So it‘s theorized that—because he couldn‘t hear correctly, he couldn‘t process sounds, articulate sounds, and that may have led to some of… his… delays. (Appendix E, pp. E2-E3) This statement suggested that she viewed her son‘s struggles from a medical angle, likely based upon the opinions of others (―...it‘s theorized that...‖), with P‘s early hearing problem caused by fluid in his ears that resulted in a problematic otitis media as the starting point. And she went on to explain: there‘s no real one… thing but that seems to be the possible…reason is because he couldn‘t hear correctly. If you can‘t hear correctly, you can‘t speak correctly. If you‘re not speaking, you‘re gonna have not developed those… synapses to… understand. (Appendix E, p. E3) She then described the extensive nature of P‘s delays in articulation and language development: ―He‘s always had about an 18-month speech delay. So, when he was 2, he still wasn‘t talking. So figure by the time he was 4, he was still 18 months delayed in his speech‖. And she reaffirmed that those delays were ―significant‖ (Appendix E, pp. E4-E5). It should be noted that she was using another in-vivo code here; by ‗speech‘ she generally meant oral language, as she affirmed upon inquiry (Appendix E, p. E5). P did have articulation problems as well, and it is noteworthy that those subsided, according to his mother, around the same time his reading tutoring services were discontinued (first or second

140

grade). Recall that ‗reading‘ is equated with ‗(accurately) reading out loud‘ both by his mother and his school; the fact that reading-out-loud intervention would cease together with articulatory problems is coherent with this definition. There is another medical aspect mentioned by P‘s mother; his ADD and the relief provided by the medication as illustrated by her account of homework struggles. Prior to taking the medicine, she said, the family would spend ―up to 3 to 4 hours on homework (…) at times where it should have been less than an hour‖, but with the medicine, ―homework (…) didn‘t take more than… an hour and he did it on his own‖ (Appendix E, pp. E15-E16). And she raised an interesting question: ―was his lack of comprehension caused—in part— due to his ADD? Or was it the other way round‖ (Appendix E, p. E16). From a perspective that views learning as an active process, it would certainly make sense to link failure to comprehend to attention issues—if there is nothing comprehensible to attend to, why focus? Conversely, a persistent attention problem might be a causal factor in the etiology of poor comprehension, if the learner is unable to focus on a text or construct meaning from it, comprehension will likely be hampered. Within the confines of this study, the causal direction of the influence could not be determined, but it is illustrative to note P‘s mother‘s take on the issue: she pointed out that despite the success of his medication, P ―still has comprehension issues‖ (Appendix E, p. E17). Thus, if there is a causal connection between the two, it would appear to run from poor comprehension to ADD, not the reverse. His mother also suggested that P had comprehension problems from very early on: ―I knew he had comprehension issues before he was (even) able to read‖ (Appendix E, p. E4), but qualified this remark later, saying ―I don‘t know. I don‘t know‖ (p. E5) and adding: ―He was too young. He couldn‘t communicate‖ (p. E5). I tried to get more information about

141

those early reading experiences from her, but P‘s mother said she could not remember much of them (Appendix E, pp. E6-E7). We can only speculate what kind of behaviors would lead to discerning comprehension problems in a preverbal child; lack of attention and interest might play a role. But this raised another question: What is the social and experiential world of a child who is treated as someone with a comprehension problem from such an early age? To gauge comprehension, we have to rely on outward behaviors of some kind, as the phenomenon itself cannot be observed directly (see the discussion of comprehension as a quale in Chapter Two; also Memo Eight, Appendix B, pp. B10-B12). Many (if not all) of such behaviors will be oral-verbal; for a child with expressive language problems, then, caregivers‘ judgment of comprehension or its absence will be influenced by the child‘s difficulties in producing expected behaviors, and may impact interaction. For example, Bruner (1986, pp. 70-78) describes mothers‘ linguistic scaffolding of their children‘s abilities during shared book reading. His work demonstrates that the mother adjusts her talk to the child‘s productions, constantly raising the bar as to what she deems an acceptable answer, which helps the child master more and more complex language forms. What does a mother do, then, with a child who does not respond with productions of increasing complexity, who does not even seem to understand the interaction? Cazden (1988) suggested that teacher‘s perceptions of students‘ expressive behaviors may create a self-fulfilling prophecy; where teachers underestimate students‘ abilities based on the ways they talk, students may come to perceive themselves as low achievers and act accordingly, leading to actual underachievement (see also McDermott & Varenne, 1995). In a case of a child who is perceived as a poor comprehender from infancy on, do perceptions of

142

the child by caregivers, peers, or educators shape their interactions with the child and possibly create an undesired feedback loop, where the child‘s self-image as someone who does not understand is constantly reinforced? It was the impossibility to address these questions that made me feel like I was getting only half the story of P‘s etiology. His social-interactional history remained in the dark. This lack of information becomes even more glaring in the light of his mother‘s remarks that she has always been an avid reader (Appendix E, p. E31). P did thus have a social model for literacy as well as opportunity to participate in reading as a sociolinguistic process (Damico, Nelson, & Bryan, 2005), and yet, this has not translated in becoming a proficient comprehender. And there is another aspect to the ‗missing half‘ of his story, namely, P‘s lived experience. As seen in Chapter Two, many theorists assign a crucial role to motivation in reading; there needs to be willingness to interact with a text if comprehension is to occur. In this light, then, what happens to such motivation if failure to comprehend is the prevalent experience from early on? Theme Four: ISSUES EXPRESSING HIMSELF. This theme includes the subordinate themes of Boredom and Aggravation from the interview with P. It has a special status in that it functions as a substitute, as it were, for what would ideally be the core of my data – P‘s own account of his lived experience as a reader. But as we will see, P‘s expressive difficulties are so pronounced that they got in the way of garnering such information. Consequently, other data were employed to try to get an insight into what might be his own experience of his development as a reader. In this section, I will discuss the expressive issues that appear to get in his way instead.

143

P‘s mother made a first statement related to this theme when talking about his early comprehension difficulties: ―He couldn‘t communicate‖ (mentioned above). She went on to explain that P has expressive difficulties up until the present moment, and that he is aware of them. As she put it: I caught him (…) trying to explain something and he… it‘s there. It‘s just not—he has a hard time making it click. And he actually said, ‗come on, [P], just spit it out. Say what you have to say‘‖ (Appendix E, p. E10) Later, his mother touched on the difficulties he encounters when retelling a story or event: ―that‘s how it is, is that you when you retell something or you explain yourself to somebody, you have to—it‘s certain things have to be present for somebody to understand what you‘re saying‖ (Appendix E, p. E14). It was not entirely clear whether she was referring to format of talk or to failure to take listener perspective into account. Either way, it would seem that in addition to word-finding issues, P‘s expressive problems are also characterized by pragmatic difficulties. And these extend beyond language proper to impact peer interaction, as his mother observed: ―Sometimes kids get frustrated with him because he has a hard time expressing what his… needs are‖ (Appendix E, p. E30). It has long been known that children with language disorders have social-pragmatic issues (Fujiki & Brinton, 2004; Snyder, 1976). This expands the question asked above: What is the lived experience of someone faced with constant failure to comprehend and be understood? When I interviewed P himself, I witnessed his expressive issues firsthand. He seemed to struggle with any question regarding his experience of reading, and resorted to using the wording of my questions in his replies. Consider the following exchange (R = researcher):

144

R: (…) So um… How do you feel about reading in general? P: Um... (pause) (...) I... uh... R: See, me, for example, I like reading. It‘s something that I do when, when I get home, (…) a long day, I‘m tired, I just lounge on my couch like this, (put a … book) or journal or something. Do you ever do something like that? P: Well… I read a book on the (couch). (pause) (Appendix E, p. E35) Subsequent interchanges followed the same pattern. P reported not being ‗aggravated‘ with reading; reading to ‗relax‘ and ‗liking‘ to read; all these expressions came out of my attempts to get him talking. When he did not have these linguistic models, he was barely able to provide coherent accounts, as in the following examples: R: What other books did you read? P: Uh, um, er, (it has) a… title… R: Come again? P: Um, it has a, the book has a title, and… a… author, who, um, does the book, and… (pause) I forgot what I was saying. (Appendix E, p. E36) R: Do you like reading at school? P: Yes:: (pause) Uh… R: (…) Go ahead. P: I like reading (pause) every day, I like reading in, in that class— R: Uh-huh.

145

P: —(too). On my desk. A::nd I read a lot of books. On my desk. Every kind of book. (Title.) A:nd that‘s it. (Appendix E, p. E40) It also bears mention that in one instance, his expressive issue extended into the grammatical aspect of language. When asked if he had seen the Harry Potter movies, he replied: ―Well, I (didn‘t) at at least see all of them… I just saw saw some of them‖ (Appendix E, p. E42, emphasis mine). To reiterate, due to his expressive issues I was not able to glean any direct insight into P‘s lived experience regarding reading from the interview. However, I did get an indirect impression on this topic, which I will discuss over the subsequent themes. Theme Five: (LACK OF) MOTIVATION & NEED FOR SCAFFOLDING/STRUCTURE. This theme brings together the subordinate themes labeled Motivation from both interviews, as well as Need for scaffolding/structure and Interests from the interview with his mother. The first time motivation was mentioned by P‘s mother was in the context of discussing what she called his ‗need for structure‘. When I asked her to clarify, she explained: If he‘s given an assignment, and he didn‘t understand (…) what the lecture was in class, (…) he‘s just staring at it, that‘s what I meant. (…) somebody needs to, you know, ―so do you have a question about what‘s go—‖, you know(…)—he has a hard time with that. He has a hard time being a self-starter when it‘s something like that. (Appendix E, p. E21) The in-vivo code ‗self-starter‘, which she used several times, appears to encode independent tackling of tasks or activities; ‗hard time at being a self-starter‘, then, could

146

translate into ‗being passive unless prompted‘. And she explained that this trait is contextbound: Especially in school (…) it‘s different when it‘s something he‘s interested in, obviously, because (he will) be more attentive to something that he‘s interested in like (…) things that he does around here. He‘s a self-starter around here, when it‘s something he is motivated to do. But when it‘s schoolwork, and it‘s a… new subject that‘s coming up, you know, that… (pause) that is what I meant by that. (Appendix E, p. E22) In other words, P has difficulties in tackling tasks on his own when they are new, when he has problems understanding them, and especially when he is not interested in them. In and by itself, this is hardly atypical for children his age. But in the context of his comprehension problems, these difficulties become exacerbated to the point of interfering with academic tasks. His mother makes it clear that she feels P lacks some sort of internal ‗structure‘ that would enable him to bring his knowledge to a task and make sense of it. The following example (my hearer signals omitted) provides additional insight in what she meant by that. First, she outlines how his ability to focus has improved due to his ADD medication. And yet, she explains, he still needs someone to direct him towards engagement with task demands: For a 9-year old, he just doesn‘t have a lot of—of the structure (…) Even after he was put on his medicine (…)—he was better at paying attention and following along in class but he still needs someone to say ―ok so do you understand?‖ (…) It‘s still hard for him to do that in class because… he just has a hard time… grasping everything all at once. He (…) does need (…) somebody to… help him (…), ―k did you get this?

147

Are you sure that you got (…)‖, you know, just somebody just to come up behind him and (…) support him. (Appendix E, p. E20) Based upon this and other statements, it appears that his mother believes that P‘s improved ability to focus, provided by his medication, is not enough for him to tackle tasks independently in class. In addition to his lack of motivation, he apparently gets easily overwhelmed by the amount of stimuli in the classroom, and possibly by time constraints (―grasping it all at once‖). Additional evidence for this comes from his mother‘s report that he does his homework independently (see above, also Appendix E, p. E19, p. E29). Thus, working at his own pace in a quiet environment appears to help him cope. That is, his mother is suggesting that P needs mediation or assistance similar to classic Vygotskian scaffolding to overcome his problems in class; someone who provides him with more advanced mental structures than his own (Bruner, 1986, pp. 70-78; Geekie, Cambourne, and Fitzsimmons, 1999, pp. 1-26; Vygotsky, 1978). It is noteworthy that P‘s mother explicitly contrasted his passivity at school with his behavior at home: ―He‘s a self-starter around here, when it‘s something he is motivated to do‖ (Appendix E, pp. E21-E22). What is it, then, that sparks P‘s interest? We might assume that it is not reading. However, this is not entirely true. First, it is interesting that while most reading does not motivate P, he does not seem to harbor specific aversions against it either, despite his continuous experience of failure. As his mother put it, ―I don‘t think he… dislikes it. (…) just don‘t think it‘s something that (…) he‘s gonna do is grab a book and, you know…‖ (Appendix E, p. E30). This is not surprising; the purpose of reading for oneself is to construct meaning from the written material. Absent the ability to do

148

so, individual reading becomes pointless. This observation fits well with P‘s own comments. Prior to the interview, he had told me that he gets bored with books; when asked to specify, he stated he usually reads only five pages in a book and then gets bored (Appendix E, p. E33). Boredom would seem to be a natural reaction to an activity that is, literally, meaningless. When analyzed, P‘s boredom and lack of motivation seem to be centered mainly on storybooks. In sharp contrast to what we have said so far, his mother reported that he does show interest in expository print; he likes topics like ‗sharks‘ or the ‗Titanic‘ so much that he researches them on the internet on occasion (Appendix E, p. E25). I was not able to find out with certainty whether she was talking about independent research or about a joint activity with his father. Either way, the fact remains that P does show interest in some written materials, depending on their topic. P‘s other interests are summed up quickly (cf. Appendix E, pp. E22-E24, E26-E27). He likes to play with remote controlled cars and video games, drawing and coloring, listening to music, watching dance videos, and he enjoys outdoor activities such as riding his bike, playing on the swings, or playing with the family dogs. As for sports, he enjoys individual activities such as tennis or swimming, but not team sports (a theme we will explore in more detail below). It is noteworthy that he seems to have found nonverbal ways to express himself, such as drawing and music (his mother mentioned that he has voiced interest in learning the guitar). Overall, the emerging picture is not one of a child that is generally passive. When tackling an activity that is meaningful to him, P clearly shows initiative and motivation. This applies even to reading-related activities such as learning about information of interest.

149

Given his record of ongoing comprehension failure and lack of appropriate scaffolding, P‘s lack of motivation and interest in most reading is not surprising. Further, the next discussed theme suggests that there is another obstacle to comprehension that emerged from the interviews – P‘s tendency to abide by others‘ expectations. Theme Six: OTHER-ORIENTATION. This theme amalgamates the following subordinate themes from the interviews: Coping mechanisms, Role of teacher, Group sports issues, Lack of peers, Passivity, and People pleaser from the interview with P‘s mother, as well as Other-orientation and Spelling = copying words from the interview with P himself. Note that the comparably low frequency and late appearance of its component themes within the course of the interviews may conceal its importance. However, this theme appears to be influential and it appears to play a pivotal role in the pedagogical situation within which P exists (more on this issue later). Unlike some of the other subordinate themes, one of the components of other-orientation, P as a ‗people pleaser‘, emerged from the very first bit of data collected regarding P. My field notes taken after the consent/assent procedure read: ―he didn‘t say much, just nodded & assented to everything. Holly (Dr. Damico) later described him as a ‗pleasant kid‘‖. Similarly, the notes from the initial assessment start with ―P was pleasant & helpful as before‖, and this is reiterated in the field notes from the first eye tracking session: ―P was as pleasant as usually‖ (all Appendix B, pp. B1-B2). This observation also made it into the first memo (Appendix B, p. B3) and was expanded, in the second memo, to the overarching label ‗other-orientation‘ with the tentative implication ―not much own drive to understand stuff for the sake of understanding?‖ (Appendix B, p. B4). This subordinate theme became the central focus of memo Five, which was written while I was coding the interviews (more on this below).

150

Another subordinate theme, ‗Role of teacher‘, links the previous theme with the current one. As mentioned previously, P‘s mother was trying to get an in-class aide to scaffold P‘s comprehension, a role that in her view cannot be fulfilled by a teacher who has to be there for 30 students simultaneously (Appendix E, p. E18). Rather, his mother believed that P needs full-time, on-the-spot guidance despite the fact that he receives comprehension intervention several times a week, by two different therapists who both work on his ability to think and express himself independently. I could not help wondering if there was more at play here than just a persistent language problem. Even in the presence of such a problem, a reader typically attempts independent processing and interpretation of visual texts at some point, especially when he is aware of his issues. P was not doing this. Why did he have such a need for prompting? The answer, which appears to be winding as a partly-hidden thread through my observations and through his mother‘s discourse, is that he has developed a habit of orienting his actions towards others‘ expectations, and an expectation to be led by others, in social and academic situations. His mother, as an illustration, described his peer relations as follows (my hearer signals omitted): He‘s at a disadvantage (…) with his (…) comprehension and his maturity level. He still doesn‘t let that stop him from [joining peers in play] but (…) kids do tend to… get the upper hand on him because he is a, you know, naïve in that sense. (…) Cause he just wants to please people. (Appendix E, p. E28) Later in the interview, when discussing other children‘s frustration with P‘s expressive problems (see above), she added that because of these issues, ―he tends to let

151

people… (…) to let other kids take the (…) role as leader‖ (Appendix E, p. E30). As mentioned above, I developed an impression of the importance of this behavior of his during my own interview with him, where he resorted to simply reiterating my wording instead of coming up with an answer on his own. It is worth noting that due to the absence of children in the neighborhood and the family‘s busy schedule there is a relative lack of peers in P‘s life (Appendix E, pp. E27-E28). In addition, P‘s mother remarked that due to his non-competitive nature, team sports have not ‗worked out‘ for him, leaving him with solitary sports like swimming or playing tennis with his father (Appendix E, p. E27). What is true for P‘s social behavior also seems true for his academic behavior, notably his reading habits. This is evidenced not only by his mother‘s talk, but also by his own. When asked why people read books, his answer was: ―Um, because, they read books… when… uh… reading starts? Uh… read, uh… when reading starts, you read a books, a lot of books‖ (Appendix E, pp. E36-E37). The construal of reading in this utterance is that reading, almost like a force of nature, is something that ‗starts‘ by itself, not something in which you actively engage. P appears to indicate that the reason people read books is because that is what you do ‗when reading starts‘; in his construal of reading, there is no point to reading other than reacting to an external task demand. In support, when asked what the teacher wants him to do when reading, he replied: they want me to rea:d… some… other kinds of books that I read in class (with)—it‘s a group and—a reading group and I read‘em and then um… I read um one book… a::nd… (pause) read another one… ‗nother one and uh that‘s it. (Appendix E, p. E43)

152

Consider also the following remark: ―I read (to) my teacher… (and)… that‘s all I do‖ (Appendix E, p. E46). This attitude results in rather cursory memory for the content of the read material. When asked what he reads, P replied ―snakes—snake books (…) a:nd (pause) sea books. (…) A:nd some houses. A house book‖ (Appendix E, pp. E43-E44, 6 turns at talk omitted). When asked directly what the point of reading at school is, his answer was: ―Um (pause) uh (pause) hm hh uh uh: (pause) I forgot what I was to say but I don‘t remember…‖ (Appendix E, p. E47). Based upon these statements and the other data within the interviews, the only goal of reading at school that P was able to talk about was the requirement to ‗get your words right‘, as seen in Theme One above. This requirement may be the reason that he says he is ‗not bored‘ with reading at school; at least there is a point in reading to the teacher, namely, getting rewarded for verbal accuracy. In sum, then, a major obstacle for his comprehension development would seem to be his orientation towards others‘ expectations. As a result, he only tackles comprehension when explicitly guided to do so. Theme Seven: TANGIBILITY. This theme subsumes Tangibility, Comprehension intervention, and Positive affect vs. tangibility from P‘s mother‘s talk. It is reminiscent of an earlier theme, Reading vs. comprehending, in that it represents a juxtaposition of behavioral phenomena with cognitive ones. From a transactional viewpoint, those are simply aspects of an overall process, but in P‘s environment of they are treated as though they were distinct. The first time P‘s mother mentioned ‗tangibility‘ was in the context of her talk about the school‘s lack of support. She talked about P being ‗kicked out‘ of the tutoring program: ―They‘re focusing on this [child with a decoding problem] because it‘s something tangible and not on him because it‘s something that—they‘re just not concerned about‖ (Appendix E,

153

p. E8). When I asked what got him into reading tutoring in the first place, she referred to his inability to read out loud, and specified that, ―they‘re expected to… be able to… meet certain benchmarks. You know, sight words, ‗I‘, ‗was‘, ‗the‘‖ (Appendix E, pp. E8-E9, my hearer signal omitted). The term ‗benchmarks‘ showed up again in her discourse on P‘s current intervention; when comparing his private to his school SLP, she explained that the school‘s approach is more um tangible well they have to… show… so there‘s certain things that they work towards that are… you know, he needs to meet benchmarks and percentages and things like that. So I think there‘s there‘s more structure there. So which we do the two different things. (Appendix E, pp. E14-E15, my hearer signal omitted) The in-vivo code ‗tangible‘ clearly referred to countable or measurable behaviors; that is, recordable and demonstrable abilities such as oral reading accuracy, reading of sight words, or answering ‗main idea‘ questions (cf. the IEP goals quoted above). This is the type of outcome the school is interested in, not the comparably vague notion of ‗comprehending‘ a text. P‘s mother herself appeared to have difficulty conceptualizing less ‗tangible‘ aspects of reading, as can be seen in her talk about the school SLP and the private SLP. At first, she stated that both do ―the same thing‖, namely, ―a lot of retelling to get him to… read something and then retell it with the coaching and the leading (…) without the leading questions but encouraging him to think… on his own‖ (Appendix E, p. E13). But then, she added that the private SLP encourages him to write. (…) I don‘t think they do that at school. [Private SLP] (…) she does the Venn diagrams and she encourages him to think out loud (…)—she

154

encourages him to do that ‗cause she feels that doing that will help him with his understanding. (Appendix E, p. E14) Finally, towards the end of her discourse on this topic, she came to a conclusion quite opposite to the initial one, explaining that the school SLP doesn‘t follow the same approach as [private SLP] does. [Private SLP] is very different with the way she does it and she does wonderful with him and he loves going and… he seems to do really well with her. I think the school‘s approach is more um tangible. (Appendix E, p. E14, my turn omitted) That is, while both SLPs are working on comprehension, one does so in a tangible (measurable) format working towards benchmarks such as correct answers to main idea questions, whereas the other focuses on scaffolding meaning construction. It also seems clear based on his mother‘s comments that P responds positively to the less tangible format; however, there is no data on his attitude towards the more tangible school intervention. To sum up, then, P‘s mother‘s talk establishes a direct connection between his school‘s interest in ‗tangible‘ results of therapy and P‘s trajectory as a reader. First, his inability to meet tangible benchmarks got him into a tutoring program for reading out loud; once he was able to meet benchmarks, this intervention was ended. While he has received intervention for most of his school career (Appendix D, p. D1), there is an implicit plaint in his mother‘s talk, seen in the (LACK OF) SUPPORT theme, that he is not getting the type of intervention he needs. Instead of providing him with in-class support to master academic tasks, the school provides him with tangible (measurable) intervention in a traditional pull-

155

out format. In the light of his stark orientation towards others‘ expectations, this begs the question of therapeutic efficacy. Therapy that is geared towards specific tasks such as ‗listening/responding to various text dependent questions‘ (as in his IEP above) may not prove helpful when trying to foster independent tackling of unfamiliar texts. Theme Eight: EMERGING AWARENESS. This theme combines Growing awareness of comprehension and Increasing confidence from the interview with P‘s mother together with Awareness issues from the interview with P. It was prompted by a mismatch between accounts of his awareness of his comprehension issues. On the one hand, his mother noted a positive development in P‘s ability to monitor his linguistic issues (expressive and receptive). Recall her description of his expressive problems, quoted above: ―he actually said, ‗come on, [P], just spit it out. Say what you have to say‘‖ (Appendix E, p. E10). She then elaborated: He knows it now. He‘s aware of it. Before, he was but I don‘t think he really understood the significance (…) but now that he‘s older (…)—he‘s completely aware of it now. And he asks questions. Before, he didn‘t. He asks questions now. He wants to know. (Appendix E, p. E10, hearer signal omitted) Thus, while P‘s comprehension problems are still present, his mother does note that he has become aware of not understanding or being able to express himself, and he has started to ask questions if comprehension fails him. This is accompanied by an increase in confidence in academic tasks (and possibly social interaction—his mother‘s discourse was not completely clear in this regard) due to the successful treatment for ADD (Appendix E, pp. E19, E28-E29). However, P‘s retells of his readings evoke a quite different picture. As we will see below, his retells convey no meaning at all, and even when given the chance to

156

go back to the text and improve his understanding, he typically fails to do so. And yet, when asked in the interview if he usually understands what he reads, he replied ―yeah-h‖ (Appendix E, p. E45) and reaffirmed later that he is ―a good reader‖ (p. 47)—which, as mentioned above, is true only when comprehension is taken out of the equation for reading. It should be noted that I did not ‗probe‘ P during his retells, as I wanted to capture his performance when asked to self-initiate interaction with print. Additionally, at that point, I did not know that P is mostly passive unless prompted. Thus, his retells during this study may be indicative of his need for scaffolding rather than a lack of metacognitive awareness. I do not have enough data to draw definite conclusions on this theme, but whatever growth of awareness his mother may see in him, one thing is certain; he does not seem to be aware that his retells do not usually result in a comprehensible rendering of what was read. Connotative theme: UNCERTAINTY. As mentioned above, the ninth theme formulated from the interviews is less directly grounded in the data than the previous ones. Instead, it arose from a general intuition of mine rather than from specific participant discourse. The fact that the nature of P‘s issues seemed to be poorly understood by his social and academic environment, the apparent failure of his school to cater to his needs, and his own lack of awareness of his comprehension issues, all of these congealed into a general sense of uncertainty that I expressed in Memo Three, written during line-by-line coding of the data, and which I quote at length here to convey the ‗flavor‘ of my intuition: There is something surrounding P‘s lack of comprehension—how to call it? Lack of awareness? Ungraspable nature? What I mean is that no one involved seems to have a good grip on the phenomenon. P doesn‘t realize he‘s not understanding. The school doesn‘t care as long as he reaches tangible benchmarks. Mom sees what‘s going on—

157

or does she? ―Isn‘t decoding and comprehension the same?‖ she asks me at one point in the interview. And there‘s the (related?) issue of his expressive language problem. Can‘t express himself ↔ people can‘t gauge whether or not he‘s comprehending. Also, if he can‘t express himself TO HIMSELF, how can he understand?  reciprocity of others understanding him and he understanding himself. (The idea of scaffolding: Understanding has to be provided, at least to some extent.) Addition on 8/10/2012: P‘s response to my question about understanding—he started talking about getting words wrong… (Appendix B, p. B5) Thinking about what might have prompted this intuition, I realized there was a common thread in the interviews, an apparent helplessness in all stakeholders, including P himself, to give P what he would need to improve his comprehension. As the quote above suggests, P himself does not know what he needs—he has just begun to show indications of realizing that he has comprehension troubles. His mother seems to have a partial grasp on his needs—hence her fight with the school over in-class support—but her puzzlement regarding the intricacies of the reading process became clear when she asked me if decoding and comprehending were the same. The school SLP, and with her the institution as a whole, tackles aspects of comprehension but appears to be missing the big picture, notably P‘s otherorientation and his failure to motivate himself to engage in meaning construction. The private SLP‘s approach appears beneficial insofar as it stimulates positive affect towards reading, but so far has failed to generalize to the academic environment. Within these circumstances, not only do P‘s needs go unmet, he is unable to express them; as expressed in the previous

158

memo, he cannot figure out for himself what he needs to understand better. No wonder, then, that he resorts to doing exactly what others want him to do, and no more. So far, this is the one strategy that has appeared to be moderately successful for him. Home Reading: Behavioral and Miscue Analysis In Chapter Three, it was discussed that ‗typical‘ Grounded Theory uses mainly interviews, processed in line-by-line coding. By contrast, for the data in this section I used incident-by-incident coding, the ‗close cousin‘ of the line-by-line method. Due to the nature of the data, not all of the themes discussed here are gleaned from physical codes; this is particularly true for those parts of behavioral analysis that were done on video data only, namely, P‘s pre-reading behaviors and behaviors during transition from text one to text two. Since video and miscue data did not lend themselves to be grouped into themes directly; the themes emerged during the discussion of the findings. Consequently, a different presentation format will be employed in this section. Instead of organizing these results by thematic groupings, the findings will be discussed in such a way that the eventual groupings will emerge via the process of empirical and theoretical explication. These emergent themes will then be linked back to the previously established themes, during or at the end of the discussion. Home readings one & two: Behavioral analysis of pre-reading/between-reading behaviors. As mentioned, P did not appear to be comfortable during the interview; however, as the field notes report, he ―cheered up quite a bit when I brought out the ‗technology‘ (camera & HQ voice recorder) to tape his reading‖ (Appendix B, p. B2). Also, the field notes report a revealing instance of trouble:

159

Some awkwardness arose around the selection of the ‗comfortable spot‘ to read: Mom didn‘t want me in P‘s room and had chosen the couch area in the living room for this portion. P, by contrast, selected his own bed as the most comfortable spot to read and when he learned he couldn‘t do that, he rejected the couch area and chose to remain at the table instead. Also, the texts weren‘t self-chosen but selected by Mom, however, he had read them beforehand. (Appendix B, p. B2) This brief event serves as a reminder that there are, in the life of a child, expectations and limitations that the child needs to accept; at the same time, P does have options other than passive acceptance in his behavioral repertoire. This applied to his interaction with me as well. As mentioned, there were several books but P assented to read from only two of those, and it took some coaxing to get him to comply with that task. Characteristics of the texts that were read are presented in Table 4.3 below.

Text #

Author, Title

Genre

AR book level

1

Jasmine Jones, Toy Story 3

Storybook

n/a

2

Anne Marie Welsh & Ryan Hobson, Heroes of the Titanic

Expository text

6.6

Table 4.3: Stimuli for P Home Reading My analysis of P‘s behaviors before the home readings notes ―a lot of bantering and excitement about the equipment before reading starts‖ and poses the question: ―is it actual excitement, or avoidance behavior to postpone reading?‖ (Appendix F, p. F1). It took a full 1.5 minutes, and five attempts, to start the reading (cf. ibid.). At my first attempt, P was

160

leaning in to the microphone across the table, with no way of seeing the book he was expected to read; when I asked him if he was ready to read, he responded with a drawn-out ‗yes‘ but remained in the same position, with no indication that he was actually going to begin. When I asked him to sit back, he interrupted me with a question on how to ‗restart the microphone‘. At my second ‗you‘re ready? he leaned back while talking and greeted his mother who had entered the room. The third time he confirmed ‗ready!‘ but then attempted to reposition the microphone. The fourth prompt was greeted with an excited ‗boop boop boop‘, leaning over to the microphone again. At this point, I may have looked a little stern, for his excited grin faded while he was looking at me. This mood shift gave me the chance for a fifth attempt, during which I was finally able to explain my expectations: ―I want you to read a little bit, and then I‘m gonna ask you to retell what you read to me. ok?‖. Upon which he asked: ―uh…can we just hear it on the… thing?‖ (Appendix F, p. F2). That is, he asked if we could just replay his reading instead of having him retell the story. I answered that we could do that after his retell, upon which he began reading. On the analysis transcript, I coded these 1.5 minutes as ‗avoidance behavior‘, reflecting the observation that poor readers often resort to all kinds of distracting actions when asked to read (Damico et al., 2008). Looking back, I am no longer sure if this is the only way of interpreting P‘s motivation. His excitement about the ‗technology‘ (camera and microphone) seemed genuine, which in itself may have been reason enough not to focus on reading. Additionally, my memos report a marked difference in P‘s behavior while reading for assessment as compared to his home reading. In Memo One, I wrote:

161

there was a clear difference between the way he read during testing vs. at home. Very careful & deliberate in the former situation, rather ‗sloppy‘ (let‘s say comfortably) [as] in the latter one. So he does seem to adjust his reading according to situation. (Appendix B, p. B3) And in Memo Two, I noted: I would like to know what happens in different situations when he‘s asked to read. The differences between the GORT and the home reading—astounding. And of course, the GORT comes with all the formality of standardized assessments (―read it as fast as you can and as well as you can‖). The home reading came with fun. What happens in other situations when the demand for reading is placed in different ways? (Appendix B, p. B4) I was initially using the GORT-4 for assessment, a test that requires participants to read out loud, much as in the data collection for miscue analysis. What these observations seem to imply, then, is a heightened sensitivity, on P‘s part, to situational demands. In the rather formal assessment situation, he was focused and deliberate; in the informal, ‗comfortable‘ home reading situation he was relaxed, including bantering and distraction. Avoidance may certainly have played a role (most notably expressed in his suggestion to simply replay the recording instead of doing a retell), but this observation would be incomplete without the complementary one that P has an intuitive understanding of the ‗seriousness‘ of different reading situations, and adjusts his behavior in accordance. In consideration of this interpretation, it is interesting to note the brief interlude between home readings, where I had to coax him into reading some more. I give the

162

interchange in its entirety here to provide the full flavor of the situation (R = researcher; for transcription conventions cf. Appendix E, pp. E48-E49): R: so, would you be up for reading another one? P: um… I‘m good… (pause) not really s um my cars and my toy sword R: um, if I ask really nicely? P: uh R: maybe the Titanic one? P: (pause) sure (grabs book) let‘s see (buh-t-t) R: (coming around the table) let‘s see let let‘s let‘s let‘s see uh how much you wanna read. How much would you wanna read of that? P: (reading title) ―Heroes of the Ti[tanic]‖

[Titanic]

―[Heroes] of the Titanic‖ [(…)] that (reading

R:

page headings) ―Titanic tragedy‖, ―Prologue‖, so= =(…) the picture

P: R: the picture is pretty… scary huh?

P: nah it‘s not (horror), just… action or… (looking down, smile fading, ‗tense‘ voice quality) comedy… R: it‘s what? P: (looking down) nothing (smiles) R: ok so where do you wanna start? P: hm, let‘s start when they hit the iceberg (flipping through pages) (where) is it? (…)

163

R: (after some flipping) I think we‘re already past the iceberg. ‗cause I saw something about a guy jumping in the water P: hm? R: (flipping back) right here P: where? R: ―Jack leaps for his life.‖ It sounds like the iceberg had already struck at that point (flipping back) ―Jack Thayer.‖(flipping back to correct page) P: (looking at the page without showing recognizing behavior) R: I think… (pointing to page) oh here it says ―Iceberg reports‖. P: (looking over, getting up and looking towards where R pointed) (…) R: what do you want P:

(…) (sitting back down again, grabbing book ) let‘s read it

This interaction contains, in a nutshell, all that has been said about P as a reader so far. At first, he showed a lack of motivation and expressed a preference for less demanding activities. But he was easily swayed, corroborating his tendencies of people pleasing and other-orientation discussed above. He then showed some actual interest in the printed material, pointing the picture of the sinking ship out to me—but in the next few seconds, his interest was dampened by his inability to share it verbally. When I acknowledged his move, saying the picture is ―pretty scary‖, he replied: ―nah it‘s not (horror), just… action or… (looking down, smile fading, ‗tense‘ voice quality) comedy…‖ Not having understood him, I asked: ―It‘s what?‖, whereupon he looked down and averted further inquiry with ―nothing‖. For a brief instant, P and I were having a literacy-related conversation, mutually engaged in a meaningful sociolinguistic process (Damico, Nelson, & Bryan, 2005). I

164

expressed a feeling about a picture in the book, and in return he explained to me that the book was not a ‗horror‘ story, but about action. And the next moment, this conversation ended when he added the book was also a ‗comedy‘ (possibly an association prompted by my reading of ‗tragedy‘). I assume that he realized the inappropriateness of this term while uttering it, as evidenced by his dropping head, fading smile, and somewhat ‗uneasy‘ tone of voice, and also by his reaction to my inquiry. His ―nothing‖ ended the conversation as effectively as if he had walked out of the room closing the door, and the brief moment of engagement was over. While analyzing this interaction, I wondered how often P may have experienced a similar interchange with a teacher or peer; not being able to express himself appropriately; being misunderstood by the other person; and ending the conversation to avoid further inquiry. No wonder he would resort to orienting towards others, instead of making his own ‗voice‘ known. In sum, behavioral analysis of pre- and between-reading behaviors corroborated some of the themes gleaned from the interviews, namely, OTHER-ORIENTATION, LACK OF MOTIVATION,

and ISSUES EXPRESSING HIMSELF. In addition, the analysis showed additional

facets of these themes: an acute sense of the ‗seriousness‘ in a reading situation, which squares well with OTHER-ORIENTATION, and a tendency to show avoidance behaviors, which would seem to be related to LACK OF MOTIVATION. It is interesting to note that the between-reading exchange seen above foreshadowed some of P‘s weaknesses in text usage. He clearly knew the Titanic book and the sequence of events, as evidenced by his announcement that he wanted to begin reading ―when they hit the iceberg‖ and his attempt to look for this page in the book. However, he was unsuccessful in

165

finding it; he flipped right past the proper spot without noticing, and recognized the page only after I pointed it out to him. (It is also possible that he did not even recognize it but just followed my lead.) Miscue analysis home reading one. For both readings, miscue analyses proceeded based on the ‗in-depth procedure‘ detailed by Goodman, Watson, and Burke (2005, pp. 131160). Reading materials were copied and magnified to provide space for marking miscues. Observed responses (ORs) were marked in accord with the in-depth procedure. However, for this investigation these analyses went beyond the standard miscue procedure. Every deviation from the expected response (ER) was marked including prosodic phenomena. Miscues were then transferred to the coding sheet provided by the authors and coded as per their instructions and they were grouped by joined patterns of meaning construction and grammatical relations. Any OR-ER deviation that did not qualify as a miscue was listed and coded separately (These miscue analyses are contained in Appendix G). Discussion of these miscue analyses, including examples for types of miscues and other ORs, will be presented followed by a brief discussion of P‘s reading strategies (predicting, confirming, correcting, initiating/sampling, and integrating, cf. Goodman, Watson, & Burke, 2005, pp. 192-195). Each part of the discussion corresponds with a type of ‗incident‘ and will be associated with a theme gleaned from it. For home reading one, P read a total of 481 words. More precisely, coding was discontinued after 481 words, as his reading became more and more ‗sloppy‘ after about 4:45 minutes into his reading. The analysis notes report a ―remarkable deterioration of his reading: a lot of omissions of entire words and unstressed syllables, sloppy [dialectal] pronunciation, text portions read with glottal fry‖ (Appendix F, p. F3) and conclude as ―overall impression:

166

he‘s just trying to get through [the text]‖ (Appendix F, p. F3). There are two possible interpretations for this phenomenon. It could either be due to P‘s general relaxation and low level of ‗seriousness‘ discussed above, or it could be indicative of fatigue. Either way, it was decided to exclude the last 3:30 minutes of his reading from analysis, as coding them would have resulted in a misleadingly high number of miscues. On the text read and analyzed, P produced 32 miscues, which equals 6.65 miscues per 100 words (MPHW). Due to the qualitative nature of miscue analysis, this number cannot be compared to a standardized norm, but it is illustrative to note that in the four cases discussed by Goodman, Watson, and Burke (2005), they found a range of MPHW from 1.9 to 12, placing P squarely in the observed range. When a more detailed analysis of the miscues is undertaken, the data show that of P‘s 32 miscues, 11 of them (34.375%) did not result in meaning loss; the remaining 21 miscues (65.625%) did result in partial (6/32) or complete (15/32) meaning loss. 17 of the miscues (53.125%) indicated grammatical weakness while the remaining 14 miscues (46.875%) indicated strength or partial strength in grammatical relations or were the result of a grammatical overcorrection. All of the 16 ORs (100%) which represented a substitution of an ER (rather than an omission, insertion, or shift in prosody) were highly similar both graphically and phonetically (14/16 and 15/16 respectively) or somewhat similar (2/16 and 1/16 respectively) to the ER. In sum, the miscue patterns for P are roughly consistent with a nonproficient reader profile (Goodman, Watson, & Burke, 2005, pp. 168-171). Table 4.4 sums up the pattern analysis for home reading one.

167

Pattern

%

%

Meaning construction No loss Partial loss Loss

34.375 18.75 46.875

65.625

Grammatical relations Strength Partial strength Overcorrection Weakness

31.25 12.5 3.125 53.125

46.875

OR-ER similarity Graphic similarity High Some None

87.5 12.5 0

Sound similarity High Some None

93.75 6.25 0

100

100

Table 4.4: Pattern Analysis of Miscues for P Home Reading One More importantly, these patterns indicate that P‘s primary concern while reading appears to be accuracy of grapheme-phoneme correspondence. As noted, 100% of his observed responses that were substitutions for his expected responses closely matched the targeted grapheme or phoneme that was missed, typically with an accompanying loss of the meaning of the text at that point. For example, none of his substitution miscues were visually or phonetically dissimilar to the ER and 12 out of the 16 miscues of this kind (75%) were nonwords that closely resembled the ER in looks and sound but not meaning. Examples are

168

$[tɹyɛt] for crept; $[wʊːld] for would; or $[ˈʌftˌsaɪd] for outside (Appendix G, pp. G1-G2). This insight yielded the first theme derived from the miscue analysis: FOCUS ON CODE. His environment and P himself may see him as a good reader, but viewed through the lens of miscue analysis, a different picture emerges. As noted above, both syntactic and semantic accuracy of P‘s reading lag behind the graphophonemic aspect. More than half of his miscues indicate weakness in grammatical relations, or, as Goodman, Watson, and Burke (2005) put it, they do not ‗sound like language‘ (cf. p. 189). In P‘s case, this descriptor phrase is particularly appropriate in that 8 out of his 17 syntactically weak miscues occurred on the level of prosody. The first of these happened at the very beginning of his reading (see Figure 4.1 below): in Ok, places, everyone, he produced the second word with falling intonation, effectively treating the nominal command places like a term of address (cf. ok, guys). While this is partially acceptable syntactically as per the in-depth procedure—the first part of the sentence sounds like actual language—it is semantically unacceptable and suggests that he is not processing the text for meaningfulness while reading out loud. Additionally he was not conscious of this since he did not attempt a correction. This miscue indicates weakness in the syntactic aspect of language use while reading as well as a lack of focus on meaning.

Figure 4.1: A Prosody-based Miscue Indicating Syntactic Weakness Another instance of this type of OR occurred on line 010, where P produced Rex clapped, following a stretch of direct speech by the characters in the text, in an intonation that would have been appropriate for Rex said in sentence-final position, whereas the actual

169

sentence continued: Rex clapped his tiny plastic tyrannosaurus hands. This miscue is shown in Figure 4.2 below.

Figure 4.2: Another Prosody-based Miscue Indicating Syntactic Weakness In both cases, P produced a sentence that, while the wording remained unchanged, did not make sense due to his uncorrected use of inappropriate prosody. He also produced syntactic miscues not based on prosody. In Figure 4.3, an omission of a pronoun is shown that renders the sentence ungrammatical, indicating, again, syntactic weakness as well as a lack of monitoring for meaning.

Figure 4.3: An Omission Indicating Syntactic Weakness P‘s lack of concern with what might be referred to as higher-level language is equally obvious in the area of semantics. Almost half of his miscues resulted in complete loss of meaning (46.8%) while another 18.75% exhibited partial meaning loss. Similar to the earlier analysis, the bulk of these meaning-deficient miscues fall into the same categories as the instances discussed above, they were either the result of a graphophonemic-based nonword, or of a syntactic alteration that impacted meaning. Notably, many of his miscues that were either partially or fully acceptable semantically are of the same type. That is, their acceptability or partial acceptability stemmed from the fact that they were either corrected, or

170

did not disrupt meaning completely. An example for the latter type can be found in line 004, where P inserted the into a sentence, rendering it partially acceptable (see Figure 4.4).

Figure 4.4: A Partially Acceptable Miscue (other miscue markings removed) In a second step of the miscue analysis, P‘s miscues were grouped by joint syntacticsemantic patterns. This type of analysis is not discussed by Goodman, Watson, and Burke (2005), but the authors do mention that there is a category of ―miscues that disrupt syntax and meaning‖ (p. 87), which they dub ‗low-quality miscues‘. This category appears to have good conceptual validity in that grouping of P‘s data revealed that all miscues that resulted in partial or full loss of meaning also indicated weakness or partial strength in grammatical relations. By contrast, all miscues that did not result in meaning loss indicated strength in grammatical relations, or, in one case, a grammatical overcorrection. When the data were organized by these syntactic-semantic patterns, it became clear that 21 out of P‘s 32 miscues (65.625%) were of low quality (i.e. they resulted in full or partial loss of meaning), and/or indicated partial strength or weakness in syntax. In other words, almost two-thirds of his miscues indicated problems in higher-level language, either in semantics or in grammatical relations. Consequently, the second theme gleaned from the data, therefore, can be termed HIGHER-LEVEL LANGUAGE ISSUES.

Table 4.5 sums up the results for grouping patterns of

semantic-syntactic alterations.

171

Pattern

%

Meaning loss, weakness in grammatical relations

37.5

Partial meaning loss, weakness in grammatical relations

15.625

% low qual.

65.625 Meaning loss, partial strength in grammatical relations

9.375

Partial meaning loss, partial strength in grammatical relations

3.125

No meaning loss, grammatical overcorrection

3.125

No meaning loss, strength in grammatical relations

31.25

Table 4.5: Home Reading One Miscues Grouped by Combined SemanticSyntactic Alterations Aside from his struggles with higher-level language, P also exhibited difficulties with articulation and prosody. He produced a large number of ORs which did not fall under any of the classifications for miscues in Goodman, Watson, and Burke (2005) but which were sufficiently different from the ER, in their phonetic and/or prosodic form, to warrant scrutiny. This observation seemed paradoxical at first, given the consensus about his decoding abilities. But it made sense in the context of his history as a language user. The ORs in question will be discussed and then explained in the light of P‘s trajectory. P produced dialectal forms such as substitutions of [t] with a glottal stop (e.g. [gɛɁ] for get in line 001); substitutions of [ð] with [d], as in [də] for the (e.g. lines 022 and 023); cluster reductions e.g. of best to [bɛs] (line 006), or coarticulations such as laminal [s] in the second [s] of space ranger (line 007). These ORs did not seem indicative of any particular reading problem. However, there was a continuum of these with what Goodman, Watson, and Burke (2005) term ‗misarticulations‘. They exhibit the following characteristics: they were either similar to dialectal processes, or indicative of developmentally earlier

172

phonological processes, or both; also, a few of these were non-English sounds. Taken together, this suggests articulatory difficulties that are absent from P‘s conversational speech. The most prevalent examples of pseudo-dialectal misarticulations were substitutions of [d] with a glottal stop, e.g. [sɛɁ ɪn] for said in in line 002, which is similar to dialectal [t  Ɂ] substitutions but is not considered a feature of dialect. These and other misarticulations of alveolar sounds indicated difficulties with alveolar placement, particularly in consonant clusters, or in environments with several alveolar consonants in rapid succession. The most telling indication for this was his omission of the past tense marker –ed, which occurred 8 times in the coded lines, and commonly after an alveolar or palatal verb-final consonant (e.g. shift for shifted, line 003, or punch for punched in line 026). Another interpretation of these ORs is that they represent colloquial speech. Productions such as [lʊkn] (looking, line 035) or [ˈɛ:mʌn] (anyone, line 046) certainly lend themselves to this conclusion. However, there were a number of productions that were counter-indicative of this explanation. For example, he produced lid as $[wɪd] (line 005), cordless as $[ko:ɹdəˈwɛs] (line 025), or creaked as $[kwik] (corrected) in line 034. Also, he produced box as $[bɑkɬ] (line 056), and the first syllable of heartbroken as $[xɑɹʔ] (line 058) and these types of productions suggest phonological processes atypical of P‘s age, as well as non-English sounds which suggest other explanations. Since all of these articulatory differences occurred only while reading (there were no indications of any articulation problems in his conversational speech), this raises the possibility that his struggles while reading are partly due to lingering articulatory difficulties, remnants of his earlier articulation problems, which according to all stakeholders were thought to have been resolved. No research appears to have been done on this phenomenon,

173

and hence my explanation is a tentative one. As previously discussed his mother stated that his reading (out loud) tutoring was ended around the same time that his articulation problems subsided and this suggests that his ability to read out loud had been hampered by these problems. It may be that close miscue analysis shows that he is still struggling with this aspect. Perhaps this indicates an additional cognitive load during reading due to the dual demand of translating graphic symbols to spoken sounds, and articulating those sounds. In this light, it would make sense that his comprehension should suffer; with all attention and energy directed towards the task of accurate reading performance, not many cognitive resources may remain to engage in construction of meaning. An alternate possibility may be that since he has been taught to focus on the graphophonemic code as a way to process visual text during reading out loud, often to the detriment of reading for meaning, his attempts to decode the visual graphemes into phonemes are guided by visual processing and/or visual analogy so that his phonemic productions do not always align with the phonological system he has developed for oral production of speech. In this case, the ―misarticulations‖ may be driven not by his phonological system or a cognitive overload or a but, rather, by a faulty processing strategy created by pedagogical focus on decoding so that he develops some idiosyncratic and ineffective oral miscues that are phonologically or prosodically inappropriate. These analyses, however, while intuitively appealing, are direct contradictions to P‘s mother‘s observation that ―reading aloud helps him (…) with his comprehension‖ (Appendix E, p. E1). As we will see below, this could not be confirmed in the eye tracking condition, where P completed a silent reading. However, an experience early in this investigation with an assessment instrument initially selected for data collection (but discarded due to its

174

inherent inconsistency of scoring and results) may provide some assistance in resolving this conundrum. As previously mentioned, P‘s score on the Gray Silent Reading Test placed him in the < 7;0 age range and at the 1.0 grade level, whereas his score on the Woodcock Language Proficiency Battery‘s word decoding subtest indicated proficiency at an adult level; thereby documenting his ―low comprehender profile‖. However, prior to the administration of these instruments, P had also completed the GORT-4. As mentioned, this instrument was later rejected as an assessment for the investigation because it yielded inconsistent results. In P‘s case, this test yielded two comprehension scores, depending on whether or not the comprehension points he achieved after reaching the comprehension ceiling but before reaching the fluency (oral reading proficiency) ceiling were counted. (The interpretation manual for this test did not indicate that two such scores could be achieved and provided no explanation for the inconsistency of results; not only for P but for another potential participant early in the investigation). For P, the lower score was at the 6;6 age level and that matched a 1.4 grade equivalency, roughly the same level as the GSRT. The higher score placed him at age 7;3 level and a 2.2 grade equivalency, respectively, which was still well below his actual age and grade level. By contrast, his fluency score, a combination of rate and accuracy, placed him at an 8;9 age level and at a 3.7 grade equivalent. In other words, his oral reading proficiency on the GORT-4 was just barely at his actual age and grade level, while his comprehension was a little higher than on the other assessment. Figure 4.5 shows the grade level results for all three assessments.

175

16 14 12 10

oral reading

8

comprehension (1)

6

comprehension (2)

4 2 0 WLPB & GRST

GORT

Figure 4.5: Grade Level Results for Oral Reading vs. Comprehension between Assessments As previously noted, P‘s articulatory struggles may serve as an additional impediment to comprehension. By contrast, his mother said that reading out loud helps with his comprehension. Her view was confirmed, but the situation turned out to be more complex than anticipated. The comparison of assessments showed a strong decrease in decoding proficiency, and a weak increase in comprehension, when reading out loud. In other words, his performances in decoding and comprehending converged a bit when reading connected text out loud with the expectation of comprehension. Performances diverged when the ability to comprehend and ability to decode were assessed independently. While this may be an artifact of the instruments and tasks provided to him in this study, it is possible that assessments of P‘s proficiency as a reader in real-life environments diverge along the same lines. In a situation where he is simply asked to read out loud but not to comprehend, he will be assessed as a good or even very good reader. Evidence for this is found in his mother‘s statement, ―He can read War and Peace. He just can‘t tell you what it‘s about‖ (Appendix E,

176

p. E4). In a situation that requires both reading out loud and comprehending, his abilities may find more realistic appraisal. We can only speculate what role this phenomenon might have played for a child whose environment views reading, first and foremost, as the ability to accurately read out loud. We know that P‘s literacy instruction was modeled, at least partly, on the Simple View. We also saw a general tendency, both in P and his environment, to place more emphasis on oral proficiency than on meaning construction. With our newly gained insight that his decoding abilities improve when he is not reading for meaning, it must be asked whether his instructors, faced with the pressure to meet ‗tangible‘ benchmarks, focused on this apparent strength while neglecting his comprehension weakness. Again, this is only speculation. If it were true, though, it would provide another important piece of the puzzle for P‘s difficulties. For our purposes, it does enable the formulation of a third theme based on these miscues, labeled STRUGGLE FOR ACCURACY. A third type of OR-ER differences can be dubbed ‗hesitations‘. This label combines ‗hesitations‘, ‗choppy reading‘, and ‗lengthening of in-word sounds‘ in Appendix G, p. G17. Each of these behaviors disrupt the flow of reading, and indicate tentativeness in P‘s oral productions, suggestive of an orientation towards oral reading accuracy. The most prevalent subtype identified in these data was ‗choppy reading‘. It was marked as ‗choppy‘ if there was any interruption of reading flow that was long enough to be noticed by a listener but did not create an actual pause. P produced 50 of such interruptions, often in close proximity to one another, resulting in a halting quality in his reading. Figure 4.6 shows an example of this particular quality. ‗Choppy‘ hesitations are marked with ¦, whereas | stands for a hesitation proper (discussed below).

177

Figure 4.6: Instances of ‘Choppy’ Reading An interruption was marked ‗hesitation‘ when it resulted in a noticeable pause, indicating processing activity. P produced 46 hesitations in reading one, 12 of which (26.1%) occurred at the line‘s end. Figure 4.7 shows two examples of end-of-line pauses.

Figure 4.7: Examples of End-of-Line Hesitations The first of these occurred within the phrase Green Army Men, which refers to a group of toy soldiers that are part of the main cast in Toy Story. The capitalization indicates that this is the proper name for this group of characters. P has read the book before, so he can be expected to be familiar with them (as we will see below, in his retell he showed considerable knowledge of the storyline). However, his hesitation at this point indicates that he is not using his prior knowledge of the text to predict the final word of the phrase. It is instructive to compare this with the next instance of end-of-line hesitation, gym | sock. Hesitation is not surprising in this instance, as the second part of the compound word is not easily predicted from prior text. Yet P treats both chunks of language, the predictable one and

178

the less predictable one, the same. This suggests a lack of predictive strategies (cf. Goodman, Watson, & Burke, pp. 192-193) Figure 4.7 also shows an instance of within-word hesitation (in|to). This hesitation is indicative of good predictive strategies, as the sentence could indeed have ended with …trooped in. Most other hesitations, however, both of the within-word and the betweenword type, are indicative of lack of prediction, as in Figure 4.8.

Figure 4.8: Examples of Between- and Within-Word Hesitations Jessie, the red-haired cowgirl, is a protagonist of the Toy Story franchise and one of the better known characters in the Disney universe (cf. Disney Wiki). P‘s difficulty in verbalizing the core descriptor phrase for this character suggests that he is not making use of any pertinent knowledge he possesses; even if he does not have any such knowledge, the hesitation after cow would still be indicative of failure to use predictive strategies since the only possible continuation of a compound word beginning with cow- and following redhaired would be either cowgirl or cowboy, as cows do not have red hair, nor do any other entities that compound with cow (e.g., cowshed or cowbell). This particular set of hesitations, as well as the end-of-line hesitation in Figure 4.7, could be linked to a lack of motivation on P‘s part. His mother mentioned that she ―tried to get him to read chapter books, like this Toy Story, myeah (…) it didn‘t do anything for him‖ (Appendix E, p. E25). Thus, his failure to predict names or descriptive phrases for main characters may be due to a lack of motivation to engage with the story. Other hesitations, by

179

contrast, are clear indicators for reading without prediction, using strategies that could be dubbed ‗piecemeal processing‘, following Garner (1981). Figure 4.9 shows instances of such processing. In lines 008-009 and 014-015, P produces various hesitations at phrase boundaries (e.g. between they just hoped and it would work) or before words that presented difficulties to him (e.g. hoisted).

Figure 4.9: Examples of Hesitations Between Phrases These particular hesitations show that P does not venture to make predictions if he is unsure of upcoming text. Instead, he pauses until he is sure that he is able to pronounce the next word accurately—except in situations where he resorts to a sounding-out strategy, as with the word hoisted in line 015. There were three corrected partials in home reading one, and five in home reading two, indicating that this is a strategy he employs successfully (if success is equated with accurate oral rendering of print). Within-word hesitations are even more manifest indicators of piecemeal processing. We saw a hesitation within a compound word in Figure 4.8; in other instances, P read words on a syllable-by-syllable basis, hesitating after each syllable until he was sure to pronounce the next one accurately. Figure 4.10 shows two such instances in the words rummaged and

180

sarcastically. (Note: part of the text in lines 058-059 has been edited out for clearer presentation.)

Figure 4.10: Examples of Hesitations between Syllables In both instances, the word in question is an unusual and complex one, and P‘s strategy of tackling them ensures that he gets through without deviating from the print—but also without using any language systems other than graphophonemics. The tentative nature of his reading here is particularly well illustrated in the lengthening of the [ɹ] sound in sarcastically. A final piece of evidence for piecemeal processing of print comes from P‘s graphophonemically-based miscues, subsumed above under FOCUS ON CODE. Productions such as $[wʊːld] for would, or $[spaɪəlt] for spilled indicate processing on a grapheme-bygrapheme basis with no recourse to other cueing systems. The latter miscue was corrected; however, this does not necessarily mean that P became aware that his production had no meaning. $[wʊːld] is a nonword just like $[spaɪəlt], but it was not corrected, suggesting that the correction on spilled may have been triggered by the grapheme sequence , in which invariably stands for a monophthong. Alternatively, we can speculate that P did realize that $[wʊːld] is a nonword but did not correct it because it sounded phonetically ‗close enough‘ to the ER. If our contention is correct that he is struggling for accuracy—either

181

because of some lingering articulation issues, or because of flawed processing strategies—it would be plausible that P should settle for a good-enough match of the ER and move on. It cannot be said that P reads without making use of prediction entirely. Many of his reading behaviors actually do show predictive strategies, albeit unsuccessful ones. For example, note the prosodic miscues shown in Figure 4.1 and 4.2. In both cases, he predicts a certain syntactic structure: Okay, [term of address] [end of sentence] in line 001; There they are, Rex [verb of vocal expression] [end of sentence] in line 010. The actual lexical item in each case is incongruent with the predicted structure (places instead of a term of address in line 001, clapped instead of a verb of vocal expression in line 010). The same applies to line 014 in Figure 4.9, where his intonation suggests that he predicted Sarge called as he [verb], whereas the actual text contains another phrase before the verb. Each of these cases appears to suggest a disconnect between syntax-based and semantics-based strategies in P‘s repertoire. On occasion, it seems, a strong expectation of a particular syntactic structure can override his habit of piecemeal processing, but his pattern of processing still interferes with meaningfulness due to lack of focus on wording, as well as a to failure to process print into larger meaningful units. At times, however, this strategy is not only a weakness; it can also be a strength. Figure 4.10 shows an interesting interplay between both aspects. Taken by itself, his reading of lines 058-059 until said is not only accurate but even expressive, with appropriate stress on that and falling intonation after said. In this particular case, P‘s predictions were accurate as regards wording—except that the sentence did not end there (and that the next word was so complicated that he produced it syllable by syllable). In sum, we can conclude that while P has begun to make use of some predictive strategies, he is not effective at doing so; consequently, his predictions actually tend to mislead him to where

182

they become incongruent with the text. Absent more effective strategies, it is not surprising that he would rely mainly on tentative, piecemeal processing of graphophonemics. In sum, the thematic insights from the preceding discussions are as follows. First, we saw earlier themes re-emerge: LACK OF MOTIVATION from the interview, which may have contributed to P‘s failure to use predictive strategies; and FOCUS ON CODE as well as STRUGGLE FOR ACCURACY.

Two new themes emerged as well, both of which were evident in

the halting quality of his reading; they can be termed PIECEMEAL PROCESSING and STRUGGLE WITH PREDICTIONS.

Both are related to HIGHER-LEVEL LANGUAGE ISSUES; where processing

occurs in small units, from phrases all the way down to syllables, prediction is difficult. No confirming strategies could be observed during reading one. As regards correcting behaviors, it is instructive to scrutinize all nine miscues that were corrected (cf. Appendix G, pp. G10-G11). Six are obvious nonwords, such as $[spaɪəlt] above, or $[blʌ] and $[mbʌ] for Buzz, indicative of P‘s FOCUS ON CODE and STRUGGLE FOR ACCURACY. Only three are actual words; one of them, shown in Figure 4.11 below (part of the text edited out for clearer presentation), is one of the few miscues that were acceptable both syntactically and semantically, and yet P corrected it, which is another indication of both themes linked with a reduced focus on meaningfulness. (Incidentally, it is also the only miscue in home reading one that is only somewhat similar to the ER both graphically and phonemically, showing once again P‘s concern with accurate rendering of print.)

183

Figure 4.11: The Only Completely Acceptable Miscue in the Sample (Corrected) P‘s sampling strategies are characterized by piecemeal processing and excessive focus on the graphophonemic aspect of print, as discussed above. Based on his reading behaviors, we can surmise that he typically samples every bit of printed information, which would be an inefficient strategy (this will be discussed further in the subsequent section on EMMA). Regarding integration of read material, it will become clear during discussion of the retell that P does not integrate read material into a meaningful whole. However, his reading is not completely devoid of integration; there are hints of it, for example, in line 015 (see Figure 4.12 below), wherein he omitted the determiner the, effectively turning the noun space ranger into a proper name. This is consistent with the prevalence of other noun-type proper names in the text (e.g. Buzz Lightyear, Sarge, Green Army Men); this may suggest some integration of textual knowledge. Similarly, line 050 (see Figure 4.12) was read in appropriate character voice; even though there were no other such instances, this suggests some integration of story structure on P‘s part. On the whole, however, such attempts at integration do not go beyond the local level.

184

Figure 4.12: Examples of Good Integrative Strategies As another data set, P‘s retell of the text (cf. Appendix G, pp. G22-G23) speaks clearly to his failure of integrating what he is reading. The most flagrant indication for this is the fact that he did not retell the passage he read, but a later text portion. (Hence, I did not score his retell as per the in-depth procedure; the score would simply have been 0.) In addition, his narrative was disjointed, most notably because he quoted direct speech out of context and his overall presentation did not accommodate for a listener without prior knowledge of the book. I quote the last turn of his retell, which contains examples for all that has been said: And then everyone was going to daycare. ―No no he was putting you in the attic, so how did we end up on the curb?—That was a mistake, Andy‘s Ma thought you were trash and (called us junk)‖. And Jessie started to say, ―Andy is going to college‖. And then Woody tried to pull the box out of the car and then Andy, Andy‘s Mom closed the back door and then and then going to daycare and then putting them in the caterpillar room. Appendix G, p. G22 P did not indicate who is talking when he quoted the first long stretch direct speech; consequently, there is a lack of cohesion between this part, the following sentence (And Jesse

185

started to say…) and the final part of the presentation. Additionally, P did not explain caterpillar room even though a hearer unfamiliar with the story would not be able to make sense of this item. Note also the lexicogrammatical oddity of the phrase started to say, which speaks to his expressive language issues. When I asked him if what he had told back to me was what he had read to me, he answered affirmatively, oblivious to the fact that nothing of what he read appeared in his retell. These observations led to a sixth and final theme for home reading one: P‘s failure to integrate printed information, both while reading and during the retell, was dubbed LACK OF INTEGRATION.

The themes gleaned from the first home reading are presented in Table 4.6.

Theme

Main indicator

FOCUS ON CODE

OR-ER similarity

HIGHER-LEVEL LANGUAGE ISSUES

2/3 of miscues indicate lexicogrammatical weakness

STRUGGLE FOR ACCURACY

Incidence of misarticulations

PIECEMEAL PROCESSING

Pauses, hesitations, ‗choppy‘ reading

STRUGGLE WITH PREDICTIONS

Expectation of syntactic structures overrides meaning

LACK OF INTEGRATION

No evidence of integrative strategies; disjointed retell

Table 4.6: Themes from Home Reading One Miscue analysis home reading two. As previously mentioned, P‘s interest in expository texts like Titanic seems to be higher than his interest in storybooks; I was therefore expecting a more constructive transaction with the text. However, that expectation was countered even before he began reading; he was not even able to find the spot in the book where he intended to begin and, as we will see, his retell was as disintegrated as the previous one. By contrast, P‘s miscues exhibited some differences that could be attributed to a higher level of comfort with and engagement in the text.

186

P read a total of 377 words, on which he produced 23 miscues, yielding an average of 6.1 miscues per 100 words, which is consistent with his prior performance. Of his 23 miscues, 10 (43.5%) did not result in meaning loss; the remaining 13 (56.5%) resulted in partial (4/23) or complete (9/23) meaning loss. This pattern is roughly consistent with previous results. P‘s performance as regards grammatical relations in home reading two is also similar to that exhibited in home reading one; 13 (56.5%) of his miscues indicated strength or partial strength in grammatical relations, or were the result of a grammatical overcorrection, while 10 (43.5%) indicated grammatical weakness. The graphic and phonemic similarities between OR and ER also show the same pattern as in reading one. In 20/22 ORs (90.9%) to which this measure applies, there was high similarity to the ER both graphically and phonemically. The FOCUS ON CODE theme was thus corroborated. However, in contrast to prior results there was one miscue with no sound similarity to the ER in this sample, and one that was neither graphically nor phonemically similar. Table 4.7 sums up the patterns in home reading two.

187

Pattern

%

%

Meaning construction No loss Partial loss Loss

43.5 17.4 39.1

56.5

Grammatical relations Strength Partial strength Overcorrection Weakness

34.8 13 8.7 43.5

56.5

OR-ER similarity Graphic similarity High Some None

86.4 9.1 4.5

Sound similarity High Some None

86.4 4.5 9.1

95.5

90.9

Table 4.7: Pattern Analysis for P Home Reading Two P seemed to be more engaged with this text than the previous one, and the differences in miscue patterns are illustrative of this. The best evidence for this comes from the two miscues with little or no resemblance to the ER, both of which also fully preserved meaning. This pattern indicates constructive engagement with the text; it shows that P is capable of making predictions that are independent of print and contextually meaningful. Figure 4.13 below shows the miscues in question (lines 050 and 052). They are presented in context to illustrate their textual fit (text and markings edited for presentation).

188

Figure 4.13: Miscues Indicative of Constructive Prediction Consider the partial phrase in the Ti-, corrected to in the hull (line 050). P‘s prediction that the sentence would read slicing small holes in the Titanic makes perfect sense, given the previous information that the iceberg grazed the Titanic‘s starboard side. Also, this sentence presents the pivotal moment in the sequence of events; it can be assumed that P has background knowledge of this text. Either way, whether it is contextual information or prior knowledge that P is using, he is making a meaningful and appropriate prediction that is graphophonemically dissimilar to the print, indicating that he is integrating information and actively processing the text. The same is true for his corrected substitution of the for his in line 052, which is as appropriate as the ER syntactically and semantically. His constructive engagement with the text at this point is vividly illustrated by his musings over the word hull, interspersed after he corrected in the Ti- to in the hull. The video data (minute 29:13-29:23 of P home reading) shows lively facial expression as he gazed from the text towards me and back. ―What‘s the—‖ he began, and then answered himself: ―oh the hull is like the, the down uh part‖, after which he resumed reading. Briefly puzzled by an unknown word, he wondered

189

out loud about its meaning, and quickly came to a conclusion that he shared with me. We cannot know whether he retrieved previously acquired information or whether he connected the pieces of information provided by the text. Either way, this brief sequence illustrates that P is capable of active integration of information and making meaningful predictions. It would seem that such engagement has an overall beneficial effect on P‘s reading strategies in this sample. Evidence for this comes from another miscue illustration (line 049). Consider his struggle with slicing small. He repeated slicing three times, even though he had read it accurately the first time (albeit with unnatural intonation—equal stress on both syllables—but as we have seen, intonation is not one of P‘s prime concerns while reading). This is in stark contrast of his usual struggle to match his production with the print, and his disregard for prosody. The best explanation for why he would repeat an accurate production three times is that he was struggling to confirm the word‘s meaning. This is in contrast to the previous reading, where he showed no indication of confirming strategies. I will return to this point below. Unfortunately, this page represents the only text portion where he demonstrated such engagement. And even here, he kept his constructive momentum in check, overcorrecting his productions in lines 050 and 052 to achieve the print-to-sound accuracy we have found to be his prime concern. Further, his other miscues have the same characteristics as the ones in home reading one, thereby corroborating previous themes. I will discuss some pertinent instances in turn. The most telling indicators for P‘s failure to actively construct meaning are his multiple uncorrected productions of nonwords with high graphic and phonemic similarity to the print. Seven of his miscues (30.5%) fall in this category: $[ꞌ tɔɹwɔɹd] for toward (line

190

009); $[ꞌ kɑlɪʒən] for collision (line 016); $[ꞌ ɛlədʒənt] for elegant (line 019); $[ꞌ tɛˑnꞌ tæˑʃ] for advantage (line 030); $fantic for frantic (line 044); $seem-in-gly for seemingly; and $[ꞌ pɔɹꞌ tʃːɪn] for portion (line 048) (Appendix G, pp. G5-G9). The fact that all of these remained uncorrected corroborates the HIGHER-LEVEL LANGUAGE ISSUES theme that emerged previously. Tellingly, P carried one of his nonwords over into the retell. When I asked him to look back at the text to get some more information for his retell, he glanced over the first page and said: ―oh the Titanic was steaming $torpord an ice field‖ (Appendix G, p. G24). That is, even during the retell he failed to attend to the fact that a word he was using did not make sense. This provides further evidence for his difficulty with higher level language. Note also how close his productions are to the ERs; this is consistent with the STRUGGLE FOR ACCURACY

and FOCUS ON CODE themes listed above.

Grouping miscues by combined semantic-syntactic patterns as previously, little more than half turned out to be of low quality in this reading (i.e. they disrupted both syntax and semantics). 13 out of P‘s 23 miscues (56.5%) resulted in full or partial loss of meaning, and/or indicated partial strength or weakness in syntax. In the previous reading, 65.625% of miscues had these characteristics. The relative improvement in higher-level language may be due to P‘s higher interest in and possibly greater familiarity with the text; further evidence for this comes from his strategy usage, to be discussed below. On the whole, however, his profile is still indicative of a non-proficient reader. The HIGHER-LEVEL LANGUAGE ISSUES theme was thus confirmed. Table 4.8 sums up the grouped patterns for home reading two.

191

Pattern

%

Meaning loss, weakness in grammatical relations

39.1

Partial meaning loss, weakness in grammatical relations Partial meaning loss, partial strength in grammatical relations No meaning loss, grammatical overcorrection No meaning loss, strength in grammatical relations

4.4

% low qual.

56.5

13 8.7 34.8

Table 4.8: Home Reading Two Miscues Grouped by Combined SemanticSyntactic Alterations The STRUGGLE FOR ACCURACY theme, already confirmed by P‘s production of nonwords, was corroborated further by phonetic alterations of the same type as in reading one, that is, dialectal productions such as [ʔ] substitutions of plosives (e.g. line 001), or misarticulations such as [ɭ] for [ɹ] in operator (line 006). His struggle with the name of a main character (Lightoller) illustrates how themes can interact; each of the three times the name occurred, he struggled with it, producing a within-word hesitation on the first instance, a dysfluency on the second, and several attempts at sounding out with eventual correction on the third. Although his first production was accurate, he sounded as if he was reading the name for the first time on each subsequent instance, suggesting he did not integrate information to the point of anticipating/retrieving, the name of a main character. At the same time, he is clearly concerned with producing it accurately; together, these observations confirm the earlier theme LACK OF INTEGRATION and show its interaction with STRUGGLE FOR ACCURACY. Figure

4.14 shows all three instances.

192

Figure 4.14: Interaction of LACK OF INTEGRATION and STRUGGLE FOR ACCURACY As for the PIECEMEAL PROCESSING theme, the same hesitations and ‗choppiness‘ in his reading could be observed as in reading one, sometimes with concurring meaning loss, as in Figure 4.15 below (miscues not pertinent to the discussion edited out).

Figure 4.15: Example for PIECEMEAL PROCESSING resulting in Meaning Loss However, there was a substantial difference in numbers. P produced 68 instances of ‗choppiness‘ and 38 hesitations in reading two, as compared to 50 and 46, respectively, in reading one. Also, the number of words was lower in the former (377 as compared to 481); this translates into 28.1 instances per 100 in reading two, and 20 instances of halting reading per 100 words in text reading one. So why should there be a higher incidence of haltingness in a text with which P is more familiar and engaged? Viewed in conjunction with the lower incidence of miscues in reading two, this observation can be tentatively explained. Recall that P‘s primary concern in reading appears to be accurate rendering of text, and that he struggles to accomplish this goal. In a text that he has less difficulty with (e.g., because he is more engaged or familiar with it), this approach

193

may result in a lower incidence of miscues; at the same time, it may result in more halting productions. The struggle is shifted from the obvious wrestling with text resulting in miscues to the subtler struggle for accuracy revealed through haltingness. The incidence of misarticulations in both texts would appear to corroborate this impression. In home reading one, P read 481 words, and produced 34 dialectal or misarticulated forms (i.e. 7.1 instances per 100 words). In home reading two, he read 377 words and had 12 struggle-type productions (i.e. 3.2 per 100 words). As with actual miscues, the observation seems to be that with greater ease of reading, his struggles do not simply decrease but are shifted to the less obvious, prosodic plane. The STRUGGLE WITH PREDICTIONS theme appears in a somewhat different light in this reading. We already saw that P made two high-quality predictions in lines 050 and 052, but there were no other indications for predictive strategies in the sample. There were, however, indications for lack of such strategies, such as inaccurate corrected partials close to the surface form of the print as in Figure 4.16 below (other miscues edited out). In both examples, P makes a prediction that is based not on meaning but on the graphic surface of the letter sequences.

Figure 4.16: Examples for STRUGGLE WITH PREDICTION In addition, P produced another prosody-based syntactic miscue, erroneously predicting a specific syntactic structure at the expense of meaning. Instead of the phrase of

194

address, ―Say, old man‖, he read ―Say, OLD man‖ with stress on old. This prosodic pattern would have been appropriate for a structure of the type ―Say x‖, where x stands for any unit of language the speaker wishes the hearer to repeat. It can be assumed that P is unfamiliar with this somewhat dated phrase, and hence his syntactic expectations once more overrode meaningfulness. In sum, the fluctuations in his predictive performance seem to warrant maintenance of the STRUGGLE WITH PREDICTIONS theme. However, we can account for the effective predictions we saw above by viewing them as indicators for emergent proficiency in integration: predictions cannot be made without integrating prior information. Also, P made use of confirmation strategies in reading two, that is, in the slicing small sequence in Figure 4.13, or when he repeated the accurately but haltingly read sequence The Captain then. All of these suggested that he was integrating textual information better than in the first reading. There is another set of behaviors in the sample that can be viewed as indicators of integration as well – paratextual comments. We discussed P‘s musings about the word hull above; two similar comments are shown in Figure 4.17 (text and markings edited for presentation).

Figure 4.17: Paratextual Comments

195

In the first example, P looked at the troublesome passage to process it (indicated by the filler uh) until he had made of sense of it (indicated by the exclamation oh!) and then proceeded to read it. This is quite different from his usual strategies of tentative piecemeal processing or sounding out; it is indicative not only of constructive engagement but also of integration of previous information; there were three clock times in the text prior to this one, which had the same format: 9:00 a.m., 1:00 p.m., and 6:00 p.m. (cf. Appendix G, pp. G5G9), suggesting that he was integrating this information to resolve his problem at this point. In the second example, P produced to ―Shut up!‖ tentatively first, and then partly repeated it in an alternative rendition of the same meaning, ―just shut up, son‖. Again, this is different from his usual piecemeal strategies, and appears to be a blend of a paratextual comment and the repetition-for-confirmation strategy discussed above. A final set of evidence for better integrative strategies comes from P‘s retell of reading two (Appendix G, pp. G24-G25). In contrast to the previous one, he actually rendered the text so that it was possible to score it as per the in-depth procedure. I quote the retell it in its entirety here, omitting a hearer signal of mine: P:

uhm the Titanic was appearing—the—iceberg appeared in the—at— Tietienec—Titanic (…) and then try to turn the ship and then the ship went (in there) first and made a turn on a side then hit and the ice break two holes in Tite—in the Titanic‘s hull—hull. And then water going in and then water— pouring into the Tine—Titanic‘s boiler rooms and then the Titanic unsink— the unsinkable Titanic uh was about to sink.

R:

wow. Anything else?

P:

(while looking at the book) uhhh… no—I think that‘s it! (pause)

196

R:

wanna have another look real quick?

P:

(pause, looks over first page of reading) oh the Titanic was steaming $torpord an ice field. Too. And that‘s it.

P‘s retell, while relatively consistent with the storyline, was far from complete; he left out the entire build-up to the central moment of the collision, and he did not mention any of the characters or how their actions contributed to the events. In addition, the retell suffers from lexicogrammatical oddities; he struggled for the appropriate verb form of appear, left out the pronoun they in try to turn the ship, produced an inaccurate verb form on break, omitted the auxiliary in water going in and water pouring, included a nonsense word ($torpord), and alternated between past and present tense throughout. The in-depth procedure retell score does not account for such phenomena, hence the score is solely composed of content factors. The authors of the miscue analysis manual suggest allotting 20 points for character recall and 20 for character development, and 60 for recount of main events. As per this guideline, P scored 24/100 points for his recall of the pivotal events. Thus, the retell showed the same ambiguity in integrative strength similarly to the strategies as discussed above. On the one hand, P clearly did better than in reading one, which can be attributed to his interest in the text; on the other, he still struggled, and his performance was far from proficient. I decided that it was warranted to change the LACK OF INTEGRATION theme to account for the observation that P does integrate information when interested in a text, albeit in a vague and inconsistent way. I renamed the theme EMERGING INTEGRATION. In sum, then, the themes gleaned from home reading one could be confirmed in reading two, except that one theme had to be changed to accommodate new findings. Table 4.9 lists the complete list of themes that emerged from home readings 1 and 2.

197

Theme

Main indicator

FOCUS ON CODE

OR-ER similarity

HIGHER-LEVEL LANGUAGE ISSUES

2/3 of miscues indicate lexicogrammatical weakness

STRUGGLE FOR ACCURACY

Incidence of misarticulations

PIECEMEAL PROCESSING

Pauses, hesitations, ‗choppy‘ reading

STRUGGLE WITH PREDICTIONS

Expectation of syntactic structures overrides meaning

EMERGING INTEGRATION

Some evidence of integrative strategies when interested in text

Table 4.9: Joint Themes from Home Readings One and Two Eye Movement Miscue Analysis A few weeks after the home reading, P was invited into University of Louisiana at Lafayette‘s eye tracking laboratory and presented with text excerpts from two storybooks and one expository text, selected for age-appropriateness using the Accelerated Reader (AR) system. I asked him to read them from a screen, wearing a Mobile Eye eye tracking device; as previously, he was informed prior to each reading that he would be asked to tell the texts back to the researcher. For the first text, he was instructed to read silently; the rationale for this was to determine any differences in eye movements between reading silently and reading aloud. Table 4.10 sums up the characteristics of the three stimuli texts and the reading condition. Text #

Author, Title

Genre

AR book level

Reading condition

1

Avi, Something Upstairs

Storybook

4.1

Silent

2

Dan Gutman, Getting Air

Storybook

4.1

Aloud

3

Sue Whiting, All About Ants

Expository text

3.8

Aloud

Table 4.10: Stimuli for EMMA

198

EMMA text one: Silent story book reading. As the first reading was silent, no miscue analysis could be effected. However, a retell was elicited and analyzed. Results show almost complete lack of integration. The text describes a boy‘s attachment to his home city Los Angeles, CA, and juxtaposes it with a description of Providence, RI, to where the boy‘s family is moving; it also discusses the boy‘s emotional response. P conveyed virtually nothing of the read text, omitting the evocative depictions of both locales and the protagonist‘s reaction to the move, and made only fleeting mention to the fact that a move was happening (―they were going somewhere on the road‖). I provide the entire portion of the retell prior to prompts to give the flavor of the rendition: uh (pause) uuhhh (pause) they were going somewhere down the street where um (pause) uh his father um said something about (pause) to his mother um um sister. They were um they were uh going somewhere on the road um to somewhere where they were going go to somewhere where they wanted to go uh that‘s it. (Appendix H, p. H7) When asked to look back at the text to provide additional information, P flipped back through the stimulus slides in reverse order, reading disconnected sentences from the screen that caught his attention (my hearer signals omitted): The street was narrow, crowded, and old. Aannd… houses were built. aaand… the smallest in the Union (…) is so small that the city of Los Angeles couldn‘t be fit into it. (pause) Kenny could, and did it—and did play baseball most days. (long pause) All his life he had lived in nothing but spring and summer weather. That‘s it. (Appendix H, p. H7)

199

In other words, his retell included a few of the textual descriptions, but other than that it bore no resemblance to the actual text. No event structure or textual cohesion could be detected. It is not unreasonable to assume that P did not understand the text in the least. His retell score reflects this at 6 out of 100 points. This is for the portion of the retell prior to the prompt (i.e. without looking back). Goodman, Watson, and Burke (2005) do not give guidelines on whether to score prior to or after the lookback. If the latter were recommended, P‘s score would be at 18 out of 100; however, this number masks the fact that his retell is characterized by the quasi-absence of a complete narrative. Interestingly, there is an indication that P did actively process the print while looking back. After glancing over the last slide of the text (the first in the lookback, as he proceeded in reverse order), he said ―not that‖ voicelessly, as if to himself, and went on to scrutinize the second to last slide. There are 6 fixations on the slide in question (cf. Appendix H, p. H6), but the amount of eye movements shown in the Mobile Eye video suggests that he glanced over text more than the fixations reveal, albeit without fixating. We have no way of knowing what made him decide that the slide did not contain useful information, but this incident indicates that he was using some kind of attempt at integration of printed information, albeit unsuccessfully. This recapitulates the ambiguity of his integrative strategies, captured in the EMERGING INTEGRATION theme

above.

During eye movement analysis of P‘s first reading, an unexpected problem arose. Due to the interaction between P‘s gaze angle and the curvatures of stimulus screen and camera lens, the recording showed a considerable amount of vertical drift: a systematic upward shift of fixation progressions and skewed progressions, resulting in fixation recordings that were higher than the actual progressions, and slanted upward toward the right-hand margin of the

200

text. I had to adjust for this error by manually altering vertical position and rotating the progressions. (Rotation was not necessary on all slides, but vertical adjustment had to be done for every stimulus in text one.) Figure 4.18 shows an example of this phenomenon. It depicts the first stimulus screen of text one prior to and after adjustment. Progressions for each line are depicted in a different color, starting with the first saccade of the return sweep (recall that the plots in this study were generated manually, as described in Chapter Three).

Figure 4.18: Fixation Progressions on Slide One before and after Adjustment The figure corroborates that this phenomenon was indeed a systematic flaw in the recording. The upper part of the picture shows fixation progressions prior to, and the lower part, after manual adjustment. This was accomplished by grouping all progressions into a single shape in PowerPoint that was then shifted five keystrokes downward and rotated two degrees in clockwise direction. It is obvious that each of the progressions in the upper portion

201

deviates from the printed lines both in vertical position and in angle, resulting in end-of-line fixations that are systematically higher than the final word of each line, and in almost complete lack of fixations on the last line. The match between progressions and lines after adjustment provides further evidence. There are as many horizontal progressions—delimited by the return sweep—as there are lines, and that the last progression is slightly shorter than the others, just like the last line. The last fixation on this slide—no. 111— is the only fixation progression that is shorter than a corresponding line. Prior to the adjustment, it appeared to have landed on the second to last line, two words before its end; there is no evidence why P‘s fixation behavior should be different for this line. Absent any other explanation, we have to conclude that fixation 111 did, in fact, land on the last word of the last line, which is where it ended up after adjustment. It could be argued that P may have produced one excess fixation progression, (e.g., by fixating the first line twice) but that still would not explain the relative shortening of the last progression as compared to all the previous ones, which scrupulously matched the length of the corresponding lines. As the exact placement of fixations could not be determined, not all measures that are usually done for eye movement recordings of reading were completed for text one. Most notably, number, percentage, and types of words fixated were not included, because they require precise information about fixation locations. Table 4.11 shows the measures that could be completed (cf. Appendix H, p. H40).

202

Measure

Result

Reading time

276.830 sec

Average fixation duration

0.257 sec

Fixation duration range

0.1—0.9 sec

# of words

425

# of fixations

755

# of fixations per 100 words

178 (rounded)

# of regressions

90

% of regressions

11.94%

Table 4.11: Eye Movement Measurements for Text One Despite their incompleteness, it is instructive to compare these results to the measures reported in the literature. Rayner and Juhasz (2006) provide an overview of developmental trends, reproduced in Table 4.12 below. The general trend for all measures is decrease from first grade to adulthood. Comparison to P‘s performance indicates that his average fixation duration (0.257 sec) is close to that of a typical fifth-grader, a little shorter than would be expected in his grade level (0.266 sec). The range of his fixation durations (0.1 to 0.9 sec) appears fairly typical. Rayner (2009) states that average values range from 0.05 to more than 0.6 sec, whereas Paulson and Freeman (2003) report fixations of almost 1.3 sec in one reader, placing P well within observed ranges. By contrast, P‘s number of fixations per 100 words (178) is higher than would be expected at his grade level (121), and falls between the values that are reported as typical for first- and second-graders, respectively. Then again, his percentage of regressions (11.94 %) is lower than typical for his grade level (26%), and falls within the 10-15% reported as typical for adults (Rayner and Juhasz; Rayner, 2009), with values close to the low adult range.

203

1

Fixation duration (sec) 355

Fixations per 100 words 191

2

306

151

26

3

286

131

25

4

266

121

26

5

255

117

26

6

249

106

22

Adult

233

94

14

Grade level

% of regressions 28

Table 4.12: Developmental Trends in Eye Movement Measures (from: Rayner & Juhasz, 2006, p. 377) This first set of eye movement data revealed a mixed picture. While P‘s average fixation duration and his range of fixation durations were both fairly typical, the other two measures were below and above expected values, respectively. He has slightly more fixations per 100 words than the average fourth-grader, with results placing him in the range typical for first- and second-grade readers. By contrast, he made fewer regressions than average for his age, so few as to place him at the lower end of typical adult range. From a transactional perspective, these findings complement each other. They can be interpreted as indications that P‘s eye movement strategies may be primarily geared towards sampling bits and pieces of visual information, as indicated by the relatively high number of fixations per 100 words. This would be consistent with the PIECEMEAL PROCESSING theme that emerged during miscue analysis. By contrast, the relatively low percentage of regressions suggests that he does not use regressions as a strategy to confirm and integrate printed information. It is commonly agreed upon that regressions are indicative of active engagement with print, particularly as regards facilitation of comprehension, and correction of miscues. (Which explains why adults make fewer regressions than children; where comprehension is

204

developed, fewer facilitative strategies are needed.) As Rayner (2009) explains, ―when comprehension is not going well or the text is particularly difficult, more long-range regressions occur to earlier words in the text‖ (p. 1460). Similarly, Paulson (2000) discusses readers‘ strategic use of regressions to deal with troublesome text passages. P‘s comprehension did not appear to ‗go well‘, at least not in light of his retell performance. If lack of comprehension sparks longer regressions, we would expect to see a large number of regressions, particularly long-range ones, in his eye-movement data. However, this is not the case. His fixation progressions are fairly linear, as seen in Figure 4.18; where they did show regressions, they showed a ‗cluster‘ pattern where regressions and re-fixations occurred in rather limited space, usually within two or three words. Figure 4.19 (cf. Appendix H, p. H2) shows an example of such a cluster, possibly prompted by the unusual syntactic structure at the beginning of the first sentence: Kenny could, and did, a rapid succession of two incomplete clauses with an auxiliary but no main verb, one of which elides the subject.

Figure 4.19: ‘Cluster’ Pattern of Regressions P did make a few saccades that looked like long-range regressions, but it was impossible to determine with certainty if they were actually used for this purpose. P had a high incidence of what could be called ‗settling in‘ behavior, as in Figure 4.19, where the actual progression along the printed line does not start until fixation 124; fixations 120-123 show an erratic pattern. P showed this behavior both on the first line of slides and during return sweeps, hence I interpreted it as a search for a starting point when tackling a new line.

205

However, some saccades during reading of lines looked erratic in the same way; some of these resulted in what appeared to be regressions, but due to the recording error, it was difficult to determine whether fixations landed on the text or not. Accounting for the double difficulty of erratic fixation behaviors and recording error, I could only discern one unambiguous long-range regression, shown in Figure 4.20 below (Appendix H, p. H6). It also happens to be the only regression that spanned more than 3 lines.

Figure 4.20: Long-range Regression The duration of the fixation in question (no. 392) was 0.37 seconds, substantially longer than P‘s average fixation duration, indicating that the lookback was deliberate and focused. And yet, this did not translate into comprehension. As seen, there was no mention of snow, Rhode Island, or Los Angeles in his retell. P‘s regressions are thus indicative of the relative weakness in integrative strategies we observed before, and termed EMERGING INTEGRATION.

Paulson (2005) suggests another set of measures to describe natural reading behaviors. He observed that eye movements while reading connected text for meaning are characterized by strong variability at every level of measurement. Figure 4.21 shows an example for fixation durations in one of his participants. Whether averaged per paragraph or

206

per sentence, or displayed individually, there is no uniformity for durations across the reading, indicating active engagement with the print. The same is true, Paulson lays out, for the percentage of fixations on words and for saccade lengths.

Figure 4.21: Variability in Fixation Durations (from Paulson, 2005, p. 350) Due to the recording error discussed above, I decided not to use percentage of words fixated for comparison to Paulson‘s (2005) measurements. Likewise, saccade length did not seem a valid comparison, as it is measured in degrees of angle, and would likely have been affected by head movements due to the head worn setup. However, average fixation duration would not be affected by either problem. To rule out fixation locations as an error source, I decided to report only individual fixations and average duration per paragraph (the same measure at the sentence level was deemed too sensitive to the recording error). With this in mind, the comparison between Paulson‘s reader and P is rather instructive. Figure 4.22 shows

207

the average duration of P‘s fixations on top, and Paulson‘s values at the bottom (graph cut off

Seconds

at 0.7 seconds in P‘s figure for comparison).

Fixation number

Figure 4.22: Comparison of Variability in Fixation Durations Both samples show substantial variability in durations, with a spread from 0.1 to 0.7 in P, and from 0.2 to 0.7 in Paulson‘s (2005) reader. P appears to have more very short fixations (0.1 to 0.2 seconds), and fewer fixations in the 0.2 to 0.5 seconds range than the comparison case, but as we do not know the age of Paulson‘s reader, we cannot interpret this observation. Overall, variability is strong in both cases. In stark contrast to this, the duration averages between paragraphs are similarly varied in Paulson‘s (2005) reader, but barely so in P. Figure 4.23 shows the average duration per paragraph of P‘s fixations on the left, and Paulson‘s values on the right.

208

Seconds

Paragraph number

Figure 4.23: Comparison of Variability in Fixation Durations per Paragraph While Paulson‘s (2005) data show a spread from just over 0.3 to 0.6 seconds, P‘s durations ranged only from 0.2 to 0.34 seconds. Due to lack of comparison data and participant information, the statistical significance of this finding could not be determined. But the interpretive import is clear. As Paulson suggests, if proficient reading is characterized by highly variable patterns of eye movements, then a ―pattern of equal fixation durations on each word would signify a lack of comprehending‖ (p. 355). Fixations per word could not be determined for P, for reasons stated above. But if the same observation holds for other units of text such as paragraphs, then Paulson‘s suggestion would be substantiated; uniformity of durations and lack of comprehension do appear to go together in P. It is too early to formulate a theme from this observation; I will take this up again when discussing the second reading. In the meantime, a final piece of data calls for scrutiny. As seen, P‘s retell did not benefit from the opportunity to look back at the text; it remained as disjointed as it was before. But there is an interesting relation between what P chose to present as part of the retell, and what he looked at in his lookback. Analysis of the relevant fixations shows that he looked at everything he added to the retell, but also that he looked at

209

more text portions than those he verbalized. Figure 4.24 shows an example of this. His verbalization of this slide was: ―The street was narrow, crowded, and old. Aannd… houses were built.‖ (Appendix H, p. H7).

Figure 4.24: Fixations during Lookback In addition to the text portions he verbalized, P fixated part of line one, the phrase could imagine, a few words in the middle of the page, and the final paragraph. This indicates that he engaged in some kind of decision process to select text portions for his retell, which is consistent with his voiceless verbalization not that on the first slide. While this process did not result in a coherent retell, it does suggest that P actively attempted integrating printed information, albeit unsuccessfully. This is, again, consistent with our EMERGING INTEGRATION theme. EMMA text two: Story book read aloud. For text two (cf. Appendix H, pp. H11H19), P read a total of 409 words, on which he produced 12 miscues (i.e., 2.94 miscues per 100 words). This number is much lower than in the home readings, and closer to the low end of ranges reported in the literature. There are two possible explanations for this. One is the

210

reading situation. As discussed above, there was a clear difference in P‘s productions in the more formal assessment situation as compared to the less formal home reading situation. In the latter, he was more relaxed, and produced more deviations from print. It can be speculated that the laboratory situation with all its technology (including the headgear he had to wear) and the detailed procedures of calibration and recording resembled a formal reading situation more than an informal one from P‘s viewpoint, and hence prompted him to read more carefully. The other possible reason is the text itself; it can be argued that the current text was easier to read due to the absence of intricate sentence structures (such as Kenny could, and did), and because it centered on action, not on description. In this second reading sample, 50% of P‘s miscues resulted in no loss of meaning, while 16.7% and 33.3%, respectively, resulted in partial or full loss of meaning. Additionally, 50% of his miscues indicated strength in grammatical relations, and 50% indicated weakness; there were no miscues that indicated partial strength, or were the result of a grammatical overcorrection. Interestingly, 90.9% of the miscues were highly or somewhat similar to the ER, both graphically and phonemically; the remaining 9.1% were not similar to the ER in this regard. The FOCUS ON CODE theme was thus confirmed. Table 4.13 sums up the miscue results for EMMA text two.

211

Pattern

%

%

Meaning construction No loss Partial loss Loss

50 16.7 33.3

50

Grammatical relations Strength Partial strength Overcorrection Weakness

50 0 0 50

50

OR-ER similarity Graphic similarity High Some None

72.7 18.2 9.1

Sound similarity High Some None

72.7 18.2 9.1

90.9

90.9

Table 4.13: Pattern Analysis for P EMMA Text Two As previously, I grouped miscues to combined syntactic and semantic patterns. Due to the low number of miscues, there were only three patterns in this sample. 50% of P‘s miscues did not result in meaning loss and indicated strength in grammatical relations. 33.3% did result in meaning loss and indicated syntactic weakness, and 16.7% indicated the same weakness but resulted only in partial loss of meaning. The insight we gained above still holds: syntactic and semantic difficulties go together in P. Therefore, the HIGHER-LEVEL

212

LANGUAGE ISSUES

theme was thus confirmed as well. Table 4.14 sums up the grouped patterns

for EMMA text two.

Table 4.14: EMMA Text Two Miscues Grouped by Combined Semantic-Syntactic Alterations Pattern

%

Meaning loss, weakness in grammatical relations

33.3

low qual. %

50 Partial meaning loss, weakness in grammatical relations

16.7

No meaning loss, strength in grammatical relations

50

P‘s miscues overall were similar to those during the home readings. He produced four nonwords close to the print: $[ꞌ finɜm] for phenom (line 014), $enc[o]ragement for encouragement (line 019), $rev[ju]lutions for revolutions (line 026), and $[ka]ping for coping, after abandoning the form copying (line 035). He also produced clammed for claimed, abandoning the correct word for the erroneous one (line 029) (Appendix H, pp. H16-H18). This latter case is particularly interesting when compared to line 035. There, he produced coping as copying first, which was semantically out of line and did not match the graphic form. He then abandoned it for a nonword that was at least graphophonemically plausible. In line 029, by contrast, he abandoned the correct form, which fit both semantically and in terms of graphophonemics, for a word that fit in neither regard. The interpretive import here is that on the one hand, he is not as completely constrained by print appearance as his strong focus on code may suggest. On the other hand, whichever processes he may be using to make reading decisions are not efficient yet, and lead him astray in various ways. Both observations are consistent with our EMERGING INTEGRATION theme.

213

HIGHER-LEVEL LANGUAGE ISSUES and EMERGING INTEGRATION occurred jointly on a few occasions, such as the uncorrected production of quiet for quite in lines 040-041 (Appendix H, p. H18). While this substitution resulted in an unacceptable OR as regards both syntax and semantics, it may have been sparked by a prior occurrence of the 3rd person singular verb form quiets in line 034, on which P struggled briefly, as evidenced by ‗choppy‘ reading prior to and within the word. This may indicate that he does form connections across bits of text, however unsuccessfully; another observation that would speak to our EMERGING INTEGRATION theme. In an interesting contrast to previous readings, P inserted a –ed past tense morpheme into the text (producing flipped for flip, line 041) instead of omitting it. The insertion did not occur after an alveolar sound (which is consistent with previous interpretations of this OR as a problem with alveolar placement), and it was corrected, which speaks, again, to EMERGING INTEGRATION—in

this case, to integration of tense information. But the fact that P produced it

in the first place, disregarding the consistent use present tense in the narrative, suggests that higher-level language aspects, such as consistency of tense information, are not his primary concern while reading. His STRUGGLE FOR ACCURACY, identified above as one his main concerns, resurfaced when he read text two. It is instructive to compare the sheer numbers of dialectal productions and misarticulations between texts. In home reading one, P produced 7.1 of such ORs per 100 words. In home reading two, the number was 3.2 per 100 words. In EMMA text two with its 409 words, he produced 61 of such forms, or 14.9 per 100 words (Appendix H, pp. H11-H13. As the text was less difficult than the prior one, there are two plausible explanations; the difference in the reading situation as mentioned above, or the fact that this text was entirely new to him. Recall that P‘s prime concern in reading is to render the text

214

accurately. When he encounters a new text and/or assumes that the environment harbors expectations regarding his performance, this may result in more struggle for accuracy, and hence in more misarticulations. A strong piece of evidence for this comes from his production of dozens in line 009 (Appendix H, p. H15): he read it as d[ə]zens first, with a strongly centralized vowel, which is consistent with his dialect. He then overcorrected this production to the more standard pronunciation d[ʌ]zens, despite the fact that his first production was accurate and intelligible. As before, there was indication for P‘s STRUGGLE WITH PREDICTIONS, as in the clause in Figure 4.25 below. His production of get for give (Appendix H, p. H14) had the potential to be appropriate until he produced on for one, creating the unconventional grammatical structure [get OBJ on]. He did correct the latter miscue to one, leaving the same ambiguous impression regarding his predictive and integrative strategies as above. This ambiguity is amplified by his other uses of predictive strategies. In line 028 (Appendix H, p. H17) he rendered some big shot scientists as some big SHOT scientists, with stress on shot, which would have been appropriate if shot had been the head noun of the phrase. In other words, his prosody evidenced erroneous prediction of a specific syntactic structure, a phenomenon we have observed before. On the other hand, he did, as previously, produce one strong prediction, meaningfully substituting a for the when he rendered off the ledge of a building as off a ledge of a building (line 047, p. H19).

Figure 4.25: Miscues indicative of STRUGGLE WITH PREDICTIONS

215

P‘s PIECEMEAL PROCESSING strategy of interacting with print, evidenced by the halting quality of his reading, could be confirmed for EMMA text two, but with an interesting twist. He produced 65 hesitations, 6 of which occurred within a word, and 6 at the end of a line; likewise, he produced 71 instances of ‗choppy‘ reading, 5 of which within-word, and 6 at the end of a line. Together, this translates into 33.3 instances per 100 words, the highest number observed so far (Appendix H, pp. H11-H13). Figure 4.26 shows an example of a sentence that was read almost phrase by phrase, including a within-word break resulting in the unconventional production moment¦arily (p. H16).

Figure 4.26: Instances of PIECEMEAL PROCESSING (line edited for presentation) We observed above that with greater ease of reading, P appears to produce fewer miscues and misarticulations, but increased haltingness. In EMMA text two, P did produce substantially fewer miscues and more haltingness than previously, but he also produced a higher number of misarticulations. We can attribute this phenomenon to the fact that on the one hand the text was less difficult than previous ones—making for greater ease of reading— but that on the other hand, P was unfamiliar with it, and may have anticipated higher environmental expectations due to the more formal situation, resulting in greater struggle for accuracy. Moving into discussion of eye movements, the first thing we notice is that the measures are quite similar to the silent condition. P read the 409 words of text 2 in 283 seconds. He produced 769 fixations with an average duration of 0.2 seconds and a range from 0.1 to 1.34 seconds; this equals 188 fixations per 100 words. 141 of his saccades were

216

regressions, making for 18.4% of all saccades. (All derived values rounded.) The latter number is the only one that differed markedly from text one, where it was 11.94%; we will discuss this further below. In contrast to text one, there was almost no skewing of eye movement recordings. No manual adjustment of fixations and progressions was needed, which enabled calculation of measures previously prevented by inaccuracy of the recordings: words fixated, and function vs. content words skipped. However, it should be noted that the recordings, once again, did not appear to be completely accurate this time. Figure 4.27 shows two stimulus slides (Appendix H, p. H20). On the left one (slide 5/6), the last line is barely touched by fixations. This is in contrast to all other slides, on which P faithfully traced each line, as seen in the right portion of the Figure (depicting slide 3/6).

Figure 4.27: Example of Possible Inaccuracies in Recording of Text Two There is another possible explanation for this phenomenon: P‘s eye movements are somewhat erratic, an observation to which we will return below. Whichever the cause of the inconsistencies, it was decided that in order to account for them, a percentage not only of words fixated but of words within the foveal window would be computed. A word was

217

counted as fixated if and only if it was touched by a fixation bubble; to counterbalance researcher expectations of high numbers of words fixated, any ‗touch‘ decision was made very conservatively, with words barely touched by bubbles not counted as fixated. By contrast, a word was counted as being within the foveal window of a fixation if the conditions for ‗fixated‘ were met, or if the corresponding fixation was a) above or below a word, within or touching the range of actual or potential ascenders and descenders of characters; or b) less than one bubble diameter to the left of a word, touching or within the word‘s ‗body‘ (characters without ascenders/descenders). Figure 4.28 gives an example of the difference between the two counts. According to the ‗fixated‘ count, only I would be counted as fixated, because it is the only word touched by a fixation bubble. In the ‗foveal‘ count, both I and know would be counted, as fixation 338 is above the word within ascender range. Why would not be counted in either case.

Figure 4.28: ‘Fixation’ vs. ‘Foveal’ Count of Fixated Words With these considerations in mind, the relation between fixations and words presents itself as follows. P fixated 55% of the words in the text as per the ‗fixation‘ count, and 73.1% as per the ‗foveal‘ count. He skipped 55.8% of function words and 36.4% of content words as per the ‗fixation‘ count; the corresponding ‗foveal‘ number is 38.7% and 18.9%, respectively. (All values are rounded.) Table 4.15 sums up the eye movement measures for text two (cf. Appendix H, pp. H20, H41).

218

Measure

Result

Reading time

282.940 sec

# of fixations

769

Average fixation duration

0.200 sec

Fixation duration range

0.1—1.34 sec

# of fixations per 100 words

188 (rounded)

% of words fixated (fixation count)

55% (rounded)

% of words fixated (foveal count)

73.1% (rounded)

% of function words skipped (fixation count)

55.8% (rounded)

% of function words skipped (foveal count)

38.7% (rounded)

% of content words skipped (fixation count)

36.4% (rounded)

% of content words skipped (foveal count)

18.9% (rounded)

% of regressions

18.4% (rounded)

Table 4.15: Eye Movement Measurements for Text Two Comparing these measures to those for the previous text and those reported in the literature, several important observations can be made. First, average fixation time for text two is notably shorter than for text one; indeed, it is so short that it falls below the adult ranges reported in Rayner and Juhasz (2006). This is particularly conspicuous as duration means tend to be longer in oral as compared to silent reading (Rayner, 2009, p. 1460). At the same time, the number of fixations per 100 words is somewhat higher than in text one, placing P even closer to a typical first-grade range than in the previous reading. In other words, P fixated on text two as much as a typical first-grader might, but with fixations shorter than those of a typical adult. P fixated around 55% of the words in text two (cf. Appendix H, p. H20). In and by itself, this measure is fairly typical. Paulson and Freeman (2003) report fixation rates of 50% to 80%, depending on the age of the reader—younger children have a higher percentage of

219

fixated words. Incidentally, they report fixation rates of 54% to 56% in fourth-grade readers (p. 12), which is P‘s range. However, this highlights a discrepancy between the different measures: the sheer number of fixations is high, as typical of a young reader; their average duration is very low, which is untypical for oral reading, and would normally indicate adult proficiency; and the percentage of words skipped is typical. This raises the question, if P does not use his high rate of fixations to look at more words than typical readers, then what does he use them for? Another discrepancy exists between P‘s measures of types of words skipped and those reported in the literature. He skipped 55.8% of function words, and 36.4% of content words as per the ‗fixation‘ count. The readers in Paulson and Freeman (2003) skipped between 62% and 46% of function words, and 17% to 20% of content words; Rayner (2009) reports even larger discrepancies of 65% skipping of function words vs. 15% of content words. Put otherwise, P treated function and content words more equally than typical readers, skipping more content and fewer function words. Notably, all but one of P‘s miscues in text two were on content words, the one exception being the syntactically and semantically appropriate substitution of a for the discussed above. In sum, the overall measures of P‘s eye movements reveal puzzling discrepancies and differences to typical readers. I will attempt to resolve these puzzles through EMMA by considering eye movements together with miscue analyses of his productions First, it should be noted that P did use his eye movements strategically to process text, giving a better insight into his reading strategies than could be observed during oral reading only. Consider the first two miscues on text two (cf. Figure 4.25), get for give (uncorrected) and on for one (corrected) in line 005. As shown in Figure 4.29 below (from Appendix H, p.

220

H20), one received a direct fixation (no. 79) when P regressed from the last word of the phrase; the eye tracking video reveals that he produced the corrected word one around the same time as fixation 79. He thus used a correction strategy, possibly prompted by the syntactic and semantic mismatch between on and last tug: he regressed to fixate one directly instead of just foveally as during first pass, and produced an accurate response. This observation is consistent with our EMERGING INTEGRATION theme established above.

Figure 4.29: Examples of Miscued Words not Directly Fixated on First Pass This observation could lead to the notion that there is a linear relation between eye movements and miscues (i.e. that words which are not directly looked are more likely to be miscued). This idea has been refuted previously (e.g. Duckett, 2002; Paulson, 2000); P‘s data corroborates the alternative concept that miscues are indicative of linguistic difficulties, and that they are not predictable by eye movements. Four of the words P miscued in text two (Appendix H, pp. 11-13) were fixated directly on first pass (cf. p. H20): big (line 010); phenom (line 014); revolutions (line 026); and claimed (line 029). big and revolutions were corrected; phenom remained uncorrected despite multiple fixations; and claimed was abandoned for the inaccurate production clammed. Thus, there seems to be no relation between lack of direct first-pass fixations and the presence of miscues, or between direct first-pass fixations and corrections of miscues. As predicted by transactional theories of reading, it is the cognitive processing of print that drives eye movements, not the eye movements that cause a particular response to print (e.g. Paulson & Freeman, 2003).

221

As an illustration, phenom is a particularly striking example of this dynamic; an unusual colloquial term that P was presumably not familiar with, it was produced inaccurately despite multiple fixations and two attempts at correction. Figure 4.30 shows the fixation progression on phenom (cf. Appendix H, p. H20).

Figure 4.30: Example of a Miscued Word Directly and Repeatedly Fixated Figure 4.31 (cf. Appendix H, pp. H11-H13, H20), shows several examples of the nonlinear relationship between miscues and fixations. The words Jimmy, my, sister, Julia, and (the second instance), David, telling, and me were not fixated; in fact, there are no fixations even within foveal range of them. And yet, none of these words were miscued. By contrast, the word Zimmerman, with the densest cluster of fixations, was produced with two corrected partials before the accurate production. The word crowd, directly fixated on first pass, was produced without any deviation from print; encouragement, which was fixated once directly and once foveally, was miscued.

Figure 4.31-1: Nonlinear Relations of Fixations…

222

Figure 4.31-2: …and Concurrent Productions Figures 4.30 and 4.31-1 also show three other features of P‘s eye movements. Consider fixations 230 and 235 in 4.30. Both constitute abrupt vertical departures from the progression, and they do not seem to serve the purpose of interacting with the word phenom, as they are too far below it for it to be in the foveal region. Similar departures occurred in Figure 4.31-1; some did not co-occur with any OR-ER disparity, such as 253 and 254 above and below Jimmy, while others were part of a struggle, such as 295-297 surrounding encouragement. Additionally, both figures feature several instances of fixation clusters (i.e. fixations that occurred on the same or almost the same spot). Examples for this are fixations 229 and 231, as well as 232-234 and 236 in Figure 4.30. In Figure 4.31-1, fixations 258, 259, and 264 are clustered together, as are 262 and 263; 268 and 269; 273 and 288; 274, 275 and 290; 279281 and 292; and 284-286. There is a closer relationship between clusters and oral productions than between vertical departures and the latter: most clusters tend to co-occur with some type of OR-ER deviation. In Figure 4.31, this included the corrected partial on

223

Zimmerman we discussed above (258, 259, and 264, 262-263, 268-269), and a halting quality after my (274 and 275) and are (285-286) as well as before friends (279-281). Finally, note the incidence of short-range regressions in Figure 4.31-1 (257-259, 262264, 266-267, 274-275, 287-289, 292-294, 295-296). With the exception of 287-289, all of these occurred during some OR-ER deviation. 257 through 267 occurred while P was struggling with Zimmerman; 274 and 275 landed on a word after which P hesitated; and 292296 occurred with his erroneous production enc[o]ragement. We saw a similar relation between short regressions and struggle with the text in Figure 4.29. In sum, P‘s eye movements show three distinct types of patterns, one of which vaguely, the other two more strongly related to OR-ER deviations. In and by itself, this observation is not surprising. All of the patterns we discussed have been reported in the literature. However, it seemed that P produced more of these patterns than comparable readers; for different reasons; and not all of them appear to be strategic. This insight is necessarily anecdotal, but it is instructive to compare his patterns to those reported. Figure 4.32 (from Paulson, 2005, p. 344) shows the features in question on a piece of text the reader was struggling with.

Figure 4.32: Eye Movement Patterns from Paulson (2005) Again, the comparison is anecdotal, particularly so as the reader is an undergraduate college student(i.e. a young adult). Nonetheless, it is instructive to juxtapose reading

224

behaviors. Struggling with a syntactically difficult sentence (...demanded that he will them his house…), the reader in 4.33 clustered and regressed on the difficult phrase he will them, as well as on the preceding verb and relative pronoun demand that, and he went back and forth between those text portions, producing a vertical departure in the process. He then proceeded in a more linear fashion. His eye movements reveal strategic attempts to make sense of the troublesome passage. Compare this to P‘s eye movements shown in Figure 4.33 below (cf. Appendix H, pp. H11-H13, H20). Eye movements on line 009 (blue) and line 010 (yellow) are shown separately for greater clarity of presentation.

009 010

009

009 010

Figure 4.33: Juxtaposition of Eye Movement Patterns and OR-ER Deviations P miscued and struggled on both lines, which is reflected in his eye movements: they show a high incidence of clusters, vertical departures, and short-range regressions. First, consider the phrase have had their turn. He predicted a different syntactic structure than is in

225

the text: have had in. (It is possible that in was actually meant to be the first syllable of enough from the common phrase have enough.) He noticed his mistake and corrected it as his fixations clustered first on had (125-128), then in foveal range below their and had (129135), with several regressions and some departure from the line in the process. In short, his fixation progression shows evidence for active processing and correcting strategies. Likewise, when he miscued on best, he first clustered on the word itself (162-164), then regressed all the way back to Because (one of the few longer-range regressions he made) and refixated the sentence in its entirety (165-170). Finally, he seemed to have struggled on and I go last more than his verbal productions revealed: he produced the clause with just one instance of choppiness (after and), but it took him 20 fixations to cope with it (138-158), including dense clustering, multiple regressions to the previous line, and some vertical departure. Even so, this set of eye movement too revealed strategic fixation behavior, such as the regression to the previous phrase had their turn, or the multiple instances of scrutinizing the troublesome clause. In sum, P makes strategic use of eye movements to cope with troublesome passages. Much like the reader from Paulson (2005), he uses them to find linguistic cues to the text portion with which he is struggling. This is consistent with the EMERGING INTEGRATION theme. However, notice the difference in appearance between both samples. P‘s progressions look ‗messier‘ and more erratic than the other reader‘s; he has much denser crisscrossing of progressions, more departures from the line, and more fixations overall. We could account for this simply by referring to the age difference. But recall that the sheer number of fixations is higher than expected even for P‘s age, and typical for much younger readers. The high incidence of clusters, departures, and regressions is linked to this atypical measure, and

226

therefore warrants attention. Also, there are three other notable characteristics in P‘s eye movements: a) he does not always use them strategically for linguistic cueing; b) the patterns show up even on text passages he does not struggle with verbally; and c) some of his eye movements actually appear counterproductive. I will explore these observations in turn. First, in Figure 4.34 (cf. Appendix H, pp. H11-H13, H20), P struggled with the word halfpipe; he rendered its first syllable as [hælf], corrected it, and produced the second syllable after a brief hesitation.

Figure 4.34: Eye Movements and Corrected Miscue on halfpipe Fixations 54-59 cluster on the troublesome first syllable, 60-62 on the second. No regressions were made from the second syllable back to the first. In addition, comparison of audio and eye movement recordings revealed that the first /p/ of pipe was produced starting around minute 0:13:09:96 (frame 00044322) of the recording. Fixation 60, the first on this syllable, occurred at minute 0:13:09:90 (frame 00044320). In other words, P finished correcting half- before he fixated -pipe for the first time. Taken together, this is strong evidence that P did not attempt to resolve his difficulties through semantic cueing, using the second syllable to figure out the word. Instead, he processed the first in isolation before moving on. This interpretation was corroborated by his subsequent productions of halfpipe. The word occurred three more times in the text. On two of these, P hesitated after the first syllable; on the remaining one, he read the word with ‗choppy‘ quality. Put otherwise,

227

throughout the text he did not integrate his knowledge gained from the first encounter. This observation corroborates the PIECEMEAL PROCESSING theme. Second, P sometimes employed clusters, regressions, and departures yet showed no sign of struggle with the print. In Figure 4.35, for example (cf. Appendix H, pp. H11-H13, H20), he produced the first word of thirteen-year-old without problems, but then hesitated before year. Fixations 227 and 228 on year and old were longer than previous ones, indicating that he was processing the words; the patterns in question, however, show on thirteen. It is unclear what P used the patterns for in this instance; based on our previous insights that he is concerned with a FOCUS ON CODE and STRUGGLES WITH ACCURACY, a reasonable assumption is that he was sampling the visual shape word thoroughly in order to avoid any production struggles.

Figure 4.35: Patterns Indicative of Struggle in Absence of Struggle Third, there are instances of departures from the line which are not only inefficient but actually counterproductive. Consider fixations 112, 122, 130, and 132 in Figure 4.33 above. 112 was part of P‘s struggle with Dozens, which he overcorrected. 122, 130, and 132 were part of his trouble with have had their turn. However, they landed on the line below (112, 122) the problem area, a foveal distance from it (132), or in between lines (130), and none of them contributed to resolving the problem. There is no information in last (112) that has any bearing on the pronunciation of Dozens; the or the best (122) do not help in processing have had their turn; and whether best or right were in P‘s foveal window on fixation 132, neither would have been helpful. We do not know exactly about fixation 130,

228

but it cannot have been more informative than what was looked at during fixations 129 or 131, which were in foveal distance of the target phrase. Despite their lack of import, however, P devoted considerable attention to these spots, as evidenced by fixation durations. Out of the fixations related to Dozens (109-116), 112 was the longest at 0.37 seconds. Of the fixations while processing have had their turn (121-137), 122 and 130 were the second longest at 0.3 seconds each; 132, equally unproductive, was the longest at 0.33 seconds. Figure 4.36 shows the fixation sequences in question (edited for presentation).

Figure 4.36: Fixation Durations on Dozens (left) and have had their turn (right) There are other instances of the same phenomenon. Consider fixations 297 in Figure 4.31, part of P‘s processing (and eventual miscuing) of encouragement. Out of the entire sequence (fixations 292-297), 297 was the longest at 0.5 seconds. And yet, it did not appear to serve any purpose, as it landed on blank space. Similarly, fixations 333 and 334 in Figure 4.37 below (Appendix H, p. H20) did not contribute, in any apparent way, to the struggle with momentarily (eventually produced as moment¦arily, cf. Figure 4.26), yet 333 was the third longest fixation in the sequence (327-335) at 0.6 seconds. And at 1.34 seconds, 334 was, in fact, the longest fixation P made during the entire reading of text two.

229

Figure 4.37: Fixation Progression and Durations on momentarily Viewed by itself, this phenomenon could be seen either as indicative of confusion, or as inefficient implementation of linguistic/contextual cueing strategies related to P‘s emerging integrative proficiency. However, in the context of his overall reading behaviors an alternative explanation is possible. We have observed that P uses clusters, short-range regressions, and departures from the line in four distinct ways: a) to sample linguistic information on text portions he is struggling with; b) to sample visual information on text portions he is struggling with; c) to visually sample text portions he is not struggling with; and d) to fixate spots on the stimulus devoid of relevant information. Taken together, this seems to indicate an intense preoccupation with sampling visual information—so intense, in fact, that at times he even searches for information where there is none available. And the uses he makes of this set of strategies—a) to d) listed above—range all the way from effective (if not always efficient) to ineffective and inefficient. If this interpretation is accurate, it would help explain the characteristics of P‘s eye movements; the high number of fixations, the comparably short average fixation duration, and the fact that the percentage of words fixated is typical but the ratio of function vs. content words skipped is not. Regarding the former two, if his main strategy in engaging with print is

230

visual sampling, we would expect a high incidence of fixations. We would also expect a low average duration, especially when reading out loud; there are time constraints if the reader wishes to maintain a relative naturalness of reading flow. This would also explain why his average durations were shorter when reading aloud as compared to silently, which is at odds with reported measures. It is instructive to consult the literature at this point. Kim, Knox, and Brown (2007) had an eight-year-old read The Boat in the Basement, a short text with inbuilt errors. Figure 4.38 shows the reader‘s eye movements during the first reading (top) and during a second reading (bottom), after the researchers had ―invited her to look for ‗errors‘‖ (p. 51).

Figure 4.38: Fixation Progressions on First Reading vs. Search for Errors

231

The authors report that the reader‘s fixations were longer on average during the first than during the second round: 0.506 seconds as compared to 0.462 seconds. The reason for this could be either the repetition (with differences in durations reflecting differences in familiarity with the text) or the difference in expectations: ‗searching‘ eye movements may differ in length from ‗reading‘ eye movements. It could also be a combination of both. Either way, P‘s eye movements were notably shorter than either measure. Importantly, some of his more conspicuous progression patterns resemble those Kim et al.‘s (2007) reader made in the second round (e.g., the crisscrossing of saccades over discovered and the various vertical departures). The import here is that P‘s eye movements look more like search behavior than like typical reading, in their progression patterns as well as in their durations. Finally, consider what I have termed ‗settling in behavior‘. As seen, P frequently does not tackle a new line simply by doing a return sweep and then progressing along the line. Instead, he fixates within the line before going to its beginning, and/or produces several short-range saccades around the beginning until he has ‗settled in‘ to the line and begins the actual progression. Figure 4.39 (cf. Appendix H, p. H20) shows slide three of text two, where he engaged in this behavior on seven out of 12 lines (lines 1, 2, 6, 8, 9, 11, and 12; the pertinent fixations are circled with a dashed line).

232

Figure 4.39: Examples for ‘Settling in Behavior’ In the light of our present discussion, it is tempting to interpret this behavior as yet another aspect of his proclivity for visual search. That is, it looks as if he were trying to find the best point to start the line from, and absent other strategies, he uses visual information to this end. In sum, it would appear that one of P‘s main strategies when engaging with text is to search for any kind of visual information that might help him in processing it. This claim is in line with the uses he makes of his eye movements, and with their overall characteristics when compared to typical measures, all of which we discussed above. Where search for visual input is predominant, we would expect more fixations and shorter durations than typical. In addition, he frequently fixates in the vicinity of words, possibly to view them from

233

different angles, and he searches for information in areas where there is none available. Both observations account for the high number of fixations in conjunction with the facts that the percentage of words fixated was typical and that there was a high number of fixations within foveal range of words. Finally, viewing P‘s efforts as a visual search strategy would explain why there was a higher incidence of regressions in text two than in text one; when faced with an oral reading task, P may intensify his search for information to help him cope with the demands of accurately rendering the print. EMMA of text two therefore enables us to formulate a new theme in P‘s reading, which I call PREOCCUPATION WITH VISUAL SEARCH. The fact that P‘s ratio of function to content words skipped is lower than typical seems to be related to this behavior. Readers usually skip a large portion of function words because they are predictable (Paulson & Freeman, 2003, pp. 16-17); it is more efficient to use one‘s own linguistic knowledge to infer these function words than to sample them visually. It would not be surprising for P to be less proficient at doing this, given his history of language difficulties; indeed, we previously established STRUGGLE WITH PREDICTION and HIGHER-LEVEL LANGUAGE ISSUES

as pertinent themes. This would explain why he fixates more on function

words than is typical, and speak to his habit of PIECEMEAL PROCESSING. There is no such clear-cut explanation for the fact the he skipped more content words than typical readers. But our observations are in line with reports by van der Schoot, Vasbinder, Horsley, and van Lieshout (2008), who found that poor comprehenders spent less time than good readers fixating important words in a text (i.e., core content words), but looked longer at anaphors (a type of grammatical relations) than controls. It would therefore seem that lower ratios of fixations on content words are a general characteristic of poor

234

comprehenders, and it is reasonable to assume that this trait is linked to their higher-level language issues, even though the connection between these two phenomena is unknown at this point. By contrast, we can draw a direct connection between PREOCCUPATION WITH VISUAL SEARCH

and the previous themes. FOCUS ON CODE, STRUGGLE FOR ACCURACY, and PIECEMEAL

PROCESSING

would seem to be natural correlates of the former: when text is processed in bits

and pieces, and high importance is placed on rendering code accurately, visual search for information becomes more important than activation of linguistic and background knowledge. Likewise, HIGHER-LEVEL LANGUAGE ISSUES, which implies STRUGGLE WITH PREDICTIONS,

may lead to higher emphasis on visual and code-based strategies; conversely,

such an emphasis may hinder development of higher-level language processing. As we saw in the EMERGING INTEGRATION theme, P shows clear signs of beginning use of integrative, higher-level strategies, possibly a result of the intervention he is receiving. But nonintegrative, lower-level approaches such as VISUAL SAMPLING or PIECEMEAL PROCESSING are still his prevalent mode of interaction with text. Another piece of evidence for this interpretation comes from the lack of variability in P‘s fixation durations, a measure we explored above for text one. Much as in the previous comparison, the variability in P‘s individual fixation durations appeared to be fairly typical. When averaged by paragraph, however, a different picture emerges. Figure 4.40 juxtaposes the average fixation duration by paragraph in P‘s reading of text two with the same average durations in the reader studied in Paulson (2005).

235

Seconds

Paragraph number

Figure 4.40: Comparison of Variability in Fixation durations per Paragraph As mentioned, Paulson‘s (2005) reader shows a spread from just over 0.3 to 0.6 seconds. P, by contrast, has an even smaller range in text two than in text one; his average durations per paragraph run only from (roughly) 0.18 to 0.24 seconds. We saw that Paulson explicitly connected lack of variability in durations with lack of meaning construction while reading. This prediction was confirmed for text 1, and it can again be confirmed for text two—as we will see with discussion of the retell below—albeit with the same caveat as made previously; it is not lack of variability in individual fixations but in average durations over parts of text, such as paragraphs, that indicates lack of comprehending. As for the retell, the same observations as previously were made. P‘s rendering of the text was disjointed, lacked story line and narrative frame, and suffered from omission of core ideas. I render it in its entirety here, omitting two hearer signals of mine. P:

um Jimmy had a girlfriend and gave her a wink, and then the crowd yelled ―I love you, Jimmy!‖. And Jimmy wasn‘t (gettin‘) second place. And that‘s it.

R:

anything else?

P:

uh-uh.

236

R:

you wanna go back to the text and have a look at it?

P:

nobody ever did three complete revolutions in the air before and came down with both feet on their board. (pause) hmmm. (long pause) It‘s the X Games and the whole world is watching. Or the whole skateboarding world. (pause) This is my ultimate fantasy. Done! (Appendix H, p. H22)

Instead of the main elements such as setting and scenario, P reported an out-ofcontext fact first (Jimmy had a girlfriend and gave her a wink), mixed up two pieces of textual information (it was the girlfriend, not the crowd, who shouted I love you), and mentioned main elements only after being prompted to look back at the text; even then, he rendered them in reverse order, namely, the order in which he looked at them, which speaks to the idea that his main strategy when interacting with text is visual sampling. However, there are three noteworthy observations regarding the retell. First, it does not feature any lexicogrammatical oddities, possibly due to the relatively low difficulty level of the text. Second, he did mention the main elements of the narrative, albeit out of order and only after prompting to look back; this is reflected in his relatively high score on the retell form (39/100, Appendix H, p. H24). And this does not appear to be a random stroke of luck. As previously, he read the text verbatim when looking back, suggesting lack of integrative engagement. However, as previously he did fixate text portions that he chose not to render as part of his retell. Figure 4.41 (cf. Appendix H, p. H21) shows the most significant of these: he scanned the entire slide systematically to discern whether or not it contained relevant information. The audio track of the eye gaze video reveals that he was uttering hmmm when looking at the third and fourth line of the text, which provides additional evidence that he

237

was actively processing it. This observation is in line with both our PREOCCUPATION WITH VISUAL SEARCH

and EMERGING INTEGRATION theme in that P does actively process text, albeit

with a heavy emphasis on visual sampling absent linguistic integration, and hence his attempts at ‗connecting the dots‘ between text portions are frequently unsuccessful.

Figure 4.41: Text Fixated During but not Included in Retell A third and final observation relates to his metatextual comments. He signaled twice during his brief retell that he had completed it, commenting and that‘s it after the first portion; done after the lookback part. After the first signal, he affirmed his wish for closure with uh-uh when I asked him if there was anything else. Previous retells included similar comments. The import of this observation would appear to be that he seems to feel a strong urge to be ‗done‘. While he can be quite persistent while reading out loud, he seems to want to finish retells as quickly as possible. This is understandable in the joint light of his integrative difficulties and his awareness issues discussed in the interview section above. There is too little data to confirm this conjecture, but we may suspect that his rush to finish retells may be linked to his EMERGING AWARENESS of his comprehension issues. That is, he

238

may be realizing that he is on uncertain territory while retelling, and trying to get out of there sooner rather than later. In sum, the themes from the home readings could be confirmed for EMMA texts one and two, and a newly emerged theme was added to the list. Table 4.16 shows the current list of themes.

Theme

Main indicator

FOCUS ON CODE

OR-ER similarity

HIGHER-LEVEL LANGUAGE ISSUES

2/3 of miscues indicate lexicogrammatical weakness (low-quality miscues)

STRUGGLE FOR ACCURACY

Incidence of misarticulations

PIECEMEAL PROCESSING

Pauses, hesitations, ‗choppy‘ reading, fixation clusters on small units of language

STRUGGLE WITH PREDICTIONS

Expectation of syntactic structures overrides meaning

EMERGING INTEGRATION

Some evidence of integrative strategies when interested in text

PREOCCUPATION WITH VISUAL SEARCH

Excessive sampling behaviors, untypical measures

Table 4.16: Joint Themes from Home Readings and EMMA Texts One and Two EMMA text three: Expository text read aloud. For text three (cf. Appendix H, pp. H25-H27), P read a total of 332 words (not including the words on the last stimulus slide, which consisted of an image with a few captions). On these, he produced 20 miscues, which equals 6.02 miscues per 100 words (MPHW). This is consistent with his home reading performance, albeit not with his performance in EMMA text two. As we will see, the bulk of his miscues came from code-based productions of genre-typical words that were presumably unfamiliar to him; the higher number of MPHW can therefore be explained with the genre difference between texts. This is also reflected in the pattern analysis. In this text, 85% of his

239

miscues resulted in meaning loss; the same percentage indicated grammatical weakness while 15% did not hamper meaning and indicated grammatical strength. All miscues for which this pattern was applicable were highly similar to the ER both visually and phonemically. Table 4.17 sums up the results of the pattern analysis for text three.

Pattern

%

Meaning construction No loss Loss

15 85

Grammatical relations Strength Weakness

15 85

OR-ER similarity Graphic similarity High

100

Sound similarity High

100

Table 4.17: Pattern Analysis for EMMA Text Three As was done previously, these patterns were grouped to determine the number of lowquality miscues. In text three, the grouping was particularly clear-cut; all miscues which led to meaning loss also indicated weakness in grammatical relations, and all miscues which did not lead to meaning loss also indicated strength in these relations. Additionally, the percentage of low-quality miscues (combined semantic and syntactic issues was higher than

240

in any previous text, speaking, again, to the genre-related relative difficulty of the vocabulary. Table 4.18 sums up pertinent findings.

Pattern

%

Meaning loss, weakness in grammatical relations

85

No meaning loss, strength in grammatical relations

15

low qual. % 85

Table 4.18: EMMA Text Three Miscues Grouped by Combined SemanticSyntactic Alterations Taken together, these findings corroborate our FOCUS ON CODE and HIGHER-LEVEL LANGUAGE ISSUES themes

listed above. Pertinent miscues include, for example, $['dɛspɪt] for

despite (line 009), $['dɹəʊni:s] for drones (line 024), $['wɛvɝ] for weaver (line 030), or $ch[æ]mber for chamber (lines 048 and subsequent). In addition to such nonsense words, P also produced an ungrammatical sentence when he rendered the phrase to let air in as to let air, omitting the preposition (line 050) (cf. Appendix H, pp. H29-H35.) P‘s STRUGGLE FOR ACCURACY and his PIECEMEAL PROCESSING are evident in various text portions, such as the first four lines shown in Figure 4.42 below (Appendix H, p. H28). The various vertical lines stand for choppiness or hesitations, indicating piecemeal processing of the text; the lengthening of the [v] in every and the production of steamy as the nonsensical phrase steam me testify to his struggle to accurately render the written words. Note also that he read over the two commas, which suggests that he does not attend to normal prosodic boundaries while reading out loud.

241

Figure 4.42: Examples for STRUGGLE FOR ACCURACY, PIECEMEAL PROCESSING An intriguing example for both themes comes from his production of the phrase The queen ant (lines 050-051) (Appendix H, p. H36). Possibly because he encountered the word ant only after the line break, he did not seem to recognize it as part of the noun phrase: while producing it, he hesitated before articulating the [t], which made it sound as if he had been articulating it as and. He then repeated the entire phrase, producing the [t] as an ejective for emphasis. The within-word hesitation speaks to the piecemeal fashion in which ant was processed: only when tackling a word one bit at a time is it possible to hesitate within it, particularly in the case of a single-syllable word like ant. Likewise, it speaks to P‘s concern with accuracy: he would not even finish the word in whichever way he was about to pronounce it, but hesitated until he had recognized the word-final letter, and then repeated the entire phrase with added emphasis on the troublesome letter to make clear to the listener that he had resolved the accuracy problem. Figure 4.43 shows the passage in question.

Figure 4.43: Another example for STRUGGLE FOR ACCURACY, PIECEMEAL PROCESSING

There were a few instances of STRUGGLE WITH PREDICTION in text three. P‘s production of nests in Most ant colonies live in nests (line 026, Appendix H, p. H32) could be

242

interpreted this way. He first read the word as nest, then attempted to correct this, unsuccessfully rendering it as $nest[ɪs]. It can be argued that his failure to predict the plural form could be subsumed under this theme, although it could also be categorized under FOCUS ON CODE.

Another incident that suggested predictive issues was P‘s production of lines 044

and 045, shown in Figure 4.44 below (Appendix H, p. H34). P struggled with the sentence, An ant‘s nest is its home, as evidenced by the hesitations he inserted after almost each word. He then produced the next phrase, Just like your home, with inappropriately weak stress on your, indicating that he did not attend to the implied contrast between an ant‘s home and your home. Further corroboration comes from his struggle with the word home, evidenced by the corrected partial h-. The structure of the sentence could have led him to predict occurrence of home, which he failed to do. Finally, he repeated an in the phrase an ant‘s nest, despite the fact that the exact same construction occurred only one line prior. This indicates, again, that he did not use previous information to predict upcoming text.

Figure 4.44: Examples for STRUGGLE WITH PREDICTION As previously, there is some evidence that P has begun to use some integrative strategies, corroborating the EMERGING INTEGRATION theme. One such instance is his correction of the nonword ['fiməl] to female (Appendix H, p. H31). Or consider the metatextual comment P made before beginning to read: he glanced at the first text slide (cf. Figure 4.42 above) and commented: ―isn‘t there another sentence after that?‖ referring to the

243

fact that the last line of the slide ends mid-sentence. This suggests active processing of language to the point of discerning that the sentence was incomplete. As regards measures of eye movements, the data for text three was once more skewed due to technical difficulties. At the beginning of the fourth stimulus slide (cf. Appendix H, p. H37), fixations and point of gaze were lost because P had displaced the monocle of the headworn eye tracker. After readjustment, the pattern of his fixation progressions changed, as can be seen in Figure 4.45 below. Prior to the incident (top left part of the figure), fixations traced the printed lines faithfully. After readjusting the monocle, progressions no longer appeared to match the lines and the oral productions; also, these deviations appeared to be idiosyncratic and unsystematic. Consider the top right part of the figure, which represents the lower part of the slide on the first line of which the incident occurred. At first glance, the fixations appear to match the printed lines; however, P struggled both on Female worker ants and on drones, and yet these pieces of text are barely touched by fixations, while there are clusters on parts of the text P did not struggle with. Together with the observation in the lower part of the figure, where the first line appears to have been passed twice but the last line remains unfixated, this suggests a recording error.

244

Figure 4.45: Evidence for Systematic Recording Error For these reasons, it was decided not to use measures that depend on accuracy of fixation location, such as percentage or type of words fixated. As in text one, only general measures independent of fixation location were used. Table 4.19 lists the results for eye movement measurements on text three. Note that of the 689 fixations in the recording, the following were not included in these calculations: no. 151-167, as they were recorded while the monocle was readjusted; and 638-689, as they occurred not on continuous text but on the picture which covered the entirety of the last stimulus slide. Fixations that occurred on pictures next to the text were included in average fixation duration, range of duration, and overall number of fixations, but not in fixations on text, fixations per 100 words, and regressions. (cf. Appendix H, p. H42).

245

Measure

Result

Reading time

308.67 sec

Average fixation duration

0.198 sec

Fixation duration range

0.1—0.87 sec

# of words

332

# of fixations (overall)

620

# of fixations on text

611

# of fixations per 100 words

184 (rounded)

# of regressions

80

% of regressions

13.1% (rounded)

Table 4.19: Eye Movement Measurements for Text Three While reading text three, P‘s average fixation duration was almost identical to the previous oral reading (0.198 as compared to 0.2 seconds). His fixation durations ranged from 0.1 (the threshold) to 0.87 seconds, which is similar to the silent reading of text one (0.1 to 0.9 seconds) but lacks the higher durations of text two (up to 1.34 seconds). He produced 184 fixations per 100 words, which is consistent with his prior performances (178 and 188, respectively). 13.1% of his saccades were regressions, a value somewhat higher than in the silent reading of text one (11.94%) but lower than in text two (18.4%). This last finding is the only one that is not fully consistent with prior ones. It was stated previously that P may have tackled the additional task demand of oral reading by engaging in more visual search strategies, as evidenced by the higher number of fixations per 100 words and the higher percentage of regressions in text two as compared to text one. The number of fixations is similar in text three but the percentage of regressions is markedly lower. One explanation for this is the relatively higher readability of text three due to larger font and shorter lines. Another hint for solving this puzzle lies in the number of miscues

246

(Appendix H, pp. H25-H27). The number of miscues per 100 words in text two was inconsistent with the home readings in that it was much lower than in the latter (2.94 as opposed to 6.1 and 6.65, respectively). MPHW in text three was consistent with the home readings. This might indicate that P was trying particularly hard to ‗get it right‘ while reading text two, to which end he may have used a higher ratio of regressions. In either case, the measures for text three show the same discrepancies as previous ones. That is, P‘s fixations were shorter, and included fewer regressions, than a typical adult‘s; at the same time, incidence of fixations as compared to number of words places him at a first-grade level. This corroborates our PREOCCUPATION WITH VISUAL SEARCH theme. Due to the technical difficulties discussed above, it was not possible to do an actual EMMA on text three. However, as text three was the only one to contain pictures, one observation could be made which confirmed the EMERGING INTEGRATION theme: P does make efficient use of eye movements to integrate picture information while reading. Figure 4.46 shows slide eight of the stimulus (Appendix H, p. H37). P fixated below the first word of the text (559), then veered downwards to the picture (560), as if to ascertain if there was any pertinent information in it. As there was not, he went back to the text until he arrived at holes (574), at which point he sampled the picture again (575, 576), possibly because it shows a hole in the ground. He then read on until fixation 596. Due to the recording error mentioned above, we cannot be sure if this fixation is indeed on queen, as it appears to be; judging by progressions, it may well be on the line below, (i.e. on chamber). However, either word would be a meaningful prompt to look back at the picture, which features both an ant (if not the queen) and something that could be a chamber. P produced three fixations (597-599) on the ant and then went back to the text. Apparently he had sufficiently sampled the picture at

247

this point to know that it would not provide helpful information in dealing with the rest of the slide, hence he read on without going back to the picture. This behavior shows efficient use of pictorial information, speaking to our related theme.

Figure 4.46: Efficient Sampling of Picture Information Finally, the retell needs to be considered. As previously, P‘s rendition of the text was incoherent, lacked structure, and did not convey core information. The retell is shown in its entirety here. P:

the uh ant queen is (pause) the most… ant that… is the king and the… king and… other ants live in homes and have spaces in their homes. And… they keep um they find these (scent)… food… where they find food where they‘re hungry. And that‘s… (let‘s go check) (looks back at text) ―The queen ant lives in one chamber.‖ (pause) ―several chambers‖ wait um ―food is stored in several chambers.‖ ―They give off different scents to communicate different messages.‖ ―Most ant colonies live in nest.‖ ―Female worker ants and male ants called ['dɹəʊni:s] live in the colony too.‖ Done.

248

R:

alright. Very good. Anything else that might be important in there?

P:

uh-uh (Appendix H, p. H38)

P mentioned the queen ant but inserted an imagined king to go with it; he did not fully grasp the idea of an ant colony (other ants live in homes implies that they live apart from king and queen); and he produced bits of information which were grammatically and semantically incomplete (they keep um they find these (scent)… food… where they find food where they‘re hungry). Interestingly, he did not wait for the prompt to look back at the text this time. Instead, he tackled this portion independently, even commenting on it. As previously, he ended his retell with a comment on its finality (Done) and claimed to not remember anything else when asked. This confirms our caution as regards his integrative abilities: they are, at best, EMERGING. This is true despite the fact that he got a relatively high score on his expository retell form (39/100, cf. Appendix H, pp. H39-H40). The retell analysis form used in the miscue analysis manual places high value on specific information, with less regard to their integration into a meaningful whole. As seen previously, P is somewhat proficient at rendering printed information in this way, in particular when allowed to look back at text. Overview and Preliminary Axial Coding of Results Based upon the data collected and analyzed, it is now possible to bring together the insights gained from the different methods of inquiry used. Table 4.20 lists the themes in question. To link them to the research questions that sparked this study, those that emerged from the interviews (with support of artifact and behavioral analysis) are subsumed under the primary aspects with which they dealt; namely, the social and psychological sides of P‘s poor

249

comprehension. The themes gleaned from miscue analyses and EMMA are listed as linguistic-behavioral. Social & Psychological

Linguistic-Behavioral

READING VS. COMPREHENDING TANGIBILITY

FOCUS ON CODE

(LACK OF) SUPPORT

HIGHER-LEVEL LANGUAGE ISSUES

HALF THE STORY

STRUGGLE FOR ACCURACY

ISSUES EXPRESSING HIMSELF

STRUGGLE WITH PREDICTION

LACK OF MOTIVATION & NEED FOR

PIECEMEAL PROCESSING

SCAFFOLDING/STRUCTURE

EMERGING INTEGRATION

OTHER-ORIENTATION

PREOCCUPATION WITH VISUAL SEARCH

EMERGING AWARENESS Connotative theme: UNCERTAINTY

Table 4.20: Overview of All Themes Derived from the Data After generating thematic categories, the next step in devising a Grounded Theory is to organize them into an explanatory whole through a process of ‗axial coding‘. As mentioned in Chapter Three, two major ways of axial coding are discussed in the literature: organization around a central axis (Charmaz, 2006, pp. 60-63), or ‗crosscutting‘ (i.e. relating themes from distinct categories through propositional logic such as is the same as or is similar to). This latter approach was discussed by Corbin and Strauss (2008, pp. 195-228). In the case of P, I began to organize my findings around a dynamic axis early in the process, organizing them along P‘s trajectory as it emerged during interview analysis, from early hearing difficulties to poor comprehension in fourth grade. I devised the first outline of such an organization while coding the interviews (cf. Memo Seven, Appendix B, p. B9), before I had completed the pertinent list of themes and before even starting analysis of the readings. 250

The result, which I called a ‗hypothetical outline of P‘s path to poor comprehension‘, is shown in Figure 4.47 below. (As it was drafted early in the analysis process, it counts as data in the tradition of Grounded Theory, and hence is discussed in this Chapter).

Failure to develop comprehension monitoring & strategies— ―needs somebody to check comprehension‖

Learning to read as ―people pleasing‖? Doing what is asked ―when reading starts‖?

Hearing problems

School concerned with reading out loud; mother concerned with comprehension, but also experience of uncertainty re problem; not enough support

Language problems— doesn‘t understand what‘s going on, can‘t contribute Passivity, immaturity; ―people pleasing‖

ADD medication helps with doing what is asked but not with independent tackling of comprehension issues

Independent activity where no higher-level conceptual thinking is needed; passivity where such thinking is needed

Figure 4.47: A Hypothetical Outline of P’s Path to Poor Comprehension The starting point of the hypothesized trajectory comes from P‘s mother‘s depiction of his early communication problems. His trouble clearly began with early hearing difficulties, leading to problems in language acquisition and to a persistent experience of communicative failure—not understanding, not being understood. Such an experience may plausibly lead to passivity in social and academic relations, and to general social immaturity and ‗people pleasing‘ behavior. If you fail to understand much of what is going on around you, and to make yourself understood, motivation may become restricted to activities that do not require communicative success, as we have indeed seen in P‘s mother‘s description of his interests. Likewise, development of social maturity, such as competencies in Theory of Mind

251

and self-awareness, requires age-appropriate social interaction and interacts with language development (e.g. Astington, 1994; Hardcastle, 2003). Where such maturity cannot develop due to communication difficulties, ‗people pleasing‘ would, again, be a plausible strategy to achieve a minimum of social involvement and success. In an environment where quantifiable behaviors are prized and expected, then, a ‗people pleaser‘ with little or no experience of meaningful interaction with print may end up fulfilling the expectations to produce such behaviors—primarily accurate oral rendering of print and verbatim recapitulations as responses to ‗comprehension questions‘, as seen during his retells—without ever feeling need or motivation to construct meaningful mental representations. Learning to read, for P, may have been limited to the continuous effort of accurately responding to others‘ expectations, to do what he is supposed to do whenever ‗reading starts‘. Importantly, such a way of engaging with print would not require comprehension or monitoring of comprehension beyond what is needed to come up with the expected response. It would make sense, in this light, that his mother should see a need for ‗someone to check his comprehension‘. If the only reason to construct meaning from text is to fulfill others‘ expectations of doing so, then meaning construction will not occur until prompted. The same applies to metacognition. Where the ultimate yardstick for success is others‘ evaluation, monitoring if comprehension has occurred for oneself is unnecessary. This might also explain why intervention has not yielded the expected results. As noted, P‘s private SLP seems to foster independent engagement with text. And yet, this has not translated into independent attempts at comprehending at school. If the assumption is correct that P has a deeply engrained habit of other-orientation, then any intervention geared towards weaning him off that habit may run into a paradox similar to the classic ‗spontaneity

252

problem‘ (Watzlawick, Beavin Bavelas, & Jackson, 1967/2011). For P, being guided towards independence through intervention may result in a need to be prompted to interact with text independently, defeating the purpose of the intervention. Much has been written about the relation of ADD/ADHD to higher-level cognitive skills and literacy problems (see Westby, 2010, for an overview). Many aspects of the disorder could be viewed as an explanation for poor reading comprehension; indeed poor comprehension is listed as one of its aspects (Westby, pp. 534-535). In P‘s case, however, his mother‘s verdict is unequivocal: While ADD medication has helped him to cope with immediate task demands (i.e. to meet others‘ expectations), it has not improved comprehension. This indicates that P‘s ADD was not causally implied in his comprehension problems; his failure to engage in higher-level thinking has persisted. It is not surprising, then, that he should engage in independent activity only as long as it does not require such thinking. If he never experiences the pleasure of understanding (cf. Gopnik, Meltzoff, & Kuhl, 1999, pp. 162-164), there is no reason to seek it out and tackle reading independently. Drafted early in the process, the ‗hypothetical pathway‘ diagram did not include the themes that emerged from the analyses of P‘s reading behaviors. It does, however, provide us with a first step towards axial coding and integration of those latter themes with those gained from the interviews and related data. It therefore concludes our exposition of results, and leads us into the discussion of the findings, which is addressed in Chapter Five.

253

Chapter Five: Discussion and Interpretation of Results No complex system, whether biological or social, can be understood without considering its history. Cilliers (1998, p. 107) The First Law of Ecology: Everything Is Connected to Everything Else. Commoner (1972, p. 33) As we move into in-depth scrutiny and interpretation of the data discussed in Chapter Four, the quotes above seem out of place in their simplicity, almost as if they were platitudes. Of course, one feels compelled to reply, every phenomenon has a past that helped shape it, and a multitude of connections to other phenomena in the present. As we saw in Chapter Two, though, these simple insights have been neglected in the study of poor comprehension, leading to confusion. In this chapter, I will attempt to show that the phenomenon is understandable (if not explainable—cf. Bransen, 2001) when psychosocial-behavioral developmental dynamics and reader-environment interactions are accounted for. To this end, I will first discuss the themes from Chapter Four in terms of P‘s developmental dynamics, and then ‗re-describe‘ (Bransen) my observations using the terminology of systems theory. I will then use an ecological metaphor in my attempt to formulate a tentative grounded theory on the phenomenon. Poor comprehension, I will argue, can be conceptualized as a psychosocial-behavioral ‗niche‘ into which readers drift naturally (Maturana & Varela, 1992) when pertinent conditions are in place. The Developmental Dynamics of P’s Poor Comprehension As seen in Chapter Four (Figure 4.47), the information gleaned from interviews with P and his mother provided insight in P‘s history early in the process of data analysis. A hypothesized trajectory could be posited which plausibly reconstructs P‘s development from early hearing problems to persistent poor comprehension. After finalizing the thematic codes

both for the social-psychological and the linguistic-behavioral aspects of P‘s problems, we can now combine this early outline with the themes gleaned from our analysis, modifying it to accommodate the latter. Figure 5.1 shows the result: a theme-based outline of P‘s developmental trajectory. It is organized as a timeline, along which the themes from Chapter Four are ordered into larger macro-themes.

Behavioral and psychological consequences HIGHER-LEVEL LANGUAGE ISSUES STRUGGLE WITH PREDICTION PIECEMEAL PROCESSING PREOCCUPATION WITH VISUAL SEARCH LACK OF MOTIVATION & NEED FOR SCAFFOLDING/STRUCTURE

The puzzle of intervention UNCERTAINTY (LACK OF) SUPPORT EMERGING AWARENESS, EMERGING INTEGRATION

Social-psychological and behavioral response OTHER-ORIENTATION FOCUS ON CODE STRUGGLE FOR ACCURACY Hearing problems

Academic experience READING VS. COMPREHENDING TANGIBILITY

Language problems & psychosocial effects HALF THE STORY ISSUES EXPRESSING HIMSELF

Figure 5.1: A Thematic Outline of P’s Trajectory Like its predecessor in Chapter Four, this trajectory starts with the obvious medical and developmental aspects: P‘s hearing issues and the ensuing language problems. As we ascertained during the interviews, those did have a major impact on his psychosocial development, resulting in ISSUES EXPRESSING HIMSELF which at the time of this study were still so pervasive that they represented a genuine barrier to insights into P‘s lived experience (the

255

HALF THE STORY

theme). Burdened with this problem complex, P entered an academic

environment that treated skills in reading out loud as separate from comprehension, and stressed tangible measurements of children‘s progress (READING VS. COMPREHENDING and TANGIBILITY).

We have to be careful, at this point, not to place unsubstantiated blame for P‘s

troubles on his school; we do not know enough about the extent and scope of comprehension instruction he experienced to draw any clear-cut conclusions. That said, the fact that he received reading intervention specifically targeted at reading out loud leaves us with the reasonable assumption that the school seems to have placed great importance on measurable skills related to oral reading proficiency. Regardless of the respective quantity and quality of oral reading and comprehension instruction, one insight seems fairly certain: P responded to the combined pressure of expressive difficulties, social issues, and academic demands with passivity and a proclivity for OTHER-ORIENTATION, both within and outside of the classroom. Insofar as reading was associated with expectations of accuracy, he developed a habit of FOCUSING ON THE CODE and STRUGGLING FOR ACCURACY,

but not of constructing meaning from print. This was enough for

him to get by; apparently, his school views the ability to render disconnected bits of information as sufficient to advance to fourth grade. As a result, P wrestles with a number of behavioral and psychological difficulties when reading. He does not make much use of HIGHER-LEVEL LANGUAGE SKILLS,

and when he does use such skills to predict upcoming text,

his PREDICTIONS are frequently wrong. It is not surprising, therefore, that he should stick to PIECEMEAL PROCESSING

of print. That is, absent more complex strategies, this is the one safe

option he has at hand. To better cope with visual information (and to make up, as it were, for the lack of linguistic strategies) he has developed a ‗VISUAL SEARCH‘ strategy of using eye

256

movements to sample as much information visually as possible. All of which allows him to get through text accurately enough and with a sufficient stock of information bits to get by, but not to construct coherent meanings that would meet the definition of actual comprehension, let alone skills related to monitoring his own comprehension. It is plausible that this should lead to a LACK OF MOTIVATION to engage with text in the absence of an explicit task demand; his mother has labeled this NEED FOR STRUCTURE and assumes he needs intensive, one-on-one SCAFFOLDING to overcome it. This leads to the final macro-theme of P‘s trajectory that might be labeled, the puzzle of intervention. I mentioned that I perceived a general sense of UNCERTAINTY around P‘s issues. Nobody in his environment seemed to understand his needs to the point of being able to provide enough help to substantially change his engrained habits. Hence, he was still struggling at the time of data collection despite the fact that he was receiving considerable and variegated SUPPORT. He did exhibit some growth in the areas of general AWARENESS of his comprehension issues as well as in strategies of INTEGRATION he was beginning to use, but the pace of his progress appeared rather slow when compared to the successful short-term interventions reported in various studies (cf. Chapter Two). The best explanation for this observation that I had at this point in the inquiry was that his engrained habits of other-orientation and of failure to construct and monitor meaning for himself create a paradoxical feedback loop, symbolized above by the final arrow going back to the social-psychological and behavioral response macro-theme. This suggests that intervention targeting comprehension, no matter how well-crafted, may founder on the fact that it inevitably includes yet more expectations placed on P from without. Thus, whichever efficient and effective strategies he may learn during intervention may simply not be carried

257

over to times where he would have to tackle text in a truly independent fashion; as long as he sticks to complying with others‘ expectations, any intervention will simply confirm this pattern, unless there is a substantial change either in his environment—such as the introduction of expectations to comprehend—or, preferably, in his own habits. If he could be alerted to the beauty of meaningfulness, my reasoning went (cf. Gopnik, Meltzoff, & Kuhl, 1999, pp. 162-164), he may be able to surpass his limitations and explore new ways of interacting with the written word. To investigate this hunch, and possibly to find further pointers as to why P did not appear to make progress in therapy, I reviewed a case of poor comprehension whom I had diagnosed two years prior to the current study, and who had effectively functioned as a sensitizing case for me (cf. the discussion of ‗sensitization‘ in Chapter Three). In addition to comparing my then assessment to the findings regarding P, I collected additional data from P‘s private SLP, who also happened to have worked with the sensitizing case (henceforth referred to as S). During informal conversations, I had learned that S, in contrast to P, had benefited from intervention; I contacted the SLP to inquire into her explanation for this differential response. Addressing the Puzzle: Comparison to a Sensitizing Case In this section, I will compare the two cases in question. First, the clinical data on S will be reviewed; next, the newly collected data from the private SLP will be added. In the Grounded Theory (GT) tradition, such deliberate inclusion of new data is termed ‗theoretical sampling‘; it precedes the formulation of the actual grounded theory and serves to ‗saturate‘ existing codes until no new insights emerge from them (Charmaz, 2006, pp. 96-122). As discussed previously, due to the particular framework of this study, both study question and

258

choice of methods of data collection and analysis were consistent with ‗theoretical sampling‘ as per GT guidelines. However, nothing in the GT literature appears to preclude the use of several layers of theoretical sampling; also, such an approach is in accord with the iterative ‗dance‘ of qualitative research (Janesick, 1994). Note that neither set of new data will be coded line-by-line. Instead, it will be analyzed by means of existing focused codes, a GT technique used to cover new or larger chunks of data after initial codes have emerged (Charmaz, 2006, pp. 42-71). Clinical evaluation of the sensitizing case. As per the clinical diagnostic report (Appendix I, pp. I1-I23), S was aged 9;2 at the time of referral to the University of Louisiana at Lafayette Speech, Language & Hearing Center, and presented with a history of language and reading problems which had led to major disturbances in his family life and academic development. He was not formally assessed for poor reading comprehension, but authentic/dynamic assessments revealed a profile consistent with this diagnosis. Reading a text at AR book level 3.2, S produced 12.12 miscues per 100 words, 47% of which were of high quality, and 53% of low quality. No formal coding was conducted but informal analysis of his miscue patterns revealed heavy reliance on the graphophonemic as well as the syntactic cueing systems. Like P, he used unconventional prosodic patterns, failing to segment written language into meaningful chunks. His retell was neither coherent nor consistent, covered only the second half of the story, and did not benefit from guiding questions and lookbacks. In a cloze assessment as per Kemp (1987), for which a storybook text was typed and every 7th word was replaced by a blank to be filled in by the client, S produced 47.5% acceptable fill-ins. 52.5% of his fill-ins were either unacceptable, or he failed to produce a suggestion, indicating problems with text coherence and integration

259

similar to those discussed in Chapter Two. Several other assessments also indicated problems with the integration of information. In sum, the following themes that emerged from P‘s data were confirmed for S: FOCUS ON CODE, HIGHER-LEVEL LANGUAGE ISSUES, PIECEMEAL PROCESSING,

and EMERGING INTEGRATION.

Regarding social-experiential and psychological factors, an ethnographic interview with S‘ parents (mother and stepfather) revealed a picture of intense struggle and frustration resulting from his reading difficulties (Appendix I, pp. I24-I32). Parents reported spending considerable time every day helping him with his homework (i.e., much more time than estimated by teachers). Specifically, they mentioned issues in comprehension and failure to retain information, as well as expressive problems. S would stop making an effort almost immediately when encountering difficulty, claim inability to complete tasks because they were ‗too hard‘, and describe himself as ‗stupid‘. Also, he would experience emotional outbursts, crying or throwing himself on the floor. These responses did not stem from parental pressure; both parents reported abstaining from any kind of pressuring, and S‘ mother expressed that she did not wish for him to be a brilliant student, but to be more comfortable with learning, and not as ‗hard on himself‘. These observations amount to stark contrast between the two cases; no pattern of psychological struggle was found in P. The import of this finding will be discussed below. Similar to P‘s case, much of S‘ literacy instruction proceeded from a bottom-up or code-based approach. His parents reported that when entering first grade, S was able to sound out individual letters but not to join them to a word. He received remedial services from Sylvan learning center, which helped him to sound out larger letter strings but not to understand their meaning. School documents revealed that assessment at school relied largely

260

on mechanical skills (e.g. spelling words in decontextualized lists, or reading of nonsense words). S also received remedial services for at least two years; intervention focused on phonemic awareness, improvement of spelling skills through phonics instruction, oral drills, and other lower-level competencies. Taken together, these findings are consistent with the TANGIBILITY and (LACK OF) SUPPORT themes from P‘s data. As in the latter case, S‘ academic environment exhibited a focus on measurable skills and provided one-sided support, which helped S with lower-level skills areas but did not enable him to develop an integrated array of reading strategies. Unlike P, S did not have a significant history of medical or social problems. Parents did report a range of linguistic issues, which led me to diagnose him with a mild language impairment, and which appeared to have some psychological ramifications. Specifically, S was very talkative, forgetful, easily distracted, awkward and clumsy, easily frustrated, had a short attention span and a lack of self-confidence. He was only ‗sometimes‘ able to understand conversations, and to be understood by strangers. He was frequently unable to find the word for a concept, using vague descriptions instead, and got easily frustrated when his interlocutors failed to understand what he tried to convey. These observations were consistent with the ISSUES EXPRESSING HIMSELF theme gleaned from P‘s data. Again, the element of FRUSTRATION found in S was not present in P‘s case. Interview with private SLP: New themes and a breakdown. The interview with the private SLP (henceforth SLP), undertaken during the writing of this discussion, corroborated many of the themes from the analysis of P‘s data and provided useful comparisons with the case of S (cf. Appendix I, pp. I33-I56). It also added new themes pertaining to the progress made by both cases, and resulted in a few ‗breakdowns‘ (Agar,

261

1986) which led to enhanced understanding of P‘s trajectory and the reasons for its differences as compared to S. I will discuss these findings in turn. Confirmed and comparison themes. The following themes resurfaced during the interview, corroborating previous findings: OTHER-ORIENTATION, NEED FOR SCAFFOLDING/ STRUCTURE,

EMERGING INTEGRATION, FOCUS ON CODE, and HIGHER-LEVEL LANGUAGE ISSUES.

Some of them also lent themselves to direct comparison with S. As regards OTHER-ORIENTATION, SLP confirmed that P ―was always pretty agreeable‖ (Appendix I, p. I33) and that he ―would usually do what you asked him to do‖ (p. I45). She affirmed that ‗people-pleaser‘ was a fitting description of his personality (p. I46), and that he was clearly different to S in this regard: ―[P] would go to greater lengths to please you than [S] would (pp. I46-I47). The latter ―did have quite a bit of avoidance behavior‖ (p. I33) and was ―really resistant, really reluctant‖ (p. I38), which she attributed to his FRUSTRATION (p. I38). P‘s NEED FOR SCAFFOLDING/ STRUCTURE did not feature as prominently in this data set as in the interview with P‘s mother. SLP did mention, though, that P tackled writing tasks in a rather unstructured way, whereas S ―understood that it was all supposed to be kinda linked together (…) P never quite grasped that idea‖ (Appendix I, p. I33). As regards the ‗scaffolding‘ aspect, SLP did convey that P ―sometimes (…) would need a little encouragement‖ (p. I38). She did mention that he seemed to make some progress as regards organization of his writing (pp. I49-I51). However, when I asked her about his lack of independent work, using P‘s mother‘s in-vivo code SELF-STARTER, she affirmed that she ―didn‘t see a whole lot of change in that‖ (p. I52).

262

P‘s EMERGING INTEGRATION featured prominently in SLP‘s talk, and her comparisons to S‘ integrative abilities were rather illustrative. She confirmed that P was mostly unable to produce coherent pieces of discourse, both in writing (a main focus of her therapy) and in processing of read texts (several turns at talk omitted): he would just write a string of sometimes unrelated, random ideas (…) he wasn‘t really able to answer um very abstract questions about what he had read and sometimes he had trouble answering even more concrete questions. (…) Sometimes, [P] was just like—dude, you just read this. (…) you can‘t remember one single thing about it? (Appendix I, pp. I33-I35) SLP also related that P‘s integrative abilities did not substantially improve in his time with her: it‘s not like (there wasn‘t) no progress, but (…) in some ways there wasn‘t a lot. You know? (you—beating his) head against the—that wall and (…) he‘s still just looking at you with that blank stare. (Appendix I, p. I52) In stark contrast to P, S seemed to have had better EMERGING INTEGRATION from the onset of working with SLP. In her words (several turns at talk omitted), he wasn‘t necessarily a good writer, but (…) he had a little more (…) of an idea of how to structure writing. (…) He sort of understood that it was all supposed to be kinda linked together and (…) there needs to be some cohesion (…) whereas [P] (…) never quite grasped that idea. (Appendix I, p. I33)

263

This account appeared to contradict my diagnosis of S as a poor comprehender. Upon probing, however, SLP did confirm that [he] would also not necessarily give a cohesive retelling (…) he just seemed to end up with more information sticking. [He] might remember some of the things (…) about it. (…) But [not] to the degree that you would consider (…) age-appropriate or gradeappropriate. (Appendix I, pp. I34-I35, several turns at talk omitted) In other words, while S had better integrative abilities at the onset of intervention, they were still EMERGING at best. And even though they improved over time, they did not progress to the point of independent grade-level functioning. SLP mentioned several times that she expected both P and S to be in constant need of support throughout their schooling (Appendix I, pp. I42, I48). Finally, P‘s FOCUS ON CODE showed up briefly in SLP‘s discourse; in her words, ―he always (…) seemed like a good reader… because he could say what the words were‖ (Appendix I, p. I33, turns at talk omitted). New themes and breakdowns. New aspects of P‘s trajectory brought up in the interview with SLP included the differences between P and S (as seen above), their respective PROGRESS and continued NEED FOR SUPPORT, as well as FAMILY issues. The interview also led to three breakdowns (Agar, 1986) of the existing interpretation, two of which were resolved, while the third led to enrichment of the existing codes. SLP related that both boys made some PROGRESS during their time with her, but in different ways. Notably, while skills had somewhat improved in both, she noted more PROGRESS

in their attitudes (Appendix I, p. I47). For P, she reported that ―the more we did

264

[writing tasks] (…) he did seem to enjoy it. To some degree‖ (p. I50). S, by contrast, ―never enjoyed it (…) but he would (…) throw up fewer obstacles. (…) He would put up with it‖ (p. I50, my hearer signals omitted). She also mentioned that S would show fewer avoidance behaviors (p. I50). Her talk about P ‗enjoying‘ literacy tasks immediately raises a question: if he enjoyed it more, why did he not make PROGRESS? This question will be discussed in a subsequent section together with her observations about P‘s MOTIVATION. As regards skills, the picture is less clear. SLP noted that while P became ―hopefully (…) a little more organized in his approach to (…) writing‖ (Appendix I, p. I49), his integrative abilities and comprehension did not improve, nor did his independence in tackling literacy-related tasks (p. I52). For S, she noted that he ―got better at [writing]‖ (p. I49), speaking to his improvement in integrative abilities, but she did not specify how significant this progress was. Either way, his progress did not to translate into overall improvements in comprehension (p. I51). When asked directly who, in her opinion, had made more progress, she unequivocally opted for S, affirming that not only had he made more progress during his time with her, but also this time had been shorter (p. I54). ‗More‘ turned out to be a relative term. According to SLP, both boys would always NEED SUPPORT PROGRESS

to keep up in school (Appendix I, pp. I42, I48). Nonetheless, the differences in

were clear enough to be of interest for the present inquiry, particularly in the light

of the explanations SLP offered for those different outcomes (i.e., the differences in the boys‘ familial backgrounds, and their respective cognitive make-up). Before considering these, however, two breakdowns should be discussed. The first related to S‘ profile. Some of SLP‘s descriptions appeared to contradict his diagnosis as a poor comprehender. This applies in particular to her statement that he

265

―struggled more with figuring out what the word said but seemed to have a little better hand [on] making sense of it. Or remembering‖ (Appendix I, p. I34). This description is consistent with dyslexia, not with poor comprehension, as seen in Chapter Two. According to Agar (1986), breakdowns occur when a ‗strip‘—a bounded phenomenon encountered in the course of investigation—cannot be made sense of with existing schemas. In the present case, based on my evaluation I expected S to have a clear poor comprehender profile. This schema broke down with SLP‘s statement. When probed, however, she affirmed that she had received the results of my assessment when starting therapy with S, and that the poor comprehender profile I diagnosed him with did fit him: ―I remember it that way. That (…) he could read but (…) (he was) kinda like [P] and (…) couldn‘t talk too much about what he read. (…) I think he definitely fits your profile‖ (Appendix I, p. I44, turn at talk omitted). The breakdown was thus resolved; the major insight appears to be that both readers, while presenting with an overall similar profile, had sufficient individual differences to cause confusion in a seasoned professional. However, this does appear to relate to the well-established heterogeneity of the population of poor comprehenders (cf. Chapter Two). The other breakdown occurred with regard to P‘s LACK OF MOTIVATION. SLP surprised me with her statement that P ―was definitely more motivated [than S] (…) give him a task to do, he‘s gonna (…) get after it…‖ (Appendix I, p. I38) (cf. also her talk of his ‗enjoyment‘ above). This appeared to directly contradict our previous findings. However, when I asked her to differentiate between motivation to do the task as opposed to doing the task for her, she affirmed she was referring to the latter: ―it wasn‘t like, oh, please let me write. No. I

266

think (…) it was (…) like I had asked him to‖ (p. I45). As mentioned above, she concurred with the description of P as a people-pleaser. These breakdowns appeared to be resolved effortlessly, but SLP also introduced a new line of interpretation; the notion that family dynamics may have been a factor in the cases‘ differential development. Specifically, she highlighted P‘s ―intact family unit‖ (Appendix I, p. I39): ―Mom and Dad, [have] always been married‖ (p. I39); P‘s mother, a stay-at-home housewife, had the time and leisure to bring him to therapy (and, as seen in Chapter Four, to vie with the school for better services). S‘ parents, by contrast, were divorced; S went ―back and forth between Mom and Dad‖ (p. I39) for a while. During the time periods in question here, he was living with his mother and stepfather; both worked, which made it difficult for them to bring him to therapy: ―It seem(ed) like it was always (…) kind of a frenzy to find a way to get him here‖ (p. I39). Any attempt at explaining the differential psychological make-up of the two cases by these differences must remain speculative, of course. When I asked SLP to do exactly that— speculate on the repercussions—she replied cautiously that in S‘ case, ―you can‘t help but think, that if you could kinda settle some of that family stuff down, (…) it might help him (…) to change his approach to academics‖ (Appendix I, p. I42). When I elaborated, musing that S‘ issue with self-esteem may partly have been due to his family situation, she concurred: ―yeah, at least [P] had one area where things were ok (…) whereas [S] had (…) some instability here, and some instability there. So it‘s like (…) no safe haven‖ (pp. I43-I44, turns at talk omitted). It should be noted that S‘ family situation was far from dysfunctional. Both my clinical field notes (Appendix I, pp. I24-I32) and SLP‘s remarks (p. I44) concur that the

267

atmosphere in his home was warm, supportive, and fundamentally stable. Instability was simply found in the fact that the family had gone through a divorce, and that both parents worked. The main import of the new FAMILY ISSUES theme was therefore a vague general impression congruent with the hunch that had prompted the collection of this new data; that is, P, in stark contrast to S, did not seem to experience suffering in his situation. However, SLP brought up another possible reason for P‘s lack of progress that challenged the psychological-experiential assumptions at the center of the analysis so far. Musing over P‘s failure to respond to intervention, she surprised me by saying ―I would wonder if [P] had some mild cognitive deficit‖ (Appendix I, p. I53). And she continued (several turns at talk omitted): I didn‘t wonder that as much with [S]. Not that [P] was like super below normal, but—maybe just, just slightly, slightly, slightly. (…) whereas (…) [S] struck me more as a child with learning disability (…) Which would also support why… you would see less progress [in P] (…) relative to the amount of intervention. Appendix I, pp. I53-I55 In other words, she suspected that P may have a mild intellectual disability, or generalized difficulty of adaptive functioning in the conceptual, social, and practical domains (American Psychiatric Association, 2013a). S‘ profile, in her appraisal, appeared to be more akin to a learning disorder; that is, a disability specific to academic achievement (American Psychiatric Association, 2013b). This new outlook was unexpected, but it might make sense in the light of their differential progress. Further, the school documentation provided by P‘s mother when viewed for this possibility might provide support for SLP‘s assertion. For example, P‘s report card for third grade indicated that none of his final grades was higher

268

than C, with the exception of ‗conduct‘, where he earned an A (speaking to his peoplepleasing behavior) (Appendix D, p. D6). Additionally, on the iLEAP, all his results were at Basic or Approaching Basic level, the two lowest passing levels (Appendix D, pp. D4-D5) and his General Student Information sheet indicated that by the end of third grade P ―need[ed] teacher support in all areas‖ (Appendix D, p. D1). Finally, along with these allencompassing academic difficulties, he does have some adaptive trouble in the social domain, as we saw in Chapter Four. Of course, there was no information regarding his practical functioning obtained during this investigation and the reasons for his poor academic performance may be myriad; however, the pervasiveness of his low academic functioning combined with his social issues makes SLP‘s speculation sufficiently plausible to be made part of our investigation and it would enable the inclusion of some type of genetic and/or cognitive variable which would increase the complexity of the model. Consequently, in the models described below, the possibility of a mild intellectual impairment will be incorporated into the discussion as a ―standby‖ variable for various types of genetic and cognitive conditions. We saw in Chapter Two that cognitive difficulties are part of the poor comprehender profile, even though they are not constitutive of it; the population is heterogeneous in this regard. Nonetheless, for our descriptive account of P‘s case, our inclusion of this very tentative assertion regarding a mild intellectual impairment acts as new information that amounts to a major breakdown (Agar, 1986): it does not fit with our interpretation so far. Resolution of the breakdown. As Agar (1986) explains, a breakdown in understanding must be resolved by adapting an existing line of explanation, or by abandoning it for a new one. So far, our interpretation has centered on the psychosocial aspects of P‘s

269

trajectory; his experience with reading, our reasoning went, has led to the maladaptive behaviors and motivational issues described above. The new information, albeit tentative, adds interesting complexity and would indeed contradict this account if it were seen as an exclusive explanation. The verstehen tradition of inquiry, however, does not presume exclusive causes for phenomena; various factors can be incorporated into an interpretation if they do not violate its plausibility. How, then, does the possibility of P having an intellectual disability fit into our account? First, it needs to be noted that while this observation is informal, it comes from a seasoned clinician, and is congruent with our data. It is therefore appropriate to take it seriously. On the other hand, SLP pointed out that if such a disability were present, it would be very mild (‗just slightly, slightly, slightly below normal‘). A less than very mild intellectual issue would warrant abandoning our account, for it would likely represent a major obstacle to comprehension, and to learning in general. A very mild impairment, however, can be integrated into the existing explanation; we can assume that while it presents an additional obstacle, it does not outweigh the other obstacles we identified. With this in mind, our analysis can proceed. Recall that according to the original trajectory, P‘s main obstacles to comprehension and to progress in therapy were found in a self-defeating feedback loop: OTHER-ORIENTATION combined with a focus on low-level reading skills led to behavioral and psychological consequences; intervention had proven unsuccessful in breaking through this dynamic. So far, we assumed the reason for this was that intervention may have created a paradoxical situation; that is, as long as any learning is prompted from without, and evaluated by an external observer, he would not be able to overcome his OTHER-ORIENTATION. We also

270

surmised that if he could be led to experience the pleasure of comprehending, more progress might be seen. This new information allows us to form a better idea of what may hold P back. If he does have an intrinsic cognitive problem (a tentative assumption on our part), he would experience even more failure than poor comprehenders without such problems – such as S – and his progress would be even slower and more laborious. In addition to early language problems and a school experience where lower-level skills were emphasized, intrinsic obstacles would add to his troubles in constructing meaning; in this amended scenario, OTHER-ORIENTATION

would be an even more plausible response to his compounded

difficulties. If the ‗others‘ towards whom he oriented rewarded him for FOCUSING ON CODE and ACCURACY, but less so for integration and comprehension, and his intrinsic obstacles prevented him from making sense of texts for himself, it would be even more plausible for him to never progress beyond low-level skills. This would then have led to the psychological and behavioral consequences we saw: intense and problem-laden low-level engagement with text combined with a LACK OF MOTIVATION to tackle what is meaningless from his viewpoint. Intervention, in this scenario, would have to break through the double barrier of habit and intrinsic obstacles. In addition, awareness of comprehension issues itself may have been hampered by his intellectual problems, further reducing any incentive to respond to intervention. If a reader does not know that he is missing out on a vital part of the reading experience, there is no apparent need for him to change the experience. With this combination of intrinsic obstacles and lack of incentive, P‘s LACK OF PROGRESS is even less surprising in this amended version. Figure 5.2 outlines his trajectory in the light of this new information.

271

Behavioral and psychological consequences HIGHER-LEVEL LANGUAGE ISSUES STRUGGLE WITH PREDICTION PIECEMEAL PROCESSING PREOCCUPATION WITH VISUAL SEARCH LACK OF MOTIVATION & NEED FOR SCAFFOLDING/STRUCTURE

Mild ID

Intervention issues UNCERTAINTY (LACK OF) SUPPORT EMERGING AWARENESS EMERGING INTEGRATION LACK OF PROGRESS

Mild ID

Social-psychological and behavioral response OTHER-ORIENTATION FOCUS ON CODE STRUGGLE FOR ACCURACY

Mild ID

Hearing problems Mild ID Academic experience READING VS. COMPREHENDING TANGIBILITY

Language problems & psychosocial effects HALF THE STORY ISSUES EXPRESSING HIMSELF Mild ID

Figure 5.2: Amended Thematic Outline of P’s Trajectory In Figure 5.2, P‘s possible mild intellectual disorder (ID) is depicted as an obstacle at every moment in his literacy experience. Note that I remain agnostic as to the onset of the ID. Our information is too rudimentary; there is simply no way of knowing if it was present at birth, or if it is a corollary of his language development issues. Either way, this is of no import to our purposes: what matters is that the ID may have gotten in the way of his literacy acquisition at all points in his development. Another change from the previous model is found in the feedback loop, which now includes P‘s academic experience. As the predecessor focused on P‘s LACK OF MOTIVATION and OTHER-ORIENTATION as crucial factors to understand his trajectory, the continuing experience at school did not seem impactful enough to be included. But as we now

272

incorporate P‘s ID as another possible obstacle, our attention needs to be redirected to include his schooling where, to the best of our knowledge, he continues to receive instruction and intervention focused on ‗tangible‘ outcomes. We saw that these did include higher-level competencies such as finding the main idea or making inferences (cf. Chapter Four); hence, his academic experience does not seem to be as exclusively focused on lower-level skills as S‘ was. Still, insofar as we may assume that he experiences more failure than success in his schooling, both because of possible intrinsic obstacles and because his needs are not met, the persistence of his difficulties becomes more plausible. It is illustrative to use the little information we have about S‘ development to devise a comparison trajectory. His mild language (and possible learning) disability alone would explain why he has difficulty overcoming the obstacles to comprehension posed by an academic environment focused on low-level skills. In addition, however, he has experienced disruption in his family environment; whether as a corollary or independently, he experiences major frustration with his issues. His need for academic support was only partly met (i.e., exclusively with more lower-level instruction), leading to similar reading behaviors as seen in P. In contrast to P, however, S is aware of his problems, as we have seen in the fact that he experiences major frustration. (Note that frustration is not the same as lack of motivation; a learner may well be frustrated with his/her failures, but still be motivated to overcome them.) Given the appropriate intervention, S has shown that he is capable of making some progress. He had better integrative abilities than P to start with, and he has honed them even further. His success gives him incentive, if not a lot, to work on his problems. Nonetheless, his LLD persists, and so does the fact that he has experienced disruption in his family life. We cannot know the extent to which either of those contributes to the persistence of his

273

issues, but we saw that SLP‘s appraisal is unequivocal: in her view, both S and P will need intervention, in one form or another, throughout their academic life. Figure 5.3 illustrates what we know about S‘ development. The upward arrow symbolizes S‘ progress: his comprehension and related literacy competencies are developing, however slowly and laboriously. Intervention issues SOME PROGRESS BETTER EMERGING INTEGRATION LESS FRUSTRATION

Academic experience NEED FOR SUPPORT TANGIBILITY (LACK OF) SUPPORT

Language/learning disability & psychosocial effects ISSUES EXPRESSING HIMSELF FRUSTRATION

Ongoing development

Behavioral response FOCUS ON CODE HIGHER-LEVEL LANGUAGE ISSUES PIECEMEAL PROCESSING EMERGING INTEGRATION

Exacerbating factor FAMILY ISSUES

Figure 5.3: A Thematic Outline of S’ Trajectory In contrast to P‘s model, this one does not include a starting point. This is simply due to the relative paucity of the data we have on S, but it highlights an important insight: ongoing comprehension difficulties can be conceptualized as a loop, where one aspect of the problem engenders another in a ‗dance‘ without beginning or end. In a final amendment, then, we can conceptualize P‘s current functioning in the same way – as a loop of mutually sustained complexes of difficulty, shown in Figure 5.4. I did not include a symbol for ongoing development, as in the model for S. As per SLP‘s report, P did make some marginal,

274

unspecified progress during his time with her, but not enough to warrant continued effort: ―as a Mom, I don‘t think I would have kept bringing him to me either (…) and uh I don‘t think it‘s anything I was doing wrong, (…) it‘s just—we were sort of treading water‖ (Appendix I, pp. I55-I56, turn at talk omitted).

Behavioral and psychological consequences HIGHER-LEVEL LANGUAGE ISSUES STRUGGLE WITH PREDICTION PIECEMEAL PROCESSING PREOCCUPATION WITH VISUAL SEARCH LACK OF MOTIVATION & NEED FOR SCAFFOLDING/STRUCTURE

Mild ID

Intervention issues UNCERTAINTY (LACK OF) SUPPORT EMERGING AWARENESS EMERGING INTEGRATION LACK OF PROGRESS

Mild ID

Mild ID

Social-psychological and behavioral response OTHER-ORIENTATION FOCUS ON CODE STRUGGLE FOR ACCURACY

Mild ID

Language problems & psychosocial effects HALF THE STORY ISSUES EXPRESSING HIMSELF

Academic experience READING VS. COMPREHENDING TANGIBILITY Mild ID

Figure 5.4: P’s Trajectory, Conceptualized as a Loop Two major insights can be taken away from this discussion. First, our comparison confirms what has been long known: poor comprehenders are a clearly heterogeneous population. Even when looking at as few as two cases, distinct differences emerge. P‘s and S‘ poor comprehension are outcomes of quite different developmental trajectories, with considerable variation in contributing factors. Similarly, both cases differ in their response to intervention; we can assume that this is due to the etiological variation behind their similar profiles. This observation, in and by itself, raises doubts that a rigid and unspecific

275

framework like the Simple View (cf. Chapter Two) can adequately capture the developmental, psychosocial-experiential, and biomedical variation behind the phenomenon. We will return to this objection in Chapter Six below. Second, while it may not be possible to explain poor comprehension – in the sense of finding the unique causal factor of which it is the purported result – we are now in a position to claim that it is possible to understand it as a plausible outcome of a young reader‘s development over time. Put otherwise, it made sense that the cases we examined should have a poor comprehender profile. When treating the various intrinsic, internal, external, and developmental factors contributing to the issue as pieces to a puzzle, rather than as an array of variables from which the one causal factor must be isolated, we arrive at a comprehensible and plausible account. But this holistic outlook is not the only difference to a causal-componential explanation. Our description also accounts for the role both of intentionality, and of unintended outcomes of intentional actions. Recall that from a verstehen perspective, ―human actions take place in order to bring certain future states of affairs about, not merely because certain past states of affairs happened‖ (Bransen, 2001, p. 6). The case has been made that reading, or indeed any type of epistemic engagement with the world, is naturally done in order to bring about understanding, learning and comprehension are simply states of the human organism. They are (…) a consequence of being alive. Their presence in human beings doesn‘t have to be explained, only their absence‖ (Smith, 2004, p. 8)

276

In other words, any account of poor reading comprehension needs to describe in how far, and ideally why, the natural urge to comprehend is stifled or fails. At the same time, such a description would have to incorporate expert knowledge to account for the role of unintended consequences of intentional actions. As we discussed in Chapter Three, a pure ‗insider‘ perspective, the ‗gold standard‘ of qualitative research, is likely to prove unfruitful where the objects of study are problems for which participants seek help from professionals; if participants had complete knowledge of the phenomenon, their actions would had led to the intended results, and they would not conclude that they have a problem. In both cases we studied, we found a variety of unintended outcomes. Both P and S have experienced instruction in lower-level reading skills that were intended to guide them towards higher-level integration, but instead fostered an undue engagement with the lower levels of written language. P is intentional in his transaction with text, as seen in his very deliberate efforts in STRUGGLING FOR ACCURACY, FOCUSING ON CODE, as well as in his extensive VISUAL SEARCH behaviors. But the outcomes are less than successful (at least from the perspective of adult stakeholders); likewise, the SUPPORT he gets does not show the intended results. In S‘ case, we saw similar dynamics, except that for him, the support he received prior to therapy with SLP did not only fail to help but positively frustrated him. In both cases, expert knowledge was needed to make sense of the dynamics involved; insider perspective alone would have been insufficient. We have thus arrived at an empathetic re-description of the factors that led to comprehension failure for both our cases. S, faced with a combination of difficulties in learning and external circumstances not conducive to it, ran into a ‗wall of failure‘ in his

277

attempts to comprehend and experienced stark frustration about this. When he was given appropriate strategies, his transactions with print became somewhat more successful. P‘s academic experience was commensurable with S‘, his intrinsic difficulties possibly more obstructive, and his social experience characterized by a more withdrawn attitude on the one hand, and a more secure family environment on the other. Crucially, he was not as aware of his failure to comprehend as S was, and whatever awareness he had did not lead to frustration. In short, then, S was aware of his troubles and frustrated by them for various reasons; P was less aware, and his troubles were more severe but hampered him less in his well-being. It is therefore perfectly understandable that a) either should have comprehension problems, and b) that S should be more successful than P in learning to comprehend while P should not make much discernible progress. That said, we are facing a problem in answering the general research question for this study: What are the qualitative characteristics of poor reading comprehension and its context? We divided this general inquiry into three sub-questions pertaining to 1) linguistic and behavioral, 2) psychological, and 3) contextual (social and developmental) characteristics. As seen in our discussion so far, we found characteristics in all three areas and were able to combine them to a plausible account of the development of poor comprehension. However, we were able to study only one poor comprehender in depth, and compare him to cursory data on another. Even if the phenomenology of both cases were isomorph, we would not be able to devise an authoritative list of characteristics from such limited data, or devise a grounded theory of poor comprehension on basis of such a list. If we want to arrive at an answer to our general question, then, we need to couch our findings in a theoretical framework that can accommodate the characteristics we found, and the

278

developmental narratives that emerged from them, as well as characteristics that might be found if we were to study more cases. The remainder of this chapter will be devoted to discussing a theoretical framework for our results, with the goal of answering our general research question. Specifically, I will place our findings within the framework of systems theory, and couch them in an ecological metaphor. Taken together, both will elucidate our understanding of the present case and the comparison case, as well as leave room for future data. Poor Comprehension as a Systemic Phenomenon The idea that P‘s development of poor comprehension was so ‗understandable‘ that it could be described as a ‗natural‘ process emerged early during this study, prompted by the plausibility of the emerging narrative. In Memo 12, drafted less than two months after data collection and analysis had begun, I mused: Wondering if the best way of describing (…) my data would be in a metaphorical ecological/evolutionary framework. The kids working themselves into a cognitivepsychological-behavioral-interactional niche—or rather, drifting there naturally (Varela/Maturana)—in interaction with their environments. (Appendix B, p. B17) ‗Natural drift‘ (Maturana & Varela, 1992) is an alternative take on Darwinian evolution that conceptualizes evolutionary processes as an ongoing unfolding of ‗goodenough fit‘ with the environment, rather than as a struggle for survival. As we will see, this metaphor lends itself neatly to capturing the dynamics that lead to and sustain poor comprehension. But we cannot simply jump from behavioral and interview data to using such a distal concept. First, we need to couch our findings within a theoretical framework which

279

can accommodate human development as well as ecological dynamics. Such a framework is found in the theoretical underpinning for Maturana‘s and Varela‘s work; that is, systems theory (von Bertalanffy, 1968), a generalized approach to viewing natural phenomena as dynamic, interconnected wholes. Systems theory: A brief overview. Systems theory is a variegated paradigm to which many fields have contributed. von Bertalanffy (1968) is frequently referenced as the seminal work (e.g. Thelen & Smith, 1994, pp. xix-xx; Winek, 2010, p. 4), although the roots of the paradigm can be traced back as far as medieval thinkers like Nicholas of Cusa, and its development was based on the convergence of ideas from various disciplines including biology, physiology, and information theory (von Bertalanffy). It has influenced such divergent epistemological outlooks as constructivism (Fosnot & Perry, 2005; Rusch, 1994) and connectionism (Cilliers, 1998), and it has been used to describe and model diverse phenomena, among others, the emergence of motor movements and basic cognition in young children (Thelen & Smith), linguistic and cognitive (Smith & Thelen, 1993a) as well as emotional development (Lewis & Granic, 2000), the evolution of cognition (Varela & Maturana, 1992), and dynamics of interpersonal relations (Winek, 2010). Systems theory is not a theory in the usual sense of the term (i.e. a ―collection[.] of sentences, propositions, statements or beliefs (…) and their logical consequences [including] explanatory and predictive laws‖ (Bogen, 2013). Rather, it represents a general paradigm in Kuhn‘s (1962) sense; a complete way of ‗doing science‘, which includes practices, methods, theoretical stances, accepted findings, and a particular view of the world (Kuhn, pp. 10-34; von Bertalanffy, 1968, pp. xvii-xxiv). von Bertalanffy subdivides the paradigm into three aspects: ‗systems science‘ (i.e. the methods used to study systems across disciplines ranging

280

from the natural to the social sciences); ‗systems technology‘ (concerned with devising or maintaining systems in the real world); and ‗systems philosophy‘ (a way of thinking about the natural world which presents an alternative to the linear-causal paradigm). As we have not used formal systems-science methods in the present study, and are not concerned with engineering a system, it is the latter aspect that is of interest to us. von Bertalanffy suggests that systems-philosophy itself has a tripartite structure. First, it comprises ‗systems ontology‘, the study of what constitutes a system. Second, it comes with ‗systems epistemology‘ that is a holistic, rather than analytical, outlook on phenomena of interest. Finally, it includes what can be termed ‗values‘, or the symbolic universes inhabited by humans, which are usually studied by the humanities. This latter assumption is of particular interest for us, as it allows for the incorporation of qualitative data within a systems framework. (As we will see, not all authors agree with von Bertalanffy‘s stance on the inclusion of ‗values‘). I will present the basic tenets of systems theory along these lines, relating them to our phenomena of interest. Systems ontology and epistemology will be discussed together; the role of values and their relation to interpretive analysis will be discussed subsequently. Systems ontology and epistemology in poor reading comprehension. Definitions of ‗systems‘ are as varied as the fields which have contributed to the theory, but they all share a common denominator: a commitment to viewing phenomena, not in terms of linear relations of cause and effect, but in terms of wholes with relations to other wholes (environments). Systems theory is particularly (but not exclusively) concerned with wholes whose properties are not reducible to causal relations between parts (von Bertalanffy, 1968). The basic definition of systems is ―sets of elements standing in interrelation‖ (p. 38). In this basic sense, the world abounds with systems: a university, an animal, a person, a tree, a

281

forest, or a football team are all systems (Meadows, 2008). In addition to such concrete systems, there are ‗conceptual‘ systems, e.g. language, music, or logic. The boundary between concrete and conceptual systems is not always clear; ecological or social systems are as concrete as forests or animals, but they cannot be as directly observed, and hence their study requires some degree of conceptual abstraction (von Bertalanffy). In the present study, several phenomena can be conceptualized as systems. First, there are the readers themselves in their individuality as ―active personality systems‖ (von Bertalanffy, 1968, pp. 192-194). Likewise, the totality of the readers, their family, and their schools, as well as the interactions between them, can be thought of as a social system (von Bertalanffy, pp. 194-197; Winek, 2010). It is this system that will be of major interest for the discussion below; I will refer to it as ‗reader-environment system‘. On a more abstract level, language and cognition are systems in their own right. Systems can be open or closed; closed systems, such as chemical reactions or the motions of the planet, do not interact with their environments, whereas open systems do. Consequently, elements of closed systems interact with each other in linear ways; their final state is ―unequivocally determined by initial conditions‖ (von Bertalanffy, 1968, p. 40). Open systems such as animals, humans, or social institutions, by contrast, are in constant interchange with their environments. In these systems, trajectories are not fixed, they may reach a final state from different initial conditions (‗equifinality‘); conversely, any given initial condition may lead to a potentially infinite number of different outcomes (‗equipotentiality‘) (von Bertalanffy, p. 39-40; Winek, 2010, p. 7). A related distinction can be made between ‗complex‘ and ‗non-complex‘ (‗complicated‘ or ‗simple‘) systems. Non-complex systems can be described in terms of the

282

interrelations of their elements; examples include cars, airplanes, or snowflakes. In theory, all of these can be ‗dissected‘ into parts; an observer provided with a complete description of these parts would understand the system, as relations between parts are linear. In complex systems, by contrast, relations between parts are non-linear; hence, such systems display characteristics that are not predictable by or reducible to the interrelations of their parts. These are termed ‗emergent properties‘ (Cilliers, p. 5; von Bertalanffy, 1968, p. 55). Understanding of such systems is impeded if the observer attempts to arrive at it by looking at their parts (Cilliers, 1998, pp. xviii-x, 1-7). For the present study, it is obvious that we are concerned with open, complex systems. Our readers, as well as the reader-environment systems in question, exchange information with their respective environments. Similarly, analyzing them into parts is not conducive to understanding them. As we have seen, in twenty years of research into the presumed component parts of poor comprehension investigators have not gotten to the heart of the matter. And looking at the present cases in the light of systems theory, it becomes plausible why that would be the case. As has been pointed out before (e.g. McDermott & Roth, 1978; Perkins, 2005), processes observed in individuals‘ behavior can be viewed as emergent properties of the interaction between individuals and their respective environments (social or otherwise). Consider the social, psychological, and behavioral processes we observed (and even the cognitive processes we inferred), all of which are integrated in Figures 5.1 through 5.4. Seen through a systems lens, it becomes clear that most of them are not simply located within the reader or the environment. Rather, they are emergent properties of the transaction between the readers and their environment (i.e. of the reader-environment system). Likewise, the

283

dynamics from which they result, as well as the dynamics they engender, are properties of this system. As an example, READING VS. COMPREHENDING is a concept experienced by P (the personality system) at his school (the social system). P‘s OTHER-ORIENTATION is a psychosocial habit of a personality system acquired in and sustained by interaction with teachers and peers. Similarly, his FOCUS ON CODE can be traced back to pertinent expectations at school. Note also how equifinality and equipotentiality play themselves out in our case comparison: P‘s and S‘s starting points and trajectories are clearly different, yet they lead to (roughly) the same result, namely, poor comprehension. Since each readerenvironment system has different properties, however, their trajectories diverged again from there. Another agreed-upon characteristic of complex systems is that they are selforganizing (i.e. their internal structure is not determined by external input, but results from processing it in consonance with prior system states). In a simple manner of speaking, complex systems strive for internal balance (‗homeostasis‘ or ‗equilibrium‘) by regulating themselves through internal feedback loops. For example, the metabolic system self-regulates by giving feedback of blood sugar levels to the brain. Similarly, artificial neural networks self-organize by means of feedback loops between neurons and groups of neurons (cf. Cilliers, 1998; von Bertalanffy, 1968; Fosnot & Perry, 2005). This concept is intuitively applicable to psychological and cognitive processes, and have, indeed, been used to describe both learning (e.g. Piaget, 1970) and relationship dynamics (e.g. Winek, 2010). Winek introduces the term ‗circular causation‘ for the same notion; a system‘s self-organization may include processes of the type ‗A causes B‘, ‗B

284

causes C‘, ‗C causes D‘, and ‗D causes A‘, forming a circle of causal relations (pp. 8-9). Not incidentally, we saw this same kind of ‗loop‘ in Figures 5.3 and 5.4. However, this is a simple description, and possibly too simplistic. First, the idea of ‗equilibrium‘ as the end goal of any systemic dynamic is misleading; self-organizing systems strive for, but never reach a state of true equilibrium. The dynamics of the interplay between their inner states and the environment leads to continuous reorganization (Fosnot & Perry, 2005; von Bertalanffy, 1968). In von Bertalanffy‘s words: ―life … is essentially maintenance of disequilibria (…) Reaching equilibrium means death and consequent decay‖ (p. 191). As we have seen in both our cases of poor comprehension, development never stops; even in P, there is some progression and change over time, however halting and ill-defined it may be. Second, it is obvious that our feedback loops do not represent causal effects in the strict sense of a recurrent contingency (cf. the discussion in Chapter Two). Rather, the relations represented in the figures cover a generalized notion of ‗leads to‘, as well as what Howe (1992) calls ‗intentionalist causation‘. Consider Figure 5.4: P‘s language problems and their psychosocial effects do not ‗cause‘ his academic experience. Rather, his problems led to an undue prominence of READING VS. COMPREHENDING and TANGIBILITY in his experience of his literacy instruction. Put otherwise, children without P‘s problems may have a different literacy experience in the same environment (e.g. by making up for the lack of comprehension instruction using their own linguistic resources). Similarly, this experience didn‘t cause P‘s response but rather led to an intentional attempt at coping, which in turn likely had unintended behavioral and psychological consequences. The concept of self-organization enables us to further pursue the idea that our observed themes are not simply located within the reader or the environment. As we said, P‘s

285

FOCUS ON CODE and his STRUGGLE FOR ACCURACY can easily be envisioned as emergent property of a classroom situation where a teacher expects accurate rendering of text and P struggles to overcome his articulation difficulties. Similar reasoning applies to the other themes, at least until we get to behavioral and psychological consequences. It is seductive to think of HIGHER-LEVEL LANGUAGE ISSUES, STRUGGLE WITH PREDICTION, or PIECEMEAL PROCESSING,

as properties of P‘s cognitive make-up; likewise, his PREOCCUPATION WITH

VISUAL SEARCH

appears to be a behavioral strategy based on these internal structures.

However, systems theory does not let us get away with such easy allocations. We will discuss this in more depth below; first, however, the friction between qualitative and quantitative approaches to systems merits consideration. Systems values: Qualitative vs. quantitative stances. The use of systems theory to descriptive ends is not without contention. There is a tension between highly formalized, mathematical approaches on the one hand, and qualitative-metaphorical ones on the other. Examples for the former can be found in neural network modeling (e.g. Cilliers, 1998), but also in the study of motor development and the beginnings of cognition in children (Thelen & Smith, 1994), and generally in all fields traditionally thought of as ‗natural sciences‘ (von Bertalanffy, 1968). The latter are used to describe psychological and interpersonal dynamics (Winek, 2010), but also human learning, including language acquisition and the development of cognition (Fosnot & Perry, 2005; Smith & Thelen, 1993a). And some authors on the formal end of the spectrum have warned against using systems theory in metaphorical ways: there is a danger (…) in the temptation (…) to merely redescribe the phenomena we have been studying for so long, and then conclude that we have explained them. (…)

286

We should apply the theory to particular phenomena (…) without trying to sidestep the theory‘s mathematical foundations. Robertson, Cohen, and Mayer Kress (1993, p. 119) Since re-description is exactly what we are doing in the present study, this critique merits a response, which will be given in a twofold manner. First, the roots of systems theory are not found in mathematics alone. As mentioned earlier, variegated fields have contributed to the paradigm, and while many of them did contribute an impressive formal-mathematical apparatus, this is by no means the only source for the systems view. von Bertalanffy (1968) cites the medieval philosopher and theologian Nicolaus Cusanus as a predecessor (p. v), and affirms early in his tome that systems thinking cannot solely rest on formalization. ‗Values‘, the meanings ascribed to the natural world that constitute the world in which humans live, have their intrinsic place in the paradigm. His explication is worth quoting at length: If reality is a hierarchy of organized wholes, the image of man will be different from what it is in a world of physical particles governed by chance events (…) The world of symbols, values, social entities and cultures is something very ―real‖; and its embeddedness in a cosmic order of hierarchies is apt to bridge the opposition of (…) science and the humanities. (pp. xxii-xxiii) This argument is directly parallel to the ‗verstehen versus erklären‘ debate. The latter centered on causation; as Bransen (2001) lays out, there is a clear difference between the contingency of cause-effect relations in the natural realm, and the intentionality with which human actions are used to bring about desired effects – the latter cannot be explained in terms of the former. Likewise, the material world that arises from contingent interactions

287

between particles is different from the symbolic world that arises from intentional actions and interpretations by humans. With this in mind, von Bertalanffy‘s (1968) plea against the exclusive use of formal methods in systems thinking becomes all the more urgent: This humanistic concern (…) makes a difference (…) I do not see that these humanistic aspects can be evaded if general system theory is not limited [sic] to a restricted and fractional vision. (p. xxiii) The ‗bridging‘ potential of systems theory has a direct import on the issues at hand. We have seen the effects of a ‗restricted and fractional vision‘ in the study of poor reading comprehension; absent a ‗humanistic‘ outlook, research relying on mechanistic thinking kept grasping at purported components and their causal relations, only to find them elusive. Thus, the first counter-argument against the warning by Robertson, Cohen, and Mayer Kress (1993) is grounded in the original spirit of systems theory itself; that is, to eschew metaphors and redescription may not only give rise to a fractured view of the world, but may actually hamper the advancement of research. What about the second part of the warning, which contends that mere re-description is not an explanation? This objection can be directly countered by harking back to Bransen (2001). As discussed in Chapter Two, verstehen-based inquiry is geared towards capturing the empathetic quality of a given phenomenon, not contingent relations between factors. To understand is, in fact, not the same as to explain; it is a different kind of knowledge, and it has its own merits in and by itself. The usefulness of systems theory lies, among other things, in its applicability to both types of research, and its ability to accommodate both types of knowledge.

288

A similar argument comes from the same volume where the warning is found. Smith & Thelen (1993b) respond to their colleagues by acknowledging the danger of overusing systems metaphors. We must take care, they concede, not to use systems theory the way a toddler uses a wooden hammer, indiscriminately banging it on all kinds of data until they are indistinguishable. That said, they continue, metaphorical applications are useful when done with caution, especially since formal treatments would require amounts of data simply not available to researchers. Importantly, systems metaphors may help overcome less useful metaphors such as notions of cognitive structures ‗in the head‘ which influence development. As they point out, not only are there no such structures; in addition, thinking about cognition in structural terms has led to partial treatment of data, exerting ―a powerful influence on what data have been collected (and sometimes what data have been dismissed) and has strongly limited theoretical advance‖ (p. 166). We have seen in the study of poor comprehension what the authors allude to here: qualitative data on experience, knowledge, and motivation was rarely collected, and did not garner much attention when it was. As a result, the preliminary theoretical grip on the issue was lost. In such a situation, re-describing a phenomenon is a viable option. As Smith and Thelen put it: Even if (…) dynamic systems is just a metaphor, it is no mere metaphor. It (…) turns empirical questions around by focusing attention on (…) the relation between stability and variability, and the process of change. It (…) asks us to shift our attention from the study of knowledge structures in development to a study of the developmental pathway itself. (p. 166)

289

The present study has done exactly that. Focusing on the development of poor comprehension, it has made visible the pathway that led into it, as well as the contributing factors. Thus, the programmatic stance expounded by von Bertalanffy (1968) has shown its empirical applicability. Systems metaphors are a useful way to re-describe puzzling phenomena in order to make sense of them where sense is lost. Going back to the behavioral and psychological consequences aspect of P‘s trajectory, it becomes clear now that simply locating them in the reader would amount to succumbing to the structural fallacy bemoaned by Smith and Thelen (1993b). It is only from a cognitive-structure viewpoint that these are ‗just‘ properties of P. From a dynamic systems viewpoint, they cannot be ascribed to P‘s cognitive-behavioral makeup without considering the context in which they emerged, the context in which they are observed and the context in which they are employed. In other words, they are not so much part of the reader system as of the reader-environment system: they emerge from prior system states, and are embedded within its ongoing self-organization. Without P‘s language issues and the ensuing psychosocial effects, without his prior experience of schooling and his attempts at fitting into this experience (response), his reading strategies could not be explained. The same is true, of course, for the issues encountered in intervention, such as the limited PROGRESS he made, or his troubles with INTEGRATION. Another illustration for this point comes from the observation of P‘s reading behaviors. Recall that behavioral and psychological consequences vary between textual transactions: the percentage of his low-quality miscues was lower in home reading two, the text he was more familiar with and more interested in than the text for home reading one. Even his assessment results varied between types of assessment (cf. Chapter Four). In other

290

words, none of our findings are simply properties of the reader. Even when the social environment is not directly involved, they emerge from transactions with a symbolic environment, which in turn are informed by the entirety of the reader‘s prior experience. This insight gives us the justification to attempt a systems-based formulation of a grounded theory of poor reading comprehension. First, however, we need to overcome one final obstacle: the temptation to conceptualize reading comprehension itself in terms of mental entities. Comprehension: Structure or Quale? The picture emerging from the discussion above is at odds with a prevailing view in the literature. That is, the equation of comprehension to a set of observable behaviors, which in turn are, implicitly or explicitly, taken to indicate localizable mental entities (or ‗structures‘, cf. Smith & Thelen [1993b]). In other words, the ontological status of poor reading comprehension is not clear. We briefly touched on this issue in Chapter Two, where we questioned the notion that comprehension can be operationalized in terms of behaviors, and discussed the possibility that it may better be thought of as a quale, an ineffable mental state. This ineffability shows glaringly in the issue of assessment. Taking a brief look at the evolution of pertinent tools, we found a trend; over time, lists of skills thought to indicate understanding became less comprehensive—as if the field had surrendered to the complexity of the matter. Occasionally, authors do mention that there is a degree of uncertainty in this regard. Nation (2005), for example, observes that some investigators treat inferencing problems as a causal factor for poor comprehension, while others equate them with it. But such moments of deeper questioning are rare, and for the most part, the field proceeds doing what Kuhn (1962) calls ‗normal science‘, without examining underlying assumptions.

291

A closer look at Cain & Oakhill (2007) as well as Oakhill and Cain (2007) reveals the origins of the divergent ideas on inferencing. The authors contend that this ability is not only as an important predictor for comprehension, but that it is causally implicated in it, shown by the fact that poor comprehenders perform worse than comprehension-age matched good comprehenders on inferencing tasks. However, this conclusion is based on the groups‘ performance on standardized measures, notably the NARA (Neale, 1997), which uses inference-tapping questions to assess comprehension (Cain & Oakhill, 1999; Ricketts, n. d.). In other words, the authors claim that the same ability that is used to gauge comprehension is also a causal factor in it—exactly the kind of circular reasoning to which Nation (2005) refers. Thus, the question whether poor inferencing is a cause, an effect, or the same as poor comprehension remains unresolved. But even absent such circularity of argumentation, any behavior one might choose to study—lower or higher-level language skills, knowledge of narrative structure, etc. (cf. Cain & Oakhill, 2007; Oakhill & Cain, 2007)—comes with the same interpretive uncertainty. In all these cases, the question whether an observed skill is a cause or an aspect of comprehension (let alone an indicator for a mental structure) cannot be empirically answered through isolated assessment, but must be considered in the light of all pertinent findings, and within the framework of a theory of comprehension. These observations prompted a lengthy deliberation on my part, captured in Memo Eight (Appendix B, pp. B10-B12). It began with a meditation on the design of the GORT-4 (the instrument I had initially used to assess participants), in which I doubted the usefulness of questions to determine comprehension: ―Technically, no comprehension (…) is necessary to answer a question about a piece of information from the text—understanding the question

292

suffices‖ (p. B10). Or at least, this is true when the questions are not worded to require text knowledge. This seems to be the case with the GORT-4. Keenan and Betjemann (2006) found that many questions on the instrument could be answered without reading the text passages. Consequently, these authors admonish the field in general for a lack of concern with this issue, indicating that not enough attention has been paid to what is actually assessed with comprehension questions. Retells are another common way of determining comprehension, and intuitively, they seem to be better suited to this end. In P‘s case, we found it plausible to assume that verbatim rendition of disconnected, garbled textual propositions is an indicator of comprehension failure. But is this really the case? After all, P has a history of expressive language problems; what if the ‗mental model‘ (Kintsch & Rawson, 2005) he constructed from the text is more coherent than the oral rendering? Note that a mental model is a holistic entity in which all subunits coexist simultaneously, whereas language proceeds in a linear, time-bound fashion, requiring proper sequencing of units. Maybe P‘s retells failed, not because he did not construct an adequate mental model but because he was unable to translate this model to a very different mode of representation. Conversely, even if he had produced coherent retells, rehashing them from memory, the argument could have been made that accurate memorization and reproduction is not the same as comprehension. As in the behaviors discussed above, it is not clear what we mean by ‗comprehension‘, or how we justify equating it with a circumscribed set of behaviors. Thusly dissatisfied with existing definitions, I took it on myself to devise a list, based on the literature and my own deliberations, of all behaviors that could be taken to indicate

293

comprehension. I render it here to convey the idea that we commonly draw on a very limited subset of those: 

Detecting syntactic, semantic etc. errors



Detecting contradictions/inconsistencies



Detecting factual errors (…) (requires previous knowledge)



Carrying out actions specified in, or associated with a text (including answering ‗comprehension questions‘)



Placing a text in context to other texts (in terms of style, content, genre)



Relating text to existing knowledge



Relating text to third party such that third party indicates, ―I understand reader‘s rendering of this text‖ [i.e., a retell]



Using text, or aspects of it, to create a piece of writing based on/incorporating either (Appendix B, p. B11)

Obviously, this list is still not exhaustive. More importantly, it simply reflects the overall abilities of someone we would call a reader. I take this to indicate that any definition and assessment of comprehension which includes only an arbitrary subset of reading competencies will face justificatory difficulties since we lack the grounds to assume that comprehending is the same as showing a limited set of expected behaviors. A related type of confusion shows in the treatment of comprehension monitoring in the literature (cf. Cain & Oakhill, 2007; also Chapter Two). Most authors assign importance to it as a component of, or factor in, comprehension – and rightly so, as it would be odd to say that a reader is comprehending without being cognizant of it. But, as I wrote in Memo 294

Eight, monitoring of one‘s own comprehension is qualitatively different from performing one of the various behaviors associated with comprehension. It is the awareness of whether any particular strategy—inferencing, use of anaphors, building of narrative structure—yields the desired result. Hruby (2009) makes the same point: The self-perception of comprehension by a subject should be distinguished from comprehension processes per se. This requires a theoretically clear distinction between self-monitoring processes and comprehension processes. (p. 210) Again, the difference between rote execution of skills and ‗actual‘ comprehension (whatever that may be) is illustrative here. It is completely conceivable that a child might excel, say, on anaphoric resolution tasks but fail to monitor whether the resolved sentences make sense to her—in which case we would not speak of comprehension. The issue culminates in the observation that the very scope of comprehension is unclear. When we move beyond circumscribed skills, its boundaries begin to give way. The reason for this is the intertextual nature of print; all written information is inextricably linked with other bits of written information. In its totality, text itself is boundless (Allen, 2000). If this is the case, then the same applies to comprehension of text. As I worded it in the memo: There is no ontological boundary to ‗comprehension‘. Take simple chemistry e.g. about the structure of water: Has the student understood the text if they can reiterate that a water atom is made of two hydrogen and one oxygen atom? Or (…) if they can render the internal structure of the atoms? The chemical processes of atomic bonding? The role of hydrogen, oxygen, and water in nature (…)? Or in the physiology of

295

living systems? And so forth (…). From this follows that ‗understanding‘ is always relative to what is situationally constructed as ‗(adequate) understanding‘. (Appendix B, p. B12) This discussion may appear philosophical in the derogatory sense of the term. But it highlights the very practical problem of assessment. We saw that the measures we blithely use to this end are ill-defined. The behaviors we equate with, or take as indicators of comprehension, are arbitrarily chosen subsets of all such behaviors. This is not to say that they are invalid, only that their validity is taken at face value, and that their constructed and limited nature is given too little consideration. In the end, I concluded, it is impossible to know, in the positivist sense of the term (cf. Damico & Ball, 2010), whether a reader comprehends, for we do not know exactly what ‗comprehension‘ means. In the same way, it is ultimately not possible to determine the ontological status of a comprehension problem. All we can say about comprehension or the lack thereof is that an expected behavior has or has not occurred: The ultimate judgment of the validity of the reader‘s behavior will have to come from a more competent reader—‗comprehension‘ will be socially constructed (…). We learn from Quine that ultimately, there is no translating between minds—(…)—there is only the assumption of agreement based on behaviors that accord with expectations based on that assumption (…). How do we know, then, that a child ―really‖ has comprehension problems (…)? Provocatively put, we may define comprehension as the ability to produce a certain behavior but from the child‘s perspective, comprehension may be something quite different (…). (Appendix B, p. B11)

296

In the light of these concerns, it is justified to treat comprehension as a quale: an ineffable, unitary experiential state that cannot be captured in any one behavior. Moreover, we cannot assume that such a quale has boundaries that can be found within a reader‘s brain, equating neural and mental structures. Rather, it emerges from the ongoing transaction between a reader and an environment, textual or social. Recently, authors in the philosophy of mind and language have begun to take seriously the idea that cognition and language do not simply happen in the brain or mind, but are located in the totality of the interaction of a mind and an environment (e.g. Noë, 2009; Thibault, 2011). In other words, what transactional theorists have said about reading decades ago (cf. Chapter Two) is now discussed with regard to all of cognition. In this light, and in the light of the ontological indeterminacy of comprehension, it is warranted to construe poor comprehension likewise as a quale: an unbounded, transactional phenomenon arising from the dynamics of reader-environment systems—an ‗emergent property‘ in the terminology of systems theory. Towards a Grounded Theory i): Poor Comprehension as an Emergent Phenomenon It is now possible to formulate an initial attempt at a grounded theory to answer our general research question. We said that while all the thematic aspects we found appear to inform our cases‘ poor comprehension (cf. Figures 5.1 through 5.4), it would not be plausible to locate poor comprehension in one or several of them. Rather, the issue arises from of all of them—more precisely, from the dynamics of their interplay over time. Thus, we have a true alternative to the causal-componential (‗structural‘) models on which most research has been predicated. Figure 5.5 attempts to capture such models, using the organization of the pertinent literature review in Chapter Two. In research based on componential models,

297

various observable behaviors are conceptualized as real-world instantiations of actual mental structures or processes such as linguistic processing, working memory, among others, each of which is treated as a possible candidate for a cause of poor comprehension. The underlying question is: ‗what caused this‘?

Text coherence and integration

Use of context information and background knowledge ?

?

Working memory

Inferencing

?

?

Metacognition and psychological factors

? Linguistic processing

? Poor comprehension

?

Language impairment and cognitive abilities

Figure 5.5: The Causal-Componential View of Poor Comprehension In stark contrast to this, a systems-theoretic outlook posits that no one component can be singled out this way. Rather, all aspects of reader-environment dynamics over time contribute to the behaviors we observe and collectively label ‗poor comprehension‘. From this perspective, the causal question is misguided. Instead of chasing after the one, elusive causal explanation, a humanistic systems view seeks to understand the issue. The underlying question in this line of inquiry is ‗what is this?‘ The answer to this question is a scientific proposition that, in contrast to its equivalent from the experimental genre, does not rely on the format ‗x causes y‘. Its details have been spelled out in narrative form above; put succinctly, it can be worded as a main proposition with two sub-statements:

298

i) Poor comprehension is an emergent property of a reader-environment system. a. It is not localizable within a reader or an environment, but emerges from the systemic dynamics of their interaction. b. It is not static but can change over time. The case for the main proposition and for sub-proposition a) has been made above; sub-proposition b) is implicit in our comparison of P and S. In the latter, change has clearly been observed—but even P changed over time, albeit in minor and unspecified ways, as both his mother and SLP testified. Figure 5.6 visually represents our conceptualization of poor comprehension as an emergent phenomenon for P‘s case.

Behavioral and psychological consequences HIGHER-LEVEL LANGUAGE ISSUES STRUGGLE WITH PREDICTION PIECEMEAL PROCESSING PREOCCUPATION WITH VISUAL SEARCH LACK OF MOTIVATION & NEED FOR SCAFFOLDING/STRUCTURE

Mild ID

Poor comprehension Intervention issues UNCERTAINTY (LACK OF) SUPPORT EMERGING AWARENESS EMERGING INTEGRATION LACK OF PROGRESS

Mild ID

Mild ID

Social-psychological and behavioral response OTHER-ORIENTATION FOCUS ON CODE STRUGGLE FOR ACCURACY

Language problems & psychosocial effects HALF THE STORY ISSUES EXPRESSING HIMSELF

Mild ID Academic experience READING VS. COMPREHENDING TANGIBILITY

Mild ID

Figure 5.6: The Systemic View: Poor Comprehension as an Emergent Phenomenon

299

Two objections likely to be raised need to be addressed at this point. The first concerns the details of the model above. Can we simply lump all our observations into the ‗emergent‘ bucket, the critic might ask? What about the HALF THE STORY or the UNCERTAINTY themes, for example? They do not capture observable properties of P or his environment, so how do they inform the overall phenomenon? The answer is that these are, indeed, qualitatively different from observable, behavioral aspects such as, say, PIECEMEAL PROCESSING. Yet they do contribute to the overall picture. HALF THE STORY is intimately linked to ISSUES EXPRESSING HIMSELF, which is unquestionably a part of what we observe and label poor comprehension. UNCERTAINTY, the connotative theme derived from my own ‗hunches‘ during the analytic process, is taken here to reflect actual uncertainty on part of all stakeholders—P, his parents, the school environment, SLP—regarding P‘s needs to foster his comprehension. Insofar as no one involved has a grasp on what P might need, UNCERTAINTY is, in fact, a factor in the emergence and maintenance of his issues. A second and more substantial objection would be that we applied an overly general model to a very specific phenomenon—harking back to Smith‘s and Thelen‘s (1993b) metaphor of a toddling researcher indiscriminately banging systems models against any kind of data. All complex phenomena have emergent properties; a critic could rightly point out that we have simply lumped a complex issue into a one-size-fits-all model where it uncomfortably coexists with completely different phenomena such as consumer behavior, bird swarm flight initiation, or neural network patterns. An account that pretends to be a ‗Theory of Everything‘ ultimately explains (or understands) nothing. This is a valid point, and it merits a twofold response.

300

First, as I have repeatedly pointed out, the theory formulated here can certainly not aspire to be the end product of our investigation into poor comprehension. The sheer paucity of our data precludes such pretense. At best, we can claim to provide a starting point for a different type of inquiry. The formulation above testifies to the feasibility and usefulness of qualitative investigation of poor comprehension, and to the re-descriptive possibilities of systems theory. It does not lay claim to ending the inquiry. Second, there are isomorphisms in systems across different fields of inquiry (cf. von Bertalanffy, pp. 80-86). Taking up our examples from above, phenomena such as consumer or bird swarm behavior do emerge from the dynamics of the systems that produce them, just as poor comprehension does according to our claim. Making this observation is not the same as drawing a facile analogy, a method von Bertalanffy warns against. He argues that equating, for example, the development of an individual with that of a group of individuals of the same species (such as a tree vs. a forest) or very different kinds of growth pattern (such as the growth of plants vs. that of crystals) is not useful; ―analogies are scientifically worthless‖ (p. 85). However, I contend that in calling poor comprehension an emergent phenomenon, we are not using analogy but homology: the comparison of phenomena in which ―the efficient factors are different, but the respective laws are formally identical‖ (p. 84). Homologies are helpful steps in arriving at explanations; von Bertalanffy cites the formalization of heat flow in terms of fluid dynamics as an example, which allows for using the same mathematical formulae in the description of both phenomena. As we are using systems theory in a metaphorical sense here, we cannot formalize our observations the same way; however, we can use what the author calls ‗logical homologies‘, giving the following example: ―If an

301

object is a system, it must have certain general system characteristics, irrespective of what the system is otherwise‖ (p. 85). Readers and their environments are systems, and in their transactions they form an overarching system together; hence, it is plausible that this readerenvironment system should display properties such as emergence. In other words, we are not trying to explain away the specific characteristics of poor comprehension by claiming it to be the same as, for example, swarm behavior. We are simply ascribing systems characteristics to what clearly is a system. The discussion could conclude at this point. However, it falls short in one crucial aspect: ‗emergence‘ is an abstract concept that so abstract that it arguably does not meet Bransen‘s (2001) requirement that verstehen-based inquiry re-describe phenomena until they make empathetic sense. It is difficult to see, in the light of our findings, how we could arrive at such an account. We were able to empathetically describe the trajectories of both our cases, but we also saw that dynamics were quite different between them. We accounted for this divergence by re-describing it abstractly, but how can we arrive at an empathetic account, given the variety we saw, and the variety we might encounter across a larger number of cases? It is possible, I believe, to capture our findings in somewhat more empathically ‗accessible‘ terms. As mentioned above, I started to explore the possibilities of an ecological metaphor early in the analytical process. While not exactly psychological or phenomenological, such a metaphor has the advantage that it captures our findings in more grounded terms than the highly abstract concepts of systems theory.

302

Towards a Grounded Theory ii): An Ecological Metaphor for Poor Comprehension As mentioned previously, while analyzing the data I found myself saying things like ‗yes, this makes sense, I‘m not surprised this kid would develop this way‘, and similar affirmations of the understandability of P‘s (and S‘) trajectory into poor comprehension. In other words, I came to perceive it as a natural process—akin to other natural processes like the stunted growth of plants when rooted in shallow or nutrient-deficient soil, or the decay of wood in a humid environment. And early on, I thought about using this perception as a conceptual tool for the formulation of a theory. Nature, in the various meanings of the word, is a concept close to the human heart and mind. To describe something as ‗natural‘ is to construe it in a sweepingly generalized yet intuitively accessible manner, invoking perceptions of an ‗essence‘ of things as well as the idea of an untouched but lawful whole (e.g. Ginn & Demeritt, 2009; Williams, 1980). As Ginn and Demeritt put it: Far from being something located ‗out there‘, nature is also something with us ‗in here‘, in the ways that our bodies, our sense of our selves and our world, and our daily routines are informed by various overlapping concepts of nature. (p. 308) In other words, human life is ‗naturally‘ conceptualized in terms of natural processes. The same, then, should be possible for human development. Trying to bring my impressionistic thoughts into a coherent theoretical form, I mulled over P‘s development in evolutionary terms, as seen in the quote from Memo 12 above. The metaphor I used was that of ‗natural drift‘, Maturana‘s and Varela‘s (1992) alternative take on Darwin‘s (2003/1859)

303

‗natural selection‘. I will briefly discuss both concepts to illustrate the weight such metaphors carry in descriptions of the human condition, and to elucidate why I chose the former. ‘Struggle’ vs. ‘drift’: A Darwinian heterodoxy. When we think about the theory of evolution and the scholar who first formulated it (Darwin, 2003/1859), the concepts that come to mind are, arguably, buzzwords such as ‗survival of the fittest‘ and ‗natural selection‘, and possibly ‗struggle for existence‘. They evoke a ruthless, merciless idea of nature, and, by extension, of human life in its natural state (Williams, 1980). It can be speculated that this unfortunate line of reasoning, amplified by some of Darwin‘s remarks on social issues made in other contexts (Paul, 2009) may be one of the reasons for the widespread rejection of the theory of evolution. What is rarely discussed in this context is the fact that Darwin used the latter two terms in a metaphorical sense. They were never meant to convey that nature actively selects individuals, or that beings fight one another daily for survival. Rather, they are the results of the necessity to describe non-anthropomorphic processes in a language evolved mainly to capture the human realm. Darwin was quite aware of this problem. Regarding ‗natural selection‘, he observed: In a literal sense, no doubt, natural selection is a false term; but who ever objected to chemists speaking of the elective affinities of the various elements? (…) Who objects to an author speaking of the attraction of gravity as ruling the movements of the planets? Every one knows what is meant and is implied by such metaphorical expressions; and they are almost necessary for brevity. (pp. 77-78, emphasis mine)

304

In discussing ‗struggle for existence‘, he is even clearer that the empirical realities he intends to cover with the term are quite different from its metaphorical connotations—so different, in fact, that they may be called incompatible: I use this term in a large and metaphorical sense including dependence of one being on another (…). A plant on the edge of a desert is said to struggle for life against the drought, though more properly it should be said to be dependent on the moisture. (….) The mistletoe is dependent on (…) trees, but can only in a far-fetched sense be said to struggle with these trees (…). Darwin (2003/1859, p. 62, emphasis mine) In other words, a term which generally invokes imagery of mutual obliteration, and by extension to the human world as ―the social jungle, the rat race, the territory-guarders, the naked apes‖ (Williams, 1980, p. 82), was originally intended to cover all kinds of organismic relationships, including those of interdependence—a profoundly ethical concept (cf. Cilliers, 1998; Maturana & Varela, 1992). It is also worth noting that ‗fittest‘ in ‗survival of the fittest‘ was never meant to signify ‗strongest‘, but refers to ―best adapted to a given and variable environment‖ (Williams, p. 82). This discussion highlights two insights of importance for present purposes. First, it confirms that metaphors of nature, especially when used to capture realities of human life, are highly charged and powerful. Second, it illustrates vividly that our choice of metaphors matters. Despite the fact that Darwin never endorsed ‗social Darwinism‘ or eugenics, both ideas were advanced based on the metaphors he used (Paul, 2009). Maturana and Varela (1992), mindful of the undesirable connotations of ‗selection‘ (p. 101), introduce an alternative metaphor for evolutionary processes. ‗Natural drift‘

305

removes any notion of an active environment. It builds on the systems-theoretic premise that systems and their environments are parts of an ongoing interaction. There is no one-way street of environmental pressure impinging on a system; their intertwined dynamics affect both. This is easy enough to see in human systems: Take, for example, P‘s mother‘s account of his social exploits, where his expressive difficulties interact with impatience on part of his peers, leading to OTHER-ORIENTATION. But how do Maturana and Varela justify using the same notion for large-scale evolutionary processes? After all, climate and vegetation zones are environments ‗given‘ to an organism, and the task of the latter can only be to adapt. Not quite, the authors respond. First, some organisms do physically change the environment. As an example, early living cells emitted oxygen, altering Earth‘s atmospheric makeup. (A similar observation could be made about contemporary humans.) More importantly, the idea that organisms evolve simply by increasing their ‗fitness‘ is misguided. Evolution cannot be described in terms of organisms maximizing use of their environment by conforming to its pressures. Living systems respond to environmental stimuli by altering their internal structure; their evolutionary path is determined as much by previous systeminternal states as it is by environmental input. Also, the organizing principle is not optimization but fit. That is, as long as self-organization is maintained, the organism‘s adaptation is ‗good enough‘ (Maturana & Varela, 1992). The authors do not actually use this latter phrase; it stems from a related discussion in Thelen and Smith (1994, pp. 145-146), who compare natural selection to the purchase of a suit. The classical reading of the ‗fitness‘ metaphor, they argue, is analogous to having a suit tailor-made: there is one, and only one best fit, with very precise specifications. More often than not, however, the buyer simply chooses from a variety of options available. There are

306

constraints on the decision, such as taste, fashion, purpose, or budget, but since there are multiple options, there is no one ‗optimal‘ solution: ―[the] final choice has the form of satisfying some very loose constraints (…) [but not] of a fit—even less so of an optimal fit— to any of these constraints‖ (p. 146). Obviously, this analogy does justice neither to the complexities of systemenvironment interactions, nor to the fact that most ‗decisions‘ made by living systems, including those of interest in the present study, are very much not like conscious picks from predetermined choices. But it is helpful to illustrate Maturana‘s and Varela‘s (1992) redescription of evolutionary theory. Living systems are not shaped in ruthless competition with one another, nor are they in a ‗race of arms‘ for superior adaptation. Rather, they maintain their structural integrity in a good-enough fashion while ‗drifting‘ through spatial and temporal environments. Evolution operates in this natural drift: it is neither the strongest nor the fittest who survive, but those who are on continuously good enough terms with their situation, maintaining an ecological ‗niche‘. Poor reading comprehension as a quasi-ecological ‘niche’. The concept of ‗niche‘ lends itself to adding an element of verstehen to our abstract theory by employing a more evocative metaphor than ‗emergence‘. Niche, roughly synonymous with ―setting, context, habitat, environment‖ (Smith & Varzi, 1999, p. 1), evokes connotations of belonging or goodness of fit, which are intuitively at odds with the popular conceptions of Darwinist struggle discussed above, but mesh well with the idea of natural drift. An organism‘s niche is its ‗home‘, not only in a spatial sense, but in a way close to the anthropomorphic meaning of the term; in the classic formulation, Smith and Varzi report, it can be thought of ―as a way of making a living in an organic community‖ (p. 4). They quote Elton (1927):

307

When an ecologist says ‗there goes a badger‘ he should include in his thoughts some definite idea of the animal‘s place in the community to which it belongs, just as if he had said ‗there goes the vicar‘. (Smith & Varzi, 1999, p. 4) This notion of ‗niche‘ does not solely draw on physical-spatial factors, but incorporates an element of ‗values‘ in von Bertalanffy‘s (1968) sense. We may speculate that this conceptual versatility helped advance its use in various disciplines ―from biology to economics‖ (Smith & Varzi, 1999, p. 1), including the study of the evolution of language (Odling-Smee & Laland, 2009), of particular interest for present purposes. The latter authors make clear that a niche is not simply a given environment into which organisms strive to fit. Similar to Maturana‘s and Varela‘s (1992) reasoning, they advance ‗niche construction theory‘: organisms, they hold, are active in constructing their habitats from environmental parameters. This notion is exemplified by the earthworm, whose habitat is a self-constructed niche of soil altered structurally and chemically by the worm‘s activity. Figure 5.7 illustrates their theory and captures all the factors that contribute to a species‘ evolution over time: genetic inheritance; environmental constraints; and the ongoing, mutual interaction between organisms and environment, conceptualized here as ‗niche construction‘ and ‗natural selection‘.

308

Figure 5.7: Niche Construction (from Odling-Smee & Laland, 2009, p. 100) It should be noted that this representation is somewhat simplified. From a systemstheoretic standpoint, it is the entirety of organism-environment interaction, including genetic and ecological inheritances, which represents the niche in question. Importantly, the authors extend this notion to social-communicative and cultural niches—which, they argue, may have provided the environment in which language evolved. We may thus note that niches can come in manifold forms and shapes: spatial, functional, conceptual, interactional and cultural among others. Elaborating on my ‗hunch‘ in Memo 12, then, we can advance the idea that poor reading comprehension might best be redescribed as a niche—more precisely, an emergent habitat made of social-environmental experiences, psychological attitudes, and concrete behaviors, into which poor comprehenders drift naturally as a function of the interaction of their system states and environmental parameters.

309

We mentioned that ‗fitness‘, in a systems view on evolution, is not an optimum but a good-enough match. With this in mind, consider how all our observations made in P‘s case fit together. We found understandable relations between all aspects we scrutinized. For example, the phonetic peculiarities of his productions (STRUGGLE FOR ACCURACY), his FOCUS on surface aspects of print, the expectations of his academic environment that literacy development be TANGIBLE, and a general agreement that COMPREHENSION and READING are different; but, all these aspects form a seamless whole which is comprehensible to an empathetic human observer. In addition, we may have found an additional obstacle to his development, namely a mild intellectual disability. Taken together, P‘s poor comprehension can be described as a good-enough fit of his behavior to his intrinsic abilities, his experiences, and his environment, that is, as a niche. To fit these observations to the model, we simply redraw it as a representation not of intergenerational dynamics but of P‘s own trajectory over time. Our thematic codes are jointly depicted as emergent properties of system-environment interactions, possibly including a mild intellectual disability as part of his genetic makeup (either in a literal sense, or metaphorically as an unchanging systemic parameter). We did not study his environment directly in this study, hence we cannot provide concrete ascriptions to this part of the model, but pertinent factors (TANGIBILITY, READING VS. COMPREHENDING) are implicit in the emergent interactions we captured above (See Figure 5.8). Of course, the mild intellectual impairment is merely surmised here and has no specific requirement as an emergent property of the interactions but any such genetic or environmental constraint will play a role in the determination of a ―good-enough‖ fit within the niche of comprehension (good or poor).

310

Emerging properties of interaction

Mild ID

Environmental factors

(social, psychological, behavioral)

Emerging properties of interaction

Mild ID

(social, psychological, behavioral)

Figure 5.8: The Niche of Poor Comprehension (for P’s case) Two likely objections come to mind. The first one pertains to the general idea of conceptualizing poor comprehension as a habitat. As it is not functional for practical purposes, should it not rather be thought of as akin to a maladaptation (Crespi, 2000)? The answer to this is a version of ‗yes, but…‘ For academic and professional life as seen from a conventional, adult standpoint, poor comprehension is certainly maladaptive: it has already hampered P‘s success at school, and is likely to continue to affect him. From an empathetic verstehen perspective, however, it makes perfect sense that P, burdened with a variety of obstacles to literacy development, should end up in this peculiar niche. It fits his systemic makeup, provides good-enough interaction with his environment, and has served him well so far. The absence of motivation and frustration we observed speaks to the idea that he is comfortable with his situation. From his viewpoint it makes for a viable habitat. Which leads to the second objection: this model may be applicable to P, but clearly not to S, who was greatly distraught by his situation. This is true; but I contend that the

311

general idea of a natural drift into poor comprehension still applies to S. The difference is that his niche was maladaptive for him: it did not make for a viable habitat from his viewpoint. We may assume that it was precisely this perception of being ‗in a bad spot‘ which provided for some motivation for change and which, together with the absence of intrinsic difficulties, allowed him to make greater progress than P. Maladaptation has not been studied as much as adaptation, and various definitions coexist, as do suggestions for evolutionary mechanisms that lead to it (Crespi, 2000; Nesse, 2005). One definition applicable to our purposes is to describe it as ―the distance of a population from the nearest adaptive peak‖ (Crespi, p. 624). In a metaphorical sense, we could think of S‘ niche as remote from the adaptive peak we might call his ‗comfort zone‘. For P, by contrast, niche and comfort zone are currently sufficiently coextensive. One of the explanatory mechanisms for maladaptation suggested by Nesse (2005) fits well with this notion and with our theoretical framework: living systems are path-dependent. That is, their trajectory is constrained by prior system states, which places constraints on their evolution. We discussed the differences in P‘s and S‘ history, speculating that cognitive makeup and family dynamics may have been factors in their divergent trajectories. S may have had less of a comfort zone than P to start with; P, burdened with intrinsic obstacles to the exploration of new territory, may have found his niche too comfortable to give it up. Having addressed potential objections, we are now in a position to formulate the second part of our tentative grounded theory, using a more evocative metaphor than ‗emergence‘ to re-describe poor comprehension. As before, our theory consists of a general statement with two sub-propositions; the second sub-proposition addresses the developmental diversity we observed.

312

ii) Poor comprehension can be construed as a psychosocial-behavioral niche. a. It is an understandable behavioral fit of a reader’s intrinsic and experiential characteristics to an environment. b. It can be adaptive or maladaptive from the reader’s viewpoint. This formulation concludes the discussion of our findings. We were able to show that poor comprehension, when conceptualized not in causal-componential but in systemic terms, and couched in an ecological metaphor, can be understood in the sense of verstehen epistemology (Bransen, 2001). The ramifications of this demonstration will be discussed in the final chapter.

313

Chapter Six: Conclusion The ingrained mechanistic way of thinking which comes into difficulties with modern scientific developments is a consequence of our specific linguistic categories and habits. von Bertalanffy (1968, p. 225) No scientific explanation of human behavior could ever be complete. In fact, no unpoetic description of the human condition can ever be complete. Berliner (2002, p. 20) Pitting Whorf against Kant in his discussion of the cultural relativity of categories, von Bertalanffy (1968) advances the notion that our quest for causal relations—which are viewed as a fundamental conceptual category by Kant—is based on verbal habits. We are so used to talking about anything, even about psychosocial processes, as analogous to material particles impinging on each other, that we cannot help looking for causal culprits when things go wrong. Non-Indo-European languages such as Hopi, by contrast, construe the material world in psychological terms; it can be speculated, the author concludes, that a Hopi science would devise propositions quite unlike those based on Aristotelian and Kantian categorizations. The main thrust of this study has been to offer an alternative way of thinking and talking about poor reading comprehension; the conclusion will therefore place prime focus on the consequences that arise from this alternative. I will begin with its theoretical ramifications, discuss some clinical suggestions, and outline the limitations of the present effort before suggesting potential future lines of research. The final part of this conclusion will explore the ‗completeness‘ of the description proposed here, couched in an epistemological discussion. In order to preempt the idea that our description is ‗true and complete‘, it will be argued that our findings are best viewed within the pragmatist framework introduced in Chapter Three.

Theoretical Implications Our discussion so far has import on three areas related to poor comprehension. First, there is the phenomenon itself. Based on our findings, it can be argued that the ways we think and talk about it may benefit from reorientation. Second, the issue of diagnostic methods will be touched upon; it will be asked how (and how far) our methods of assessing and diagnosing the phenomenon are linked to a particular way of thinking about it. Finally, I will discuss in how our findings are compatible with the Simple View, which has informed so much of our thinking on poor comprehension. The concept of poor comprehension. In our review of the literature (Chapter Two), we discussed the ‗rift‘ between the publications prior to and those after Yuill and Oakhill (1991). Earlier authors were not hesitant in interpreting their findings; poor comprehension, they argued, arises when readers pay undue attention to lower-level aspects of print. Failure to construct coherent mental models, and ensuing lack of motivation, are due to deficient schemata for reading. Several authors unequivocally trace these issues back to children‘s instructional experience, blaming educational focus on decoding for the phenomenon. Later authors, by contrast, look for causal connections between presumed cognitive components, leaving children‘s environment largely out of the picture. On occasion, publications from these two decades do touch on wider factors such as motivation (Cain, 1994, referenced in Oakhill & Yuill, 1996) or exposure to print (Stanovich et al., 1996). Most, however, stick with a cognitive-componential framework. It would be tempting to claim that the present study has corroborated the validity of earlier investigations. But that would amount to a coarse simplification of our findings. From a systemic view, we cannot simply reverse the presumed causal direction, replacing an

315

intrinsic with a behavioral or environmental explanation, and assert that children‘s poor comprehension is caused by a focus on decoding, or by inadequate instruction. We saw how numerous factors contribute to the emergence of poor comprehension; some are readerintrinsic, others properties of the reader‘s environment; yet others emerge from the interaction of reader-intrinsic and environmental factors, and their emergence occurs dynamically over time. In this sense, all our observations can be said to contribute to the phenomenon. Note also that if there were a true causal relation from instruction to behavior to lack of comprehension, all children in P‘s and S‘ schools would have to be poor comprehenders. We do not have any data to back up or refute such a claim, but it is highly implausible that this should be the case. In sum, causal explanations fail when applied to the experiential realm, just as they do on the cognitive-componential plane. How, then, does our study fit with existing findings? The answer is that it offers a descriptive alternative that both accommodates and transcends earlier findings. Construing poor comprehension as a systemic, emergent phenomenon resolves puzzles that have baffled researchers for decades. The elusiveness of causal factors is no longer a problem. The surprise at the fact that most poor comprehenders respond well to any kind of intervention evaporates—from a systemic view, it is expected that changes in reader-environment dynamics engender changes in emergent properties of these dynamics. The vexing fact that the population has wildly divergent etiologies is accommodated by the systems-theoretic notion of equifinality. Differences in responses to intervention, such as those found in this study, can be accounted for by the notion of equipotentiality. In sum, our systems-theoretic re-description allows for a plausible and understandable account of the phenomenon, and accommodates existing data. Causal-componential and

316

causal-experiential approaches have failed in both regards. In this sense, our account goes beyond existing explanations, providing a truly novel angle of view on the issue. This insight has important implications for the way we view poor comprehension. Insofar as we think about it as a disorder (which is the prevalent view, despite the fact that it is not included in the latest edition of the DSM, [cf. Snowling & Hulme, 2012]), our findings construe it as qualitatively different from various other disorders of communication. We identified it as an emergent phenomenon, placing it in the same category as bird flight initiation or neural network patterns; a property of complex reader-environment systems that arises from, but is not reducible to, the interaction of aspects of those systems. This conclusion establishes an unambiguous delineation to other conditions that are not emergent. A brief glance at any textbook in the field (e.g. Owens, Farinella, & Metz, 2015) reveals various conditions with agreed-upon causes and established etiologies, such as aphasia, apraxia, or dementia, among others. All of those have in common that they can be explained by reference to causal factors (even though the different forms they take in individuals could be described in terms of emergence). Poor comprehension, by contrast, is not explainable in causal terms, whether we look at it from a cognitive-componential or an experiential angle: it is ontologically and etiologically different from causally-based disorders. This outlook is novel and unusual for the field of communication disorders, but it meshes well with recent developments in the study of (disordered) language and cognition, where these phenomena are tackled from an ecological, systemic viewpoint. Fowler and Hodges (2011) as well as Thibault (2011) conceptualize ‗languaging‘ as an embodied phenomenon embedded in and emerging from ongoing patterns of interaction. It is not found

317

within any particular interlocutor but ‗distributed‘ across systemic interactions and their contexts. As Thibault puts it: Language (…) is spread across diverse spatiotemporal scales ranging from the neural to the cultural. It is not localizable on any one of them, but it involves complex interactions between phenomena on many different scales. (p. 1) The same can be said even of the so-called ‗components‘ of language. Pragmatics, for instance, has been described as the emergent outcome of the interaction of ―a range of linguistic, cognitive, sensorimotor and sociocultural elements‖ (Perkins, 2005, p. 15). Both views can effortlessly be applied to comprehension: as comprehension is based on language use, it can equally be said to be spread across the levels invoked by Thibault (2011), or the elements listed by Perkins (2005). This outlook is consistent with transactional perspectives on reading, and the general systemic view espoused in the previous chapter. The same, then, should apply to poor comprehension, and this alternative view has important implications for the assessment of poor comprehension, and for our theoretical modeling of reading in general. Each of these issues will be addressed in turn. Assessment of poor comprehension. In Chapter Five, we made the argument that comprehension, and consequently poor comprehension, should be thought of as a quale: an ineffable mental state whose various aspects are ‗distributed‘ (Thibault, 2011) across comprehenders and their environment. If this perspective is viable, it raises the question whether the assessments we use to diagnose the phenomenon do justice to its non-bounded nature. Recall that poor comprehension is typically assessed using two instruments, one that measures decoding of unrelated words, and one that measures comprehension. We discussed

318

the difficulties associated with the assessment of comprehension in the previous chapter; we are now in a position to subject the overall procedure to criticism. I mentioned previously that I administered two rounds of assessment to P. In the first one, I used the GORT-4 (Wiederholt & Bryant, 2001) to address comprehension and decoding, and the GSRT (Wiederholt & Blalock, 2000) to compare results from the readaloud measure to silent reading. Because results from the GORT-4 were ambiguous due to its dual ceiling requirements, possibly exacerbated by the design of the comprehension questions (Keenan & Betjemann, 2006), I subsequently administered a decoding-only test (WLPB, Woodcock, 1991) and based my diagnosis of P on the latter and the GSRT. I discovered a curious disconnect between the GORT-4 and the WLPB. P‘s fluency score on the GORT-4, a combination of rate and accuracy, placed him at a 3.7 grade equivalent. By contrast, his score on the WLPB decoding task indicated adult-like proficiency. The WLPB does not address reading rate, whereas the GORT-4 does; it could be argued that the former simply did not capture P‘s slow decoding and gave an exaggerated picture of his decoding abilities. This is a distinct possibility, especially in the light of his STRUGGLE FOR ACCURACY. However, I did not note any particularly laborious or slow articulations during decoding. Hence, another speculation is worth entertaining here. The reason for P‘s differential performance may be found in the different task demands. The WLPB assesses decoding using a list of nonwords. The only requirement is to articulate them. The GORT-4, by contrast, presents the reader with an actual text, and the requirement to answer comprehension questions after reading it. If we take seriously the idea that language is a distributed phenomenon, then it is to be expected that different types of reader-text transactions engender different linguistic productions.

319

Figure 6.1 illustrates this finding using the stock taxonomy of reading proficiency based on the Simple View (Catts, Adlof, & Weismer, 2006; poor comprehension is termed ‗specific comprehension deficit‘ in this schema). The triangle represents P‘s schematic position as per the eventual set of assessments (the WLPB and GSRT): his comprehension is poor, while his word recognition is extraordinarily developed. The square assigns his place according to the original assessment (GORT-4), leaving out, for present purposes, the differences between GSRT and GORT-4. According to this measure, his comprehension is still poor but his decoding is much less advanced. In other words, on the former set of measures, P‘s profile is an extreme version of a poor comprehender profile; on the latter, his profile much less extreme, if still within the range of poor comprehension.

Figure 6.1: P’s Places in a Simple Taxonomy of Reading Difficulties Recall also P‘s STRUGGLE WITH ACCURACY as observed when he read authentic texts. In those contexts, he appeared much more similar to a ‗mixed deficit‘ type reader, especially

320

when taking into account the subtle phonetic and prosodic deviations he produced. Which begs the question: which of these profiles is true? From a systemic, ecological view, they all are, for they all represent P‘s abilities in a certain systemic context. However, this calls into question whether poor comprehension is really best thought of as a category, a ‗disorder‘ with clear boundaries to other disorders and to proficient reading. This is not to say that it does not exist as a distinct profile. However, there might be merit in thinking of this profile, not as a ‗bucket‘ with boundaries, but as being located on an open continuum with other profiles. Models of reading. The observations we have made so far have important implications beyond the study of poor comprehension proper. They have the potential to inform the ways reading in general is modeled and assessed. Recall that the main theoretical divide found in the literature is between the Simple View on the one hand, and the various transactional and integrated models on the other. The Simple View holds that reading is a function of two independent yet correlated skills, decoding and comprehending: the former is thought to be specific to reading, the latter a function of general linguistic abilities. Transactional and integrated models, by contrast, posit intricate, non-linear interactions between all aspects of language; grapheme-phoneme relations, in this view, are seen as merely one aspect part of written language, no more central than any other aspect. P‘s results show that both models can be used to describe the reading process. However, we get a substantially different picture, depending on which angle of view we adopt. In the assessment modeled on the Simple View, with comprehending and decoding separated, P excelled in decoding and was below expectations in comprehending. In an assessment modeled on natural reading, his performance in both areas converged as his

321

decoding score dropped dramatically. Further, an in-depth analysis of his behavior in (somewhat) authentic reading tasks revealed performance even closer to that of an overall poor reader, as his accuracy deteriorated in subtle ways not captured in the assessments. Thus, it seems that decoding, when theoretically and descriptively disconnected from other language skills, takes on a ‗life on its own‘, yielding a picture quite different from descriptions that integrated with other aspects of reading. The question, of course, is whether one way of conceptualizing reading is better than the other. From a pragmatist viewpoint—the epistemology espoused in this study—this question can only be answered with reference to what we want to achieve with our descriptions. If we wish to describe reading in a way that captures oral reading performance disconnected from comprehending, the Simple View seems to be a good enough tool. However, as mentioned at the onset of Chapter Two, scholars across the theoretical spectrum agree that comprehension is the ultimate goal of reading instruction. In this light, it can be questioned whether the Simple View is suited to the task. First, it does not appear to coherently describe what happens during the reading of actual texts (i.e., a type of reading where comprehension is paramount). Second, it engenders erroneous reasoning that illustrates lack of understanding of the processes involved as they are revealed through indepth description. These claims will be illustrated using quotations from the literature. Consider the following statements from Snow and Juel (2005). The authors claim that, ―there is now abundant evidence that the prediction model of reading is incorrect‖ (p. 507). In other words, they dismiss a basic tenet of the transactional model, namely, prediction of upcoming text by means of linguistic cues, as wrong. And they quote Perfetti (1985):

322

The main failing of [Goodman‘s] approach (…) is that it does not recognize that one of the ―cueing systems‖ is more central than the others. A child who learns the code has knowledge that can enable him to read no matter how the semantic, syntactic, and pragmatic cues might conspire against him. (p. 507) This proposition encapsulates the main tenet of the Simple View; a disconnect between decoding and other language skills, and implies that the former is more foundational than the latter are. Note also the somewhat odd wording that cues do not ‗conspire‘. This phrasing suggests a ‗passive reader‘ model (cf. Chapter Two), according to which the reader‘s task is to correctly recognize external stimuli (code or cues). Our in-depth miscue analysis in Chapter Four presents an alternative way of describing what happens during reading. Take Figure 4.1 and 4.3 as illustrations. In these examples, P produced Okay, places, everyone. as Okay, places. Everyone. Likewise, he rendered ―There they are!‖ Rex clapped his tiny plastic tyrannosaurus hands… as ―There they are!‖ Rex clapped. His tiny plastic tyrannosaurus hands… In both cases, the resulting sentence was nonsensical. The prosodic-syntactic alternation cannot be adequately conveyed in writing, but as mentioned in Chapter Four, P produced places with an intonation curve that would have been adequate if the word had been a term of address. Likewise, the prosodybased shift in syntactic structure of the second example yielded the ungrammatical, verb-less sentence His tiny plastic tyrannosaurus hands as Sarge and two small Green Army men trooped into Andy‘s room. We cannot know exactly in what way P comprehended or failed to comprehend these sentences. Either way, his verbal productions were clearly inaccurate. A theoretical account

323

of this observation which does not incorporate prediction and the use or misuse of syntactic cues cannot explain (or understand) how this inaccuracy came about. Based on the Simple View, all we can say is that P‘s decoding was erroneous. But this statement falls short in two ways. First, P‘s decoding in the strict sense of the word—translation of graphemes into phonemes—was accurate: he did not produce any phonemic deviations. Second, unless we wish to dismiss his productions as ‗noise‘ or ‗random deviation‘, we have to explain their occurrence. They could be dismissed if they were dissimilar to actual language. But as they do resemble actual prosodic forms of actual syntactic constructions, the best way of accounting for their occurrence is to infer that P was making an erroneous prediction regarding the upcoming syntactic structure. Note that there is no way of ‗proving‘ (or rather attempting to refute) this statement in a controlled, experimental way. There is no one syntactic structure that would prompt a significant proportion of readers to incorrectly predict an alternative structure. It is not even predictable that P would produce this structure the same way if I asked him to read the text again. However, interpreting it from a verstehen viewpoint enables us to understand its occurrence. If nothing else, such a viewpoint shields us from making erroneous assumptions such as the one in the quotes above. Our analysis shows that knowing the code, in and by itself, does not ‗enable a child to read‘. If the goal of reading is comprehending, then P is not reading, and his knowledge of code is not sufficient for him to read. A host of linguistic strategies, of which decoding is but one, have to be used in order to predict upcoming print and make sense of it. Another statement which warrants scrutiny comes from Nation (2005). Drawing on earlier accounts of poor comprehension, she asserts:

324

It seems unlikely that such children are devoting excessive resources to word identification and decoding, or that their reading comprehension is severely compromised by inefficient word-identification processes. (p. 256) Note the implicit assumptions behind this claim: the Simple View looms large in the juxtaposition of comprehension and decoding. Also, the author is lumping together Cromer‘s (1970) ‗difference‘ and ‗deficit‘ perspectives, or an ‗active‘ and a ‗passive reader‘ view. Research on poor comprehension since Cromer has indeed shown that poor comprehenders are good decoders (i.e. their ‗word identification processes‘ are efficient). But this observation is different from the observation that they use lower-level reading strategies at the expense of higher-level resources. The latter phenomenon was described in the older literature prior to Yuill and Oakhill (1991); as mentioned, it was based on an ‗active reader‘ view geared towards describing what readers do when they read. From a ‗passive reader‘ perspective, which conceptualizes reading as the reader-internal response to incoming stimuli, both statements appear to make the same claim. But they are, in fact, empirically unrelated. As established during assessment, P‘s decoding was efficient, and his word identification appeared to be intact as well. But consider P‘s PIECEMEAL PROCESSING behaviors, illustrated in Figures 4.6 through 4.10. All of these show, not undue attention to decoding in the strict sense of the term (i.e. processing of words one grapheme at a time), but an exclusive focus on parts of words and sentences—morphemes, syllables, words and phrases. Take his production of The red-haired cowgirl as The red-haired | cow|girl. Each individual word (including the two stems of the compound term cowgirl) is identified and

325

produced accurately. And yet, the two hesitations turn this simple descriptor phrase into an unrelated string of words. This is even more significant insofar as he has read the text before, and should be familiar with the phrase, as it is the frequently occurring name of a main character. Similar observations can be made for P‘s recurrent struggle with the name Lightoller (Figure 4.15), and for his failure to use the second stem of the compound word halfpipe to predict the first (Figure 4.35). It is not clear how these observations can be accounted for within a theoretical perspective based on the Simple View and ‗passive reader‘ models. P‘s decoding was sufficiently advanced (if varying by context) to place him in the poor comprehender category; his word identification is intact as regards single, simple words. What he lacks is the ability to construct morphosyntactic wholes from the single words he processes, and the ability to predict upcoming structures, even when they are recurrent or when he is familiar with them. It would be possible to discount the notion that prediction is involved as mere speculation. But this would lead to a double gap in our understanding of what is going on. First, we would be unable to account for why P fails to comprehend despite adequate lowerlevel skills; operating from the Simple View, we could only make a vague and general reference to linguistic issues, and we would have to disregard the conspicuous patterns of PIECEMEAL PROCESSING in the data. Second, we would have difficulty accounting for the fact that P‘s decoding abilities in isolated nonword tasks do not match his word recognition skills in authentic reading. This observation is neatly captured in the proposition that a mere decoding task is qualitatively different from reading precisely in that the latter requires integration and prediction. If we wish to forego this interpretation, the only alternative would

326

be to claim that P may not be a poor comprehender after all, but a poor reader, as his decoding is deficient. Yet his assessment results are clear, so this claim is unsustainable. In addition, the VISUAL SEARCH behaviors we observed in his eye movements pose a problem for a ‗passive reader‘ perspective and for the Simple View. Neither provides an explanatory framework for the fact that P scanned text in highly inefficient ways, using more fixations with shorter durations than typical readers his age, and occasionally even searching for information where there was none. From the transactional model espoused in this study, the interpretation of these findings is straightforward. P fails to predict upcoming information, which leads to halting, bit-by-bit processing of print, and is understandably connected to comprehension failure: if no integrative strategies are used to arrive at a coherent representation of linguistic input, comprehension cannot occur. His VISUAL SEARCH behaviors are understandable within the same framework: in the absence of linguisticpredictive strategies, with no resources other than print, it is unsurprising that P should resort to excessive scanning. In this sense, then, poor comprehenders do ‗devote excessive resources to word identification‘, and to dealing with low-level aspects of print in general. Importantly, this theory-based inference is consistent with all existing data on poor comprehension. In fact, some findings proffered by researchers who operate within the Simple View can be better explained with a model that embraces prediction. An example is the work of Nation and Snowling (1998), who found that poor comprehenders are less effective at using background and context information (cf. Chapter Two). This brings us to a final statement that merits consideration. Summing up their discussion of best practices in literacy instruction, Snow and Juel (2005) resolve:

327

In our view, then, the findings from a wide array of sources (…) converge on the conclusion that attention to small units in early reading instruction is helpful for all children, harmful for none, and crucial for some. (p. 518) The present study offers insights which merit reconsideration of this blanket proposition. Consider our findings regarding instructional focus on lower-level aspects of print and TANGIBILITY, and the underlying Simple View philosophy on which these foci are based. In a limited sense, it can be said that both P and S were taught to be poor comprehenders. This did not happen intentionally, of course: it is a result of the interaction of institutional emphasis on ‗small units‘ with our cases‘ own intrinsic difficulties and developmental pathways. It is also clear that this emphasis did not help P and S to move beyond small units, but got them ‗stuck‘ on them, which in turn hampered their progress. Thus, in order to reflect our findings, the above statement would have to be reworded to read: ‗attention to small units in the absence of thorough instruction in higher-level strategies and meaning construction is not helpful for all children, and may be harmful in the presence of obstacles to typical literacy development‘. In sum, then, we have to conclude that the Simple View is a coarse conceptual tool, ill-suited for capturing the intricacies and complexities of reading. At best, it provides us with a handy shortcut to classifying reading difficulties (with the limitations we discussed above). At worst, it skews authors‘ perspectives, leading their reasoning astray. Recall Perfetti‘s statement above, quoted in Snow and Juel (2005). Based on our findings, it is clear that if a child knows the code but the other cues continue to ‗conspire‘ against him, his pathway to becoming a reader is blocked. This is not a final verdict, as most children in this situation

328

respond well to intervention specifically geared towards helping them comprehend. But as long as knowledge of the code is the only strategy available to the young reader, as long as other cueing systems are absent, the child in question is not a reader, but a poor comprehender. A final illustration for the infelicitous conceptual consequences of the Simple View is found the works of its very originators. Gough, Hoover, and Peterson (1996) use the following vignette to illustrate their view of the reading process: Our favorite example is drawn from the life of John Milton. In his dotage, Milton wished to reread the Greek and Latin classics, but he was going blind. So he taught his daughters to decode Greek and Latin. They read the classics aloud while he listened to them. Between them, there was reading comprehension. (p. 3) Unfortunately, the conclusion that the vignette describes reading comprehension is misguided. Reading comprehension does not happen between persons, but between a reader and a text. At least this is true for the mature form of reading, the goal of literacy instruction. The best we can say about the scenario above is that it resembles a preliminary or ancillary exercise for reading where students read to the teacher texts that are meaningless to them (akin to, say, a nonword assessment). A less benign interpretation is possible, though. Note that Milton‘s daughters have been painstakingly taught to attend to lower levels of print, but not to comprehend what they decode. And due to intrinsic and experiential factors—lack of familiarity with and instruction in the languages in question—they are unable to construct meaning for themselves. As a result, only Milton comprehends what is read. This is very similar to the situation I

329

encountered with P, and we may speculate that it is also representative of multiple other situations in P‘s life. From this angle, then, the vignette is by no means illustrative of reading in the conventional sense. What it shows is an instance of poor reading comprehension. Clinical Implications The clinical insights gained from the present study are less straightforward than the theoretical ones. In fact, the picture seems less clear now than it did prior to this inquiry. Recall that any type of intervention aimed at fostering higher-level language skills generally proved successful with the population (cf. Chapter Two). By contrast, the present study revealed a case where intervention remained largely unsuccessful. We do not have the data to explain (or understand) this finding, hence it is difficult to make pertinent suggestions. The best we can do is to offer some plausible speculations. I will do so, first gearing my recommendations specifically towards P, and subsequently discussing some general insights pertinent to the overall population. P’s case from a clinical angle: The pitfalls of isolated intervention. Our overarching finding was that poor comprehension can be described as a psychosocialbehavioral ‗niche‘ into which readers drift naturally. For readers who feel ‗maladapted‘ in that niche, as in S‘ case, any type of appropriate intervention should be successful, as it taps into the reader‘s motivation for change. We saw this mechanism play out for S as described by SLP (even though she was doubtful about the extent of his progress). By contrast, readers such as P, who find their ‗niche‘ quite hospitable, may not easily be motivated to overcome the multiple barriers stacked up against them. The most obvious strategy to improve intervention for P, then, would be to alter his environment to where his ‗niche‘ is no longer comfortable for him. Based on my familiarity with SLP, I deem it safe to assume that she did

330

provide such an environment during clinical sessions. However, if his ‗real‘ environment outside of therapy does not communicate expectations of comprehending, clinical efforts may well evaporate. Authors who operate from integrated and transactional perspectives have long argued that literacy development is not simply a matter of acquisition of skills, but a socialinteractional process that involves much more than learning how to interact with print. As Smith (2004) puts it, identification with people is the prime vehicle of all knowledge acquisition: learning, including learning to read, happens when learners are motivated to become readers like the people around them. Similarly, Damico, Nelson, and Bryan (2005) conceptualize literacy as a ‗sociolinguistic process‘ that is acquired in authentic interactions where a proficient reader models for the learner what it means to be a reader. Based on our data, we have reason to assume that P has lacked such opportunities for learning. Early efforts by his mother were stifled by his lack of comprehension, likely due to his recurrent ear infections and the unfolding language problem, and possibly also because of innate cognitive obstacles. And his schooling does not seem to provide much pertinent modeling. In sum, P is missing out on the social aspect of comprehending. It is not surprising that acquisition of related skills should be hampered if the reader is not taught what it is to comprehend and has no opportunity to share his or her comprehension with others. By this reasoning, it would make sense that isolated clinical sessions such as the ones provided by SLP will not suffice. Not because the strategies taught or the methods used are inadequate, but because they do not do justice to the systemic, distributed nature of P‘s issues. As mentioned before, isolated demands to comprehend are prone to simply reinforce his habit of OTHER-ORIENTATION—expectations of independent meaning construction, absent

331

motivators for him to do so, amount to other-induced demands for self-propelled action; this results in a paradoxical message. This conundrum could only be transcended if P‘s entire social and academic environment were to communicate expectations of comprehension. In other words, the main burden of intervention would have to shift from P‘s clinicians to his school and home. The school would need to alter its instructional approach to include clear expectations of comprehending, and provide P with the tools he needs to meet them; namely, instruction in higher-level reading strategies and meaning construction. At the same time, considerable effort would have to be devoted to familiarize him with the pleasure of comprehending (Gopnik, Meltzoff, & Kuhl, 1999), and the power of story (Bruner, 1986). We did not learn much about the current home reading situation during the interviews, but it can be assumed that P‘s mother would have reported if there were a substantial amount of literacy-related interaction. If this assumption is correct, then harnessing his parents‘ influence might prove beneficial. Sharing the satisfaction of reading with immediate family is a powerful motivator (Smith, 2004). If P could learn from teachers, peers and parents what it means to be a reader, his niche of comprehension failure might be replaced by a larger sociolinguistic dynamic which gives his budding abilities the grounds and space to bloom. General clinical implications. As we have investigated only one case of poor comprehension, it is difficult to give clinical recommendations for the population as a whole. Based on our Discussion (Chapter Five), however, we can touch on some general points likely to be beneficial. Consider first the systemic nature of P‘s poor comprehension. If this finding holds for the entire population, then intervention might benefit if clinicians took a broad perspective on

332

each case. This would include assessment not only of clients‘ reading strategies and ability to comprehend, but also the investigation of their social-experiential and psychological background to ferret out any non-linguistic, non-cognitive obstacles to comprehension. Notably, clinicians should be trained to look for infelicitous interactional dynamics which hamper comprehension acquisition, such as heavy emphasis on reading accuracy at school, or lack of literacy-related interaction at home; they should also look for indications of lack of motivation and deficient schemas for reading. Finally, they should remain open to the possibility of undiagnosed but pertinent issues—in P‘s case, a potential cognitive delay which was noticed by SLP only well into their therapeutic relationship. Early detection of such issues would allow for better targeted intervention. Obviously, clinicians cannot be expected to study each reader in as much detail as provided in this investigation. And it is not necessary to do so in order to arrive at an understandable picture. The professional literature abounds with suggestions for authentic/dynamic assessment (e.g. Brinton & Fujiki, 2010; Owocki & Goodman, 2002; Goodman, Watson, & Burke, 2005; Westby, 1990), all of which provide methods and tools similar to those used in this study but in a form amenable to clinical use. However, this presupposes that clinicians have been trained in using them. If they do not feel comfortable or competent enough to do so, they may miss out on the opportunity of better understanding their clients. Ultimately, then, it is the responsibility of educators in the field to provide their students with pertinent tools. Thus, clinical education might benefit from adopting a general systemic perspective, as has been done in branches of psychology (Winek, 2010). If clinicians were taught the systemic aspects of communicative disorders as an integral part of their client‘s functioning,

333

they might be in a better position to understand each individual case they are faced with, as opposed to relying on ‗labels‘ (cf. Damico, Müller, & Ball, 2010). A final suggestion pertains to clinicians‘ training in literacy intervention. Any such training should incorporate integrative perspectives on reading, lest students get caught up in the pervasive assumption that there is a disconnect between decoding and comprehending. As this study has shown, it is relatively easy to teach children lower-level literacy skills: it can be done successfully even in the presence of language and cognitive issues, and such skills can readily be assessed to show measures of progress. However, this does not ensure that children will be readers. Directions for Future Research The present study provides several starting points for further inquiry. Harking back to our collection of themes in Table 4.19, we can divide them into two general areas: socialpsychological characteristics of poor comprehension on the one hand, and linguisticbehavioral ones on the other. In addition, the question why some poor comprehenders do not appear to make progress in intervention should be addressed. Future research: Social-psychological aspects. Most of the themes we found in this area are likely unique to P: comparison with S revealed that even between only two poor comprehenders, social-psychological aspects may vary considerably. The one aspect our participants had in common was the heavy emphasis on lower-level reading in their academic experience. Both went through intensive tutoring in decoding; we do not know much about the instructional focus at their schools but P‘s mother‘s talk about TANGIBILITY lends itself to the suspicion that a similar emphasis was present in this context. One promising area for further research would therefore be the investigation of poor comprehenders‘ experience with

334

literacy instruction. We already know that the population comprises children with both linguistic and cognitive issues, but their academic environment has not been systematically investigated. It is noteworthy that allusions to pedagogical practices in the earlier literature (prior to Yuill & Oakhill, 1991) were not based on empirical evidence; thus, there is dire need for further study before we can arrive at any general conclusions about the role of instruction. Another area that warrants more scrutiny is motivation. As seen, the present study did not arrive at a coherent account of this aspect. We found clear affective differences between S and P: the former was highly frustrated with his predicament, the latter rather impassive. SLP reported that both may have made progress in their attitudes towards literacy, although it remains doubtful that either of them discovered the pleasures of reading and writing—by her account, S simply became less reluctant, and P possibly more comfortable working with (more precisely; for) her. I said above that S‘ motivation for change was higher than P‘s, because he was frustrated with his situation—but being motivated to work for academic success is different from being motivated to read for the sake of reading, and according to SLP, S showed no indications that he started to enjoy transacting with print. Thus, while it seems clear that motivation does play a role in poor comprehenders‘ trajectories, we are far from understanding exactly how it interacts with the other factors involved. Methodologically, studies of instructional experience and motivational factors should be qualitative. For the former, trained field researchers should investigate poor comprehenders‘ academic environment using ethnographic interviews, participant observation, and artifact analysis. Additional data should be collected from parents and other stakeholders to investigate the interaction of instructional experience with issues intrinsic to

335

the young readers or inherent in their home environment. Researchers should expect individual trajectories to be diverse and likely quite different from those outlined in this study; the goal would be to determine if instructional experience is a common denominator that unites this heterogeneous population. Similar formats of inquiry should be used to investigate motivation, except that emphasis should be on interviews with parents and other stakeholders. In addition, such studies could benefit from reading tasks similar to those in the present study, with reading materials geared towards eliciting interest and/or making use of familiarity. If poor comprehenders show different reading behaviors based on whether or not they enjoy texts or are familiar with them, the role that motivation and its absence play in the issue may become clearer. Finally, poor comprehenders‘ schemata for reading should get more attention. As seen in Chapter Two, early studies found clear differences in how this population understands the reading process, as compared to typical readers. This area could not be addressed in this study, as P simply did not convey any useful information in this regard (another aspect in which he differed from other poor comprehenders). Assuming this is an idiosyncratic phenomenon, pertinent studies should employ the interview techniques used in this study and/or those reported in the literature. It seems that most children are able to communicate their understanding of reading adequately enough. It needs to be noted that any research as discussed here should not operate from a causal-componential but from a systemic viewpoint: its guiding assumption would not be that instruction, motivation, or schemata for reading cause poor comprehension, but that they may be a crucial factor in its emergence. Hence, the same lines of research should also

336

investigate interacting factors that contribute, in each individual poor comprehender, to the emergence of the issue. Obviously, these too can be expected to be quite divergent. Again, research so far has shown that some individuals have language and/or cognitive problems while others do not. From a systemic viewpoint, this is explained by the notion of equifinality; the same outcome can be reached from a variety of starting points. We may suspect that home literacy experience, general home situation, personality traits, socioeconomic status and a variety of other issues might play a role in each case‘s trajectory. In-depth case studies like the present one appear to be the prime way to gain insight into the different pathways into poor comprehension. Future research: Linguistic-behavioral aspects. The wealth of observations we made on P‘s reading behaviors allows for more specific recommendations for future studies. Some of those will be based on the themes we devised, others on specific behaviors, as many behaviors that were interpreted as indicators of an underlying theme can be fruitfully studied independently. Quality and type of miscues. As seen, a large proportion of the miscues produced by P (between 50% and 85%) were of low quality, disrupting meaningfulness. It would appear promising to compare this percentage to typical readers. Such a study should follow an experimental design using authentic texts, and include both age-matched and comprehensionage matched controls to determine whether incidence of low-quality miscues is simply an indicator of delayed development, or a factor which differentiates the population. If the latter were found to be the case, it would indicate a systematic absence of attention to meaning in poor comprehending readers. Along the same lines, it should be studied whether the population makes more miscues that are graphically and phonemically highly similar to the

337

expected response. If so, this would indicate a generalized bottom-up approach to print, and provide an explanation for the low quality of miscues. Erroneous prediction of syntactic structures. One of the more intriguing findings of this study was that P sometimes mistakenly predicts upcoming syntactic structures in a way that renders the ensuing production meaningless (e.g. ―There they are!‖ Rex clapped. where the actual sentence continues after the verb). Two lines of investigation appear promising. First, poor comprehenders should be experimentally compared to age- and comprehensionage matched controls to determine whether this behavior is unique to their population, or simply indicative of delayed literacy development. Second, the qualitative aspects of these miscues should be analyzed and compared to similar miscues made by typical readers. Specifically, it should be determined a) whether the erroneous structures are syntactically simpler than the expected response, and b) whether they result in low-quality miscues (such as clapped where only a term of verbal expression would make sense). The former could indicate that poor comprehenders operate from a bottom-up approach, attempting to construct language from simple building blocks. The latter would be illustrative of how meaningfulness is overridden in that process. This line of research should, again, be done using actual texts (or parts of texts). Additionally, eye movement measures could be used to determine whether poor comprehenders look ahead to resolve syntactic ambiguities, and whether they do so in a way different from controls. Piecemeal processing behaviors. Another notable feature of P‘s reading behaviors was his tendency to articulate print bit by bit, resulting in choppy-sounding productions and hesitations, some of which occurred within words. We took these behaviors to be indicative of an excessive focus on print and a failure to predict upcoming language; it would be fruitful

338

to investigate if they are commonly found in poor comprehenders, and if so, whether they occur in significantly larger amounts than in typical readers. The experimental design would be the same as in the studies proposed above. If corroborated, this would be yet another indicator that the population operates from a bottom-up approach to print. Reading flow. Flurkey (1997, 2008) found that reading of connected text proceeds in ebbs and flows, speeding up when the reader is easily able to process the print and slowing down when he or she is struggling with an aspect of the text. In other words, variability of rate, termed ‗reading flow‘, is a hallmark of engagement with text. This is in line with Paulson‘s (2005) observation that proficient reading is characterized by variability in fixation durations. P‘s reading was characterized by a high number of hesitation and ‗choppy‘ productions, but his reading flow was not investigated in the present study. Future research should investigate poor comprehenders‘ flow; it can be hypothesized that reading rate is less variable in this population than it is in good readers, indicating lack of active and engaged processing of text while reading. Lack of integration. The failure to integrate read information to a coherent whole is, of course, the defining feature of the population, as evidenced over decades of research through incoherent retells, failure to detect inconsistencies or resolve pronominal anaphors, etc. But as we saw in P‘s data, this feature can be pinpointed even in small-scale aspects of oral production. Examples included P‘s piecemeal processing of the name of a main character in home reading one (i.e., a bit of language he should have been familiar with), and his struggle to accurately produce the name of a character in home reading two even after this name occurred for the third time. To investigate this phenomenon, texts should be chosen to

339

include such recurrent character names or other complicated target words, and poor comprehenders‘ production should be compared to typical readers as outlined above. Measures of eye movements. P‘s eye movements differed from those of typical readers in multiple ways. The number of fixations he made was higher than expected, and typical of much younger readers. By contrast, the average duration of his fixations was shorter than typical, and he had a lower percentage of regressions; in both regards, his measurements were adult-like. In addition, he used his eye movements inefficiently, searching for visual information where there was none, and failing to scan upcoming information even within words he struggled on. In conjunction with the other reading behaviors we observed, we interpreted these findings to indicate a preoccupation with visual sampling and an absence of linguistic strategies of tackling print. Also, P showed less variability of fixation durations per paragraph than typical readers, indicating he processed print in a linear fashion, which has been theorized to suggest that the reader is not reading for meaning (Paulson, 2005). All of the above lends itself to experimental investigation as outlined previously (i.e. comparison of the target population with typical readers matched for age and comprehension age). A cautionary remark is in order. If our interpretation of P‘s eye movements, that he uses them to sample as much visual information as possible, is correct, we may not find the same patterns in other poor comprehenders. Visual sampling can take other forms than the SEARCH

patterns we observed. For example, it is conceivable that a reader may stare at each

word for a long time to decode it. In such a case, the numerical values for average fixation duration and fixations per 100 words would be quite different.

340

The other two measures, by contrast, can be hypothesized to be similar across poor comprehenders. Percentage of regressions is associated with strategic lookbacks, which are used, among other purposes, to confirm or disconfirm constructed meanings (e.g. Paulson, 2000, 2002). Where no meaning construction happens, the proportion of regressions is expectably low. A similar rationale applies to variability in duration. As Paulson (2005) explains, shorter fixations are associated with aspects of text that are easy for the reader to process; longer durations occur when the reader tries to comprehend difficult stretches of print. Since most texts comprise both easy and difficult aspects for any given reader, variability of fixation durations is expected to be high in active reading for meaning, but low in linear, code-based reading. Our assumption is that the latter form is typical of poor comprehenders; if this is correct, we should see similar lack of variability in durations across the population, regardless of whether they use visual search behaviors, staring behaviors, or some other form of linear visual sampling. Finally, it would be beneficial to do qualitative analyses of eye movements on any of the studies outlined above—in other words, to analyze the data using EMMA—to determine how poor comprehenders handle difficult stretches of text. It can be assumed that they would show similar behaviors to P (i.e., that they would neither look back as much as typical readers, nor look ahead to aid predictive strategies). Again, it should be noted that none of the behaviors discussed in this section could be taken to be the causes of poor comprehension. Rather, they would have to be seen as aspects of it. Future research: Lack of response to intervention. A final area of study should take as its starting point the surprising observation that not all poor comprehenders benefit

341

from intervention. This is at odds with the literature both from the earlier and the later ‗eras‘ of research, and our study has not yielded a satisfactory account of this phenomenon. Future research should therefore strive to achieve such an account. As we have no concrete pointers as to which factors in poor comprehension could be operationalized experimentally, such studies should ideally be qualitative: broad-based, comprehensive investigations just like the present one but with the specific goal of finding factors that could help understand nonresponsiveness to intervention. To that end, such studies should scrutinize both responsive and non-responsive poor comprehenders to determine differences between them. Limitations The present study does not only suffer from the typical limitations endemic to any scientific endeavor. It was also fraught with difficulties pertaining to data collection; in addition, due to its nature as an in-depth case study with the goal of examining the usefulness of an alternative descriptive and theoretical framework, its very scope is different from typical investigations, and needs to be delineated carefully. Note first that the present study comprises only one case (plus some additional data from a comparison case), and this case is unusual. In contrast to more typical poor comprehenders, our participant did not respond well to intervention. In addition, there were technical problems during part of the data collection, which introduced a degree of uncertainty regarding location of fixations during eye movement measures. As a result of all these factors, the present study comes with a substantial degree of uncertainty regarding the applicability of its findings. It is conceivable that there may be undetected factors in P‘s trajectory that make him so different from more typical poor comprehenders that our findings would not be applicable to much of the population. A less

342

substantial uncertainty pertains to the technical difficulties: it is possible that some of our findings regarding eye movements might not hold if we used a different technology. For these reasons, the results of this study need to be viewed as preliminary. We cannot claim substantiality for them until they can be shown to hold across a number of poor comprehenders. Importantly, we can emphatically not claim to present a generalized picture of the population. What this study does show is much more limited in scope, and of a more abstract nature. Consider the factors of P‘s poor comprehension (social-experiential, psychological, and linguistic-behavioral). It cannot be expected that the very same factors—the specific behaviors, numerical values, psychosocial dynamics and so on—will be present in other poor comprehenders. What can be expected is that there will be factors on the social, psychological, and behavioral level that interact to contribute to the emergence of poor comprehension. We cannot know their exact nature, or the nature of their interaction, but we can assume their presence. Again, the present study does not offer an account that would hold across the population. It does offer conceptual (and methodological) tools that can be used to describe other cases, and possibly eventually yield a generalizable picture. In this regard, it needs to be pointed out that our grounded theory itself—the description of poor comprehension as an emergent niche—is to be seen as such a conceptual tool, not as a definitive account of the phenomenon. It is useful as a starting point for further investigation, but it is completely conceivable that it may have to be rewritten in the light of new data. Finally, limitations pertinent to our methodological approach must be addressed. First, it is obvious our conclusions are interpretive statements, and are not typically considered experimentally substantiated hypotheses. In other words, readers who equate

343

scientific research with controlled attempts at falsification will be disappointed. This is not a limitation per se: qualitative findings are simply not based on the same assumptions as experimental ones. But it is worth pointing out this difference to preclude objections to our findings based on expectations of experimentalist standards. More importantly, qualitative research itself suffers from limitations as regards its capability to capture its phenomena of interest; also, these limitations are compounded in our study. Geertz (1973) points out that qualitative descriptions have a peculiar status with regard to the ‗reality‘ they purportedly convey: they are interpretations, not one-to-one renderings. In other words, they are fictions: ―in the sense that they are ‗something made,‘ ‗something fashioned‘—the original meaning of fictiō—not that they are false, unfactual, or merely (…) thought experiments‖ (p. 15). This means that their capacity to help us understand the phenomena we investigate is limited. In ethnography—Geertz‘s area of study—this limitation is counterbalanced by the prescription that interpretations be ‗actor-oriented‘. To Geertz, this means that interpretations of cultures ―must be cast in terms of the constructions we imagine [members of those cultures] to place upon what they live through, the formulae they use to define what happens to them‖ (p. 15). Note the double ‗removal from reality‘, as it were. Ethnographic descriptions are interpretations of interpretations: scientific accounts of participant meanings, or even scientific ‗imaginations‘ of the latter. Thus, the study of people‘s lifeworlds—the classic object of inquiry in qualitative research—is a re-constructive, creative exercise (and much debate in the field revolves precisely around the question in how far it is possible or even desirable to attempt such re-construction—cf. e.g. Denzin & Lincoln, 2005). Thus, our study

344

suffers from the same vulnerability as all qualitative research: it is, ―to borrow W. B. Gallie‘s by now famous phrase, ‗essentially contestable‘‖ (Geertz, 1973, p. 29). To this general limitation, our study adds the problem of expert knowledge (cf. Chapter Three). The phenomena we are interested in are different from those studied in classic ethnography: we assume that there is more to them than is accessible to participants. In other words, we contend that we are construing a plausible account, not only of participants‘ lifeworlds, but also of even deeper layers of their experience hidden from their own view (as in our interpretations of P‘s reading behaviors, or the role of his OTHERORIENTATION in

the development of his poor comprehension). Our justification is our claim to

expert knowledge on reading; this claim makes us suspect in most quarters of the qualitative research community where the ethics and politics of talking about the ‗other‘ are treated as matters of great import, and highly controversial ones at that (Denzin & Lincoln, 2005). Thus, due to its dual nature as an account of participant perspective combined with expert knowledge, our study is even more contestable than ‗typical‘ qualitative works: we rely, not only on interpretations of participant interpretations—which could be fine-tuned through lamination (Agar, 1986)—but on interpretations ‗brought into‘ participants‘ worlds from our own, preconceived theoretical perspective. Our fictiō—the story crafted from our observations—rests on theoretical and methodological decisions made before we even encountered P. However plausible our interpretations may seem from this perspective, there is no way of determining if P would agree to them; likewise, taking an expert perspective prevents us from ‗digging deeper‘ into the subject matter, from going beyond the angle of view which made the inquiry possible in the first place. This compounds the problem that the ‗real reality‘ of our phenomenon of interest, the ‗truth‘ of the interplay of lived experience

345

and cognitive-linguistic strategies, remains as elusive as the ethnographer‘s attempt at a ‗complete‘ description of the symbolic world she is describing. The conundrum we are facing is not the same as the one captured in the classic anecdote about Indian cosmology, but can be illustrated with the same story: An Englishman who, having been told that the world rested on (…) the back of an elephant which rested in turn on the back of a turtle, asked (…) what did the turtle rest on? Another turtle. And that turtle? ―Ah, Sahib, after that it is turtles all the way down.‖ Geertz (1973, pp. 28-29) The moral is that no matter how deep one‘s interpretation ‗digs‘ into its subject matter, it will never get to where the researcher can claim to have exhausted it. Our problem is somewhat different. The reality we are exploring may well consist of a bedazzling variety of turtles all the way down. Yet our inquiry is limited due to its very starting point: we contend ourselves with exploring those turtles we can reach from our spot atop the elephant. Concluding Remarks: Ontological Non-Commitments and a Poetic Pragmatist Stance As we near the end of the study, it is worth harking back to the epistemological discussion in Chapter Three. Recall that I opted for a pragmatist stance instead of a realist one, despite the fact that the latter is more commonly used in qualitative studies of reading. I justified this with a better fit of pragmatism to Grounded Theory; by now, it has become clear that our findings, too, are better accommodated by this framework than they would be by realism. Recall that the latter posits a threefold model of reality comprised of observations, events, and structures/processes. Our empirical findings—our observations— are substantial, yet they apply only to one case; our interpretations are therefore somewhat

346

bold (some might say ‗heady‘). If I were to defend them from a realist viewpoint, I would have to commit to their ontological reality: the interacting dynamics of our themes, the emergence of P‘s poor comprehension, and its distributed nature would all have to be viewed as ‗events‘ instantiated by the ‗real‘ evolutionary dynamics of reader-environment systems. This seems to be too stark a claim, given that it is based on one case only, and fraught with the limitations discussed above. From a pragmatist viewpoint, these limitations are not an issue. Recall that the main concern of this epistemological stance is to find what ‗works‘. The purpose of this study was to try out an alternative description of poor reading comprehension, and our findings show that this description worked: it provided us with an understandable account of the case we studied. This is not to say we have explained the phenomenon, only that we may have found a set of tools for describing it, and possibly a better one than those used in the past. This set includes a kit of methodologies, a theoretical framework, and a preliminary interpretation (our twofold grounded theory of poor comprehension as an emergent ‗niche‘). Note that we cannot be sure that all of these tools will be useful for all cases we encounter: New data may surprise us in this regard. Consider also that ultimately, all scientific endeavors are based on ‗linguistic habits‘ (von Bertalanffy, 1968). This includes our own. Descriptive accounts, particularly those steeped in the language of non-linearity, systems, etc., are seductive ways of talking which can easily take on facile forms (Bricmont & Sokal, 2001; Sokal, 1996). And verstehen is incomplete unless researchers leave the realm of science and add an element of poetry to their descriptions (Berliner, 2002). It is therefore judicious to refrain from assertions about

347

the ontological reality of our interpretations: the best we can claim to have done is to describe our data with as much interpretive rigor as possible, and with the amount of poetry necessary. Darwin (2003/1859) concludes his Origin of Species exclaiming, ―There is grandeur in this view of life‖ (p. 507). Our ‗niche‘ metaphor cannot exactly claim grandeur, but it offers an accessible way of understanding (verstehen) the phenomenon, framing it in terms of the natural world as poetic descriptions of the human condition are wont to do. And even if our interpretations turned out to be too bold, and our poetry to be inappropriate, our empirical findings would still have value. As Geertz (1973) reminds us, the ultimate task of qualitative research is not to find definitive answers, but ―to make available to us answers that others (…) have given, and thus to include them in the consultable record of what man has said‖ (p. 30). At the very minimum, this is what the present study has achieved: to make available what we have been given, contributing to the consultable record of how poor comprehenders live and what they do.

348

Bibliography Adler, P. A., & Adler, P. (1994). Observational techniques. In N. K. Denzin & Y. S. Lincoln (Eds.), Handbook of qualitative research (pp. 377-392). Thousand Oaks, CA: Sage. Adlof, S. (2010). Morphosyntactic skills of poor comprehenders. Unpublished doctoral dissertation, University of Kansas. Retrieved January 12, 2012 from ERIC database via UL Lafayette online library catalog. Agar, M. (1986). Speaking of ethnography. Newbury Park, CA: Sage. Agar, M. (1996). The professional stranger. New York: Academic Press. Allen, G. (2000). Intertextuality. New York: Routledge. American Psychiatric Association (2013a). Intellectual disability [fact sheet]. Retrieved May 7, 2014 from http://www.dsm5.org/Documents/Intellectual%20Disability%20Fact%20Sheet.pdf. American Psychiatric Association (2013b). Specific learning disorder [fact sheet]. Retrieved May 7, 2014 from http://www.dsm5.org/Documents/Specific%20Learning%20Disorder%20Fact%20Sheet.pdf. Applied Science Laboratories (2008). Operation manual MobileEye. Bedford, MA: Applied Science Laboratories. Applied Science Laboratories (2009). Eye tracker systems manual ASL Eye-Trac 6: D6 Desk Mounted Optics. Bedford, MA: Applied Science Laboratories. Applied Science Laboratories (2010a). Eye tracker systems tutorial: ASL Results file analysis tool. Bedford, MA: Applied Science Laboratories. Applied Science Laboratories (2010b). Eye tracker systems manual: ASL Results file analysis tool. Bedford, MA: Applied Science Laboratories.

Astington, J.W. (1994). Children‘s developing notions of others‘ minds. In J. Duchan, L. Hewitt, & R. Sonnenmeier (Eds.), Pragmatics: From theory to practice. (pp. 72-87). Englewood Cliffs, NJ: Prentice-Hall. Atkinson, P., & Hammersley, M. (1994). Ethnography and participant observation. In N. K. Denzin & Y. S. Lincoln (Eds.), Handbook of qualitative research (pp. 209-219). Thousand Oaks, CA: Sage. Avi (1997). Something upstairs. Logan, IA: Perfection Learning Corporation. Baker, A. (2010). Simplicity. In E. N. Zalta (Ed.), The Stanford Encyclopedia of Philosophy (Summer 2011 Edition). Retrieved January 22, 2013 from http://plato.stanford.edu/entries/simplicity/. Beck, I. L. (1981). Reading problems and instructional practices. In G. E. Mackinnon & T. G. Waller (Eds.), Reading research. Advances in theory and practice (pp. 53-95). New York: Academic Press. Berliner, D. C. (2002). Comment: Educational research: The hardest science of all. Educational Researcher, 31, 18-20. Bilheimer, M. I. (1992). A comparison of the macro-, micro-, and superstructural levels of expository and narrative text comprehension of seventh-grade good and different poor comprehenders. Unpublished doctoral dissertation, Lehigh University. Bishop, D (1983). Test for reception of grammar. Manchester, UK: University of Manchester Department of Psychology. Bogen, J. (2013). Theory and observation in science. In E. N. Zalta (Ed.), The Stanford Encyclopedia of Philosophy (Summer 2014 Edition). Retrieved May 7, 2014 from http://plato.stanford.edu/cgi-bin/encyclopedia/archinfo.cgi?entry=science-theory-observation.

350

Booth, J. R., MacWhinney, B., and Harasaki, Y. (2000). Developmental differences in visual and auditory processing of complex sentences. Child Development, 71, 981-1003. Borella, E., Carretti, B., & Pelegrina, S. (2010). The specific role of inhibition in reading comprehension in good and poor comprehenders. Journal of Learning Disabilities, 43, 541552. Bowyer-Crane, C., & Snowling, M. J. (2005). Assessing children‘s inference generation: What do tests of reading comprehension measure? British Journal of Educational Psychology, 75, 189-201. Bransen, J. (2001). Verstehen and erklären, philosophy of. In N. J. Smelser & P. B. Baltes (Eds.), International encyclopedia of the social and behavioral sciences (pp. 16165-16170). Oxford: Elsevier. Retrieved December 12, 2008 from http://www.ru.nl/publish/pages/536984/verstehen.pdf Bricmont, J., & Sokal A. D. (2001). Defense of modest scientific realism. Retrieved May 28, 2014 from http://www.physics.nyu.edu/faculty/sokal/bielefeld_final_rev.pdf. [Presented at the Bielefeld-ZiF conference on ―Welt und Wissen – Monde et Savoir – World and Knolwedge, June 18, 2001, Bielefeld, Germany.] Brigandt, I., & Love, A. (2008). Reductionism in Biology. In E. N. Zalta (Ed.), The Stanford encyclopedia of Philosophy (Fall 2008 Edition). Retrieved January 30, 2012 from http://plato.stanford.edu/archives/fall2008/entries/reduction-biology/ Brinton, B., & Fujiki, M. (2010). Principles of assessment and intervention. In J. S. Damico, N. Müller, & M. J. Ball (Eds.), The handbook of speech and language disorders (pp. 131150). Malden, MA: Wiley-Blackwell. Brown, J. (1964). Flat Stanley. New York: Harper & Row.

351

Brown, J., Kim, K., & O‘Brien Ramirez, K. (2012). What a teacher hears, what a reader sees: Eye movements from a phonics taught second grader. Journal of Early Childhood Literacy, 12, 202-222. Brown, A. L., & Palincsar, A. S. (1982). Inducing strategic learning from texts by means of informed, self-control training. Topics in Learning & Learning Disabilities, 2, 1-17. Bruner, J. (1986). Actual minds, possible worlds. Cambridge, MA: Harvard University Press. Bryant, A., & Charmaz, K. (2007a). Introduction—Grounded Theory research: Methods and practices. In A. Bryant & K. Charmaz (Eds.), The SAGE handbook of Grounded Theory (pp. 1-28). Thousand Oaks, CA: Sage Publications. Bryant, A., & Charmaz, K. (2007b). Grounded theory in historical perspective: An epistemological account. In A. Bryant & K. Charmaz (Eds.), The SAGE handbook of Grounded Theory (pp. 31-57). Thousand Oaks, CA: Sage Publications. Cain, K. (1996). Story knowledge and comprehension skill. In C. Cornoldi & J. Oakhill (Eds.), Reading comprehension difficulties: Processes and intervention (pp. 167-192). Mahwah, NJ: Lawrence Erlbaum Associates. Cain, K. (2006). Individual differences in children‘s memory and reading comprehension: An investigation of semantic and inhibitory deficits. Memory, 14, 553-569. Cain, K., & Oakhill, J. V. (1998). Comprehension skill and inference-making ability: Issues of causality. In C. Hulme & R. M. Joshi (Eds.), Reading and spelling. Development and disorders (pp. 329-342). Mahwah, NJ: Lawrence Erlbaum Associates. Cain, K., & Oakhill, J. V. (1999). Inference making ability and its relation to comprehension failure in young children. Reading and Writing: An Interdisciplinary Journal, 11, 489-503.

352

Cain, K., & Oakhill, J. V. (2006). Profiles of children with specific reading comprehension difficulties. British Journal of Educational Psychology, 76, 683-696. Cain, K., & Oakhill, J. V. (2007). Reading comprehension difficulties. Correlates, causes, and consequences. In K. Cain & J. V. Oakhill (Eds.), Children‘s comprehension problems in oral and written language. A cognitive perspective (pp. 41-75). New York: Guilford Press. Cain, K., & Oakhill, J. V. (2011). Matthew effects in young readers : Reading comprehension and reading experience aid vocabulary development. Journal of Learning Disabilities, 44, 431-443. Cain, K., Oakhill, J. V., Barnes, M. A., & Bryant, P. E. (2001). Comprehension skill, inference-making ability, and their relation to knowledge. Memory & Cognition, 29, 850859. Cain, K., Oakhill, J., & Lemmon, K. (2005). The relation between children‘s reading comprehension level and their comprehension of idioms. Journal of Experimental Child Psychology, 90, 65-87. Cain, K., Patson, N., & Andrews, L. (2005). Age- and ability-related differences in young readers‘ use of conjunctions. Journal of Child Language, 32, 877-892. Calfee, R. C. (2009). Foreword. In S. E. Israel & G. G. Duffy (Eds.), Handbook of research in reading comprehension (pp. xi-xv). New York: Routledge. Canney, G., & Winograd, P. (1979). Schemata for reading and reading comprehension performance. Technical Report No. 120. Washington, D.C.: National Institute of Education. Retrieved January 13, 2012 from ERIC database via UL Lafayette online library catalog.

353

Catts, H. W., Adlof, S. M., & Weismer, S. E. (2006). Language deficits in poor comprehenders: A case for the Simple View of reading. Journal of Speech, Language, and Hearing Research, 49, 278-293. Catts, H. W., Hogan, T. P., & Adlof, S. M. (2005). Developmental changes in reading and reading disabilities. In H. W. Catts & A. G. Kamhi (Eds.), The connection between language and reading disabilities (pp. 25-40). Mahwah, NJ: Lawrence Erlbaum. Cazden, C. (1988). Classroom discourse: The language of teaching and learning. Portsmouth, NH: Heinemann. Chakravartty, A. (2013). Scientific Realism. In E. N. Zalta (Ed.), The Stanford Encyclopedia of Philosophy (Summer 2013 Edition). Retrieved November 6, 2013, from http://plato.stanford.edu/archives/sum2013/entries/scientific-realism/ Chall, J. S. (1983). Stages of reading development. St. Louis: McGraw-Hill. Charmaz, K. (2003). Grounded theory. In J. A. Smith (Ed.), Qualitative psychology (pp. 81110). Thousand Oaks, CA: Sage Publications. Charmaz, K. (2005). Grounded theory in the 21st century: Applications for advancing social justice studies. In N. K. Denzin & Y. S. Lincoln (Eds.), The SAGE handbook of qualitative research (3rd ed., pp. 507-535). Thousand Oaks, CA: Sage. Charmaz, K. (2006). Constructing grounded theory. A practical guide through qualitative analysis. Thousand Oaks, CA: Sage Publications. Cilliers, P. (1998). Complexity & postmodernism. Understanding complex systems. New York: Routledge. Clay, M. M. (2001). Change over time in children‘s literacy development. Portsmouth, NH: Heinemann.

354

Commoner, B. (1972). The closing circle. Nature, man & technology. New York: Alfred A. Knopf. Coots, J. H., & Snow, D. P. (1980). Understanding poor reading comprehension: Current approaches in theory and research. Information Analyses (070), Southwest Regional Laboratory for Educational Research and Development. Corbin, J., & Strauss, A. L. (2008). Basics of qualitative research (3rd ed.). Thousand Oaks, CA: Sage. Cornoldi, C., Carretti, B., & De Beni, R. (2001). How the patterns of deficits in groups of learning-disabled individuals help to understand the organization of working memory. Issues in Education, 7, 71-79. Cornoldi, C., De Beni, R., & Pazzaglia, F. (1996). Profiles of reading comprehension difficulties: An analysis of single cases. In C. Cornoldi & J. Oakhill (Eds.), Reading comprehension difficulties: Processes and intervention (pp. 113-136). Mahwah, NJ: Lawrence Erlbaum Associates. Cragg, L., & Nation, K. (2006). Exploring written narrative in children with poor reading comprehension. Educational Psychology, 26, 55–72. Crago, M. B. (1990). Development of communicative competence in Inuit children: Implications for speech-language pathology. Journal of Childhood Communication Disorders, 13, 73-84. Crespi, B. J. (1999). The evolution of maladaptation. Heredity, 84, 623-629. Cromer, W. (1970). The difference model: A new explanation for some reading difficulties. Journal of Educational Psychology, 61, 471-488.

355

Cronbach, L. J. (1975). Beyond the two disciplines of scientific psychology. American Psychologist, 30, 671-684. Damico, J. S., & Augustine, L. E. (1995). Social considerations in the labeling of students as attention deficit hyperactivity disordered. Seminars in Speech and Language, 16, 259-274. Damico, J. S., & Ball, M. J. (2010). Prolegomenon: Addressing the tyranny of old ideas. Journal of Interactional Research in Communication Disorders, 1.1., 1-29. Damico, J. S., Müller, N., & Ball, M. J. (2010). Social and practical considerations in labeling. In J. S. Damico, N. Müller, & M. J. Ball (Eds.), The handbook of speech and language disorders (pp. 11-37). Malden, MA: Wiley-Blackwell. Damico, J. S., & Nelson, R. L. (2010). Reading and reading impairments. In J. S. Damico, N. Müller, & M. J. Ball (Eds.), The handbook of speech and language disorders (pp. 267-295). Malden, MA: Wiley-Blackwell. Damico, J. S., Nelson, R. L., & Bryan, L. (2005). Literacy as sociolinguistic process for clinical purposes. In M. J. Ball (Ed.), Clinical sociolinguistics (pp. 242-249). Malden, MA: Wiley-Blackwell. Damico, J. S., Nelson, R. L., Damico, H., Abendroth, K., & Scott, J. (2008). Avoidance strategies in an exceptional child during unsuccessful reading performances. Clinical Linguistics & Phonetics, 22, 283-291. Damico, J. S., & Simmons-Mackie, N. N. (2003). Qualitative research and speech-language pathology: A tutorial for the clinical realm. American Journal of Speech-Language Pathology, 12, 131-143. Darwin, C. (2003/1859). The origin of species. 150th anniversary edition. New York: Penguin Signet Classics.

356

De Beni, R., Palladino, P., Pazzaglia, F., & Cornoldi, C. (1998). Increases in intrusion errors and working memory deficit of poor comprehenders. The Quarterly Journal of Experimental Psychology, 51A, 305-320. De Beni, R., & Palladino, P. (2000). Intrusion errors in working memory tasks: Are they related to reading comprehension ability? Learning & Individual Differences, 12, 131-144. Denzin, N. K., & Lincoln, Y. S. (2005). Introduction: The discipline and practice of qualitative research. In N. K. Denzin & Y. S. Lincoln (Eds.), Handbook of qualitative research (3rd ed., pp. 1-32). Thousand Oaks, CA: Sage. de Sousa, I., & Oakhill, J. (1996). Do levels of interest have an effect on children‘s comprehension monitoring performance? British Journal of Educational Psychology, 66, 471-482. de Saussure, F. (2001/1916). Grundfragen der allgemeinen Sprachwissenschaft [Course in general linguistics]. Berlin: Walter de Gruyter. Disney Wiki (n. d.). Jessie. Retrieved December 30, 2013, from http://disney.wikia.com/wiki/Jessie. Duckett, P. d. B. (2001). First-grade beginning readers‘ use of pictures and print as they read: A miscue analysis and eye movement study. Unpublished doctoral dissertation, The University of Arizona. Duckett, P. d. B. (2003). Envisioning story: The eye movements of beginning readers. Literacy Teaching and Learning, 7, 77-89. Elliott, C. D., Smith, P., & McCulloch, K. (1996). British Ability Scales Second Edition (BAS-II). Windsor, UK: National Foundation for Educational Research.

357

Farnill, D., & Hayes, S. C. (1995). Retest reliability of the Woodcock Language Proficiency Battery. Perceptual and Motor Skills, 81, 1147-1152. Feilzer, M. Y. (2010). Doing mixed methods research pragmatically: Implications for the rediscovery of pragmatism as a research paradigm. Journal of Mixed Methods Research, 4, 616. Fleisher, B. M. (1990). Self-corrections of poor comprehenders. Journal of Reading, Writing, and Learning Disabilities, 6, 105-115. Floyd, R. G., Bergeron, R., Alfonso, V. C. (2006). Cattell-Horn-Carroll cognitive ability profiles of poor comprehenders. Reading and Writing: An Interdisciplinary Journal, 19, 427456. Flurkey, A. D. (1997). Reading as flow: A linguistic alternative to fluency. Unpublished doctoral dissertation, University of Arizona. Flurkey, A. D. (2008). Reading flow. In A. D. Flurkey, E. J. Paulson, & K. S. Goodman (Eds.), Scientific realism in the studies of reading (pp. 267-304). New York: Lawrence Erlbaum Associates. Fosnot, C. T., & Perry, R. S. (2005). Constructivism: A psychological theory of learning. In C. T. Fosnot (Ed.), Constructivism. Theory, perspectives, and practice (2nd ed., pp. 8-38). New York: Teachers College Press. Fowler, C. A., & Hodges, B. H. (2011). Dynamics and languaging: Toward an ecology of language. Ecological Psychology, 23, 147-156. Freeman, A. E. (2001). The eyes have it: Oral miscue and eye movement analyses of the reading of fourth grade Spanish/English bilinguals. Unpublished doctoral dissertation, University of Arizona.

358

Fujiki, M., & Brinton, B. (2004). Pragmatics. In R. D. Kent (Ed.), The MIT encyclopedia of communication disorders (pp. 372-374). Cambridge, MA: The MIT Press. Gagné, E. D. (1981). Comprehension and the long-term recall of information. Final report. Washington, D.C.: National Institute of Education. Retrieved January 12, 2012 from ERIC database via UL Lafayette online library catalog. Garner, R. (1981). Monitoring of passage inconsistency among poor comprehenders: A preliminary test of the ―piecemeal processing‖ explanation. Journal of Educational Research, 74, 159-162. Garner, R., & Kraus, C. (1981). Good and poor comprehender differences in knowing and regulating reading behaviors. Educational Research Quarterly, 6, 5-12. Garner, R., & Reis, R. (1981). Monitoring and resolving comprehension obstacles: An investigation of spontaneous text lookbacks among upper-grade good and poor comprehenders. Reading Research Quarterly, 16, 569-582. Garner, R. (1982). Resolving comprehension failure through text lookbacks: Direct training and practice effects among good and poor comprehenders in grades six and seven. Reading Psychology, 3, 221-231. Garner, R., Wagoner, S., & Smith, T. (1983). Externalizing question-answering strategies of good and poor comprehenders. Reading Research Quarterly, 18, 439-447. Gavelek, J., & Bresnahan, P. (2009). Ways of meaning making: Sociocultural perspectives on reading comprehension. In S. E. Israel & G. G. Duffy (Eds.), Handbook of research in reading comprehension (pp. 140-176). New York: Routledge. Geekie, P., Cambourne, B., & Fitzsimmons, P. (1999). Understanding Literacy Development. Staffordshire, UK: Trentham Books.

359

Geertz, C. (1973). Thick description: Toward an interpretive theory of culture. In C. Geertz (Ed.), The interpretation of cultures: Selected essays (pp. 3-30). New York: Basic Books. Gerstl-Pepin, C. I., & Woodside-Jiron, H. (2005). Tensions between the ―science‖ of reading and a ―love of learning‖: One high-poverty school‘s struggle with NCLB. Equity & Excellence in Education, 38, 232–241. Geva, E., & Massey-Garrison, A. (2012). A comparison of the language skills of ELLs and monolinguals who are poor decoders, poor comprehenders, or normal readers. Journal of Learning Disabilities, 46, 387-401. Ginn, F., & Demeritt, D. (2009). Nature: A contested concept. Retrieved May 19, 2014 from http://www.uk.sagepub.com/upm-data/24131_17_Hollway_Ch_17.pdf. [Published in N. Clifford, S. Holloway, S. P. Rice, & G. Valentine (Eds.), Key concepts in geography (2nd ed., pp. 300-311). Thousand Oaks, CA: Sage Publications. Gladstone, C. J. (1991). The nature of good and poor comprehenders‘ ability to construct meaning during reading. Unpublished doctoral dissertation, Northwestern University. Glaser, B. G., & Strauss, A. L. (1967). The discovery of Grounded Theory: Strategies for qualitative research. New York: Aldine de Gruyter. Glaser, B. G. (2007). Doing formal theory. In A. Bryant & K. Charmaz (Eds.), The SAGE handbook of Grounded Theory (pp. 97-113). Thousand Oaks, CA: Sage Publications. Glaser, B. G. (2008). Doing quantitative Grounded Theory. Mill Valley, CA: Sociology Press. Goldkuhl, G., & Cronholm, S. (2010). Adding theoretical grounding to grounded theory: Toward Multi-Grounded Theory. International Journal of Qualitative Methods, 9, 187-205.

360

Golinkoff, R. M. (1975). A comparison of reading comprehension processes in good and poor comprehenders. Washington, D.C.: National Institute of Education. Retrieved January 12, 2012 from ERIC database via UL Lafayette online library catalog. Goodman, K. S. (1967). Reading: a psycholinguistic guessing game. Retrieved May 30, 2013 from http://readbysight.com/images/reading_a_psycholinguistic_guessing_game.pdf. Originally published in Journal of the Reading Specialist, 6, 126–135. Goodman, K. S., & Burke, C. L. (1968). Study of children‘s behavior while reading orally. Retrieved May 30, 2013 from http://files.eric.ed.gov/fulltext/ED021698.pdf Goodman, K. S. (1994). Reading, writing, and written texts: A transactional sociopsycholinguistic view. In R. B. Ruddell, M. R. Ruddell, & H. Singer (Eds.), Theoretical models and processes of reading (4th ed., pp. 1093–1130). Newark, DE: International Reading Association. Goodman, K. S. (2008). Miscue analysis as scientific realism. In A. D. Flurkey, E. J. Paulson, & K. S. Goodman (Eds.), Scientific realism in the studies of reading (pp. 7-21). New York: Lawrence Erlbaum Associates. Goodman, Y. M., & Goodman, K. S. (1994). To err is human: Learning about language processes by analyzing miscues. In R. B. Ruddell, M. R. Ruddell, & H. Singer (Eds.), Theoretical models and processes of reading (pp. 101-123) Newark, DE: International Reading Association. Goodman, Y. M., Watson, D. J., & Burke, C. L. (2005). Reading miscue inventory. From evaluation to instruction (2nd ed.). Katonah, NY: Richard C. Owens Publishers. Goodwin, C. (1995). Co-constructing meaning in conversations with an aphasic man. Research on Language and Social Interaction, 28, 233-260.

361

Gopnik, A., Meltzoff, A. N., & Kuhl, P. K. (1999). The scientist in the crib. What early learning tells us about the mind. New York City: Harper. Gough, P. B., Hoover, W. A., & Peterson, C. L. (1996). Some observations on a simple view of reading. In C. Cornoldi & J. Oakhill (Eds.), Reading comprehension difficulties: Processes and intervention (pp. 1-13). Mahwah, NJ: Lawrence Erlbaum Associates. Grice, H. P. (1975) Logic and conversation. In P. Cole & J. L. Morgan (Eds.), Syntax and Semantics. Vol. 3: Speech Acts (pp. 41-58). New York, NY: Academic Press. Gruen, J. R., & Nelson, N. W. (2011). Advances in Research: Diagnosis and Treatment of Dyslexia. Presented at the Specialty Recognition Board Katharine G. Butler Symposium in Child Language, 2011 ASHA Convention, San Diego, CA. Gutman, D. (2007). Getting air. New York: Simon & Schuster. Hahn, A. L., & Smith, T. (1983). Students‘ differentiation of reader-based and text-based questions. Journal of Educational Research, 76, 331-334. Halliday, M. A. K., & Matthiessen, C. M. I. M. (2004). An introduction to Functional Grammar (3rd ed.). London: Arnold. Hardcastle, V.G. (2003). The development of self. In G.D. Fireman, T.E. McVay, Jr., & O.J. Flanagan, (Eds.), Narrative and consciousness. Literature, psychology, and the brain. (37-50). Oxford: Oxford University Press. Hare, V. C., & Borchardt, K. M. (1985). Good and poor comprehenders‘ detection of errors revisited. Journal of Educational Research, 78, 237-241. Hess, A. M. (1982). An analysis of the cognitive processes underlying problems in reading comprehension. Journal of Reading Behavior, XIV, 313-333.

362

Hoffman, J. V. (2009). In search of the ―Simple View‖ of reading comprehension. In S. E. Israel & G. G. Duffy (Eds.), Handbook of research in reading comprehension (pp. 54-66). New York: Routledge. Hookway, C. (2010). Pragmatism. In E. N. Zalta (Ed.), The Stanford encyclopedia of Philosophy (Spring 2010 Edition). Retrieved June 2, 2013 from http://plato.stanford.edu/archives/spr2010/entries/pragmatism/ Hoover, W. A., & Gough, P. B. (1990). The simple view of reading. Reading and Writing: An Interdisciplinary Journal, 2, 127-160. House, E. R. (1991) Realism in research. Educational Researcher, 20, No. 6, 2-9+25. Howe, K. R. (1992). Getting over the quantitative-qualitative debate. American Journal of Education, 100, 236-256. Howe, K. R. (2004). A critique of experimentalism. Qualitative Inquiry, 10, 42-61. Hruby, G. G. (2009). Grounding reading comprehension in the neuroscience literatures. In S. E. Israel & G. G. Duffy (Eds.), Handbook of research in reading comprehension (pp. 189223). New York: Routledge. Huey, E. B. (1968/1908). The psychology and pedagogy of reading. Cambridge, MA: MIT Press. Idol, L. (1987). Group story mapping: A comprehension strategy for both skilled and unskilled readers. Journal of Learning Disabilities, 20, 196-205. Idol-Maestas, L. (1985). Getting ready to read: Guided probing for poor comprehenders. Learning Disability Quarterly, 8, 243-254. Louisiana Department of Education (2007). The integrated Louisiana educational assessment program. Retrieved October 17, 2013 from http://www.doe.state.la.us/lde/uploads/11494.pdf

363

Itzkoff, S. W. (1996). Children learning to read. A guide for parents and teachers. Westport, CT: Praeger. Janesick, V. L. (1994) The dance of qualitative research design: Metaphor, methodolatry, and meaning. In N. K. Denzin & Y. S. Lincoln (Eds.), Handbook of qualitative research (pp. 209-219). Thousand Oaks, CA: Sage. Johnson-Glenberg. M. C. (2005). Web-based training of metacognitive strategies for text comprehension: Focus on poor comprehenders. Reading and Writing, 18, 755-786. Johnson-Laird, P. N. (2004). The history of mental models. In K. Manktelow (Ed.), Psychology of reasoning: Theoretical and historical perspectives (pp. 179-212). London: Psychology Press. Retrieved October 8, 2012 from http://mentalmodels.princeton.edu/papers/2005HistoryMentalModels.pdf Jones, J. (2010). Toy story 3. New York: Random House/Disney. Kaschak, M. P., Jones, J. L., Coyle, J. M., & Sell, A. (2009). Language and body. In R. K. Wagner, C. Schatschneider, & C. Pythian-Sence (Eds.), Beyond decoding. The behavioral and biological foundations of reading comprehension (pp. 3-26). New York: Guilford Press. Keenan, J. M., & Betjemann, R. S. (2006). Comprehending the Gray Oral Reading Test without reading it: Why comprehension tests should not include passage-independent items. Scientific Studies of Reading, 10, 363-380. Kemp, M. (1987). Watching children read & write. Melbourne: Nelson Australia. Keller, H. R., & Sabers. D. L. (2003). Test review of the Gray Silent Reading Tests. In B. S. Plake & J. C. Impara (Eds.), The fifteenth mental measurements yearbook in print. Lincoln, NE: Buros Center for Testing.

364

Kelso, K., Fletcher, J., & Lee, P. (2007). Reading comprehension in children with specific language impairment: An examination of two subgroups. International Journal of Language and Communication Disorders, 42, 39-57. Kendall, J. R., & Hood, J. (1979). Investigating the relationship between comprehension and word recognition: Oral reading analysis of children with comprehension or word recognition disabilities. Journal of Reading Behavior, XI, 41-48. Kim, K., Knox, M., & Brown, J. (2007). Eye movement and strategic reading. In Y. Goodman & P. Martens (Eds.), Critical issues in early literacy. Research and pedagogy (pp. 47-57). Mahwah, NJ: Lawrence Erlbaum Associates. Kinney, J. (2007-present). Diary of a wimpy kid. New York: Scholastic. Kintsch, W., & Rawson, K. A. (2005). Comprehension. In M. J. Snowling & C. Hulme (Eds.), The science of reading. A handbook (pp. 209-226). Malden, MA: Blackwell Publishing. Kletzien, S. B. (1991). Strategy use by good and poor comprehenders reading expository text of differing levels. Reading Research Quarterly, 26, 67-86. Kovarsky, D., & Crago, M. (1990-1991). Towards the ethnography of communication disorders. National Student Speech Language Hearing Association Journal, 18, 44-55. Kuhn, T. S. (2012/1962). The structure of scientific revolutions: 50th anniversary edition. Chicago: The University of Chicago Press. Lakoff, G., & Johnson, M. (2003/1983). Metaphors we live by. Chicago: The University of Chicago Press. Levin, J. R. (1971). Comprehending what we read: An outsider looks in. Journal of Reading Behavior, 4, 18-28.

365

Lewis, M. D., & Granic, I. (2000). Emotion, development, and self-organization. Dynamic systems approaches to emotional development. Cambridge, UK: Cambridge University Press. Long, D. L., & Chong, J. L. (2001). Comprehension skill and global coherence: A paradoxical picture of poor comprehenders‘ abilities. Journal of Experimental Psychology: Learning, Memory, and Cognition, 27, 1424-1429. Lysynchuk, L., Pressley, M., & Vye, N. J. (1989). Reciprocal instruction improves standardized reading comprehension performance in poor grade-school comprehenders. Paper presented at the Annual Meeting of the American Educational Research Association, March 27-31, 1989. Retrieved January 12, 2012 from ERIC database via UL Lafayette online library catalog. Marshall, C. M., & Nation, K. (2003). Individual differences in semantic and structural errors in children‘s memory for sentences. Educational and Child Psychology, 20, 7-18. Maturana, H. R., & Varela, F. J. (1992). The tree of knowledge. The biological roots of human understanding. Revised edition. Boston: Shambhala. Maxwell, J. A. (2004). Causal explanation, qualitative research, and scientific inquiry in education. Educational Researcher, 33, 3-11. McDermott, R. P. (1987). The explanation of minority school failure, again. Anthropology & Education Quarterly, 18, 361-364. McDermott, R. P., & Roth, D. R. (1978). The social organization of behavior: Interactional approaches. Annual Review of Anthropology, 7, 321-345. McDermott, R., & Varenne, H. (1995). Culture as disability. Anthropology and Education Quarterly, 26, 323-348.

366

Meadows, D. H. (2008). Thinking in systems. A primer. White River Junction, VT: Chelsea Green Publishing. Miller, G. A. (1956). The magical number seven, plus or minus two: Some limits on our capacity for processing information. The Psychological Review, 63, 81-97. Miller, J. W., & Isakson, R. L. (1976). The effect of syntactic and semantic violation on high and low reading comprehenders. Paper presented at the annual meeting of the American Educational Research Association, April 19-23, 1976. Retrieved January 12, 2012 from ERIC database via UL Lafayette online library catalog. Morgan, D. L. (2007). Paradigms lost and pragmatism regained: Methodological implications of combining qualitative and quantitative methods. Journal of Mixed Methods Research, 1, 48-76. Morse, J. M. (2007). Sampling in grounded theory. In A. Bryant & K. Charmaz (Eds.), The SAGE handbook of grounded theory (pp. 229-245). Thousand Oaks, CA: Sage Publications. Nation, K. (2005). Children‘s reading comprehension difficulties. In M. J. Snowling & C. Hulme (Eds.), The science of reading. A handbook (pp. 248-265). Malden, MA: Blackwell Publishing. Nation, K. (2009). Reading comprehension and vocabulary: What‘s the connection? In R. Wagner, C. Schatschneider, & C. Phythian-Sence (Eds.), Beyond decoding. The behavioral and biological foundations of reading comprehension (pp. 176-194). New York: Guilford Press. Nation, K.. Adams, J. W., Bowyer-Crane, C. A., & Snowling, M. J. (1999). Working memory deficits in poor comprehenders reflect underlying language impairments. Journal of Experimental Child Psychology, 73, 139-158.

367

Nation, K., Clarke, P., Marshall, C. M., & Durand, M. (2004). Hidden language impairments in children: Parallels between poor reading comprehension and specific language impairment? Journal of Speech, Language, and Hearing Research, 47, 199-211. Nation, K., Clarke, P., & Snowling, M. J. (2002). General cognitive ability in children with reading comprehension difficulties. British Journal of Educational Psychology, 72, 549–560. Nation, K., Cocksey, J., Taylor, J. S. H., & Bishop, D. V. M. (2010). A longitudinal investigation of early reading and language skills in children with poor reading comprehension. Journal of Child Psychology and Psychiatry, 51, 1031-1039. Nation, K., Marshall, C. M., & Snowling, M. J. (2001). Phonological and semantic contributions to children‘s picture naming skill: Evidence from children with developmental reading disorders. Language and Cognitive Processes, 16 (2/3), 241-259. Nation, K., & Snowling, M. J. (1998). Semantic processing and the development of wordrecognition skills: evidence from children with reading comprehension difficulties. Journal of Memory and Language, 39, 85-101. Nation, K., & Snowling, M. J. (1999). Developmental differences in sensitivity to semantic relations among good and poor comprehenders: evidence from semantic priming. Cognition, 70, B1-B13. Nation, K., Snowling, M. J., &Clarke, P. (2007). Dissecting the relationship between language skills and learning to read: Semantic and phonological contributions to new vocabulary learning in children with poor reading comprehension. Advances in Speech– Language Pathology, 9, 131-139. Neale, M. D. (1989). The Neale Analysis of Reading Ability-Revised (NARA-II). Windsor, UK: NFER-Nelson.

368

Nelson, N. (2010). Language and literacy disorders: Infancy through adolescence. Boston: Allyn & Bacon. Nelson, R. L., & Damico, J.S. (2006). Qualitative research in literacy acquisition: A framework for investigating reading in children with language impairment. Clinical Linguistics and Phonetics, 20, 631-639. Nelson, R. L., Damico, J. & Smith, S., (2008). Applying eye movement miscue analysis to the reading patterns of children with language impairment. Clinical Linguistics and Phonetics, 22, 293-303. Nesse, R. M. (2005). Maladaptation and natural selection. The Quarterly Review of Biology, 80, 62-70. Noë, A. (2009). Out of our heads. Why you are not your brain, and other lessons from the biology of consciousness. New York: Hill and Wang. Noerager Stern, P. (2007). On solid ground: Essential properties for growing grounded theory. In A. Bryant & K. Charmaz (Eds.), The SAGE handbook of grounded theory (pp. 114-126). Thousand Oaks, CA: Sage Publications. Oakhill, J. (1993). Children‘s difficulties in reading comprehension. Educational Psychology Review, 5, 223-237. Oakhill, J., & Cain, K. (2007). Introduction to comprehension development. In K. Cain & J. Oakhill (Eds.), Children‘s comprehension problems in oral and written language. A cognitive perspective (pp. 3-40). New York: Guilford Press. Oakhill, J., & Garnham, A. (1988). Becoming a skilled reader. New York: Blackwell.

369

Oakhill, J., Hartt, J., & Samols, D. (2005). Levels of comprehension monitoring and working memory in good and poor comprehenders. Reading and Writing: An Interdisciplinary Journal, 18, 657-686. Oakhill, J., & Patel, S. (1991). Can imagery training help children who have comprehension problems? Journal of Research in Reading, 14, 106-115. Oakhill, J., & Yuill, N. (1996). Higher order factors in comprehension disability: Processes and remediation. In C. Cornoldi & J. Oakhill (Eds.), Reading comprehension difficulties: Processes and intervention (pp. 69-92). Mahwah, NJ: Lawrence Erlbaum Associates. Odling-Smee, J., & Laland, K. N. (2009). Cultural niche construction: evolution‘s cradle of language. In R. Botha & C. Knight (Eds.), The prehistory of language (pp. 99-121). New York: Oxford University Press. Owocki, G., & Goodman, Y. M. (2002). Kidwatching: documenting children‘s literacy development. Portsmouth, NH: Heinemann. Palincsar, A. S., & Brown, A. L. (1984). Reciprocal teaching of comprehension-fostering and comprehension-monitoring activities. Cognition and Instruction, 1, 117-175. Palladino, P., Cornoldi, C., De Beni, R., & Pazzaglia, F. (2001). Working memory and updating processes in reading comprehension. Memory & Cognition, 29, 344-354. Paul, D. B. (2009). Darwin, social Darwinism, and eugenics. In J. Hodge & G. Radick (Eds.), The Cambridge companion to Darwin (2nd ed., pp. 219-245). New York: Cambridge University Press. Paulson, E. J. (2000). Adult readers‘ eye movements during the production of oral miscues. Unpublished doctoral dissertation, University of Arizona.

370

Paulson, E. J. (2002). Are oral reading word omissions and substitutions caused by careless eye movements? Reading Psychology, 23, 45-66. Paulson, E. J. (2005). Viewing eye movements during reading through the lens of chaos theory: How reading is like the weather. Reading Research Quarterly, 40, 338-358. Paulson, E. J., Flurkey, A. D., Goodman, Y. M., & Goodman, K. S. (2002). Eye movements and miscue analysis: Reading from a constructivist perspective. 52nd Yearbook of the National Reading Conference (pp. 345-355). Retrieved February 10th, 2012 from http://ericpaulson.org/yahoo_site_admin/assets/docs/Paulson_NRCYBK.7141052.pdf Paulson, E. J., & Freeman, A. E. (2003). Insight from the eyes. The science of effective reading instruction. Portsmouth, NH: Heinemann. Paulson, E. J., & Goodman, K. S. (2008). Re-reading eye-movement research: Support for transactional modes of reading. In A. D. Flurkey, E. J. Paulson, & K. S. Goodman (Eds.), Scientific realism in the studies of reading (pp. 25-47). New York: Lawrence Erlbaum Associates. Pearson, P. D. (2009). The roots of reading comprehension instruction. In S. E. Israel & G. G. Duffy (Eds.), Handbook of research in reading comprehension (pp. 3-31). New York: Routledge. Peirce, C. S. (1878). How to make our ideas clear. Popular Science Monthly, 12, 286-302. [Retrieved May 27, 2013 from http://www.peirce.org/writings/p119.html] Perfetti, C. A., Marron, M., A., & Foltz, P. W. (1996). Sources of comprehension failure: Theoretical perspectives and case studies. In C. Cornoldi & J. Oakhill (Eds.), Reading comprehension difficulties: Processes and intervention (pp. 137-165). Mahwah, NJ: Lawrence Erlbaum Associates.

371

Perkins M. R. (2005). Pragmatic ability and disability as emergent phenomena. Clinical Linguistics and Phonetics, 19, 367-377. Persson, U.-B. (1994). Reading for understanding: An empirical contribution to the metacognition of reading comprehension. Unpublished doctoral dissertation, Linköping University (Sweden). Retrieved January 12, 2012 from ERIC database via UL Lafayette online library catalog. Piaget, J. (1970). Structuralism. New York: Harper & Row. Pimperton H., & Nation, K. (2010). Suppressing irrelevant information from working memory: evidence for domain-specific deficits in poor comprehenders. Journal of Memory & Language, 62, 380-391. Pimperton H., & Nation, K. (2012). Poor comprehenders in the classroom: Teacher ratings of behavior in children with poor reading comprehension and its relationship with individual differences in working memory. Journal of Learning Disabilities. Advance online publication. doi: 10.1177/0022219412454172. Quine, W. V. O. (1951). Two dogmas of empiricism. The Philosophical Review, 60, 20-43. Quine, W. V. O. (1960). Words and objects. Cambridge, MA: MIT. Rasinski, T. V. (1989). Fluency for everyone: Incorporating fluency instruction in the classroom. The Reading Teacher, 42, 690-693. Rayner, K. (1998). Eye movements in reading and information processing: 20 years of research. Psychological Bulletin 134, 372-422. Rayner, K. (2009). The 35th Sir Frederick Bartlett lecture: Eye movements and attention in reading, scene perception, and visual search. The Quarterly Journal of Experimental Psychology, 62, 1457-1506.

372

Rayner, K., & Juhasz, B. (2006). Reading processes in adults. In R. E. Asher & M. M. Y. Simpson (Eds.), Encyclopedia of Language and Linguistics, vol. 10 (pp. 373-378). Amsterdam: Elsevier. Rayner, K., & Slattery, T. J. (2009). Eye movements and moment-to-moment comprehension processes in reading. In R. K. Wagner, C. Schatschneider, & C. Pythian-Sence (Eds.), Beyond decoding. The behavioral and biological foundations of reading comprehension (pp. 27-45). New York: Guilford Press. Renaissance Learning, Inc. (2012). AR BookFinder. http://www.arbookfind.com/default.aspx. Rickards, J. P., & Hatcher, C. W. (1977-1978). Interspersed meaningful learning questions as semantic cues for poor comprehenders. Reading Research Quarterly, 13, 538-553. Ricketts, J. (n. d.). Reading: assessment and research. Retrieved May 18, 2014 from http://www.yarcsupport.co.uk/documents/reading-research.pdf. Ricketts, J., Bishop, D. V. M., & Nation, K. (2008). Investigating orthographic and semantic aspects of word learning in poor comprehenders. Journal of Research in Reading, 31, 117135. Ritter, S., & Idol-Maestas, L. (1986). Teaching middle-school students to use a test-taking strategy. Journal of Educational Research, 79, 350-357. Robertson, S. S., Cohen, A. H., & Mayer-Kress, G. (1993). Behavioral chaos: beyond the metaphor. In L. B. Smith & E. Thelen (Eds.), A dynamic systems approach to development: Applications (pp. 119-150). Cambridge, MA: MIT Press.

373

Rosenblatt, L. M. (1994). The transactional theory of reading and writing. In R. B. Ruddell, M. R. Ruddell, & H. Singer (Eds.), Theoretical models and processes of reading (4th ed., pp. 1057–1092). Newark, DE: International Reading Association. Ruddell, R. B., & Unrau, N. J. (1994). Reading as a meaning-construction process: The reader, the text, and the teacher. In R. B. Ruddell, M. R. Ruddell, & H. Singer (Eds.), Theoretical models and processes of reading (4th ed., pp. 996-1056). Newark, DE: International Reading Association. Rusch, G. (1994). Kommunikation und Verstehen [Communication and understanding]. In K. Merten, S. J. Schmidt, & S. Weischenberg (Eds.), Die Wirklichkeit der Medien [Media reality] (pp. 60-78). Opladen, Germany: Westdeutscher Verlag. Schmidt, S. J., & Weischenberg, S. (1994). Mediengattungen, Berichterstattungsmuster, Darstellungsformen [Media genres, patterns of reporting, and modes of presentation]. In K. Merten, S.J. Schmidt, S. Weischenberg (Eds.), Die Wirklichkeit der Medien. Eine Einführung in die Kommunikationswissenschaft [Media reality. An introduction to communication science] (pp.212-236). Opladen, Germany: Westdeutscher Verlag. Seeberger, S., Nelson, R., & Curtis, N. (2009). Investigating Dandy-Walker syndrome: Integrating Conversation Analysis and reading eye-movements. Asia Pacific Journal of Speech, Language, and Hearing, 12, 235-242. Share, D. L., & Leikin, M. (2004). Language impairment at school entry and later reading disability: Connections at the lexical versus supralexical levels of reading. Scientific Studies of Reading, 8, 87-110.

374

Simmons-Mackie, N. N., & Damico, J. S. (2003). Contributions of qualitative research to the knowledge base of normal communication. American Journal of Speech-Language Pathology 12: 144-154. Skeat, J., & Perry, A. (2008). Grounded theory as a method for research in speech and language therapy. International Journal of Language and Communication Disorders, 43, 95109. Smith, F. (1998). The book of learning and forgetting. New York: Teachers College Press. Smith, F. (2004). Understanding Reading. A psycholinguistic analysis of reading and learning to read (6th ed.). Mahwah, NJ: Lawrence Erlbaum Associates. Smith, J. A., & Osborn, M. (2003). Interpretative phenomenological analysis. In J. A. Smith (Ed.), Qualitative psychology (pp. 51-80). Thousand Oaks, CA: Sage Publications. Smith, L. B., & Thelen, E. (Eds.) (1993a). A dynamic systems approach to development: Applications. Cambridge, MA: MIT Press. Smith, L. B., & Thelen, E. (1993b). Can dynamic systems theory be usefully applied in areas other than motor development? In L. B. Smith & E. Thelen (Eds.), A dynamic systems approach to development: Applications (pp. 151-170). Cambridge, MA: MIT Press. Smith, B., & Varzi, A. C. (1999). The niche. Nous, 33, 198-222. Snow, C. E., & Juel, C. (2005). Teaching children to read: What do we know about how to do it? In M. J. Snowling & C. Hulme (Eds.), The science of reading. A handbook (pp. 501520). Malden, MA: Blackwell Publishing. Snowling, M. J., McLean, J., & Stothard, S. E. (1996). Graded Nonword Reading Test. London: Thames Valley Test Company.

375

Snowling, M. J., & Hulme, C. (2012). Annual Research Review: The nature and classification of reading disorders—a commentary on proposals for DSM-5. Journal of Child Psychology and Psychiatry, 53, 593-607. Snyder, L. S. (1976). The early presuppositions and performatives of normal and language disabled children. Papers and Reports on Child Language Development, 12. Abstract retrieved October 29, 2013 from http://eric.ed.gov/?id=ED161301 Sokal, A. D. (1996). Transgressing the boundaries: Towards a transformative hermeneutics of quantum gravity. Retrieved May 28, 2014 from http://www.physics.nyu.edu/faculty/sokal/transgress_v2/transgress_v2_singlefile.html#405. [Published in Social Text 46/47, 217-252.] Spiro, R. J., Vispoel, W. P., Schmitz, J. G., Samarapungavan, A., & Boerger, A. E. (1987). Knowledge acquisition for application: Cognitive flexibility and transfer in complex content domains. University of Illinois at Urbana-Champaign: Technical Report No. 409. Retrieved August 16, 2012 from https://www.ideals.illinois.edu/bitstream/handle/2142/17527/ctrstreadtechrepv01987i00409_ opt.pdf?sequence=1 Stanovich, K. E., West, R. F., Cunningham, A. E., Cipielewski, J., & Siddiqui, S. (1996). The role of inadequate print exposure as a determinant of reading comprehension problems. In C. Cornoldi & J. Oakhill (Eds.), Reading comprehension difficulties: Processes and intervention (pp. 15-32). Mahwah, NJ: Lawrence Erlbaum Associates. Stothard, S. E, & Hulme, C. (1995). A comparison of phonological skills in children with reading comprehension difficulties and children with decoding difficulties. Journal of Child Psychology & Psychiatry & Allied Disciplines, 36, 399-408.

376

Strauss, A. L., & Corbin, J. (1994). Grounded theory methodology: An overview. In N. K. Denzin & Y. S. Lincoln (Eds.), Handbook of qualitative research (pp. 273-285). Thousand Oaks, CA: Sage Publications. Swanson, H. L., Howard, C. B., & Sáez, L. (2007). Reading comprehension and working memory in children with learning disabilities in reading. In K. Cain & J. Oakhill (Eds.), Children‘s comprehension problems in oral and written language. A cognitive perspective (pp. 157-189). New York: Guilford Press. Tarlow, M.-C. (1990). Differences between good and poor comprehenders in comprehension processing of expository text. Unpublished doctoral dissertation, University of California, Berkely. Retrieved January 11, 2012 from ProQuest database via UL Lafayette online library catalog. Thelen, E., & Smith, L. B. (1994). A dynamic systems approach to the development of cognition and action. Cambridge, MA: MIT Press. Thibault, P. J. (2011). First-order languaging dynamics and second-order language: The distributed language view. Ecological Psychology, 23, 210-245. Tong, X., Deacon, S. H., Kirby, J. R., Cain, K., & Parrila, R. (2011). Morphological awareness: A key to understanding poor reading comprehension in English. Journal of Educational Psychology, 103, 523`-534. Torppa, M., Tolvanen, A., Poikkeus, A.-M., Eklund, K., Lerkkanen, M.-K., Leskinen, E., & Lyytinen, H. (2007). Reading development subtypes and their early characteristics. Annals of Dyslexia, 57, 3-32.

377

Tye, M. (2009). Qualia. In E. N. Zalta (Ed.), The Stanford Encyclopedia of Philosophy (Summer 2009 Edition). Retrieved October 1, 2012 from http://plato.stanford.edu/archives/sum2009/entries/qualia/ van der Schoot, M., Vasbinder, A. L., Horsley, T. M., & van Lieshout, E. C. D. M. (2008). The role of two reading strategies in text comprehension: An eye fixation study in primary school children. Journal of Research in Reading, 31, 203-223. van der Schoot, M., Vasbinder, A. L., Horsley, T. M., Reijntjes, A., & van Lieshout, E. C. D. M. (2009). Lexical ambiguity resolution in good and poor comprehenders: An eye fixation and self-paced reading study in primary school children. Journal of Educational Psychology, 101, 21-36. van der Schoot, M., Horsley, T. M., & van Lieshout, E. C. D. M. (2010). The effects of instruction on situation model construction: An eye fixation study on text comprehension in primary school children. Educational Psychology, 30, 817-835. von Glasersfeld, E. (1985). Konstruktion der Wirklichkeit und des Begriffs der Objektivität [Construction of reality and of the idea of objectivity]. In H. Gumin & H. Meier (Eds.), Einführung in den Konstruktivismus [Introduction to constructivism] (pp. 9-39). Munich, Germany: Piper. Van Gulick, Robert (2011). Consciousness. In E. N. Zalta (Ed.), The Stanford Encyclopedia of Philosophy (Summer 2011 Edition). Retrieved October 1, 2012 from http://plato.stanford.edu/archives/sum2011/entries/consciousness/ von Bertalanffy, L. (1968). General system theory. New York: George Brazillier. Vygotsky, L.S. (1978), Mind in society. The development of higher psychological processes. Cambridge, MA: Harvard University Press.

378

Wade, S. (1990). Using think alouds to assess comprehension. The Reading Teacher, 43, 442-451. Wagner, R. K., Schatschneider, C., & Pythian-Sence, C. (2009a). Preface. In R. K. Wagner, C. Schatschneider, & C. Pythian-Sence (Eds.), Beyond decoding. The behavioral and biological foundations of reading comprehension (pp. xi-xii). New York: Guilford Press. Wagner, R. K., Schatschneider, C., & Pythian-Sence, C. (Eds.) (2009b). Beyond decoding. The behavioral and biological foundations of reading comprehension. New York: Guilford Press. Watzlawick, P., Beavin Bavelas, J., & Jackson, D. D. (2011). Pragmatics of human communication. A study of interactional patterns, pathologies, and paradoxes. New York: W. W. Norton & Company. (Original work published 1967.) Weaver. C. (1994). Parallels between new paradigms in science and in reading and literacy theories: An essay review. In R. B. Ruddell, M. R. Ruddell, & H. Singer (Eds.), Theoretical models and processes of reading (4th ed., pp. 1185-1203). Newark, DE: International Reading Association. Wechsler, D. (1990). Wechsler Objective Reading Dimensions. London: Psychological Press. Weekes, B. S., Hamilton, S., Oakhill, J. V., & Holliday, R. E. (2008). False recollection in children with reading comprehension difficulties. Cognition, 106, 222-233. Welsh, A. M., & Hobson, R. (2011). Heroes of the Titanic. New York: Scholastic. Westby, C. (1990). Ethnographic interviewing: Asking the right questions to the right people in the right ways. Communication Disorders Quarterly, 13, 101-111.

379

Westby, C., & Watson, S. (2010). ADHD and communication disorders. In J. S. Damico, N. Müller, & M. J. Ball (Eds.), The handbook of speech and language disorders (pp. 529-555). Malden, MA: Wiley-Blackwell. Whiting, S. (2006). All about ants. Washington, DC: National Geographic. Wiederholt, J. L., & Blalock, G. (2000). Gray Silent Reading Tests. Austin, TX: Pro-Ed. Wiederholt, J. L., & Bryant, B. R. (2001). Gray Oral Reading Tests—Fourth Edition. Austin, TX: Pro-Ed. Williams, R. (1980). Ideas of nature. Retrieved May 19, 2014 from http://www.religionandnature.com/bron/courses/pdf/Williams--NatureIdeas.pdf. [Published in R. Williams, Problems in materialism and culture (pp. 67-85). New York: Verso.] Winek, J. L. (2010). Systemic family therapy. From theory to practice. Thousand Oaks, CA: Sage Publications. Wittgenstein, L. (1984/1953). Philosophische Untersuchungen [Philosophical investigations]. In L. Wittgenstein, Tractatus logico-philosophicus, Tagebücher 1914-1916, Philosophische Untersuchungen. Werkausgabe Band I [Logical-philosophical treatise, Diaries 1914-1916, Philosophical investigations. The complete works Vol. I] (pp. 225-618). Frankfurt/M., Germany: Suhrkamp. Woodcock, R. W. (1991). Woodcock Language Proficiency Battery – Revised. Itasca, IL: Riverside Publishing. Yuill, N. (1996). A funny thing happened on the way to the classroom: Jokes, riddles, and metalinguistic awareness in understanding and improving poor comprehension in children. In C. Cornoldi & J. Oakhill (Eds.), Reading comprehension difficulties. Processes and intervention (pp. 193-220). Mahwah, NJ: Lawrence Erlbaum Associates.

380

Yuill, N. (2007). Visiting Joke City: How can talking about jokes foster metalinguistic awareness in poor comprehenders? In D. S. McNamara (Ed.), Reading comprehension strategies: Theories, intervention, and technologies (pp. 325-345). Mahwah, NJ: Lawrence Erlbaum Associates. Yuill, N. (2009). The relation between ambiguity understanding and metalinguistic discussion of joking riddles in good and poor comprehenders: Potential for intervention and possible processes of change. First Language, 29, 65-79. Yuill, N., & Joscelyne, T. (1988). Effect of organizational cues and strategies on good and poor comprehenders‘ story understanding. Journal of Educational Psychology, 80, 152-158. Yuill, N., & Oakhill, J. (1988a). Understanding of anaphoric relations in skilled and less skilled comprehenders. British Journal of Psychology, 79, 173-186. Yuill, N., & Oakhill, J. (1988b). Effects of inference awareness training on poor reading comprehension. Applied Cognitive Psychology, 2, 33-45. Yuill, N., & Oakhill, J. (1991). Children‘s problems in text comprehension. An experimental investigation. Cambridge, UK: Cambridge University Press. Yuill, N., Oakhill, J, & Parkin, A. (1989). Working memory, comprehension ability and the resolution of text anomaly. British Journal of Psychology, 80, 351-361.

381

Kroll, Tobias A. Magister Artium, University of Münster, Spring 2007; Doctor of Philosophy, University of Louisiana at Lafayette, Fall 2014 Major: Applied Language and Speech Sciences Title of Dissertation: Reading behaviors, psychological characteristics, and experiential background of one poor comprehender: A Grounded Theory case study Dissertation Director: Dr. Jack S. Damico Pages in Dissertation: 398; Words in Abstract: 335 ABSTRACT This study investigates the emergence of poor reading comprehension in one nineyear old, using the qualitative method of Grounded Theory. Poor reading comprehension has been studied with experimental methods for over 20 years; no causal explanation for the phenomenon has been found. The goal of this study is to examine an alternative methodology, and to take first steps towards an interpretive account of the issue. Inquiry follows a transactional model of reading: social and experiential background of readers; psychological factors such as attitude, motivation, and understanding of reading; and reading behaviors are viewed as necessary aspects of the reading process. Data includes interviews with the poor comprehender, his mother, and his speech-language pathologist; school documentation; linguistic productions while reading and eye movement data; and documentation of reading-related behaviors. Analysis follows the coding procedures of Grounded Theory: data is scrutinized line by line and incident by incident, and emerging codes are chunked into larger themes, which are then combined to an interpretive account. Data converge to yield a coherent picture of a child who, due to early delays in speech and language development, developed a passive personality geared towards fulfilling others‘ expectations. In conjunction with a school environment that places prime importance on oral reading accuracy, but less so on comprehension, it is plausible why the participant has not learned to view reading as an activity done to gain meaning from text. His reading behaviors are consonant with these findings: both eye movements and oral reading data show

a strong orientation towards accuracy of oral production, not construction of meaning. In addition, his reading performance, including his comprehension, varies between contexts. Findings are interpreted within the framework of systems theory, and couched in an ecological metaphor: it is argued that the participant has found his literacy ‗niche‘, i.e. a good-enough fit between his abilities and environmental expectations, and that his poor comprehension is an emergent property of that ‗niche‘. Implications for intervention as well as theoretical conclusions are discussed; directions for future research are given.

383

BIOGRAPHICAL SKETCH Tobias A. Kroll was born December 24, 1972, in Munich, Germany, son to Karl Ernst Alf and Ursula Kroll, and older brother to Daniel J. Kroll. He grew up in Düsseldorf, Germany, and attended the University of Münster, Germany, where he earned a M.A. (Magister Artium) degree in Linguistics. He is married to Anuradha Chandrasekaran and works as an Assistant Professor in Speech-Language & Hearing Sciences at the Texas Tech University Health Sciences Center, Lubbock, TX. Tobias is a member of Phi Kappa Phi and was named Outstanding Doctoral Student in 2009-2010.