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

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 cJCIRC/Datoouopss-p. 15

INHIBITORY LANGUAGE DEFICITS 1N ATTTENTION
DEFICIT/HYPERACTIVITY DISORDER AND READING DISORDER: A
CANDIDATE SHARED DEFICIT

By

Lisa G. Blaskey

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Psychology
2004

ABSTRACT
INHIBITORY LANGUAGE DEFICITS IN
ATTENTION DEFICIT/HYPERACTIVITY DISORDER AND READING DISORDER:
- A CANDIDATE SHARED DEFICIT
By

Lisa G. Blaskey

Few studies have examined whether inhibition impairments in Attention
Deficit/Hyperactivity Disorder (ADHD)—well-established in the motor output domain—
extend to higher-order cognition such as language processing. This study explored
whether inhibitory mechanisms that protect language comprehension are impaired in
ADHD, and whether such inhibitory deficits, if they exist, clarify the etiology of the high
rate of co-occurrence between ADHD and Reading Disorder (RD). Two language
comprehension tasks from the cognitive literature believed to probe inhibitory
mechanisms were used: 1) a sentence-level lexical ambiguity meaningful judgment task
that measured interference control, and 2) a syntactic garden path revision task which
examined ability to use semantic cues to revise misinterpretations “on-line” during
sentence comprehension.

97 children aged 7 to 14 with ADHD or RD participated in the study after
thorough diagnostic assessment with structured diagnostic interview and parent and
teacher ratings. Both (a) higher levels of inattention and (b) ADHD diagnosis were
. associated with poorer inhibitory control of irrelevant information during lexical
ambiguity resolution (e.g., 5:26, p<.05), as well as with decreased ability to revise or

suppress misinterpretations after syntactic garden paths (e.g., _r=.32, p<.01). Girls with

ADHD were more impaired than boys (interaction: p<.01). For children with RD, lower-
order language problems (e.g., word decoding, understanding complex syntax) affected
higher—order inhibitory processes during language comprehension, yielding findings
similar to those found in children with ADHD. Taken together, the pattern of results
suggested that ADHD and RD do share deficits in inhibitory mechanisms that protect
higher—order language comprehension, but that these may develop via different pathways.
In ADHD, these impairments may develop directly from core cognitive inhibition
deficits, whereas in RD they may arise secondarily to language automatization failures
and subsequent requirements for more effortful processing of language (which leaves
fewer resources for inhibition in higher-level language tasks). This was the first study to
use these tasks to assess inhibitory mechanisms in language comprehension in ADHD.
Identification of these difficulties may help clarify the role of language in the self-
regulatory and executive functioning problems that characterize the disorder.
Implications for understanding the etiology of the high co-occurrence of ADHD and RD

are discussed.

ACKNOWLEDGMENTS

This dissertation could never have happened without the steadfast support of
countless people. To my advisor, Dr. Joel Nigg, thank-you for your support and guidance
throughout this process and for staying the course with me on this long road. The scope
and depth of your knowledge and your high standards for intellectual and research
excellence have continually inspired me. I have learned a tremendous amount from you
throughout my graduate school experience and know that I will carry this with me in all
my future endeavors.

I would also like to thank my committee members, Drs. Fernanda F erreira, Jeffrey
Marler, and Norman Abeles, for their comments and suggestions. Your knowledge and
expertise were incredible assets and truly enhanced the quality of this thesis. Also, to the
children and families who participated in this Study and the undergraduate research
assistants who helped with data collection, thanks for your dedication, interest, and
willingness to participate; without people like you this research would never have been
possible.

To my family and fi‘iends who supported me through this process: your
encouragement and support mean the world to me. Thank-you for helping me meet my
goals, for standing by me through trying times, and for being as tolerant as possible about
my long work hours and lack of availability. A Special thanks to my dissertation support
group, Sally Theran, Cheryl-Lynn Podolski, and Anat Barlev who provided countless
' hours of support, constructive suggestions and assistance, and lots of “cheer-leading”

along the way.

iv

Finally, and most importantly, to Matt. I could never have done this without you. I
feel blessed to have someone as caring and supportive as you in my life, someone who
inspires me intellectually, knows the answers to all my “why do you suppose...”
questions, with whom I can always laugh, and for being someone I continue to love more

and more each day. Now, on to a real life!

TABLE OF CONTENTS

 

 

 

 

 

 

 

LIST OF TABLES -- -viii
LIST OF FIGURES xi
Inhibitory Language Deficits in Attention Deficit/Hyperactivity Disorder and
Reading Disorder: A Candidate Shared Deficit - - -_ - - 1
Background on ADHD and RD ....................................................................................... 3
ADHD+RD Comorbidity Findings ................................................................................. 7
Empirical Evidence for Shared Deficits ....................................................................... 16
Inhibition: New Candidatefor a Shared Deficzt7 .............. 28
Inhibition Mechanisms and Language Processing ....................................................... 40
Conclusion .................................................................................................................... 50
METHODS - - -- 52
i Participants ................................................................................................................... 52
Procedure ...................................................................................................................... 56
Measures ....................................................................................................................... 5 7
Experimental Hypotheses and Predictions ................................................................... 70
RESULTS-Part I - - 72
Sample Description ....................................................................................................... 72
Ambiguous Words Task ................................................................................................ 78
RESULTS-Part II _- - - -- -- -- 108
Garden Path Task ....................................................................................................... 1 08
Dimensional Analyses of Inhibitory Eflects on Garden Path, Plausibility, and ......... 115
Verb-Eflect Difference Scores in Relation to Reading Ability and Attention Problem
Scores .......................................................................................................................... I 15
DISCUSSION - -- - - -- ..... -- 134
Interference Control .................................................................................................... 13 7
Revision of Misinterpretations .................................................................................... I 48
Integration and Proposed Theory for the C o-Occurrence of ADHD and RD ............ 158
Limitations .................................................................................................................. 165

vi

Conclusions ................................................................................................................. 1 70

APPENDIX A - - - - _ - -- - - - - - - -- - - 171

 

 

 

 

Demographics and Group Descriptions ..................................................................... 171
Ambiguous Words Task ............................................................................................... 178
APPENDIX B ----- - 202
Preliminary Analyses of Facilitation Ejfects across Groups ...................................... 202
APPENDIX C - - -‘ - 215
Ambiguous Words Task Instructions .......................................................................... 215
Garden Path Instructions: .......................................................................................... 216
APPENDIX D -- - - 217
Ambiguous Words Task Stimuli .................................................................................. 21 7
Garden Path Stimuli ................................................................................................... 219
REFERENCES.-- ..... - - - - - 224

 

vii

LIST OF TABLES

Table 1. Demographics Description: Means (SD). ........................................................... 52
Table 2. Example Sentences by Condition for the Garden Path Task. ............................ 65
Table 3. Demographics Description: Means (SD). ........................................................... 73
Table 4. lnattention, Hyperactivity, ODD, and CD by group. ......................................... 74
Table 5. Achievement, Language, and Learning: Means by group. ................................ 77
Table 6. Example Sentences and Condition Error Rates Across Groups. ....................... 78

Table 7. Ambiguous Words Task Means for Accuracy and Reaction Time Across
Groups for Total Sample, Visual, and Auditory Task Versions. .............................. 83

Table 8. Correlations for Accuracy and RT Cost Scores across Groups ....................... 86

Table 9. Correlations for Prime-Target Changes in Accuracy and RT for Non-RD
Children in the Different and Nonsense Conditions of the Visual Task (N=75) ...... 87

Table 10. IQ, Achievement, and Language Correlations for Accuracy and RT Cost
across Non-RD Groups on the Visual Task (N=75). ................................................ 89

Table 11. RT Cost Regressions for ADHD and Control Children on the Visual
Ambiguous Words Task (N =75) ............................................................................... 91

Table 12. Accuracy Cost Regressions for ADHD and Control Children on the Visual
Ambiguous Words Task (N =75) ............................................................................... 92

Table 13. Group Means for Accuracy and RT on the Visual Ambiguous Words Task .. 94

Table 14. Achievement/IQ Correlations for Accuracy and RT Cost Scores for RD
Children on Visual and Auditory Tasks .................................................................... 98

Table 15. Inattention and Hyperactivity Correlations for Accuracy Cost for RD children
on the Visual and Auditory Task ............................................................................ 101

Table 16. Accuracy Cost Regressions for the Visual Ambiguous Words Task: RD (N =8)
................................................................................................................................. 102

Table 17. RT Cost Regressions for the Visual Ambiguous Words Task: RD (N=8) 102

Table 18. Accuracy Cost Regressions for Auditory Ambiguous Words Task (N=12) . 103

viii

Table 19. RT Cost Regressions for Auditory Ambiguous Words Task (N=12) ............ 103
Table 20 Auditory Task ................................................................................................. 105
Table 21 Visual Task ..................................................................................................... 106

Table 22. RD Group Means for the Auditory and Visual Versions of the Ambiguous
Words Task ............................................................................................................. 107

Table 23. Example Sentences and Condition Error Rates Across Groups for the Visual
and Auditory Garden Path Task. ............................................................................. 108

Table 24. Derived Index Scores for Garden Path and Plausibility Effects for Visual and
Auditory versions of the Task: ................................................................................ 116

Table 25. Garden Path Effects Correlations for IQ, Achievement, and Language ........ 118

Table 26. Effect of Verb Type on Garden Path Effects in Plausible Conditions ........... 119

Table 27. Correlations between Plausibility Effects and Indices of Achievement and
Language. .............................................................................. _ .................................. 120

Table 28. Garden Path Effects: Correlations with Inattention and Hyperactivity. ........ 122

Table 29. Correlation between Plausibility Effects and Inattention and Hyperactivity. 123
Table 30. Total Garden Path Effect (across Plausibility and Verb Types) .................... 125
Table 31. Condition Means by Diagnostic Groups ........................................................ 132
Table 32. Derived Index Scores for Garden Path and Plausibility Effects by Group:... 133
Table 33 Mean Parent Behavioral Ratings by Group (T-Scores). .................................. 173
Table 34 Comparison of RD children with and without situational inattention. ........... 177

Table 35 Accuracy Correlations Among Ambiguous Words Task Conditions and Age for
Both Primes and Targets. ........................................................................................ 181

Table 36. Reaction Time Correlations Among Ambiguous Words Task Conditions for
Primes and Targets (Visual Version: N=83). .......................................................... 182

" Table 37. Reaction Time Correlations Among Ambiguous Words Task Conditions for
Primes and Targets (Auditory: N=14). ................................................................... 183

Table 38. Inattention Correlations with Accuracy and Reaction Time by Condition. .. 188

ix

Table 39. Inter-Condition Correlations Among Garden Path Task Conditions (Across

Visual and Auditory: N=97). .................................................................................. 193
Table 40. Inter-Condition Accuracy Correlations Among Garden Path Task Composites
(across Visual and Auditory Task) ......................................................................... 194
Table 41. Accuracy Correlations Among Garden Path Task Conditions (Auditory Task
Only: N=14). ........................................................................................................... 195
Table 42. Plausibility Effect on Garden Pathing (across verb type). ............................. 197
Table 43. Effect of Verb Type on Garden Pathing (across Plausibility Types)—NonGP-
GP for OT vs. RAT ................................................................................................. 199
Table 44. Comparison of Plausibility and Verb Type on Garden Path Effect
(Verb*Plausibi1ity*GP Interaction) ........................................................................ 201
Table 45. Auditory vs. Visual for Gemsbacher Task .................................................... 202
Table 46. IQ, Achievement, and Language Correlations for Accuracy and RT Benefit
across Non-RD Groups on the Visual Task (N=75). .............................................. 204
Table 47. Inattention and Hyperactivity Correlations for Accuracy and RT Benefit across
Non-RD Groups (N=75) ......................................................................................... 204
Table 48. RT Benefit Regressions for Visual Ambiguous Words Task (N =75) ........... 206
Table 49. Mean Benefit (Facilitation) Scores by Diagnostic Group ............................. 207

Table 50. IQ, Achievement, and Language Correlations for Accuracy and RT Benefit for
RD children ............................................................................................................. 209

Table 51. Inattention and Hyperactivity Correlations for Accuracy and RT Benefit for RD
children on the Visual and Auditory Task .............................................................. 210

Table 52. RD Accuracy Benefit Regressions for Auditory Ambiguous Words Task
(N=12) ..................................................................................................................... 212

Table 53. RD Benefit RT Regressions for Visual Ambiguous Words Task (N=8) ....... 214

LIST OF FIGURES

Figure 1. Group by Sex Interaction for Accuracy Cost in ADHD vs. Control ................ 93

Figure 2. Comparison of Prime-Target RT Differences in the Nonsense vs. Different
Conditions ................................................................................................................. 96

Figure 3. Garden Path by Plausibility Interaction for RAT and OT Verb Conditions in
Control Children. .................................................................................................... 11 1

Figure 4. Garden Path x Group Interaction .................................................................... 127

Figure 5. Comparison of Garden Path Effects for Control vs. ADHD-C children for OT

verbs. ....................................................................................................................... 128
Figure 6. Comparison of Garden Path Effects for ADHD vs. Control Children on RAT
verbs. ....................................................................................................................... 129
Figure 7. Conceptual Model for the Relationship of Reading Ability to Interference
Control in Children with RD and Normal-Reading Controls. ................................ 144
Figure 8. Theoretical Model of Possible Pathways from ADHD to RD and RD to ADHD.
................................................................................................................................. 163
Figure 9. Group by Stimulus Interaction for Control vs. ADHD-C in the Nonsense
Condition............... .................................................................................................. 190
Figure 10. Group by Stimulus Interaction for Control vs. ADHD—C in the Different
Condition ................................................................................................................. 190

Figure 11. Verb x Plausibility Interaction for Control Subjects on the Garden Path Task.
................................................................................................................................. 196

xi

Inhibitory Language Deficits in Attention Deficit/Hyperactivity Disorder and Reading

Disorder: A Candidate Shared Deficit

Attention deficit-hyperactivity disorder (ADHD) and Reading Disorder (RD) are
among the two most commonly diagnosed psychological disorders of childhood. ADHD
affects some 3-5 percent of school-aged children while RD affects approximately 3-9
percent (Offord et a1., 1987; Shaywitz, Shaywitz, Fletcher, & Escobar, 1990; Rutter,
1978). Both disorders can be highly impairing and often lead to school failure, low self-
esteem, oppositional behavior, and other emotional or psychological problems (Frank,
1996). Considerable empirical evidence supports the existence of distinct cognitive and
neuropsychological deficits in the two disorders, with the most robust of these being the
double dissociation between (a) phonological processing deficits in RD and (b) executive
function deficits in ADHD (Pennington, Groisser, & Welsh, 1993). Findings of distinct
deficits do not, however, explain a puzzling characteristic of the two disorders—the high
frequency with which they co—occur (Beitchman, Hood, & Inglis, 1990). Based on their
prevalences, the expected rate of ADHD+RD comorbidity in the community would be
approximately 1 percent if they are causally unrelated, even assuming upper bound
estimates. Yet, studies consistently find a 10 to 90 % incidence of comorbidity between
the disorders, with conservative estimates putting the overlap at 30-50% depending on
the sample examined (Beitchman & Young, 1997; Hinshaw, 1992).

One logical explanation for this high overlap might be that RD children develop
behavioral problems resulting from school frustration, whereas ADHD children fail to
achieve due to inattention in school. Empirical evidence to date suggests, however, that

at best, this “phenocopy” explanation accounts for only some of the cases of ADHD+RD

co-occurrence (Spreen, 1989). One potential alternative explanation is that the risk for
developing both disorders increases as a result of an as yet unidentified shared cognitive
deficit. The interface between language processing and inhibitory control is one
theoretically promising but largely unexplored possibility for such a shared deficit.
Language and inhibitory processes have been examined separately in both ADHD and in
RD, but few studies have examined their interaction. Indeed, mechanisms by which
irrelevant information is suppressed are one of the leading contenders for a “core deficit”
in ADHD, yet are largely unstudied in the language domain—the domain in which one
would expect a link to RD to occur.

The current study sought to examine inhibitory-reliant language processes in
children with ADHD and with RD, with the aim of exploring whether the disorders share
a common cognitive deficit that could help explain their high rates of co-occurrence.
Information on a number of topics relevant to the two disorders is provided here so as to
establish a framework within which readers can fit the current study. Background
regarding diagnostic criteria, symptomatology, and limitations inherent to the DSM-IV
(American Psychiatric Association, 1994) classification of ADHD and RD will be
reviewed, followed by a consideration of theories regarding the etiology of ADHD+RD
comorbidity, evidence that the two disorders have distinct cognitive deficits, and
candidate domains for locating a shared deficit. Finally, discussion of the role of
inhibition in language processing will be provided in the context of its relevance to
ADHD+RD comorbidity as well as its importance to a deeper understanding of ADHD

per se.

Background on ADHD and RD
Background on ADHD

ADHD is among the most common reasons for referral to pediatric clinics and
other mental health professionals (Offord et al., 1997). The DSM-IV (American
Psychiatric Association, 1994) describes ADHD as developing in very early childhood
and occuning four to nine times more ofien in males. As currently defined, the syndrome
is diagnosed in the presence of six behavioral symptoms of either hyperactivity/
impulsivity or inattention/disorganization that are extreme for age and begin prior to age
seven. Impairment from these symptoms must be present in at least two settings for a
period of at least six months. Three sub-types are defined: predominantly inattentive,
predominantly hyperactive-impulsive, and combined types. Symptoms such as excessive
motor activity often manifest themselves from toddlerhood and usually remain stable and
severe throughout childhood. Children with ADHD ofien experience impaired academic
achievement, peer rejection, and difficult family relationships (Barkley, 1996). Although
many sufferers experience a waning of hyperactive/impulsive symptoms in late
adolescence and adulthood, many others experience poor long-term outcomes such as
antisocial behavior and substance abuse, traffic and other accidents, employment
problems, and marital problems (Carroll & Rounsaville, 1993; Loeber & Keenan, 1994).

Although the DSM-IV’s (American Psychiatric Association, 1994) differentiation
of the ADHD syndrome into three subtypes has produced enhanced diagnostic
specificity, considerable within-group symptom heterogeneity continues to pose
problems for establishing the disorder’s construct validity. For example, the majority of

children with ADHD—68 percent according to one survey (Offord et al, 1987)—

receives one or more additional, or comorbid, diagnoses (Jensen, Martin, & Cantwell,
1997). Commonly co-occurring disorders include other disruptive behavior disorders
(e.g., Oppositional Defiant and Conduct Disorder), anxiety, depression, and learning
disabilities (Biederrnan, Newcom, & Sprich, 1991; Jensen, Martin, & Cantwell, 1997;
Faraone et al., 1993; Nottleman & Jensen, 1995; Hinshaw, 1992). Reading disorder is a
particularly important yet poorly understood comorbid condition as noted earlier.
Background on Reading Disorder

Reading disorder, as currently defined, is diagnosed in the presence of a
significant discrepancy (usually defined as 1.5 standard deviations) between normal
intellectual abilities and below average reading achievement. Effects can persist into
adolescence and adulthood and have a negative impact on interpersonal relationships,
earning power, and school or professional achievement (Culbertson & Edmonds, 1996;
Aram, Ekelman, & Nation, 1984). As is the case for ADHD, the current diagnosis of RD
suffers from a number of problems with its construct validity. The DSM-IV and
empirical literature differentiate between RD and “garden-variety” poor reading, which is
characterized by below average functioning in both reading achievement and IQ. Some
recent empirical evidence disputes this distinction, however, finding no differences
between the two groups’ reading impairments (Beitchman & Young, 1997; Shaywitz,
Fletcher, & Shaywitz, 1994). In addition, after extensive work in a variety of cognitive
domains (including visual processing deficits and “surface” dyslexia), a growing
consensus in the empirical literature favors phonological processing impairments as
central to most manifestations of RD (Morris et al., 1998; Pennington, 1991). Yet, this

impairment is not incorporated into current diagnostic criteria.

Children with RD are also a heterogeneous group. One recent study, for instance,
identified seven different subtypes of RD. Although most of these subtypes shared a
phonological processing deficit, they differed according to deficits in verbal short-term
memory, processing speed, and rapid serial naming (Morris et al., 1998). Children with
RD also commonly experience psychiatric problems. ADHD is the most common of
these, with reported rates of co-occurrence between 15 and 53 percent in both clinical and
community samples (Baker & Cantwell, 1987; Cantwell & Baker, 1991; Riccio & Hynd,
1993; Beitchman, Hood, & Inglis, 1990; Prizant, Audet, Burke, Hummel, Maher, &
Theadore, 1990; Wood & Felton, 1994; Shaywitz, Fletcher, & Shaywitz, 1994).

Conceptual Diagnostic Issues

Ultimately, characterizing groups according to cognitive or neuropsychological
deficit patterns, if it can be done, could set the stage for uncovering neurobiological or
other etiologies and, in turn, for developing more effective preventions or treatments
(Shaywitz, Fletcher, & Shaywitz, 1994). Such information would, at the same time, help
determine whether the separation of ADHD from key comorbid disorders in the DSM-
IV’s behaviorally-based taxonomy is valid from the point of view of associated
neuropsychological deficits. Yet, doing so is not straightforward.

Due to the considerable heterogeneity within ADHD and RD samples,
distinguishing these disorders’ primary causes from their secondary symptoms is
difficult. The usual practice of studying DSM-IV or other behaviorally-defined clinical
groups underscores this heterogeneity problem, in that a diagnosed group may well
consist of children with similar behavioral symptoms but different neurobiological

etiologies. From a neuropsychological perspective, then, DSM-IV diagnostic

categorization may sometimes lump distinct cognitive dysfunctions into a single
diagnosis or, alternatively, differentiate related dysfunctions into separate diagnoses. In
its tacit adoption of a descriptive medical syndrome model, the DSM-IV also admittedly
fails to address developmental or causal pathways to disorder. For instance, the same
symptoms and diagnostic criteria (i.e., number of symptoms) are required for diagnosis
regardless of age.

These issues are especially salient to questions regarding the causes of
comorbidity. For instance, comorbidity between ADHD and RD could result from: (1) a
distinct syndrome with a unique etiology (e.g., ADHD+RD has a unique etiology
compared to ADHD or RD alone; Rowe & Rowe, 1992), (2) the independent or chance
co-occurrence of multiple disorders (unlikely in ADHD+RD as noted earlier), or (3) a
single shared underlying neuropsychological deficit that produces vulnerability towards
different manifestations of disorder (i.e., different phenotypes), and leads to greater than
expected co-morbid cases (Willcutt et al., 2001). Because DSM-IV classification is based
solely on a behavioral level of analysis, attempting to differentiate syndrome-specific
from generalized or shared cognitive deficits (i.e., deficits common to both disorders)
may be a key strategy for clarifying the etiological relationship between ADHD and RD.

Despite the DSM-IV’s limitations, the validity of its diagnoses (and the symptom
clusters upon which they are based) is extensively supported empirically and provides a
common framework for describing and classifying children across studies (Lahey et al.,
1994). Furthermore, although there is considerable heterogeneity among individuals with
ADHD or RD diagnoses, the use of these classifications has yielded well-replicated

group-specific cognitive and neurological deficits, notably executive function deficits in

ADHD and phonological processing deficits in RD. Thus, DSM-IV categorization
provides an essential tool for categorizing and describing psychological problems both
for clinical and empirical purposes.
Summary

Despite the benefits of the DSM-IV (e. g., its wide-spread use across studies and
empirically validated symptom clusters), it does not shed light on etiology, so improved
understanding of the cognitive and neuropsychological deficits associated with ADHD
and RD may be one useful approach for clarifying the relationship between the two
disorders.

ADHD+RD Comorbidity Findings

A note on terminology

Before reviewing key aspects of RD and ADHD comorbidity, a note about some
of the distinctions in this literature is necessary. The empirical literature on the co-
occurrence of ADHD and language or learning problems varies widely according to how
the impairments are classified and which specific problem is being studied. Studies on
the relationship of ADHD to RD, learning disability (a broader category which subsumes
RD), or language impairments (an overlapping but also broad category) could reasonably
be discussed as distinct bodies of literature. Broad language impairment, for instance, is
a sufficient although not a necessary condition for producing a reading deficit. However,
though they are distinct concepts, RD, learning disability, and language impairments are
also highly inter-related. RD, for instance, is often preceded by language impairment and
is the most common learning disability associated with language impairment (Riccio &

Hynd, 1993; Riccio & Jemison, 1998). Moreover, language impairments and learning

disabilities are often distinguishable primarily by the age of the child, with language
impairments during preschool more likely to be manifested as learning disabilities at
school-age (Kahmi & Catts, 1986).

A second reason for including findings regarding learning disability and language
impairment is the current study’s emphasis on language functioning as the possible
shared mechanism in RD and ADHD. Language impairments can act as a mediator
between attention and reading problems (Riccio & Jemison, 1998) and, conversely, RD
may mediate links between language impairment and disruptive behavior problems
during development (Tomblin, Zhang, Buckwalter, & Catts, 2000). Language impairment
therefore may play a crucial role in understanding the relationship between RD and
ADHD.

Thus, my review of findings on ADHD+RD includes studies of learning disability
(LD) or language impairments (LI). Whenever possible, or relevant, the specific
syndrome under study will be identified and distinctions noted.

Background on ADHD +RD Comorbidity

Depending on the criteria used, on average, ADHD children have comorbidity
rates of between 30 to 50 % for formal learning disorders (Tomblin et al., 2000; Semrud-
Clikeman et al., 1992; Hinshaw, 1992). A number of explanations exist for the wide
discrepancies in comorbidity estimates between studies. These include sampling biases
(e. g., use of clinical versus community samples or RD clinic-recruited versus ADHD
clinic- recruited samples) and the criteria used to diagnose RD (e. g., IQ—achievement
discrepancies, diagnosis by achievement deficits without comparisons to IQ, or exceeding

cut-offs on any one of several different diagnostic measures; Semrud-Clikeman et al.,

1992). Thus, studies have reported academic difficulties or LD diagnoses in a large
percentage of ADHD children, with one study finding over one-half in need of tutoring
and one-third placed in special classes or repeating a grade (F araone et al., 1993;
Cantwell & Baker, 1991; Hinshaw, 1992; Love & Thompson, 1988; Carlson, Lahey, &
Neeper, 1986; Barkley, DuPaul, & McMurray, 1990; Lamminmaki, Ahonen, Nirhi,
Lyytinen, & deBarra, 1995; Trautman, Giddan, & Jurs, 1990; Elbert, 1993).

With regards to children with RD, more subclinical symptoms of ADHD are
exhibited compared to nonimpaired children (Willcutt etal., 2001). Voeller (1999)
concluded that the occurrence of attention problems in RD may be much higher when
examined from a dimensional rather than a categorical perspective (i.e., allowing “sub-
diagnostic” but impairing symptoms to be noted) and when classification is based on
neurological/cognitive tasks rather than behavioral ratings. Voeller argued that most RD
children have some degree of cognitively-measurable attentional deficit.

Some evidence suggests that learning and language problems in ADHD may vary
according to subtype—specifically, that they may be more common in the inattentive
subtype (Tirosh & Cohen, 1998; Lamminmaki, Ahonen, Narhi, Lyytinen, & deBarra,
1995; August & Garfinkel, 1989; Shaywitz, Fletcher, & Shaywitz, 1994). August and
Garfinkel (1989) differentiated between a behavioral and cognitive subtype of ADHD
and found the cognitive subtype to be associated with inattention, linguistic information
processing deficits, and academic underachievement. One study in Chile also found
reading problems to be more prevalent in inattentive than in combined type children,

though combined-type children also demonstrated more reading problems than control

children (Lamminmaki et al., 1995). Thus, language and learning problems are found in
both subtypes of ADHD children.

In addition to subtype differences, learning disabilities and language impairment
in ADHD children increase with age (or underlying learning disabilities or language
impairments can be more easily identified as academic demands increase in later school
grades; Tirosh & Cohen, 1998). For instance, a combination of hyperactivity, inattention,
and language delay during preschool may be a marker for later diagnosis with ADHD
(Beitchman, Tuckett, & Batth, 1987; McGee, Partridge, Williams, & Silva, 1991). A
longitudinal study on Israeli school children found that 80 percent of preschoolers with
this triad of symptoms had a diagnosis of combined or inattentive subtypes of ADHD
when followed-up between 7 and 14 years of age (Omoy, Uriel, & Tennenbaum, 1993).
Furthermore, Baker and Cantwell (1992) reported that, wheareas 20 percent of 6 and 7
year-olds with ADHD + history of speech/language disorder were also diagnosed with a
learning disability, this rate increased to 86 percent of children aged 11-15 (Baker &
Cantwell, 1992).

Sex of a child is also an important factor in the incidence of comorbidity between
ADHD and RD. Although ADHD and RD are both more common in boys than girls,
Beitchman, Hood, and Inglis (1990) reported that girls with speech/language disorders
were twice as likely as boys to have a co-occurring psychiatric disorder. Girls are
therefore more highly represented in groups of children with concurrent ADHD and
language disorders than in samples of ADHD—only children (Berry, Shaywitz, &
Shaywitz, 1985; Love & Thompson, 1988; Tirsoh & Cohen, 1998). This difference may

be especially true of girls in the predominantly inattentive subtype (Love & Thompson,

10

1988). Girls with concurrent ADHD and language impairments also demonstrate more
severe language deficits, lower verbal IQ’s, and poorer academic performance than
ADHD boys with language impairments (Berry, Shaywitz, & Shaywitz, 1985;
Beitchman, Hood, & Inglis, 1990). These sex differences may be an artifact of
ascertainment biases (i.e., only severely impaired girls are referred for treatment due to
demonstration of fewer behavior problems), or may be due to polygenetic transmission
factors (i.e., girls must carry a higher genetic load and meet a higher threshold before
phenotypic impairment becomes evident) (Berry, Shaywitz, & Shaywitz, 1985). In either
event, consideration of possible sex differences is important in studies of ADHD+RD
comorbidity.

Now that a summary of overarching background issues relevant to understanding
ADHD+RD comorbidity has been provided, I next examine findings on the cognitive and
neuropsychological deficits associated with comorbid ADHD+RD. Findings in this
domain will lay a foundation upon which to describe candidates for shared deficits that
potentially could explain this comorbidity, enabling description of the rationale for the
current study. Most comorbidity studies to date have focused on determining whether the
cognitive and neuropsychological deficits associated with each disorder co-occur in their
comorbid form or whether one of the disorders is a phenocopy. Evidence for each of
these will now be reviewed.

Cognitive and Neuropsychological Findings on ADHD+RD Comorbidity
Phenocopy Findings.
A number of studies have supported the idea that comorbid ADHD+RD has only

one disorder’s neuropsychological impairments (e. g., phonological processing or

11

executive functioning deficits), while the symptoms of the other disorder behaviorally
mimic, or are phenocopy symptoms, of true neuropsychological deficits (e.g.,
hyperactivity only as a sequelae of RD-related school frustration; poor reading
achievement as a sequelae of ADHD-related inattention at school; Zentall, 1993;
Cantwell & Baker, 1991; Pennington, Groisser, & Welsh, 1993; Frith, 1995). Some
empirical findings suggest that comorbid ADHD+RD involves RD-specific cognitive
deficits (McGee, Williams, Moffitt, & Anderson, 1989), while ADHD symptoms result
from school frustration and other negative effects of early reading difficulties (Tomblin et
al., 2000; McGee & Share, 1988; Stanovich, 1986). A study by Shaywitz and colleagues
(1995), for example, reported significant impairment in ADHD+RD children on measures
of phonological processing and teacher ratings of activity, attention, and behavior, but not
on a cognitive measure of visual attention (an ADHD-related deficit measured by an

' underlining task). Similar findings of phonological but not executive function deficits in
an ADHD+RD group were reported by Pennington, Groisser, and Welsh (1993). Results
of these studies therefore suggest that comorbid ADHD+RD consists of RD children with
secondary behavior problems mimicking ADHD.

However, other studies have found the cognitive deficits of ADHD to be primary
in the comorbid group, whereas symptoms of RD are secondary results of school failure
(i.e., no problems with phonological processing are found). Rowe and Rowe (1992), for
example, reported a reciprocal relationship between primary inattention in school and
subsequent poor reading achievement that then leads to further exacerbation of
inattentiveness in school. F ergusson and Horwood (1992) found that attention influenced

reading achievement even though reading achievement did not influence attention.

12

These studies suggest that ADHD+RD is not “truly” comorbid (i.e., only one
disorder has the underlying cognitive deficit of the pure disorders). Although this finding
is probably supported in some cases of ADHD+RD comorbidity, the majority of recent
research has not supported the phenocopy hypothesis, instead finding that the comorbid
group has distinct cognitive deficits associated with both pure disorders. These findings
will now be reviewed.

Distinct Deficits Findings

Some studies examining ADHD+RD comorbidity have found that
neuropsychological deficits characteristic of both “pure” disorders co-occur in the
comorbid form. Tirosh et a1. (1998), for example, found that parent and teacher
behavioral report of attention and learning problems both loaded on the same factor in
ADHD+LD children. Several studies have reported a double dissociation between
ADHD and RD on measures of executive functioning (ADHD) and phonological
processing (RD) (Pennington, Groisser, & Welsh, 1993; Shaywitz, Fletcher, & Shaywitz,
1995, Nigg, Hinshaw, Carte, & Treuting,l998; Ackerman, Dykman, & Gardner, 1990;
Reader, Ham's, Schuerholz, & Denckla, 1994; Willcutt et al., 2001; Klorrnan etal.,
1999). In addition, ADHD-specific deficits in effortful executive control processes and
RD-specific deficits on more automatic processes have also been documented (Carte,
Nigg, & Hinshaw, 1996; Kemp & Kirk, 1993; Ackerman, Anhalt, Holcomb, & Dykman,
1986; Ackerman, Anhalt, Dykman, & Holcomb, 1986). Certain language deficits also
appear to be specific to diagnosed language impairments or learning disabilities,
regardless of comorbid disorders (Vallance, Im, & Cohen, 1999). Cohen et al. (2000), for

instance, reported that receptive and expressive language deficits were specific to

13

diagnosed language impairment and were not impaired in children with “pure” ADHD
relative to those with other comorbid psychiatric diagnoses, thus suggesting that any
language deficits in ADHD resulted from co-occum'ng language impairment.
Severity Findings

Other findings have suggested that the comorbid ADHD+RD group not only has
the cognitive symptoms of both disorders, but is also more globally impaired than those
with the pure forms of the disorders. For instance, auditory memory (Kataria, Hall,
Wong & Keys, 1992; Tamowski, Prinz, & Nay, 1986) and working memory (Shaywitz et
al., 1995; Willcutt et al., 2001) are more impaired in children with ADHD plus language
impairments than in either ADHD-only or RD-only children. ADHD plus language
impaired children have also been found to be more distracted by external stimuli than
either pure ADHD or language impaired children (Cotugno, 1987) and to have greater
impairment in word decoding and sequential memory (August & Garfinkel, 1989).
Final Methodological Note

Several methodological inconsistencies in studies of ADHD+RD comorbidity
make it difficult to generalize findings about the co-occurrence of the disorders. For
instance, few studies (with several notable exceptions: e.g., Pennington, Grossier, and
Welsh, 1993; Purvis & Tannock, 1995) have simultaneously utilized a sample including
children with the “pure” form of both disorders as well as children with their comorbid
form. Thus, the majority of studies have used samples of children with either RD or
ADHD and compared them to those with ADHD+RD (e. g., Halperin, Gittelrnan, Klein,
& Rude], 1984). They also tend to study functioning in only one domain (e. g., language

or executive function). Sampling biases are an additional concern in many of these

14

studies, since the comorbid children were often drawn from primarily RD or primarily
ADHD samples (e. g., children from ADHD or RD clinics rather than from community
samples). Finally, the criteria used to define RD or language impairment have varied
widely across studies. Some studies, for instance, required larger IQ-achievement
discrepancies than others did, while other studies required poor achievement without any
consideration of discrepancy from IQ (Semrud-Clikeman et al., 1992; Purvis & Tannock,
2000).
Conclusions

Despite these methodological problems, findings provide considerable evidence
for the co-occurrence in ADHD+RD of cognitive deficits characteristic of the respective
“pure” disorders. Though successful in describing the cognitive and neuropsychological
impairments in ADHD+RD, research has not advanced nearly as well the question of why
these two disorders co-occur so frequently. One potential answer is that, in addition to
their unique cognitive dysfunction, both disorders also share another cognitive or
neuropsychological deficit (such as inattention, disinhibition, or working memory
impairment) that may predispose an individual to development of both disorders and
which may lead to greater than chance co—occurrence (Love & Thompson, 1988;
DeLong, 1995; Purvis and Tannock, 1999; Willcutt et al., 2001; Spreen, 1989; Denckla,
1996). This hypothesis of a shared deficit, though not well-explored, provides an
intriguing possibility for explaining the etiology of ADHD+RD comorbidity. If this
shared deficit theory is true, then both pure disorders should have a common deficit.

Attention to the existence of possible shared cognitive deficits in each of the “pure”

15

forms of the disorders (RD alone, ADHD alone) may thus provide clues regarding the
etiological underpinnings of ADHD+RD comorbidity.

Evidence for this hypothesis will now be considered. In each section, findings of
studies that directly compared ADHD and RD will be reported first, followed by findings
for each disorder individually. Because the shared deficit hypothesis is not yet well-
explored, most empirical findings are from studies examining one of the “pure” disorders.
Again, because of methodological disparities across studies, this review includes findings
for LD and language impaired children, in addition to those for RD.

Empirical Evidence for Shared Deficits
Genetics Studies

Genetic studies have provided mixed evidence regarding the etiology of ADHD
and RD. Faraone and colleagues (1993) reported evidence for independent genetic
precursors of the two disorders (perhaps through nonrandom mating). In a twin sample
selected for ADHD, children with ADHD + LD (learning disability) had a greater number
of relatives with a learning disability compared to ADHD-only children and an equal
proportion of relatives with ADHD. However, other studies have supported a shared
etiology for the two disorders. In a sample of twins selected for the presence of RD, 45
percent of the proband deficit in reading was accounted for by genetic factors that also
contributed to hyperactivity. Due to the significant bivariate heritability between ADHD
and RD, common genetic influences were concluded to contribute to the co—occurrence of
the two disorders, even if nonrandom (cross-assortative) mating was also a contributing

factor (Light, Pennington, Gilger & DeFries, 1995; Willcutt et al., 2001 ).

l6

Anderson (1999) reported results of studies in rodents that provide one
preliminary theory regarding the genetics of ADHD+RD co—occurrence. These studies
examined the effects of mutations involving the Dlx-l and Dlx-2 genes, which derive
from the subventricular zone of the lateral geniculate eminence and play a role in cell
migration and differentiation in the striatum and lateral forebrain. Mutations of the genes
caused a proliferation of cells in the striatum that never migrated to neocortex. In
particular, a 75 percent reduction in GABAergic intemeurons of the neocortex occurred
in those rodents with a Dlx-l or Dlx-2 mutation. Given hypothesized inhibitory deficits
(often related to functioning of GABAergic neurons) and striatal involvement in ADHD
(discussed below) as well as some evidence linking RD with alterations in the lateral
geniculate nucleus and its magnocellular pathways, these findings provide one theoretical
mechanism of genetic transmission.

Neuroimaging

Imaging studies have also provided evidence for shared deficits. Hynd et a1.
(1990) found several shared neuroanatomical differences in ADHD and RD individuals
versus controls. For instance, both ADHD and RD groups demonstrated smaller MRI-
measured widths of right anterior brain regions compared to control subjects. RD subjects
also exhibited a significantly shorter insular region bilaterally, while those with ADHD
fell between RD and control subjects, but were nonsignificantly different from both
groups (Hynd et al., 1990). A RD-related right>lefi planum temporale length was the
main distinguishing characteristic differentiating ADHD and RD individuals (Hynd,

Semrud-Clikeman, Lorys, Novey, Eliopulos, & Lyytinen, 1990).

17

This pattern of partially overlapping neural abnormality also extends to findings
involving striatal activation. Although decreased striatal activation has previously been
associated with ADHD (Lou, Henriksen, Bruhn, Burner & Nielsen, 1989), a recent study
suggests this trait may not be specific to the disorder. Lou, Andresen, Steinberg,
McLaughlin, and F riberg (1998) studied the relationship between verbal awareness and
striatal functioning in a' group of ADHD and comparison children using functional
imaging to measure regional cerebral blood flow (rCBF). As in previous studies, ADHD
children exhibited decreased rCBF in the right striatum compared to controls in semantic
processing conditions. Findings also indicated that striatal and infero-frontal activation
were related to the language demands of the experimental tasks (semantic processing
under passive listening to a series of animal names) for both groups, while activation of
the anterior cingulate resulted from higher-order attentional demands (detection of
“dangerous animal” targets). Lou and colleagues concluded that the activation of
inferofrontal and striatal regions supports their role in verbal awareness. Given that
striatal activation is associated with passive language processing and children with
ADHD demonstrate decreased striatal activation under these conditions, these findings
provide a possible association between ADHD-related brain impairments and deficits in
language processing.

Wood and Flowers (1999) conducted a factor analysis of PET activation findings
to examine covariation in neuroanatomical regions of interest. They identified nine
neuroanatomical factors and compared activation in these factors to behavioral measures
of reading and attention. Three distinct phenotypes based on deficits (cut scores

identifying 15 percent of population) in phonemic awareness, phonological decoding, and

18

single word reading were created. The phenotypes were related to activation in inferior
temporo-occipital, right central-frontal, and thalamic areas, respectively. Interestingly,
both phonemic awareness and inattention were related to activation on the
bilingual/hippocampal (inferior temporo-occipital) factor, suggesting a possible common
substrate to ADHD and RD deficits (or at least for inattention found in those with RD).
The relationship between phonological decoding and right hemisphere activation was
hypothesized to be related to requirements for integration and synthesis of phonemes.
With regards to findings for other symptoms related to ADHD, hypoactivation of
bifrontal and biorbital factors was related to impulsivity and D-prime (i.e., discriminating
targets from non-targets) on a continuous performance test. The authors also speculated
that the findings for inattention, perhaps as found in predominantly inattentive ADHD,
may be associated with a more posterior, sensory-perceptual deficit.
Conclusions about Genetic and Neuroimaging Findings

Overall, neuroimaging and genetics data provide fairly strong support for some
shared biological correlates in ADHD and RD, particularly shared genetic influences and
shared neural abnormalities in right frontal, inferior temporo-occipital, and possibly
striatal regions. Some findings in the cognitive literature also provide support for a
relationship between the disorders. Auditory processing, working memory, language,
and inhibitory processing are all candidate domains that may provide specification for a
putative cognitive overlap between ADHD and RD. Evidence for each candidate

cognitive domain will now be discussed.

l9

Neurocognitive Deficits

Auditory Processing

Impairments in auditory processing mechanisms provide one potential common
pathway of comorbidity in ADHD and RD. Riccio and colleagues (1994) reported that
differences in behavioral symptom profiles between ADHD children with and without
language impairments disappeared when central auditory processing was used as the
measure of language impairment classification. Thus, the behavioral correlates of central
auditory processing disorder and ADHD appeared to be similar. Children with ADHD
have also been found to demonstrate reduced N40 evoked potential amplitudes similar to
non-impaired younger children. Dysfunction (or immaturity) in the brainstem thalamo-
cortical projections underlying the lowest level of the auditory system (which controls
sound localization, noise suppression, and signal detection/target selection) are
hypothesized to underlie this dysfimction (McPherson & Davies, 1995). Inhibitory
mechanisms (which will be discussed in greater detail later) may influence this thalamo-
cortical processing mechanism, thereby suggesting a potential cause of auditory
processing problems in ADHD.

Auditory attention and auditory processing have also been found to improve in
children with ADHD following Ritalin administration. This improvement was
demonstrated by better performance on hit rate and false alarms for an auditory
continuous performance test and on a receptive language test requiring following
increasingly lengthy and complex commands (Token Test). ADHD children did not
show improvement on an auditory figure-ground test that measured ability to

discriminate words in the presence of competing background noise, suggesting the

20

possibility of a more severe deficit or a deficit in a pathway not sensitive to Ritalin
treatment (Keith & Engineer, 1991).

Interestingly in light of the ADHD findings, auditory deficits in children with RD
are also supported. On an auditory continuous performance task comparing responses to
rhymed words versus the lowest volume tone in a group, children with RD demonstrated
anterior temporal activation on both tasks in contrast to the temporoparietal and posterior
frontal activation pattern of control children (Rumsey et al., 1992). Children with RD
also demonstrated increased metabolism in the medial temporal lobes bilaterally
compared to controls on an auditory continuous performance task that required response
to a target syllable (Hagman, Wood, Buchsbaum, Tallal, Flowers, & Katz, 1992).
Examination of ERP activation during a tone discrimination task revealed an absence of
the right hemisphere lateralized activation pattern of control subjects, as well as impaired
ability to discriminate deviant tone patterns. No differences were found for discrimination
of pairs of tones (i.e., those without patterns). These deficits were interpreted to support
a basic nonlinguistic auditory processing deficit that occurs under requirements for
discrimination of temporal patterns within the context of other sounds (e. g., like
phonemes in words) (Kujala, Myllyviita, Tervaniemi, Alho, Kallio, Naatanen, 2000).

Summary. Overall, then, evidence supports the existence of auditory processing
deficits in both ADHD and in RD. Few, if any studies, appear to have directly compared
the deficits of the two groups or their comorbid form, however. Although auditory
processing is thus a viable candidate for a shared deficit in the two disorders, this deficit
could also occur subsequent to another, more primary, shared impairment such as in

inhibition (as suggested above by McPherson & Davies, 1995).

21

Working Memory

Working memory—itself a multi-component construct-- is one of the most
widely studied domains of evidence for common deficits in ADHD and RD. Both ADHD
and RD children demonstrate deficits on verbal and spatial span tasks (Shaywitz, et al.,
1995, Siege] & Ryan, 1987; Cohen et al., 2000), on linguistic and semantic fluency tasks
(Felton, Wood, Brown, Campbell & Harter, 1987), and on confrontation and rapid
automatized naming tasks (Ackerman, Anhalt, Holcomb, & Dykman, 1986; Tannock,
Marinussen, & F rijters, 2000). However, results have varied considerably across studies.
For instance, verbal and nonverbal working memory and rapid naming impairments are
more often associated with RD (Siegal & Ryan, 1989; Cohen et al., 2000; Tirosh &
Cohen, 1998; F elton et al., 1987; August & Garfinkel, 1990, Nigg et al., 1998), and are
typically more severe than in ADHD -- especially for RD+ADHD (Shaywitz et al., 1995;
Willcutt et al., 2001; Cohen et al., 2000).

Age and type of task may provide some explanations for discrepant results across
studies. Deficits in children with ADHD and children with RD, for example, often begin
to appear as both task demands and requirements for controlled processing increase when
they begin elementary school. In addition, younger ADHD children are more likely to
demonstrate working memory impairments (Mariani & Barkley, 1997; Siegal & Ryan,
1989). Delayed development of the mechanisms underlying working memory (perhaps in
inhibition of competing information, which will be discussed later) could account for
these age-related differences. Given the possibility for development delays,

consideration of age is important in the study of ADHD+RD phenomenology.

22

Relationship to phonological processing and reading. In addition to the direct
findings that support working memory deficits in both ADHD and RD children, some
research supports a connection between working memory and phonological processing
abilities. ADHD children’s performance on phonological processing tasks becomes
impaired as the working memory demands of the task increase. Thus, whereas ADHD
children do not show impairment in nonword decoding, their performance on phoneme
segmentation and deletion tasks-- while not as impaired as in children with RD-- is
deficient versus control children in some studies, although not yet widely replicated
(Purvis & Tannock, 2000; Willcutt et al., 2001). Furthermore, Mariani and Barkley
(1997) found that preschool children with ADHD demonstrated deficient word
reading/decoding ability on a simple word-recognition task that loaded strongly on a
working memory factor. The authors proposed that early word recognition abilities may
depend on working memory in children with ADHD rather than phonological processing
factors. Working memory deficits in ADHD children might then serve as an early
cognitive vulnerability for development of RD.

Research in nonclinical groups has also shown that individual differences in
working memory predict reading ability (Daneman & Carpenter, 1980; Daneman &
Carpenter, 1983). Text comprehension can be predicted by a text’s working memory
demands (Britton & Gulgoz, 1991) as well as an individual’s working memory capacity
(Lorch & van den Broek, 1997). Cantor and Engle (1993) hypothesized that working
memory constraints rather than problems with lower-order linguistic processes accounted
for impaired inferential abilities in less skilled readers. In one study, adults with low

working memory spans were unable to incorporate probabilistic information to resolve

23

syntactic ambiguities while reading, despite adequate knowledge of that information
(Pearlrnutter & MacDonald, 1995). As discussed later, such language tasks may be
potentially useful in examining ADHD and RD.

Summary. Considerable evidence therefore supports the existence of working
memory impairments in both ADHD and in RD, as well as a possible connection between
working memory and phonologiCal processing/reading ability. Examination of language
impairments in the two disorders will provide further information regarding mechanisms
by which ADHD and RD may interact.

Language

Language deficits are a central component of RD. In addition to phonological
processing deficits, RD children demonstrate deficits in syntactic, semantic, and
pragmatic aspects of language processing (Purvis & Tannock, 1997; Tannock &
Schachar, 1996; Bender, 1996; Cohen, Vallance et al., 2000). Some RD-specific deficits
have included problems with referential cohesion (understanding and utilization of
pronouns, possessive adjectives, demonstratives, and comparatives), over-use of
ambiguous references in oral narratives (Roth, Spekman, & Fye, 1995; Purvis &
Tannock, 1997), and giving less task-relevant information to a listener under instructional
conditions (Spekman, 1981). Relative to children with ADHD, children with RD
demonstrate problems with semantic/receptive language and story recall that occur
despite application of effortful learning strategies (which are not used by children with
ADHD) (Purvis & Tannock, 1997; Tannock and Schachar, 1996; O’Neil & Douglas,
1991). Syntactic morphology and complex language deficits have also been reported to

be unique to RD (Javorsky, 1996). However, at least one other study has found syntactic

24

difficulties in ADHD children as well (Cohen, Davine, Horodezky, Lipsett, & Issacson,
1992)

Language deficits in children with ADHD that are independent of comorbid
learning disability status have also been documented. Findings typically support deficits
in the pragmatics of language, but not in the receptive/expressive domains of language,
which include phonological, syntactical, and semantic subsystems. Compared to children
with RD, for example, children with ADHD exhibit more extemalized and less
internalized self-talk and less effective use of language to control their behavior (Berk &
Landau, 1993; Berk and Potts, 1991). List-leaming and list-recall impairments are also
ADHD-specific, suggesting the appearance of deficits when organizational requirements
increase (F elton et al., 1987; Chang et al., 1999). ADHD children have also been found
to have problems adjusting language to the listener and for specific social or
communicative contexts (Laundau & Milich, 1988; Whalen et al., 1979) and to fail to
answer questions or comply with commands during parental interactions (Tarver-
Behring, Barkley, & Karlsson, 1985; Barkley, Cunningham, & Karlsson, 1983).

Pragmatics deficits are also pronounced (Humphries, Malone, Koltun, & Roberts,
1994; Cohen et al., 2000; Tannock, Purvis & Schachar, 1993; Tannock and Schachar,
1996). Westby and Cutler (1994) asserted that a number of DSM—IV ADHD symptoms
are consistent with pragmatic language deficits. Although these deficits are not evidence
of a pragmatic language deficit in and of themselves, symptoms include talking
excessively, difficulty awaiting turns, interrupting others, blurting out answers to
questions before they are finished being asked, and not listening. Other commonly

observed ADHD behaviors, such as switching topics abruptly, speaking unconnected

25

thoughts, and intruding into conversations at awkward moments were also theorized to be
consistent with pragmatic language impairment.

Summary. Executive function deficits, specifically in organization and self-
monitoring, have been hypothesized to produce the pragmatics language deficits in
children with ADHD. This interpretation is consistent with other research in ADHD that
has found executive functioning (including in the domains of planning, organization, and
motor output) as a major replicated and potentially primary element in the ADHD
syndrome (Barkley, 1997). The exact mechanism producing these language-related
executive function deficits has not been well explored, however. In addition, little
research has examined whether children with ADHD might demonstrate receptive
language deficits under increased requirements for executive or inhibitory control. At the
same time, there has been little exploration of whether RD-related deficits are related to
(or cause) problems in inhibitory functions.

Theories in the cognitive literature provide one possible means by which to
understand the relationship among phonological processing, working memory, and
executive dysfunction. Baddeley and Hitch (1994) hypothesized that working memory
holds words, sentences, numbers, or phrases by means of a “phonological loop” under the
control of a “central executive.” Denckla (1996) suggested that phonological processing
problems in RD lead to dysfunction in the phonological loop and, thus, a failure to
provide language to the central executive for the performance of executive functions. In
contrast, executive function impairments in ADHD produce an inability to efficiently
utilize and organize information from the phonological loop, leading to skill acquisition

problems that may affect the processes involved in learning to read. According to this

26

theory, then, executive control and language processes are inextricably linked. The
interface between language and executive/inhibitory processes therefore may be a
promising place in which to look for a shared deficit in ADHD and RD.
Conclusions about Neurocognitive Deficits

Empirical support for shared deficits in ADHD and RD in the auditory
processing, working memory and language domains suggest that there may be some
other, primary, shared impairment that produces secondary deficits in these domains. One
possibility is that inhibitory mechanisms play a role in the appropriate functioning of
working memory, auditory processing, and language and that impairments in such
mechanisms can produce secondary deficits in these domains. Recent findings that are
suggestive of inhibitory deficits in RD create further intrigue about whether weakness in
inhibitory mechanisms that help support language processing may be a potential Shared
mechanism underlying ADHD and RD comorbidity (Purvis & Tannock, 2000).
Examining the role played by inhibition in language processing may be especially helpful
for determining whether there is a shared inhibitory deficit in ADHD and RD, because
both of these impairments are central to one or the other disorder (i.e., language in RD,
inhibition in ADHD). Furthermore, although empirical findings in the cognitive literature
support the role of inhibitory mechanisms in language processing, for the most part this
idea has been largely unexplored in clinical samples.

In order to fully address the possible connection between ADHD and RD-related
inhibitory deficits and language impairments, the empirical literature on inhibitory
functioning in ADHD and RD will now be reviewed and will be followed by findings on

the relationship between inhibition and language processing. The cognitive literature

27

makes two important distinctions about types of inhibitory mechanisms -- effortful vs.
automatic inhibition and behavioral versus cognitive inhibition. These distinctions and
their relevance to ADHD and RD will now be reviewed.
Inhibition: New Candidate for a Shared Deficit?
Theoretical Background

Efforfiul versus Automatic Inhibitory Processes

Following established traditions in the literature, Nigg (2001) defined effortful
inhibitory control as the “top down” active suppression of a cognition or behavior in the
service of attaining a goal (as internally represented in working memory).
Neuroanatomical substrates hypothesized to underlie effortful inhibition include the
anterior cingulate as well as dorsolateral- and orbito- prefrontal cortex and associated
basal ganglia structures (Casey, 2001; Castellanos et al., 1996). Automatic inhibition has
been described as including attentional processes responsible for preconscious selection
of information that enters consciousness. Selection of context-appropriate meanings of
homographic words is cited as an example of automatic inhibitory control, because it
involves suppressionof previously activated cognitive contents that occurs without
apparent intention or conscious awareness and seems to happen automatically
(Hamishfeger, 1995). Inhibition in terms of attentional functioning and working memory
is both automatic (suppression of unattended locations) and deliberate (suppression of
competing distractors). These inhibitory functions have been well-described in relation
to visual-motor processes (Nigg, 2000), but less well described in language processes.

On one hand, inhibitory processes of one kind or another may distinguish ADHD

and RD. Some theories suggest that children with ADHD are likely to experience

28

failures in cognitive or behavioral processes requiring effortful or deliberate inhibitory
control, whereas children with RD demonstrate difficulties with automatic suppression
mechanisms (Ackerman, Anholm, Dykman et al., 1986; Ackerman, Anholt, Holcomb et
al., 1986). One study, for instance, found deficits on an automaticity/ language factor of
executive functioning to be associated with RD and deficits on an attention/maintenance
factor Of executive functioning to be associated with ADHD (Kemp & Kirk, 1993).
Furthermore, recent findings suggest that impaired rapid naming of colors (which
requires effortful semantic processing) is associated with ADHD, while impaired naming
of numbers (which requires more automatic processes) is associated with RD (Tannock,
Martinussen & Frijters, 2000; Nigg et al., 1998). Evidence that ADHD (at least the
combined type) is related to impairments in frontally-mediated neural mechanisms and
RD to temporoparietal neural deficits also supports the idea of effortful versus automatic
inhibition deficits as distinguishing the two disorders. The mechanisms underlying the
inattentive subtype of ADHD have received less investigation and are less clear. Due to
the nature of the attentional impairments exhibited by inattentive subtype children, some
evidence has suggested that, similar to RD, the inattentive subtype may involve deficits
in parietally-mediated or automatic attentional mechanisms (Schaughency & Hynd, 1989;
Goodyear & Hynd, 1992).

Although it may be possible to distinguish ADHD and RD according to
impairments in effortful versus automatic inhibitory processes, it is also important to
recognize that, just as distinctions between effortful and automatic processes are relative
(i.e., on a continuum) rather than absolute, the same may be the case with the inhibitory

deficits of children with ADHD and RD. In other words, if children with ADHD and

29

RD are vulnerable to an inhibitory deficit in language, then their effortful and automatic
inhibitory deficits are likely to be on a continuum that is a matter of degree rather than a
dichotomous deficit. According to this logic, ADHD and RD children should
demonstrate deficits in both automatic and effortful inhibitory processes, although they
may experience more severe deficits on one or the other end of the continuum (e. g., more
severe automatic inhibition deficits in RD and more severe effortful inhibition deficits in
ADHD).
Behavioral versus Cognitive Inhibition

Hamishfeger and Bjorkland (1993) defined cognitive inhibition as the
suppression of previously activated cognitive contents or processes and resistance to
interference from irrelevant information. Cognitive inhibition is responsible for keeping
irrelevant information from entering or being maintained in working memory and for
preventing disruption of cognitive processes by extraneous information. As just noted, it
can be intentional and conscious (i.e., effortful) or unintentional and unavailable to
consciousness (i.e., automatic) (Hamishfeger, 1995). Definitions of cognitive inhibitory
processes differ somewhat across studies and theories. Gemsbacher (1997), for instance,
distinguishes between cognitive inhibition, defined as the blocking of activation, and
suppression, defined as the dampening of existing activation. According to Gemsbacher,
interference control requires suppression, but not inhibition. The distinction between
suppression and inhibition is not crucial to the present study, so the terms will hereafter
be used interchangeably.

In contrast to cognitive inhibition, behavioral inhibition involves the intentional

control of behavior, i.e., motor output (Hamishfeger, 1995) and has been associated with

30

size of right prefrontal and basal ganglia structures. Casey (2001) theorized that
inhibitory projections of basal ganglia thalamocortical circuits produce inhibition of
behavior, whereas excitatory projections from frontal cortex maintain information or
representations upon which to act. Orbitofrontal and right anterior cingulate activation
has been associated with performance on a response execution (behavioral inhibition)
task (Casey, 2001). Cognitive and behavioral inhibition may interact, in that efficient
cognitive inhibition can facilitate behavioral control by suppressing inappropriate
thoughts or impulses. Because language-related processes like internalized speech (which
likely is moderated by a cognitive suppression component) are directly related to self-
control, deficiencies in cognitive inhibitory mechanisms that affect language could lead
to behavioral regulation problems like those in ADHD.

Inhibitory deficits, especially in relation to motor control or behavioral responses,
have been the focus of a number of prominent theories on the etiology of ADHD
impairments (Barkley, 1997; Quay, 1997; see Nigg, 2001 for review). Building on the
general theories of cortical functioning by Bronowski (1977) and F uster (1989), Barkley
(1997) proposed that the main deficit in ADHD stems fi'om failures in behavioral
inhibition. Functionally localized bilaterally to the prefrontal and motor cortex,
behavioral inhibition was proposed to create a delay in motor activity for further direction
from the executive functions. Failures in behavioral inhibition would then cause
executive function failures that are often manifested by difficulties in self-control and
goal directed behavior. Barkley proposed that working memory, self-regulation of affect-
motivation-arousal, internalization of speech (to be discussed later), and reconstitution

rely on behavioral inhibition.

31

Evidence for deficits in cognitive inhibition in ADHD are less well studied and
have mixed support (N igg, 2001). Casey (2001) found ADHD-related deficits on a
stimulus selection task requiring suppression of previously attended stimulus attributes
and proposed that these deficits may be associated with impairments in dorsolateral
prefrontal cortex.
Development of Cognitive and Behavioral Inhibitory Control

Response inhibition is sufficiently developed by 2"‘1 grade to detect individual
differences across children (Hamishfeger & Bjorkland, 1993). Relative delays in
development of frontal inhibitory mechanisms in 2nd graders would then cause a child’s
behavior to be more heavily influenced by prepotent responses than that of non-delayed
peers (Dempster, 1993; Barkley, 1997). With regards to development of cognitive
inhibition, Dempster (1993) asserted that resistance to interference by both external and
internal sources is a major factor in cognitive development and is associated with frontal
cortex development. Children may have inefficient inhibitory mechanisms that make
them vulnerable to interference by irrelevant information in working memory. As a
result, they utilize more mental resources in the service of information processing. As
children develop, increased efficiency of inhibitory mechanisms allows them to devote
more resources to storage and cognitive strategies (Hamishfeger, 1995; Hamishfeger &
Bjorkland, 1993).

Control over interference does not develop uniformly across all domains of
processing and can be said to develop from an “outward to inward” direction. For
instance, children initially develop the ability to suppress motoric interference, then later

perceptual interference, and, finally, develop ability to suppress linguistic interference

32

(Dempster, 1993). Linguistic interference is greater than motoric interference beginning
around 6 years of age (for instance, Stroop interference generally does not occur before
age 6), thus making it a key candidate for a shared deficit domain in RD and ADHD—
both of which are usually first diagnosed at around age 6. Hamishfeger (1995) reported
that inhibitory processes in negative priming do not appear until at least age 8, while
ability to ignore irrelevant information doesn’t peak until early adolescence.

Hamishfeger and Bjorkland (1993) proposed that children improve information
processing capacity and efficiency through the development of direct retrieval
mechanisms. These direct retrieval mechanisms are thought to develop by activation of
one stimulus and simultaneous inhibition of any other competing stimuli, leading to the
eventual development of processes such as direct lexical access. Failure of a stimulus to
reach the activation threshold for selection leads to reliance on a piece-meal processing
strategy (e.g., phonological decoding rather than direct lexical retrieval). One theory for
RD deficits posits delays or failures in the development of skill automatization (Moores
& Andrade, 2000). Cognitive inhibition impairments provide one potential explanation
for this deficit, in that inability to sufficiently suppress competing “noise” could lead to
failures in the development of automatic or direct retrieval mechanisms. Automatization
failures (i.e., the continued need for conscious allocation of effort for execution of a
routine task) could then create an increased reliance on effortful inhibitory processes for
skills that are normally automatic.

Using reading as an example, failure to develop automatic word decoding could
increase the processing requirements/effort needed to read a passage, thereby resulting in

limited resources available for other cognitive processes, like reading comprehension.

33

Such resource limitations would also then be likely to produce additional cognitive
deficits (Crain & Shankweiler, 1988; Shankweiler et al., 1999). Some support for this
assertion is found in the neuroimaging literature. Brunswick, McCrory, Price, Frith, &

F rith (1999) reported that college students with RD demonstrated reduced activation in
left posterior inferior cortex and increased activation in a pre-motor region of Broca’s
area while reading pseudowords, despite similar accuracy to non-impaired comparison
students. The authors attributed this result to problems with highly automated aspects of
reading (specifically, lexical retrieval) and an associated increased need for effortful
compensatory strategies.

To summarize, although speculative, if automatic cognitive inhibition deficits
cause automatization failures that reduce resources for higher level cognitive processes
(i.e., those that typically require greater effort), then performance on tasks that require
effortful inhibition will likely also be impaired. Such an explanation could provide a
rationale for the co-existence of both automatic and effortful inhibition impairments and,
consequently, the co-occurrence of both types of inhibition deficits in ADHD and RD.
Importantly, if children with RD have genuine core impairments in automatic inhibition,
then they would also be expected to demonstrate problems with nonverbal tasks requiring
inhibition. To further explore the role of inhibition in these disorders, discussion now
moves to review of empirical findings on inhibitory deficits of various kinds in ADHD
and RD. These findings can be broadly classified according to effortful/behavioral
inhibition findings and automatic/cognitive inhibition findings. These findings highlight
the relevance of studying inhibitory mechanisms as a means of clarifying the relationship

between ADHD and RD.

34

Empirical Findings on Inhibitory Mechanisms in ADHD and RD

Ejfortful Inhibitory Control and Motor Inhibition

There is considerable empirical support for the existence of an executive-type
inhibitory deficit in ADHD that impairs suppression of motor response, though it may or
may not be primary (i.e., may be secondary to impairments in arousal, activation, and
motivation; see Nigg, 2001 for review). Behavioral inhibition deficits, as measured by
Logan’s Stop task (Logan & Cowan, 1984), have been demonstrated in a number of
studies on children with ADHD. On this task, ADHD children require increased time to
stop a prepotent (primed, prepared, or already initiated) response (rapidly hitting “X” and
“0” keys) when a “stop signal” (an auditory beep) occurs (Schachar & Logan, 1990;
Oosterlaan, Logan, & Sergeant, 1998; Nigg, 1999a). Impaired performance on tasks
assessing motor inhibition (specifically, Go—No-Go types of tasks) in ADHD has been
associated with structural differences in orbitofrontal and right anterior cingulate brain
regions and may be indicative of abnormalities in frontalstriatal circuitry (Casey, 2001).

Although inhibitory deficits have been shown to be specific to ADHD in several
studies, they may not be unique to children with ADHD—milder inhibitory deficits may
occur in RD also. One study found deficits in RD children on Logan’s Stop task (Purvis
& Tannock, 2000). These findings were replicated in another recent study, but the effect
was not robust to covariance of IQ and ADHD symptoms simultaneously (Willcutt et al.,
2001). N igg (1999) reported a correlation between reading achievement and response
inhibition on the Stop task, even though covarying reading ability did not eliminate the
ADHD effect. In contrast to these supportive findings, no RD-related response inhibition

deficit (measured by commission errors) was found on a task that was similar to the Stop

35

task and required inhibition of response to green circles followed at varying intervals by a
red square (Moores & Andrade, 2000).

Studies using Go-No-Go types of tasks provide support for inhibitory deficits in
RD but only on tasks that have verbal or phonemic processing requirements, suggesting
that they may be confined to the language domain. Moores and Andrade (2000) found
that teenagers with RD demonstrated deficits (as measured by commission errors) on a
go-no-go type task that used digits, but evidence was less clear when nonverbal stimuli
were used (e. g., squiggles). On the nonverbal version of the task, teenagers with RD
made significantly more commission errors than comparison children only after reaction
time was covaried. Participants reportedly did not have ADHD, but inattention and
hyperactivity symptoms were not controlled. Willcutt et a1. (2001) found increased
commission errors for children with RD on a continuous performance task (requiring
response to a white “S” followed by a blue “T”); this finding was not robust to covariates,
however. In addition, Kupietz (1990) reported that children with RD demonstrated
sustained attention deficits and increased commission errors on a continuous performance
task that were characteristic of inattention or short-term memory deficits and decreased
with age. Purvis and Tannock (2000), on the other hand, found no evidence of a
response inhibition deficit for RD on the Conners Continuous Performance Test
(requiring responding to all letters except X). Possible alternative explanations for RD
findings on response inhibition tasks like the Stop, Go-No—Go, and Continuous
Performance tests include automaticity deficits (Moores & Andrade, 2000) or problems
processing two stimuli in rapid succession (Fitch, Miller, & Tallal, 1997; Purvis &

Tannock, 2000). For instance, failures to make certain functions (e. g., phonological

36

coding) automatic could lead to a requirement for more effortfirl/controlled processing of
stimuli that limits availability of resources for other tasks. Exploration of this association
using tasks other than Continuous Performance tests or the Stop task is sorely needed.

Summary. Deficits in effortful inhibitory motor control currently represent one of
the most promising domains for identification of a neuropsychological “marker” of
ADHD. Recent findings with regards to RD present a potential problem for theories
asserting that certain kinds of inhibitory deficits are unique to ADHD. Due to some of the
possible alternative explanations for deficits on the Stop Task, it and other commonly
used response inhibition and sustained attention paradigms (e.g., continuous performance
tests) will not resolve the issue of whether RD children demonstrate inhibitory deficits
like those in ADHD. Alternative methods are therefore necessary. One possibility is that
the two groups demonstrate a continuum of inhibitory deficits, such that the severity or
type of inhibitory deficit determines predisposition toward the development of one or
both disorders. The continuum between automatic and effortful inhibitory processes is
one interesting possibility. A continuum of language abilities which may interact with
dimensional inhibitory abilities is another such possibility.

Discussion now moves to empirical findings for automatic inhibition and
attentional deficits in ADHD and RD. Although not a focus of this study, execution of
attention relies, at least in part, on inhibitory processes. Since research in the RD domain
has focused more on attentional than inhibitory functioning, this literature may be useful
for understanding the relevance of studying inhibitory mechanisms in RD. Attentional

deficits in ADHD may also be relevant to relatively automatic inhibitory processes.

37

Automatic Inhibitory Control and Inhibition for Attention

Halperin and colleagues (1997, 1993) found evidence for two distinct groups of
children with ADHD. An ADHD+RD group had increased plasma levels of a
noradrenergic metabolite in comparison to an ADHD-only group. The authors
hypothesized that increased noradrenergic activity in the ADHD+RD group may be
indicative of dysfunction in the posterior attention system and regions of posterior cortex
and thalamus that are related to auditory reception and language processing. The ADHD-
only group was hypothesized to have deficits in the anterior attention system, leading to
executive function and inhibitory failures. A study by Conners and colleagues provided
support for this theory. Children classified as having automatic attention deficits on a
covert orienting task that required similarity judgements for pairs of letter arrays (i.e.,
deficits at the 100 ms interstimulus interval) did not demonstrate deficits on an executive
function or a vigilance task. In contrast, those classified as having a deficit in
voluntary/effortful attention on the covert orienting task (i.e., deficits at the 800 ms
interstimulus interval) performed more poorly only on the vigilance and executive
function tasks. Automatic attention deficits were therefore asserted to be associated with
impaired rapid processing of lexical arrays (unpublished study reported in Schulte,
Conners, & Osborne, 1999).

Studies also support attentional deficits in RD. Segalowitz, Wagner, & Menna
(1992) reported that poor readers’ Contingent Negative Variation attentional ERP,
collected during a vigilance task, was related to reading comprehension. This effect was
proposed to result from problems recruiting attentional resources to coordinate the

linguistic and graphemic processes necessary to learn to read. Immaturity of pathways in

38

the frontal lobe/limbic system was hypothesized as one potential mechanism underlying
this finding.

Dichotic listening studies in children with RD are also supportive of attentional
dysfunction. Stringer and Kershner (1993) reported that children with RD show the
typical right ear advantage for reporting dichotically presented digits orally, but show a
weaker right ear advantage for written reporting. The authors asserted that this finding is
consistent with a RD-related vulnerability to attentional interference, such that the
increased effort/attentional demands associated with a phoneme-to-grapheme processing
step interfered with the execution of other cognitive processes. Attentional interference
could, in turn, be prevented by an inhibitory mechanism. Children with RD have also
been found to demonstrate deficits in attentional shifting in dichotic listening (Kershner
& Graham, 1995; Kershner & Morton, 1990). Kershner and Graham (1995) speculated
that problems with attentional shifting could produce impairments in the ability to deal
with the rapidly changing cognitive requirements for reading. Importantly, although
children with RD did not have significantly lower IQ’s, their performance in these
dichotic listening studies mirrored that of reading achievement-matched controls,
suggesting the possibility of delayed development of the mechanism underlying
performance on this task. Inhibitory mechanisms that protect attention are still maturing
in young children (Dempster, 1993) and thus provide a plausible explanation for these
results.

Summary. Evidence in both the ADHD and RD literatures supports the existence
of deficits in attention or interference control. The possibility of deficits in an inhibitory

mechanism that protects attention in RD has been hypothesized and warrants

39

consideration. Impairments in cognitive inhibitory mechanisms could account for these
attentional deficits and would be consistent with problems with interference control.
These findings on attentional functioning in both disorders highlight the relevance of
studying inhibitory mechanisms, while also demonstrating the relative lack of knowledge
or specification of the exact nature of these hypothesized deficits.

Due to the well-documented occurrence of language impairment in RD,
examining language processes reliant on inhibition may be one useful area for attempting
to clarify whether there are inhibitory deficits in RD and may help explain the high
comorbidity between RD and ADHD. Furthermore, examining whether inhibitory
impairments in ADHD extend to the language domain may further explain the high rate
of co-occurrence between the two disorders. Linkage of ADHD with inhibitory
mechanisms during language processing has yet to be attempted. The next section
therefore examines these possibilities with regards to language production and language
comprehension. Literature that provides evidence for importance of inhibitory processes
in language processing will be reviewed first, followed by the limited studies in ADHD
and RD that are relevant to potential weaknesses in inhibitory mechanisms that protect

language processing.

Inhibition Mechanisms and Language Processing
Developmental Considerations
Dempster (1993) suggested that difficulty with text processing in young children
may be due to inefficient inhibitory mechanisms. Given enough time, older children will
not only have activated all meanings of an ambiguous word, but will also have inhibited

its nondominant meaning (as evidenced by slower responding to a nondominant

40

meaning) and focused attention on the dominant meaning. Children who are in 2nd and
4th grade, however, do not show suppression of the nondominant meaning of the
ambiguous word and continue to show facilitation of all meanings at even a 2000 msec
delay (Simpson & Foster, 1986; Simpson & Lorsbach, 1983). One possible reason for
these findings is that, though automatic spreading activation in lexical access is thought
to mature by age 4, control of the conscious allocation of attention doesn’t fully mature
until 5th or 6th grade.

Much theoretical (and some empirical) work has focused on the connection
between language development and development of inhibitory control. Verbal self-
regulation is thought to develop between the ages of 1.5 and 5.5 years of age
(Hamishfeger & Bjorkland, 1993) and internalized speech subsequently contributes to
behavioral self-control. Shelton & Barkley (1994) asserted that the prefrontal
impairments theorized to produce disinhibition, inattention, and overactivity in ADHD
may also contribute to deficits in syntax, fluency and verbal production. Riccio and
Jemison (1998) reported that language impairments mediated the association between
inattention and reading ability, thereby suggesting that inattention may have a direct
effect on language processing impairments. Although linguistic factors, such as
phonological processing, explained the most variance in reading ability in one study,
moderate-sized correlations between inattention and reading achievement (including
phonological processing and comprehension) were also found (Shaywitz, Fletcher, &
Shaywitz, 1994). The idea of language impairment as a common link between ADHD
and RD thus appears to be plausible, though it has not been directly tested. One

possibility is that impaired development of a central inhibitory mechanism, such as one

41

that might support effortful and strategic control of cognition and behavior generally,
creates a vulnerability for the development of language processing impairments and,
subsequently, reading disorders or the development of ADHD. The development of
internalized speech is one area in which hints at such processes are apparent.
Internalized Speech

Current theories of behavioral self-regulation focus on the importance of
internalized speech for controlling behavior (Barkley, 1997). Skinner (1953, as described
in Barkley, 1997) suggested that behavioral self-control develops through a language-
based process that begins with behavioral control by the language of others, proceeds to
control of behavior by private speech, and, finally, results in the use of internalized
speech for problem-solving. Berk and Potts (1991), furthermore, propose that the
development of private speech is reciprocally related to the development of self-control
and is moderated by inhibitory mechanisms. Inhibitory control is therefore crucial to the
development of internalized speech, which, in turn, leads to the development of greater
self-restraint and self-guidance as well as problem-solving and planning abilities (i.e.,
executive functions).

Inhibition deficits in ADHD may lead to failures in internalized speech
development that exacerbate problems with behavioral self-control. At the same time,
children with speech and language impairments (including RD) may be delayed in the
use of language to control their behavior, placing them at risk for the development of self
control problems, including poor ability to inhibit inappropriate behaviors (Riccio &

Hynd, 1993).

42

Children with ADHD have been found to demonstrate more private speech
(defined as audible talking not directed towards another person) than comparison children
during free play (Copeland, 1979). Furthermore, one study linked poorer response
inhibition in ADHD to apparent delays in internalized speech development. Thus, 90
percent of children with ADHD engaged in some kind of observable mediating behavior
(such as counting out loud or tapping a finger) during performance on a response
inhibition task versus only 45 percent of comparison children. In contrast, 80 percent of
comparison children reported using a cognitive mediator (e.g., counting in their head)
compared to only 30 percent of children with ADHD. Thus, findings suggested that
children with ADHD may have been unable to utilize internalized speech to control their
behavior (Gordon, 1979).

In addition to the relationship between internalized speech and behavioral control,
Damico, Damico, and Armstrong (1999) posited that internalized language also
influences the development of cognitive maps/schematas of the world that produce
complex thought and abstract thinking. ADHD and language disorders (along with RD),
consequently, may share deficits in metacognition and its linguistic or representational
underpinnings. Alternatively, as discussed earlier, deficits in executive/inhibition
systems may interfere with language use, thereby leading to risk of RD.

Evidence for the relationship between inhibitory mechanisms and language
processing will now be discussed in more detail.

Role of Behavioral Inhibition in Language Processing: Language Production

The majority of research on language processing in ADHD has focused on

language production. This research has found that ADHD children demonstrate excessive

43

verbal output in unstructured communicative interactions (Zentall, 1988), decreased
verbal output and increased verbal dysfluencies as language task organization
requirements increase (Zentall, 1988; Ludlow, Rapoport, Bassich, & Mikkelsen, 1980),
and problems maintaining and changing topics and in conversational tum-taking
(Tannock & Schachar, 1996). Unsurprisingly, in view of their well-known executive and
response output problems (Barkley, 1997), the narrative speech of ADHD children is
marked by disorganization, lack of information, poor cohesion, and greater inaccuracies
than the narratives of unimpaired comparison children (Hamlett, Pellegrini, & Connors,
1987). One study found that ADHD children used fewer descriptive instructions and
exhibited increased child-initiated and disruptive speech when required to instruct a
listener on the construction of simple block designs. Disruptive speech on this task was
manifested by increased interactive communication with the examiner (e.g., by asking
more questions and making more commands than necessary for task performance), but
was not manifested by increased non-task directed speech. ADHD children also
demonstrate problems with self-monitoring of language output. Purvis and Tannock
(1997) found significant deficits in event sequencing, original text misinterpretations, and
word substitutions on a story re-telling task in comparison to nondisordered control
subjects.
Summary

Considerable research supports the existence of inhibitory-related deficits in
language production in ADHD that seem to be related to already established problems
with executive control of motor and behavior output. What is less clear is whether

children with ADHD experience problems with language comprehension, a cognitive

44

processing realm more closely related to the demands of reading and the deficits in RD.
Just as little research has yet to explore whether inhibitory deficits in ADHD extend into
the area of cognitive inhibition, little is known about whether inhibitory deficits in
ADHD could produce impairments in language comprehension. Yet, the possibility
clearly exists—if inhibitory systems regulating language use are unitary, they may also
break down in comprehension. If so, this could provide a link to RD.

Research on the role of inhibitory mechanisms in language comprehension is
prominent in the cognitive literature, but, for the most part, has not been extended to
clinical research. Linking these bodies of work is thus quite timely. Examination of
inhibitory-related language comprehension mechanisms in ADHD and RD has two
powerful advantages. First, it provides sorely needed exploration of how inhibition
deficits in ADHD might extend beyond well-documented problems in the behavioral
domain to impact cognitive functioning. And, second, it may provide insights into the
etiology of ADHD+RD comorbidity. A focus on language comprehension thus deserves
further consideration.

Role of Cognitive Inhibition in Language Processing: Language Comprehension

Dempster (1993) asserted that text processing is highly reliant on interference
control. He described a number of text characteristics that can produce interference,
including unrelated statements about the same topic, involvement of the same set of
characters in different activities, ambiguous messages with more than one meaning, and
irrelevant or contradictory text information. Research on the relationship between
cognitive inhibition and language processes is prominent in the cognitive literature

(Gemsbacher & Faust, 1991; Faust & Gemsbacher, 1996; Dell & O’Seaghdha, 1994;

45

Eberhard, 1994; Dempster, 1993). Inhibitory mechanisms have been most well studied
in the domain of lexical ambiguity. Neill, Valdes, and Terry (1995) reported that
competition among various meanings of ambiguous words involves the same
mechanisms underlying stroop and negative priming effects. Lexical ambiguity is
proposed to create two incompatible mental representations that compete for control of
behavior. The context-appropriate meaning of the word is theorized to receive activation,
while the context-inappropriate meaning is suppressed. The result is the eventual
selection of the appropriate word meaning. Gemsbacher and colleagues performed a
number of lexical ambiguity studies that support the role of suppression in language
comprehension. Specifically, they found that response times to targets that were
contextually inappropriate meanings of ambiguous words were slower at a short delay
than at a long delay (Gemsbacher, Vamer, & Faust, 1990, Experiment 4).

This process is promising for better understanding deficits in poor readers.
Indeed, a relationship between weak interference control/cognitive suppression and
poorer reading comprehension has been demonstrated by a number of studies (Tompkins,
Baumgaertner, Lehman, & Fassbinder, 2000; Gemsbacher, Vamer & Faust, 1990;
Gemsbacher & Faust, 1991, Gemsbacher, 1997). Less skilled comprehenders have been
found to have poorer access to recently comprehended information and to generate too
many mental substructures during comprehension (Gemsbacher, Vamer, & Faust, 1990).
Gemsbacher and colleagues hypothesized that the creation of too many mental
substructures results from incorporation of irrelevant information into mental
representations of text. Lorch and van den Broek (1997) speculated that identifying and

creating representations of causal relationships is the core mechanism involved in

46

comprehension. Comprehension and recall of text is influenced by the number of causal
connections shared by events in the narrative and whether the event lies along a “causal
chain.” If impaired suppression mechanisms prevent the development of this causal
chain, comprehension deficits may be expected to follow. Supporting this notion, less
skilled readers demonstrated recall of fewer important story details (Purvis & Tannock,
1997) and deficits in their ability to draw inferences from text, despite intact sentence-
level structural representations and adequate knowledge about sentence content. They are
also unable to integrate information from different parts of a story or elaborate their
structural representations with topic-related information (Long, Oppy, & Seely, 1997,
1994). It was hypothesized that a bottleneck occurs in less skilled readers at the discourse
level of processing.

Findings of impaired suppression in the language domain have been documented
in older adults and adults with right hemisphere brain damage, both of whom are
hypothesized to experience weakened cognitive inhibitory control. In a series of studies
on individuals with right—hemisphere brain damage, Tompkins and colleagues found
evidence for a relationship between impaired comprehension and inhibitory dysfunction.
On Gemsbacher’s ambiguous word paradigm, right hemisphere brain damaged adults
failed to demonstrate the expected reduction in reaction time to context inappropriate
targets of ambiguous words from short- to long-delay probe intervals (Tompkins et a1,
2000). This relationship was specific to right-hemisphere brain damaged individuals
when compared to individuals with left hemisphere brain damage (Tompkins, 1991).
Individuals with right hemisphere brain damage also tend to maintain incorrect inferences

about information later negated in text, while simultaneously incorporating correct

47

inferences (Tompkins, Lehman, & Baumgaertner, 1999). Thus, discourse comprehension
deficits are particularly apparent under conditions requiring resolution of multiple or
competing interpretations. Similar findings have been reported in older adults (Hasher &
Zacks, 1988). Hamm (1992) found that older adults continued to incorporate corrected
misinformation into their inferences about a passage, despite also incorporating the
corrected information.

Although the performance of children with ADHD on such comprehension-
suppression tasks is unknown, there is reason to think that they may show weakened
suppression processes. For one thing, behavioral and cognitive deficits in ADHD have
often been compared to those of individuals with right hemisphere brain damage (e. g.,
Heihnan, Voeller, and Nadeau, 1991). As such, findings of impaired discourse
comprehension in right brain damaged individuals may provide a model for language
dysfunction in children with ADHD. Although very little research has examined either
text or language comprehension in ADHD, those findings that do support comprehension
deficits are consistent with possible inhibitory impairments.

Further, some studies suggest that ADHD children make less use of causal
connections in story narratives, leading to poorer overall story comprehension and
retention of minor story details at the expense of more important ones (Lorch et al.,
1999). Other studies, however, report no significant differences for ADHD children in the
recall of story narratives (Purvis & Tannock, 1997). Processing demands may influence
whether impairments in the mechanisms underlying reading comprehension are
manifested in comprehension impairments. For instance, children with ADHD

demonstrated impaired recall of television stories under conditions of divided attention

48

and derived even less benefit from causal connections (Lorch et al., 1999). Lorch and
colleagues (1999) hypothesized that such deficits may result from weakness in inhibitory
mechanisms, allowing interference with selection of important story units as a result of
failure to suppress irrelevant ones. Though one study found children with ADHD to have
poor reading comprehension despite intact word identification and phonological
processing skills (Brock & Knapp, 1996), another study indicates no significant
differences for ADHD children in the recall of story narratives (Purvis & Tannock,
1997)

In addition, the use of more ambiguous references by children with ADHD (as
well as by children with RD) on a story re-telling task has been posited to support deficits
in the construction of mental representations of comprehended material. These
ambiguities were speculated to support problems in the organizational cohesion of
information across sentences. These problems were found for the reproduction of the
stories (Purvis & Tannock, 1997). As a result, evidence seems to support the idea that
inhibitory deficits may create difficulties for developing mental representations that are
necessary for reading comprehension. Although findings of unimpaired comprehension
appeared in some studies of ADHD, the possibility exists that traditional methods of
assessing reading comprehension may not be sensitive enough to detect ADHD
impairments, because they do not specifically assess cognitive inhibition mechanisms.
“On-line” methods of measuring comprehension in children with ADHD may prove

useful for identifying more subtle effects of inhibitory deficits, if they exist.

49

Conclusion

Several distinct facts feed into the current research focus: (a) the proposed
relationship between internalized speech (a language function) and behavioral self-
control, (b) the well-documented incidence of inhibitory failures in response output in
ADHD, and (c) evidence for the importance of cognitive suppression mechanisms in
effective language comprehension, as well as (d) clear comprehension problems in RD,
and (e) possible comprehension problems in ADHD. Together, these facts suggest that
studying language comprehension functions that require inhibition may provide one
avenue for better understanding cognitive impairments in ADHD and RD. In particular,
little research has addressed the role of inhibitory functioning in language comprehension
in ADHD or, for that matter, in RD (despite all the studies using informal classifications
of “poor comprehenders”). This absence occurs despite considerable support for a
relationship between language comprehension and inhibitory functions in the cognitive
literature. Since comprehension is a cognitive rather than a production-oriented process,
the type of inhibition utilized in this domain may differ from that used in discourse
production and may be more related to interference control and cognitive suppression.
This direction departs from the usual and rather well-established output problem in
ADHD to look at an unstudied domain of ADHD cognition.

Delineating the primary language deficits in each disorder is important, because it
may reveal areas of “overlap” in the cognitive impairments (or “impairment profile”) in
each disorder and thus provide clues as to the etiology of their comorbidity. Which
deficits exist in either ADHD or RD or both could further define the nature of cognitive

impairments in the two disorders as well as provide an explanation for the high incidence

50

of comorbidity between them. In contrast, finding no indications of deficient cognitive
suppression in ADHD children would help narrow the scope of inquiry regarding the
nature of language impairments in ADHD. It would also provide further support for those
theories and empirical findings suggesting that ADHD is primarily a disorder of
behavioral (output) disinhibition—even in the domain of language. Or, alternatively, it
would help rule out the language domain as a possibility for shared deficits, leaving other
domains (e.g., auditory processing and working memory) for future study. Thus, studying
this domain in relation to ADHD is also important in its own right, regardless of RD.

The potential contributions of this study are as follows. 1) Studying the
performance of children with ADHD on inhibitory language tasks may help to further
clarify the nature of their language impairments and, by extension, clarify how self
regulation fails in that disorder. 2) Clearer delineation of the mechanisms by which
inhibition and language comprehension interact in ADHD may help to identify areas of
commonality to RD-related language impairments and thus may provide hints as to the
etiology of the frequent co-occurrence of the two disorders. 3) Examining whether
children with RD have impairments on inhibitory-reliant language tasks may provide

additional evidence as to whether there are inhibitory deficits in this disorder.

51

METHODS

Participants
Participants were 97 children between the ages of 7 and 14 (57 boys and 40 girls)

with a mean age of 10.2 i 1.2 years.I Sample characteristics are summarized in Table 1.

Table 1. Demographics Description: Means (SD).2

 

 

 

 

Control ADHD-Only RD- RD+ADHD
ADHD-C ADHD-I Sub- Only ADHD-C ADHD-1
Threshold/
Situational
N 27 28 7 14 13 4 3
Age 10.3 9.9 10.0 10.9 10.5 11.0 10.3
(1.4) (1.0) (1.0) (1.1) (1.3) (1.2) (1.2)
Sex
(% Male) 41.0 79.0 57.0 50.0 54.0 75.0 67.0
Ethnicity
(0/. White) 58.0 69.0 83.0 50.0 100 100 67.0
Estimated 107.0 100.4 98.1 107.2 103.5 98.3 102.7
1Q (16.5) (15.4) (13.9) (11.5) (16.0) (8.2) (30.1)

 

Children were community volunteers recruited through invitation letters sent to
parents of children in the regional schools and through newspaper and radio
advertisements. Children whom the school district had classified as ADHD or RD, as
well as regular education children, were identified by the school to receive invitation
letters without revealing their identity to the investigator. Children with neurological
impairments, other serious medical or psychiatric conditions, estimated Performance IQ
below 75 (or Full Scale IQ below 70), uncorrected visual or hearing impairments, and

those who did not speak English as their first language, were excluded from the study.

 

' The targeted age range was 9 and older; thus, only one 7 year-old and three 8 year-olds were included in
the total sample.

2 Note that 1 child not included in this table met criteria for the Hyperactive-subtype of ADHD. In
addition, 5 RD children met criteria for ADHD by parent report only (i.e., “situational ADHD”), but were
included in the pure RD group based upon their pattern of reading achievement, language, and learning
problems (to be further described below).

52

 

 

lilt

Note that 2 ADHD children did not meet IQ cut-off ‘s, so were excluded from all
cognitive task analyses.
Diagnostic Procedure

ADHD Classification

Diagnosis of ADHD was made by: 1) parent report on the Diagnostic Interview
Schedule of Childhood-IV (DISC-IV; Shaffer et a1, 1993), a structured diagnostic parent
interview assessing DSM-IV symptomatology of ADHD (and other DSM—IV disorders),
and 2) was supplemented by an “OR-algorithm” that followed the DSM-IV field trials
validity data (Lahey et al., 1994). This “OR” algorithm summed the total number of
parent-rated symptoms on the DISC-IV with non-overlapping teacher-rated DSM—IV
symptoms on the ADHD Rating Scale (DuPaul, 1998).

Even if full criteria (i.e., 6 inattention / 6 hyperactivity symptoms) were met by
parent report on the DISC-IV, final diagnosis also required teacher support (by exceeding
clinical cut—offs) on at least one common ADHD rating instrument (including the DuPaul
ADHD Rating Scale [DuPau1, 1998], short form of the Conners Rating Scale [Conners,
1997], and the Behavioral Assessment System for Children [Reynolds & Kamphaus,
1992])3. Symptoms also had to have persisted for more than six months, been present in
both home and school settings, appeared by seven years of age, and have produced
functional impairments (determined by DISC-IV impairment items). DISC-IV interviews
were conducted by trained interviewers during either an initial half-hour phone call

(which was then completed during the child’s visit) or during an initial two-hour office

 

3 Children with a prior ADHD diagnosis who were always stimulant medicated during school hours were
included without teacher questionnaire support if full criteria were met on the DISC-IV.

53

visit.4 Parents and teachers were always asked to make ratings according to the child’s
behavior when unmedicated (if possible). Thus, every effort was made to assure that
children met DSM-IV criteria for ADHD, including use of multiple informants and a
structured diagnostic interview.
RD Classification

Following the suggestion of Sattler (1992), RD diagnosis was made utilizing a
regression equation incorporating IQ and the mean of word recognition, spelling, and
phonological decoding achievement. The regression method is more accurate than a
“simple-discrepancy” method of diagnosing RD, because it accounts for regression to the
mean with higher and lower IQ’s. The equation used in the current study was a modified

version of that used by Pennington, Groisser, and Welsh (1993).

Reading Quotient6= (WIAT Reading SS+WIAT Spelling SS+Word Attack SS)/3
-- (Expected IQ for Age [100] + Estimated F SIQ)/2

RD was diagnosed for cut-off score’s less than 0.85 (i.e., mean achievement one standard
deviation below the average of actual IQ and population IQ average), which was more
conservative than Pennington et al.’s cut-off of 0.90. Note that, consistent with
Pennington et al’s procedure, this regression equation and cut-off score allowed some

high IQ individuals with reading scores within the average range to be included as RD.7

 

‘The interviewer was always blind to child performance on tasks, although not always to study hypotheses.
5 It may not be possible for interviewer to be blind to study hypotheses, but the interviewer will be blind to
child performance on tasks.

6 Use of Standard Scores (SS) rather than age-equivalents is the main difference between this equation and
that used by Pennington et a1.

7 2 RD children had reading scores in the average range.

54

Supplemental reading measures assessed reading comprehension skills and expressive
and receptive language (described below).

Classification of “Pure ” RD vs. RD+ADHD Children. One complication in
determining group classifications was the very high rate of subclinical attention problems
in many RD children. A number of these children (N=5) met criteria for ADHD-I by
mother report on the DISC, but teacher ratings did not support an ADHD diagnosis (i.e.,
these children had “situational” ADHD). These children were ultimately included in the
“pure RD” group based upon their pattern of reading achievement, language, and learning
problems, which suggested that they had more severe reading problems than did RD
children who did not meet criteria for situational ADHD. By this rationale, higher
mother-rated inattention may stem from these children’s increased academic struggles,
likely manifested by greater avoidance behavior around academic tasks (e.g., homework)
in the home setting. Consistent with this interpretation, RD children with situational
ADHD accounted for the higher incidence of ODD in RD children (refer to Table 4 in
Results section, p. 72).

The statistical significance of differences among pure-RD children with and
without situational ADHD was difficult to evaluate due to the small sample sizes (N=5
and N=8, respectively). However, means for these groups are reported in Table 33 in
Appendix A, along with a more detailed discussion of trends in this data (p. 171).
Non-Disordered Control Group C lassification

The comparison group met all inclusion criteria, but had 4 or fewer ADHD
symptoms in either domain by the “OR” algorithm (to exclude borderline ADHD cases)

and a Reading Quotient above 0.85. Incidence of oppositional and conduct problems and

55

anxiety and depression were also measured and children with any of these diagnoses
were excluded from the Control group. A number of children (N=l4) were
“subthreshold” for ADHD diagnosis (i.e., they were 1 or 2 symptoms short of diagnosis
for ADHD-I or ADHD-C or met criteria situationally only) and were consequently
excluded from group analyses. This subthreshold group was, however, included in
dimensional analyses.
Procedure

Children completed the research battery at Michigan State University during one
2.5 to 3 hour visit to campus. A portion of children completed this visit as the first of a 3-
visit, larger-scale, study of child ADHD. After parental consent was obtained, children
assented to the testing procedure and were administered the tests in a standardized order
(the key experimental tasks were administered in a randomly counter-balanced order).
Frequent breaks were interspersed within the testing session. Those children taking
stimulant medications were tested after a minimum of a 24- or 48-hour wash-out period
for short-acting and long-acting stimulant medications, respectively. Such a washout
period is generally considered adequate in the literature in view of the half-life of these
medications (on the order of 2-6 hours). Time since last medication dose was recorded
and was available for secondary analyses. Following the completion of the testing
session, children were paid $10 and were given some free-time to play computer games.
Mothers received $40 and were given written or oral feedback on the child’s test results.
Teachers were paid $15 for completion of behavioral ratings. Those families screened

out after completing the initial packet of behavioral ratings were paid $5, while families

56

screened out after the second screening stage (telephone DISC—IV interview or 2-hour

office visit) were paid $20 for participation to that point.

Measures
Descriptive and Background Measures

Intelligence

Wechsler Intelligence Scale for Children—III. A five-test short form of the
Wechsler Intelligence Scale for Children-III was used to estimate intelligence. This
short-form has a reliability of .95 and validity of .86 (Sattler, 1992), and consists of the
Vocabulary, Similarities, Block Design, Information, and Picture Completion subtests.
Achievement

Wechsler Individual Achievement Test (WIA T). Word recognition and spelling
were assessed using the Wechsler Individual Achievement Test (The Psychological
Corporation, 1992), which is co- normed with the WISC-III normative sample.8

Woodcock-Johnson Psycho-Educational Battery-Revised (WJ-R). The WJ-R
Word Attack and Passage Comprehension subtests (Woodcock & Johnson, 1989,1990)
were used to measure phonological decoding (i.e., nonword reading) and reading
comprehension.9

The WJ-R Passage Comprehension test assesses reading comprehension via
ability to identify a missing key word in the context of increasingly semantically and

syntactically complex passages (i.e., high cloze passages). This measure provided an

 

8 For the reading subtest, average split-half reliability for children between the ages of 5 and 17 (N=3899)
is .92.

9 Internal consistency (split-half reliability) of the Word Attack subtest ranges from .88 to .95 in children
between the ages of 6 and 18. Median reliability of the Passage Comprehension subtest for children
between ages 5 and 19 is .83.

57

index of general comprehension skill for comparison to experimental task performance.
Independence from reading comprehension ability would provide support for the
hypothesized inhibitory effects on experimental tasks. Secondary analyses examined
correlations between reading comprehension and experimental task performance
(regardless of diagnostic classification) to ascertain whether Gemsbacher and colleagues’
(2001) hypothesis about the relationship between comprehension and suppression skill
extends to a child population. In validity studies for the WJ-III (McGrew & Woodcock,
2001), children with ADHD had poorer comprehension on this subtest than non-impaired
comparison children.

Language

Clinical Evaluation of Language F undamentals-3 (CELF-3). Estimates of
receptive and expressive language ability were obtained from the Concepts and
Directions and Recalling Sentences subtests of the Clinical Evaluation of Language
Fundamentals-111(Semel, Wing, & Secord, 1995). The Concepts and Directions subtest
assesses ability to interpret and follow commands of increasing length and complexity
(e.g., “Point to the little triangles. Then point to the big circles. "). On the Recalling
Sentences subtest, children must repeat increasingly longer and more complex sentences
verbatim (e. g., "The catcher caught the ball and the crowd cheered loudly. ").

Both tests also have high working memory demands. The Recalling Sentences
subtest, because of its requirements for reconstituting sentences for verbatim oral
reproduction, was believed to have a particularly strong working memory requirement
(Barkley, 1997) and will therefore often be referred to as a measure of both expressive

language and working memory.

58

Experimental Tasks

The two experimental tasks used in this study were measures of language
comprehension hypothesized to challenge cognitive inhibition. The Ambiguous Words
Task served as a measure of interference control (relying upon both automatic and
effortful suppression mechanisms), whereas the Garden Path Task served as a measure of
ability to revise incorrect syntactic representations and their resultant misinterpretations
(believed to rely upon more effortful suppression mechanisms).

An important caveat with regards to most measures of inhibitory function also
applies to these tasks, however. For both of these tasks, alternative explanations can be
made for the mechanisms underlying task performance. For instance, decay of activation
or boosting of activation could also provide a reasonable explanation for task effects.
This debate remains unresolved in the cognitive literature (Dagenbach & Carr, 1994).
Given that it would be impossible to find any task that has a proven inhibitory
mechanism underlying performance, the current tasks were chosen because of their
hypothesized inhibitory processes and the fact that their data are consistent with such an
explanation.

Visual and Auditory Task Versions. WIAT word reading norms (The
Psychological Corporation, 1992) indicate that a child aged 10:0 (the target age of
children recruited for this study) who reads 1 ‘/2 standard deviations below average for
his/her age (SS=85) would be reading at a mid-second grade level. All stimuli were
therefore designed to be readable at an early second grade level. Stimuli were piloted and

tested for readability in six children from the ages of seven to ten10 to determine the

 

'0 Children were believed to have had average intelligence and average to below average reading abilities,
although this was not formally tested.

59

relative ease with which stimuli could be decoded. Stimuli were adjusted until easily
decoded and comprehended by seven-year-olds.

Contrary to initial expectations, most of the RD children included in the study had
reading levels below the second grade level, however. Since it was not possible to
administer auditory task versions to all subjects (many subjects had already been
collected on the visual version), an additional auditory version of both tasks was created.
Children who read below a second grade level (identified by a cut-off raw score of 25 on
the WIAT word recognition subtest) were then administered the auditory version of the
task, while all other children took the visual version of the task. The auditory task
versions utilized pre-recorded stimuli (to maintain standardized administration) that were
presented via computer in conjunction with the visual stimulus. The auditory task was in
all other ways analogous to its visual version.

Measure of Interference Control: The Ambiguous Words Task (Gemsbacher, Robertson,
& Werner, 2001)

The Ambiguous Words Task employed sentences containing homonyms to
measure interference control during lexical ambiguity resolution. Lexical ambiguity
studies that require plausibility judgments find increased reaction times and error rates to
homonym targets following their altemate-meaning primes (Simpson & Kang, 1994;
Gemsbacher, Robertson, & Werner, 2001). This relative “Cost” in processing different-
meaning targets has been argued to reflect functioning of automatic inhibition
mechanisms. In contrast, performance is facilitated when the homonym sentences
maintain the same meaning over the two presentations. Gemsbacher, Robertson, and

Werner (2001) reported that college-aged poor comprehenders demonstrated significantly

60

reduced “costs” (i.e., inhibitory deficits) versus good comprehenders when processing
such sentences and interpreted these differences to reflect impairments in inhibitory
control. A rough estimate of effect sizes produced with this measure for the Cost
comparison in average comprehenders is d=1.26 (f=.63) for RT and d= .98 (f=.49) for
accuracy. These are “large effects” according to Cohen, 1992.

Stimuli. Modeled on that of Gemsbacher, Robertson, and Werner (2001), the task
used in the current study consisted of 37 pairs of experimental sentences ending in a
homonym (12 “Different,” 12 “Same,” and 13 “Nonsense”) and 60 pairs of filler
sentences (also ending in a homonym) on which children were required to make a
meaning judgment regarding whether the sentence “makes sense.” The first sentence of
each pair acted as a prime and the second sentence as the target. For example, the target
sentence “She blew out the match” would be preceded by one of 3 prime sentences: a
same-meaning prime (e.g., “She lit the match”), a different-meaning prime (e.g., “She
won the match”), and a nonsense-meaning prime (e.g., “She sang the match”). Same-
meaning primes were designed to produce a benefit in reaction time or to decrease the
percentage of errors compared to nonsense-meaning primes. Different-meaning primes
were designed to produce a “cost” in reaction time or an increase in percentage of errors.
Given the “yes”-biased correct answers on both primes and targets in the Same and
Different conditions, filler sentences balanced out the required response patterns to
prevent children from developing response strategies. Filler sentences therefore
consisted of 12 “no-yes” items, 24 “yes-no” items, and 24 “no-no” items. Response

times and accuracies (recorded via computer) following nonsense-meaning prime

61

sentences (which were designed to bias neither homonym meaning) were compared to
both same-meaning and different—meaning primes.

In creating the stimuli, homonyms were selected from the University of Alberta
norms for homograph frequency published by Twilley, Dixon, Taylor, and Clark (1994),
as well as from stimuli used by Gemsbacher, Robertson, and Werner (2001). Stimuli
were created using homonyms that were unbalanced (i.e., the alternate meanings of the
homonym had unequal frequencies). Thus, one homonym meaning was “dominant” and
the other meaning was “subordinate”. In the current task, the target sentence aim used
the subordinate homonym meaning. “Different” condition stimuli therefore consisted of
a dominant-meaning prime sentence and a subordinate-meaning target. Stimuli in the
“Same” condition consisted of a subordinate-meaning prime and subordinate-meaning
target, while stimuli in the “Nonsense” condition consisted of a no-meaning prime and
subordinate-meaning target.

Based on findings from Gemsbacher and colleagues’ (2001) study, the rationale
for using subordinate targets was related to the desire to maximize potential “Costs” with
lower frequency words, while at the same time preserving the possibility of detecting
“Benefits”.n Several studies of homograph resolution have argued for functioning of
inhibitory mechanisms even under these conditions (Simpson & Adamopoulos, 2001).

Use of subordinate targets did, however, have some potential pitfalls with regard
to measuring inhibitory control. In particular, use of a dominant, high-frequency,
homonym in the prime sentence might produce very high levels of activation that would

carry into target processing (or make it unnecessary to suppress the context-inappropriate

 

” In Gemsbacher et al’s studies, the use of balanced homonyms produced somewhat smaller costs for the
Nonsense versus Different condition, whereas use of dominant targets did not produce benefits for the
Nonsense versus the Same condition.

62

subordinate meaning during prime processing). Slower responding or poorer accuracy to
targets could then also be interpreted as resulting from interference from the highly
activated dominant prime rather than to inhibition of the subordinate-meaning homonym
during prime processing. This potential confound in experimental manipulations will be
kept in mind when interpreting results. As will be addressed in the Discussion, even if
findings better match an activation-interference interpretation, it might be possible to
detect inhibitory functioning during target rather than prime access.

Task Construction. Words were selected for ease of reading, and all sentences
were designed to be appropriate for children reading at a second grade level. Stimuli
were initially piloted on 6 children who were normal readers (ages 7-10, grades 2—4) to
determine appropriateness of stimuli difficulty, and a number of stimuli were then
modified prior to finalization of the task.

When experimental stimuli were created, sentence pairs for each condition (i.e.,
Same, Nonsense, Different) were constructed for every homonym and were then
randomly assigned to one of three lists. Ultimately, however, one list was chosen for
administration to all children, such that all children received exactly the same items
(although their order of presentation was randomized). This approach prevented
examination of specific stimulus item effects across all subjects. However, despite this
task limitation, using one list was ultimately preferred to maximize statistical power in
this small sample.

Procedure. The task procedure was as follows: To ensure that children
understood the concept of a “nonsense” sentence, task instructions were presented with

training and feedback on how to determine whether a sentence made sense. Task

63

instructions can be found in Appendix C (p. 215). Stimuli were presented via computer
and children were required to press a “yes” or “no” key to indicate their meaning
judgment. Prime sentences appeared in their entirety following a brief (750 ms) “+” cue
and remained on the screen until the child made a response. The sentence was then
replaced with another brief “+” cue, followed by presentation of the target sentence. '2
Children first completed a practice block of 5 sentence pairs and were offered accuracy
feedback after each sentence (this feedback was given only for practice sentences).
Experimental sentences were presented in 5 blocks of 20 pairs (with the exception of the
final block which had 17 pairs), and children were offered a break after each block.
Entire task performance took approximately 20 minutes.
Measure of Revision: Garden-Path Sentences Task (Christianson, Hollingworth,
Halliwell, & Ferreira, 2001)

The Garden Path Task was used to examine ability to revise misinterpretations
arising from a misleading structural (syntactic) ambiguity and to arrive at a correct
sentence interpretation during language comprehension (both written and spoken). The
effect of semantic plausibility and syntactic structure on the ease with which
misinterpretations could be revised was also examined. Very little research to date has
examined sentence processing in children. Trueswell, Sekerina, Hill and Logrip (1999)
examined resolution of temporary syntactic ambiguities in kindergarten-aged children by
examining eye movements during response to spoken instructions. They found that these
children showed extreme resistance to revising incorrect interpretations of temporarily
ambiguous sentences and hypothesized that resource limitations (which may rely on

inhibitory mechanisms) were responsible for this pattern of performance. Thus,

 

'2 On the auditory task, sentences were presented both visually and auditorily with the same procedure.

64

inhibition failures would be manifested by persistence of misinterpretations despite cues
regarding their incorrectness.

Stimuli. A full list of task stimuli can be found in Appendix D (p. 217).
Methodology was based upon that of Christianson, Hollingworth, Halliwell, and F erreira
(2001) but was simplified for use in a child population. Experimental stimuli consisted of
28 experimental garden path (subordinate-matrix) sentences and 28 control non-garden
path (matrix-subordinate) sentences. Conditions varied along two dimensions, one of

which was semantic and the other of which was syntactic.

Table 2. Example Sentences by Condition for the Garden Path Task.

 

Condition Item Example
. As Joe walked the do na ed.
Plausrble Q: Did Joe walk thegdogrzlp
As the ape ate the tree grew.
Q: Did the ape eat the tree?
As Betty woke up Daddy was sleeping in the den.
Q: Did Betty wake up Daddy?
As the man shaved the cat sat in the tree.
Q: Did the man shave the cat?
The water cooked as Deb drank.
Q: Did Deb drink the water?
The worm swam in the lake as Jack ate.
Q: Did Jack eat the worm?
Plausible The baby was in bed as the boy washed.
Q: Did the boy wash the baby?
RAT , The kids swam in the pool as Frank dried off.
Implausible Q: Did Frank dry off the kids?

As Daddy made the food the kids played in the den.
Q: Did Daddy make the tape?

The boys took their pens as they left for jail.
Q: Did the boys leave for jail?

As Pam was feeding the fish she saw a UFO.
Q: Did Pam see a UFO?

OT: Optional Transitive; RAT=Reflexive Absolute Transitive

 

 

 

OT
Implausible

 

Plausible

 

Garden Path

Implausible

 

Plausible
OT

 

Implausible

 

 

Non-Garden Path

 

Filler

 

 

 

 

 

 

65

The semantic dimension manipulated “plausibility” of garden path interpretations.
Implausible condition sentences were those in which 1) the subordinate verb and noun
phrase had an implausible relationship (i.e., “ate the tree “shaved the cat ”), app 2) the
resulting garden path misinterpretation was itself also implausible: (e. g., “As the ape ate
the tree grew. "). In contrast, Plausible condition sentences had 1) a plausible relationship
between the subordinate verb and noun phrase (e.g., “ walked the dog”, “woke up
Daddy ”), but 2) a resulting garden path misinterpretation that was still implausible (e. g.,
“As Joe walked the dog napped ”—In this sentence, the garden path interpretation is still
implausible because Joe can’t walk the dog and have it nap simultaneously). Note that for
both plausible and implausible conditions, the matrix clause itself i_s_ plausible (i.e., a dog
napping; a tree growing).

Conditions also varied syntactically based upon Optional Transitive (OT) and
Reflexive Absolute Transitive (RAT) verb types. RAT verb conditions differed
syntactically from OT verb conditions in that RAT verbs are understood reflexively (even
without an object). Thus, if a RAT garden path sentence is parsed correctly, it can’t be
inferentially linked to an unspecified object as an OT verb can. For example, the RAJ:
sentence “As the man shaved the cat sat in the tree, ” if parsed correctly, cannot produce
an inference that the man shaved the cat; he must shave himseu In contrast, in the IE
sentence, “As the ape ate the tree grew, ” one might still infer that the ape ate the tree due
to the inference that the ape must be eating something and thus was likely eating the
banana. Since RAT verbs can also retain their transitive structure without a direct object
(or can utilize their implicit reflexive object—cg, himself/herself), garden path

sentences utilizing RAT verbs do not need to be completely syntactically revised to allow

66

for semantic reinterpretation or verb-noun phrase thematic role re-assignment. Put
another way, RAT verbs are not as resistant to reinterpretation—perhaps because the verb
and object are not as strongly linked (from an activation perspective, for example, the
causal connection linking a RAT verb and object may be weaker—perhaps 'due to
competition from an alternate implicit object [e.g., herself/himselfl).

There were also 36 filler sentences comprised of both subordinate-matrix and
matrix-subordinate sentences that differed from garden path sentences by the addition of
an extra noun phrase. These filler sentences also differed according to whether the
comprehension questions probed: a) the object of the subordinate clause or the subject of
the matrix clause (e.g., in a sentence such as As Pam was feeding the fish she saw a UFO
the question might be: 1) Did Pam feed the fish? or 2) Did Pam see a UF 0?), and b) a
“true” subject or object of the sentence or a “foil” (e.g., Did Pam feed the lion?). Finally,
filler sentences also varied according to their pragmatic plausibility, with some sentences
requiring a “yes” answer to implausible relationships (e. g., The mom played the socks as
she watched the news. Q: Did the Mom play the socks?)

Task Construction. Items were piloted on 6 children (not the same children on
which the Ambiguous Words Task were piloted) prior to task finalization and modified
as necessary. As in the Ambiguous Words Task, two item lists were initially created by
randomly assigning garden path sentences and their non-garden path pair to one list or the
other. Two maximize statistical power, one list was then chosen for administration to all
children.

Procedure. Children viewed sentences on a computer screen and answered

yes/no comprehension questions about the sentence. Task instructions are presented in

67

Appendix C (p. 215). Prior to task administration, children were coached on
distinguishing pragmatic inferences from literal sentence interpretation and encouraged to
make sure they interpreted sentences literally (e. g., to answer “no” to The water cooked
as Deb drank since the sentence does not directly state that she has).

Sentences appeared on the computer screen in their entirety. Children controlled
the presentation rate and pressed a key to advance to the comprehension question.l3 Total
administration time was approximately 20-30 minutes, with opportunity for breaks
interspersed throughout. Errors served as the primary variable of interest. Inhibitory
deficits on this task would be manifested by increased errors versus comparison children.
Since sentence length among conditions was not controlled, RT’s were not used in
analyses.

Plan of A nalysis

Data for the Ambiguous Word and Garden Path tasks were analyzed separately in
two stages: 1) Dimensional analyses (both correlation and regression) exploring the
relationship of reading, attention, and hyperactivity to task performance, and 2)
Categorical Analyses of group effects using analyses of varience to compare 1) ADHD
and Control groups, 2) RD and Control groups, and 3) ADHD and RD groups. Details of
these analytic comparisons will be presented in more detail within the Results section.

Results were always re-checked after separately covarying for missing data
replacement, IQ, age, and symptoms of other commonly co-occuning disorders (e. g.,
oppositional defiant and conduct disorder). Exploration of sex effects was examined by

including sex as a between subjects factor. Using the procedure of many ADHD+RD

 

'3 On the auditory task, children did not control presentation duration of the sentence but had unlimited
time to answer the comprehension question. It is therefore possible that the memory demands of the task
were higher for these children.

68

comorbidity studies (Pennington, Groisser & Welsh, 1993, Willcutt et al., 2001, Nigg et
al., 1998), descriptive characteristics of the comorbid ADHD+RD group (e.g., language
and reading achievement) were addressed by re-running analyses using a 2x2 design
(ADI-ID present, RD present).

To increase power to detect differences in this small sample, composite indexes
were used to examine task effects where possible. Overall, effect sizes for the various
experimental manipulations in this study were similar to those reported in the literature
for executive functioning tasks (typically d=.60 or eta2 = .08 or larger) comparing ADHD
versus Control children (Pennington & Ozonoff, 1996). Thus, the utility of these
cognitive tests for examining functioning in ADHD children is equivalent to those of
traditional executive function measures. With the sample size and analytic design used
here, power to detect effects in all key comparisons was above 0.80.

Although the large number of analyses used in this study somewhat increased the
risk of a Type-1 error, it is proposed that correcting for number of analyses by adjusting
the significance level (e.g., with a modified Bonferoni correction setting alpha = .01)
would not be helpful. Due to the study’s small sample size, adjusting the significance
level would lower power and thus increase Type-2 errors. Furthermore, specific
predictions are defined a priori in accordance with theoretical proposals, arguably
reducing the need to control for chance findings. Therefore, it was decided that the
significance level not be altered from alpha = .05, although the possibility of Type-l

errors will be considered in task interpretations.

69

Data Reduction and Data Cleaning

Missing data was imputed using the estimation maximization (EM) algorithm,
which is one form of maximum likelihood estimation (Shafer & Graham, 2002). This
imputation method is believed to provide better estimates of missing data than mean
substitution or regression-based replacement methods. Across all variables (cognitive
tasks as well as parent and teacher behavioral ratings), 5.8 % of data was missing. To
compensate for the potential effects of estimation, a missing data variable was created
and covaried in all subsequent analyses as recommended by Cohen and Cohen (1983),
enabling the effects of substitution to be both assessed and removed. Covarying for the

effect of missing data did not affect the results.

Experimental Hypotheses and Predictions
Hypothesis:
If children with ADHD have impairments in cognitive inhibition then they would be
likely to experience problems in higher-order language comprehension (e. g., drawing
inferences from text and revising or updating text interpretations). Furthermore, if
cognitive inhibition impairments contribute to the language processing problems of
children with RD, then they should also demonstrate impairments in higher-order
language comprehension. RD and ADHD should therefore be associated with similar
patterns of impairment in higher-order language processing that will provide clues as to

whether they share a common underlying neurocognitive deficit.

70

Predictions:

1) ADHD children will demonstrate deficits on both the Garden-Path and Ambiguous
Words tasks. Likewise, higher levels of inattention and hyperactivity will be
dimensionally associated with poorer performance on these tasks. .

a) ADHD group differences or dimensional associations between inattention/
hyperactivity and performance on the Ambiguous Words Task would reflect interference
control problems suggestive of impairment in more automatic cognitive inhibition.

b) ADHD group differences or dimensional associations between
inattention/hyperactivity on the Garden Path Task would reflect impairments in
correction of misinformation and in the more effortful aspects of cognitive inhibition.

c) If ADHD diagnosis and inattention/hyperactivity are not associated with such
deficits, results would be consistent with the idea that ADHD is simply a motor
inhibition/output deficit and does not involve problems with either cognitive inhibition or
language comprehension.

2) RD children and dimensional measures of reading ability will also be associated with
deficits on the Ambiguous Words and Garden-Path tasks. RD group differences will
reflect: a) lower-order language impairments as well as b) higher-order inhibition deficits.
Support for higher-order inhibition deficits will be reflected in failure of language or
reading impairments to entirely account for group differences.

ADHD and RD children will show both similar and different patterns of deficits
on these tasks. RD children will show patterns of performance suggestive of both lower-
order and higher-order language processing deficits whereas ADHD children will have

only higher-order processing problems.

71

RESULTS-Part I
Note about the Organization of Sections
Extensive amounts of data are reported in the following section. To orient the
reader to the results, the general structure of this section follows. The first section
describes sample characteristics in some detail, due to their direct bearing on later data
interpretation regarding ADHD and RD (and ADHD+RD comorbidity). The second and
third sections report results for the Ambiguous Words and Garden Path tasks,
respectively. These sections are organized according to: 1) an exploration of task effects
(e.g., whether the task worked as expected), 2) dimensional analyses (both correlational
and regression) that address hypotheses concerning the relationship of inhibition to
attention problems and reading, and 3) group analyses (ANOVA’S) that address
hypotheses regarding whether ADHD and RD have overlapping cognitive deficits. To
streamline and focus results, findings peripheral to the primary hypotheses were placed in
Appendix A as background. The reader will be referred to the Appendix for further
information when appropriate.
Sample Description
Demographics
Sample characteristics are in Table 3 (below). Group differences on
demographics variables were examined for key diagnostic groups (i.e., Control, ADHD-
C, ADHD-I, pure RD), using 3x2 univariate ANOVA’s with ADHD group (Control,
ADHD-C, ADHD-I) and RD group (RD yes, no) serving as between-subjects variables.
RD group and ADHD group were used as separate, between subjects, factors to examine

comorbid ADHD+RD effects in a preliminary manner. Post-hoe analyses compared

72

discrete groups (rather than RD and ADHD group interactions) and significant findings

are detailed via superscripts as explained in the table footnotes.

Table 3. Demographics Description: Means (SD).’4

 

 

 

 

Control ADHD-Only RD- RD+ADHD
ADHD-C ADHD-1 Sub— Only ADHD-C ADHD-1
Threshold/
W
N 27 28 7 14 13 4 3
Age 10.3 9.9 10.0 10.9 10.5 11.0 10.3
(1.4) (1.0) (1.0) (1.1) (1.3) (1.2) (1.2)
Sex 8 AB A
(“A Male) 41 79 57 so 54 75 67
Ethnicity C D DC
(% White) 58 69 83 so 100 100 67
Estimated 107.0E 100.4E 98.1 107.2 103.5 98.3 102.7
[Q (16.5) (15.4) (13.9) (1 1.5) (16.0) (8.2) (30.1)

 

Agar, 41) = 2.62, p=.05; ”x2 (1, 55) = 8.20, p<.01; C12 (1, 41) = 2.62, p<.01; ”x20, 38) = 4.68, p<.05;
F(1,54)=2.38,p=.13

Overall, demographics analyses indicated prototypical ADHD and RD samples
and supported diagnostic classifications. Groups did not differ in age. There were more
boys in the ADHD-C group than in the control or pure RD groups, while the RD group
had less ethnic minority representation than the Control and ADHD-C groups. Contrary
to frequent findings in ADHD samples, the IQ of the ADHD children in this sample
(either ADHD-C or ADHD-I) was not significantly lower than that of Control children.
Likewise, the IQ’s of RD children also did not significantly differ from Control or

ADHD groups.

 

'4 Note that 1 child not included in this table met criteria for the Hyperactive-subtype of ADHD. In
addition, 5 RD children met criteria for ADHD by parent report only (i.e., “situational ADHD”), but were

included in the pure RD group based upon their pattern of reading achievement, language, and learning
problems (to be further described below).

73

Behavioral Characteristics
Detailed information regarding group differences and statistical analyses
concerning behavioral indices of inattention, hyperactivity, and other externalizing and
internalizing behaviors (by both maternal and teacher ratings) can be found in Appendix
A (p. 171). A brief overview of these results is provided here and is summarized in Table

4.

Table 4. lnattention, Hyperactivity, ODD, and CD by group.

 

 

Control ADHD-Gnu RD- RD+ADHD
N=27 Combined Inattentive m ADHD-C ADHD-1
N=28 N=7 N= 1 3 N=4 N=3
Total Inattention* 0.6 ABC 85"" 7.9BE 4.9CDIE 9.0 8.0
Symptoms (1.0) (0.8) (1 . 1) (3.3) (0.0) (1.7)
(0R Algorithm)
Total Hyperactivity* 0.8 "F 8.1 "CD 3.0 9” 1.2C 8.3 3.3
Symptoms (1 .2) (1.2) (2.9) (1.4) (1.5) (1.2)
(OR Algorithm)
DISC 0.6 8.0 6.9 4.6 8.3 6.3
Inattention Sympts.* (1 .0) (1.3) (1.7) (3.5) (1.0) (2.5)
DISC 0.7 6.7 2.4 0.9 6.8 2.3
Hyperactive (1.1) (2.2) (3.2) (1.0) (2.2)
Sympts.* (0.6)
ODD (%) 4.0“ G 60.7A 28.6E 30.0G 75.0 33.3
(N=l) (N=l7) (N=2) (N=3) (N=3) (N=l)
CD (%) 4.0 10.7 14.3 0.0 25.0 0.0
(N=l) (N=3) (N=1) IN=01 (N= 1) (N=0L

 

NOTE: Matching letters within a row are significantly different from each other.
A p<.001; B p<.001; C p<.001; D p<.001; E p<.05; Fp<.01; G p<.10, " Maximum number of symptoms=9.

As commonly found in ADHD samples, children with ADHD had higher levels of
behavioral symptoms across a range of indices, including depression, anxiety, and
oppositional and aggressive behavior. Also consistent with some prior studies, children
with RD had higher levels of inattention, depression, and anxiety than did Control

children. Findings of high levels of inattention in both ADHD and RD children [ADHD-

74

RD group interaction: F (2,76)=10.74, p<.001] provided preliminary support for

behavioral phenotypic overlap between the two groups, while distinctions in terms of

behavioral hyperactivity supported that these were clearly non-identical groups.
Achievement, Language, and Learning

Patterns of performance by group for reading achievement, receptive and
expressive language ability (CELF-3: Concepts and Directions and Recalling Sentences
subtests), and developmental and current learning problems are shown in Table 5 and
briefly summarized here. More detailed description can be found in Appendix A (p.
171)

Findings here again were consistent with prior empirical findings and supported
the diagnostic classifications. As expected, children with RD (regardless of comorbid
ADHD status) had poorer performance across a range of achievement and language tasks
and were rated by parents and teachers as having both developmental and current
learning problems. Children with ADHD did not have weaknesses in basic reading skills
(e.g., word recognition, phonological decoding, or spelling). However, as often found in
ADHD samples, there were some indicators of achievement difficulties, particularly in
reading comprehension and parent/teacher ratings of developmental and current
reading/leaming problems.’5 Both RD and ADHD children therefore demonstrated some
form of reading achievement or learning problems, yet were readily distinguished by the
RD group’s greater severity and phonological decoding weakness. Note that qualitative
analysis of means for the RD+ADHD children suggested that their deficits were not

additive——that is, the comorbid group did not have more severe language deficits than did

 

’5 Thus, children with ADHD-C had poorer reading comprehension than Control children, children with
ADHD-I had more developmental reading problems, and both ADHD groups were rated as having greater
current learning problems than Control children.

75

the “pure RD” group. Overall, the preliminary sample characteristics suggested that
groups were validly defined and set the stage for testing of hypotheses with the cognitive

tasks.

76

Sovee 88.8.. .838 e .886 .886 wove e wove e
.528 soao EB: “sang: 3:52:6me o8 32 a 555 88.2 32082 05.02

Amt: ”28¢ emanate“

 

 

 

 

 

8.: an 8.8 2 8.: SH 8.: Z 8.: on 8.: M: Gm:

938.: aha—58.850

8.: 8.: 8.: 03 he 8.: g e 8.: SN 8.: «.2 a 8.: a: 832$ 2:. messes.

Aoueowrb

8.2: 8.? 8.: nae 88.: 8.8 U8.8: E 18.: 4% 88.: Se use—gem mafia..— om<m .388...

8.:: 3: A: : 8.2 8.:: 28 8.2: 3- 8.:: no- 8.:: 2. 8: 853285 86832859

9233. P8? :5. .95.on .9533:

8.2: 8.8 8.: 3: 2.: 8.2 3.2: 8.2: 8.:: SS 88.: 3.2 8568234 8263:

A933. 3.33

8.: 8.8 8.: 3 3.8.: 3 8.: 8.8 e 8.: S: e 8.: ~.: 883:8 mam—.33. 338

A983. 3.33

8.: oh 8.: 3 8.: no 8.: 8.2 8.: 2: 8.: 2: 828er 23 82.8.80 mid—u

3.33 3.6.35

8.: 8.3 8.: new as 8.2: New e8.2: 3.2 e .88: 32 abs: 2.: 8858....an omens— ~33

8.38 952.33

8.8: on 8.: new at 88: ME 5. 8.8: :8 d 88: 22 e 8.:: e82 eased. 29$ :55

A983 FEE—83

8.2: 8.: 8.: of. E... 8.8: 3: 68.: 8.8 .88: 38 h 8.:: 8.8” ”5:28 385

A988 Eat-.83

8.2: Sm 8.: SK ea 8.: 4.8 e:.m: an: e8.:: 32 88; : 5.2 nausea 55$

:8: 5.2 8.: $8 88: an: 8.2: :8 8.2: 4.82 88: 8.:: 9 28m :5— 885%....—

mnz fz Ruz £12

:53 0.9:? anz 822.335 8:555 Ruz
angina N1:51.12 i.=o-a=n< _eeeeu

 

 

96% 3 382 ”manhood was .ommanmq 4582633“. .m 2an

77

Ambiguous Words Task
Description of Conditions
As described in the Methods, the Ambiguous Words Task consisted of 3 primary
conditions: 1) Same (the meaning of the ambiguous word in both the prime and target
sentences was the same, less frequent/subordinate meaning), 2) Different (the meaning
of the ambiguous word in the prime sentence was the dominant meaning and in the target
sentence was the subordinate meaning), and 3) Nonsense (the meaning of the ambiguous
word in the prime sentence had no specific meaning—i.e., the sentence had a “nonsense”
meaning that didn’t prime either the dominant or subordinate meaning of the word—
while the target sentence contained the subordinate meaning). The following scores were
then derived: “Cost” (i.e., interference) and “Benefit” (i.e., facilitation) as described

below. Examples of the stimuli in each condition are shown in Table 6.

Table 6. Example Sentences and Condition Error Rates Across Groups.

 

 

 

 

 

 

 

_Condition W W
Vis./Aud.
s Prime She threw out the pit. 15% / 20%
ame
Target She spit out the pit. 13% / 14%
. Prime She stole from the bank. 12% / 9%
Different
Target She fished from the bank. 37% / 42%
Prime She hoped the drill. 14%/ 16%
Nonsense . .
Target She marched dunng the drill. 16% / 20%

 

 

 

 

 

 

Comparison of the Visual and Auditory Task
Mean accuracies across groups for the visual and auditory versions of the task as

well as for the entire sample (across task versions) are reported in Table 7 (page 83). In

78

general, the pattern of effects suggested that the experimental manipulations worked as
expected for both task versions; see Tables 6 and 7). However, the tasks did appear to
differ qualitatively (but not statistically significantly) in the magnitude of errors, with
somewhat more errors (more interference) in the auditory version, along with lack of a
RT interference effect (in contrast to the visual task). Since it was not possible to
determine whether these findings were true task differences (i.e., systematic differences
in the visual versus auditory task) or trends resulting from diagnostic group differences
(most of the children administered the auditory task had an RD diagnosis), auditory and
visual task data were analyzed separately. Due to the very small size of the group
receiving the auditory task, more emphasis will be placed on visual task results.
Task Validation

Appendix A (p. 171) presents extensive background on task characteristics and
validations, including an item analysis, inter-condition correlations for accuracy and RT,
direct accuracy—RT correlations, age—effects (across groups) and corrections, and
confirmation of basic task effects (i.e., interference effects across groups).

Overall, findings supported the effectiveness of experimental manipulations in
each condition and the expected Cost (i.e., interference or inhibition) and Benefit (i.e.,
facilitation) effects for the visual as well as auditory versions of the Ambiguous Words
Task. Accuracy data produced the clearest findings. However, RT data were also
moderately supportive—especially with regard to interference effects on the visual task.

Despite the overall effectiveness of the key experimental manipulations, two
' minor issues and one substantial issue in the data must be noted.16 First, diagnostic

groups had different responses to the Nonsense condition. To handle this confound,

 

'6 For more detailed discussion of these task inconsistencies, see Appendix A (p. 190).

79

secondary analyses compared group differences on interference in the Different condition
only (rather than relative to a Nonsense condition baseline). Second, RT “Costs,”
although representing relatively slowed responding to Different condition targets (versus
other conditions), were not actually “absolute” Costs (i.e., target RT was actually
facilitated vs. prime RT). The relevance of this finding will be addressed in the
Discussion. Third, and most important, differences among conditions in prime
accuracy/RT had to be statistically corrected. That issue merits more explanation here.

The issue here was that, contrary to expectations, prime accuracy and prime RT
for the entire sample were n_ot equivalent among the experimental conditions (Same,
Nonsense, Different) (Table 7).‘7 Differences among conditions on prime accuracy and
RT presented a potential problem, because the relevant “Cost” and “Benefit” composite
scores were initially calculated (Gemsbacher, Robertson, & Werner, 2001) from
performance on target sentences only—in effect assuming prime sentences to be
equivalent across conditions.

This issue was handled by creating “corrected” difference scores as outcome
variables, in a two-step process as follows. First, an “accuracy-corrected” and “RT-
corrected” score was created for each experimental condition by taking the difference
between primes and targets (e.g., “Corrected Same Accuracy” = ame Tagget

Accuracy-_S_ame Prime Accuracy). Second, using these scores as a basis, composite Cost

 

’7 Results held when the visual task was examined in isolation, when Control children were examined in
isolation [Same-Different primes: t( l ,33)=2.23, p<.04], and when the auditory task was examined in
isolation. The difference between Nonsense and Difference primes became statistically significant when
the analysis was constrained to the performance of Control children [t(l,33)=2.21, p<.04].

80

and Benefit scores were calculated by combining the conditions separately for accuracy

and RT data as follows”:

a) Cost (i.e., Interference) = Nonsense (Corrected Accuracy or Corrected RT)
- Different (Corrected Accuracy or Corrected RT)
b) Benefit (i.e., Facilitation) = Same (Corrected Accuracy or Corrected RT) —

Nonsense (Corrected Accuracy or Corrected RT).

The results that follow deal entirely with the Cost index, a measure of interference
control or lack thereof. Because facilitation effects were not the focus here, those data
and analyses can be found in Appendix B. They did not alter interpretations of the Cost
variable analyses that follow. All analyses that refer to “Costs” refer to the Corrected
difference scores iust described.l9 Higher scores mean greater Costs or more
interference.

Summary on Task Validity

Despite task inconsistencies, these problems were felt to be relatively minor with
regard to overall task effects and could be accommodated by statistical corrections or
post-hoc interpretations. However, given the presence of these issues compared to
Gemsbacher and colleagues’ (2001) findings, it seems possible that the task is measuring

somewhat different processes in children than it did in their adult sample. These

 

18 Note that in these composite scores the Nonsense condition serves as a theoretical “no-priming”
baseline.

’9 Note that corrected accuracy Cost scores were slightly higher than those obtained when comparing
accuracy for targets alone (see Table 3-5).

81

differences will be addressed further in the context of data interpretations in the
Discussion.
Preliminary Analyses of Inhibition Effects across Groups

Preliminary analyses indicated that the experimental manipulations of interest
worked as expected, producing Costs (i.e., interference/inhibition effects) in this child
sample (see Table 7) that were confirmed by paired samples t-tests [Different vs.
Nonsense condition: Accuracy: t(96)=-11.12, p<.001; RT: t(82)=5.82, p<.001]. The small
group of children administered the auditory version of the task had a non-significantly
higher accuracy Cost compared to those who received the visual version of the task
[F(1,95)=0.8I].

In Gemsbacher, Robertson, and Wemer’s (2001) college sample, participants
incurred a 9% accuracy cost and 166 ms RT cost on this task. This contrasts with a
greater than 20% accuracy cost and 700 ms RT cost in the current. child sample (see
Table 7). Current findings therefore suggested that interference control was far weaker in
children than in adults. For the auditory task, however, RT did not reliably predict

008118.20

 

20 On the auditory task, children incurred no significant RT cost for the Nonsense compared to the Different
condition [t(l,l3)=-0.10; change=~48 ms] or from prime to target in the Different condition alone
[t(l,l3)=0.95].

82

 

 

 

 

 

88:33” 2,835.3. 5.88.8 238.8 .583
84.3: 2.5. 8.8:: v.8. 8:: 88 dd: 2.8 has:
teach—con: $809300 Quench—con: flown-OLLOU
08E. :ezoaom mow—53‘

 

 

 

 

 

 

 

89:2: ”many 25:. .8323“ can 22.—cue. EUPCeOéD can “.83.—.80 imam. 2:35 28: 5:5.

NOVQ m ._oo.Vm oJOOWQ o .o _ .VQ m . Son < .350 some :5: EEfltw bEScEwfi Ba .38 8 £53, BEE chBNE ”mHOZ

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

82: a 88. _ em: 83%: 88.8: 83 5 8N3:
8.3: 8.32 9.. _ E 828; 8.3% 888% .5
8:: 88.8: 8:: 82: 8:: 8::
82 h 8.8 88 8.8 3.8 63.8 8.2.52
512: 88:6:
8m. _ R : 8822 : 8:8: 83%? 88.83 : 838:
82mm 2.3% 8.82 83.3% 3.38 62.83 am
8:: 8:: 8:: 8:: 8:: 8::
$8 .._ $8 8.8 88 8.8 has 82334
8812: .555
88.2: : 8:89 : 8888: 8883? 8:2 : 82mm:
88mm 8.3% 828“” 888m 8288 8.2% em
8:: 8:: 8:: 8:: 8:: 8::
$8 a. $8 8.8 8 £8 $8 :88 8.282
8612: 295m .35
law slid mam slid mam 25.8
E232: 8:852 25%

 

 

 

 

 

 

.m:o_meo> xmfl.
ESE=< can ._m=£> .295m :38. no: 8:80 mmos< 25H cozomom can 523884 8.:. 8:32 xwfl. 3.83 msosmEE< s 638%

83

Brief Recap of Hypotheses

Analysis now moves to specific hypothesis tests. Hypotheses involved both
dimensional and categorical predictions. In brief recap of these predictions, poorer
reading/language ability and higher inattention were both predicted to be related to
weaker interference control/cognitive inhibition on the Ambiguous Words Task.
Likewise, ADHD and RD children were both predicted to show poorer interference
control than Control children. Note that when results refer to gecuracv or RT “Costs”,
these costs often will be referred to as interference. As used here, interference is defined
more broadlv as__indicative of rekitive slowing in RT or decreases in acgurgcv (rather than
necessafrily referring specifically to competition between two activated homonym
meanings)?

Note about Plan of Analysis

Data for the Ambiguous Words task is hereafier analyzed in 3 parts: 1) visual task
data for non-RD children (i.e,. Control and ADHD children: N=75), 2) visual task data
for RD children (N=8), and 3) auditory task data for RD children (N=12). The reasons
for this plan of analysis will become clearer as subsequent analyses and findings are
presented. Briefly, however, in addition to findings presented above, correlational
analyses revealed that patterns of results on the auditory and visual versions of the task
were not equivalent. 22 To capture these differences, the data for RD children were

analyzed separately on both task versions. Due to the very small sample sizes in these

 

2‘ In other words, it would be possible for the term interference to be used to refer to a finding in which a
meaning was less active (i.e., inhibited/ less competitive), but in which processing costs were reflected in
pzerformance. See further discussion of this interpretation in my Discussion section later.

Comparisons of RD children who took the visual versus auditory tasks revealed dramatically different
patterns of results on the two tasks.

84

comparisons, as well as the clear confounds related to administration of multiple task
versions, analyses and results for RD children are necessarily viewed as preliminary.

In the results that follow, dimensional (correlational and regression) analyses are
presented first, followed by analyses based on diagnostic group (ANOVA’s). Regression
analyses were used to determine whether indices of achievement versus inattention
uniquely or additively predicted interference control/inhibitory ability (defined as Cost on
the Ambiguous Words Task) when controlling for effects of other variables such as
estimated IQ and age. These effects were also controlled in ANCOVA analyses.

Test of Interference Control I: ADHD/Control Comparisons
Dimensional Analyses for ADHD and Control Children on the Visual Task

Inattention, Hyperactivity, and Learning. Correlation and regression analyses
examined the relationship between parent and teacher ratings of inattention,
hyperactivity, learning and accuracy/reaction time Cost (Tables 8 and 9 below).
Consistent with hypotheses, greater inattention was predictive of weaker interference
mpg! in terms of RT Cost across a range of indices (See Table 8).23 Consequently, a
high inattention symptom score was associated with more competition among homonym
meanings and slower response to a context-appropriate, altemate-meaning, target. This
pattern suggests failures in interference control, since successful suppression of a
competing meaning should allow the context-appropriate meaning to become available
more quickly. Controlling for age, IQ, and reading effects did not alter the overall pattern

(Table 11, page 91). Findings for teacher ratings were particularly robust.

 

23 Total inattention symptoms (by the “OR” algorithm) predicted RT Cost with marginal significance
(p_<.O7, [3:21)

85

Table 8. Correlations for Accuracy and RT Cost Scores across Groups
ADHD/Control, Visual Only, N=75

 

 

 

 

 

 

 

 

 

 

 

 

Cost
(Nonsense-Different)
Accuracy Reaction Time

Total Inattentive Symptoms:c -.010 .212+
Total Hyperactivity Symptoms: -.008 .182
Mother-Ratings

ARS Inattentive Raw .040 .136
ARS Hyperactivity Raw .058 .147
Teacher-Ratings

ARS Inattentive Raw .036 .273*
ARS Hyperactivity Raw .025 .150
BASC Attention .218+ .227*

 

 

 

 

 

"p<.01; * p<.05 (2-tailed); + p_<_.08 (2-tailed); larger correlations=greater Costs
1 Mother+Teacher symptoms by the “OR algorithm”

These effects failed to emerge for the accuracy Cost composite (Table 8). As a
secondary check of hypotheses, correlation and regression analyses individually
examined the Nonsense and Different conditions (Table 9). Supporting hypotheses of
weaknesses in interference control/suppression, attention problems were a marginally
significant predictor of weaker interference control in the Different [E(1,73)=3.31, p<.08,
R2=.04, [3:21; not robust to iq/age control] and Nonsense condition [E(1,71)=4.05,
p<.05, AR2 =.05, [3:24 after iq/age control]. Thus, indices of inattention were associated
with greater interference (i.e., higher error rates/Cost) when accessing a previously
context-irrelevant competing meaning of an ambiguous word (i.e., subordinate target
following dominant-meaning prime). Inattention was also associated with increased cost
(i.e., greater interference) to subordinate targets following no-meaning primes (in the

Nonsense condition).

86

Table 9. Correlations for Prime-Target Changes in Accuracy and RT for Non-RD
Children in the Different and Nonsense Conditions of the Visual Task (N=75)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Different Condition Nonsense Condition
Cost
(Prime-Target) grime-Target)
Accuracy Reaction Accuracy Reaction
Time Time
Total Inattentive SymptomsI .208+ .024 .211+ .216+
Total Hyperactivity SymptomsI -.l39 .181 .142 .044
Mother-Ratings
ARS Inattentive Raw .240* .053 .182 .l 11
ARS Hyperactivity Raw .237* .221+ .158 .044
Teacher-Ratings
ARS Inattentive Raw .185 .105 .161 .200+
ARS Hyperactivity Raw .1 l 1 .119 .064 .075
BASC Attention .300“ .1 18 .030 .150

 

”p<.01; * p<.05 (2-tailed); + p_<_.08 (2-tailed); larger correlations=greater Costs; I Mother+Teacher

symptoms by the “OR algorithm”

Consistent with hypotheses, then, slower RT as well as worse accuracy was

associated with higher levels of inattention. Thus, despite slowing down, those with high

levels of inattention still had poorer accuracy.

Secondajv checks on validity of findings. To ensure that these correlational

findings did not simply result from decreased task effort that could result from

inattention, additional accuracy/RT correlational analyses were conducted for the

individual conditions (i.e., Nonsense, Same, Different; See Table 38, p. 191, in Appendix

A for the relevant correlations and inattention indices). Analyses revealed significant

correlations between accuracy and indices of inattention for both primes and targets

across most conditions (r’s from -.26 to -.40), but no significant correlations with RT. .

 

The dissociation between accuracy and RT correlations with inattention suggested
' that children with high levels of inattention had more difficulty with this task as a whole,
but that these difficulties were not associated with impulsive responding or decreased

effort. Such an interpretation was supported by failure to find significant correlations (or

87

even close to significant correlations) between inattention and RT on global task
performance—i.e., when including RT on incorrect as well as correct responses.
Consistent with the validity of the primary hypothesis test, then, these secondary
checks suggested that findings related to inattention on this task represent valid cognitive
impairments rather than artifacts of speed-accuracy trade-offs or decreased task effort.24
Note also that accuracy findings were always reported as relative differences (i.e., change I

from prime to target), so that globally poorer accuracy differences should not affect

 

interpretability of findings.

Achievement and Language. Correlations and regressions for measures of
achievement, IQ, language and accuracy/RT Cost (i.e., interference) are shown in Tables
10 and 11. Contrary to hypotheses, reading achievement was not related to interference in
lexical ambiguity resolution (as measured by either accuracy or RT Cost or individually
for the Different and Nonsense conditions) for these non-reading impaired children.
Thus, across ADHD and Control groups, for those receiving the visual version of the task
(N =75), correlations between measures of reading achievement and accuracy or RT Cost
(i.e., magnitude of interference) were all non-significant (See Table 10).

Findings were robust to age and IQ controls (Table 11 and 12, p. 91-92).
However, regression analyses revealed a marginally significant interaction between mean
reading achievement and inattention symptoms (with IQ controlled; F (1,70)=3.58, p=.06;

R2A=.04), such that better reading ability was associated with more interference/higher

 

' 24 One exception to this fm’ding was a significant correlation between hyperactivity and both prime accuracy
(r=-.24, p=.04) and RT (r=-.4l, p<.02) in the Same condition (i.e., subordinate prime) as well as hyperactivity
and target accuracy (r=-.28, p<.02) and target RT (r=-.42, p<.02). Thus, it is possible that impulsive
responding and decreased effort contributed to poorer performance when primes were subordinate-meaning
homonyms, but these findings are not relevant to hypotheses regarding interference control (although they
would be applicable to questions regarding facilitation effects—not the current focus).

88

accuracy Cost (Mean Reading Score: r=.38, p<.10; Word Reading: 5:139, p<.05) in
Control children but less interference in ADHD children (—.28, p=.12). This finding thus
suggests that inattention may moderate the relationship between reading and interference

control.

Table 10. IQ, Achievement, and Language Correlations for Accuracy and RT Cost
across Non-RD Groups on the Visual Task (N=75).

 

 

 

 

 

 

 

 

 

 

 

 

Accuracy Reaction Time
Cost Cost
Age .060 .134
Full Scale IQ (Estimated) -.277** .030
Nonverbal 10 (Estimated) -.277** .002
Mean Reading Score -.026 .104
Passage Comprehension (SS) -.149 .009
Concepts and Directions (53) .049 .021
Recalling Sentences (ss) -.238* .097

 

*p<.05 (2-tailed); ** p<.02 (2-tailed); + p<.10 (2-tailed)

Working Memory and IQ, In children without reading impairments, better
working memory capacity and higher intelligence were associated with better
interference control, i.e., with more success in selecting context—appropriate meanings of
homonyms.25 These effects provided encouraging validation of the tasks as measures of
the relevant higher-order interference control construct intended. Given findings in the
empirical literature concerning the relationship between working memory capacity and
reading comprehension ability, this finding was of particular interest.

Specifically, weaker expressive language/working memory (on the Recalling

Sentences test) was associated with greater interference (i.e., larger accuracy Cost but not

 

25 1f intelligence is thought of as relating in some respects to information processing capacity and
efficiency, this finding for intelligence seems appropriate. The empirical literature also indicates that

working memory capacity is an important factor in reading comprehension (to be discussed later in more
detail).

89

RT Cost). Conversely, then, larger working memory capacity was predictive of reduced
interference from competing homonym meanings.26 Although this effect was no longer
significant after controlling for IQ, IQ accounted for only approximately half of this
variable’s effect ([3: .12 after IQ controlled; IQ-Recalling Sentences correlation: p=.58,
p<.001; Table 12, pg. 92).27 RT was not related to interference control on this measure.
Summary. Key findings from dimensional analyses can be summarized as
follows: 1) Higher inattentive (but not hyperactive) ADHD symptom score was
associated with more interference (support of hypothesis); 2) Reading achievement was
related to interference only after accounting for the moderating relationship of attention
(variation on hypothesis or partial support); 3) Smaller working memory capacity (or

weaker expressive language) was associated with more interference (support of

hypothesis 1.

 

26 Note that total inattention symptoms was not significantly correlated with performance of Recalling
Sentences (_r:=-.06), suggesting that working memory capacity and inattention are independent predictors or
interference control.
‘ 27 In addition, while reading achievement and Recalling Sentences performance were moderately correlated
(_r=.43, p<.001), controlling for reading ability did not influence this test’s predictive power (see Table 12,
pg 92), providing fiirther indication that performance on the Recalling Sentences test may depend on
factors in addition to language ability (i.e., working memory capacity) and that these independent factors,
not language ability per se—e.g., working memory capacity» might account for the prediction of
interference control with this measure.

90

Table 11. RT Cost Regressions for ADHD and Control Children on the Visual

Ambiguous Words Task (N=75)

 

 

 

 

 

 

mane K Mg“. 4. E 11* I
_R_ R2 for R2 A
1 Mean Achievement .01 -.00 .01 0.79 (1,73) .10 0.89
2
Step 1 Full Scale IQ (estimated) .00 -.01 .00 0.07 (1,73) -.1 1 -0.83
Step 2 Age (Months) .02 -.01 .02 1.54 (1,72) -.12 -0.92
Step 3 Mean Achievement .03 -.01 .01 0.66 (1,71) .12 0.81
1 Total Inattention Sympts. .05 .03 .05 3.44 (1,73) " .21 135*
2
Step 1 Full Scale IQ (estimated) .00 -.01 .00 0.07 (1,73) -.03 -0.26
Step 2 Age (months) .02 -.01 .02 1.54 (1,72) -.l l -0.95
Step 3 Total Inattention Sympts. .06 .02 .04 2.65 LL71) .19 1.63
1 Teacher Inattention .07 .06 .07 5.86 (1,73) ‘ .27 2.421
(TARS)
2
Step 1 Full Scale [Q .00 -.01 .00 0.07 (1,73) .01 0.08
Step 2 Age .02 -.01 .02 1.54 (1,72) -.10 -0.85
Step 3 Teacher Inattention .09 .05 .06 4.90 (1,71) ‘ .26 2.21‘
(TARS)
1
Step 1 Mean Achievement .06 .02 .06 1.61 (3,71) .15 1.13
Total lnattention Sympts. .22 1.90+
Full Scale IQ (estimated) -.07 -0.50
Step 2 Interaction .07 .02 .01 0.73 (1,70) .10 0.85

(Achiev.*Inatt.)

‘13 after final step; "p _<_ .001; *pg .05; + p308; positive B’s=larger Costs.

91

Table 12. Accuracy Cost Regressions for ADHD and Control Children on the Visual
Ambiguous Words Task (N=75)

 

 

 

 

 

 

rename a2 as} a E if :r.
.13. R2 For R2 A
1 Mean Achievement .00 -.01 .00 0.05 (1,73) -.03 -0.22
2
Step 1 Full Scale IQ (estimated) .08 .06 .08 6.04 (1,73) . -.34 -2.66"
Step 2 Age (Months) .08 .05 .00 0.00 (1,72) .04 0.36
Step 3 Mean Achievement .09 .06 .02 1.35 (1,71) .16 1.16
1 Total Inattention Symptoms .00 -.01 .00 0.01 (1,73) -.01 —0.09
2
Step 1 Full Scale IQ (estimated) .08 .06 .08 6.04 (1,73) ‘ -.29 -2.41.
Step 2 Age (Months) .08 .05 .00 0.00 (1,72) -.01 -0.08
Step 3 Total Inattention Symptoms .08 .04 .00 0.17 (1,71L -.05 -0.42
1
Step 1 Mean Achievement .09 .06 .09 2.50 (3,71) + .16 1.25
Total Inattention Symptoms -.07 0.57
Full Scale IQ (estimated) -.38 4.95“
Step 2 Interaction (Achiev.*Inatt.) .14 .09 .04 3.58 (1,70)+ -.21 -1.89+
1 Recalling Sentences .06 .04 .06 4.40 (1,73) ‘ -.24 .2.10‘
2
Step 2 Full Scale IQ .09 .06 .01 0.73 (1,72) -.21 -1.51
Recalling Sentences -.12 -0.86
3
Step 2 Mean Achievement .06 .04 .06 4.87 (1,72) ' .09 0.74
Recalling Sentences -.28 -2.21‘.

 

I B after final step in the model; ** p_<_ .01; *p<.05; + p507

Categorical Analyses for ADHD and Control Children on the Visual Task

To evaluate hypotheses regarding group differences (apart from dimensional

associations on the variables of interest) a limited number of analyses of variance were

performed. AN OVA’s combined ADHD-C and ADHD-I children into a single group

(because most correlation and regression results were related to inattention and the

inattentive-only group was small) and compared this group to a group of Control

. children.28

 

28 Findings were checked for the ADHD-C group when ADHD-1 children were excluded and results
remained unchanged in all cases.

92

Accuracy. The 2 (Group: ADHD vs. Control) x 2 (Sex: Boys vs. Girls) ANOVA
for accuracy Cost was significant [F(1,55)=9.25, p=.004, n2 =.144; Figure 1], such that
girls with ADHD demonstrated much more interference (higher Costs) from competing
homonym meanings [mean: 0.46 i 0.18; F(1,23)=9.25, p=.006, n2 =.287] than ADHD
boys (mean: 0.18 _-_+-_ 0.18) or Control girls or boys (means: 0.24 i 0.17 and 0.26 i 0.17,
respectively). This effect was robust to age and IQ covariance and also held when
performance was examined in the Different condition alone [F(1,55)=12.14, p=.001, n2

=.l8].

Figure 1. Group by Sex Interaction for Accuracy Cost in ADHD vs. Control

 

 

   

 

 

 

 

0.55 a
0.5 1
4;) 0.45 '1
:3
>1 0.4 '
o
a:
S 035-
8 ' —-0-—Girls
<2: 0 3 d —I 'Boys
g .
:3
o
.2 -
2 o 5 \ \
0.2 ¢ \
0.15 -
0.1

 

Control - ADI-II)

93

Table 13. Group Means for Accuracy and RT on the Visual Ambiguous Words Task

 

 

 

Control ADHD-C ADHD-Only
+ ADHDJ: ADHDJ
ADHDJ

Accuracy
Same

Prime 0.91 (0.1) 0.80 (0.2) 0.82 (0.2) 0.74 (0.2)

Target 0.95 (0.1) 0.82 (0.1) 0.82 (0.1) 0.80 (0.1)
Nonsense

Prime 0.85 (0.2) 0.86 (0.1) 0.87 (0.1) 0.83 (0.1)

Target 0.89 (0.1) 0.82 (0.1) 0.81 (0.2) 0.82 (0.1)
Different

Prhne 093(0J) 087(02) 085(02) 092(01)

Target 0.73 (0.1) 0.57 (0.2) 0.56 (0.2) 0.61 (0.2)
Filler

Prime 0.92 (0.1) 0.88 (0.1) 0.87 (0.1) 0.89 (0.1)

Target 0.89 (0.1) 0.85 (0.1) 0.84 (0.1) 0.90 (0.1)
Reaction Time
Same

Prime 3601.0 (1018.2) 4075.0 (1368.5) 3982.7 (1339.9) 4417.8 (1527.6)

Target 2659.9 (812.0) 3189.9 (1248.6) 2940.5 (966.0) 41 16.1 (1776.6)
Nonsense

Prhne 39865(13969) 38106(12258) 38291(12996) 37420(9846)

Target 2769.6 (911.8) 3135.4 (1161.0) 3038.9 (1049.8) 3493.7 (1551.6)
Different

Prime 3377.2 (1196.0) 3512.8 (1226.4) 3563.4 (1328.0) 3375.0 (958.7)

Target 3074.0 (786.3) 3338.4 (1233.2) 3255.7 (1157.3) 3697.9 (1593.9)
Filler

Prhne 36100(10076) 37698(10969) 37473(1065J) 38536(1296J)

Targg 3191.4 (959.9) 3221.3 (1016.5) 3192.3 (969.7) 3329.1 (1255.1)

 

Interference Scores

 

Control M ADHD-Only
+ ADHD-C ADHD-r
ADHD-I
Cost (Corrected)
(Different — Nonsense)
Accuracy 0.25 0.25 0.24 0.30
(0.17) (0.22 (0.22) (0.21)
Reaction Time 898.9 500.9 481.9 571.1
(760.7) (1086.1) (1217.7) (325.3)

 

 

94

Reaction Time- To examine the effect of interference on reaction time, an
additional 2 x 2 (Group x Sex) ANOVA examined group differences in RT Cost. The
group by sex interaction was not significant [F(1,55)=0.12]. Since there were no sex
differences, data for girls and boys were combined and an additional univariate AN OVA
examined RT Cost across sex. This analysis revealed a marginally significant group
main effect [_E(1,58)=2.59, p=.11] that was weakened by age and IQ covariance.

Secondary comparisons of RT changes from prime to target in the Nonsense
condition for Control versus ADHD children was significant [F(l,58)=5.29, p=.03,
n2=.08], such that ADHD children showed sinillg reductions in RT from prime to target
(i.e., increased interference/Cost) than did Control children. In contrast, the test of prime-
target RT differences in the Different condition was non-significant [F (1,58)=0.3 5]
indicating no differences in relative change in RT between groups (i.e., no interference
differences).

Contrary to accuracy findings, then, comparisons of RT changes for ADHD and
Control children suggested that primary differences in interference control occurred when
a subordinate target followed a no-meaning prime (i.e., Nonsense condition), whereas
both groups experienced similar slowings of RT (and thus similar interference as
measured in RT) to subordinate targets following a dominant-meaning prime (i.e.,
Different condition).29

Summary of Categorical Analyses. Comparisons of ADHD and Control groups

. revealed a sex-specific impairment in interference control for girls with ADHD in terms

 

29 It should be recalled (as mentioned earlier) that all subjects experienced facilitation in RT (i.e. got faster)
from prime to target. Thus, “costs” in reaction time actually reflect differences in relative RT benefit
between prime and target and findings could instead be interpreted as reflecting activation failures.

95

of accuracy Cost. All children with ADHD also had relatively slower response times to
targets following a no-meaning netural prime; (i.e., Noensense condition) that provided

some, albeit weaker, support for the hypothesis.

Figure 2. Comparison of Prime-Target RT Differences in the Nonsense vs. Different
Conditions

- Prime
4000 _ m Target

   

i

3000 7

Mean Reaction Time [msec]

 

Control ADHD Control ADHD
Nonsense Different

Summary of Key Findings for ADHD and Control Children on the Visual Task

Overall, consistent with predictions, both high levels of inattention and ADHD
diagnosis were associated with more interference from competing homonym meanings,
consistent with weakness in inhibition mechanisms that protect language comprehension.
Secondly, somewhat differently than predicted, inattention influenced the relationship
, between reading and interference control such that high levels of inattention attenuated
the association between reading ability and interference. Finally, smaller working

memory capacity was also associated with weaker interference control.

96

Test of Inhibition Effect 1]: RD Comparisons for the Visual versus Auditory Tasks

Dimensional Analyses for RD Children on the Visual versus Auditory Task}0

For children with RD, correlation and regression analyses were conducted
separately for subjects receiving the visual and auditory versions of the task (for reasons
explained earlier). Although this sample was quite small (making the power to detect
statistical differences low), it is notable that the correlations between task performance
and measures of reading achievement and inattention/hyperactivity were in Opposite
directions for children administered the visual versus auditory versions of the tasks.
Important for interpreting the results that follow is that children taking the visual version
of the task were higher functioning with regard to reading performance (thus they. read
well enough to take the visual task) than children taking the auditory version of the task. »

Achievement and Language. For RD children who took the auditory task (due to
low reading ability), mean reading achievement (word recognition, spelling, and
phonological decoding) significantly predicted accuracy Cost, such that highs;
achievement was associated with decreased interference from competing homonym
meanings (Table 14, p. 98). 3' This finding was not robust to IQ and age control but the
effect was still quite strong (A R2=.20, fi=-.59).32 There was no effect for RT.

In contrast, for RD children on the visual task (N=8), lower reading achievement

and history of more severe developmental reading problems were associated with less

 

3° Group means are in Table 22, pg. 107

‘ 3' Note, however, that these scores likely have a very truncated range because reading achievement for all
of these RD children was very low.

32 Paradoxical 1y, higher teacher-rated learning problems also marginally significantly predicted reduced

interference after age and IQ were controlled (A R2 =.25). However, learning problems appeared to be
more related in all analyses to inattention and this finding would be consistent with those results.

97

interference (lower accuracy costs; see Table 14)——or, conversely, higher reading
achievement was associated with greater interference (higher accuracy costs or increased
competition between alternate meanings of ambiguous words) for accuracy but not for
RT.33 Poorer reading performance in these “functional” RD readers may thus be
suggestive of interference control problems, manifested by normal activation of
homonym meanings but failure to constrain access to the context-appropriate meaning.

Age and IQ did not significantly affect this relationship.

Table 14. Achievement/IQ Correlations for Accuracy and RT Cost Scores for RD
Children on Visual and Auditory Tasks

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

VISUAL (N=8) AUDITORY (N=12)
Accuracy Reaction Accuracy Reaction
Time Time
Age -.275 .636+ £41 .142
Full Scale IQ (Estimated) .391 L588, -.274 .4_24
Nonverbal IQ (Estimated) .404 mg -.045 .444
Mean Reading Score .635+ -.378 -.615* .055
Passage Comprehension LSS) 5L5 -.076 -.878** -.084
WIAT Reading(SS) .765** -.332 -.582* -.022
Concepts and Directions (ss) .087 -.348 -.275 .002
Recalling Sentences (ss) .108 -.053 -.649* ._44_0

 

** p<.02 (2-tailed); * p<.05 (2-tailed); + p<. 10 (2-tai1ed); larger correlations=13rger costs

Working Memory and IQ. On the auditory task, as found in the ADHD/Control
group, weaker expressive language/ working memory but not receptive language was
associated with greater interference (r_'=-.65) in accuracy. Thus, as expected by the
empirical literature on lexical ambiguity resolution, larger working memory capacity was
predictive of reduced interference from competing homonym meanings, even afier

controlling for age and IQ (Table 18, pg. 101). This effect also remained marginally

significant above and beyond the effect predicted by reading achievement (RZA =.18; [3:-

 

33 There was, however, a marginally significant association between older age and increased interference
(RT Cost: [=.64).

98

.47).34 The model incorporating reading achievement, inattention symptoms, and
expressive language/working memory was a statistically significant predictor of
interference, with higher working memory/reading abilities and inattention all predicting
reduced interference [F (3,8)=4. l 7, p<.05; R2 =6 l ].

In contrast, on the visual task, performance on the expressive language/working
memory measure (Recalling Sentences) did not predict interference (for accuracy or RT).
Controlling IQ and reading achievement did not affect this finding. Thus, one possibility
is that expressive language or working memory deficits are an additional (i.e.,
independent) factor that independently affects the functioning of children with RD, such
that children who have this deficit are more severely affected than those who do not
(thereby explaining the findings for RD children on the auditory task); alternatively,
poorer working memory capacity (and, consequently, poorer interference control) could
be a byproduct of a more severe RD impairment.

lnattention, Hyperactivity, and Learning, Correlations between Cost scores (i.e.,
interference) and symptoms of parent and teacher-rated inattention and hyperactivity for
children with RD on the visual versus auditory tasks are shown in Table 15.

For RD children Who took the auditory task, inattention/hyperactivity was not
associated with accuracy Cost, even after controlling for age and IQ effects. There was,
however, a significant correlation between mother-rated attention problems on the BASC
and RT Cost, such that high levels of inattention were associated with reduced

interference (_r=-.61). This finding Was not robust across multiple attention indices.

 

3‘ Thus, as with findings for Control/ADHD children, this finding provides further indication that
performance on the Recalling Sentences test may depend on factors in addition to language ability (i.e.,
working memory capacity) and that these independent factors, not language ability per se might account for
the prediction of interference control with this measure.

99

Reading and total inattention symptoms did not interact to predict interference
(i.e., Cost) on the auditory task (Table 18). However, the interaction of inattention and
_agp was a marginally significant predictor of accuracy Cost, such that older children with
attention problems experienced less interference from competing homonym meanings
than those without attention problems. Both groups had higher levels of interference than
younger RD children, for whom inattention did not affect level of interference. flrjp
finding suggests that lower functioning children with RD (i.e., those who could not read
well enough to take the visual tafiexperience increased interference with agepbut thit
mattention moderaLtes and attenuates this effect. High levels of inattention may thus
interfere with a developmental trajectory for maturation of inhibitory mechanisms that
constrain activation to the context-appropriate meaning of the ambiguous word.

On the visual task, greater mother-rated inattention was associated with a pattern
of higher RT Cost (i.e., greater interference) across multiple indices. Again reciprocal to
findings for the auditory task, symptoms of inattention significantly positively predicted
RT Cost when age was controlled (R2 = .76, RZA = .36, p 5 .05; Table 17), such that
increased inattention was associated with increased interference in high functioning RD
children. The strong positive predictive power of age on RT Cost suggests that
interference increased with age independently of inattention effects.35 Accuracy Cost
was not correlated with any indices of inattention or hyperactivity, and there was no
interaction between reading achievement and inattention in predicting interference

control for either accuracy or RT.

 

35 Note that this finding matched that for the auditory task, paradoxically reflecting the development of
inhibitory mechanisms that ultimately constrain activation to the context-appropriate meaning of the
ambiguous word as children age.

100

Table 15. Inattention and Hyperactivity Correlations for Accuracy Cost for RD children
on the Visual and Auditory Task

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

VISUAL (N=8) AUDITORY (N=12)
Accuracy Reaction Accuracy Reaction
Time Time
Total Inattentive Symptoms -.249 .30] -.032 .450
Total Hyperactivity Symptoms .096 .281 -.248 -.036
Mother
ARS Total (Raw) -.1 16 .576 -.070 .015
ARS Inattentive (Raw) -.156 .627+ -.070 -.072
ARS Hyperactivity (Raw) -.033 .421 -.066 .092
BASC Attention (T-Score) -.188 .4_6§ -.158 -.607*
‘ Teacher
ARS Total (Raw) .375 .065 -.178 -.083
ARS Inattentive (Raw) .173 .154 -.268 -.014
ARS Hyperactivity (Raw) .594 .064 -.050 -.125
BASC Attention (T—Score) .398 .369 -.358 -.247
BASC Learning LT—Score) -.092 .283 -.456 .008

 

 

+ p<.08 (2-tailed)

Conclusions. For all RD children, reading ability was associated with level of
interference, although the pattern of findings differed for the visual versus auditory. tasks.
RD children administered the auditory task demonstrated more interference as a function
of both weaker reading and smaller working memory capacity, whereas RD children on
the visual task showed an association between lower reading achievement and less
interference (or stronger reading ability and more interference). For both groups, data are
consistent with the idea that higher levels of inattention disrupt a developmental
trajectory towards more constraint-based lexical access that could be associated with
development of inhibition-driven ambiguity resolution (and, consequently, with an

attention-related inhibition deficit).

101

 

Table 16. Accuracy Cost Regressions for the Visual Ambiguous Words Task: RD (N=8)

 

 

 

 

 

 

 

Model Variable E Ag], A R2 _1; If 1
_g R2 A
- =
1 Mean Achievement .40 .30 .40 4.05 (1,6)+ .63 2.01+
2 Full Scale IQ (estimated) .41 .17 .25 2.12 (1,5) -.06 -0.14
Mean Achievement .68 1.46
3 Age (Months) .45 .23 .37 3.36 (1 ,5) -.21 -0.63
Mean Achievement .61 1.83
1 Total lnattention Symptoms .06 -. 10 .06 0.40 (1,6) £5 -0.63
2 Full Scale IQ (estimated) .16 -.18 .01 0.04 (1,5) .35 0.76
Total lnattention Symptoms -.09 -0.19
3 Age (months) .19 -. 14 .1 1 0.70 (1 ,5) -.37 -0.89
Total lnattention Symptoms -.35 -O.84
1
Step 1 Mean Achievement .41 .17 .41 1.71 (2,5) .70 1.69
Total lnattention Symptoms -.08 0.85
Step 2 Interaction (Achiev.*Inatt.) .45 .04 .05 0.33 (1,4) ._2; 0.59
l Recalling Sentences .01 -.15 .01 0.07 (1,6) .11 0.27
2 Age (months) .12 -.23 .05 0.28 (1,5) -.36 -0.80
Recalling Sentences ._2__4 0.53

 

‘13 after final step; +p5.10

Table 17. RT Cost Regressions for the Visual Ambiguous Words Task: RD (N=8)

 

 

 

 

 

 

Model Variable _R: A4], A33 E 13* 1
.11. R2 A
E
1 Mean Achievement .14 .00 .14 1.00 (1,6) -.38 -1.00
1 Total lnattention Symptoms .16 .02 .16 1-l5 (1.6) ~40 1-07
2 Full Scale IQ (estimated) .37 .1 l .02 0.17 (1,5) -.51 -l.28
Total lnattention Symptoms .16 0.41
3 Age (months) .76 .67 .36 7.46 (1,5)* .8] 3.55*
Total lnattention Symptoms .62 2.73*
1 Mother Inattent st. (MARS) .39 .29 .39 3.89 (l,6)+ .63 l.97+
2 Age (months) .85 .79 .45 15.24 (1,5)* .68 3.95*
Mother Inattent st. (MARS) .67 3.90*
1
Step 1 Mean Achievement .23 -.08 .23 0.75 (2,5) -.46 -1.10
Total Inattention Symptoms .37 0.93

Step2 Interaction (Achiev.*1natti .44 .03 .21 1.52(1.4) -.50 4.23

 

‘13 after final step; ‘p _<_ .05; +p5.10; positive B’s=1arger Costs.

102

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 18. Accuracy Cost Regressions for Auditory Ambiguous Words Task (N=12)
Model Variable 33 _A_dj, ABE E ff 1
E R2 A
_— E
1 Mean Achievement .38 .32 .38 6.10 (1,10)* -.62 -2.47*
2
Step 1 Full Scale IQ (estimated) .08 -.02 .08 0.81 (1,10) .09 0.28
Step 2 Age (Months) .19 .01 .12 1.29 (1,9) .13 0.40
Step 3 Mean Achievement .39 .17 .20 2.69 (1,8) —.59 -l .64
1 Total lnattention Symptoms .00 -.10 .00 0.01 (1,10) -.03 -0.10
2 Full Scale IQ (estimated) .22 -.08 .03 0.34 (1,8) -. 12 -0.36
Age (months) .43 1.21
Total lnattention Symptoms fl -0.59
1 Age .48 .29 .27 4.14 (1,s)+ .62 2.24+
Total lnattention Symptoms £1 -1.29
Interaction (Age*Inattention) -.55 -2.03+
1
Step 1 Mean Achievement .41 .28 .41 3.15 (2,9)+ -.79 -2.95*
Total lnattention Symptoms -.28 -1.08
Step 2 Interaction (Achiev.*Inatt.) .53 .35 .11 1-92 (1,3) -37 1-39
l Recalling Sentences .42 .36 .42 7.27 (1,10)* -.65 -2.70*
2 Full Scale IQ .56 .40 .37 6.86 (1,8)* .43 1.31
Age (months) .34 1.30
Recalling Sentences -.83 -2.62*
3 Mean Achievement .61 .46 .61 4.17 (3,8)* -.46 -1.86+
Recalling Sentences -.49 -2.02+
lnattention _-_2___2 -0.98
i B after final step; Tp 5 .001; Mp501; *pg .05; + p510
Table 19. RT Cost Regressions for Auditory Ambiguous Words Task (N=12)
Model Variable 13: app _A_Rj _15 B} '_r_
_Rj R2 A
1 E
1 Mean Achievement .06 -.10 .00 0.03 (1,10) .06 0.18
2
Step 1 Full Scale IQ (estimated) .16 -.15 .00 0.00 (1,8) .41 1.06
Step 2 Age (Months) .33 0.83
Step 3 Mean Achievement .02 0.06
1 Total lnattention Symptoms .20 .12 29 2.54 (1,10) -.45 -1.59
2 Full Scale IQ (estimated) .42 .20 .26 3.51 (l,8)+ .39 1.31
Age (months) .48 1.55
Total lnattention Symptoms -.53 -1.87+
Step 2 Interaction (Achiev.*lnatt.) .34 -.03 .07 0.71 (1,7) .31 0.84
2
Step 1 Mean Achievement .31 .05 .31 1.19 (3,8) .00 0.01
Total lnattention Symptoms -.6l -1.89+
Age (months) .32 0.85
Step 2 Interaction (Achiev.*Inatt.) .39 .04 .08 0.91 (1,7) .31 0.95

 

:13 after final step; +p§.10; positive B’s=larger Costs.

103

Secondary Comparison of Current Findings to those of
Gemsbacher, Robertson, and Werner (2001)

In addition to experimental hypotheses, an additional question of interest was
how the current findings in this child sample compared to those of Gemsbacher and
colleagues’ college sample of poor versus good comprehenders, and whether current
findings supported a relationship between inhibitory deficits and reading comprehension.
The relationship between passage comprehension and accuracy/RT Cost was therefore
examined separately for the auditory and visual tasks.

On the auditory task, passage comprehension significantly negatively predicted
accuracy Cost above and beyond effects of overall reading achievement, IQ, and age
(Table 54, below). Thus, higher readingcomprehension was associated with r_eg_u_c_e_d
interference (i.e., decreased Cost).36 Together, reading achievement and passage
comprehension explained 82% of the variance in interference control, with passage
comprehension accounting for the majority of this effect (Bcompmhcnsiofi -.76 vs. Brewing = -
.23).

To further explore how well interference control (i.e., Cost) predicted reading
comprehension above and beyond reading ability, Cost and reading achievement were
entered as predictors in a regression with passage comprehension as the outcome. This

analysis revealed that interference control accounted for the vast majority of predictive

 

3" Note that in Gemsbacher et al’s study, decreased Cost was taken as a reflection of inhibition failures
(which, according to Gemsbacher’s Structure Building Framework, should produce poorer
comprehension). As a result, if interpreted according to Gemsbacher et al’s schema, the current findings
would actually suggest that better reading comprehension predicted an inhibition deficit. An alternative
interpretation of this finding will be reviewed in the Discussion.

104

power for passage comprehension performance (Bean: -.91 vs. Breading = -.05)37, such
that—consistent with Gemsbacher’s hypothesis—poorer interference control (increased
Cost) was associated with poorer reading comprehension. In interpreting these findings,
it is important to note that given the very low reading abilities of children in this group,
the range of performances on this task was truncated. Despite the supportive findings for
accuracy results, passage comprehension was not associated with RT interference

(contrary to Gemsbacher’s findings).

Table 20 Auditory Task

 

r

 

 

Model Variable B.’ A_d]°, A31 13 p; _T_
B— .1319.

1 Passage Comprehension .77 75 .77 33.53 (1,10)? -.88 -5.79t

2 Full Scale IQ .79 .71 .60 22.45 (1,8)T -.13 -0.75
Age -.12 -0.56
Passage Comprehension -.91 4.741

3 Mean Achievement .81 .77 .43 20.24 (1,9)? -.23 -1.35
Passage Comprehension -.76 4.501

4 DV=Passage Comprehension

Step 1 Mean Achievement .26 .19 .26 3.49 (1,10)!- .51 1.87-1-

Step 2 Mean Achievement .77 .72 .51 20.24 (1,9)? -.05 -0.25
Accuracy Cost -.91 4.501-

 

On the visual task, similar to other differences in auditory versus visual task
performance, there was a trend towards a positive predictive relationship between passage
comprehension and accuracy Cost, such that mrer reading comprehension in this group
was associated with reduced interference. Thus, one possibility that would be consistent
with Gemsbacher’s theory is that these individuals have poorer interference control or an

" inhibitory deficit (i.e., “less cost”). It is notable, however, that this effect was largely

 

37 Accuracy cost accounted for 50% more variance than reading ability in predicting performance.

105

accounted for by overall reading achievement (see Table 55, below) and the two
constructs were highly related (_r=.88, p<.01) in this group. Changes in response speed

(i.e., RT Cost) were not predicted by Passage Comprehension performance in this group.

Table 21 Visual Task

 

 

Model Variable g2 _A_dj, A_Rz p 51 I
.133 .319.
l Passage Comprehension .33 .22 .33 2.96 (1,6) pp 1.72
2 Full Scale IQ .33 .06 .18 1.33 (1 ,5) .02 0.04
Passage Comprehension ,_5_6 1.15
3 Age .47 .25 .39 3.68 (1 ,5) -.38 -1.13
Passage Comprehension p; 1.92
4 Mean Achievement .40 .17 .00 0.01 (1,5) ,_5_7_ 0.79
Passage Comprehension .07 0.10

 

106

 

 

 

 

 

 

 

 

 

m. :3 9.8: 8.22. 23.3 SE 2.28 2a: 2.58 228 328 N38 .lowc:
338 8.88 Sam 8.308 8.23. 3.8: cam: 6.38 4.82 2.. _ m8 :22 est:
8::
meme. 829.: M58. 2328 «.28. $.88 2.22 258 ~22 6.58 as: m can;
35: 3.38 2.252 8.2:”: foam 2.2.2: 4.84:. 2.8:: 8:? 2.2.3: 09.? 25.2
.5382
Emma 8.2 8 ~22 22;: SE 2.28 338 8.88 0.82 2238 28m 82:;
N. :32 6.88 8.3% acme: 2c: 2 .38 f: m 228 do: m 2.88 38m 25...
02.8222
3: m 3.8. .8 4.28. 8.5: 2.82 2.128: .82 8.38 8.33” 8.58 Sham 28:;
38% 2.88 .882.“ 3.8va :2; 228 2a; 9:8 «.38 8.88 3:: 25...
«Sam
22:. 22.23%
83 2.8 83 2.8 83 a8 8.8 2.8 2.2 2.8 83 case
83 2.8 83 a8 :3 2.8 a; 8.8 23 2.8 3.8 as...
3::
$2 2 .8 8.2 3.8 8.2 2.8 ms 2 .8 and 2 .8 who .82.?
No.8 2.8 a; :8 Po 2 .8 c3 2 .8 as 2 .8 8.8 252
929.2885
$8 9.8 a; 3.8 NS 8.8 $8 2.8 8.8 2 .8 23 3a.;
2.3 3.8 :3 2 .8 Po 2 .8 a; 2.8 cad a8 3.0 25.2
omcomccz
No.8 3.8 a; 88 $2 3.8 83 a8 a: 2.8 a; 8:;
No.2 3.8 _3 2.8 who 2.8 m; 2.8 a: 2.8 a; as...
22.5
a
72 ~12 mnz Nuz N12 212
_-2:2< o-2:2< fillaoém _-2:2< o-2:2< »|_=o-2:
2:252: 2:252:
A<Dm~> 855.553.

 

 

 

 

:38. 2:03 m202w58< 05 .80 meofiao> 335 :22 200293 05 28 $282 2220 Ox .3 2an

107

RESULTS-Part 11
Garden Path Task

Description of Conditions

In brief recap of Garden Path Task methodology, conditions varied along two

dimensions. Plausible-implausible conditions addressed the influence of semantic

factors on garden path revision—in particular, sensitivity to semantic cues signaling an

incorrect interpretation or “comprehension error”. Reflexive Absolute Transitive

(RAT) versus Optional Transitive (OT) verb conditions examined the influence of

syntactic factors on garden path revision—i.e., whether syntactic complexity and

requirements for syntactic “editing” affect success at revising misinterpretations.

Experimental manipulations also varied according to the incorrect interpretation’s

“strength” or “resistance to revision.” Thus, semantically plausible and OT verb

sentences are more difficult to revise than implausible or RAT verb sentences.

Table 23. Example Sentences and Condition Error Rates Across Groups for the Visual
and Auditory Garden Path Task.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Condition Item Example Accuracy
Visual/Audit.
Plausible As .loewalked the dog nap7ped. 0.6] 0.44
5 OT Q. Did Joe walk the dog. (0.27) (0.29)
a Implausible As the ape ate the tree grew. 0.82 0.75
a Q: Did the ape eat the tree? (0.24) (0.27)
§ Plausible As Betty woke up Daddy was sleeping in the den. 0.82 0.64
6 RAT Q: Did Betty wake up Daddy? £0.26) (0.31)
Implausible As the man shaved the cat sat in the tree. 0.86 0.70
Q: Did the man shave the cat? (0.29 LOSOL
g Plausible The water cooked as Deb drank. 0.73 0.69
‘5'; OT Q: Did Deb drink the water? (0.27) £0.30)
2; lm lausible The worm swam in the lake as Jack ate. 0.93 0.94
g p Q: Did Jack eat the worm? (0.19) (0.09)
3'5 Pl b] The baby was in bed as the boy washed. 0.93 0.93
“g RAT “5‘ c Q: Did the boy wash the baby? (0.21) (0.18)
£ Implausible The kids swam in the pool as Frank dried off. 0.93 0.82
Q: Did Frank dry off the kids? £119) (0.25)
OT: Optional Transitive, RAT: Reflexive Absolute Transitive

108

 

Comparison of the Auditory and Visual Task

In the case of this task, visual and auditory task data were analyzed together for
several reasons: 1) Overall error patterns were the same in both; 2) Correlations (below)
revealed similar patterns for both task versions; and 3) Performance on control sentences
was similar in the two versions [F(l, 95)=l .30, n.s.; Auditory mean:0.85: 0.14; Visual
‘ mean: 0.89 i 0.13].38

Task Validation

Appendix A presents extensive background on task characteristics and validity,
including an item analysis, inter-condition correlations for accuracy, age-effects (across
groups), and confirmation of basic task effects (i.e., garden path effects across groups).
Refer to the Appendix A (p. 171) for more detailed information regarding these findings.

Overall, those data supported the effectiveness of the experimental manipulations
in each condition and the adequacy of the individual items. Mean accuracy across
subjects for each condition is shown in Table 23. Task effects, worked as expected, with
higher error rates in Garden Path than Non-Garden Path, Plausible than Implausible, and
OT than RAT cells. Age was weakly correlated with accuracy in several cells (OT-
Plausible, OT-lmplausible, RAT-Plausible; r’s=.18). As might be expected, IQ was also
correlated with accuracy in most conditions but particularly so in experimental conditions
(r’s ranging from .29 to .48). Age and IQ were therefore controlled in regression and
ANOVA analyses. Note that confinement of age and IQ correlations to the experimental

(high interference) conditions could suggest the operation of well-developed,

 

38 One potential complication, however, were the higher error rates on filler sentences in the auditory task
[F(] ,95)=6.05, p<.02]. Results were therefore also checked using only visual task data. In all cases, the
pattern of results did not change when checked in the visual task only.

109

automatized, processes underlying control condition performance, whereas the
experimental conditions could reflect processes that are still manning.
Preliminary Analysis of Task Eflects in Control Children

A 2 (Garden Path vs. Non-Garden Path) x 2 (Verb Type: RAT vs. Optional
Transitive) x 2 (Plausibility: Plausible vs. Implausible) repeated measures analysis of
variance was used to determine the effectiveness of experimental manipulations in
Control children. Briefly, this analysis revealed the expected garden path (interference39)
effects and showed utilization of both syntax and semantics to help in revision of garden
path misinterpretations, leading to better performance in RAT verb conditions as well as
semantically implausible conditions relative to their counterparts (see Figure 3 below).

Thus, this analysis revealed a marginally significant 3-way Garden Path*Verb
Type“ Plausibility interaction [F (l,26)=3.43, p<.08, n2=.12]. The Garden Path *
Plausibility interaction was significant for OT verb sentences [F(1,26)=10.21, p<.01] but
not for RAT verb sentences [F (1 ,26)=1 .40, p=ns]. The pooled two-way Garden Path *
Plausibility interaction was also significant [F (I ,26)=9.71, p<.01, n=.27], whereas the
Verb Type *Garden Path interaction was not [F (1 ,26)=1 .77, 112806].40 Finally, there
were main effects of Verb-type [F (l, 26)=32.32, p<.001, n=.55], Plausibility
[F(l,26)=16.78, p<.001, n2=.39], and Garden Path [F(l,26)=14.81, p=.001, n2=.36], and
significant 2-way interactions of Verb*Plausibility [F(l ,26)=35.32, p<.001, q2=.58;

Figure 10 in Appendix A]. All of these effects essentially indicated that the children

 

39 Note that the term “interference” again refers- not neceflin'lv to competition between eguivalently-

active meanings (i.e., egual competitors)-- but to the ability to handle or suppress inappropriate activation.
40Accuracy differences were significant in OT sentences and, in RAT sentences, was marginally worse than

accuracy in non-garden path RAT control sentences [F( l ,26)=2.94, p<. l0].

 

 

110

were responding to the task as expected and that the manipulations appeared to be

challenging interference control mechanisms as intended.

Figure 3. Garden Path by Plausibility Interaction for RAT and OT Verb Conditions in
Control Children.

- Garden Path V
m Non—Garden Path

Mean Accuracy

 

 

Plausible Implausible Plausible Implausible

Optional Transitive RAT

Data Reduction and Review of Planned Comparisons
For ANOVA, relevant interactions were examined; for regressions, composite
accuracy scores were created to reduce the total number of these analyses (and
familywise Type I error). Table 32 (pg. 174) shows means of derived indexes by group.“
Three types of effects were of interest here, leading to creation of three main composite
scores and/or interaction effects in those respective analyses, as follows.

1) “Garden Path Effects” (Non-Garden Path minus Garden Path accuracy

score or Garden Path x Non-Garden Path interactions). Here, Garden Path (subordinate-

 

4| Correlations among these various derived indexes are shown in Table 40 in Appendix A.

III

matrix syntactic structure—see example in Table 23,) versus Non-Garden Path (matrix-
subordinate syntactic structure—see example in Table 23) comparisons explored ability
to “inhibit” or “revise” incorrect syntactic representations over and above that of
incorrect “pragmatic inferences” (i.e., those drawn from lexical-semantic activation or
associations; See example in next paragraph). Put another way, these analyses compared
inferential errors resulting from stimulus-driven semantic associations in non-garden path
sentences (e.g., between “ate” and “banana” in “The banana fell as the ape ate” where ate
and banana are not linked syntactically) to those resulting from syntactic (as well as
semantic) associations in garden path sentences (e. g.,”As the ape ate the banana fell”
where “ate” and “banana” are syntactically linked as part of a subordinate-matrix clause).
Experimental manipulations of plausibility and verb-type then examine the success of
top—down revision or suppression on these bottom-up associations.

Compare, for example, the garden path sentence “As Deb drank the water
cooke .” to the non-garden path sentence “The water cooked as Deb drank. ” Lar er
error differences between such sentences (the “Garden Path Effect”) indicates less ability
to overcome (i.e., inhibit) syntactically-established causal relationships (e.g., to causally

“disconnect” water from drank) relative to nonspecific semantic associational

 

relationships (i.e., semantic associations between water and drank). Prediction: Deficits
in strategic inhibition mechanisms should be associated with larger “garden path effects”
of this sort.

2) “Plausibility Effect” (Implausible accuracy-Plausible accuracy and
Misible x flausible intflztions). Plausibility comparisons provided an index of

sensitivity to semantic cues signaling a misinterpretation and ability to deploy

112

semantically-driven inhibition mechanisms to suppress or revise them (relevant indices:
Non-Garden Path Plausibility Effect: OT/RAT, Garden Path Plausibility Effect:
OT/RAT). Plausibility comparisons therefore address hypotheses regarding ability to
draw upon semantically-driven (yet still strategic or effortful) inhibition mechanisms.
_Iiirggdifferences indicate greater success at utilizing strategic inhibition. Prediction:
Functioning of top-down, controlled, inhibitory processes should produce re-
interpretation of implausible “misinterpretations” (and, consequently, larger differences
between Implausible and Plausible accuracies). Thus, opposite to “garden path effects”
larger scores indicate better inhibition.

3) Verb effects (RAT accuracy-OT agcurarcy; RAT x OT interactionp):
Comparisons of RAT and OT verb manipulations provided an index of sensitivity to
syntactic complexity and ability to revise misinterpretations when requirements for
effortful processing are reduced-- in effect when the revision threshold is lowered
(perhaps due to competition from the implicit reflexive object [himself/herself] that
weakens the strength of the garden pathed syntactic subordinate verb-noun phrase
association). _Iirgg differences between RAT and OT verb accuracies reflect improved
performance when inhibition requirements are reduced—i.e., successful inhibition.
Prediction: Improved accuracy for RAT verbs compared to OT verbs would indicate
successful inhibition. Individuals with strategic inhibition impairments may therefore
show reduced improvement to RAT verb sentences (i.e., show larger verb effects).
Summary

In brief recap of diagnosis-specific hypotheses, poorer reading/language ability

and higher inattention were both predicted to relate to failure to revise garden path

113

misinterpretations (i.e., inhibitory deficits or failure to inhibit inappropriate meanings
once activated), specifically when effortful or strategic control was required. Likewise,
both ADHD and RD children were predicted to show inhibitory deficits (i.e., higher
revision failures) on this task in comparison to Control children. Thus, impaired strategic
or effortful inhibitory control should lead to one or more of the following: (a) large
garden path scores, (b) small plausibility scores, (c) and small verb effect scores.
Plan of Data Analysis

Analyses will be presented as follows: 1) Dimensional analyses (both
correlational and regression) address hypotheses concerning the relationship of reading
and attention problems to inhibition and interference effects. Regression analyses
examined whether indices of achievement versus inattention uniguely or additively
predicted inhibitory ability when controlling for effects of other variables such as
estimated IQ and age. 2) Group/categorical analyses (ANOVA’s) addressed hypotheses
regarding whether ADHD and RD have overlapping cognitive inhibition deficits. These
analyses were conducted with condition means rather than derived indexes. Note that to
streamline and focus results on the hypotheses, many non-significant findings that were
peripheral to primary hypotheses have been placed in Appendix A. The reader will be
referred to the Appendix for further information when appropriate.

Note about Reaction Time

Following the procedures of Christianson et al., (2001), reaction time (RT) data

were considered supplementary to key questions and were not examined in the current

study, because sentence and question length were not controlled within or across

114

conditions. Ideally, eye movement data would be utilized to examine RT effects on
garden pathing. Thus, all data described here are accuracy (or error) data.

I now move to specific hypothesis tests. Note that tables of correlation analyses
display correlations for numerous indices of a given construct (e.g., reading or
inattention). These multiple indices of the same construct are displayed merely to show
that findings are robust across multiple measurements. Interpretation of task effects is
based upon patterns of results across these indices only. The most important
inattention/hyperactivity score is the “Total lnattention” and “Total Hyperactivity”

scores, which sum mother and teacher symptoms.

Dimensional Analyses of Inhibitory Effects on Garden Path, Plausibility, and
Verb-Effect Difference Scores in Relation to Reading Ability and Attention Problem
Scores

IQ, Achievement and Language

Global Inhibition Effects (across verb type and plausibility; Table 24).
Dimensional analyses first explored the existence of inhibitory effects across verb type
and plausibility to differentiate whether dimensional measures of reading/language could
globally predict a failure to “inhibit” or “adjust” incorrect syntactic representations versus
whether effects result solely from incorrect “pragmatic inferences” (i.e., inferences drawn
from bottom-up lexical-semantic associations). Poorer reading achievement (average of
word recognition, spelling, and pseudoword decoding) predicted a larger inhibition
deficit (the total garden-path effect across verb type and plausibility). Although age and

IQ also strongly negatively predicted the garden-path effect, reading achievement

115

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2N8 2.28 328 828 $28 828 2&8 228 22:32 .35
So- a; 82 Re So so 2 o 22
23.8 5.8 328 3.28 2.28 2&8 32.8 23.8 236.2
:2- 88 8.2 22 22 2.2 3.8 mg .
28.8 28.8 3.8 32.8 3N8 22.8 $8 ~38 32>
8.? m; 8.2 8.2 85 mod 2.2 2.2
2.5. 2.0 as. so 2.52 he :2 .5
5a.— ..aEeoéez 5a.. e 25 e323: a: 252.22

aseaaadsaaaiéfl 33.. 8250-5252555

.8222 232.23: .822 5a.— 5.50

 

 

 

 

 

”x23. 08 mo 22382, Egg/x was .2225 28 floobm 322633 23 £3 20230 28 328m 525 33.35 .vm 03m...

116

continued to be a significant predictor in the model afier these variables were controlled
[R2A=.05, p=-.25, p<.05].

An even stronger predictor of global garden path inhibitory deficits, however, was
performance on the measure of expressive language/working memory (Recalling
Sentences). Performance on this measure still negatively predicted the global garden
path effect after age, IQ, and reading achievement were controlled [R2A=.06, B=-.32,
p<.01]. Perhaps not surprisingly, reading no longer significantly predicted the garden
path effect when language was also entered as a predictor.

Semantic and Syntactic Influences on Garden Path Eflects. To differentiate
whether semantic or syntactic experimental manipulations produced unique influences on
inhibition, analyses next examined garden path effects (Non-Garden Path - Garden Path
accuracy) separately by plausibility and verb-type and these are depicted in Table 25.

These correlations revealed strong, significant, negative associations between
measures of reading and language and garden path effects for OT/implausible (but not
OT/plausible) sentences. Thus, poorer reading and language abilities were associated
with less effectiveness at utilizing semantic information to drive top-down, strategic
inhibition of garden path misinterpretations (i.e., larger garden path effects), especially
under conditions of higher processing demands (i.e., when the causal structure is more
resistant to revision in OT verbs).42

Globally poorer language ability was also associated with failure to benefit from
reduced inhibition demands from RAT verb sentences (lower revision threshold)

regardless of semantic plausibility. This effect was robust to age and IQ control (Table

 

42 Note that all children had difficulty revising semantically plausible garden path misinterpretations.

117

25). Thus, poorer language abilities were associated with either 1) difficulty utilizing
syntactic rules/structures to recover from garden path errors (e. g., due to a lack of implicit
knowledge or proficiency with RAT verbs), or 2) a more severe inhibition deficit that
prevents inhibition of misinterpretations even when the causal structure is less resistant to
revision. Given that semantic plausibility played no role in ameliorating this effect (i.e.,
it was present for RAT verbs in both the plausible and implausible condition), it seems
more likely that language-specific impairments in processing complex syntax rather than
an inhibitory deficit produced this finding.

Taken together, findings suggest that language and reading problems are
associated with both a) decreased effectiveness at utilizing semantic information to
inhibit garden path misinterpretations and b) failure to benefit from reduced inhibition
requirements of RAT verbs (stemming either from lack of knowledge about the syntactic
features of RAT verbs or from a more severe inhibition deficit). Findings were robust to

age and IQ covariance.

Table 25. Garden Path Effects Correlations for IQ, Achievement, and Language.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Garden Path Effect: Garden Path Effect:
Plausible Implausible

Optional RAT Optional RAT

Transitive Transitive
Age -.064 -.1 16 -.071 -.115
Full Scale IQ (Estimated) -.123 -.225* -.293** -.081
Mean Reading Score —.l30 -.093 -.350*** -.124
Passage Comprehension (SS) -.l38 -.O48 -.310** -.l37
Concepts and Directions (ss) .090 -.001 -.349** -.204*
Recalling Sentences (ss) .007 -.330*** -.339*** -.262**
LBQ Total Score .034 .035 .323" .010

 

on P300]. "I P<~0L *p<,05 (2-tailed), + p<.10 (2-tailed)

118

Table 26. Effect of Verb Type on Garden Path Effects in Plausible Conditions.

 

 

 

_Variable a2 Mi. A yo: a I
L R2 R’A
— — —!
l Recalling Sentences .07 06 .07 7.07 (1,95)" .26 2.66“
2
Step 1 Age .00 - 01 .00 0.22 (1,95) .11 1.08
Step 2 IQ .01 - 01 .01 0.99 (1,94) -.08 ~0.64
Step 3 Recalling Sentences .09 06 .07 7.39 (1,93)M .33 2.72"
3
Step 1 Age, IQ .01 -.01 .01 0.61 (2,94) .08 -031
-.04 0.77
Step 2 Reading Average .02 -.02 .00 0.22 (1,93) -. 16 -1.34
Step 3 Recalling Sentences .10 .06 .09 9.03 (1,92)M .38 3.01""r

 

Comparison of Garden Path versus Non-Garden Path Effects (Inhibition versus
Activation Effects). As a final step, correlational analyses also examined plausibility
effects (Implausible — Plausible accuracy) separately for garden path and non-garden path
sentences to determine whether effects were related solely to performance in one
condition or the other.

This analysis found the strongest effects on non-garden path sentences. Thus,
lower reading achievement, language abilities, and IQ, were globally associated with
higher sensitivity to semantic plausibility in non-garden path sentences. In other words,
those with poorer abilities in these areas continued to rely upon pragmatic inferential
aspects of the sentence, answering based upon bottom-up lexical-semantic associations,
even in non-syntactically complex sentences. No such finding occurred in the RAT/Non-
Garden Path condition (where the RAT verb all but prevents misinterpretation of verb-
object relationships).

Language and reading comprehension abilities were also associated with

performance on garden path sentences. Interestingly, findings suggested that lower

119

reading, language, and IQ were also associated with increased reliance on bottom-up
lexical-semantic associations or even a possible stronger reliance on semantic
implausibility (i.e., increased utilization of top-down, semantically-driven suppression—
for revision of syntactically complex OT garden path interpretations). This finding
suggested that language and reading problems may be associated with global difliculties
processing complex syntax and a subsequent reliance on inferences from bottom-up

lexical-semantic activation (i.e., verb-noun phrase associations).

Table 27. Correlations between Plausibility Effects and Indices of Achievement and
Language.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Plausibility Effect: Plausibility Effect:
Garden Path Non-Garden Path
Optional RAT Optional RAT

Transitive Transitive
Age -.060 -.051 -043 -.029
Full Scale IQ (Estimated) -.274** -.140 -.348*** .052
Mean Reading Score -.110 .084 -.245* .057
W] Word Attack (SS) -.064 .061 -.262** .030
Passage Comprehension (881 -.l78+ .116 -.263** .020
Concepts and Directions (ss) -.006 .114 -.384*** -.104
Recalling Sentences (Si -.l90+ -.028 -.448*** .090

 

 

”* p5.001, ** p<.01, *p<.05 (2-tailed), + p<.10 (2-tailed); Plausibility Effect":Implausible-Plausible

Summary. Taken together, dimensional analyses suggested that language and
reading problems are associated with both a) global difficulties processing complex
syntax with a subsequent reliance on inferences from bottom-up lexical-semantic
activation (i.e., verb-noun phrase agreement), and b) decreased effectiveness at utilizing
semantic information to inhibit garden path misinterpretations (perhaps as a byproduct of

problems with complex syntax).

120

Inattention, Hyperactivity, and Learning

Global Garden Path Effects (across verb type and plausibility; Table 28).
Dimensional analyses first explored the existence of inhibitory effects across verb type
and plausibility to differentiate whether dimensional measures of inattention/
hyperactivity globally predict a failure to “inhibit” or “adjust” incorrect syntactic
representations versus whether effects are driven by “pragmatic inferences.” (i.e.,
inferences based upon bottom-up lexical-semantic associations). As shown in Table 28,
hyperactivity significantly predicted the global garden path effect, but this effect was not
robust to age and IQ control.

Semantic and Syntactic Influences on Garden Path Effects. To differentiate
whether semantic or syntactic experimental manipulations might have obscured garden
path inhibition effects, analyses next examined garden path effects (Non-Garden Path -
Garden Path accuracy) separately by plausibility and verb-type. These are depicted in
Table 28. lnattention and hyperactivity were not associated with garden path effects in
either the OT or RAT plausible condition, suggesting that all children experienced
difficulty revising misinterpretations in these conditions.

lnattention and hyperactivity were, however, associated with larger garden path
effects in the OT/implausible condition, such that more inattention and hyperactivity was
associated with less effectiveness at utilizing semantic information (i.e., implausibility) to
drive top-down, strategic inhibition of garden path misinterpretations (i.e., larger garden
path effects), especially under conditions requiring more inhibition for success (i.e., OT
verb’s greater resistance to revision). Likewise, hyperactivity ratings (but not inattention

ratings) were also associated with larger garden path effects (strategic inhibition

121

problems) in the RAT/implausible condition, which might reflect consequences of more

severe inhibitory deficits. 43

Table 28. Garden Path Effects: Correlations with lnattention and Hyperactivity.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Garden Path Effect: Garden Path Effect:
Plausible Implausible
Optional RAT Optional RAT
Transitive , Transitive
Total Inattentive SymptomsI -.098 .112 244* .083
Total Hyperactivity Symptomsf .018 .060 .266“ .222*
Mother-Ratings
ARS Total Raw-Mother -.059 .056 .254* .143
ARS Inattentive Raw -.085 .072 .247* .114
ARS Hyperactivity Raw -.023 .030 .230* .150
Teacher-Ratings
ARS Total Raw .016 .102 .311** .226*
ARS Inattentive Raw .009 .090 .299“ .174+
ARS Hyperactivity Raw .021 .099 .281“ .249*
Conners ADHD Index T -.055 .019 .354*** .158
BASC Learning .051 .093 .392*** .177+

 

 

***p<.001, **p<.01, * p<.05, + p<.10 (2-tailed)
: Mother+Teacher symptoms by the “OR algorithm”

Comparison of Inhibition (garden path) versus Semantic Activation (pragmatic

inference) Effects in Relation to Semantic Plausibility and Verb Type. As a final step,

correlational analyses also examined plausibility effects (Implausible — Plausible

accuracy) separately for garden path and non-garden path sentences to determine whether

effects were related solely to performance in one condition or the other. Correlations

between measures of inattention/hyperactivity and plausibility effects are shown in Table

29 below.

 

‘3 Findings of no association between inattention and hyperactivity and garden path effects in the plausible
conditions reflects the fact that all subjects experienced un-revised garden path misinterpretations (for
clarification, refer to means for Control children in Table 31, ).

122

 

These analyses revealed a differential effect of hyperactivity and inattention.

Higher levels of hyperactivity were marginally associated with smaller plausibility

effects (thus, worse interference control) in the garden path/OT condition, indicating

failure to utilize semantics to overcome garden path misinterpretations under highly

“revision-resistant” conditions (i.e., OT verb sentences).

44945

Table 29. Correlation between Plausibility Effects and lnattention and Hyperactivity.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Garden Path Non-Garden Path

PlausibilitLEffect Plausibility Effect

Optional RAT Optional RAT

Transitive Transitive

Total Inattentive Symptoms.I -.157 -.016 .165 -.059

Total Hyperactivity SymptomsI -.186+ -.075 .038 .089
Mother-Ratings

ARS Total Raw-Mother -.073 -.087 .206* .003

ARS Inattentive Raw -.039 -.047 .256* -.010

ARS Hyperactivity Raw -.100 -.121 .125 .017

BASC Attention -.001 .014 .282" -.049
Teacher-Ratings

ARS Total Raw -.l79+ -.065 .082 .055

ARS Inattentive Raw -.1 l6 -.O77 .134 .004

ARS Hyperactivity Raw -.220* -.043 .017 .100

BASC Attention -.O7O -.083 .277” -.128

BASC Learning .009 -.115 .275** -031

 

**p<.01; * p<.05; + p<.08 (2-tailed)

I Mother+Teacher symptoms by the “OR algorithm”

Symptoms of inattention, on the other hand, were correlated with plausibility

 

effects in the non-garden path/0T condition, such that higher levels of inattention were
related to more reliance on sealantic inferential reasoning (i.e., bottom-up, lexical-

semantic associations) in less syntactically complex conditions (i.e., control, non-garden

 

‘4 This finding is opposite to findings for language/reading.
‘5 Note that this is a weaker test of the hypothesis, since it does not take into account over-reliance on
bottom-up semantic mechanisms controlled by the comparison to Non-Garden Path sentences.

123

path sentences, for example, “The banana fell as the ape ate.”). High levels of inattention
were thus associated with a greater tendency to be influenced by “bottom-up” semantic
activation of the verb-object relationship, even when such a relationship was not
syntactically licensed. Such difficulties might relate to either a) insensitivity to bottom
up syntactic inputs or b) failure of automatized suppression mechanisms that should
control interference from irrelevant lexical-semantic activation in order to support
construction of accurate text-level representations (e. g., as proposed by Gemsbacher’s
(1997) Structure Building Framework)“

Summary. lnattention and hyperactivity were both associated with less
effectiveness at utilizing semantic information (i.e., implausibility) in the form of top-
down, strategic inhibition of activated misinterpretations. There was some suggestion,
however, that for inattention this effect was driven by over-reliance on semantic
inferential reasoning (i.e., bottom-up, lexical-semantic associations) reflective of possible
failures of a more automatic mechanism that should block irrelevant activation in order to
allow more efficient construction of accurate text-level representations. In contrast,
hyperactivity may be more associated with failure of effortful and strategic mechanisms.

1 now turn to findings for categorical/group differences.

 

‘6 Note that the key test of this association: “Total Inattentive Symptoms” was not significant. Thus, this
result must be interpreted with caution.

124

Table 30. Total Garden Path Effect (across Plausibility and Verb Types)

 

 

 

 

 

 

 

 

 

 

 

_Variable sf an A ring 6‘ I
E, R’ R2 A
1 Mean Achievement .08 .07 .08 7.68 (1,95) -.27 4.77“-
2
Step 1 Age .18 .15 .05 5.46 (1,93)‘ -.27 -2.76“
Step 2 Full Scale IQ (estimated) -.23 -2.18'
Step 3 Mean Achievement -.25 -2.34'
1 Total lnattention Symptoms .02 .01 .02 1.66 (1,95) .13 1.29
1
Step 1 Age .13 .11 .13 7,18 (2,94)? -23 -212’
Full Scale IQ -27 -267“
Step 2 Mean Achievement .18 .15 .05 2.70 (2,92)+ -.25 -2.25'
Total lnattention Symptoms .01 0.07
Step 3 Interaction (Achiev.*lnatt.) .18 .14 .00 0.00 (1,91) -.00 -0.04
_1_ Total Hyperactive Symptoms .05 .04 05 4.99 (1,95)' .22 2.23'
2
Step 1 Age (months) .16 .13 .02 2.62 (1,93) -20 -201‘
Step 2 Full Scale IQ (estimated) -.30 -3.09"
Step 3 Total Hyperactive Symptoms .16 1.62
l Recalling Sentences .15 .14 .15 16.85(1,95)’r -.39 4.111
2 Age (months) .23 .20 .09 11.21093): -.26 -232“
Full Scale IQ -.13 -l.16
Recalling Sentences -.38 -3.35"
3
Step 1 Age .24 .21 .06 7.75 (1,92)' -.30 .110“
Full Scale IQ -.09 -0.79
Step 2 Mean Achievement -.l6 -l.50
Step 3 Recalling Sentences -.32 -2.78"
i Passage Comprehension .06 .05 .06 6.16 (1,95)‘ -.25 -2.48‘
2 Age .15 .12 .02 1.68 (1,93) -.23 -2.32°
Full Scale [Q -.25 -2.20'
Passage Comprehension -.15 -1.30
1p<.001,“p<.01, rp<.05

125

Categorical Hypotheses: Analysis of Group Differences on Garden Path (i.e., Inhibition)
Effects

In addition to dimensional analyses on the variables of interest, categorical
analyses evaluated hypotheses regarding group differences. Such comparisons are
important for determining whether patterns of impairment differ at the level of clinical
diagnosis (i.e., when symptoms reach a certain level of severity) versus whether
impairments become evident along a continuum as symptom severity increases (i.e., as
seen in dimensional analyses). To evaluate categorical hypotheses, 6 additional repeated
measures analyses of variance were performed. These analyses relied upon condition
means rather than the earlier index scores, because findings are conceptually easier to
understand and to depict graphically. To maximize inclusion of participants, ADHD-C
and ADHD-I children were combined into one ADHD group (based upon findings in
regression analyses related primarily to inattention but not to hyperactivity). Also, those
with ADHD+RD were included in the RD group but not in the ADHD group.

First, to examine performance for all conditions and all groups, a 2 x 2 x 2 x 3
omnibus repeated measures ANOVA was used, with Garden Path Condition (Garden
Path versus Non-Garden Path), Plausibility (Plausible vs. Implausible), and Verb Type
(RAT vs. Optional Transitive) as within subjects factors and diagnostic group (Control,
ADHD, RD) as the between subjects factor. This omnibus analysis revealed a main
effect of group [£(2,79)=6.26, p<.01], a significant 2-way Group x Garden Path condition
interaction [F(2,79)=4.00, p<.03, n2=.09; Figure 4], and a significant 3-way Verb x
Plausibility x Group interaction [F (2,78)=4.68, p<.02, 112:.11]. There was also a main

effect of verb type [E (l , 79)=44.34, p<.001, n2 =.36]. The 4-way interaction and the

126

Verb x Garden Path x Group and Plausibility x Garden Path it Group interactions were
non-significant.

Taken together, then, this omnibus analysis revealed global differences in ability
to revise garden path misinterpretations among groups and was indicative of ADHD and
RD impairments in effortful inhibition and with utilization of top—down, semantically-
driven inhibition mechanisms.

As a secondary comparison to determine whether verb or plausibility had any
effect on susceptibility to making errors on garden path sentences alone (i.e., not in
comparison to non-garden path sentences), an additional 2 x 2 x 3 omnibus repeated
measures ANOVA was employed. Plausibility (Plausible vs. Implausible) and Verb
(RAT vs. OT) served as the within subjects factors and Group (Control, ADHD, RD)
served as the between subjects factor. This analysis revealed no significant interactions

with group.

Figure 4. Garden Path )1 Group Interaction.

 

1I Non-Garden Path
LEltga‘tarden Path" J

 

 

 

Estimated Accuracy

 

 

 

 

 

 

Control ADHD RD

 

127

To deconstruct omnibus findings, a series of additional planned 2 (Garden Path
vs. Non-Garden Path) x 2 (Plausible vs. Implausible) x 2 (Group) post—hoe repeated
measures ANOVAs explored: 1) ADHD vs. Controls on a) RAT verb sentences and b)
OT verb sentences and 2) RD vs. Controls on a) RAT verb sentences and b) OT verb
sentences. Sex, age, and IQ effects were also explored, but these did not substantially
alter any findings.

Control versus ADHD. For OT verb sentences, the 2 x 2 x 2 ANOVA revealed a
marginally significant 3-way interaction [F(], 53)=3.26, p<.08, 112 =.06], in which ADHD
and Control children demonstrated similar differences between garden path and non—
garden path sentences in the plausible condition, but ADHD children continued to show a
much stronger garden path effect in the implausible condition (Figure 6). Thus, for OT
verb sentences, there was modest support for an ADHD-related deficit in effortful
inhibition related to an inability to use semantics to aid in revision of garden path

misinterpretations.

Figure 5. Comparison of Garden Path Effects for Control vs. ADHD-C children for OT
verbs.

 

I Garden Path
E1 Non-Garden Path

Accuracy

 

ADHD Control ADHD Control

Plausible Implausible

 

 

 

128

For RAT verb sentences, the 2x2x2 ANOVA revealed no significant 3—way
PIausibility*Garden Path*Group interaction [F(l ,53)=0.02]. However, the Group x
Garden Path interaction was significant [F(1,53)=7.17, p=.01, n2 =.12], such that ADHD
children continued to show garden path effects (in the plausible and implausible
conditions) for RAT verb sentences, whereas Control Children did not experience garden
pathing for these verbs (Figure 6). RAT verb-type was therefore associated with failure
to revise or adjust incorrect garden path interpretations in ADHD children (regardless of
the plausibility of the misinterpretation). Thus, as is also evident in Figure 6, ADHD
children continued to receive no benefit from semantic implausibility (although the large

standard deviations indicate that the group’s performance was quite variable).

Figure 6. Comparison of Garden Path Effects for ADHD vs. Control Children on RAT
verbs.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1 1 learden Path
' IE1 Non-Garden Path
1
> 0.9 —
O
E
3
0
O
< 0.8 —
0.7 —
0.6 ~
ADHD Control ADHD Control
L Plausible Implausible

 

129

Control vs. RD. RD versus Control differences were also examined with the 2

 

(Garden Path vs. Non-Garden Path) x 2 (Plausible vs. Implausible) x 2 (RD vs. Control)
ANOVA’S separately for OT and RAT verb types.

For OT verb sentences, the 3-way interaction narrowly missed significance
[F (l,45)=2.40, p=.13, n2=.05] but showed a similar pattern to that found in the
comparison of ADHD and Control children (see Figure 5). Thus, as with ADHD
children, for OT verb sentences, RD was associated with a possible trend towards an
effortful inhibitory deficit at the semantic/inferential level of interpretation.

For RAT verb sentences, the 2x2x2 ANOVA revealed no significant 3-way
interaction [F (1 ,45)=0.00]. However, the Group x Garden Path interaction was
significant [F(1,45F3.95, p=.05, n2 :08], such that RD children continued to Show
garden path effects for RAT verb sentences in the plausible as well as implausible
condition (whereas Control children exhibited substantially reduced garden path effects
in the plausible condition). Thus, similar to ADHD children, RD children continued to
have difficulty revising garden path sentences even when the syntactic structure afforded
by the RAT verb prevented them from having to make syntactic revisions in order to
revise semantic misinterpretations (thematic role assignments). Some evidence from
correlational analyses in RD children suggested that they may have trouble utilizing
syntactic rules to mitigate garden path effects.

ADHD vs. RD Commrison. A final post-hoe repeated measures ANOVA
explored the differences between ADHD and RD (including RD+ADHD) children. This
comparison revealed no Significant group differences. When the RD group was

constrained to exclude comorbid ADHD+RD children results remained unchanged. Note

130

that the informal check on differences between the comorbid ADHD+RD group using
two between group factors (ADHD: yes/no, RD: yes/no) revealed no significant between
group factor interaction.
Summary of Findings on the Garden Path Task Overall

Taken together, results for this task and the three main types of interference
effects examined (garden path effects, plausibility effects, and verb type effects) suggest
that children with ADHD and RD both have problems utilizing semantically-driven,
effortful inhibition to revise misinterpretations. These inhibition failures may, however,
have different sources. Thus, RD impairments may derive fi'om difficulties with complex
syntactic analysis and resultant resource limitations. On the other hand, ADHD
impairments may derive directly from failures in two kinds of inhibition (or suppression)
mechanisms. Which mechanism is involved may differ by which ADHD symptom
domain is at issue. Thus, it appeared that behavioral symptoms of inattention-
disorganization were associated with failure of automatic suppression mechanisms that
control “bottom up” interfering information before it has to be cognitively processed.
Hyperactivity was associated with weakness in a more effortful mechanism that should

enable suppression of information at the level of cognitive processing. .

131

 

 

 

 

 

 

 

 

 

 

 

 

 

 

222.8 22.2 222.8 22.2 22.8 22.2 222.8 22.2 222.8 22.2 222.8 22.2 .22...
222.8 22.2 22.8 22.2 222.22 22.2 24.8 22.2 222.8 22.2 2 2.8 22.2 22223.25.
82.8 22.2 222.8 22.2 222.8 22.2 8.8 22.2 222.8 22.2 222.8 22.2 22.23:
3:80-25.
222.8 22.2 2.28 22.2 22.8 22.2 222.8 22.2 22.8 22.2 222.8 22.2 22.23.25.
2228 22.2 22.8 22.2 62.8 22.2 222.8 22.2 82.8 22.2 22.8 22.2 22.23:
$9.
5.8 22.2 222.8 22.2 222.8 22.2 222.8 22.2 22.8 22.2 5.8 22.2 2223222:
228 22.2 22.8 22.2 222.8 :2 62.8 22.2 22.8 22.2 62.8 22.2 22.22.:
25822.8
222.8 22.2 222.8 22.2 2.28 22.2 222.8 22.2 222.8 22.2 222.8 3.2 23232.25
5.8 22.2 222.8 22.2 222.8 22.2 222.8 22.2 222.8 22.2 222.8 22.2 2.22.3.2
2.22
Now—=8“
Tan—42 Diana—d2 212 95283:— 352820 mmuZ
2:22.22 11.5-2.2 alli_eo-2:2< .8289 22222.25

 

 

 

 

 

 

22:80 unmoawaa B 2282 2222:2220 .2 032.2

132

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Azania-22322.28

 

 

2:22 525-5225252222
.2222 522 22:20

 

 

 

52:22.23 a.
5.2m eon-:20

 

 

222.8 222.8 22.8 222.8 22.8 222.8 222.8 22.8 222.8 224 8 2-2:2<
:2- 24.2 22.2- 22.2 :2 24.2 222 22.2 22.2 S 2-
222.8 22.8 22.8 222.8 22.8 222.8 22248 222.8 222.8 222 8 o-2:2<
22.2 22.2 22.2 22.2 22.2 :2 22.2. 22.2 22.2- 22 2-
222522
222.8 222.8 222.8 22.8 82.22 22.22 222.8 222.8 222.8 222.8
22 .2- 22.2 22.2 22.2 22.2 22.2 222 22.2 22 .2 22.2 22
222.8 222.8 a: .8 82 .8 222.8 222.8 22.8 222.8 32. 8 222.8 2-22222
22.2- 2.22 22.2 22.2 22.2- 22.2 22.2 22.2 22.2 22 2-
22.8 2212.8 222.8 2228 2 2.8 22222 222.8 222.8 222.8 22.8 o-2:2<
22.2- 22.2 22.2 22.2 22.2 22.2 22.2 22.2 22.2 22.2
22.8 22.8 A 2 2.8 222 .8 22.2% 2 2.8 d 2 .8 22 2.18 2:8 222.8 .2225
22.2 22.2 22.2 22.2 222 22.2 22.2 :2 22.2 E 2
:2 ma :2 Po :2 be :2 so :2 .20
52 22:22.82 5a: .5an 22.2.2 :2 222.22
422133

 

 

”2:80 3 2282.5 25322232 222 522 22220 22.2 228m 25222 2222.45 .Nm 22.2.2-

133

DISCUSSION

Controlling interference from competing thoughts or information in one’s mind is
crucial for accurately interpreting and functioning in the world. Children operate under
constant requirements for such suppression—for example, suppressing the thought of the
ice cream they were told they couldn’t have to help them control the behavior of asking
their mother for it over and over again, tuning out thoughts of what they will do afier
school to attend to their teacher’s instructions, or ignoring the television playing in the
next room to focus on their homework. In addition, the ability to identify and correct
one’s mistakes (i.e., self-monitoring), revise expectations to adapt to change, and even to
“forgive and forget” are crucial components to social-emotional functioning.

The inattention and disorganization that subsume many of these behaviors are key
diagnostic markers of ADHD and typical behavioral descriptors of these children.
Puzzlingly, however, empirical evidence for associated neurocognitive deficits has been
difficult to document within a laboratory setting (contrary to the extensive empirical
support for behavioral dysregulation and “output” deficits—cg, response disinhibition
on Logan’s Stop Task—that are related to behavioral impulsivity, emotional reactivity,
and disinhibition).

Cognitive inhibition (i.e., the capacity to resist interference from irrelevant
information and suppress activated cognitive contents) is one theoretically promising but
understudied candidate for attention and organizational impairments in ADHD. The
distinction between automatic (i.e., unconscious and resource-independent) versus
strategic (i.e., conscious, effortful, and resource-dependent) inhibition impairments is also

a key area for clarification. Of the few studies that have explored cognitive inhibition in

134

ADHD, most have largely failed to support such impairments (e.g., using directed
forgetting paradigms; Hamishfeger, 1995). Few studies had heretofore examined whether
inhibitory impairments could be detected in higher-order cognition, particularly in the
language domain. This lack of research is surprising since language plays a critical role
in self-regulation (Barkley, 1997) and social adjustment, ADHD has a high rate of
comorbidity with language/reading problems, and the cognitive field has produced a
modest body of literature on interference control in language processing.

Dempster (1993) posited that control over interference does not develop
uniformly across all domains of processing and, instead, develops from an “outward to
inward” direction, with children initially developing the ability to suppress motoric
interference, then perceptual interference, and, finally, linguistic interference (proposed
to dominate beginning around 6 years of age). Because most ADHD studies focus on
elementary school-aged children (ages 6-10), measures of interference in the language
domain may be most sensitive to the effects of inhibitory impairments on cognition
during this developmental stage. I

This study sought to utilize the enhanced sensitivity provided by cognitive
language comprehension tasks reliant upon putative inhibitory mechanisms“ (i.e., those
that involve resolving competition among competing cognitive contents or protecting
working memory from extraneous or irrelevant information) to examine four aims: 1)
whether children with ADHD exhibit problems with higher-order language

comprehension (even in the absence of a learning disability), 2) if inhibitory deficits in

 

‘7 Ability to suppress or overcome irrelevant information in cognition is the primary measure of interest in
this study. Whether this mechanism of suppression is truly inhibitory is a matter that continues to receive
widespread debate but is beyond the scope of this study. For current purposes, this process will be referred
to as either inhibition or suppression with the understanding that the true underlying nature of this
suppression (i.e., whether it is activation or inhibition-related) remains unresolved.

135

ADHD (well-established in the motor output domain) extend to the area of cognitive
processing and, in particular, emerge during language comprehension; 3) whether
children with RD also show evidence of inhibitory-related language impairments (beyond
generalized language deficits), as is suggested by the cognitive literature, and 4) whether
impairments on inhibitory language tasks, if they exist in ADHD and RD, clarify the
etiology of the high rate of co-occurrence between the two disorders.

These aims were explored with two cognitive tasks of language comprehension
believed to require inhibition and which varied according to their requirements for
automatic versus strategic/effortful suppression. The Ambiguous Words Task examined
utilization of both automatic and effortful inhibition for interference control in language
comprehension. The Garden Path Task, on the other hand, assessed ability to utilize
strategic and effortful control to revise misinterpretations.

Briefly, results of the current study supported the existence of language-related
inhibition impairments in children with ADHD in both expected and surprising ways. As
predicted, higher levels of inattention (or ADHD diagnosis) were associated with poorer
utilization of “top-down,” effortful, inhibitory control of irrelevant information during
language comprehension. Somewhat less expected, girls but not boys with ADHD
exhibited additional difficulties with relatively automatic inhibition of irrelevant
information during language processing.

Although analyses for RD children were necessarily preliminary and exploratory,
findings suggested that primary problems with lower-order language produced patterns of
impairments on inhibitory-related language processing much the same as those found in

ADHD children. Compared to Controls, children with RD exhibited weaknesses in both

136

automatic and effortful inhibition that were dimensionally related to reading ability (but
not attention) within that group. Preliminary findings therefore suggested that lower-
order language deficits might affect higher-order inhibition of irrelevant information in
language comprehension in those children.

Discussion now moves to elaboration of experimental findings, their implications
for understanding ADHD and RD, and future studies. Findings for interference control on
the Ambiguous Words Task will be addressed first and will then be followed by findings
for revision of misinterpretations on the Garden Path Task.

Interference Control
Relationship of lnattention to Interference Control

Current findings revealed that children with ADHD do indeed have difficulty
resolving interference from competing verbal information during higher-order cognition.
Functionally, this difficulty affects reading and listening comprehension—specifically,
the ability to draw conclusions or make connections about material within text (or
discourse).

Thus, on the Ambiguous Words Task, findings for sentence-level lexical
ambiguity resolution in the sample of ADHD and Control children revealed an
association between higher inattention (or ADHD diagnosis, but not hyperactivity) and
weaker interference control‘g—manifested by both more errors (i.e., higher Costs) and
relatively slower response speed”. Such impairments are consistent with deficits in

“cognitive inhibition,” in that previously activated cognitive contents remain in working

 

‘8 Following different-meaning as well as no-meaning/nonsense primes (in contrast to Control children
who showed modest facilitation—i.e., decreased error rates). Performance in the nonsense condition
therefore actually also may be a byproduct of inhibitory deficits.

‘9 Recall that RT for all subjects was facilitated for targets (compared to primes) but ADHD children
experienced less facilitation.

137

memory and then interfere with subsequent processing (Hamishfeger, 1995). If results
are, in fact, a byproduct of poor cognitive suppression, then the pattern of accuracy and
RT costs can perhaps be understood in the following way. Consistent with exhaustive
access models of lexical ambiguity resolution, all groups (regardless of clinical
diagnosis) experienced continued activation of both homonym meanings afier prime
processing (demonstrated by relative facilitation in mean RT to altemate-meaning
targets). Sustained activation of both homonym meanings produces competition during
target presentation. The success with which the now context—inappropriate dominant
meaning is suppressed during subordinate-meaning target access then influences the
resolution of this competition and is dependent on the efficiency of interference control
mechanisms (Neill, Valdes, & Terry, 1995). In the context of weaker suppression
abilities, the result would be slower ambiguity resolution and more errors.50

Using a “horse race” analogy, relatively equivalent levels of activation for both
the prime (dominant) and target (subordinate) meanings of the homonym should lead to
increased chances of incorrectly selecting the dominant meaning for the target sentence
(i.e., the horse race ends prematurely with the incorrect dominant meaning reaching the
activation threshold for selection first or with the correct subordinate meaning never
achieving a high enough activation for selection) or to taking more time to access the
target meaning (i.e., a longer time for the subordinate meaning to achieve selection
thresholds; recall that RT data were used only when both prime and target responses were

correct).

 

5° Note that by this logic, the very mild costs obtained by ADHD children to targets following no-meaning
primes could be consistent with similar mechanisms at work in over-coming the slight advantage likely
received by the dominant homograph meaning (due to its greater frequency) in the absence of specific
context priming. Control children did not experience such costs in accuracy.

138

While deficits in interference control have never been studied in relationship to
language functioning in ADHD, such impairments are not without precedence in other
domains. Casey (2001), for example, found children with ADHD to exhibit deficits on a
stimulus selection task requiring suppression of previously attended stimulus attributes
and proposed that these deficits may be associated with impairments in dorsolateral
prefrontal cortex.

ADHD-related interference control problems in the domain of language
comprehension have widespread implications for the language functioning of these
children. Dempster (1993), for example, described a number of text characteristics that
can produce interference during reading, including unrelated statements about the same
topic, involvement of the same set of characters in different activities, ambiguous
messages with more than one meaning, and irrelevant or contradictory text information.
Thus, interference control problems would have a pervasive impact on the reading
comprehension abilities of children with ADHD and may explain prior empirical findings
as well as anecdotal clinical evidence of reading comprehension difficulties and sub-par
achievement in this population (Brock & Knapp, 1996; Lorch et al., 1999; Fomess,
Youpa, Hanna, Cantwell, & Swanson, 1992). Problems with language-related
interference control could also explain problems with organization, effectiveness of
internalized language in guiding or regulating behavior, and perhaps even to problems
with emotion regulation.

One primary means by which interference control problems may affect language
comprehension is through a reduced ability to make inferences. Mclnnes and colleagues

(2003) found children with ADHD to have impairments in the inferential aspects of

139

auditory passage comprehension (i.e., the ability to make connections among text ideas
and to “read-between-the-lines”) despite intact retention of concrete facts. Thus, current
findings extend those of Mclnnes and colleagues by suggesting that similar inference
difficulties (i.e., determining whether a sentence makes sense) can occur under relatively
non-demanding language processing conditions (i.e., sentence-level rather than passage-
level comprehension) and under relatively minimal demands for “off-line” retrieval of
information from memory. In particular, as posited by Gemsbacher’s (1997) Structure
Building Framework, failure to suppress irrelevant information when constructing
sentence-level representations (in this case the context-inappropriate meaning of a
homonym) may impair construction of accurate causal chains and lead to impaired
reading comprehension and inference failures (e.g., ADHD-related problems with
retaining irrelevant story details at the expense of more important ones [Lorch et al.,
1999]).

Other prior studies on language or reading comprehension in ADHD also find
impairments under conditions most likely to involve interference control. Lorch and
colleagues (1999) found that ADHD children had impaired recall of television stories
only under divided attention processing conditions and retained irrelevant story details at
the expense of more important ones. These authors hypothesized that failure to suppress
irrelevant details interfered with selection of important story units. Current findings are
consistent with this interpretation.

A final means through which cognitive inhibition and interference control
impairments would influence language functioning in ADHD children is through their

impact on internalized speech, which is thought to be critical for behavioral control (Berk

140

& Potts, 1991) and which is deficient in ADHD (Copeland, 1979). Internalized language
is thought to influence executive functioning and the development of cognitive
maps/schematas of the world that produce complex thought and abstract thinking
(Damico, Damico, and Armstrong, 1999; Barkley, 1997). Problems in interference
control could, for instance, affect this “internal language system” in much the same way
as it does language comprehension. Competition and interference from competing
thoughts could produce difficulties in the maintenance of coherent representations of
internal goals or in construction of a causal chain of steps to achieve these goals.
Interference might also produce difficulties in the use of language to reason through
problems (i.e., constructing a coherent, causally-structured, inner dialogue) and in the
verbal mediation of problem-solving or behavior. Thus, problems with interference
control in internalized language—if replicated—could have significant implications for
understanding executive functioning impairments in ADHD.

The current study raised the additional possibility of ADHD-related gender
differences in interference control impairments. A number of recent studies have
demonstrated that ADHD girls have the same executive functioning deficits as boys
(Nigg, Blaskey, Huang-Pollock, & Rappley, 2002; Houghton et. al., 1999), but few
studies have directly compared boys and girls. In this study, girls with ADHD
experienced the most severe interference, demonstrating twice the error rate (costs) for
alternate-meaning targets as non-ADHD children (boys and girls) and ADHD boys.5 I
These robust interference effects in ADHD girls are particularly intriguing, since they

suggest the possibility of an etiological mechanism unique to girls with the disorder.

 

5 ' Boys did, however, show evidence of problems with interference control in the no-meaning priming
condition and in RT data.

141

One possibility is that girls with ADI-ID have more severe inhibition impairments
than boys with ADHD. Nigg and colleagues (2002), for instance, found that ADHD-I
girls but not boys had response inhibition deficits, suggesting the possibility of more
severe deficits in diagnosed girls. Other evidence for wider-spread or more severe
cognitive impairments in girls with ADHD is findings of increased vulnerability to co—
occurring language/reading disorders--especially in those with ADHD-I (Berry,
Shaywitz, & Shaywitz, 1985; Love & Thompson, 1988; Tirsoh & Cohen, 1998).
Alternatively, ADHD girls with certain cognitive deficits could represent a biologically
distinct subgroup of ADHD children (Doyle, Faraone, DuPre, & Biederman, 2001).

Findings of more global or more severe inhibition deficits in girls with ADHD
would suggest the possibility of ascertainment biases (i.e., only severely impaired girls
are referred for treatment due to demonstration of fewer behavior problems) or
polygenetic transmission factors (i.e., girls must carry a higher genetic load and meet a
higher threshold before phenotypic impairment becomes evident; Berry, Shaywitz, &
Shaywitz, 1985) that influence the expression of the disorder in girls. Given the very
small group of girls in the current study, these findings will need to be replicated.

Overall, then, ADHD and inattention were associated with language-related
problems in interference controls2 Still unanswered, however, is the secondary question
of whether reading abilities have a similar relationship to interference control as does
ADHD, which might provide clues as to why ADHD and RD co-occurr so often. Because
data were analyzed separately for Control/ADHD and RD children, two questions have

relevance to understanding the relationship of reading and language to inhibitory control.

 

52 It is notable that these findings do not mirror the exact pattern of results found by Gemsbacher and
colleagues (2001) in their college-aged sample of poor comprehenders. Possible reasons for this
discrepancy will be discussed later in the Limitations section.

142

First, is reading related to suppression failures for normal variations in reading ability?
Second, is reading related to suppression failures in a language-disordered, RD, sample?

Although findings for children with RD must be regarded as preliminary, their
performance on this task offers some information regarding the relationship between
reading and inhibitory processes that may be useful in guiding future, larger studies.
Note that while the implications of RD findings could be discussed extensively in their
own right (e.g., as to their relevance to theories on language comprehension in the
cognitive literature), given the secondary nature of these analyses, most of the current
discussion will focus on how findings for RD and reading were similar or dissimilar to
findings for ADHD and attention problems.

Relationship of Reading and Language to Interference Control

The patterns of effects produced for reading and language abilities were clearly
distinct from those related to ADHD/inattention. Overall, when results for RD children
on both the auditory and visual task were considered together, these children appeared to
demonstrate a U-shaped developmental pattern of interference control on the Ambiguous
Words Task (see Figure 7 below). Thus, RD children with both the weakest and
strongest reading abilities experienced the most interference from competing cognitive
contents, whereas RD children with intermediate levels of reading ability (still impaired
relative to others their age), experienced dimensionally lower interference than their other
RD counterparts. One important caveat regarding interpretation of RD findings is the
possibility that differences in task demands inherent to the auditory versus visual version
of the task produced differences in response patterns. This possibility will need to be

examined further in future studies.

143

One possibility is that differences in resource allocation are associated with
relative differences in reading abilities. Thus, RD children who were the poorest readers
appeared to experience difficulty activating the subordinate, less-frequent, meaning of the
homonym (likely due to weaker language abilities and vocabulary), which consequently
led to greater interference when required to retrieve this meaning over the dominant

meaning.5 3

Figure 7. Conceptual Model for the Relationship of Reading4Ability to Interference
Control in Children with RD and Normal-Reading Controls.

Development of
Subordinate Activation

Cost

Auditory RD
Subjects

 

 

 

Reading Ability

 

’3 This interpretation is supported by findings of smaller facilitation effects on the auditory task (reported in
the appendix) when the less-frequent homonym meaning was primed in both the prime and target sentence
(i.e., “Benefits” for sentences presented in the “Same" condition).

5‘ Note that this figure depicts a conceptual integration of the positive and negative correlational patterns
demonstrated by RD children on the auditory vs. visual task versions and the positive correlation for
Control children.

Findings of reduced interference for RD children with intermediate reading
ability (still impaired relative to others their age in reading), suggested more accessibility
to both word meanings than the other RD children, along with a failure to utilize context
to constrain competing homonym meanings. Simpson, Krueger, Kang, and Elofson
(1994) suggested that such a finding may be attributable to a developmental transition
involving greater familiarity with less-frequent homonym meanings but lack of
knowledge for dominance hierarchies (i.e., relative frequencies) that results in activation
of both meanings.

Higher levels of interference as reading ability becomes stronger suggest
utilization of selective access to the context-appropriate homonym meaning (perhaps
through an automatic suppression mechanism; Gemsbacher, Robertson, & Werner,
2001). Note that normal, non-ADHD, children in this study demonstrated the same,
albeit weaker, association between better reading ability and more interference (i.e., the
right side of Figure 8).

Taken together, RD children’s performance on this task can be characterized as
proceeding fiom (a) high-frequency-driven lexical access, to (b) exhaustive lexical access
for both meanings (regardless of context), and then to (c) selective access of the context-
appropriate meaning (producing processing costs when required to then access the
context-inappropriate meaning). Since associations between interference and reading
ability were not related to absolute reading level (i.e., raw scores) and because Control
children showed a similar associational pattern, this finding may reflect differences in
resource allocation associated with reading ability (or vice versa) rather than with a

specific developmental trajectory.

145

The right half of the curve shown in Figure 8 is consistent with the findings of
Gemsbacher and colleagues’ (2001) study of college students (and as noted was also
found in Control children in this study). Findings thus partially replicated those of
Gemsbacher et a1 and extended them to a child sample. Given the similar patterns of
findings in both children and adults across their study and this one, current findings again
may reflect not merely a developmental trajectory towards constrained lexical access but,
instead, innate differences in resource allocation in strong versus weak readers. Thus, part
of what makes a good reader may be use of suppression to block out irrelevant or
competing meanings during comprehension (or more automatic decoding which then
frees up resources to direct towards suppression). Current findings of no association
between absolute reading level (i.e., raw scores rather than age-standardized scores) and
interference control suggests that this may be the case (although would certainly not rule
out a more subtle developmental effect).

Relationship of Reading and Attention Interaction to Interference Control

How then do findings for interference control in reading relate to those for
inattention? First, there was some suggestion that reading ability and attentional control
interacted to determine degree of interference control in language comprehension for I
children with normal reading (i.e., Non-RD). Thus, after removing the effects of IQ, the
interaction between attention problems and reading ability marginally predicted
vulnerability to interference on the Ambiguous Words Task.

The interaction between reading and attention suggested that attention problems
mediated the relationship between reading and interference control. Thus, in the absence

of attention problems the relationship between reading and interference control for

146

normal readers followed the pattern just discussed (see right side of Figure 8). However,
strong reading ability in the context of high inattention was associated with less
interference from competing meanings (i.e., suppression impairments)——similar to that
found in poor readers. For good readers, attentional abilities therefore appeared to
mediate interference from competing semantic representations, suggesting that effective
allocation of attention may in fact play a role in effective text comprehension. Good
readers likely have more automatic word decoding, but if they have high levels of
inattention they too experience reduced resources to allocate for suppression of irrelevant
information (thus performing like poor readers—with or without attentional
impairments).

What these findings suggest is that 1) reading ability influences interference
control--or vice versa-— and 2) a possible developmental trajectory exists for reading
ability and resource allocation in which automatic suppression mechanisms are not
utilized until more advanced reading development.55 lnattention influenced interference
control independently of reading effects, yet also interacted with them. Note that weaker
verbal working memory (on the expressive language task), like high levels of inattention,
was also strongly and consistently associated with interference from competing meanings
across clinical groups. Like inattention, then, smaller working memory capacity
(regardless of clinical diagnosis), was associated with difficulty controlling interference

from competing cognitive contents.

 

55 Despite no interference findings associated with absolute reading level or age, most children in this
sample activated both homonym meanings (as evidenced by mean facilitation in RT from prime to target in
all conditions for all groups). Thus, at this developmental stage—with the exception of the best readers-
most children (regardless of their clinical diagnosis) may not yet have developed automatic suppression
mechanisms that constrain lexical access.

147

In summary, one possibility suggested by current findings is that suppression
failures evidenced by children with ADHD and RD have a different locus. Thus, children
with ADHD or individuals with lower working memory capacities may have difficulty
resolving interference due to impairments in strategic inhibitory processes that require
utilization of context to suppress competition. In contrast, RD children may experience
problems with more automatic suppression (which only appear at a more advanced level
of reading proficiency—perhaps after the act of decoding and reading attains high enough
levels of automaticity—and occurs as a byproduct of lexical selection) or possibly
activation (i.e., ability to quickly enhance and retrieve semantic or lexical information),
but they do not have difficulty using context to strategically or effortfully suppress
activation once it occurs.

Revision of Misinterpretations

Performance on the Garden Path Task examined a second aim of this study: the
ability of ADHD and RD children to consciously and effortfully revise or update
assumptions when given cues that they had made a misinterpretation. Just as interference
control is necessary for individuals to develop and maintain coherent thoughts and ideas
and to regulate and direct their behavior (e. g., through internally represented goals), an
individual’s ability to revise or update assumptions is also crucial to cognition and
behavioral regulation, and impairments in this ability would be likely to produce
difficulties in social interaction and oppositional behavior (which might actually stem
from deficits in cognitive processing rather than “true” oppositionality). Everyone, for
example, has moments of speech misperception that produce problems in understanding a

speaker’s intended meaning. Often in such situations, one realizes the error from the

148

subsequent context of the conversation and is able to revise the initial misinterpretation to
produce a coherent understanding of the discourse. In contrast, inability to adapt to these
kinds of conversational “updates” would cause an individual to seem perpetually socially
“out-of-step”—a common descriptor of ADHD children.

The process of revising misinterpretations differs somewhat from the interference
control construct just discussed in that it requires 1) more conscious allocation of
resources towards effortfirl or strategic inhibition and 2) inhibition at the syntactic as well
as semantic level for correct performance. Thus, unlike the Ambiguous Words Task,
performance does not rely merely on overcoming semantic activation. Rather, it requires
actually altering the syntactic representation (or causal chain) from which the semantic
activation and interpretation derives and therefore examines the role of inhibition at a
higher level of cognition.

Relationship of lnattention/ADHD Diagnosis to Revision of Misinformation

Comparisons of ADHD and Control children revealed that both groups produced
a large number of errors when misled into a semantically plausible sentence
misinterpretation (e. g., answering yes when asked, “Did Joe walk the dog?” following
“As Joe walked the dog napped.”).56 However, ADHD children were much less able to
utilize semantic implausibility (i.e., seemingly illogical sentence meanings) to overcome
misinterpretations arising from syntactic garden paths. Such a finding suggests that
Control children utilize effortful, strategic, or executive control to suppress an initial
interpretation when given more contextual information to do so (i.e., when the

interpretation arising from the garden path cannot be semantically licensed). In contrast,

 

56 Recall that even “plausible” sentences produced semantically implausible misinterpretations but that the
verb—object relationship was plausible.

149

higher levels of hyperactivity, inattention and ADHD diagnosis were associated with
deficits in these strategic processes and were robust to age and IQ control.

The perseverance of these implausible misinterpretations in children with ADHD
was quite remarkable. Thus, children with ADHD continued to maintain
misinterpretations under highly illogical circumstances— for instance, answering “yes”
when asked, “Did Meg play the Shoes?” following the sentence, As Meg played the shoes
got shined. What’s more, it is unlikely that these findings in ADHD children can be
attributed to globally decreased effort or to over-reliance on pragmatic inferences (i.e.,
general reasoning), since to answer ‘yes” incorrectly to the question, “Did the ape eat the
tree?” or “Did Meg play the shoes?” is more illogical and in greater violation of a general
reasoning strategy than to answer “no” (which would have been the correct answer).
Thus, the most likely explanation for these findings is that failures in strategic
suppression mechanisms affect children with ADHD’s ability to revise causal
relationships (and the resultant interpretations) once they have been established. One
important caveat to this interpretation, however, is that—to prevent strategy effects and to
encourage effortful processing—filler sentences occasionally required “yes” answers to
similarly implausible assumptions (As the girl played the fox her mom left the room. Did
the girl play the fox?) Experimental design might therefore partially account for children
with ADHD’s decreased sensitivity to semantic implausibility, but does not explain their
poorer performance relative to Control children (i.e., Controls did benefit from semantic
implausibility) nor does it explain their decreased benefit from RAT verbs (i.e., under

conditions with a lower revision threshold).

150

Findings regarding difficulties revising misinterpretations are relevant to the
functioning of ADHD children in a multitude of areas. Consider, for example, the
ADHD child who gets bumped while standing in line. Such a child, upon receiving an
apology and being told that the offense was accidental, would have an exceedingly hard
time revising an initial interpretation that he/she had been purposefully pushed and might
also fail to benefit from external cues in the environment that might demonstrate that the
push had been accidental (e.g., incorporating perceptual information about the other child
tripping prior to the bump). These kinds of difficulties in children with ADHD might
therefore contribute to the development of hostile attribution biases, a precursor to
oppositional defiant or conduct disordered behavior (Crick & Dodge, 1994). Future
studies could examine this hypothesis by determining whether deficits in revision of
misinterpretations in the language domain are associated with higher levels of
oppositional or conduct disorder behaviors.

As well as examining the effect of contextual cues on ADHD children’s
implementation of strategic suppression, this study also examined whether ADHD
children’s performance would improve when inhibitory processing demands were
reduced (i.e,. for sentences using Reflexive Absolute Transitive [RAT] verbs, which were
easier to revise than sentences with Optional Transitive [OT] verbs). For instance, in the
RAT-verb sentence, “As Daday shaved the cat sat in the tree, ” the relationship between
“shaved” and “cat ” should be easier to alter or “delete” than the relationship between
“walked” and “dog ”in the OT-verb sentence, “As the man walked the dog napped” (for

reasons to be discussed shortly).

151

Findings revealed that both groups of children found it easier to revise
misinterpretations when syntactic demands were less complex (e. g., in the “shaved the
cat ” RAT-verb sentence above), but that ADHD children benefited less than Control
children (thus continuing to show large inhibition failures). There was some indication
that this effect was specific to hyperactivity rather than inattention symptoms, suggesting
that more severe deficits in effortful inhibition might underlie this finding. Alternatively,
as with findings for the Ambiguous Words Task, inattention may have been the primary
factor associated with difficulty in revision of misinterpretations (i.e., interference
control), but high levels of hyperactivity may have been related to decreased processing
eflort (e. g., more superficial processing of sentences), leading to continued inhibition
failures even when syntactic and inhibitory demands were reduced (as in the RAT
sentences).

One interpretation for why all children improved their performance on RAT
garden path sentences (despite the lesser benefit for ADHD compared to Control
children), is that improved performance may reflect the relatively greater ease with which
the inappropriate interpretation (i.e., verb-noun thematic role assignment) can be
overcome or “deleted” when either a) the subordinate verb is more weakly connected to
the object (perhaps through competition from the implicit object [i.e., himself/herself]
during syntactic parsing) or b) when the incorrect thematic role assignment is not
necessary to maintain the causal structure of the sentence—perhaps because the initial
thematic role assignment can be replaced with an implicit reflexive object

(herself/himself)57 For example, consider again the sentence, “As Daddy shaved the cat

 

’7 Since “plausible” condition sentences continued to produce implausible misinterpretations (e.g., As Betty
woke up Daddy was sleeping in the den). Thus, effortful and semantically-driven inhibitory mechanisms

152

sat in the tree. ” It might be possible to delete the association between “shaved” and
“cat ” by altering the sentence to “As Daddy shaved [himself] the cat sat in the tree. ”

Taken together, then, current findings suggest that: 1) Children with ADHD are
less successful than Control children in using either implausibility or RAT verbs to revise
garden path misinterpretations; and 2) the locus of these revision failures may relate to
ineffectiveness of effortful or strategic inhibition, such that ADHD children find it
difficult to “delete” incorrect thematic role assignments or to revise verb argument
structures in conditions in which Control children can do so.58 These findings thus
replicate many of the findings from the Ambiguous Words task and suggest a consistent
relationship between inattention, weaker cognitive inhibition, and poorer language
procesSing. Findings on the Garden Path Task additionally suggested that hyperactivity
might—not surprisingly—be associated with decreased processing effort under more
difficult or more complex processing conditions (leading to continued inhibitory deficits
even under conditions with reduced processing demands).

These findings have broader implications in the context of language
comprehension or reading comprehension deficits in ADHD, as well as for possible
methods of remediating these difficulties. With regard to the more global implications

for the effects of inhibitory deficits in language comprehension, Johnson and Seifert

(1994) examined the persistence in memory of inferences based upon corrected

 

could still be drawn upon. In contrast, Christianson et al.’s study, which had a condition that induced truly
plausible misinterpretations—e.g., As Betty woke up Daddy stretched his arms—showed continued garden
path effects for plausible RAT verb sentences that were similar to those found for OT verbs (contrary to
current findings).

58 An alternative interpretation might be that RAT inhibition involves a non—effortful process—perhaps
more syntactically and automatically driven-- and ADHD children then don’t receive any additional benefit
from contextual-level inhibition. This interpretation does not fit with the fact that ADHD children clearly
receive some benefit from contextual level inhibitory processes (these are just not as large as for Control
children).

153

misinformation during comprehension of “garden path passages” (i.e., passages that
presented misinformation, which was later corrected). Consistent with current findings,
misinformation was maintained in passage memory when it afforded causal structure to
the inference (e.g., when a character mentioned as a crime suspect is later said to have
been cleared of wrong-doing but no new suspect is named). This effect was mitigated, but
still persisted, when the misinformation was replaced by other causal information, rather
than when it was merely negated (for instance, when the initial crime suspect is absolved
of the crime and a new perpetrator is named).

Clearly, then, impairments in strategic inhibitory mechanisms that affect language
comprehension would have wide-spread implications for general functioning. At the
minutest level, such difficulties would seriously impact ability to draw correct inferences
(i.e., meanings/conclusions) from both written and spoken language and might also
contribute to problems with language production (e. g., to processing disfluencies in
natural speech). At a more global level, social interaction and communication would,
perhaps, be most affected, due to their fluidity and constant demands for “on-line”
accommodation to the communications of others. As mentioned earlier, such
impairments would also have direct implications for drawing accurate conclusions about
others’ behavior (e. g., incorporating the contextual information about the boy tripping to
understand that his pushing you was not intentional) and for updating and changing one’s
conclusions when new information becomes available. Functional impairments in these
processes put a child with ADHD at- risk for a host of other problems, including
development of hostile attribution biases associated with Oppositional Defiant and

Conduct Disorders (as mentioned earlier; Crick & Dodge, 1994) and school failure, and

154

could also explain many ADHD-related executive functioning deficits (e. g., in self-
monitoring). With regard to implications for remediation, these findings suggest that
children with ADHD should be taught to question their interpretations about events and
to search for possible “replacement” information (e. g., the information about the boy
tripping) that might help them to revise incorrect interpretations.

’ Discussion now moves to findings regarding the role of reading and language to
revision of garden path misinterpretations.

Relationship of Reading and Language to Revision of Misinformation

Overall, as on the Ambiguous Words Task, the pattern of results in RD children
on the Garden Path Task showed striking similarity to the overall ADHD pattern, but
appeared to derive from different core deficits. For the most part, reading achievement
per se was not related to effortful suppression on the Garden Path Task. However, broad
estimates of language abilities and RD diagnosis were related to suppression failures and
to difficulties revising misinterpretations.

Like findings in ADHD, language problems/RD diagnosis was associated with
poorer ability to utilize effortful inhibitory control to suppress an initial misinterpretation.
Thus, RD children were much less able to utilize semantic implausibility (i.e., seemingly
illogical sentence meanings) to overcome misinterpretations arising from syntactic
garden paths. Such a finding suggests failures in utilization of effortfirl inhibitory control
to suppress an initial interpretation when given increased incentive to do so (i.e., when
the interpretation arising from the garden path cannot be semantically licensed). Also

like ADHD children, RD children benefited less from reduced syntactic demands (i.e.,

155

RAT verb manipulation) in revision of misinterpretations (thus continuing to show large
garden path effects).

However, RD children demonstrated two important differences in their
performance from ADHD children. First, RD children were more sensitive to the effects
of plausible verb-object relationships in garden path conditions (regardless of verb type)
than Control children (something that was not true of the ADHD group). Consequently,
they were more vulnerable to being misled into a garden path misinterpretation when the
verb-object relationship was plausible (e. g., “the ape ate "). Such a difference may
suggest that underlying language-related difficulties in processing complex syntax leads
to a concomitant reliance on a more piece-meal, general pragmatic reasoning strategy that
is based on analysis of verb-object relationships to aid comprehension. This interpretation
is supported by a second primary difference between the performance of ADHD and RD
children: increased sensitivity to implausibility in comprehension of non—garden path
control sentences. Thus, RD children were able to “recoup” their prior impairments
when they could rely on a general pragmatic inference reasoning strategy (based on verb-
object implausibility) afforded by non-complex or misleading syntax.59

Taken together, then, RD and language findings are associated with 1) reduced
benefit from implausibility in correcting garden path misinterpretations, suggestive of
impairments in effortful or strategic inhibition, like ADHD, and 2) increased
vulnerability to being misled by semantically plausible misinterpretations, reflecting a
possible reliance on more piece-meal, general pragmatic inferencing strategies (e. g.,

regarding verb-object relationships) under conditions requiring comprehension of

 

59 Note that dimensional analyses found similar findings related to inattention, however these results did not
hold during categorical group analyses, so may not have been specific to ADHD.

156

complex syntax, and 3) increased reliance on contextual cues (e.g., semantic

implausibility vs. plausibility) to compensate for language processing weaknesses.

Conclusions and Preliminary Interpretations.

Taken together, findings for ADHD and RD children provide some preliminary
ideas regarding the co-occurrence of the two disorders. Failure to effectively utilize
semantic information to recover from garden path misinterpretations was the primary
finding associated with both inattention and reading problems on this task. Both ADHD
and RD children thus exhibited impairments in effortful inhibitory control on this task.
However, RD children also seemed to have more generalized problems parsing and
comprehending complex syntax (i.e., they were more vulnerable to making garden path
misinterpretations) and seemed to rely more heavily on semantic implausibility and
pragmatic inferencing to help them compensate for these difficulties. Thus, one
possibility is that RD children had to allocate more resources to syntactic parsing and
analysis, leaving fewer resources to suppress and revise misinterpretations. Such an
interpretation, if borne out in future studies, would suggest that RD children are more
likely to exhibit inhibition impairments when their processing resources are taxed by
more lower-order language deficits. Such processing capacity “overloads” would cause
them frequently to demonstrate behavior that looks inattentive or to produce problems
with planning and organization analogous to those found in ADHD. As a final element to

discussion of current findings, further consideration will now be given to findings

157

regarding the overlap of ADHD and RD in the current sample and to a proposed theory
for their co-occurrence.

Integration and Proposed Theory for the Co—Occurrence of ADHD and RD

C o- Occurrence of Inattention and Reading Problems

At a behavioral or phenotypic level of analysis, those with high levels of
inattention were at risk for reading problems. Thus, with the stringent diagnostic criteria
employed in this community-recruited sample, 17% of ADHD children also met criteria
for RD. This comorbidity rate was lower than many prior studies but still far above
chance. Children with ADHD-I were more than twice as likely to meet criteria for RD as
children with ADHD-C (30% vs. 13%, respectively). Moreover, ADHD children without
an RD diagnosis had poorer reading comprehension than did Control children, although
groups did not differ in basic reading skills (e.g., word recognition, spelling, and
phonological decoding) or language abilities.

Findings also supported high rates of inattention in those with reading problems.
Thus, 35% of RD children also met criteria for ADHD, with approximately equal co-
occurrence rates for ADHD-C versus ADHD-I (20% vs. 15%, respectively). “Pure” RD
children (i.e., those without an ADHD diagnosis) also had much higher levels of
inattention than did Control children (on average, 5 of 9 DSM-IV inattention symptoms).

Proposed Theory for Future Study

Taken together, then, expected ADHD-RD group distinctions (e.g., in behavioral
hyperactivity and lower-order reading and language skills) underscored the etiological
independence of the two disorders. They were also distinguished by a clearly different

pattern on some types of interference control measures in this study. However, findings

158

also revealed possible reasons for the frequent behavioral phenotypic overlap between
ADHD and RD. What emerged was the possibility that inhibitory impairments may
produce problems with higher-order language processes in ADHD children. Likewise,
lower-order language problems (e. g., in phonological processing) may tax the cognitive
resources of children with RD, producing impairments in inhibition-reliant language
processes and secondary attention problems.

One primary contribution of this study is that it highlights the fact that different
core deficits in ADHD and RD can also lead to similar cognitive outcomes (or secondary)
deficits. What remains to be answered is whether these core deficits themselves can
actually produce the other disorder (rather than common secondary symptoms). Although
this current study cannot directly answer this question, it is interesting to speculate about
pathways by which such comorbidity of ADHD and RD could occur.

Pathway from ADHD to RD

With regard to a possible pathway from ADHD to RD co-occurrence, attention
problems in ADHD might create working memory/capacity limitations that—combined
with even weak phonological processing or phonological awareness skills—could cause
significant problems with phonological decoding and manipulation of phonemes. If so,
ADHD children should show increased phonological decoding deficits as working
memory/effortful processing demands increase but have no problems with easier
phonological awareness tasks. It should also be possible to see dimensional differences
in phoneme awareness in ADHD children, with the tendency for RD diagnosis to increase
as phoneme awareness skills decline. One other possibility is that children with ADHD

spend less time engaged in language or reading-related activities (due to their attentional

159

difficulties), and this lack of “experience” or practice with language (which would
prevent development of greater automaticity in language functions) leads to requirements
for more effortful processing which children with ADHD are unable to deploy because of
their inhibition deficits. As a result, inhibition and decreased language facility would
interact directly without the need to posit working memory impairments as an
intermediary (MacDonald & Christiansen, 2002).

If all of these factors were borne out, one might expect that ADHD children
would have a reduced threshold for developing RD, such that weaknesses in phonological
awareness (which are believed in the empirical literature to occur along a continuum)—
even if not as severe as those found in “pure RD”—and their interaction with the severity
of attentional impairments could produce a true reading disorder or dyslexia in ADHD
individuals. Future studies can test these hypotheses by using more sophisticated
measures of phonological processing (Fitch, Miller, & Tallal, 1997) and examining
whether the performance of ADHD children changes according to working memory and
effortful processing demands. It may also be interesting to examine these processes in
much younger children who are just beginning to develop reading in order to gain clues
as to developmental trajectories.

Pathway from RD to ADHD

Could a similar pathway lead from RD to ADHD? Current findings suggest that
inhibition and suppression deficits in RD children might occur only on language-
mediated or perhaps also temporal processing tasks (Fitch, Miller, & Tallal, 1997)).60 In

this scenario, RD children may experience capacity or resource limitations during

 

6° Assuming that RD doesn’t also involve generalized problems with automaticity and sequencing that are
not solely language-dependent. In the cognitive literature on RD deficits, this debate has not yet been
resolved

160

language processing that would be manifested in secondary deficits in effortful
processing and inhibition and might not be expected to occur when language processing
was not required. Thus, behavioral diagnosis of ADHD might be made due to the
secondary manifestations of attention and organization problems that—rather than
deriving from a primary attentional or inhibition deficit—arise from lower-level language
processing impairments (e. g., phonological decoding) that produces resource limitations.
If that is the case, then this pathway would produce real inhibition and attentional
impairments in children with RD but only under specific processing conditions (e. g.,
language-dependent tasks) and would therefore not be a “true ADHD”——i.e., would not
involve a core inhibition impairment.

Most studies finding inhibition deficits in RD children have utilized tasks with
language demands (e.g., CPT’s and Logan’s Stop Task—both of which require speeded
and automatic processing of letters; Purvis & Tannock, 2000; Willcutt et al., 2001; Nigg,
1999). Only one known study has examined response inhibition in RD children using
nonverbal stimuli, and this study produced equivocal results (Moores & Andrade, 2000).
To determine whether RD children demonstrate primary inhibition deficits independent
of those produced by language processing weakness, future studies should further
examine these processes using nonverbal tasks (e. g., antisaccade or covert orienting
tasks).

Summary

These independent pathways can be summarized and integrated in the following

ways (see Figure 8 below). Increased effort required for word decoding in individuals

with reading problems likely leads to resource limitations (i.e., reduced processing

161

capacity or “bottlenecks”) in other areas. For instance, increased effort applied towards
reading decoding would leave less available resources for reading comprehension and
would also limit the available working memory “space” for such additional operations
(e.g., creating and holding a coherent semantic and representational structure, per
Gemsbacher’s (1997) Structure Building Framework). Children with ADHD, in contrast,
experience these resource limitations directly from their attentional/inhibitory deficits
which would also limit the size of working memory space and could produce processing
bottlenecks and interference with efficient language comprehension or reading
comprehension. Thus, both groups do experience true suppression failures, despite the

fact that the origin of their suppression

162

 

 

     

Eco: B...
“.3323ch A|1lrv

:éwieccna—

 

 

2“..wa
+DEQ<

 
         

Q2 +
DEA—AV

  

mESA—Emm
.83.—822.:

 

       

.
cu
..
.
.-
‘-
1
.‘I‘
a
.a
a
1
a
.-
.
..

 

=82:
«wan—9:3 £0qu
mm 5533:—

Q=D<

 

 

 

.9222 a 92 2a 92 a 9:2 822 measure 227.82 22 3222 _aueoofi .2 as»;

163

deficits differ. Both groups also experience end difficulties in language processing
despite the fact that they derive from different sources.

Given that reading deficits in dyslexia do not occur in a vacuum (i.e., additional
cognitive deficits—particularly in phonemic awareness—underlie these reading
problems), it is also important to consider the role played by inhibition in language
development and the manner through which inhibition impairments might then affect
reading development. One possibility is that deficits in phonemic awareness and
phonological processing could lead to problems in orthographic decoding—at least in
part—via inhibition failures.

In children with RD, difficulties deciphering phonemes auditorily may lead to a
general lack of knowledge about the properties of phonemes in speech and about how
they combine to form words. Developmentally, this difficulty might then lead to
increased processing requirements/effort needed for the linguistic aspects of language
processing and result in limited resources available for other higher-order cognitive
processes like learning and storing phoneme-grapheme correspondences, orthographic
decoding, development of automatic word recognition, and reading comprehension
(Crain & Shankweiler, 1988; Shankweiler et al., 1999). Later in development these
resource limitations may appear as frank inhibition impairments, as children with RD will
have limited resources to utilize for other higher-level cognitive language processes (not
just related to reading) such as interference control. Some support for this assertion is
found even for dyslexic adults in the neuroimaging literature. Brunswick, McCrory,
Price, Frith, & Frith (1999), for example, reported that college students with RD

demonstrated reduced activation in left posterior inferior cortex and increased activation

164

in a pre-motor region of Broca’s area while reading pseudowords, despite similar
accuracy to non-impaired comparison students. The authors attributed this result to
problems with highly automated aspects of reading (specifically, lexical retrieval) and an
associated increased need for effortful compensatory strategies.

Similar processes might also lead to language and reading problems in
individuals with core inhibition deficits. For instance, inhibition deficits also lead to
cognitive resource limitations that might make it hard to hold and manipulate phonemes
when decoding an orthographic representation (even in the absence of diminished
knowledge about phoneme-grapheme correspondences or linguistic properties of words),
leading later in development to poorer word decoding, failures in automatic retrieval, and
deficits in reading comprehension—despite adequate underlying language abilities.

Limitations

A number of limitations inherent to the current study must be considered.
Ambiguous Words Task

The locus of inhibition impairments for ADHD children on the Ambiguous
Words Task differed from those predicted by Gemsbacher, Robertson, and Wemer’s
findings (2001) upon which the current study was based. Several factors are important in
considering this discrepancy, including specific aspects of task construction, variability
among studies in the lexical ambiguity literature regarding what pattern should be i
“expected,” and the possibility of developmental effects.

This study used only pairs of dominant-frequency prime stimuli with subordinate-
frequency target stimuli. In hindsight, using these polarized homonyms produced a series

of confounds for interpreting results. Simpson and Adamopoulos (2001), for instance,

165

suggested that inhibition may not be required for processing dominant-meaning
homonyms when the homonym occurs late in a sentence (i.e., afier the biasing context, as
occurred here), because there is relatively little competition between meanings. Dixon
and Twilley (1999) also argued that lexical ambiguity resolution is a function of both
bottom-up stimulus driven frequency activation effects and top-down contextual
inhibition effects, with inhibitory mechanisms playing less of a role in access to high
frequency homonym meanings.

These studies therefore suggest that the dominant prime/subordinate target stimuli
used in the current study, as well as the possibility of stronger contextual biases of current
sentences, might produce different patterns of effects than those found by Gemsbacher et
a1.“ Processing of “nonsense-meaning” primes also created problems for children in this
study and confounded measurements of “Cost” from a neutral baseline. To measure
inhibition effects confidently, in future it may be better to use balanced-frequency
homonyms and to compare them to truly neutral meaning prime sentences (e. g., She
threw the spade). As pertains to current findings, it seems possible that—rather than
measuring automatic suppression that occurs as a byproduct of prime activation—the
current study measured somewhat more effortful interference control abilities. Thus, it
seems likely that the current task continued to measure inhibition in these children, but
not in the way initially anticipated.

An additional confound—at least for directly comparing effects between the
present study and those of Gemsbacher et. al.—concems likely developmental

differences between adults and children in mechanisms of lexical ambiguity resolution.

 

6' Note, however, that Gemsbacher, Robertson, and Werner (2001) did utilize a dominant prime
and subordinate target condition and found no differences in the overall pattern of results versus
their other experimental manipulations. ‘

166

Prior studies in the cognitive literature have revealed that, on the one hand, 2nd and 4th
grade children have unconstrained lexical access in the absence of a biasing context (e. g.,
they show continued facilitation for both dominant and subordinate homonym meanings
even after a 2000 msec delay), whereas older children inhibit the less-frequent meaning
(Simpson & Foster, 1986; Simpson & Lorsbach, 1983). Other evidence suggests that
younger children (5th grade or younger) rely more heavily on sentence-level semantic
context to constrain lexical access than do older, lZ-year-old, children (who rely more
heavily on frequency to determine lexical access). In fact, Simspon and colleagues
(1994), reported a “U-shaped” developmental curve for development of lexical ambiguity
that was similar to the one reported here. Taken together, then, current findings appear
consistent with the literature on lexical ambiguity resolution, despite their difference from
those of Gemsbacher and colleagues.

Finally, perhaps the most important limitation and qualification to current
findings for the Ambiguous Words Task must include an acknowledgement that a role for
inhibitory mechanisms in language processing is far from established in the cognitive
literature. Thus, it is impossible to rule out activation-decay or parallel distributed
processing model explanations of ambiguity resolution and interference control
(MacLeod, 1991). Put another way, interference findings for ADHD and RD children on
this task could also reflect differences in relative activations of competing homonym
meanings, differences in the time-course of decay, and the occurrence of processing
bottle-necks that do not stem from inhibition impairments.

Garden Path Task

167

Overall, findings on the Garden Path Task were not subject to many of the
methodological and developmental confounds that occurred in the Ambiguous Words
Task. Thus, overall, children in this study responded very similarly to Christianson et
al.’s (2001) college-aged sample. Current findings do, however, merit one important
caveat. This task is a very difficult one—even for normally functioning young adults—
and likely relies upon many cognitive processes for successful performance (not just
inhibitory processes). Both ADHD and RD children demonstrated a strong reliance on
inferential and pragmatic reasoning strategies (i.e., the plausibility of verb-object
relationships). Consequently, it is important to consider the possibility that current
findings result, not from true inhibition impairments, but from generalized factors
associated with task difficulty. In this light, ADHD and RD differences among different
conditions may stem simply from the fact that these experimental manipulations made the
task easier. If this is the case, then current findings may actually result from non-specific
variation in task difficulty and to over-reliance on inferential processing to compensate
for these difficulties, rather than from specific inhibitory deficits.

Paradoxically, contrary to potential problems with task difficulty for ADHD and
RD children, garden path manipulations were too easy for Control children, causing
performance in a number of conditions to reach ceiling levels. Due to these ceiling
effects, it was difficult to determine whether some of the significant effects resulted from
true group differences or from an artificial limitation in range for higher-functioning

subjects.

168

Additional Considerations

A number of broader methodological limitations are relevant to the study. One
obvious limitation was the use of both a visual and auditory task version which makes all
comparisons to RD children who received the auditory task of questionable
generalizability. In addition, from a statistical standpoint, the very large number of
analyses performed in this study increases the likelihood that some findings are the result
of Type I error. High variability in performance for all groups (evidenced by large
standard deviations) made it difficult to detect group differences. Furthermore, unequal
group sizes raise the possibility that some effects may have been under- or over-
estimated. The use of mathematical corrections to control for task inconsistencies (such
as the “corrected Cost” score computed in the Ambiguous Words Task) may also have
altered the validity of current findings.

One final limitation concerns this study’s small sample size. This limitation was
particularly true as it pertained to the number of ADHD girls and limits the
generalizability of the gender effect on the Ambiguous Words Task. Small sample size
also prevented analysis of age-group differences. Given the wide age-range used in this 1
study (ages 8-14), it would have been preferable to have divided children into groups of
younger and older children, thereby allowing for a more systematic exploration of
developmental effects. It is also relevant to note that the average age of this sample was
actually younger than was ideal (average age in all groups was 10), making it possible
that these children hadn’t fully developed the inhibitory mechanisms that were being

explored (Hamishfeger, 1995).

169

Conclusions

Despite these limitations, the current study breaks new ground in at least two
important ways. One, it shows that cognitive tasks of inhibitory language processing can
be adapted for use in a child population and are sensitive to differences in inhibitory
functioning. Two, this study provided new findings regarding language processing
problems in children with ADHD that are related to inhibitory control mechanisms
previously identified for motor control. Importantly, in contrast to null findings in many
previous studies of cognitive inhibition in ADHD, the current study indicated that these
impairments can be detected in higher-order language processing in these children. Thus,
null findings in prior studies of interference control and cognitive suppression could
reflect the modality in which suppression is being measured (e. g., perceptual, motor, or
language), the level to which attentional selection is directed (e.g., Balota, Cortese, &
Wenke, 2001), and the requirements for top-down, effortful, and semantically-driven
processing. Finally, similarities in ADHD and RD deficits suggest areas for future study

that may explain the disorders’ high rates of co-occurrence.

170

APPENDIX A

Demographics and Group Descriptions
lnattention and Hyperactivity Symptoms
Inattention
The interaction between ADHD group and RD group was significant for total
number of inattention symptoms by the “OR” diagnostic algorithm [E (2, 76)=10.74,
p<.001]. As expected, children with ADHD-C and ADHD-I had more symptoms of
inattention (regardless of their RD status) than did Control and pure RD children [ADHD
main effect: F (2, 76)=89.87, p<.001]. Also consistent with prior findings in RD samples,
RD children without a diagnosis of ADHD had more symptoms of inattention than did
Control children [RD main effect: F(1,76)=11.30, p=.001]. This effect held even after
excluding the five RD children who had met criteria for situational or subthreshold
ADHD [F(1,32)=10.67, p<.01, mean RD inattention symptoms: 2.3 :1; 1.9]. Similar
patterns of results were found in comparisons of DISC lnattention symptoms (i.e., only
parent-rated inattention symptoms; refer to main text) and maternal and teacher
behavioral ratings of inattention (Table 33).
Hyperactivity
There was no significant interaction between ADHD and RD group for
hyperactivity symptoms [F(2,76)=0.03], nor was there a significant main effect of RD
group [F ( l ,76)=0.47]. As shown in Table 2, both ADHD-C and ADHD-I children had
more symptoms of hyperactivity than non-ADHD children [F(2,76)=122.93, p<.001] and,

as expected, ADHD-C children had higher levels of hyperactivity than ADHD-I children.

171

Teacher ratings, shown in Table 33, did not support higher levels of hyperactivity in
ADHD-I or RD children versus control children.
Oppositional, Aggressive, and Conduct Disorder Symptoms

ADHD-C and ADHD-I children had higher rates of Oppositional Defiant
Disorder (ODD) (see Table 2). Children with RD also had marginally higher rates of
' ODD than Control children, although this higher incidence of ODD was accounted for
entirely by those RD children with high levels of situational ADHD symptomatology (by
mother report on the DISC). The rates of Conduct Disorder (CD) in the sample did not
significantly differ among any of the key diagnostic groups.

As seen in Table 33, similar results were obtained for maternal behavioral ratings
of aggressive and oppositional behavior. Teacher behavioral ratings supported increased
levels of oppositional, aggressive, and conduct problem behaviors in children with
ADHD-C but not in ADHD-I or RD (Table 33).

Anxiety and Depression Symptoms

Differences among groups on maternal and teacher behavioral ratings of anxiety
and depression are shown in Table 33. Children with ADHD-C and ADHD-I had high
levels of depression compared to Control children, while children with RD also had more
depressive symptomatology than did Control children. With regard to maternal ratings of
anxiety, both ADHD-C and RD children had higher levels of anxiety than did the other
groups. The interaction between ADHD and RD groups [F (2,76)=2.1 1, p=.13] showed a
trend suggestive of higher levels of anxiety in RD children with comorbid ADHD.
Analyses on teacher ratings of anxiety and depression indicated significantly higher

levels of anxiety and depression in ADHD-C versus control children (Table 33).

172

.Amo. v3 QM ES.“ 2:98.286 random—=92

dd Bob 8:20.886 Ewe—mama 33592820 .8818
_- QID< 808 0252286 Ewe-«Ema 385928... AS. v3 Toma/8. Bob 2:22.286 2252-28882 .AA: .v& 8280 Bob 8882.86 2:886»; memE—ao
82. v8 .850 822 322228 22522828 2 .222 v8 8280 :52 .8588 222828222 8

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

8.8 2.28 8.2 8 2.22 8.8 2.28 8.8 2.28 2.28.28 2.22 8.8 2.88 .225
8.8 2.82 8.28 2.22 28.28 2.22 28.28 2.22 158 2.8 8.8 2.28 .2222
5:28.529 Uw<m
828 2.22 8.8 2.28 a 8.8 2.8 8.28 2.22 2 A2 8 2.82 8.8 2.28 .223
8.28 :2 8.28 2.22 8.28 2.22 8; 8 2.22 8.28 2.22 2.28 _22 .5522
2.25:. 0225
8.28 2.22 8.88 :2 8.28 2.8 8.8 2.22 2a2.28 2.22 8.8 2.28 .223
8.8 2.8 8.5 2.22 28.8 2.8 2 8.88 2.22 12.28 2.22 8.8 2.28 :52:
25.22.. .2680 022.2
A: 8 2.22 8.28 2.22 8.8 2.28 8.8 2.28 6.. 8.88 2.8 8.8 8.28 .225
8.8 2.28 2.28 2.22 8.28 2.28 28.28 2.22 028.88 2.22 8.8 2.28 .2522
52829.34. Uw<m
8. _ 8 2.22 8.28 2.22 8.8 2.28 8.8 2.28 m9.828 2.22 8.8 2.28 .223
8.2 8 2.82 8.28 2.22 28.8 2.22 m 8.28 2.8 0.888 2.22 8.8 2.28 :52:
3:283:30 2.3—=30
8.88 2.8- 8.8 2.22 m 8.28 2.22 8.8 2.22 8.. 8.28 8.22 8.8 2.28 382-2
8.28 2.22 8.8 2.22 < 8.88 2.82 18.82.22 (.828 2.22 8.8 2.28 22:22
5.2.: Q=Q< 95580
8; 8 2.22 82 8 2.: 8.8 2.22 8.8 2.8 8.. 8.28 2.22 8.8 2.28 .2835
8.8 2.22 8.88 2.22 8.28 2.28 28.28 2.82 222.28 :2 8.8 2.88 .2222
22:28:22: 02%.
8.2 8 2.22 A28 8 2.22 8.28 2.22 8.28 2.22 298.28 2.22 8.8 2.28 .223
8.8 2.22 8.28 2.22 m 8; 8 2.22 12.8828 208888 2.82 8.8 2.28 $2.22
83:09.22»: 2.5—:50
8.8 2.22 8.8 2.: 284 8 2.22 2 8.28 2.8 2.88 2.22 8.8 2.88 .225
8.8 2.22 8.8 2.22 «8.28 2.22 8.8 2.: 8.8 2.22 8.8 2.28 .2522
25.282 22252:. 028.2
2uz 8nz Tz 22uz
3:98 0-5:? flnz 222.3225 22.3520 Ruz
5555. 21.20-52. »_.=o..-oI=1n< 85:20

 

 

.AmBOQmHQ macho Np mmfig

 

.m fiSSEom 258m :82 mm 032.2.

173

Achievement, Language, and Learning
Reading Achievement
Group means on tasks of reading achievement, language, and learning are

displayed in Table 4. Differences among groups for reading achievement measures were
explored using a 3x2 univariate ANOVA with ADHD group (Control, ADHD-C, ADHD-
I) and RD group (yes, no) as the between subjects variables. Regardless'of their ADHD
classification (i.e., Control, ADHD-C, ADHD-I), children with RD had lower word
recognition (mean: 80.2 i 9.4; [F(l, 76)=33.08, p<.001]), spelling (mean: 77.7 3; 9.2;
F(1,76)=29.26, p<.001), phonological decoding (mean:79.9 : 10.0; F(1,76)=21.64,
p<.001), and reading comprehension (mean: 87.5 i 11.7, F(1,76)=15.80, p<.001) scores.

There were no significant ADHD x RD group interactions on any reading
achievement measures. Thus, ADHD-C and ADHD-I children did not differ from
Control children in word recognition, spelling, or phonological decoding achievement.
The reading comprehension of ADHD-C children was, however, significantly weaker
than Control children’s and significantly better than RD children’s, consistent with
expectations of some impairment in academic achievement secondary to ADHD.
Children with comorbid ADHD+RD-mot surprisingly-- performed like pure RD children
on these measures.
Language and Learning

Receptive and Expressive Language. The 3 (ADHD group) x 2 (RD group)
ANOVA examining performance on a broad measure of receptive language (CELF
Concepts and Directions) yielded no interaction between ADHD and RD group

[F(2,76)=0.83]. Although the main effect of RD group was significant [F (1, 76)=4.60,

174

p<.04], further post-hoc analyses revealed that pure RD children did not have lower
scores on this measure than Control children [F(1,38)=0.6l; See Table 4]. Rather, the
RD main effect was accounted for primarily by RD children with comorbid ADHD, and
the performance of these children was significantly worse than that of pure RD children
[F(1,32)=5.24, p<.03].

On the broad measure of expressive language (CELF Recalling Sentences), scores
for RD children were significantly lower than scores for non-RD children [F (1, 76)=5.34,
p<.03] and, in this case, both pure RD and comorbid RD+ADHD children significantly
differed from control children [Pure RD: F(1,38)=6.85, p<.02; RD+ADHD:
F(1,32)=5.15, p=.03].

Ratings of Learning Problems. As expected, children with RD had a history of
more developmental indicators of difficulties learning to read (maternally-rated on the
Learning and Behavior Questionnaire) than those without RD [F(1,76)=29.43, p<.001].
The severity of this history did not differ between comorbid RD+ADHD and pure RD
children.

Children with ADHD did not, as a group, have a history of more developmental reading
problems [F (2,76)=l .20], and there was no significant interaction between ADHD and
RD groups [F (2,76)=0.28]. However, children with ADHD-I (with no diagnosis of RD)
were reported to have a greater history of developmental indicators of problems learning
to read than did Control children. The severity of these problems was intermediate
between children with pure RD and control children.

With regard to teacher ratings of current learning problems on the BASC

Learning Problems scale, the ADHD and RD group main effects were both significant

175

[ADHDz F(2, 76)=5.44, p<.Ol; RD: F(1,76)=23.43, p<.001], but there was no significant
interaction between ADHD and RD group [F (2,76)=0.70]. ADHD and RD children did
not significantly differ in teacher-rated learning problems. Qualitatively, as shown in
Table 4, comorbid ADHD+RD children appeared to have higher levels of current
learning problems.

Note about Classification of Pure RD Children

Despite the very small N’s, compared to RD children whose parents did not report
clinically significant attentional problems, RD children with situational ADHD had
significantly poorer performance on a broad measure of receptive language
[F (1,1 l)=6.70, p<.03, n 2 =38] (but not on a broad measure of expressive language
[F (1,1 l)=0.30]). These children also had non-significant trends for poorer reading (1] 2
=. 12), spelling (n 2 =.O7), phonological decoding (n 2 = .18), and reading comprehension
(n 2 =.l 1; see Table 34). Children with RD and situational ADHD additionally had
marginally significantly more severe developmental histories of reading problems on the
Learning and Behavior Questionnaire compared to RD children without such inattention
problems [F(l,l l)=4.23, p<.07, n 2 =28]. Differences in IQ were not statistically
significant [F (l, ll)=O.53, n 2 =.05] and the effect sizes for IQ differences were much
smaller than achievement differences.

Taken together, these results suggest that the RD children with clinically
significant attention problems by mother report had more severe impairments in reading
and language than did children without such mother-rated attention problems. Given the
significant findings on the measure of receptive language, these children may have more

severe (or more global) receptive language impairments. Due to these patterns of more

176

 

severe reading and receptive language impairment in RD children with situational

ADHD, but no differences in the severity of teacher-rated attentional problems

[F(l,l l)=0.31, n2 =.03] , these children were included in the pure RD group.

Table 34 Comparison of RD children with and without situational inattention.

 

 

RD+ADHD g);

lnattentive- (no inattention)

Situational N=8

N=5
BASC Attention Problems T-Score Mother 72.0 (6.4) 52.3 (14.3)
Teacher 55.3 (3.1) 60.4 (14.2)
Conners ADHD Index T-score Mother 73.6 (8.9) 59.4 (15.3)
Teacher 55.7 (5.1) 59.7 (15.3)
Estimated Full Scale IQ 99.4 (16.0) 106.1 (16.5)
WIAT Reading (standard score) 75.6 (8.6) 81.8 (9.2)
WIAT Spelling (standard score) 75.0 (5.8) 80.5 (12.0)
WJ-R Word Attack (standard score) 73.8 (12.5) 83.8 (8.3)
WJ-R Passage Comprehension (standard score) 81.6 (9.1) 90.8 (15.7)
CELF—3 Concepts and Directions (scaled score) 7.4 (2.3)A 10.8 (2.3)A
CELF—3 Recalling Sentences (scaled score) 7.8 (4.3) 8.8 (2.8)
Average Achievement 74.8 (7.9) 81.7 (9.2)
(Reading, Spelling, and Word Attack)
IQ-Average Achievement Discrepancy (SS) 24.6 (17.4) 24.6 (10.7)
Teacher BASC Learning Problems T-Score 60.5 (4.2) 65.0 (14.5)
Learning and Behavior Questionnaire (Total) 33.6 (4.7)” 25.8 (7.6) B
LBQ (average score: 0-5) 4.2 (0.6) 3.4 (1.0)

 

NOTE: Matching letters within a row are significantly different from each other.

A p<.05, B p<.07

177

Ambiguous Words Task
Preliminary Data Checks

Task Validity

Overall, the task performed as predicted. Thus, children produced more errors to
targets in the Different condition, suggesting that the manipulation to test inhibition
worked as expected. Error rates also decreased from prime to target in the Same
condition, providing evidence that the facilitation manipulation also worked as expected.
Item Analysis

Individual items were examined to determine whether failure rates on any prime
sentences were particularly high (and therefore represented bad items). For prime
sentences, all items were responded to with better than 60% accuracy. Of 85 prime
sentences, 2 Nonsense, 1 Same, 1 Different, and 3 Filler items had accuracies in which 25
to 37 % of subjects responded incorrectly to them. Eighteen target-sentence items had
high failure rates (failed by greater than 25% of subjects). However, given the confound
of experimental manipulations on target accuracy (e. g., nearly half of the sentences with
high error rates were targets in the Different condition), these were not necessarily bad
items. Rather, these varying rates suggested some true differences in difficultly between
conditions, as intended.

When examining items for high failure rates on both the prime and target
sentence, only one item met this criterion (failed by 35% and 31%, respectively),
suggesting that children may not have been familiar enough with the particular word

(pit). Given that over 60% of subjects responded correctly to one or both sentences,

178

however, this item was retained in analyses. Overall, then, items were deemed, at least
qualitatively, to have acceptable validity.
Inter-Condition Accuracy and RT Correlations

Inter-condition correlations for accuracy are reported in Table 35, while inter-
condition correlations for reaction time are reported in Table 36 for the visual version of
the task (N=83) and in Table 37 for the auditOry version of the task (N=14). As can be
seen in the tables, accuracy and reaction time among conditions was highly correlated.
Accuracy-Reaction Time Correlations

Correlations between RT and Accuracy were examined to ensure that accuracy
effects did not result solely from patterns of impulsive responding or low effort.
Significant correlations between RT and Accuracy were found for primes as well as
targets in the Same (r=-0.27, p=.02 for primes and targets) and Nonsense conditions (r=-
0.39 and r=-0.38, p<.001, primes and targets, respectively) but not in the Different
condition (r=-0.003, r=-0. 12 respectively). For prime to target change scores (i.e., the
“adjusted” condition scores used to calculate cost and benefit), the correlation between
accuracy and reaction time was significant for the Nonsense condition only (r=-.338, p<
.01).

Since correlations between RT and Accuracy did not occur across all conditions,
findings suggest that--at least across groups—the correlations in the Same and Nonsense
conditions resulted from the increased difficulty of the stimuli (see above findings) rather
than from inattentive or impulsive responding (These will be examined by groups below).
This finding is important, as it provides a foundation for interpreting any group

differences as stemming from “real” cognitive effects rather than simply from lack of

179

effort, impulsivity, or carelessness (often possibilities when interpreting the performance
of children with ADHD).
Comparison to findings in Gemsbacher, Robertson, and Werner (2002)

Current error rates were very similar to those found in Gemsbacher, Robertson,
and Wemer’s (2001) sample of college students. In that sample, error rates to target
sentences were as follows: 18% for the Same condition, 22% for the Nonsense condition,
and 32% for the Different condition. It is interesting to note that the current sample
actually performed somewhat better than Gemsbacher and colleagues’ sample in the
Same and Nonsense conditions but more poorly in the Different condition. The sentences
used in the current study were intentionally simplified and shortened in comparison to the
sentences used by Gemsbacher et al. to be more appropriate for a child population.

Better performance in the current study is thus not entirely surprising and suggests that
the cognitive process underlying task performance in these two conditions may be fully
developed in school-age children.62 The higher error rates compared to the Gemsbacher
sample in the Different condition is therefore particularly striking and suggests a possible

developmental effect related to inhibition.

 

62 Reaction times were, not surprisingly, significantly slower than those of college students by—on
average—approximately 1500 ms.

180

829.8 .25. 22.2 25 a 2:85:92 2.222285 2._ 822.228 <
8229.8 .26. 22.2 as a 282892 2 822.88 +
898.8 .96. 22.2 25 a 2:82.292 2_ 82.2.2.8 .
8229-8 .28. 22.2 25 .2 885292 a 828.950 :

 

.222: .2232

 

owo. . 22:22:22.. 229:.
2882. 3.228. . 25%.. «292,—.
mm _ .- 3.228. 1.28. . 222.222.251.292...
.1222. 3222. I222. 228. . 22.—El 25...—
:222. .. _ 82. 8222. 228.- 3.222. . 82:22:22: 25...—
82_. 2.828. 3.22. 22. . 282ml 25.5
8822. 1.228. 1.228. 1.222. 8.822. 2 2. 2.228. . 25.32.25: «BE
+28. +222. 22. 222. +22 2. +82. 22.8
22.2.2.2... .anflmy. mum .ufiw 8.22-2.2 .2...qu .52.... 8%“;

 

 

2632.2. 98 225E 50m 28 29.. 852 22886250 2.23. 2883 macawfig $8888. 225822200 8228:8428 mm 2322.

181

2.2.2.2 222 8.2 22. a $022.22. 2 82.2232 ..

2.2.2.2 222 :22 22. .. 2:822:22. 2 8222.8 1

 

182

 

2:22! 82.8.2.
1.222. 2.222877 82.25.
1.822. 1.222. 22:22.. 222.25.
1.222. 1.222. 1.222. 25.825: 82.3.2.
1.222. 1.222. 1.222. .1222. 2522.. 25.5
1.228. 1.222. 1.222. 1.228. 1.222. 2:22.277 25.5
1.222. 1.222. 1.222. 1.222. 1.222. 1. 222. 25.-2| 25.5
1.222. 1.222. 1.282. 1. 2 22. 1.222. 1.222. 1.222. 25.28.25.055
222.- .222; ...222.- 22 2 .- 222.. 22 2 .- .282; .222: 222.
-22. 2.2.22.2. .22-.222. -........ .222... .-._..-...... .22-.2...

 

 

.222HZ ”=o2m2o> 2222252 32222.2. 252 825.5 22.2 2822222280 2222.2. 2283 2220222252.. 22852. 852222.80 25.2. 22222832 .22 2222.2.

A3578 .96. mod 2: a Emoscgm 2 832.8 .
325$ .96. Ed 9: a 28535 2 832.8 :

 

 

255.. 893,—.
ES. . 85282.. 393,—.
08. Km. . Sam: $9:
.. $3.. 2. _. com. . EoBEA—I «out;
3:9. 13. NE. *wvfi . 2.5925...—
mom. mmm. _L G. 1.3. mmv. . 8:03:21 «St.—
NE. So. 3:.w. mfl. 29. >9. . bah—I 25...—
_oo. m-. _e_. . Noe. h~oc p_o; _h_.
3.5 8:852 3:3 2.93:5 «Sam 25...— 85292 3:...— Bogota
.owhah 33¢... “manna. 39:. 25.:— 25...— 25...—

 

 

.AEHZ ”bozvaxxv 3033. 93 88cm Ho.“ £83280 fish €83 msoswfig wq08< 98:28.80 08:. cosoaom .nm 038.

183

Task Inconsistencies and Statistical Controls
Group Inconsistencies in Nonsense Condition Responses

Inconsistencies in group performance on the Nonsense condition are discussed in
more detail in categorical analyses. However, for current purposes, it is relevant to note
that inhibition hypotheses were often secondarily checked by exploring performance in
the Different condition only (i.e., by exploring changes in performance from prime to
target) independently of the Nonsense condition baseline. Findings for these alternate
analyses are reported where relevant.

Problems with RT “Costs ”

As can be seen from examination of RT means in Table 36, RT to targets
following a different-meaning prime in the Different condition were not, on average,
actually slower than prime RT. Thus, technically, RT was facilitated (or, at minimum,
was equivalent to prime RT). When compared to prime-target RT changes in the other
conditions, change in RT for the Different condition was still smaller than for the Same
and Nonsense conditions. Thus, for the purposes of interpretation, RT changes in the
Different condition can still likely be regarded as weak support for functioning of
interference control or inhibitory mechanisms. However, they cannot be interpreted as
representing full inhibition of non-selected homograph meanings. These RT findings will
be important to account for when interpreting the overall pattern of results.

Prime Accuracy and Prime RT Inconsistencies Among Conditions

Contrary to expectations, paired samples t-tests examining prime accuracy and

prime RT in each experimental condition (Same, Nonsense, Different) for the entire

sample (across groups and across task version) revealed that accuracy for prime

184

sentences among the experimental conditions was not equivalent. Primes in the Same
condition had the lowest accuracy, followed by primes in the Nonsense condition, then
primes in the Different condition. Thus, accuracy for primes in the Same condition was
significantly worse than accuracy for prime sentences in the Different condition
[t(l,96)=3.34, p<.001]. In addition, mean accuracy for prime sentences in the Nonsense
condition (across groups and task versions) was intermediate between Same and
Different prime accuracy and was marginally worse than accuracy in the Different
condition [t (l,96)=1.73, p<.09]. This difference became statistically significant when
the analysis was constrained to the performance of Control children [t(1,33)=2.21,
p<.04].

Reaction time (constrained to the 83 subjects who received the visual version of
the task; see Table 36) was also not equivalent across prime conditions and was
significantly longer for primes in the Same and Nonsense conditions versus the Different
condition [t(l,82)=4.48, p<.001; t(1,82)=3.40, p=.001, respectively]. These results did
not hold for the auditory task-- likely because of the very large standard deviation in
prime RT in the Different condition (see Table 37 ).

Condition variability in prime sentence responses likely resulted directly—
although unintentionally-- from the experimental design. As noted above, the Different
condition employed dominant-meaning prime sentences, whereas the Nonsense and Same
conditions utilized no-meaning and subordinate-meaning primes, respectively. These
experimental manipulations were intended to maximize potential inhibitory effects.
However, proficiency with lower-frequency (i.e., subordinate) homonyms, as well as in

interpretation of “nonsense” sentences, was clearly more difficult (although subtly so, as

185

evidenced by mean error rates and accuracies in Tables 36 & 37) than interpretation of

hi gh-frequency primes.‘53

 

 

 

 

 

 

 

 

Accuracy Reaction Time
Corrected [Uncorrected Corrected I Uncorrected
Auditory 0.29 (.21) |o.22 (0.17) 48.4 (1779.9 |427.13 (540.49)
Visual 0.23(.21) I 0.21 (0.17) 713.8 (1419.1) I 349.1 (880.0)

 

Age E fleet Inconsistencies Among Conditions and Age-Corrections

Correlation analyses revealed modest age effects which were covaried in all
analyses so that slight differences in age between groups would not account for effects.
Correlations between age and accuracy on primes and targets varied from r=.11 to .27
(see Table 35). A regression analysis with age as a predictor also indicated that age
significantly predicted accuracy for nonsense condition targets, marginally predicted
accuracy for same targets, and did not significantly predict different condition targets.
Correlations between reaction time and age on primes and targets varied from r=-.16 to -
.26 (see Tables 36 & 37).

As a secondary check of analyses, age-corrected standardized residual scores
were created for primes and targets in each of the conditions by regressing age on
accuracy scores. Results remained unchanged in all analyses when utilizing this
approach. Thus, only results for analyses in which age was covaried will be reported

here.

 

63 Whether this manipulation produced similar problems in Gembacher et al.’s sample was not documented
in their study. Thus, whether this was a developmental effect is unclear.

186

Preliminary Analyses of Inhibition Effects across Groups
Paired samples t-tests were utilized to determine whether the predicted
experimental inhibition effects occurred in the task overall (ignoring groups). As seen in
Table 35, across all groups (and with auditory and visual versions of the task combined),
children incurred a significant 24% accuracy Cost when comparing the adjusted
accuracy in the Nonsense condition (mean: 0.02 i 0.17) to the adjusted accuracy in the
Different condition (mean: 0.26 i 0.17) [t(1,96)=-l 1.12, p<.001]. With regard to RT
findings, for the visual version of the task (across all groups), children incurred a
significant 842 ms Cost when comparing the change in reaction time from prime to target
in the Nonsense (mean change in RT=-94l ms) versus Different (mean change in RT:-
99ms) conditions [t(l ,82)=5.82, p<.001].‘r’4
Dimensional Analyses: Additional Results for Visual Task
Note that the significant IQ-Reading Discrepancy/Accuracy Cost correlation
likely was largely an artifact of the IQ correlation (partial correlation between Cost and
Discrepancy with Full Scale IQ controlled: r=-.l3, n.s). Regression analyses also
revealed that Achievement-IQ Discrepancy (i.e., difference between achievement and
aptitude) significantly predicted accuracy Cost. This finding was no longer significant
after controlling for IQ, although IQ only accounted for approximately half of this effect

(IQ-Discrepancy correlation: r=.64, p<.001).

 

(’4 Children responded significantly faster to targets than to primes in the Nonsense condition [t( 1 ,82)=8.04,
p<.001], but not in the Different condition [t( 1 ,82)=0.94] (despite the decrease in mean RT from prime to
targets in the Different condition).

187

Table 38. lnattention Correlations with Accuracy and Reaction Time by Condition.

 

 

 

 

 

Total ARS Teacher Teacher
lnattention Teacher BASC Conners
S m toms lnattention Attention ADHD
y p Raw Problems Index
Accuracy Different -.262* -.223+ -.120 «.128
Same -.308** -.333** -.399*** -.276“
-.018 -.018 -.227* -.097
. Filler -.267* -.369*** -.310**
Prime
RT Different .073 . 150 .105
Same .178 .189 ' .104
Nonsense -.008 .037 -.020
Filler .072 .079 . 127 .036
Accuracy Different -.386*** -.335** -.374*” -.29l"
Same -.357** -.300** -.360" -.220+
Nonsense -.299*"‘ -.232* -.265* -.162
Filler -. 179 —.222+ -.267* -.258"'
Target
Different .062 .065 .079 -.023
.179 .153 .140 -.036
Nonsense .129 .160 .171 .065
Filler -.005 .056 .137 .063

 

 

 

 

*** p<0.001 (2-tai1ed); ** p< 0.01 (2-tai1ed); * p< 0.05 (2-tai1ed), + p<.06 (2-tailed)

Note that symptoms of hyperactivity did not significantly predict RT Cost (see
Table 38), nor did mean reading achievement, or the interaction of reading achievement
and inattention. The strongest prediction of RT Cost was obtained from teacher-rated
learning problems, which remained significant after controlling for age and IQ.
Importantly, when teacher-rated learning problems and inattention symptoms were used
to predict RT Cost, learning problems were a much stronger predictor of performance.
Thus, learning problems appeared to moderate the relationship between inattention and
RT Cost.

Categorical Analyses for ADHD and Control Children on the Visual Task

There were significant 2-way interactions for Group*Condition [F (l,60)=4.06,

p<.05], Group“ Stimulus [F(1,60)=4.34, p<.05], and Condition*Stimu1us [F

188

(1,60)=96.30, p<.001], but the 3-way Group*Condition*Stimulus interaction was not
significant. In the Group*Condition Interaction, ADHD children had much poorer
accuracy in the Different condition (averaged across primes and targets) versus Control
children, but did not significantly differ in the Nonsense condition. Across all subjects,
accuracy to targets in the Different condition was far worse than accuracy to primes,
whereas accuracy to primes and targets in the Nonsense condition was similar. Finally,
across conditions, performance of ADHD children was worse to targets than primes,
whereas Control children did not have significant differences in accuracy for primes
versus targets.
Explanation of Group Differences in Nonsense Condition Performance

Examination of the means for performance on primes and targets in the Nonsense
condition revealed that ADHD-C and Control groups did not respond equivalently to the
experimental manipulation in the Nonsense condition [F(1,51)=3.38, p<.08; see Figure
10], such that Control children’s accuracy improved from prime-target (Prime-Target
difference: 0.04 + .18), whereas ADHD children’s accuracy decreased (Prime-Target
difference: -0.06 + .17).. Given that the Nonsense condition was designed to act as the
“baseline” from which to determine Cost to targets in the Different condition, findings of
group differences violates assumptions necessary for using the Cost score to examine
group differences in interference control and suggests that comparing prime-target
changes in the Different condition may provide a more valid check of group differences

in interference control.

189

Figure 9. Group by Stimulus Interaction for Control vs. ADHD-C in the Nonsense
Condition.

 

   

 

—O-— Control
\ - I- ADHD

 

 

 

Mean Accuracy
0
a)
OI

 

 

 

 

 

Figure 10. Group by Stimulus Interaction for Control vs. ADHD-C in the Different
Condition.

 

 

 

 

. - -O- Control
‘\ +ADHD
‘ 0.9 - \ ”fl" ’
3
E 0.8
3
0
O
<
C
3 0.7
E
0.6 ~

 

 

 

 

 

 

 

I 0.5 r
1 Prime Target

190

Additional RD Findings on the Ambiguous Words Task

For RD children who took the auditory task, higher IQ was associated with a
trend towards increased interference (RT Cost; non-significant trend). In addition,
opposite to accuracy findings in these children, there was a positive associational trend
between the measure of expressive language/working memory and RT Cost, such that
weaker working memory was associated with decreased interference.

With regard to RT data, for RD children who received the visual version of the
task, high levels of oppositional/aggressive behavior were significantly associated with
reduced RT Costs (see Table 36). Note also that increased aggressive/oppositional
behavior, although non-significant, was also associated with reduced RT Benefits.

Teacher ratings of inattention showed a nonsignificant trend towards predicting
reduced accuracy Cost when IQ and age were controlled. Analyses revealed a strong
positive effect of age (B=.48) which obscured the negative relationship between higher
teacher-rated inattention and reduced accuracy Cost (B=—.SO). Thus. increased tegcher—
Lated svmptorgs of inattention in this group was associated with fleduced cost in
accugrcy when 2326 effects were controlled (signaling a potenti_al inhibitory deficit—i.e.,
both meanings of the word were more active). i

With regard to RT, performance on the expressive language/working memory
measure (Recalling Sentences) also did not strongly predict RT interference in isolation.
However, when inattention, reading achievement, and Recalling Sentences scores were
simultaneously used to predict RT Cost, there was a nonsignificant trend towards good
prediction (R2=.39), with inattention (I3=-.43) and reading achievement (B=-.24)

negatively predicting cost and Recalling Sentences positively predicting RT Cost (B=.47).

191

Thus, like findings for inattention, poorer performance on the expressive
language/working memory measure was associated with reduced RT Cost.

In RD children receiving the visual task, there was a trend toward an interaction
between reading achievement and inattention in predicting RT Cost; however, given the

very small number of subjects this interaction is not interpretable.

Garden Path Task
Preliminary Data Checks

Item Analysis

Based upon analysis of control and filler item accuracies, all items had good
reliability across subjects and were appropriate for inclusion in analyses. Individual
items were examined to determine whether failure rates for any particular sentences were
particularly high. Accuracies were examined separately for auditory and visual versions
of the task. On the visual version of the task, on control sentences (i.e., non-garden path
Reflexive Absolute Transitive [RAT] and Optional Transitive, subordinate-matrix
sentences), all items were responded to with better than 65% accuracy across subjects.
The mean accuracy across all control items was 89%, while mean accuracy on Filler
sentences was86%. Thus, these figures suggest that individual control items had good
reliability across subjects and that all task items were appropriate for inclusion in
analyses.
Inter-Condition Correlations

Accuracies on the various garden path conditions were, not surprisingly, highly

inter-correlated (See Table 39).

192

325-3 2 .va + .8253 move... .88 r. .8qu :3

0363—95 3280 5am coeuaOéoZ “$-88

”93633 ”_obcoo 5mm aovSOéoZ #7280 6333295 “a: 65335 i ”25655. 3:038 "HO 6355.5. 828.3. onocoM ”.52

 

mun—IO
.3...an
. muho
ilmm 495.80
. . Ann-Hg
1.3m 5mm 3.3—EU
. . muhé
$3 3% .25—80
1.14%. .12..on asavnv. anion? mun—.0
***wmm. aaaNNm. aaafimm. *BWN. ***5N©.
muHO
aava. aaavmv. 3.1mm. anammv. aaamwo. aaaovW. mnuhé
aaohm. *aaomm. comm. aaaovW. *aaaoo. muhé
owe. ***wwm. *GVN. +vw~. .12..me aaawnv. 3:.wa aaamwm. 0— 030m :9.—
om H. mm _. oNo. coo. +3. #. +9. _. m3. 1:2. em<
knho mud—EJ— muhg n .. .. ..
.2280 43:30 495:8 495.80 5 HO m HO 3 .59— m Hg

 

 

.A 3.1.2 Hbc:v:< use 135$ 380$ 222280 xmfl. 5am macaw mecca 80:32.80 uoEvcoULSE .am 03E.

193

 

50.32

 

3o. ".85 £3.32.—

“so

.5

_m _ .- m2. :85 bias“...

#5.

.8

m8; E. as. ”seam bees...—

ab

253295

.32. EN. ...wmm.- 8o. ".85. 5am EEG

".59.

0.5329:—

os. .15... 87 as..- .SN. :85. e...— 5.50

".5

03.23...

*fiiwnvr *iwhm. 2:359 {*wwN. _v_. bNfi 800C”.— n—uah Boy—ac

£5.

«3.2:...—

vooc 3......3; 95.- :53. as. NS. ".85. 5.... 525

£0

N8. $23.- o: .- :EN; :5; :8? .08.- m2 .- 2 23m :5

o8: 25.- 3o; 89. 2 T ET 2 7 $9- a?

.552 .552 ._o .5 253295 25253.5 03.32.. 2522.—
suotm 38.—hm acute 38cm— "uootm scuba soot”.— “Scum
£3.82.— baemia bees“: 3552;... 5a.. 5an 5.... 5.25 5.... 5.50 5...— EEG
F5. ab #5. “.5 5a "ho c.5— so

23H .0963». 98 335 3808 8:89:00 33H 55 59.5 $88.... macaw—280 .8839... 523009.85 .ov 2an

194

625-5 2.5+ .35.. .55 :.
o_£m:m_qfi~ 35:00 53 $950.82 $205.50
65535 ”3980 53 5580-52 ”52838 mo_£m§_afi~ x: 63535 ”m 63:555. .2230 ”HO 63:88.5 828.2 268mg ”.32

 

 

3m. +53. 52. +33. 5:. 5%. .555. ........Www
+30 . M: 5. SN. 1%. :3. 2%. .555. 4....”pr
+2 5. M: _. . NS. +3.... 52. .2. SM. fibfiw
3.. SN. N2. . .25. E... :8. E. 49%,...”
+33. .25. +3... .55. . .85. :55. .58. .55
$3.. :3. com. 3... .35. . ma. 25$.
5%. 2.... EN. .25. 1:5. 52. . .58. 52.3—
2.52. .555. SM. 5. .53. 35$. .58. . .35—
.fiflm hum; “an...“ .wcwmw .:-._.o .30 53.3. 5-52

 

 

.AEH Z ”Eco xmmh b02593 maoEwaoU x3... 53 8580 30:2 macaw—880 5858.... ._v 033.

195

Figure ll. Verb x Plausibility Interaction for Control Subjects on the Garden Path Task.

 

 

0.95

0.9 ~

 

0.85 ~ 1'

Mean Accuracy

0.8 r I’

 

0.75 A .I

0.7 T
Plausible Implausible

 

 

 

 

 

 

Effect of Plausibility on Garden Path Inhibition (across verb types; Table 42)

To evaluate whether the plausibility of the garden path influences revision or
adjustment of the resulting misinterpretation, the effect of plausibility on garden pathing
(i.e., whether garden path misinterpretations were plausible or implausible) was
compared to plausibility effects on non-garden path control sentences (across verb types).
This analysis revealed a different pattern of results than was obtained for global garden
path effects. Mean achievement and expressive language/working memory (Recalling
Sentences) performances were 991 predictive of garden path plausibility effects. In

contrast, the strongest predictor of garden path plausibility effects was performance on

196

 

the measure of receptive language (Concepts and Directions). Whereas higher [Q was

related to stronger inhibition effects (larger differences between plausible and implausible

garden path effects), poorer receptive language ability (Concepts and Directions) was

associated with smaller differences in plausible versus implausible garden path effects,

indicating a failure to utilize implausibility (and therefore effortful inhibition) to direct

revision of garden path misinterpretations.65 Teacher-rated learning problems and total

hyperactivity symptoms were similarly predictive of plausibility effects (i.e., inhibition

failures manifested by sustained garden path effects in the implausible condition).66

 

 

 

 

 

 

 

Table 42. Plausibility Effect on Garden Pathing (across verb type).
@2919 E Adz A EL): 1* :r.
if R2 R2 A
1 Mean Achievement .01 -.00 .01 0.70 (1,95) .09 0.84
1 Total lnattention Symptoms .02 .Ol .02 1.59 (1,95) -.l3 -l.26
1
Step 1 Age .00 -.02 .00 0.09 (1,94) -.11 -0.95
Full Scale 10 —.01 -0.08
Step 2 Mean Achievement .03 -.01 .03 1.38 (2,92) .10 0.80
Total lnattention Symptoms -. 14 -1.27
Step 3 lnteractioMAchievflnattJ .04 --02 -00 0-41 (L91) .07 0-64
1 Total Hyperactive Symptoms .03 .02 .03 3.25 (1’95)+ -. 18 -l.80+
2
Step 1 Age (months) .04 .01 .04 3.65 (1,93)+ -.o4 .o.41
Step 2 Full Scale IQ (estimated) -.08 -O.74
SteLS Total Hyperactive Symptoms -.20 -1 .91+
1 Learning Problems (Teacher) .03 .02 .03 2.91 (1,95); -.17 -l.7l+
2 Age (months) .04 .01 .04 4.10 (1,93)‘ -.04 .034
Full Scale [Q -.13 -l.l7.
Learning Problems (Teacher) -.22 -2.03

 

 

65 The logic for this is as follows: This comparison takes Plausible GP Effect-Implausible GP Effect. Smaller index
scores (differences) indicate large garden path effects for both plausible and implausible conditions (or small GP
effects for plausible and implausible conditions—unlikely given overall task effects and condition means), whereas

larger differences on this index score indicate a large GP effect in plausible condition and small GP effect in

implausible condition (this would be the expected “inhibitory pattern”)
66 Note that the direction of this effect is in the opposite direction to receptive language findings because

greater problems on this measure are associated with higher scores.

197

 

 

Model Variable R2 Adi. A F E 1

 

 

 

R. _Rj R’A
l Recalling Sentences .01 -.00 .01 0.66 (1,95) .08 0.81
l Passage Comprehension ..Ol -.00 .01 0.94 (1,95) .10 0.97
1 Concepts and Directions .09 .08 .09 9.37 (1,95)" .30 3.06"
2 Age .14 .11 .13 14.29 (1,93)’ .04 0.43
Full Scale [Q -.23 -2.12'
Concepts and Directions .42 3.78’
3
Step 1 Age .00 -.02 .00 0.09 (2,94) .05 0.50
Full Scale [Q -.25 -2.12'
Step 2 Mean Achievement .02 -.02 .01 1.20 (1,93) .04 0.33
Steg3 Concepts and Directions .14 .10 .12 12.90 (1,92)i .41 3.59f

 

*p<.001, "p<.01, ‘ p<.05; NonGP-GP for P vs. 11>

Effect of Verb Type on Garden Pathing (across Plausibility Types, Table 43 )67
To determine whether inhibitory effects are influenced by verb type (specifically,

whether RAT verbs—like implausible sentences—can deploy inhibitory mechanisms to
reduce garden path effects), garden path effects were examined for RAT versus OT verbs
(across plausibility). As shown in Table 43, neither hyperactivity, inattention, reading
achievement, nor the interaction between inattention and reading achievement
significantly predicted the effect of verb type on garden-pathing (across plausibility
conditions), nor did reading comprehension or receptive language (Concepts and
Directions) abilities. Thus, at least when plausibility effects were n_ot taken into account,
problems with inattention or reading achievement did not interfere with the typical and
expected reduction in garden path effects from RAT verbs. Such a finding would go
against the idea of a global inhibitory deficit in mfic revision/reanalysis of garden

path misinterpretations and instead leaves open the possibility that inhibitory effects

 

67 NonGP-GP for RAT vs. OT.

198

occur at the level of semantic/inferential interpretations (i.e., re-analysis occurs when

implausible inferences drive inhibition or revision of the garden path misinterpretation).

Table 43. Effect of Verb Type on Garden Pathing (across Plausibility Types)—NonGP-

GP for OT vs. RAT

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Variable E A_dj, A R2 _1: m 1
J}: For R2 A
‘ —
1 Mean Achievement .01 -.00 .01 0.75 (1,95) -.09 -0.87
1 Total lnattention Symptoms .00 791 .00 0.38 (1,95) -.06 -0.62
1 Teacher lnattention (BASC) .00 -.01 .00 0.00 (1,95) -.00 -0.02
1
Step 1 Age .01 -.02 .01 0.23 (2,94) .03 0.26
Full Scale 1Q .03 0.29
Step 2 Mean Achievement .02 -.O3 .01 0.53 (2,92) -.10 -0.84
Total lnattention Sympts. -.07 -0.65
Step 3 Interaction (Achiev.*lnatt.) O2 ..... .22-”.295“ .00 0-05 (1,91) --02 '0-22
-1... Total Hmras_tiy,g§ymmmL--.:99..-, .6-01, , -_MQQ--.__9: 12.0295), --04 -0-35
1 Learning Problems (Teacher) .00 -.01 00 0.13 (1,95) .04 0.36
l Recalling Sentences .04 .03 .04 4.12 (1,95)‘ .20 2.03‘
2 Age (months) .08 .05 07 1110,93) “ .12 1.13
Full Scale 1Q -.17 -l.42
Recalling Sentences .33 2.67"
3
Step 1 Age .01 -.02 .01 0.23 (2,94) .08 0.74
Full Scale 10 -.12 -l.01
Step 2 Mean Achievement .0] -.02 .01 0.62 (1,93) -.20 4.68+
Step 3 Recalling Sentences .10 .06 .09 9.41 (1,92)" .39 3.07"
l Passage Comprehension .01 -.00 .01 0.90 (1,95) -. 10 -0.95
1 Concepts and Directions .00 -.01 .00 0.02 (1,95) .01 0.13

 

*p<.001, "p<.01, ‘p<.05

In contrast, to these findings, the measure of expressive language/working

memory (Recalling Sentences) did predict verb type effects on garden pathing, such that

poorer performance was associated with smaller differences between RAT and OT

garden path effects (i.e., RAT verbs did not provide the expected reduction in garden path

effects). Thus, expressive language/working memory ability was predictive of large

garden path effects in both RAT and OT verb conditions, indicating the possibility of

199

more global inhibitory deficits at the level of syntactic re-analysis for those with
expressive language/working memory problems (at least when plausibility effects are not
considered).

Effect of Verb and Plausibility on Garden Pathing.

In order to examine further whether inhibitory deficits were localized to the
semantic inference versus syntactic re-analysis levels, additional regression analyses
simultaneously considered the effects of verb type and plausibility on garden path effects.
lnattention was weakly negatively predictive of verb by plausibility differences on garden
pathing across a number of indices of inattention. Thus, effects were consistent for RAT
and OT verbs, and signified that inhibition failures (specifically, failure for implausible
sentences to drive inhibition of semantic misinterpretation) were weakly globally present
for both RAT and OT verbs“. This finding offers weak support for a syntactic locus of
inhibitory deficits that could operate in addition to a semantic inferencing inhibitory
effect. Results do, however, provide the strongest support for an inhibitory effect that is
primarily inference—driven.

Reading achievement did not predict this composite score. However, as in
findings reported above for verb effects on garden-pathing, performance on the measure
of expressive language/working memory (Recalling Sentences) did weakly predict the
current composite score, such that global garden path effects for both RAT and OT verbs
were consistent in both the plausible and implausible conditions (again indicating

reduced effectiveness of implausibility-driven activation of inhibitory mechanisms).

 

68 Note that this interpretation is based upon additional knowledge gleaned from earlier findings of garden
path plausibility effects. In the absence of that prior information, it would also be possible for this result to
be interpreted as representing consistency of inhibitory effects (rather than deficits) across both verb types.
See Table XXX (p.XXX) of ADHD group effects on individual conditions for further illustration of the
current interpretation of this effect.

200

Thus, inference-driven inhibition deficits, caused garden path effects in both RAT and

OT verbs to be maintained. Thus, these findings suggest that expressive

language/working memory difficulties are associated with both syntactic and semantic

level inhibition deficits. Note that none of these results were robust to IQ and age

controls.

Table 44. Comparison of Plausibility and Verb Type on Garden Path Effect
(Verb*Plausibility*GP Interaction)

 

 

 

 

 

 

 

 

 

 

 

 

Variable E Ad'. A _E p_* '_I‘_
1!- R2 for R1 A
1 Mean Achievement .01 -.01 .01 0.46 (1,95) .07 0.68
1 Total lnattention Symptoms .04 .03 .04 3.41 (1,95? -.19 4.85“
2 .
Step 1 Age (months) .00 -.01 .00 0.00 (1,95) .01 0.07
Step 2 Full Scale IQ (estimated) .03 .01 .03 2.52 (1,94) .13 1.21
Step 3 Total lnattention Symptoms .05 .02 .02 2.32 (1,93) -.16 -1.52
1 Teacher lnattention (BASC) .04 .03 .04 4.40 (1.95)‘ -.21 -2.10‘
2
Step 1 Age (months) .01 0.08
Step 2 Full Scale IQ .10 0.94
Step 3 Teacher lnattention Sympts. .05 .02 .03 2.68 (1,93) -.18 -1.64
1
Step 1 Age .03 .01 .03 1.26 (2,94) .00 0.01
Full Scale IQ .13 1.16
Step 2 Mean Achievement .05 .01 .02 1.16 (2,92) -.02 -0.15
Total lnattention Symps. -. 16 -1.51
-_Step 3 Interaction (Achievzflnatt) .05 -.00 00 0.00 (1,91) -.01 -0.06
1 Total Hyperactive Symptoms .00 -.Ol .00 0.04 (1,95) -.02 -0.19
1 Learning Problems (Teacher) .01 .00 .01 1.00 (1,95) -.10 -1.00
1 Recalling Sentences .05 .04 05 4.63 (1,95)‘ .22 2.15?
2 Age (months) .05 .02 03 2.51 (1,93) .06 0.56
Full Scale IQ .06 0.46
Recalling Sentences .20 1.58
3
Step 1 Total lnattention Symptoms -04 -03 ~04 3.41 (1.95)+ ~16 4.59
Step 2 Recalling Sentences .07 .05 .04 3.72 (1,94)* .19 1.93+
l PassaggflCogmprghension .00 -.01 .00 0.01 ( 1,95) .01 0.11
1 Concepts and Directions .01 -.00 .01 0.73 (1,95) .09 0.86

 

lp<.001, "p<.01, ' p<.05; [(contp-plaus)(contip—implausn -[(conratp-ratp)-(conratip-rati)]./
GPVeerleff=GPPIausOT-GPPlausRAT

201

 

APPENDIX B

Preliminary Analyses of Facilitation Effects across Groups
Facilitation (i. e., Benefit) Effects across Groups
Across groups, a significant 5% Benefit in accuracy from prime to target
occurred for the Same condition (mean: 0.03 i 0.13) compared to the Nonsense condition
[t(1,96)=2.19, p=.03]. This accuracy benefit was exactly equivalent to the 5% benefit
incurred by Gemsbacher, Robertson, and Werner’s (2001) college sample. Note that
children who took the auditory version of the task incurred a non-significantly higher
Benefit (mean: 0.10 i 0.30) than those that received the Visual version of the task [mean:

0.05 i 0.22; F(l,95)=0.69].

Table 45. Auditory vs. Visual for Gemsbacher Task

 

 

 

 

 

 

flditory Version Visual Version

, =14 N=83
Accuracy-Benefit 0.06 (0.12) 0.03 (0.15)
Corrected 0.10 (0.30) 0.04 (0.21)
Accuracy-Benefit
RT-Benefit -263.53 (404.35) -32.3 (929.2)
Corrected ' 103.9 (655.2) ' 84.9 (1188.3)
RT-Benef'it

 

 

 

 

 

With regard to RT, contrary to predictions, there was no significant facilitation in
reaction time when comparing change in response time from prime to target in the
Nonsense versus the Same (mean change in RT=-860 ms) conditions for the Visual task
[t(l,82)=-0.59] or the auditory task [mean change in RT=+104 ms, t(1,13)=-0.59]. In
contrast, Gemsbacher and colleagues’ participants did demonstrate a RT Benefit (176

ms). It is important to note, however, that when children were compared in the Same

202

condition only they @ show significant facilitation from prime to target [Visualz 860 ms,
t(1,82)=9.39, p<.001; Auditory: 265 ms, t(l,13)=2.09, p<.03, l-tailed] and likewise on
the Nonsense condition only [Visualz 941 ms, t(l,13)=8.04, p<.001; Auditory: 369 ms,
t(l,13)=3.50, p<.01, l-tailed]. The task thus clearly produced priming and facilitation
effects, but these facilitation effects did not significantly differ for RT findings in the
Same versus the Nonsense condition (although such differences did occur for accuracy
comparisons as described above).

To summarize these preliminary facilitation effects: Findings supported the
effectiveness of experimental manipulations as well as the effectiveness of accuracy and
RT corrections and revealed Benefit (i.e., facilitation) effects for the visual as well as
auditory versions of the Ambiguous Words Task. Accuracy data produced the clearest
findings.

Facilitation Effects 1: Visual Task Findings in Non-RD (ADHD/Control) Children
Dimensional Hypotheses: Correlations

IQ, Reading Achievement, and Language (Table 46). Full Scale IQ was not
significantly correlated with accuracy or RT Benefit. Correlations between measures of
reading achievement and accuracy/RT Benefit were all non-significant. However,

teacher-rated learning problems did significantly correlate with accuracy and RT Benefit.

203

 

Table 46. IO, Achievement, and Language Correlations for Accuracy and RT Benefit
across Non-RD Groups on the Visual Task (N=75).

 

 

 

 

 

 

 

 

 

 

Accuracy Reaction
Time
Benefit Benefit

Age -.084 .152
Full Scale IQ (Estimated) .086 .072
Nonverbal lQ (Estimated) .201+ .009
Mean Reading Score .010 .093
Passage Comprehension (SS) .01 l .120
Concepts and Directions (ss) .054 .088
Recalling Sentences (ss) .105 .049
BASC Learning Problems .227* -.361**
(Teacher-rated T-Score)

 

 

 

 

 

*p<.05 (2-tailed); ** p<.02 (2-tailed); + p<.10 (2-tailed)

lnattention and Hyperactivity (Table 47). lnattention and hyperactivity were not
significantly correlated with accuracy Benefit. However, inattention showed a
marginally significant negative correlational trend with RT benefit, such that higher
levels of inattention were associated with reduced facilitation effects in RT. Teacher-
rated hyperactivity symptoms were also marginally significantly associated with reduced

RT facilitation.

Table 47. lnattention and Hyperactivity Correlations for Accuracy and RT Benefit across
Non-RD Groups (N=75)

 

 

 

 

 

 

 

 

 

 

 

 

 

Accuracy Reaction Time

Benefit Benefit
Total lnattentive Symptoms .146 -.193+
Total Hyperactivity Symptoms .095 -.120
Mother-Ratings
ARS Total Raw-Mother .105 -.092
ARS lnattentive Raw .103 -.106
ARS Hyperactivity Raw .094 -.066
Teacher-Ratings
ARS Total Raw .155 -.226+
ARS lnattentive Raw .052 -.220+
ARS Hyperactivity Raw .1 10 -.208+
BASC Attention .057 -.231*

 

 

 

 

* p<.05 (2-tailed); + p<.08 (2-tai1cd)

204

Dimensional Hypotheses: Regressions

In regression analyses, reading achievement, total inattention symptoms, and total
hyperactivity symptoms did not predict accuracy Benefit, and these results were
unchanged when controlling for IQ and age. The interaction of inattention and reading
achievement also did not significantly predict accuracy Benefit. Using correlations from
the prior set of analyses as a guide, a limited number of additional regressions were
performed to examine specific correlational findings. Using this method, teacher-rated
learning problems were found to significantly positively predict accuracy Benefit over
and above IQ and age effects (13:32, 1:2.53, p<.02).

Findings for RT facilitation mirrored those for inhibition-RT data on the visual
task in Non-RD children. Thus, inattention (both total symptoms and teacher-rated) was
predictive of a reduced Benefit in RT and ranged from significant to marginally
significant, depending on the index of inattention that was used. Teacher hyperactivity
ratings also marginally predicted reduced RT benefit. Subsequent analyses revealed that
ratings of neither hyperactivity nor inattention had very much predictive power over and
above one another (r2 A=.00), although inattention produced a stronger effect on RT
Benefit (B=—.15) than did hyperactivity (B=-.09). These ratings thus likely tapped the

same construct (lnattention-Hyperactivity correlation: r=.79, p<.001).

205

Table 48. RT Benefit Regressions for Visual Ambiguous Words Task (N=75)

 

 

 

 

 

 

 

 

 

 

Model Variable 33 Ag; A R2 _l: 51 t
.1; R2 A
1 Mean Achievement .01 -.01 .01 0.63 (1,73) .09 0.80
1 Total lnattention Symptoms .04 .02 .04 2.83 (1,73)+ -.19 -1.68+
2 Full Scale IQ (estimated) .01 -.01 .01 0.39 (1 ,73) .08 0.71
Age (months) .04 .01 .03 2.22 (1,72) .15 1.24
Total lnattention Symptoms .06 .02 .03 1.87 (1,71) -.16 -l .37
1 Teacher lnattention (BASC) .05 .04 .05 4.13 (1,73) ‘ -.23 -203‘
2 Full Scale IQ .07 .03 .03 2.55 (1,71) .02 0.15
Age (months) .13 1.05
Teacher lnattention (BASC) -.21 -1.60
1
Step 1 Mean Achievement .04 .00 .04 1.08 (3,71) .09 0.64
Total lnattention Symptoms -.20 -1.70+
Full Scale IQ (estimated) .01 -0.04
Step 2 Interaction (Achiev.*Inatt.) .06 .01 .02 1.53 (1,70) -.14 -1.24
1 Teacher Hyperactivity (TARS) .04 .03 .04 3.29 (1,73) " -.21 4.82“
2 Age (mths) .07 .03 .03 2.36 (1,71) .15 1.26
Full Scale IQ (estimated) .08 0.67
Teacher Hyperactivity (TARS) -. 18 -1.54
3 Teacher Hyperactivity (TARS) .05 .03 .01 0.62 (1,72) -.09 -0.48
Teacher lnattention (TARS) -.15 -0.79
4 Teacher lnattention (TARS) .05 .03 .00 0.23 (1,72) -.15 -0.79
Teacher Hyperactivity (TARS) -.09 -0.48
1 Learning Problems (Teacher) .13 .12 .13 10,91(1,73)l -.36 45,30“
Full Scale IQ .15 .12 .12 10.02 (1,73)“ -.06 0.51
Age (months) .13 1.17
Learning Problems (Teacher) . -.39 -3.17“
3 lnattention (T ARS) .13 .11 .08 6.84 (1,72) -.04 -0.28
Learning Problems -.34 -2.62'

 

* B after final step; Tp _<_ .00]; "p301; "p_<_ .05; + p_<_.10

In contrast to accuracy findings for this group, teacher-rated learning problems
significantly predicted a reduced RT Benefit that continued to be significant after
controlling for age, IQ, and inattention. Note that inattention did not predict RT Benefit
over and above that predicted by learning problems [r2 A=.00; F(1,72)=0.08]. Despite the
findings for teacher-rated learning problems, actual reading achievement (as measured
experimentally) did not significantly predict RT Benefit (even after controlling for IQ and

age), nor did the interaction between inattention and reading achievement.

206

Categorical/Group Hypotheses: ANO VA ’3

Accuracy. A 2 x 2 x 2 repeated measures analysis of variance with Condition
(Nonsense versus Same) and Stimuli (Prime vs. Target) as the within subjects factors and
Group (Control vs. ADHD-C) as the between subjects factor was used to examine
Accuracy Benefit (i.e., facilitation). The 3-way interaction was not significant.
However, there was a significant Group & Condition interaction [F(1,51)=6.64, p_=.OI],
such that averaged across prime-target stimuli, ADHD children showed no accuracy
differences in the Nonsense compared to the Same condition, whereas Control children
improved their accuracy in the Nonsense compared to the Same condition.

Due to the earlier finding concerning group differences in performance in the
Nonsense condition (see group comparisons of inhibitory effects in prior section; a 2
(Prime vs. Target) x 2 (Control vs. ADHD-C) repeated measures ANOVA also explored
differences in accuracy Benefit (i.e., facilitation) in the Same condition. While this
analysis revealed a main effect of ADHD [F(1,51)=14.53, p<.001 ], the Group“ Stimulus
interaction was not significant. Thus, there was no support for differences in facilitation

for accuracy between ADHD and Control children.

Table 49. Mean Benefit (Facilitation) Scores by Diagnostic Group

 

 

Control ADHD-Only
Combined Inattentive
Benefit (Corrected)
(Nonsense - Same)
Accuracy 0.01 (0.17) 0.06 (0.27) 0.07 (0.21)
Reaction Time 261.44 (982.4) -252.0 (1231.2) -53.4 (707.8)

 

207

 

Reaction Time. The 2 (Group) x 2 (Condition) x 2 (Stimulus) repeated measures
ANOVA revealed a significant Stimulus main effect [F(1,51)=58.06, p<.001] and
significant Stimulus * Condition interaction [F (1,51)=24.72, p<.001] but no significant
interactions involving Group and no significant 3-way interaction. The follow-up 2
(Group) x 2 (Stimulus) repeated measures ANOVA for the Same condition also revealed
no significant Group main effect nor significant interaction, although the Stimulus main
effect was significant [F(1,51)=94.55, p<.001]. Thus, RT data also supported no group
differences in facilitation.

Facilitation Effects 11: RD Findings on the Auditory and Visual Tasks
Dimensional Hypotheses: RD Correlations for Auditory vs. Visual Task

1Q, Achievement, and Language (Table 50). As shown in Table 50, for'RD
children administered the auditory task (due to low reading ability), lower reading
comprehension ability was associated with less facilitation/Benefit in accuracy. Weaker
expressive language/working memory was also associated with decreased facilitation
effects. (Or, alternatively, better reading comprehension and expressive
language/working memory were associated with accuracy facilitation in the Same
compared to the Nonsense conditions). In contrast, no strong trends or significant results
existed for indices of achievement or language and RT facilitation.

RD children who took the visual task (N=8) showed no relationship between
indices of IQ, reading, and language and accuracy Benefit. However, the correlation
between mean reading achievement (average of word recognition, spelling, and
phonological decoding) and RT Benefit was marginally significant (r=.64), such that

higher reading scores in these RD children were associated with increased

208

facilitation/benefit in RT. Likewise, these children showed an associational trend

between younger age and increased RT facilitation.

Table 50. IQ, Achievement, and Language Correlations for Accuracy and RT Benefit for

 

 

 

 

 

 

 

 

 

 

 

 

 

RD children
VISUAL (N=8) AUDITORY (N=12)
Accuracy Reaction Accuracy Reaction

Time Time
Age .097 -.632+ -300 .162
Full Scale IQ (Estimated) -.202 .44_7 .070 -.226
Nonverbal IQ (Estimated) -.475 .200 -.046 -.345
Mean Reading Score -.306 .678+ £62 -.113
Passage Comprehension (SS) -.244 .347 .608* .196
WIAT Reading (SS) -.362 fl £51 .099
Concepts and Directions (ss) .325 -. 141 .096 -.134
Recalling Sentences (SS) -.331 -.091 .684“ -.O42
BASC Learning (T-Score) L524 -.318 .354 .357

 

 

 

 

** p<.02 (2-tailed); ‘ p<.05 (2-tailed); + p<.10-(2-tailed)

Inattention and Hyperactivity. Correlations between Benefit scores and

symptoms of parent and teacher-rated inattention and hyperactivity for children with RD

who received the visual versus auditory version of the task are shown in Table 51. For

RD subjects who were administered the visual version of the task, there were no

significant correlations for any indices of inattention or hyperactivity and facilitation

(either accuracy or RT), although the correlation between teacher-rated hyperactivity on

the BASC and accuracy Benefit was marginally significant (r= .70).

For those RD children who received the auditory version of the task, there were

also no significant correlations between accuracy Benefit and ratings of inattention or

hyperactivity. Note, however, that opposite to the pattern in RD children administered

the visual task, these RD children showed a consistent negative correlational trend

between higher ratings of inattention and hyperactivity and reduced accuracy Benefit.

209

 

However, high levels of hyperactivity were associated with reduced RT facilitation on
multiple mother and teacher indices (statistically significant or marginally significant on

a number of indices).

Table 51. Inattention and Hyperactivity Correlations for Accuracy and RT Benefit for RD
children on the Visual and Auditory Task

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

VISUAL (N=8) AUDITORY (N=12)
Accuracy Reaction Accuracy Reaction
Time Time
Total Inattentive Symptoms .332 .253 -.297 .187
Total Hyperactivity Symptoms .260 .135 -.139 -.517+
Mother
ARS Total (Raw) .241 .135 -.370 -.571+
ARS lnattentive (Raw) .230 .154 -.404 -.470
ARS Hyperactivity (Raw) .234 .085 -.3 13 -.620*
BASC Attention (T-Score) .143 .184 -.289 -.102
Teacher
ARS Total (Raw) -.037 .334 -.229 -.005
ARS lnattentive (Raw) .208 .323 -.053 .137
ARS Hyperactivity (Raw) -.361 .302 -.333 -.134
BASC Attention (T-Score) .160 .4_8; .042 .257

 

 

 

*p<.05 (2-tailed); + p<.08 (2-tailed)

Conclusions. Overall, there was no association between facilitation effects and
measures of inattention, although there was some support for a relationship between
decreased facilitation and increased hyperactivity for RD children receiving the auditory

task.
Dimensional Hypotheses: RD Regressions for Auditory vs. Visual Task

Auditory Task There was a nonsignificant trend for reading achievement to
predict accuracy Benefit for RD children who received the auditory version of the task
. (R2 = .22, RZA =.17 after controlling age and iq), such that higher reading achievement
was associated with increased benefit in this group (B=.54 after controlling IQ and age).

Inattention symptoms did not significantly predict accuracy benefit by mother or teacher

210

 

report, although controlling for inattention did improve the predictive power of reading
achievement (since the two were in opposite directions; (Brewing .56 vs. Bmancmion = -.17;
R2=.30 with iq and inattention controlled).

Mirroring findings on the auditory task for accuracy Cost, performance on the
expressive language/working memory measure (Recalling Sentences) positively
predicted accuracy Benefit, such that higher performance on this measure resulted in
greater accuracy Benefit (Table 52). This effect remained after controlling for reading
achievement, age, and inattention and became extremely strong after controlling for IQ.

Also inversely related to findings for accuracy Cost, passage comprehension
performance in RD children receiving the auditory version of the task significantly
positively predicted accuracy Benefit, with higher reading comprehension associated with
increased accuracy Benefit (i.e., facilitation from prime to target; Table 52). This finding
was not entirely (or even predominantly) accounted for by reading achievement.
Accuracy facilitation effects also showed a trend towards predicting passage

comprehension ability above and beyond total reading achievement (RZA =.18; (Bmdinf

.29 vs. [33mm = .47). Thus. in low functioning children with RD, facilitation m_av be
associated with improved reading comprehension.

In contrast to the findings for facilitation effects in accuracy, for RT, neither
reading achievement, inattention, nor the interaction of reading achievement and
inattention showed a trend towards predicting RT facilitation (i.e., Benefit) after age and
IQ were controlled. Passage comprehension and expressive language/working memory
(i.e., Recalling Sentences) performance was also not predictive of RT facilitation (i.e.,

Benefit).

2]]

Table 52. RD Accuracy Benefit Regressions for Auditory Ambiguous Words Task
(N=12) '

 

 

 

 

 

 

Model Variable E Adj: A R2 _E fit 3
_R_2 R2 A

l Recalling Sentences .47 .42 .47 8.82 (1,10) .68 2.97*
Full Scale IQ .82 .75 .72 31.81 (1,8)? -.83 —3.88**
Age (months) Q -1.64
Recalling Sentences 1.2 5.641-

3 Mean Achievement .51 .40 .29 5.28 (1,9)* ._2_2 0.83
Recalling Sentences .59 2.30*

4 Mean Achievement .53 .35 .27 4.66 (1,8)+ .19 0.68
Inattention -.15 -0.62
Recalflig Sentences .58 2.16+

l Passage Comprehension .37 .31 .37 5.88 (1,10)* .61 2.43*

2 Full Scale IQ .37 .14 .28 3.57 (1,8)+ -.06 -0.18
Age .01 0.02
Passage Comprehension .62 1.89+

3 Mean Achievement .40 .27 .19 2.80 (1,9) ,2_l 0.71
Passage Comprehension .50 1.67

4 DV=Passage Comprehension

Step 1 Mean Achievement .26 .19 .26 3.49 (1,10) .51 l.87+

Step 2 Mean Achievement .44 .31 .18 2.80 (1,9) .29 1.01
Accuracy Benefit .47 1.67

 

:B after final step Tp _<_ .001; “p301; *pg .05; + p510

Visual Task. For RD children who received the visual version of the task,
accuracy Benefit showed no trend for prediction by reading achievement, inattention, or
an interaction between reading and inattention on accuracy Benefit. The primary finding
was a non-significant trend for teacher-rated learning problems to positively predict
accuracy Benefit (R2 =.28) even after controlling for IQ and age, such that increased
learning problems were also associated with increased facilitation in accuracy (which was
opposite to associations between reading and accuracy Benefit in RD children on the
auditory task).

With regard to RT data, reading achievement significantly predicted RT Benefit

after controlling for age and IQ, with higher reading achievement associated with

212

increased facilitation in RT in these high functioning RD children (see Table 51). There
was also a non-significant trend for inattention to predict RT Benefit (i.e., facilitation)
after age and IQ were controlled, so that higher inattention was associated with increased
RT facilitation. Together, reading achievement and inattention significantly predicted
70% of the variance in RT facilitation [F (1,5)=5.80, 9:05] for these RD children, with
reading achievement producing the strongest effect (but inattention also serving as a
marginally significant predictor in the model). There was no predictive interaction

between reading achievement and inattention.

213

 

Table 53. RD Benefit RT Regressions for Visual Ambiguous Words Task (N=8)

 

 

 

 

 

 

Model Variable E .A_dj: 9E E g: t
_R_2 RM
1 Mean Achievement .46 .37 .46 5.11 (1,6)+ .68 226+
2 Age (Months) .78 .69 .38 8.59 (1,5)* -.57 -2.69*
Mean Achievement .62 293*
1 Total Inattention Symptoms .06 -.09 .06 0.41 (1,6) g; 0.64
Full Scale IQ (estimated) .47 .26 ._2_7 2.52 (1,5) .72 1.95
Total Inattention Symptoms .58 1.59
3 Age (months) .41 .17 .01 0.06 (1,5) -.61 -1.70
Total Inattention Symptoms .09 0.24
4 Full Scale IQ .78 .62 ._2_4 4.35 (1,4) -.06 -0.19
Age -.57 -2.42+
Total Inattention 8x3. .67 2.09
1
Step 1 Mean Achievement .70 .58 .70 5.80 (2,5)* .95 3.61*
Total Inattention Symptoms .48 1.94
Step 2 Interaction (Achiev.*Inatt.) .78 .62 .08 1.47 (1,4) .31 1.21
2 Mean Achievement .70 .58 .24 3.97 (l,5)+ .84 3.25*
Total Inattention Symptoms .52 l.99+
3 Total Inattention Symptoms .70 .58 .64 10.54 (1,5)* .52 l.99+
Mean Achievement .84 3.25*

 

:[3 after final step; *pg .05; + p310

214

APPENDIX C

Ambiguous Words Task Instructions

Set-up directions:

1) Open “Lisa Dissertation” folder on desktop.

2) Open “Gemsbacher Task” folder.

3) Double click on icon for appropriate list (right now there is only one list—List C—
soon there will be an A and a B list which we will rotate randomly—the cover page
will tell you which list to administer when that occurs).

4) Click on the run button/

5) Enter subject id# and session #(always=1).

6) Make sure the Y and N stickers are on the 1 and 3 keys and the “blank” sticker
covers the N key used for another task).

Directions:

Say:
I would like you to read some sentences silently to yourself. All of the sentences you will
read end in a word that can have more than one meaning.

You know how the word “glass” can either be the thing you drink out of gr the thing in
your window? These sentences use words like glass. After you read the sentence, you
need to decide whether each sentence makes sense for at least 9% of the meanings of the
word. So, for example, if you read “He drank from the glass”——you would answer
“yes”—the sentence makes sense. If you read “He broke the window glass,” you would

9

also answer “yes’ —the sentence makes sense.

Now, in addition to these sentences that make sense, some of the sentences are “silly”
sentences that don’t make any sense at all—no matter which meaning of the word you
use. So, for example, if you read, “He ate the glass”—you would answer “no”—that the
sentence does not make sense—it is a “silly” sentence.

Press the “Y” key if the sentence makes sense and the “N” key if the sentence does not
make sense. Be sure to respond as quickly and as accurately as you can.

Do you have any questions? [If child does not understand any aspects of the task, try to
explain fiirther].

Administer practice items—if child gets any of the items wrong—explain why they got it
wrong and try to make sure they understand why it was wrong. Use this as a way to
make sure the child understands the purpose/directions of the task.

215

Afier the practice say, “That was good. Now we’re going to try the game ‘for real’”
Make sure to press the “Y” key if the sentence makes sense and the “N” key if the
sentence does not make sense. Remember to answer as quickly and as accurately as you
can.

There will be 5 blocks of sentences and you will have an opportunity to take a short break
after each block.

Garden Path Instructions:

Read these before the instructions on the screen:

Say: You will read some sentences to yourself and will answer a question about each ‘-
sentence. The most important thing for you to remember when you answer the questions i
is that your answer should be based grily on the meaning of that specific sentence. You

should m assume any information outside of the sentence itself So, for example, if a

question asks whether Jane baked a cake, you should answer yes 9112 if the sentence

specifically states that. If you think based on the sentence that Jane might have baked the

cake sometime before, but the sentence doesn 't specifically say that she did, then you

should answer no, that Jane did not bake the cake.

Also, be aware that many (but not all), of these sentences are “tricky ” sentences—that is,
they have two parts that are blended together. Because the 2 parts of the sentence are
blended together, the sentence may seem to mean one thing, but really mean another.
Your job is to figure out which parts go together and then answer the question based on
that. You need to be careful and pay very close attention, because sometimes the answer
that seems “obvious ” is not actually the Ltgh_t answer. Also, keep in mind that sometimes
the correct answer to the sentence may seem a little silly.

Let ’3 try some for practice.

If the child gets any of the practice items wrong, ask him/her to explain why they thought
their answer was right. Explain to them, using the directions, why that answer is not
correct. E.g., point out that they made an inference outside the sentence context or that
they were tricked by the way the sentence parts were blended.

216

APPENDIX D

Ambiguous Words Task Stimuli

PrimeList

He had a sick organ.

He swung with the bat.

He walked out onto the deck.
She heard the bark.

She hurt her foot.

She lost her key.

She made it to the top.

She put more jam on the toast. She held her cup for the toast.

She put on the pretty ring.
She sent the letter.

She stole from the bank.
She wrote with the red pen.
He ate on the calf.

He ate the mint.

He bit his lip.

He broke his rib.

He closed the jar.

He did into the tie.

He fed to a spade.

He had to pay the bill.

He lay the box on its side.
He let the bird go free.

He put the books in a pile.
He put the rake in the shed.
He ran in the race.

He sang to the cape.

He sat by the fire.

He sat to fake the tag.

He sent his mom a card.
He tied the dog to the post.
He used his brush.

He wanted to get a pet.

He was a big fan.

He went to the game.

He went to the park.

She ate the ham.

Target

He played the organ.

He saw the flying bat.
He sorted the deck.

She cut the bark.

She jumped a foot.

She used the answer key.
She tried to spin the top.

She heard the loud ring.
She sang every letter.

She fished from the bank.
She threw food into the pen.
He fed the calf.

He swam the mint.

He let his lip.

He led his rib.

He wrote the jar.

He put on his new tie.

He played a Spade.

He wanted to smell the bill.
He gave him for the side.
He sent the car to free.

He met the cats to a pile.
He fed the cup in the shed.
He planted the race.

He had on a red cape.

He sang on the fire.

He went out to play tag.
He baked up the card.

He ran to fan the post.

He shot onto his brush.

He cut to rake a pet.

He ended the fan.

He saved to the game.

He saved to the park.

She went the ham.

217

Condition

Different
Different
Different
Different
Different
Different
Different
Different
Different
Different
Different
Different
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler

She ate the wake.

She baked up the state.
She came the spot.
She fed the bay.

She fell with the deep.
She had gone the fine.
She handed the face.
She killed the store.
She landed the will.
She led the chip.

She met the stick.

She met to the frame.
She played the well.
She raked the gas.
She ran a cap.

She rang on the band.
She rang the plant.
She sanded the bug.
She sang the fly.

She sat the drop.

She shut the arm.

She shut the box.

She smelled a grade.
She spoke the nut.
She spoke the tip.
She spun the land.
She swam in the pool.
She swam the mass.
She tamed a file.

She used the yarn.
She wanted to hide.

She wanted to make a trade.

She went the fall.
She went the ship.
She woke the glass.
She wrote the dip.
He called the pipe.
He looked the case.
He mopped the play.
He walked the club.
She fell a date.

She grew the ruler.
She hiked the jam.
She hoped the drill.

She sat the wake.

She saw a map of the state.
She rang the spot.

She sang the bay.

She backed with the deep.
She sanded the fine.
She let by the face.
She went to the store.
She baked the will.
She fell the chip.

She broke the stick.
She let the frame.

She did not feel well.
She stamped the gas.
She meta cap.

She rode the band.
She cut the plant.

She went the bug.

She came the fly.

She asked the drop.
She broke her arm.
She blamed the box.
She got a good grade.
She met for the nut.
She fed to the tip.

She rang the land.

She called on the pool.
She fell the mass.

She rode on a file.

She sat the yarn.

She hit by the hide.
She sanded to rope a trade.
She lay the fall.

She baked the ship.
She went the glass.
She played the dip.

He got water from the pipe.

He got ready for the case.

Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Filler
Nonsense
Nonsense

He won the game with the play. Nonsense

He hit him with the club.
She went out on a date.

Nonsense
Nonsense

She saw the throne of the ruler. Nonsense

She ate the jam.

She marched during the drill.

218

Nonsense
Nonsense

She lived a note.
She sang a seal.

She spoke the spring.

She told the nail.
She went the punch.

NULL
NULL
NULL
NULL
NULL

He broke the water pitcher.
He cut off the cow's horn.
He got a shorter jail sentence.

He mixed the batter.
He tied the red bow.

He went into the bar.

She got dressed for the ball.
She hoped to be a big star.

She lit the match.

She lived on the tenth story.
She sipped from the straw.
She threw out the pit.

Condition

control-1P
control-1P

control-1P

control-1P
control-1P
control-P
control-P

control-P
control-P

control-P
coord-cont-IP

Sentence

She played a note. Nonsense
She closed the letter with a seal. Nonsense
She broke the bed spring. Nonsense
She bit her nail. Nonsense
She was hurt by the punch. Nonsense
NULL NULL
NULL NULL
NULL NULL
NULL NULL
NULL NULL
He filled up the pitcher. Same

He looked at the ram's horn. Same
He was serving a life sentence. Same

The pup grew as Joe was cutting.
The robber was locked up as the kids

played.

The rock got passed as Fran was writing.

The snow got cold as Fred cooked.

The worm swam in the lake as Jack ate.
The bus did not run as Sam drove.

The frog swam in the lake as the snake

ate.

The girl was at the store as Tom kissed.
The notes burned as Jill was reading.

The water cooked as Deb drank.
Ken talked to the boss and that's why the Did Ken talk to the men?
men were not at work.

219

He made the batter. Same

He undid the bow. Same

He got a drink at the bar. Same

She danced at the ball. Same

She wanted to be a big star. Same

She blew out the match. Same

She went up to the tenth story. Same

She drank from the straw. Same

She spit out the pit. Same

Garden Path Stimuli

Question
Was Joe cutting the pup?
Did the kids play the
robber?

Was Fran writing the
rock?

Did Fred cook the snow?
Did Jack eat the worm?
Did Sam drive the bus?
Did the snake eat the
frog?

Did Tom kiss the girl?
Was Jill reading the
notes?

Did Deb drink the water?

coord-cont-IP

coord-cont-P

coord-cont-P

coordination-I
coordination-I

coordination-I

coordination-I

coordination-P

coordination-P

coordination-P

coordination-P

filler

filler
filler

filler

filler

filler

filler

filler

filler

filler

filler
filler

filler

filler

Sue kissed her mom and that's why her Did Sue kiss her dad?
dad sat in the den.

The girl played with the baby and that's Did the girl play with her
why her mom helped. mom?

The star kissed the fan and that's why her Did the star kiss her date?
date got mad.

Bob fed the cat and the dog was at the vet. Did Bob feed the dog?
Mommy called the doctor and the nurse Did Mommy call the

was out sick. nurse?
The girl played with the baby and her Did the girl play with her
mom went out. mom?

The star kissed the fan and her date stayed Did the star kiss her date?
home.
Ken talked to the boss and the men did Did Ken talk to the men?

good work.
My sister bumped into the mailman and Did my sister bump into
my mom told her to be careful. my mom?

Peg hit the boy and the teacher told her to Did Peg hit the teacher?
sit in the hall.

Sue kissed her mom and her dad gave her Did Sue kiss her dad?

a hug.

As Daddy made the food the kids played Did Daddy make the

in the den. tape?

As Dan drove the car he sang to the tape. Did Dan drive the bus?
As Flo was jumping rope she hurt her Was Flo jumping the log?
foot.

As Jake shot the bow the bird flew into the Did Jake shoot the

tree. bunny?

As Jane ate the food her mom did the Did Jane eat the food?
dishes.

As Jane was making the cookies her mom Did Jane's mom fly to
flew to Mars. Mars?

As Mom told the story the boy lay in his Did Mom tell the cops?
bed.

As Pam was feeding the fish she saw a Did Pam feed the fish?
frog.

As Pam was feeding the fish she saw a Did Pam see a UFO?
UFO.

As Sam played the drums his mom left the Did Sam play the drums?
room.

As the boy hit the ball the dog ran to get it. Did the boy hit the ball?
As the boys were making the fire they put Were the boys making

on more wood. the cake?
As the boys were making the kite they put Did the boys put on the
on the tires. tires?

As the dog went for the cat the boy pulled Did the dog go for the

220

filler
filler
filler
filler
filler
filler
filler
filler
filler
filler
filler
filler
filler
filler
filler
filler

filler
filler

filler

filler
filler

filler
implausible

implausible

on its rope.

cat?

As the girl played the fox her mom left the Did the girl play the fox?

room.
As the kids sang the song the teacher
played the piano.

As the man drove the cow he sang to the
moon.

As the man shot the gun the ape flew into

the tree.

Bob played in the den as his mom was
baking the cake.

Spot ate the bone as the boy was feeding
the cat.

The baby drank the milk as it sat in the
crib.

The bird flew into the tree as it looked for

the nest.

The boys took their books as they were
leaving for school.

The boys took their pens as they lefi for
jail.

The cops saved the dog as the snake ate
the home.

The dog hid the bone as it filled up the
hole.

The fireman saved the cat as the fire
burned the house.

The girl got the food as Tom sat at the
table.

The girl hid the bone as she made the
cake.

Did the kids sing the
song?

Did the man drive the
cow?

Did the ape fly into the
tree?

Was Bob's morn baking
the ham?

Did the boy feed the dog?

Did the baby sit in the
crib?

Did the bird look for the
cat?

Were the boys leaving for
the play?

Did the boys leave for
jail?

Did the snake eat the
home?

Did the dog fill up the
hole?

Did the fire burn the
house?

Did Tom sit at the table?

Did the girl make the
cake?

The girl went to the park as she was riding Was the girl riding the

her bike.
The man had a dog as he left the bank.

bike?
Did the man have a dog?

The man took the cash as he left the bank. Did the man leave the

home?

The mom played the socks as she watched Did the mom play the

the news.

socks?

The mom sewed socks as she watched tv. Did the mom watch tv?
The pup drank the ink as it sat in the cage. Did the pup drink the

ink?

The wife washed the dishes as the man cut Did the man out the note?

' the grass.

As Ben was writing the baby was on tv.

Was Ben writing the
baby?

As Bill was eating the spoon baked in the Was Bill eating the

221

implausible
implausible
implausible
implausible
implausible
implausible
implausible
implausible
plausible
plausible
plausible
plausible
plausible
plausible
plausible
plausible
plausible
plausible
RAT-cont-IP
RAT-cont—IP
RAT-cont-P
RAT-cont-P
RAT-I
RAT-I

RAT-I
RAT-I

RAT-P

OVCD.

As Buffy walked the bunny napped.
As Dan baked the milk shake tasted good.

As Deb rode the bear was sleeping in its

den.

As Fred threw the cat sat still.
As Meg played the shoes got shined.
As Ned was reading the door was closed.

As the ape ate the tree grew.
As the bird ate the tire lay in the dirt.
As Ann was eating the cake baked.

As Bill was running the race did not start

due to rain.

As Bob hunted the deer hung on the wall.
As Joe walked the dog napped.

As Pam baked the cake got stale.

As Ron drove the car sat in the lot.

As the cat chased the rat sat still.

As the man cut the grass got long.

As the storm blew the boat sat in the shed.
As Tom cooked the hot dog got cold.
The game was on tv as Mommy washed.

The kids swam in the pool as Frank dried

off.

The baby was in bed as the boy washed.

The cat sat on the bed as the man shaved.

spoon?

Did Buffy walk the
bunny?

Did Dan bake the milk
shake?

Did Deb ride the bear?

Did Fred throw the cat?
Did Meg play the shoes?
Was Ned reading the
door?

Did the ape eat the tree?
Did the bird eat the tire?
Was Ann eating the
cake?

Was Bill running the
race?

Did Bob hunt the deer?
Did Joe walk the dog?
Did Pam bake the cake?
Did Ron drive the car?
Did the cat chase the rat?
Did the man out the
grass?

Did the storm blow the
boat?

Did Tom cook the hot
dog?

Did Mommy wash the
game?

Did Frank dry off the
kids?

Did the boy wash the
baby?

Did the man shave the
cat?

As Anna dressed the dog played ball in the Did Anna dress the dog?

yard.

As Daddy hid the baby was lost.

As Daddy hid the kids looked for him.
As the man shaved the cat sat in the tree.

As Betty woke up Daddy was sleeping in

the den.

222

Did Daddy hide the
baby?

Did Daddy hide the kids?
Did the man shave the
cat?

Did Betty wake up
Daddy?

RAT-P As Frank dried off the kids got out of the Did Frank dry off the

pool. kids?

RAT-P As Jan dried off the cat jumped onto the Did Jan dry off the cat?
bed.

RAT-P As Mommy washed the cat came into the Did Mommy wash the
room. cat?

223

 

REFERENCES

Ackerman, P.T., Anhalt, J ., Dykman, R.A., & Holcomb, P. (1986). Effortful

processing deficits in children with reading and/or attention disorders. Brain and
Cognition, 5, 22-40.

Ackerman, P.T. & Dykman, RA. (1982). Automatic and effortful information-
processing deficits in children with learning and attention disorders. Topics in
Learning Disabilities, 2, 12-22.

Ackerman, P.T., Dykman, R.A., & Gardner, M.Y. (1990). ADD students with and
without dyslexia differ in sensitivity to rhyme and alliteration. Journal of
Learning Disabilities, 23, 279-283.

American Psychiatric Association. (1994). Diagnostic and statistical manual of mental
disorders (4th ed.). Washington, DC: Author.

Anderson, SA. (1999). Dlx genes, the striatal subventricular zone, and the development
of neocortical intemeurons. In: D.D. Duane (Ed.). Reading and Attention
Disorders: Neurobiological Correlates. Baltimore: York Press, 1-16.

Aram, D., Ekelman, B., & Nation, J. (1984). Pre—schoolers with language disorders: 10
years later. Journal of Speech and Hearning Research, 2 7, 232-244.

August G.J. & Garfinkel, B.D. (1990). Comorbity of ADHD and reading disability
among clinic-referred children. Journal of Abnormal Child Psychology, 18, 29-
45.

August, G.J. & Garfinkel, B.D. (1989). Behavioral and cognitive subtypes of ADHD.
Journal of the American Academy of Child and Adolescent Psychiatry, 28, 739-
748.

Baddeley, A.D. & Hitch, G.J. (1994). Developments in the concept of working memory.
Neuropsychology, 8, 485-493.

Baker, L. & Cantwell, DP. (1992). Attention deficit disorder and speech/language
disorders. Comprehensive Mental Health Care, 2(1), 3-16.

Baker, L. & Cantwell, DP. (1987). A prospective psychiatric follow-up of children with
specific language disorders. Journal of the American Academy of Child and
Adolescent Psychiatry, 26 546-553.

Barkley, RA. (1997). Behavioral inhibition, sustained attention, and executive functions:
Constructing a unifying theory of ADHD. Psychological Bulletin,121(1), 65-94.

224

Barkley, RA. (1996). Attention-Deficit/Hyperactivity Disorder. In: El. Mash and RA.
Barkley (Eds). Child Psychopathology. New York: Guilford Press.

Barkley, R.A., Cunningham, C., and Karlsson, J. (1983). The speech of hyperactive
children with their mothers: comparisons with normal children and stimulant
effects. Journal of learning Disabilities, 16, 105-110

Barkley, R.A., DuPaul, G.J., & McMurray, MB. (1990). Comprehensive evaluation of
attention deficit disorder with and without hyperactivity as defined by research
criteria. Journal of Consulting and Clinical Psychology, 58, 775-789.

Beitchman, J .H., Hood, J ., Inglis, A. (1990). Psychiatric risk in children with speech and
language disorders. Journal of Abnormal Child Psychology, I 8, 283-296.

Beitchman, J.H., Hood, J., Rochon, J., Peterson, M., Mantini, T., and Majumdar, S.
(1989a). Empirical classification of speech/language impairment in children 1:

Identification of speech/language categories. Journal of American Academy of
Child and Adolescent Psychiatry, 28, 112-1 17.

Beitchman, J .H. & Young, AR. (1997). Learning disorders with a special emphasis on
reading disorder: A review of the past 10 years. Journal of the American
Academy of Child and Adolescent Psychiatry, 36, 1020-1032.

Beitchman, J.H., Hood, J ., Inglis, A. (1990). Psychiatric risk in children with speech and
language disorders. Journal of Abnormal Child Psychology, 18, 283-296.

Beitchman, J ., Tuckett, M., & Batth, S. (1987). Language delay and hyperactivity in
preschoolers: Evidence for a distinct group of hyperactives. Canadian Journal of
Psychiatry, 32, 683-687.

Bender, W.N. (1996). Learning disabilities. In: R]. McLaughlin & P. Wehman (Eds).
Mental Retardation and Developmental Disabilities. Austin, TX: Pro-Ed, 259-
279.

Berk, L.E. & Landau, S. (1993). Private speech of learning disabled and
normally achieving children in classroom academic and laboratory contexts. Child
Development, 64, 556-571.

Berk, L. & Potts, M. (1991). Development and functional Significance of
private Speech among attention-deficit hyperactivity disordered and normal boys.
Journal of Abnormal Child Psychology, 19, 357-377.

Berry, C.A., Shaywitz, S.E., & Shaywitz, BA. (1985). Girls with attention

deficit disorder: A silent minority? A report on Behavioral and cognitive
characteristics. Pediatrics, 76, 801-809.

225

Biederman, J ., Newcom, J ., & Sprich, S. (1991). Comorbidity of attention deficit
hyperactivity disorder with conduct, depressive, anxiety, and other disorders.
American Journal of Psychiatry, 148, 564-577.

Britton, B.K. & Gulgoz, S. (1991). Using Kintsch’s computational model to improve
instructional text: Effects of repairing inference calls on recall and cognitive
structures. Journal of Educational Psychology. 83, 329-345.

Brock, S.E. & Knapp, PK. (1996). Reading comprehension abilities of children with
attention-deficit/hyperactivity disorder. Journal of Attention Disorders, 1, 173-
185.

Brunswick, N., McCrory, H, Price, C.J., Frith, C.D., & Frith, U. (1999). Explicit and
implicit processing of words and pseudowords by adult developmental dyslexics:
A search for Wemicke’s Wortschatz? Brain, 122, 1901-1917.

Cantor, J. & Engle, R.W. (1993). Working-memory capacity as long-tenn memory
activation: An individual-differences approach. Journal of Experimental
Psychology: Learning, Memory, and Cognition, 19, 1101-1114.

Cantwell, D.P. & Baker, L. (1991). Association between attention deficit-hyperactivity
disorder and learning disorders. Journal of Learning Disabilities, 24(2), 88-95.

Carlson, C.L., Lahey, B.B., & Neeper, R. (1986). Direct assessment of the cognitive
correlates of attention deficit disorder with and without hyperactivity. Journal of
Psychopathology and Behavioral Assessment, 8,_69-86.

Carroll, K.M. & Rounsaville, BJ. (1993). History and significance of childhood
attention deficit disorder in treatment-seeking cocaine abusers. Comprehensive
Psychiatry, 34(2), 75-82.

Carte, E.T., Nigg, J .T., & Hinshaw, SP. (1996). Neuropsychological functioning, motor
speed, and language processing in boys with and without ADI-ID. Journal of
Abnormal Child Psychology, 24(4), 481-498.

Casey, BJ. (2001). Disruption of inhibitory control in developmental disorders: A
mechanistic model of implicated frontostriatal circuitry. In: J .L. McClelland and
RS. Siegler (Eds). Mechanisms of Cognitive Development: Behavioral and
Neural Perspectives. The Carnegie Mellon Symposia on Cognition, vol. 28.
Mahwah, NJ: Lawrence Erlbaum, pp. 327-349.

Casey, B.J., Castellanos, F.X., Giedd, J .N . et al. (1997). Implication of right
' frontostriatal circuitry in response inhibition and attention-deficit/hyperactivity

disorder. Journal of the American Academy of Child and Adolescent Psychiatry,
36, 374-383.

226

 

Castellanos, F.X., Giedd, J .N., Marsh, W.L., et al. (1996). Quantitative brain magnetic
resonance imaging in attention-deficit hyperactivity disorder. Archives of
General Psychiatry, 53, 607-616.

Chang, H.T., Klorrnan, R., Shaywitz, S.E., Fletcher, J .M., Marchione, K.E., Holahan,
J .M., Stuebing, K.K., Brumaghim, J .T., Shaywitz, BA. (1999). Journal of
Abnormal Child Psychology, 27(3), 237-245.

Christianson, K., Hollingworth, A., Halliwell, J ., & Ferreira, F. (2001). Thematic roles
assigned along the garden path linger. Cognitive Psychology, 42, 368-407.

Cohen, J. & Cohen, P. (1983). Applied multiple regression/correlation analysis for the
behavioral sciences. Hillsdale, NJ: Erlbaum.

Cohen, N.J., Vallance, D.D., Barwick, M., Im, N., Menna, R., Horodezky, N .B.,
Isaacson, L. (2000). The interface between ADHD and language impairment: An

examination of language, achievement, and cognitive processing. Journal of
Child Psychology and Psychiatry, 39, 853-864.

Cohen, N., Davine, M., Horodezky, N., Lipsett, L., & Isaacson, L. (1993, 1992?).
Unsuspected language impairment in psychiatrically disturbed children:
prevalence and language and behavioral characterisitics. Journal of the American
Academy of Child and Adolescent Psychiatry, 32, 595-603.

Conners, CK. (1997). Conners Rating Scales-Revised. Toronto: Muli-Health Systems,
Inc.

Copeland, A. P. (1979). Types of private speech produced by hyperactive and
nonhyperactive boys. Journal of Abnormal Child Psychology, 7, 169-177.

Cotugno, A.J. (1987). Cognitive control functioning in hyperactive and nonhyperactive
learning disabled children. Journal of Learning Disabilities, 20, 563-567.

Crain, S. & Shankweiler, D. (1988). Syntactic complexity and reading acquisition. In:
A. Davison, G.M. Green at al. (Eds). Linguistic Complexity and Text
Comprehension: Readability Issues Reconsidered. Hillsdale, N.J.: Lawrence
Erlbaum, 167-192.

Crick, N. R. & Dodge, K.A. (1994). A review and reformulation of social information-
processing mechanisms in children’s social adjustment. Psych. Bull, 115, 74-
101.

Culbertson, J .L. & Edmunds, J.E. (1996). Learning disabilities. In: Adams et al. (Eds).
Neuropsychology for Clinical Practice: Etiology, Assessment and Treatment of
Common Neurological Disorders. Washington, DC: American Psychological
Association.

227

 

Dagenbach, D. & Carr, T. Inhibitory Processes in Attention, Memory, and Language.
San Diego: Academic Press, 409-453.

Damico, J .S., Damico, S.K., & Armstrong, MB. (1999). Attention-deficit hyperactivity
disorder and communication disorders: Issues and clinical practices. Child and
Adolescent Psychiatric Clinics of North America, 8 (l ), 37-60.

Daneman, M. & Carpenter, P. (1983). Individual differencews in integrating information
between and within sentences. Journal of Experimental Psychology: Learning,
Memory and Cognition, 9, 561-584.

Daneman, M. & Carpenter, P. (1980). Individual differences in working memory and
reading. Journal of Verbal Learning and Verbal Behavior, 8, 407-428.

Dell, GS. and O’Seaghdha, PG. (1994). Inhibition in interactive activation models of
linguistic selection and sequencing. In: D. Dagenbach and T. Carr (Eds).

Inhibitory Processes in Attention, Memory, and Language. San Diego: Academic
Press, 409-453.

DeLong, R. (1995). Medical and pharmacologic treatment of learning disabilities.
Journal of Child Neurology, 10(Suppl. 1), 892-895.

Dempster, RN. (1993). Resistance to interference: Developmental changes in a basic
processing mechanism. In: M.L. Howe and R. Pasnak (Eds). Emerging Themes
in Cognitive Development: Volume I: Foundations. New York: Springer-Verlag,
pp. 3-27.

Denckla, MB. (1996). Biological correlates of learning and attention: What is relevant
to learning disability and attention-deficit hyperactivity disorder? Developmental
and Behavioral Pediatrics, 17(2), 1 14-119.

DuPaul, G. J ., Power, T.J., Anastopoulos, A.D., & Reid, R. (1998). ADHD Rating Scale-
IV. New York: Guilford Press.

Eberhard, K. (1994). Phonological inhibition in auditory word recognition. In: D.
Dagnebach & T. Carr (Eds). Inhibitory Processes in Attention, Memory, and
Language. San Diego: Academic Press, 383-409.

Elbert, J.C. (1993). Occurrence and pattern of impaired reading and written language in
children with attention deficit disorders. Annals of Dyslexia, 43, 26—43.

Faraone, S.V., Biederman, J., Lehman, B.K., Spencer, T., Norman, D., & Seidman, L].
et al. (1993). Intellectual performance and school failure in children with
attention deficit hyperactivity disorder and in their siblings. Journal of Abnormal
Psychology, I 02(4), 616-623.

228

 

Faust, M.E. & Gemsbacher, M.A. (1996). Cerebral mechanisms for suppression of
inappropriate information during sentence comprehension. Brain and Language,
53, 234-259.

Felton, R.H., Wood, F.B., Brown, 1.8., Campbell, S.K., & Harter, MR. (1987). Separate
verbal memory and naming deficits in attention deficit disorder and reading
disability. Brain and Language, 31, 171-184.

Ferguson, D.M. & Horwood, LJ. (1992). Attention deficit and reading achievement.
Journal of Child Psychology and Psychiatry, 33 , 375-385.

Fitch, R.H., Miller, S., & Tallal, P. (1997). Neurobiology of speech perception. Annual
Review of Neuroscience, 20, 331-353.

Fomess, S.R., Youpa, D., Hanna, G.L., Cantwell, D.P., & Swanason, J .M. (1992).
Classroom instructional characteristics in attention deficit/hyperactivity disorder:
Comparison of pure and mixed subgroups. Behavioral Disorders, 17(2), 115-125.

Frank, Y. (1996). Learning disabilities: Classification, clinical features, and treatment.
CNS Drugs, 5(1), 24-36.

Frith, U. (1995). Dyslexia: Can we have a shared theoretical framework? Educational
and Child Psychology, 12, 6-17.

Gemsbacher, M.A. (1997). Group differences in suppression skill. Aging,
Neuropsychology, and Cognition, 4, 175-184.

Gemsbacher, M.A. & Faust, ME. (1991). The mechanism of suppression: A component
of general comprehension skill. Journal of Experimental Psychology: Learning,
Memory, and Cognition, I 7, 245-262.

Gemsbacher, M.A., Robertson, R.R.W., & Werner, N.K. (2001). The costs and benefits
of meaning. In: D. Gorgein (Ed.), 0n the consequences of meaning. Washington,
DC: APA.

Gemsbacher, M.A., Varner, K.R., & Faust, ME. (1990). Investigating differences in
general comprehension Skill. Journal of Experimental Psychology: Learning,
Memory, and Cognition, 16, 430-445.

Goodyear, P. & Hynd, G.W. (1992). Attention-deficit disorder with (ADD/H) and

without (ADD/W O) hyperactivity: Behavioral and neuropsychological
differentiation. Journal of Child Psychology, 21(3), 273-305.

229

Gordon, M. (1979). The assessment of impulsivity and mediating behaviors in
hyperactive and nonhyperactive children. Journal of Abnormal Child
Psychology, 7, 317-326.

Hagman, J .O., Wood, F., Buchsbaum, M.S., Tallal, P., Flowers, L., & Katz, W. (1992).
Cerebral brain metabolism in adult dyslexic subjects assessed with positron
emission tomography during performance of an auditory task. Archives of
Neurology, 49, 734-739.

Halperin, J .M., Newcom, J .H., Koda, V.H., Pick, L., McKay, K.E., & Knott, P: (1997).
Noradrenergic mechanisms in ADHD children with and without reading
disabilities: A replication and extension. Journal of the American Academy of
Child and Adolescent Psychiatry, 36, 1688-97.

Halperin, J .M., Newcom, J .H., Schwartz, S.T., McKay, K.E., Bedi, G., Sharma, V.
(1993). Plasma catecholamine metabolite levels in ADHD boys with and without
reading disabilities. Journal of Clinical Child Psychology, 22(2), 219-225.

Halperin, J., Gittelman, R., Klein, D., & Rude], R. (1984). Reading disabled hyperactive
children: A distinct subgroup of attention deficit disorder with hyperactivity?
Journal of Abnormal Child Psychology, 12, 1-14

Hamlett, K.W., Pellegrini, D.S., & Connors, CK. (1987). An investigation of
executive processes in the problem solving of attention deficit disorder-
hyperactive children. Journal of Pediatric Psychology, 12, 227-240.

Hamm, V. (1992). Age and the availability of inferences. Psychology and Aging, 7, 56-
64.

Hamishfeger, K.K. (1995). The development of cognitive inhibition: Theories,
definitions, and research evidence. In: F.N. Dempster and OJ. (Eds).
Interference and Inhibition in Cognition. New York: Academic Press, pp. 175-
201.

Hamishfeger, K.K. & Bjorkland, DP. (1993). The ontogeny of inhibition mechanisms:
A renewed approach to cognitive development. In: M.L. Howe and R. Pasnak
(Eds). Emerging Themes in Cognitive Development: Volume I: Foundations.
New York: Springer-Verlag, pp. 29-49.

Hasher, L. and Zacks, RT. (1988). Working memory, comprehension, and aging: A
review and a new view. In: G.G. Bower (Ed.). The Psychology of Learning and
Motivation. San Diego, CA: Academic Press, pp. 231-237.

230

Hellman, K.M., Voeller, K.K.S., & Nadeau, SE. (1991). A possible pathophysiologic
substrate of attention deficit hyperactivity disorder. Journal of Child
Neurology,6(Suppl), S74-S79.

Hinshaw, SP. (1992). Academic underachievement, attention deficits, and aggression:
Comorbidity and implications for intervention. Journal of Consulting and
Clinical Psychology, 60, 893-903.

Houghton, S., Douglas, G., West, J., Whiting, K., Wall, M., Lansford, 8., Powell, L.,
Carroll, A. (1999). Differential patterns of executive function in children with

attention-deficit hyperactivity disorder according to gender and subtype. Journal
of Child Neurology, 14, 801-805.

Humphries, T., Koltun, H, Malone, M., & Roberts, W. (1994).
Teacher-identified oral language difficulties among boys with attention problems.
Developmental and Behavioral Pediatrics, 15, 92-98.

Hynd, G.W., Semrud-Clikeman, M., Lorys, A.R., Novey, E.S., Eliopulos, D. (1990).
Brain morphology in developmental dyslexia and attention deficit
disorder/hyperactivity. Archives of Neurology, 47, 919-926. Don’t have article.

J avorsky, J. (1996). An examination of youth with attention-deficit/hyperactivity
disorder and language learning disabilities: A clinical study. Journal of Learning
Disabilities, 29(3), 247-258.

Jensen, P.S., Martin, D., & Cantwell, D. (1997). Comorbidity in ADHD: Implications
for research, practice, and DSM-IV. Journal of the American Academy of Child
and Adolescent Psychiatry, 36(8), 1065-1079.

Kahmi, A. & Catts, H. (1986). Toward an understanding of developmental language and
reading disorders. Journal of Speech and Hearing Disorders, 51, 337-347.

Kataria, S., Hall, C., Wong, M. and Keys, G. (1992). Learning styles of learning disabled
and non-leaming disabled attention deficit hyperactivity disorder children.
Journal of Clinical Psychology, 48, 371-378.

Keith, R.W. & Engineer, P. (1991). Effects of methylphenidate on the auditory
processing abilities of children with attention deficit-hyperactivity disorder.
Journal of Learning Disabilities, 24(10), 630-636.

Kemp, S.L. & Kirk, U. (1993). An investigation of frontal executive dysfunction in
attention deficit disorder subgroups. Annals of the New York Academy of
Sciences, 782, 363-365.

Kershner, J .R. & Graham, NA. (1995). Attentional control over language lateralization
in dyslexic children: Deficit or delay? Neuropsychologia, 33, 39-51.

231

Kershner, J .R. & Morton, LL. (1990). Directed attention dichotic listening in reading
disabled children: A test of four models of maladaptive lateralization.
Neuropsyhchologia, 28, 181-198.

Klorman, R., Hazel-Femandez, L.A., Shaywitz, S.E., Fletcher, J .M., Marchione, K.E.,
Holahan, J .M., Stuebing, K.K., & Shaywitz, BA. (1999). Executive functioning
deficits in attention-deficit/hyperactivity disorder are independent of oppositional
defiant or reading disorder. J. Am. Acad. Child Adolesc. Psychiatry, 38(9), 1148-
1155.

Kujala, T., Myllyviita, K., Tervaniemi, M., Alho, K., Kallio, J ., & N°°t'nen, R. (2000).
Basic auditory dysfunction in dyslexia as demonstrated by brain activity
measurements. Psychophysiology, 37, 262-266.

Kupietz, SS. (1990). Sustained attention in normal and in reading-disabled youngsters
with and without ADHD. Journal of Abnormal Child Psychology, 18, 357-373.

Lahey, B.B., Applegate, B., McBumett, K., Biederman, J ., Grenhill, L., Hynd, G.W. et al.
(1994). DSM-IV field trials for attention deficit hyperactivity disorder in children
and adolescents. American Journal of Psychiatry, 15], 1673-1685.

Lamminmaki, T., Ahonen, T. Nahri, V., Lyytinen, H., & deBarra, HT. (1995). Attention
deficit hyperactivity disorder subtypes: Are there differences in academic
problems? Developmental Neuropsychology, 11(3), 297-310. '

Landau, S. & Milich, R. (1988). Social communication patterns of attention-deficit-
disordered boys. Journal of Abnormal Child Psychology, 16, 69-81.

Light, 1.6., Pennington, B.F., Gilger, J .W., DeFries, J.C. (1995). Reading disability and
hyperactivity disorder: Evidence for a common genetic etiology. Developmental
Neuropsychology, 11(3), 323-335.

Loeber, R. & Keenan, K. (1994). Interaction between conduct disorder and its comorbid
conditions: Effects of age and gender. Clinical Psychology, 14(6), 497-523.

Long, D.L., Oppy, B.J., & Seely, MR. (1997). Individual differences in readers’
sentence- and text-level representations. Journal of Memory and Language, 36,
129-145.

Long, D.L., Oppy, B.J., & Seely, MR. (1994). Inidvidual differences in the time course
of inferential processing. Journal of Experimental Psychology: Learning,
Memory, and Cognition, 20, 1456-1470.

Lorch, E.P., Sanchez, R.P., van den Broek, P., Milich, R., Murphy, E.L., Lorch, R.F.,
Welsh, R. (1999). The relation of Story structure properties to recall of television

 

stories in young children with attention-deficit hyperactivity disorder and
nonreferred peers. Journal of Abnormal Child Psychology, 27(4), 293-309.

Lorch, R.F. Jr. & van den Broek (1997). Understanding reading comprehension: Current
and future contributions of cognitive science. Contemporary Educational
Psychology, 22, 213-246.

Lou, H.C., Andresen, J ., Steinberg, B., McLaughlin, T., & Friberg, L. (1998). The
striatum in a putative cerebral network activated by verbal awareness in normals
and in ADHD children. European Journal of Neurology, 5, 67-74.

Lou, H.C., Henriksen, L., Bruhn, P., Bomer, H, & Nielsen, J .B. (1989). Striatal

dysfunction in attention deficit and hyperkinetic disorder. Archives of Neurology,
46, 48-52.

Love, A.J. & Thompson, M.G.(1988). Language disorders and attention deficit disorders
in young children referred for psychiatric services. American Journal of
Orthopsychiatry, 58, 52-63.

Ludlow, C.L., Rapoport, J .L., Bassich, C.J., Mikkelsen, E.G. (1980). Differential effects
of dextroamphetamine on language performance in hyperactive and normal boys.
In: Treatment of Hyperactive and Learning Disordered Children. Ed: R.M.
Knights, DJ Bakker. Baltimore: University Park Press, pp. 185-205.

Mariani, M.A. & Barkley, RA. (1997). Neuropsychological and academic functioning in
preschool boys with attention deficit hyperactivity disorder. Developmental
Neuropsychology, 13(1), 111-129.

MacDonald, M.C. & Christiansen, M.H. (2002). Reassessing working memory:
Comment on Just and Carpenter (1992) and Waters and Caplan (1996).
Psychological Review, 109, 35-54.

MacLeod, C. (1991). An Integrative Review, Psych. Bull, 109, 163-203.

McGee, R., Partridge, F., Williams, S., Silva, RA. (1991). A twelve year follow-up of
preschool hyperactive children. Journal of the American Academy of Child and
Adolescent Psychiatry, 30, 224-232.

McGee, R. & Share, D.L. (1988). Attention deficit disorder-hyperactivity and academic
failure: Which comes first and what should be treated? Journal of the American
Academy of Child and Adolescent Psychiatry, 27, 318-325.

”McGee, R., Williams, S., Moffitt, T.E., & Anderson, J. (1989). A comparison of boys

with attention and/or reading disorders on neuropsychological measures. Journal
of Abnormal Child Psychology, 17, 37-53.

233

 

McGrew, K.S. & Woodcock, R.W. (2001). Woodcock-Johnson-III Tests of Achievement.
Itasca, IL: Riverside Publishing.

Mclnnes, A., Humphries, T., Hogg-Johnson, S., Tannock, R. (2003). Listening
comprehension and working memory are impaired in Attention-Deficit

Hyperactivity Disorder irrespective of language impairment. Journal of Abnormal
Child Psychology. 31(4), 427-443.

McPherson, D.L. & Davies, K. (1995). Preliminary observations of binaural hearing in
an attention-deficit pediatric population. F iziologiia Cheloveka, 21(1), 47—53.

Moores, E. & Andrade, J. (2000). Ability of dyslexic and control teenagers to sustain
attention and inhibit resonses. European Journal of Cognitive Psychology, 12,
520-540.

Morris, R.D., Stuebing, K.K., Fletcher, J .M., Shaywitz, S.E., Lyon, G.R., Shankweiler,
D.P. et al. (1998). Subtypes of reading disability: Variability around a
phonological core. Journal of Educational Psychology, 90, 347-373.

Neill, W.T., Valdes, L.A., & Terry, K.M. (1995). Selective attention and the inhibitory
control of cognition. In: F.N. Dempster and CJ. Brainerd (Eds). Interference
and Inhibition in Cognition. Academic Press, 207-259.

Nigg, J .T. (2001). Is ADHD a Disinhibitory Disorder? Psychological Bulletin, I 2 7, 571-
598.

Nigg, J .T. (2000). On inhibition/disinhibition in developmental psychopathology: Views
from cognitive and personality psychology and working inhibition taxonomy.
Psychological Bulletin, 126, 220-246.

N igg. J .T. (1999). The ADHD response-inhibition deficit as measured by the Stop Task:
Replication with DSM-IV combined type, extension, and qualification. Journal
of Abnormal Child Psychology, 27, 393-402.

Nigg, J .T., Blaskey, L., Huang-Pollock, C., & Rappley, MD. (2002).
Neuropsychological executive functions and ADHD DSM-IV subtypes. Journal
of the American Academy of Child and Adolescent Psychiatrv. 41, 59-66.

N igg, J.T., Butler, K.M., Huang-Pollock, C., & Henderson, J .M. (2002). Inhibitory
processes in adults with persistent childhood onset ADHD. Journal of Consulting
and Clinical Psychology, 70(1), 153-157.

Ni gg, J .T., Hinshaw, S.P., Carte, E.T., & Treuting, J .J . (1998). Neuropsychological
correlates of childhood attention-deficit/hyperactivity disorder: Explainable by

comorbid disruptive behavior or reading problems? Journal of Abnormal
Psychology, 107 (3), 468-480.

234

 

Nottleman, E.D. & Jensen, RS. (1995). Comorbidity of disorders in children and
adolescents: Deve10pmental perspectives. Advances in Clinical Child
Psychology, 17, 109-155.

Offord, D.R., Boyle, M.H., Szatmari, P., Rae-Grant, N.I., Links, P.S., Cadman, D.T. et al.
(1987). Ontario Child Health Study: H. Six-month prevalence of disorder and
rates of service utilization. Archives of General Psychiatry, 44, 832-836.

O’Neill, M. & Douglas, V. (1991). Study strategies and story recall in attention deficit
disorder and reading disabilities. Journal of Abnormal Child Psychology, 1 9,
671-679.

Oosterlaan, J ., Logan, G.D., & Sergeant, J .A. (1998). Response inhibition in AD/HD,
CD, comorbid AD/HD-l-CD, anxious, and control children: A meta-analysis of
studies with the stop task. Journal of Child Psychology and Psychiatry, 39, 411-
425.

Omoy, A., Uriel, L., & Tennenbaum, A. (1993). Inattention, hyperactivity
and speech delay at 2—4 years of age as a predictor for ADD-ADHD syndrome.
Israel Journal of Psychiatry and Related Sciences, 30, 155-163.

Pearlmutter, N .J . & MacDonald, M.C. (1995). Individual differences and probabilistic
constraints in syntactic ambiguity resolution. Journal of Memory and Language,
34, 521-542.

Pennington, BE. (1991). Diagnosing Learning Disorders: A Neuropsychological
Framework. New York: Guilford Press.

Pennington, B.F., Groisser, D., & Welsh, M.C. (1993). Contrasting cognitive deficits in
attention deficit hyperactivity disorder versus reading disability. Developmental
Psychology, 29(3), 511-523.

Pennington, B.F. & Ozonoff, S. (1996). Executive functions and developmental
psychopathology. Journal of Child Psychology and Psychiatry, 37(1), 51-87.

Prizant, B.M., Audet, L.R., Burke, G.M., Hummel, L.J., Maher, S.R., & Theadore, G.
(1990). Communication disorders and emotional/behavioral disorders in children
and adolescents. Journal of Speech and Hearing Disorders, 55, 179-192.

Purvis, KL. & Tannock, R. (2000). Phonological processing, not inhibitory control,

differentiates ADHD and Reading Disability. Journal of the American Academy
of Child and Adolescent Psychiatry, 39, 485-494.

235

Purvis, KL. & Tannock, R. (1997). Language abilities in children with attention deficit
hyperactivity disorder, reading disabilities, and normal controls. Journal of
Abnormal Child Psychology, 25(2), 133-144.

Quay, BC. (1997). Inhibition and attention deficit hyperactivity disorder. Journal of
Abnormal Child Psychology, 25, 7-13.

Reader, M.J., Harris, E.L., Schuerholz, L.J., Denckla, MB. (1994). Attention deficit
hyperactivity disorder and executive dysfunction. Developmental
Neuropsychology, 10(4), 493-512.

Reynolds, C.R. & Kamphaus, R.W. (1992). Behavioral assessment system for children.
Circle Pines, MN: American Guidance Service.

Riccio, C.A., Hynd, G.W., Cohen, M.J., Hall, J. et al. (1994). Comorbidity of central
auditory processing disorder and attention-deficit hyperactivity disorder. Journal
of the American Academy of Child and Adolescent Psychiatry, 33 , 849-857.

Riccio, C.A. & Hynd, G.W. (1993). Deve10pmental language disorders in children:
Relationship with learning disability and attention deficit hyperactivity disorder.
School Psychology Review, 22(4), 696-709.

Riccio, C.A. & Jemison, SJ. (1998). ADHD and emergent literacy: Influence of
language factors. Reading and Writing Quarterly: Overcoming Learning
Difficulties, 14, 43-58.

Roth, F.P., Spekman, N.J., & Fye, EC. (1995). Reference cohesion in the oral narratives
of students with learning dsabilities and normally achieving students. Learning
Disability Quarterly, 18, 25-40.

Rowe, K.J. & Rowe, KS. (1992). The relationship between inattentiveness in the
classroom and reading achievement (Part B): An explanatory study. Journal of
the American Academy of Child and Adolescent Psychiatry, 31, 357-368.

Ramsey, J .M., Andreason, P. Zametkin, A.J., Aquino, T., King, C., Hamburger, S.D. et
al. (1992). Failure to activate the left temporoparietal cortex in dyslexia: an

oxygen 15 positron emission tomographic study. Archives of Neurology, 49, 527-
534.

Rutter, M. Prevalence and types of dyslexia. In: A.L. Benton and D. Pearl (Eds).
Dyslexia, an Appraisal of Current Knowledge. New York: Oxford University
Press, 3-28.

Sattler, J .M. (1992). Assessment of children (3rd Ed.). San Diego: Author.

236

 

Schachar, R. & Logan, GD. (1990). Impulsivity and inhibitory control in normal
development and childhood psychopathology. Developmental Psychology, 26,
710-720.

Schaughency, E.A. & Hynd, G.W. (1989). Attentional control systems and the attention
deficit disorders (ADD). Learning and Individual Differences, I , 423-449.

Schulte, A.C., Conners, C.K., & Osborne, SS. (1999). Linkages between attention
deficit disorders and reading disability. In: D.D. Duane (Ed.). Reading and

Attention Disorders: Neurobiological Correlates. Baltimore: York Press, 161-
184.

Segalowitz, S.J., Wagner, WJ. & Menna, R. (1992). Lateral versus front ERP predictors
of reading skill. Brain and Cognition, 20, 85-103.

Semel, E., Wiig, E.H., & Secord, W. (1995). Clinical Evaluation of Language
F undamentals-Third Edition. San Antonio: The Psychological Corporation.

Semrud-Clikeman, M., Biederman, J., SpriCh-Buckminster, S., Lehman, B., Faraone, S.,
& Norman, D. (1992). Comorbidity between ADHD and learning disability: A
review and report of a clinically referred sample. Journal of the American
Academy of Child and Adolescent Psychiatry, 31, 439-448.

Shafer, J .L. & Graham, J .W. (2002). Missing data: Our view of the state of the art.
Psychological Methods, 7, 147-177.

Shaffer, D., Schwab-Stone, M., Fisher, P. et al. (1993). The Diagnostic Interview
Schedule for Children-- Revised Version (DISC-R): 1. Preparation, field testing,
interrater reliability, and acceptability. Journal of the American Academy of
Child and Adolescent Psychiatry, 32, 643-650.

Shankweiler, D., Lundquist, E., Katz, L., Stuebing, K.K., Fletcher, J.M., Brady, S. et al.
(1999). Comprehension and decoding: Patterns of association in children with
reading difficulties. Scientific Studies of Reading, 3, 69-94.

Shaywitz, B.A., Fletcher, J .M., Holahan, J.M., Shneider, A.E., Marchione, K.E.,
Stuebing, K.K. et al. (1995). Interrelationships between reading disability and
attention-deficit/hyperactivity disorder. Child Neuropsychology, 1(3), 170-186.

Shaywitz, S.E., Fletcher, J .M., & Shaywitz, BA. (1994). Issues in the definition and
classification of attention deficit disorder. Topics in Language Disorders, 14, 1-
25.

Shaywitz, S.E., Shaywitz, B.A., & Fletcher, J .M., and Escobar, MD. (1990). Prevalence

of reading disability in boys and girls: Results of the Connecticut longitudinal
Study. Journal of the American Medical Association, 264, 998-1002.

237

 

Shelton, T.L. & Barkley, RA. (1994). Critical issues in the assessment of
attention deficit disorders in children. Topics in Language Disorders, 14, 26-41.

Siege], L.S. & Ryan, EB. (1989). The development of working memory in normally
achieving and subtypes of learning disabled children. Child Development, 60,
973-980.

Simpson, G.B. & AdamOpoulos (2001). Repeated homographs in word and sentence
contexts: Multiple processing of multiple meanings. In: D.S. Gorfein (Ed.). 0n
the Consequences of Meaning Selection: Perspectives on Resolving Lexical

Ambiguity. Washington, DC: American Psychological Association, pp. 105-117.

Simpson, G.B. & Foster, MR. (1986). Lexical ambiguity and children’s word
recognition. Developmental Psychology. 22, 147-154.

Simpson, G.B. & Lorsbach, TC. (1983). The development of automatic and conscious
components of contextual facilitation. Child Development, 54, 760-772.

Simpson, G. B., Krueger, M.A., Kang, H., & Elofson, A. (1994). Sentence context and
meaning frequency effects in children's processing of ambiguous words. Journal
of Research in Reading, 17(1), 62-72.

Spekman, N. (1981). Dyadic verbal communication abilities of learning disabled and
normally achieving fourth and fifth grade boys Learning Disability Quarterly, 4,
193-201.

Spreen, O. (1989). The relationship between learning disability, emotional disorders,
and neuropsychology: Some results and observations. Journal of Clinical and
Experimental Neuropsychology, 11, 117-140.

Stanford, L.D. & Hynd, G.W. (1994). Congruence of behavioral symptomatology in
children with ADD/H, ADD/WO, and Learning Disabilities. Journal of Learning
Disabilities, 2 7(4), 243-253.

Stanovitch, KB. (1986). Matthew effect in reading: Some consequences of individual
differences in acquisition of literacy. Reading Research Quarterly, 21, 360-406.

Stringer, R. & Kershner, J. (1993). Possible attentional origins of word decoding deficits
in dyslexia. In: SF. Wright and R. Groner (Eds). Facets of Dyslexia and its
Remediation. Amsterdam: Elsevier Publishers, pp., 357-370.

Tannock, R., Martinussen, R., & Frijters, J. (2000). Naming speed performance and
stimulant effects indicate effortful, semantic processing deficits in attention-
deficit/hyperactivity disorder. Journal of Abnormal Child Psychology, 28, 237-
252.

238

 

Tannock, R., Purvis, K. & Schachar, R. (1993). Narrative abilities in children with
attention deficit hyperactivity disorder and normal peers. Journal of Abnormal
Child Psychology, 21, 103-117.

Tannock, R. & Schachar, R. (1996). Executive dysfunction as an underlying mechanism
of behavior and language problems in attention deficit hyperactivity disorder. In:
J.H. Beitchman, NJ. Cohen et al. (Eds). Language, Learning, and Behavior
Disorders: Developmental, Biological, and Clinical Perspectives. Cambridge:
Cambridge University Press, pp. 128-155.

Tamowski, K., Prinz, R., & Nay, S. (1986). Comparative analysis of attentional deficits
in hyperactive and leaming-disabled children. Journal of Abnormal Psychology,
95, 341-345.

Tarver-Behring, S., Barkley, R.A., & Karlsson, J. (1985). The mother-child interactions
of hyperactive boys and their normal siblings. American Journal of
Orthopsychiatry, 55(2), 202-209.

Tirosh, E., Berger, J ., Cohen-Ophir, M., Davidovitch, M., Cohen, A. (1998). Learning
disabilities with and without attention-deficit hyperactivity disorder: Parents’ and
teachers’ perspectives. Journal of Child Neurology, 13(6), 270-276.

Tirosh, E. & Cohen, A. (1998). Language deficit with attention-deficit disorder: A
prevalent comorbidity. Journal of Child Neurology, 13, 493-497.

Tomblin, J .B., Zhang, X., Buckwalter, P., & Catts, H. (2000). The association of reading
disability, behavioral disorders, and language impairment among second-grade
children. Journal of Child Psychology and Psychiatry, 41, 473-482

Tompkins, CA. (1991). Automatic and effortful processing of emotional intonation after
right hemisphere brain damage. Journal of Speech and Hearing Research, 34,
820-830. '

Tompkins, C.A., Baumgaertner, A., Lehman, M.T., & Fassbinder, W. (2000).
Mechanisms of discourse comprehension impairment after right hemisphere brain
damage: Suppression in lexical ambiguity resolution. Journal of Speech,
Language, and Hearing Research, 43, 62-78.

Tompkins, C.A., Lehman, M.T., & Baumgaertner, A. (1999). Suppression and inference
revision in right brain-damaged and non brain-damaged adults. Aphasiology, 13,
725-742.

Trautman, R., Giddan, J., Jurs, S. (1990). Language risk factor in emotionally disturbed

children within a school and day treatment program. Journal of Child
Communication Disorders, 13, 123-133.

239

 

Trueswell, J.C., Sekerina, 1., Hill, N.M., Logrip, ML. (1999). The kindergarten-path
effect: Studying on-line sentence processing in young children. Cognition, 73,
89-134.

Twilley, L.C., Dixon, P., Taylor, D., & Clark, R. (1994). University of Alberta norms of
relative meaning frequency for 566 homographs. Memory & Cognition, 22(1),
111-126.

Vallance, D.D., Im, N., & Cohen, N .J . (1999). Discourse deficits associated with
psychiatric disorders and with language impairments in children. Journal of
Child Psychology and Psychiatry, 40, 693-704.

Voeller, K.K.S. (1999). Neurological factors underlying the comorbidity of attentional
dysfunction and dyslexia. In: D.D. Duane (Ed) Reading and Attention
Disorders: Neurobiological Correlates. Baltimore: York Press, 185-212.

Westby, C.E. & Cutler, SK. (1994). Language and ADHD: Understanding the bases and
treatment of self-regulatory deficits. Topics in Language Disorders, 14, 58-76.

Whalen, C.K., Henker, B., Swanson, J .M., Granger D., Kliewer, W., Spencer, J.
(1987). Natural social behaviors in hyperactive children: Dose effects of
methylphenidate. Journal of Consulting and Clinical Psychology, 55, 187-193.

Wood, F.B. & Felton, RH. (1994). Separate linguistic and attentional
factors in the development of reading. Tapics in Language Disorders, 14, 42-57.

Wood, F.B. & Flowers, D.L. (1999). Functional neuroanatomy of dyslexic subtypes. In:
D.D. Duane (Ed.). Reading and Attention Disorders: Neurobiological Correlates.
Baltimore: York Press, 129-160.

Willcutt, E.G., Pennington, B.F., Boada, R., Ogline, J.S., Tunick, R.A., Chhabildas, N.A.,
& Olson, R.K. (2001). A comparison of the cognitive deficits in reading
disability and attention-deficit/hyperactivity disorder. Journal of Abnormal
Psychology, 110, 157-172.

Zentall, SS. (1993). Research on the educational implications of attention deficit
hyperactivity disorder. Exceptional Children, 60, 143-153.

Zentall, SS. (1988). Production deficiencies in elicited language but not in the

Spontaneous verbalizations of hyperactive children. Journal of Abnormal Child
Psychology, 16, 657-673.

240

 

   

llrllllzljllltljlllllljjrl