Articulator involvement in naming

MISSING IMAGE

Material Information

Title:
Articulator involvement in naming a test of the articulatory feedback hypothesis of naming
Alternate title:
Test of the articulatory feedback hypothesis of naming
Physical Description:
xi, 146 leaves : ill. ; 29 cm.
Language:
English
Creator:
Lu, Lisa Hsiao-Jung, 1971-
Publication Date:

Subjects

Subjects / Keywords:
Feedback   ( mesh )
Names   ( mesh )
Knowledge   ( mesh )
Articulation Disorders   ( mesh )
Department of Clinical and Health Psychology thesis Ph.D   ( mesh )
Dissertations, Academic -- College of Health Professions -- Department of Clinical and Health Psychology -- UF   ( mesh )
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph.D.)--University of Florida, 2000.
Bibliography:
Bibliography: leaves 141-145.
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Lisa Hsiao-Jung Lu.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 024732629
oclc - 53182203
System ID:
AA00022281:00001

Table of Contents
    Title Page
        Page i
    Acknowledgement
        Page ii
    Table of Contents
        Page iii
        Page iv
        Page v
    List of Tables
        Page vi
        Page vii
        Page viii
    List of Figures
        Page ix
    Abstract
        Page x
        Page xi
    Introduction
        Page 1
        Page 2
        Page 3
        Page 4
        Page 5
        Page 6
        Page 7
        Page 8
        Page 9
        Page 10
        Page 11
        Page 12
        Page 13
        Page 14
        Page 15
        Page 16
        Page 17
        Page 18
        Page 19
        Page 20
        Page 21
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
        Page 27
        Page 28
        Page 29
        Page 30
        Page 31
    Methods
        Page 32
        Page 33
        Page 34
        Page 35
        Page 36
        Page 37
        Page 38
        Page 39
        Page 40
        Page 41
        Page 42
        Page 43
        Page 44
        Page 45
        Page 46
        Page 47
        Page 48
        Page 49
        Page 50
    Results
        Page 51
        Page 52
        Page 53
        Page 54
        Page 55
        Page 56
        Page 57
        Page 58
        Page 59
        Page 60
        Page 61
        Page 62
        Page 63
        Page 64
        Page 65
        Page 66
        Page 67
        Page 68
        Page 69
        Page 70
        Page 71
        Page 72
        Page 73
        Page 74
        Page 75
        Page 76
        Page 77
        Page 78
        Page 79
        Page 80
        Page 81
        Page 82
        Page 83
        Page 84
        Page 85
        Page 86
        Page 87
        Page 88
        Page 89
        Page 90
        Page 91
        Page 92
        Page 93
        Page 94
        Page 95
        Page 96
    Discussion
        Page 97
        Page 98
        Page 99
        Page 100
        Page 101
        Page 102
        Page 103
        Page 104
        Page 105
        Page 106
        Page 107
        Page 108
        Page 109
        Page 110
        Page 111
        Page 112
        Page 113
        Page 114
        Page 115
        Page 116
        Page 117
        Page 118
        Page 119
        Page 120
        Page 121
        Page 122
        Page 123
        Page 124
    Appendix 1. Articulatory awareness test
        Page 125
        Page 126
        Page 127
        Page 128
        Page 129
        Page 130
        Page 131
        Page 132
        Page 133
        Page 134
    Appendix 2. Attention-deficit/hyperactivity disorder interview
        Page 135
        Page 136
        Page 137
        Page 138
    Appendix 3. NAPM stimuli
        Page 139
        Page 140
    References
        Page 141
        Page 142
        Page 143
        Page 144
        Page 145
    Biographical sketch
        Page 146
        Page 147
        Page 148
        Page 149
        Page 150
Full Text









ARTICULATOR INVOLVEMENT IN NAMING:
A TEST OF THE ARTICULATORY FEEDBACK HYPOTHESIS OF NAMING














By

LISA HSIAO-JUNG LU


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY, OF FLORIDA


2000













ACKNOWLEDGMENT


I would like to thank each of my committee members for their guidance: Eileen

Fennell for her clinical wisdom, conceptual inquiry, and generous support; Ken Heilman

for the opportunity to approach neuropsychology through clinical hypothesis testing;

Bruce Crosson for his belief in my abilities; Duane Dede for helping me think broadly;

and Jamie Algina for his patient and thorough teaching of statistics. I have been very

lucky in crossing the paths of mentors who have such vigor and integrity both

professionally and personally. I would also like to thank Ann Alexander, Linda

Lombardino, Tim Conway, and the American Psychological Association for their support

of this project. To my family, especially to my husband, I would like to say thank you

for your undying support through these most challenging years.













TABLE OF CONTENTS


Page
ACKNOWLEDGMENT .......................................................................... ii

LIST O F TA BLES ..................................... ...................... ......... ........ ... vi

LIST OF FIGURES .. ...... ......................... ......... ......... ...... ix

A B STR A C T ...................................... ..................... .......................... x

IN TRO D U CTIO N ................................ .................. ............................. 1

Development of the Hypothesis ...................................................... 1
Liberman's Motor Theory of Speech Perception .......................... 1
Heilman's Motor-Articulatory Feedback Hypothesis ..................... 3
Anatomy of the Articulatory Feedback System ........................... 4
Proposed Hypothesis: Articulatory Feedback Hypothesis of Naming ........... 6
Developmental Phonological Dyslexia .............................................. 8
Definition of Developmental Phonological Dyslexia ..................... 8
Nature and Extent of Naming Deficit ....................................... 14
Role of Phonological Awareness ............................................. 18
Anatomical Evidence of Anomalies ......................................... 20
Co-morbidity with Attention-Deficit/Hyperactivity Disorder ............ 26
R research Q questions ..................................................................... 27
What Is the Correlation Between Articulatory Knowledge and
N am ing? ........................................................ ...... .. 28
Do Dyslexics Have Worse Articulatory Knowledge? ..................... 28
Is There Support for the Articulatory Feedback Hypothesis of
N am ing? ........................................................ ...... .. 29
What Is the Relationship Between Articulatory Knowledge and
Phonological Awareness? ............................................ 30

M ETH O D S ......................................... ............................................ .. 32

Subjects ............................................ ................................. .... 32
Descriptive M measures ................. ............... ........................... ...... .. 38
Articulatory Awareness Test .................................................. 38
N am ing ...................................... ........................... ...... .... 40
I'







Phonological Awareness ..................................................... 41
Attention-Deficit/Hyperactivity Disorder ................................... 41
Experimental Measures ................................................................. 42
Naming Assessed via Phoneme Match (NAPM) ........................... 42
V isual M atch ..................................................................... 45
Phoneme Match ................................................................. 46
N am ing Test ................................... ................................ .. 47
Procedures ......................................... .............................. ...... .... 48

R E SU LT S ............................................ ..................... ............... ...... .... 51

Articulatory Knowledge ................................................................. 51
Phonologically Impaired vs. Controls ................................................ 53
Articulatory Awareness Test .................................................. 54
Descriptive Measures .......................................................... 55
Experimental Measures ........................................................ 57
Predictors of Articulatory Knowledge ....................................... 68
Predictors of Phonological Awareness ...................................... 70
Developmental Phonological Dyslexics vs. Adequate Readers with
Poor Phonology vs. Controls .................................................. 72
Articulatory Awareness Test .................................................. 73
Descriptive Measures .......................................................... 75
Experimental Measures ........................................................ 76
Predictors of Articulatory Knowledge ....................................... 81
Predictors of Phonological Awareness ...................................... 84
Poor vs. Adequate Articulatory Knowledge .......................................... 85
Descriptive Measures .......................................................... 86
Experimental Measures ........................................................ 88

DISCUSSION ...................................... ......................... .................. 97

Review of Hypothesis .................................................................. 97
Correlation Between Articulatory Knowledge and Name Retrieval ............. 100
Group Differences in Articulatory Knowledge ..................................... 103
Group Differences on Naming Measures .......................................... 104
Reaction Time and Response Accuracy ................................... 106
Interference Movement Frequency ......................................... 110
Attention-Deficit/Hyperactivity Disorder .......................................... 113
Relationship Between Articulatory Knowledge and Phonological
A w areness ...... ............ .................... ... .. .... .. ... .... ............ 114
Articulatory Feedback Hypothesis of Naming .................................... 118
Lim stations .............................................................................. 12 1
Summary of Findings ................................. ........................... 123
Correlation Between Articulatory Knowledge and Naming .............. 123
Dyslexics Do Not Have Worse Articulatory Knowledge ................. 123
Relationship Between Articulatory Knowledge and Phonological







A w areness ....................... ...................................... 124
Modification of the Articulatory Feedback Hypothesis of Naming .... 124

APPENDIX 1 ARTICULATORY AWARENESS TEST ................................ 125

APPENDIX 2 ATTENTION-DEFICIT/HYPERACTIVITY DISORDER
IN TER V IEW .......................... ........................................ 135

APPENDIX 3 NAPM STIMULI ............................................................. 139

REFER EN CES ................................... .................. ............................ 141

BIOGRAPHICAL SKETCH .................................................................. 146













LIST OF TABLES


Table Page

1. Summary of grouping criteria ..................................................... 36

2. Summary of demographics and grouping criteria scores ....................... 38

3. AAT scores of the Morris Center population ................................... 40

4. Sample of the chart for determining the order of task, interference,
and stimulus set for each subject ................................................. 49

5. Order of test administration .......................... .............................. 50

6. Pearson correlations between the Articulatory Awareness Test (AAT)
score and reaction time on experimental measures ............................. 53

7. AAT and AAT-R scores obtained by Phonologically Impaired (PI)
and Control (CTRL) groups ........................................................ 54

8. Pearson correlations between the AAT score and reaction time on
experimental measures for PI and CTRL groups ............................... 55

9. PI and CTRL groups' performance on descriptive measures ................. 56

10. Means and standard deviations of reaction time (RT in milliseconds)
and accuracy (% Correct) on the Phoneme Match Test....................... 58

11. Reaction time and accuracy on the NAPM and Visual Match Tests
for PI and CTRL groups............................................................. 59

12. Means and standard deviations of reaction time for each block .............. 61

13. Response accuracy (percentage) reflecting the
Task X Interference X Block interaction ...................................... 62

14. Comparison of overall findings with Block 1 and Block 2 findings ......... 63







15. Block 1 reaction time and accuracy on the NAPM and Visual
Match Tests for PI and CTRL groups ........................................... 64

16. Block 2 reaction time and accuracy on the NAPM and Visual
Match Tests for PI and CTRL groups ........................................... 64

17. Reaction time and accuracy of the Non-ADHD and ADHD subgroups
on the NAPM and Visual Match Tests ......................................... 66

18. Interfering movement frequency index (i.e., number of movements
per second) for the PI and CTRL groups ........................................ 68

19. Pearson correlations between the AAT score and interfering movement
index for PI and CTRL groups ................................................... 69

20. Pearson correlations between the AAT score and variables entered
into stepwise regression analysis for PI and CTRL groups ................... 70

21. Pearson correlations between the LAC score and variables entered
into stepwise regression analysis for PI and CTRL groups ................... 71

22. Means and standard deviations of AAT scores obtained by DPD,
ARPP, and CTRL groups ........................... .... ... .. .... .. .... ............ 73

23. Pearson correlations between the AAT score and reaction time on
experimental measures for DPD, ARPP, and CTRL groups .................. 74

24. Means and standard deviations on descriptive measures for the
DPD, ARPP, and CTRL groups ................................................... 75

25. Reaction time and accuracy on the Phoneme Match Test for
DPD, ARPP, and CTRL groups ................................................... 76

26. Reaction time and accuracy on the NAPM and Visual Match Tests
for DPD, ARPP, and CTRL groups ............................................. 77

27. Reaction time and accuracy of the phonologically impaired
Non-ADHD and ADHD subgroups .............................................. 79

28. Interfering movement frequency index for the DPD, ARPP, and
C TR L groups ................................................................. ..... .. 80

29. Pearson correlations between the AAT score and interfering movement
frequency index for DPD, ARPP, and CTRL groups ......................... 82







30. Pearson correlations between the AAT score and variables entered
into stepwise regression analysis for DPD and ARPP groups ................ 83

31. Pearson correlations between the LAC score and variables entered
into stepwise regression analysis for DPD and ARPP groups .................. 85

32. Demographics of the Poor Articulatory Knowledge (PAK) and
Adequate Articulatory Knowledge (AAK) groups ............................. 86

33. Means and standard deviations on descriptive measures for the
PAK and AAK groups ............................. ................................. 87

34. Reaction time and accuracy on the Phoneme Match Test for the
PA K and AAK groups ............................. ................................. 88

35. Reaction time and accuracy on the NAPM for PAK and AAK groups.
Numbers represent data without the covariate extracted ....................... 89

36. Reaction time and accuracy on the Visual Match Test for PAK
and A A K groups .............................................................. ..... 93

37. Interfering movement frequency index for the PAK and AAK groups ....... 95













LIST OF FIGURES


Figure Page

1. A simplified model of reading from Ellis and Young (1988)................ 11

2. Block 1 NAPM reaction time, plotted against the ability to match end
phone es ................................... .................. ...................... 91

3. Block 2 NAPM reaction time, plotted against the ability to match end
phone es ...................................... ..................................... 92

4. Formula for calculating the effect size reflecting the Group X Task X
Interference interaction .......................... ......................... ........ 96













Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

ARTICULATOR INVOLVEMENT IN NAMING:
A TEST OF THE ARTICULATORY FEEDBACK HYPOTHESIS OF NAMING

By

LISA HSIAO-JUNG LU

August, 2000

Chair: Eileen B. Fennell
Major Department: Clinical and Health Psychology

The articulatory feedback hypothesis of naming posited that articulatory feedback

facilitates name retrieval. This was tested using an interference paradigm. Naming

performance during a condition that allowed for articulatory feedback was contrasted

with a condition that interfered with articulatory feedback by providing inappropriate

articulatory feedback. Because Montgomery found that dyslexic children had impaired

articulatory knowledge, performance of phonologically impaired readers was contrasted

with those of normal readers. Subjects were also grouped by their level of articulatory

knowledge, and performance between knowledgeable and unknowledgeable groups was

compared. One assumption of the hypothesis was that those with adequate articulatory

knowledge would benefit from articulatory feedback while those with poor articulatory

knowledge would not. The hypothesis predicted that interfering with articulatory

feedback would affect subjects who have articulatory knowledge by removing the







facilitation effects provided by articulatory feedback. Results did not directly support the

hypothesis. For individuals with articulatory knowledge, naming latency during the

condition that allowed for articulatory feedback was not better than the condition that

interfered with feedback. Subjects did not spontaneously use articulatory feedback to

assist name retrieval. However, other data did suggest a relationship between articulatory

knowledge and name retrieval. Among individuals with poor articulatory knowledge,

inappropriate articulatory feedback and name retrieval interfered with each other and

competed for neural resources. This suggested a neural connectivity between articulatory

knowledge and name retrieval that was not evident between articulatory knowledge and a

nonverbal control task. Those with articulatory knowledge appeared to have processed

name retrieval automatically and efficiently, and they had sufficient extra neural

resources to process extraneous information such as interfering feedback. In contrast,

those with poor articulatory knowledge retrieved names less efficiently. They had

limited capacity to simultaneously process interfering information while engaging in

name retrieval. It was also found that articulatory knowledge and phonological

awareness were dissociable phenomena. Both normal and phonologically impaired

readers demonstrated a wide range of articulatory knowledge, and dyslexic children did

not have worse articulatory knowledge.













INTRODUCTION


The problem of name retrieval is one that has received extensive attention in the

neuropsychology literature. By focusing on this limited aspect of language, researchers

hope to generalize the knowledge learned here to other aspects of linguistic functioning.

The term name retrieval has been defined differently by different research groups. Here,

name retrieval refers only to the activation of phonological representation of a word and

does not include activation of motor patterns to produce such a representation. The

current project proposes to test a new hypothesis of name retrieval, the articulatory

feedback hypothesis of naming. First, the development of this hypothesis from

Liberman's motor theory of speech perception and Heilman's theory of motor-articulatory

feedback will be presented. Then the articulatory feedback hypothesis of naming will be

proposed. This hypothesis was tested with a population of children who have

phonological dyslexia of the developmental type. Therefore a discussion of dyslexia,

related name retrieval issues, neurological anatomy of this population, and co-morbidity

with Attention-Deficit/Hyperactivity Disorder (ADHD) will follow.


Development of the Hypothesis


Liberman's Motor Theory of Speech Perception


Liberman, Cooper, Shankweiler, and Studdert-Kennedy first proposed their motor

theory of speech perception in 1967. Extensive research following the first proposal of







their theory has led to subsequent revisions of their theory, the most recent of which was

presented in Liberman and Mattingly, 1985. Their current theory on speech perception

was based on two tenets: 1) The object of speech perception is the intended gestures of

the speaker; 2) speech perception and speech production are innately (i.e., biologically)

linked, not learned.

The first tenet of their theory speaks to why this theory is a "motor" one instead of

a "sensory" one. Unlike auditory theories, which posit that perception of speech depends

on an analysis of auditory signals, Liberman and Mattingly (1985) proposed that the goal

of speech perception is not to uncode the auditory signals, but to infer the intended

gestures of the speaker's vocal system. They argued that the uncoding of auditory cues

cannot be sufficient for speech perception because there is no correspondence between

acoustic signals and phonemic categories. Acoustic signals for the same phonemic

category vary by speaker, prosodic tone, and context. Though acoustic signals are

different in these different conditions, the same phonemic percept is perceived.

Conversely, the exact same acoustic signal under different contexts can yield different

phonemic percepts. The lack of a relationship between acoustic signals and phonemic

categories suggests that acoustic signal by itself is not sufficient for the perception of

speech. Furthermore, that visual feedback of oral gestures can influence the perception

of a speech sound (McGurk & MacDonald, 1976; MacDonald & McGurk, 1978)

suggests that both visual and acoustic signals are merely cues for the object of perception.

Liberman and Mattingly (1985) argued that the object of speech perception is the

intended gesture, or the actual motor movement, of the speaker. The object is the







intended gesture because much of the gesture takes place inside the speaker's oral cavity

and out of sight of the perceiver.

These motor theorists propose that the speech perception system is able to decode

intended gestures from auditory signals because speech perception system is a specialized

neural module evolved to perform such linguistic functions. They assumed that the

development of motor control over the vocal tract preceded the evolution of speech.

Adaptations made coarticulation of rapid phonetic gestures possible. A perceiving

system developed concomitantly, and this system is specialized to take into account

complex acoustic consequences. Because the perception system developed

concomitantly with the production system, they are biologically linked. The calculations

necessary to perceive speech are done automatically via hardwired neural structures that

connect the production and the perception parts of the system. This system is one

specialized for linguistic functions. It is a linguistic module that operates independently

from the general auditory system that processes other non-linguistic signals.


Heilman's Motor-Articulatory Feedback Hypothesis


Heilman, Voeller, and Alexander (1996) elaborated on Liberman's theory and

proposed a motor-articulatory feedback hypothesis of speech perception. They

emphasized that the perception of spoken words is associated with the production of

intended articulatory gestures. As an infant learns to perceive words, s/he imitates the

sounds heard by replicating intended articulatory gestures of the speaker with his/her own

articulators. As the infant fine-tunes this imitation of sounds, s/he associates each

phoneme with a movement of his/her articulators. Feedback from the articulators to the







neural module is essential for the individual to be aware of this relationship between

phoneme and articulatory gesture. Understanding of this relationship constitutes

articulatory knowledge, which facilitates parsing spoken words down into phonemic

parts.

Heilman et al. (1996) applied their theory to the problem of grapheme-to-

phoneme conversion in reading. They pointed out that when first learning to read, one

needs to break words down into letters clusters and associate them with their respective

phonemes. The motor-articulatory feedback hypothesis posits that children use the

articulatory apparatus when learning to associate specific graphic representations with

phonemic representations. Learning to read includes using the articulatory knowledge

learned earlier to associate phonemes with graphemes. Without this articulatory

knowledge as a mediating tool, the means by which written words are broken down into

letter clusters may seem arbitrary. Reading thus involves associating already established

phoneme-articulatory relationships to graphemic representations. Beginning readers

often move their lips and tongue even when reading silently. Adults also engage the

articulators when reading novel or hard-to-pronounce words. These findings suggest that

engaging the articulators facilitates reading. The motor-articulatory feedback hypothesis

proposed that it is the feedback that articulators provide which facilitates the learning of

grapheme-to-phoneme conversion.


Anatomy of the Articulatorv Feedback System


Central to the human language system are two major neural regions, Wernicke's

and Broca's areas in the left cerebral hemisphere of most right-handed individuals (Kolb







& Whishaw, 1990). Wemicke's area, which is located in the left posterior portion of the

superior temporal gyrus, and surrounding regions are sometimes referred to as posterior

language areas that process the perception of language. Classically, a lesion in

Wemicke's area results in fluent aphasia, characterized by fluent speech but impaired

comprehension, repetition, and naming. Information from these posterior perisylvian

regions travels anteriorly to Broca's area via the arcuate fasciculus. Broca's area, located

in the left inferior frontal gyrus, is conceptualized as an area specialized for motoric

programming of speech. Lesion of Broca's area results in nonfluent aphasia characterized

by intact comprehension but effortful, nonfluent speech.

The motor-articulatory hypothesis proposed by Heilman et al. (1996) emphasizes

the learned association between a phoneme and an articulatory gesture. When an infant

hears a novel word, his/her auditory and auditory association cortex (Wernicke's area)

analyze the sounds in the word. Pars opercularis, triangularis, and the foot of the motor

cortex (Broca's area) execute a complex motor program to approximate the heard word.

Primary motor cortex activates the articulators in the oral cavity. During word imitation,

as the articulators move, they send sensory feedback (i.e., proprioceptive and tactile) to

the primary sensory cortex and sensory association cortex. These sensory cortices

presumably connect with the frontal areas involved in motor planning (e.g., Broca's area),

thus providing linkage back to the articulatory system. The sensory areas also project to

polymodel sensory cortex in the temporal-parietal region that stores auditory

representations of words (e.g., Wernicke's area). Eventually, through this connectivity,

the infant learns that each phoneme is associated with one motor pattern for articulation,

and that words are associated with a series of articulatory patterns.








Proposed Hypothesis: Articulatory Feedback Hypothesis of Naming


Articulatory knowledge assists reading presumably because it facilitates retrieval

of information. In the case of reading novel words where it is necessary to use

grapheme-to-phoneme conversion, the subject sees a grapheme, and this grapheme

triggers the phoneme associated with it and the articulatory motor pattern used to produce

that phoneme; execution of this motor pattern results in the production of the phoneme

associated with the target grapheme. Movement of the articulators may not be necessary

for the retrieval of the phoneme, and feedback may even be intracerebral.

If articulatory feedback assists reading, it is likely that it also assists name

retrieval, a more basic language function that came into use much earlier in the

evolutionary process than reading. The articulatory feedback hypothesis of naming states

that articulatory feedback facilitates name retrieval. It does not state that articulatory

feedback is necessary for naming or sufficient for naming. Rather, articulatory feedback

assists other retrieval systems, making the process faster and more efficient.

Many of our everyday experiences suggest that activation of motor patterns can

facilitate retrieval. The tip-of-the-tongue phenomenon, where one experiences problems

retrieving a word, may sometimes be overcome by articulating the beginning sound.

When one cannot recall a phone number, pretending that one is dialing, thereby engaging

the motor system controlling the fingers, sometimes assist the recall of phone numbers.

Whereas retrieval may not need motor activation, if some part of the retrieval system is

compromised, motor activation may provide the extra input necessary to activate a

representation.







Heilman et al. (1996) posited that in order to benefit from articulator feedback,

one must have articulatory knowledge. Articulatory knowledge refers to the ability to

locate the position of the articulatory structures, such as the tongue and the lips, during

phoneme production. This knowledge could be conceptualized as a neural association

between the articulatory system, the sensory/proprioceptive system, and the phonemic

representation system. Without this link, activation of the articulators may not coactivate

neural patterns representing phonemic percepts, which would limit the articulatory

system's ability to facilitate retrieval. The spontaneous use of articulatory knowledge will

be referred to as articulatory awareness.

Montgomery (1981) showed that children with dyslexia have impaired articulator

knowledge compared to children without dyslexia. She presented cartoons of sagittal

drawings through the oral cavity that illustrate the position of tongue, teeth, and lips, then

asked which of the cartoons matched phonemes that she produced. She encouraged the

subjects to repeat the phonemes as much as they want and to think about the location of

their tongue, teeth, and lips. The non-dyslexic children were able to correctly identify

their articulatory positions better than dyslexic children. All children were able to repeat

the phonemes; thus the dyslexic children's deficit cannot be explained by an auditory

perceptual deficit. They appeared to be unknowledgeable about the position and

movement of their articulators. They lacked articulatory knowledge.

Montgomery's (1981) work suggested that dyslexic children lack sensory

feedback about their articulators' position and movement. This population could be

instrumental in testing the articulatory feedback hypothesis of naming. The hypothesis

stated appropriate articulator movements (and the sensory feedback concomitant with







those movements) facilitate lexical retrieval in individuals with articulatory knowledge.

In the normal population, inhibiting appropriate articulator movements while asking

subjects to name objects should impede their naming. This was done in this study by

introducing an interference task. Subjects were asked to engage in mouth movements

that interfered with the articulation of object names. Thus the articulatory system could

send articulatory feedback to the central nervous system that was appropriate to and

facilitated name retrieval. Individuals with impaired articulatory knowledge should

respond differently. They could differ in one of two ways. First, because these

individuals may not be as knowledgeable of their articulator movements as normal

individuals, they may not use an articulatory strategy to assist naming. Therefore

interfering with the articulators during a naming task may not impede the naming process

of these individuals as much as it does in controls. However, individuals with impaired

articulatory feedback may have no other alternative strategy to assist naming. Because

they already have impaired articulatory feedback, adding an interference could impair

their naming ability even more, making them less capable of retrieving names than

controls.


Developmental Phonological Dyslexia


Definition of Developmental Phonological Dyslexia


The terms dyslexia, learning disability in reading, and reading disability have

been used interchangeably in the literature. All three refers to problems with reading, but

a clarification of terms is in order. The termdyslexia has been used in research

attempting to understand the neurological or neuropsychological deficits underlying the







disorder. Dyslexia can be categorized into developmental versus acquired. Acquired

dyslexia refers to those individuals who acquired the disorder through insults to the

central nervous system after a period of normal reading development (Coslett, 1997).

Developmental dyslexia refers to those individuals who demonstrate problems in the

development of reading. Researchers have posited the existence of many types of

dyslexia, including phonological, surface, and deep dyslexia, to name a few (Ellis &

Young, 1988). The Orton Dyslexia Society Research Committee has defined dyslexia as:

one of several distinct learning disabilities. It is a specific language-based
disorder of constitutional origin characterized by difficulties in single word
decoding, usually reflecting insufficient phonological processing abilities. These
difficulties in single word decoding are often unexpected in relation to age and
other cognitive and academic abilities; they are not the result of generalized
developmental disability or sensory impairment. Dyslexia is manifest by variable
difficulty with different forms of language, often including, in addition to
problems reading, a conspicuous problem with acquiring proficiency in writing
and spelling. (Shaywitz, Fletcher, & Shaywitz, 1995, p. S51)

This committee, composed of representatives from the National Institute of Child Health

and Human Development, defined dyslexia as a subtype of learning disability. Thus

learning disability is an umbrella term encompassing many types of problems with

learning, including reading and math. Reading disability is an abbreviated term for

learning disability in reading. This is a legal term used to identify individuals who meet

legal criteria to receive special education services. Dyslexia is used interchangeably with

reading disability, but it is a theoretical/research term, and its use implies neurological

abnormalities within the language system that underlie difficulty with reading processes.

Dyslexia constitutes 80% of children diagnosed with learning disability

(Shaywitz, Fletcher, & Shaywitz, 1995). Of the different types of dyslexia, phonological

dyslexia is the most common form. Phonological dyslexia refers to reading problems







secondary to phonological processing deficits. Specifically, impairment in the

grapheme-to-phoneme conversion system has been implicated. A simplified version of

Ellis and Young's (1988) model for oral reading is depicted in Figure 1. According to

this model, written words are processed initially by the visual system, which processes

visual stimuli by analyzing each individual letter component. After visual analysis,

reading can be achieved by three mechanisms or routes: 1) Results of the visual analysis

system enter into the orthographic input lexicon, which contains visual representations of

words an individual has learned. The selected lexical representation enters into the

semantic system and activates the meaning of the word. From the semantic system, the

appropriate auditory representation of the word is activated in the speech output lexicon.

Then speech is produced by activation of motor patterns. Most proficient readers are

thought to use this lexical system because of its efficiency and completeness compared to

the other two systems. 2) The second method for reading is similar to the first except that

the semantic system is bypassed, so that words can be read without accessing the

meaning of words. These two lexical, or whole-word reading routes are important for

reading irregular and ambiguous words, and cannot be used to read nonwords or

pseudowords. 3) The third method, the phonological route, is a labor-intensive system

used when reading novel words (i.e., words without representation in the orthographic

input lexicon). The visual system analyzes words and parses words into letter

components. The letter or letters graphemee) are converted to the sounds they represent

(phoneme). The phonemes are blended to produce the phonological sequence for the

entire word. Phonological dyslexia results when this grapheme-to-phoneme conversion

link is defective.




























Semantic
System


11

Written word




Visual
Analysis
System

/

Visual
Input
Lexicon




Grapheme-Phoneme
Conversion


Speech
Output
Lexicon


Phoneme
Level


Speech


Figure 1. A simplified model of oral reading from Ellis and Young (1988).







One problem with dyslexia research is that research groups often do not specify

the type of dyslexia subjects demonstrate. This is especially true with treatment-focused

research or service, where the primary goal is to improve patients' reading skills,

regardless of whether patients meet criteria for certain theoretical subtype of dyslexia.

This was true for Montgomery's (1981) work, which reported that subjects were

"dyslexic" without specification of subtype. The present study, in an attempt to strive for

theoretical clarity, will limit dyslexic participants to those with developmental

phonological dyslexia, defined as individuals who have impaired grapheme-to-phoneme

conversion, because this is the largest, most common subtype of dyslexia. Thus an

assumption of the present study is that a large percentage of participants in the

Montgomery (1981) study were phonological dyslexics, and that phonological dyslexics

have decreased awareness of their articulator position and movement.

The identification of developmental phonological dyslexia is problematic for two

reasons. First, dyslexia falls under the broad category of specific learning disability

under Individuals with Disabilities Education Act (IDEA, 1997; Public Law 105-17),

which requires that educational institutions provide special services to meet the

educational needs of individuals with disabilities. Because the law does not require the

specification of the subtype of reading disability, most clinical organizations do not

specify if a patient's reading disability fits the phonological subtype in diagnostic

evaluations. Second, the IDEA specified that reading ability should be discrepant from

intellectual aptitude, but it did not state how such discrepancy should be measured.

Different groups have used intelligence quotient (IQ)-achievement discrepancy

(Ackerman & Dykman, 1993; Cornwall, 1992), chronological age-reading age







discrepancy (Fawcett & Nicolson, 1994; Felton, Naylor, & Wood, 1990; Felton, Wood,

Brown, Campbell, & Harter, 1987; Wolf& Goodglass, 1986; Wolf& Obregon, 1992),

arbitrary cutoff scores on tests of achievement or based on teacher/school referrals

(Bowers & Swanson, 1991; Denckla & Rudel, 1976; Korhonen, 1995; Manis,

Seidenberg, Doi, McBride-Chang, & Petersen, 1996; Mattis, French, & Rapin, 1975;

Swan & Goswami, 1997), or regression approaches that control for the intercorrelation

between achievement and IQ measures (Fletcher, Francis, Rourke, Shaywitz, &

Shaywitz, 1992; Fletcher, Schaywitz, Shankweiler, Katz, Liberman, Stuebing, Francis,

Fowler, & Shaywitz, 1994; Pennington, Gilger, Olson, & DeFries, 1992; Shaywitz,

Escobar, Shaywitz, Fletcher, & Makuch, 1992; Shaywitz, Fletcher, Holahan, & Shaywitz,

1992; Shaywitz, Shaywitz, Fletcher, & Escobar, 1990). Consequently, the literature on

this population is fraught with inconsistent diagnostic criteria for reading disability.

The IQ-achievement discrepancy method has been shown to be problematic in

diagnosing reading disability among minority populations. Duckworth (1999) showed

that among a sample of college students referred for evaluation of learning disability,

African Americans score on average 12 points lower on the Wechsler Adult Intelligence

Scale-Revised (WAIS-R) intelligence quotients than their European-American

counterparts. Although psychologists have attempted to design culture-free intelligence

tests in recent years, Duckworth's data suggest that a commonly used intelligence

measure, WAIS-R, is still biased against minority populations. The IDEA specified that

learning disability cannot be due to mental retardation, which is defined as IQ scores of

below 70. If the normal distribution of African Americans' IQ scores is downshifted by

12 points, then the difference between the mean IQ score of the African-American







population and the mental retardation cutoff of 70 is decreased by 12 points, which in

effect, decreases the potential number of African Americans who can meet diagnosis for

learning disability using a simple difference discrepancy method.

An alternative to this problem is to calculate expected achievements scores based

on intellectual aptitude via a regression method. This method controls for the inter-

correlation between achievement and intelligence measures and the regression of

achievement scores toward the mean intelligence score, and minimizes problems of over-

identifying high-IQ subjects and under-identifying of low-IQ subjects as learning

disabled. Using the regression formula,

Y'= [r, (S, / Sy )(IQ X)] + Y

(where Y' is the expected achievement score for a given IQ, ry is the correlation between

the IQ and the achievement test, S, is the standard deviation of the achievement test, S" is

the standard deviation of the intelligence test, IQ is the achieved intelligence score, X is

the mean of the intelligence test, and Y is the mean for the achievement test), Duckworth

showed that among those referred for a learning disability evaluation, more African

Americans would be classified as learning disabled than using a simple discrepancy

difference method (51% vs. 28%, respectively), while the method of classification used

does not significantly affect the number of European Americans classified as learning

disabled (27% vs. 30%, respectively).


Nature and Extent of Naming Deficit


A basic question relates to the existence of bonafide naming deficits in the

dyslexic population. Because dyslexia is a reading disability, could their naming deficits







be attributable to a lack of vocabulary? If they do have bonafide naming problems, do

they have problems retrieving names of symbols that compose written language? Or do

they have a general retrieval problem that implicates naming of other targets? Fawcett

and Nicolson (1994) examined naming performance of 35 dyslexic children (defined as

having at least an 18 month discrepancy between chronological and reading age, with full

scale 1Q of at least 90) and 32 chronological age controls (CA). They found that

dyslexics have impaired naming of letters, digits, colors, and pictures compared to the

CA group, with the picture naming task being the most robust measure differentiating

groups. The dyslexics' discrepant performance on color and picture naming suggested

that their deficit was not limited to grapheme-to-phoneme translation. They have actual

problems with name retrieval.

A critique of Fawcett and Nicolson's (1994) study was that their groups were not

matched on intellectual aptitude and therefore retrieval differences may be explained by

intelligence differences between dyslexics and controls. Swan and Goswami (1997)

recruited a dyslexia group (n = 16), a CA control group (n = 16), a reading age (RA)

control group (n = 16), and a garden-variety poor reader control group (GV; n = 16). All

groups had matching IQ scores except the GV (101-105 vs. 79), and all groups had

matching reading age except the CA (112-116 mo. vs. 139 mo.). Swan and Goswami

(1997) found the following pattern of performance on a picture naming test (percentage

correct score based on the total number of items familiar to each subject):

CA>RA>GV=dyslexics. Dyslexics performed as well as GV. Both groups' naming

scores were worse than younger but reading age matched controls (RA). All of these

groups performed worse than the CA controls. However, dyslexics were more accurate







16
than any other group in correctly recognizing targets on a follow-up multiple choice task

composing of items they failed to name spontaneously. (Subjects' scores on this test were

coded as a proportion of their total error score, so scores were not inflated for those with

more errors). Dyslexics' ability to correctly identify targets in a recognition paradigm

argued against a vocabulary deficit. Rather, it supported that they have problems of

retrieval. In contrast, GV were found to have poorer vocabulary on a test of receptive

vocabulary. Swan and Goswami (1997) concluded that while GV's poor picture naming

performance was due to poor vocabulary, dyslexics' was due to problems with name

retrieval.

Wolf and Obregon (1992) found similar results using a multiple-choice paradigm

with items on the Boston Naming Test (BNT) that were missed. Their selection criterion

for dyslexia was better defined than the Swan and Goswami (1997) study: Dyslexics

were 2 or more years below expected reading level as assessed by the Gray Oral Reading

Test. Compared to an average reader control group (n = 42), dyslexics' (n = 8) naming

was worse, but dyslexics were more accurate on identifying the correct target in a

multiple-choice format compared to controls. They also concluded that dyslexics'

naming errors were reflective of a retrieval deficit.

These studies showed that dyslexics have lexical retrieval problems on formal

neuropsychological tests. Murphy, Pollatsek, and Well (1988) questioned 1) if dyslexics'

retrieval problem was one of general processing deficit or was it specific to language, and

2) whether dyslexics' retrieval deficit can be seen in their natural/spontaneous use of

language. They reasoned that if dyslexics have a general processing deficit, they should

be slower on tasks not involving the explicit use of language, such as a simple reaction







time task requiring them to move their finger to the side where a visual target appeared,

and on a picture categorization task requiring them to indicate if a picture is an exemplar

of a target category. If their retrieval deficit was specific to language, they should show

deficient performance on tasks of oral expressive and receptive language as well as on

formal neuropsychological measures. They tested dyslexics identified by poor Rapid

Automatized Naming (RAN) performance and who were at least two years below their

expected reading level (n = 14). Controls were matched for age and IQ (n = 14). They

found no difference between groups on basic motor reaction time and picture

categorization, which ruled out the general processing deficit hypothesis. Dyslexics

performed worse than controls on both formal (BNT) and informal language measures.

On informal, expressive language measures, dyslexics generated fewer words in retelling

stories and had slower verbal output. On informal, receptive language measures, they

were slower at categorizing spoken words. The authors concluded that dyslexics' name

retrieval problem reflected a specific linguistic deficit, and not a general processing

deficit, and their name retrieval problem manifested in their oral language as well as on a

formal neuropsychological measure.

The retrieval problems that dyslexic children demonstrate in childhood have been

shown to persist into adulthood. Korhonen (1995) followed a small group (n = 8) of

children who had problems in rapid automatic naming and in word retrieval, and tested

them approximately 9 years later at 18 years of age to examine the persistence of naming

deficits identified during childhood. These children were originally identified by their

teachers as learning disabled children who demonstrated special problems in reading.

Korhonen comparing these individuals' performance to controls matched on age, sex, IQ,







parent SES at nine years of age, and education level at 18 years of age (n = 10).

Korhonen found that learning disabled individuals were slower and made more errors on

rapid color naming and rapid object naming, and on another test of rapid alternating

stimulus naming. The findings were not as robust as at nine years of age; nevertheless

they were present. Fawcett and Nicolson (1994) tested dyslexics from eight to 17 years

of age and also found naming deficits in their 17-years old dyslexic group (n = 13).

Felton, Naylor, and Wood (1990) followed 115 children with dyslexia into adulthood.

They defined dyslexia as a discrepancy of 1.5 years between chronological and reading

age. They found persistent problems in rapid naming, nonword reading, and

phonological awareness. These findings of persistent naming problems suggested a

deficit model of dyslexia, which conceptualized dyslexia as a deficit that does not "catch

up" with maturation.


Role of Phonological Awareness


A hypothesized deficit underlying dyslexia is an impaired sense of phonological

awareness (Liberman & Shankweiler, 1985). Swan and Goswami (1997) used a picture

naming paradigm to study the role of phonological processing in dyslexics. They

hypothesized that if a phonological deficit underlies dyslexia, dyslexics' naming

performance would be worse for longer words of low frequency. Longer words have

more phonemes to encode and retrieve, and thus were more demanding on the

phonological system. Low frequency words occur less often in language, making them

less familiar to the phonological system. They found a Group X Frequency X Length

interaction, where with frequency controlled, dyslexics (n = 16) named short words better







19
than long words. This pattern was not seen in the CA, RA, or GV controls. Lower level

interactions also showed expected findings: Dyslexics named short words better than

GV, but their naming of long words was worse than GV and RA. Dyslexics named high

frequency words better than GV, but their naming of low frequency words was worse

than RA. Swan and Goswami (1997) further posited that dyslexics' picture naming

would be worse than word naming because in word naming, letters were available to

assist the phonological system. In picture naming, no cues were present to assist the

phonological system. They did find impaired picture naming compared to word naming

for dyslexics but not for RA and CA.

A natural question that arose with evidence of phonological and naming deficits

in dyslexia regards the relationship between these processes. Two studies have addressed

this issue but with incongruent results. Cornwall (1992) used a regression analysis to

examine if phonological awareness and rapid automatized naming contributed unique

variances to reading disabled children's scores on academic achievement (n = 54; reading

disability was defined by >= 16 standard point discrepancy between Wide Range

Achievement Test, Revised Reading subtest and WISC-R FSIQ, with WISC-R FSIQ >=

90). If phonological awareness and rapid automatized naming contributed unique

variances, then they were likely independent processes affecting the dyslexic population.

With age, SES, behavioral, and intelligence factors controlled, she found that

phonological awareness (as assessed by Auditory Analysis Test [AAT], a phonemic

deletion test) and rapid naming did contribute unique shares of variance to achievement

scores. Phonological awareness contributed to nonword reading, spelling, and

comprehension. Naming contributed to single-word reading and passage reading speed.







Bowers and Swanson (1991) also conducted regression analyses to examine the same

issue. They found that most variance on nonword reading (after controlling for the

WISC-R Vocabulary score) was explained by the score on the Auditory Analysis Test,

and most variance on a single-word reading test was explained by the score on Odd Word

Out, another phonological awareness measure (a multiple-choice test requiring

identification of the non-rhyming word). In contrast, most variance on comprehension

was explained by rapid automatized naming. A problem with Bowers and Swanson's

(1991) study was that they combined poor readers (n = 19; defined by Woodcock

Reading Mastery Test, Word Identification subtest standard score at or below the 25h

percentile for age) with average readers (n = 19) in their regression analyses. It was

possible that subjects with different reading abilities have different patterns of

relationship between phonological awareness and naming. That is, phonological

processes and naming abilities may contribute differently to the reading achievement

scores of average and disabled readers. The lack of well-controlled studies in this area

rendered the contributions of phonological awareness and naming to reading achievement

equivocal.


Anatomical Evidence of Anomalies


Galaburda and colleagues (Galaburda, Sherman, Rosen, Aboitiz, and Geschwind,

1985; Galaburda, 1989) examined eight post-mortem brains of individuals identified by

the Orton Society as dyslexic. They found abnormal symmetry of the planum temporale

in these eight brains as well as ectopic neurons in the molecular layer of the perisylvian

cortex. The planum temporale lies just posterior to the Heschl's gyrus on the superior







21
surface of the temporal lobe. These and other structures surrounding the Sylvian fissure

compose the language system, which includes reading. The findings ofGalaburda et al.

(1985) suggested that neurodevelopmental abnormalities may contribute to the

symptomatology of dyslexia.

Geschwind and Levitsky (1968) found that among 100 post-mortem samples,

approximately 65% showed a left greater than right plana difference. Approximately

25% had symmetrical plana, and only 10% had a right planum that was larger than the

left. Rumsey, Dorwart, Vermess, Denckla, Kruesi, and Rapoport (1986) measured

temporal lobe volume from magnetic resonance (MR) images and found data consistent

with the results ofGalaburda et al. (1985). Nine of the ten men with documented reading

disability in psychoeducational evaluations from their childhood demonstrated

symmetrical temporal lobes. However, Rumsey et al. (1986) did not measure the planum

temporale specifically. Given the extent of the temporal lobe, it was possible that other

aspects of the temporal lobe contributed to the symmetry rather than the planum

temporale.

In 1990, two independent groups reported on the symmetry of the planum

temporale among dyslexics. Hynd, Semrud-Clikeman, Lorys, Novey, and Eliopulos

(1990) compared plana length and insular length of 10 developmental dyslexic children

with 10 non-dyslexic controls. The average age of their dyslexic children was 10 years,

and dyslexia was defined by normal or better intellectual ability (WISC-R Full Scale IQ

>= 85), reading achievement significantly below their FSIQ (>= 20 standard score points

lower than FSIQ on Woodcock Reading Mastery Test--Revised, Word Attack and

Passage Comprehension subtests), and no co-morbid diagnosis of ADHD. The average







age of their normal controls was 12 years, and they must have normal or better

intellectual ability (WISC-R FSIQ >= 85), no reportedly family history of learning

problems, no significant deficit in achievement, and no reported or observed medical,

educational, social, or emotional difficulties. From MR images, this research group

found that dyslexics have bilaterally shorter insula compared to non-dyslexic controls,

and that 90% of dyslexics have a left planum length that was shorter than their right

planum length. There was no difference between dyslexic and control groups on right

planum length. Dyslexics' overall left planum was shorter than controls' left planum.

They suggested that the nature of plana symmetry in dyslexia was due to a smaller left

planum temporale. Larsen, Hoien, Lundberg, and Odegaard (1990) also examined plana

length from MR images and found symmetrical plana among dyslexic children (n = 19;

dyslexic subjects were identified from a school psychology service and had poor word

recognition in the presence of normal intelligence). However, comparing dyslexics' MR

images to those of age- and intelligence-matched controls' (n = 19), they found that plana

symmetry among dyslexics was due to increased right planum length rather than the

decreased left planum that Hynd et al. (1990) reported.

The above studies did not differentiate between the temporal and parietal banks of

the planum temporale. Leonard, Voeller, Lombardino, Morris, Hynd, Alexander,

Andersen, Garofalakis, Honeyman, Mao, Agee, and Staab, (1993) suggested that

examining the different banks of the planum may explain some of the contradictions in

the literature. They measured the length of these two banks from MR images. Their

subjects were adults previously diagnosed with dyslexia by pediatrician, pediatric

neurologist, or learning disability specialists (n = 9), the dyslexics' biological relatives (n







= 10), and normal controls (n = 12). In contrast to the previous studies, Leonard et al.

(1993) did not find abnormal symmetrical plana in the dyslexic population. All groups

demonstrated a greater left temporal bank compared to the right temporal bank and a

greater right parietal bank compared to the left parietal bank. When only the left

hemisphere was considered, all subjects except two dyslexics had longer temporal bank

than the parietal bank. When only the right hemisphere was considered, the controls also

had longer temporal bank than the parietal bank, but 55% of dyslexics and 40% of

relatives had longer parietal bank compared to the temporal bank. They suggested that

dyslexics had reduced right intrahemispheric asymmetry (between temporal and parietal

banks) compared to controls due to the transfer of planar tissue from the temporal to the

parietal bank.

The same group also examined the structure of the Sylvian fissure. Among

controls, the left Sylvian fissure usually ended in a bifurcation into small ascending and

descending branches. Variations to this typical pattern included no bifurcation and/or

extra gyri in the parietal operculum anterior to the termination of the Sylvian fissure.

There were also variations in the number ofHeschl's gyri present. Normally, there was

one Heschl's gyrus in the left hemisphere that was visible on a mesial section of the MR

image. On more lateral sections, Heschl's gyrus moved anteriorly and dissolved into a

number of convolutions in the superior temporal gyrus. Leonard et al. (1993) found that

every subject with dyslexia showed at least one of the above anomalies. Six (66%) had

bilateral anomalies. Biological relatives had the next highest number of anomalies;

seventy percent had at least one anomaly while 20% had bilateral anomalies. In contrast,

only 17% of control subjects had one anomaly and none had bilateral anomalies. These







24
findings suggested a genetic etiology for dyslexia. The greatest number of anomalies was

found among dyslexics, the group with the next greatest number of anomalies was

biological relatives of the dyslexics, and normal controls without family histories of

dyslexia had the fewest number of anomalies. These anatomical studies indicated that

reading difficulties experienced by dyslexics may have an anatomical basis. However, a

word of caution regarding the Leonard et al. (1993) study is in order. Their subjects were

either professionals or from high functioning professional families. They described their

dyslexic subjects as "recovered dyslexics" who have been able to compensate so well that

there was much overlap between dyslexic and control groups on a measure of

phonological awareness. Thus their dyslexic subjects may not be a representative sample

of the dyslexic population.

Hynd et al. (1990), in addition to finding shorter left planum length among

dyslexic children, also compared the MR images of dyslexics (n = 10) with MR images

of children with attention deficit/hyperactivity disorder (ADHD; n = 10). Their ADHD

subjects had average or better intellectual ability (WISC-R FSIQ >= 85), no reported

family history of learning problems, no significant deficit in reported or measured

achievement, documented behavioral deficits consistent with a Diagnostic and Statistical

Manual of Mental Disorders, Revised Third Edition (DSM-III-R) diagnosis of ADHD,

who responded favorably to stimulant medication. They found that dyslexics and

ADHDs both have smaller right frontal width compared to controls. However, dyslexics'

planum temporale was shorter on the left while ADHDs showed the typical pattern of left

greater than right planum. These findings showed that while frontal anomalies may be







implicated in both groups, anomalies of the planum temporale may be specific to

dyslexia.

Imaging data from cerebral blood flow studies also supported the evidence of

structural anomalies in the dyslexic population. Rumsey, Andreason, Zametkin, Aquino,

King, Hamburger, Pikus, Rapoport, and Cohen (1992) examined cerebral blood flow

differences between dyslexic adults and normal subjects during a rhyming task. Their

dyslexic subjects all had Wechsler Adult Intelligence Scale-Revised (WAIS-R) Verbal

or Performance IQ scores of at least 89 and met DSM-III-R criteria for developmental

reading disorder. All received some special education service while in school. Subjects

were presented word pairs aurally and pressed a button if the word pair rhymed. They

found that dyslexics had decreased activation of left temporal-parietal and midtemporal

areas that corresponded to the angular gyrus and Wernicke's area. This finding of

hypometabolism in temporal parietal and midtemporal areas corresponded well to the

structural abnormalities of planum temporal and Heschl's gyrus reported by Leonard et al.

(1993).

Paulesu, Frith, Snowling, Gallagher, Morton, Frackowiak, and Frith, (1996)

conducted a different rhyming task during a positron emission tomography (PET) study

and found similar results. Their rhyming task involved visual presentation of letters.

Subjects moved a joystick to letters that rhymed with the letter "B." Their subjects were

five dyslexic adults who were university students, postgraduates, or self-employed

entrepreneur, identified from records of a dyslexia clinic, and five education-matched

controls. The non-dyslexic subjects activated left Broca's and Wernicke's areas and the

left insula. Dyslexics showed decreased activation in the left Wernicke's area and a







26
greater decrease in the left insula. On a short-term memory task where subjects judged if

a target letter was present in a previous sequence of English letters, the normals activated

the above areas plus the left supramarginal gyrus. The dyslexics activated the same areas

as the controls except for the left insula. On these two tasks, dyslexics activated the same

major language areas as controls (i.e., Broca's and Wernicke's) while attending to and

judging phonological stimuli. However, they did not activate these areas in concert as

controls. Dyslexics' lack of activation of the insular cortex suggested that the insula was

not necessary for phonological processing. The authors suggested that perhaps the insula

acted as a "bridge" between the Broca's area and the supramarginal gyrus. Though it may

not be necessary for the processing of phonological information, it provided the

connection between posterior and anterior regions. Dyslexics' anatomical anomalies and

their lack of activation of this region during phonological analysis tasks suggested that

disconnection between important regions for phonological analysis may underlie their

problems with phonological processing.


Co-morbidity with Attention-Deficit'Hyperactivity Disorder


While reading disability and ADHD have very different symptoms, they do

overlap much more than one would expect from independent random distributions of

these disorders. Approximately 30-50% of individuals with reading disability have a co-

morbid diagnosis of ADHD (Felton et al., 1987). This high co-morbidity rate has led

researchers to speculate ifattentional problems limit a child's ability to develop

automated processing skills necessary for reading. Felton et al. (1987) aimed to

disentangle the neuropsychological deficits contributed by attention deficit disorder







(ADD) and by reading disability (RD). They formed four groups from two factors, RD

and ADD: RD with ADD, RD with non-ADD, non-RD with ADD, and non-RD with

non-ADD. Using age and receptive vocabulary score as covariates and controlling for

family-wise error rates, Felton et al. (1987) found that RD and non-RD groups differed

on a visual confrontation naming test (BNT) and on a rapid automatized naming test.

There was no main effect of ADD on these measures. The ADD and non-ADD groups

did differ from each other on a test of supraspan verbal memory (RAVLT). In contrast,

there was no main effect of RD on this task. These findings showed that RD and ADD

contributed to different aspects of neuropsychological deficits. If ADD contributed to

impaired reading skills among the RD children, one would expect some overlap of

impaired areas. The findings of Felton et al. (1987) provided indirect support for the

independence of reading disability and attention deficit disorder.


Research Questions


The literature on the dyslexic population indicated that 1) dyslexic individuals

have problems with name retrieval; 2) dyslexic individuals have problems with

phonological processing; and 3) their language difficulties likely have an anatomical

basis. This combination of findings rendered the phonological dyslexic population to be

of special interest to this study, because the anatomical areas identified to be abnormal

(i.e., Wernicke's area) were also implicated by the articulatory feedback hypothesis of

naming. This hypothesis posited that articulatory awareness facilitates naming.

Presumably, sensory feedback received by the primary sensory cortex from articulators

has connectivity with both the Wemrnicke's area and Broca's area. This connectivity






28
allows for articulatory feedback to trigger phonological representations of object names,

and to trigger motor patterns to execute the articulation of those names. Phonological

dyslexic subjects and normal readers should yield a range of naming abilities by which to

examine articulatory knowledge and the relationship between name retrieval and

articulatory knowledge. The aim of this study was to test the articulatory feedback

hypothesis of naming. To achieve this aim, the following questions were asked:


What Is the Correlation Between Articulatory Knowledge and Naming?


If articulatory knowledge and naming are related, better articulatory knowledge

should be associated with either faster name retrieval latency or better name retrieval

accuracy. This relationship should hold for all subjects. Reading achievement status

may put subjects at the lower end of the continuum of naming ability. If articulatory

feedback facilitates naming for all subjects, dyslexics' naming ability will be correlated

with their articulatory knowledge in the same manner as normal readers.


Do Dvslexics Have Worse Articulatory Knowledge?


A secondary aim of this study was to replicate Montgomery's (1981) finding that

dyslexics have impaired articulatory knowledge. This study differed from Montgomery's

(1981) study in some respects. One, it was unclear how Montgomery's dyslexic subjects

were defined. This study included only those who have impaired phonological skills as

measured by impaired grapheme-to-phoneme conversion. Second, because

Montgomery's (1981) version of the articulatory awareness test was unavailable, the







present study used an alternative but similar version of the test, which was based on

Montgomery's task.


Is There Support for the Articulatory Feedback Hypothesis of Naming?


Prediction for individuals with adequate articulatorv knowledge. The articulatory

feedback hypothesis of naming stated that having articulatory feedback appropriate to

naming facilitates name retrieval. This study tested this via an interference experimental

design. If having articulatory feedback appropriate to naming facilitates name retrieval,

interfering with that appropriate articulatory feedback should reduce facilitation effects.

Prediction for individuals with impaired articulator knowledge. The hypothesis

implied that those with poor articulatory knowledge will retrieve names less efficiently

than those with good articulatory knowledge. In an interference paradigm, where

subjects were asked to engage in another task that produced articulatory feedback

incompatible with the naming task at hand, those with poor articulatory knowledge were

predicted to perform differently than controls. Whereas the controls' naming should be

de-facilitated, those with poor articulatory knowledge may respond in one of two ways.

One, because they may not rely in the articulatory feedback system to facilitate name

retrieval in the first place, interfering with articulatory feedback may not produce de-

facilitation effects as expected with controls. Or possibly, because their articulatory

feedback system was already poor, adding another task with demands on the articulatory

feedback system may exacerbate the difficulty these subjects experience, leading to even

worse naming performance than controls' de-facilitated naming performance.







30
Group differences. If Montgomery's (1981) finding is supported and those with

phonological impairments have worse articulatory knowledge compared to controls, then

performance of phonologically impaired subjects can be compared to the performance of

controls. Even if Montgomery's finding is not supported, there is theoretical interest in

comparing the articulatory knowledge of these two groups as it has been well

documented that dyslexic individuals have name retrieval difficulties.

Those with phonological impairment can be further divided into two subgroups:

those with impaired reading skills and those with adequate reading skills. Performance of

these two subgroups can be compared to examine if these subtypes show different

patterns on articulatory knowledge and name retrieval, or if they differ only in the degree

of severity.

To test the hypothesis most directly, subjects can be grouped according to their

performance on a measure of articulatory knowledge. These three ways of grouping

subjects (i.e., phonologically impaired vs. controls; phonologically and reading impaired

vs. phonologically impaired with adequate reading vs. controls; poor articulatory

knowledge vs. adequate articulatory knowledge) may yield performance patterns that

further elucidate the relationship between articulatory knowledge and name retrieval.


What Is the Relationship Between Articulatory Knowledge and Phonological Awareness?


Another secondary aim of this study was to elucidate the relationship between

articulatory knowledge and phonological awareness. Much about phonological

awareness among the dyslexic population has been studied, but little is known about

articulatory knowledge. Are they related or independent of one another? What are the







31
factors that relate to or predict the level of articulatory knowledge and phonological

awareness?













METHODS


Subjects


Three groups of subjects totaling 41 children were recruited from the Gainesville,

Florida and Chicago, Illinois metropolitan areas. Subjects were recruited from offices of

psychologists, speech pathologists, and neurologists, and from flyers distributed

throughout the community. All subjects' parents gave written informed consent and all

subjects gave oral assent to participate in this study in accordance with the requirements

of the Institutional Review Board of the Health Science Center of the University of

Florida and of the University of Chicago Hospitals.

Inclusionary criteria for subjects included:

Age 7-12

Right-handed

Intelligence quotient between 70 and 130

English is first and primary language

A lower limit of 7 years of age was selected because reading disability is often not

apparent until school age; a large percentage of children have age-appropriate, limited

reading skills before that time. An upper age limit of 12 was selected because beyond

this age, children with developmental phonological dyslexia have had several years of

struggling with reading. They may have received special services or developed other

skills on their own in order to compensate for their impaired reading. While older

32







phonological dyslexic children may still demonstrate naming problems, their retrieval

deficit is often mitigated by late adolescence (Korhonen, 1995; Fawcett & Nicolson,

1994; Felton et al., 1990). Forms of compensation may confound the contribution of

articulatory feedback to name retrieval. The age range was limited between 7-0 and 12-

11 in order to include individuals in the early years of developmental phonological

dyslexia. Right-handedness was selected as a predictor of typical language organization

so that results from subjects in this study can be generalized to the population.

Approximately 98% of right-handed individuals are left hemisphere dominant for

language. The intelligence criterion was constrained to the middle 96% of the

population. Individuals at the extremes of the continuum may not process linguistic

information in the same way as most individuals. The intelligence criterion was set so

that results can be generalized to the population. English was required as subjects' first

and primary language in order to rule out reading problems due to socio-cultural or

environmental factors.

Exclusionary criteria included:

History of neurological disorders

Previous treatment in Lindamood or Orton-Gillingham programs

Family history of learning disability

A history of neurological disorders, such as cerebral palsy, epilepsy, or Tourette's

Syndrome, increases the probability of atypical brain organization. Therefore individuals

with neurological histories were excluded. Individuals who have participated in reading

treatment programs described as or based on the Lindamood or Orton-Gillingham

programs were also excluded. These treatment programs include direct or indirect







training of articulatory awareness via training of articulatory gestures. A history of

participation in these programs may confound results because subjects' articulatory

feedback system may no longer reflect its naturalistic connectivity. Family members of

individuals with a learning disability were also excluded because of anatomical studies

suggesting a genetic basis to learning disability (Leonard et al., 1993).

Subjects meeting criteria for the following three groups were recruited:

Developmental phonological dyslexia (DPD)

Adequate reader with poor phonology (ARPP)

Normal reader controls (CTRL)

Group membership was distinguished by performance on three reading

achievement subtests of the Woodcock Reading Mastery Test (WRMT; Woodcock,

1987): Word Identification, Passage Comprehension, and Word Attack. The Word

Identification subtest consisted of single English words that subjects were asked to read.

This subtest assessed subjects' oral reading of real words, but no comprehension of word

meaning was required. The Passage Comprehension subtest consisted of a short sentence

or paragraph with one missing word. Subjects were required to read the entire passage

and come up with one word that would fill the missing blank. This subtest assessed

subjects' comprehension of written material. The Word Attack subtest consisted of

nonwords that followed the rules of pronunciation in the English language. Subjects

were required to read these words aloud. This subtest required subjects to use the

grapheme-to-phoneme conversion route to read. Raw scores were converted to age-

corrected standard scores for each of these three measures.






35
Subjects in the DPD group had impaired reading achievement scores on all three

subtests in comparison to that expected given their intellectual aptitude, as assessed with

the Test of Nonverbal Intelligence, 2nd Edition (TONI-2; Brown, Sherbenou, & Johnsen,

1990). Impairment was operationalized as actual achievement score falling at least one

standard deviation (i.e., 15 standard score points) below the expected score, with the

expected score calculated using the formula

Y' = [r,. (S, / Sy XIQ X)] + Y

(Y' = expected achievement score, r,. = estimated correlation between the TONI-2 and

the WRMT, S, = standard deviation of WRMT [15], Sy = standard deviation of TONI-2

[15], IQ = obtained TONI-2 Quotient, X = mean of TONI-2 [100], and Y = mean of

WRMT [100]).

Subjects in the ARPP group had impaired phonological skills, operationalized by

impaired actual Word Attack score in comparison to the expected score based on TONI-2

Quotient, but non-impaired reading skills as defined by commensurate actual and

expected Word Identification and Passage Comprehension scores. These subjects, as

subjects in the DPD group, were recruited as poor readers. The categorization into DPD

or ARPP groups was done after each subject's completion of participation.

Subjects in the CTRL group were matched to the other two groups on age and

intelligence. The CTRL group's expected reading achievement scores based on

intellectual aptitude were all commensurate with actual achievement scores. The CTRL

group was not matched to the other two groups on reading age because the primary

purpose of this study was to evaluate name retrieval ability. The ability to retrieve names







36
may be affected by age and intelligence. Thus all three groups were matched on age and

intelligence. Table 1 summarized the grouping criteria.

Table 1. Summary of grouping criteria._
Group Word Attack Word Identification Passage Comprehension

DPD ASS < ESS ASS < ESS ASS < ESS

ARPP ASS < ESS ASS = ESS ASS = ESS

CTRL ASS = ESS ASS = ESS ASS = ESS

Note: DPD = Developmental Phonological Dyslexia; ARPP = Adequate Reader with
Poor Phonology; CTRL = Controls; ASS = actual standard score; ESS = expected
standard score based on the formula, Y' = [ry (S, / Sy )(IQ X)] + Y; IQ = TONI-2
Quotient.


The TONI-2 was selected as the measure of intellectual ability for this study

because of its relative lack of dependence on verbal abilities. A problem with measuring

dyslexic children's general intelligence is that most common measures of intelligence

rely heavily on verbal abilities. Because dyslexic children have impaired reading skills,

they may obtain intelligence scores that are lower than their "true" intellectual capability.

To minimize this problem, intelligence quotient of the TONI-2 was used as the measure

of intelligence for this study. The TONI-2 stimuli consisted of visual patterns with a

missing piece. Subjects were required to choose a pattern to fit the visual sequence from

multiple choices. While good performance on this test may still benefit from

verbalization, the TONI-2 has less verbal demands than most other measures of

intelligence. The TONI-2 can be used with subjects from age 5 to 85. Normative data

was collected from over 2,700 subjects in these age ranges. The TONI-2 Form B's

correlation with the WISC-R FSIQ ranged from .75 to .94. Its correlation with WISC-R






37
VIQ ranged from .63 to .73, and its correlation with WISC-R PIQ ranged from .60 to .87.

Form B was selected to be used in this study because of its relatively more stable

correlation with WISC-R indices compared to Form A. Raw scores on the TONI-2 were

converted to age-corrected TONI-2 Quotients, which have a mean of 100 and a standard

deviation of 15.

Table 2 summarized each group's demographic data and grouping criteria scores.

All three groups were matched on age (F = 0.35,p = 0.71), intelligence score (F = 1.50,p

= 0.24), and grade (F = 0.74,p = 0.48). More males were represented in the DPD group

in comparison to the other two groups. The three groups did differ from each other on

the three reading achievement subtests (F = 7.30, p = 0.00). For each of the three reading

achievement subtests, the DPD group scored uniformly lower than the other groups

(Word Attack: DPD vs. ARPP, t = 2.67, p = 0.02, DPD vs. CTRL, t = 9.65, p = 0.00;

Word Identification: DPD vs. ARPP, t = 4.86,p = 0.00, DPD vs. CTRL, t = 9.36,p =

0.00; Passage Comprehension: DPD vs. ARPP, t = 6.48, p= 0.00, DPD vs. CTRL, t =

8.10, p = 0.00). The ARPP group scored lower than the CTRL group on Word Attack (t

= 7.14,p = 0.00) and Word Identification (t = 4.60,p = 0.00), but not on Passage

Comprehension (t = 1.99,p = 0.06). Within the DPD and CTRL groups, there was no

difference between any of the three achievement scores (DPD, F = 2.82, p = 0.11; CTRL,

F = 1.24,p = 0.32). The ARPP group, however, demonstrated better Passage

Comprehension compared to Word Identification (t = 7.40, p = 0.00), which in turn was

better than Word Attack (t = -8.71, p = 0.00).







Table 2. Summary of demographics and grouping criteria scores.
DPD ARPP CTRL
. . . . . . . . . . . . . . ...... ..... . .. .. ..... .... ....... .. . ....,. -. .... ..... .. ..... .. .. ..... ..- .-.- .. ... . .. .... . . . . . . .
N 11 10 20

Age 9.0(1.1) 9.4(1.9) 9.5(1.7)

Grade 3(1) 4(2) 4(2)

M:F Ratio 9:2 6:4 12:8

ADHD M:F Ratio 5:1 3:0 3:0

TONI-2 IQ 103(13) 110(7) 106(8)

Word Attack SS 70(11)8 81 (8)b 104 (8)c

Word Identification SS 68 (12)a 89 (7)d 105 (10)c

Passage Comprehension SS 70 (12)Y 96 (5)C 103 (10)Y

Note: DPD = Developmental Phonological Dyslexia; ARPP = Adequate Reader with
Poor Phonology; CTRL = Controls. Scores with different superscripts indicate
statistically significant difference.


Descriptive Measures


Subjects were administered a battery of relevant tests to compare performance

between groups and for comparison with other findings in the literature. Data from the

following descriptive measures were used to describe the groups on relevant

neuropsychological variables.


Articulatory Awareness Test (AAT)


The Articulatory Awareness Test (AAT) is a non-published, experimental

instrument modeled after Montgomery's (1981) task. The AAT stimuli consisted of eight

picture cards with cartoon drawings of the sagittal view of the oral cavity (see Appendix







1). Each picture represented one or more phoneme. The examiner produced a target

phoneme with her mouth obstructed from subject's view, then asked the subject to

identify one of out three pictures that corresponds to subject's articulatory gesture as s/he

produced that phoneme. Subjects were encouraged to repeat the target phoneme as many

times as necessary, and no time limit to responding was imposed. Three practice items

were given before test items were administered, and the examiner went over a sagittal

cartoon drawing to identify each articulator (i.e., tongue, teeth, lips) at the introduction of

the task. In the event that subject produced an atypical articulatory gesture in producing a

phoneme, the subject's gesture was used in scoring accuracy.

The AAT was produced by the Morris Center of Gainesville, Florida and used as

part of their standard evaluation for reading disability. The AAT consisted of 10 trials,

with 10 additional trials added and piloted during this study (AAT-R). Data on the

original 10 trials were available from 93 patients from the Morris Center. Seventy

percent of these subjects were male (n = 65). Thirty percent were female (n = 28).

Patients' ages ranged from 6 to 22, with an average of 11 years (standard deviation = 4).

Of these 93 subjects, 86 were classified as dyslexic, 7 were classified as borderline

dyslexic; 81 of these subjects were diagnosed with co-morbid ADHD. Table 3 shows

group means on the AAT (out of possible 10 trials). Dyslexic subjects performed more

poorly on the AAT than borderline dyslexics (t = 2.03,p = 0.04). AAT scores did not

differ between ADHD and non-ADHD groups (t = 1.34, p = 0.18). Because the Morris

Center population did not include normal readers, it was not known whether including

normal readers would yield a bimodal distribution of AAT scores, as Montgomery (1981)

showed with her version of this task. The data available from the Morris Center







40
suggested a normal distribution of scores on the AAT from the dyslexic population. This

test was administered to all subjects in the present study to estimate subjects' level of

knowledge about articulator position during phoneme production. The number of

accurate responses in 10 trials (AAT) and in 20 trials (AAT-R) was recorded for analysis.


Table 3. AAT scores of the Morris Center population.
n Mean AAT score (SD)

Dyslexic 86 7 (2)

Borderline Dyslexic 7 8(1)



ADHD 81 7(2)

Non-ADHD 12 8(1)



Total 93 7(2)



Naming


Subjects were administered the Boston Naming Test (BNT; Kaplan, Goodglass,

Weintraub, & Segal, 1983) as a measure of visual confrontation naming. Z-scores

calculated from the norms published by Spreen and Strauss (1998) were recorded for

analysis. Subjects were also administered the Rapid Color Naming and Rapid Object

Naming subtests of the 1997 experimental version of the Comprehensive Test of

SPhonological Processing, which was same as the 1999 published version of the same test

(Wagner, Torgesen, & Rashotte, 1999). Z-scores based on normative data collected in

1997 by the research group developing this battery were calculated and recorded for
Ns







analysis. These measures assessed subjects' rapid naming ability and allowed for

comparison of data from the present subject groups to other findings reported in the

literature.


Phonological Awareness


The Lindamood Auditory Conceptualization Test (LAC; Lindamood &

Lindamood, 1979) was administered to all subjects to yield an index of phonological

awareness. This test assessed subjects' phonological awareness by asking subjects to

manipulate color blocks, with each color representing one phoneme. Phoneme patterns

changed in degrees of difficulty, and subjects manipulated blocks to demonstrate their

perception of how phoneme patterns changed. Raw scores from the LAC were recorded

for analysis.


Attention-Deficit/Hyperactivity Disorder


Subjects' parents were interviewed using a semi-structured interview for

symptoms of ADHD based on DSM-IV criteria. This questionnaire asked about each

symptom listed in the DSM-IV, and if the parent endorsed six or more symptoms of

either the inattention and/or hyperactivity/impulsivity cluster, follow-up questions about

age of onset, duration of symptoms, situations where symptoms are exhibited, and extent

of symptoms' disturbance on functioning were asked. Based on parents' response to this

questionnaire, subjects were categorized into ADHD or Non-ADHD groups. Because of

the high co-morbidity rate between dyslexia and ADHD, this interview allowed for the







description of ADHD rate in the current study groups. The form used during this

interview is presented in Appendix 2.


Experimental Measures


The aim of this study was to test the hypothesis that articulatory feedback

facilitates naming. According to this hypothesis, feedback from appropriate articulator

movements facilitates name retrieval. An interference paradigm was implemented to test

this hypothesis. While subjects attempted to name objects, they were asked to engage in

another task designed to interfere with articulatory movements appropriate to the naming

task.


Naming Assessed via Phoneme Match (NAPM)


During the experimental task, NAPM, subjects were required to look at two

pictures, name those pictures to themselves, and determine if those names end in the

same phoneme. They engaged in two interference conditions while performing this

naming task. During the Mouth Interference condition, subjects were asked to engage

their mouth in the following movement sequence: Lips together (as if making the /m/

sound)-tongue between teeth (as if making the /th/ sound). These movements were

demonstrated without accompanying phonemic sounds. Because subjects' articulators

were engaged in this interference movement, they could not orally name the objects seen.

Therefore naming was assessed by asking subjects to decide if the names of the two

objects seen during each trial terminated in the same sound. They indicated their






43
response by pressing designated buttons. In order to perform this task, subjects must first

name the two objects, then judge if those names have matching end phonemes.

During the Foot Interference condition, subjects were asked to move their left foot

in a rocking movement alternating between heel and toe, while naming pictures and

deciding if the names' end phoneme matched. This condition was implemented to control

for the attentional demands of engaging in an interference task. Subjects engaged in

Mouth Interference while performing the NAPM task during half of the trials and

engaged in Foot Interference during the other half of the trials. The order of the

interference condition was counterbalanced across subjects.

Each interference condition consisted of 32 trials. Half of the trials had word

pairs with matching end phonemes and the other half had word pairs with non-matching

end phonemes. The 64 word pairs were divided into two stimulus sets and are presented

in Appendix 3. The two stimulus sets were balanced on word frequency (Francis &

Kucera, 1982), number of syllables, and grade level by which the word is taught

(Thomdike & Lorge, 1972). Simple black and white line drawings eliciting each target

were drawn from the Snodgrass and Vanderwart picture set (1980), Peabody Picture

Vocabulary Test (Dunn & Dunn, 1981), and Boston Naming Test (Kaplan et al., 1983).

In some instances, a lack of available drawings necessitated the use of locally produced

drawings, which were produced to be of similar visual complexity level as pictures from

above mentioned sets. Each stimulus set was used during Mouth Interference half the

time and during Foot Interference half the time.

These stimuli were presented to subjects on a laptop computer via a program

written with PsychLab v.6.0.2. Each trial began with a fixation mark lasting 1000 msec.






44
Then an auditory cue alerted subjects to the onset of pictured stimuli, which remained on

the screen until subjects pressed one of two acceptable keys. The computer recorded

subjects' response and reaction time from the onset of the stimuli presentation to key

press. The examiner monitored subjects' interference movement and recorded the

number of movement cycles completed during each trial. One movement cycles during

the Mouth Interference was defined as lip closure followed by intrusion of the tongue

between teeth. One cycle during the Foot Interference was defined as toe touching the

floor followed by heel touching the floor. A different auditory cue followed subjects'

response and marked the end of a trial. The screen then remained blank until the

examiner pressed one of two keys marking that trial as valid or invalid. No time limit on

response time was imposed. Because the next trial did not begin until the examiner

pressed one of two keys, the examiner controlled the pace of the testing and implemented

breaks as appropriate for each subject.

The NAPM began with six practice trials, during which subjects performed the

NAPM task without any interference. Each interference condition (Mouth and Foot

Interference) began with a demonstration of the interference task, followed by four

practice trials with interference. Subjects were instructed to engage in the interference

movement before the onset of each trial. After the practice trials, subjects were informed

that testing will begin. The first two trials were used as buffer trials (i.e., data were not

recorded) without subjects' knowledge. The 32 experimental trials followed. Data

recorded during each trial included response reaction time, response (to calculate

accuracy percentage), and number of interfering movements produced (for calculating







45
average frequency of movement). Time taken to complete the NAPM ranged between 15

to 20 minutes.


Visual Match


A visual match task was implemented to control for potentially different

attentional demands of Mouth and Foot Interference. This was a nonverbal, visual match

task requiring subjects to determine if one of four pictures matched a target. Subjects

engage in Mouth and Foot Interference during this task as well. If subjects' performance

during the Mouth and Foot Interference conditions differed on this non-verbal task, that

would suggest that mouth movements and foot movements have different levels of

interfering effect.

Similar to the NAPM, Visual Match also had 32 trials for each interference

condition. Each trial composed of one target picture at the top of the screen and four

other pictures at the bottom of the screen. Subjects were instructed to press one of two

keys indicating whether there was a match between the four pictures on the bottom and

the target on top. Half of the trials had matching pictures and the other half had non-

matching pictures. Pictures were taken from the Test of Visual-Perceptual Skills (non-

motor)-Revised (Gardner, 1996), and were selected for their difficulty to be verbalized.

One set of stimuli was constructed first. The second set was constructed by changing the

target and/or the ordering of the four pictures on the bottom.

The Visual Match Test was presented to subjects on a laptop computer via a

program written with PsychLab v.6.0.2. Experimental parameters mirrored the NAPM

parameters as much as possible. Each trial began with a fixation mark lasting 1000 msec.







46
Then an auditory cue alerted subjects to the onset of pictured stimuli, which remained on

the screen until subjects pressed one of two acceptable keys. No time limit to response

was imposed. The computer recorded subjects' response and reaction time. The

examiner recorded the number of interference movement cycles completed during each

trial. A different auditory cue followed subjects' response and marked the end of a trial.

The screen then remained blank until the examiner pressed one of two keys marking that

trial as valid or invalid.

Similar to the NAPM, the Visual Match Test also began with six practice trials,

during which subjects performed the Visual Match task without any interference.

Interference conditions (Mouth and Foot Interference) then followed. Each condition

began with a demonstration of the interference task, followed by four practice trials with

interference. Subjects were instructed to engage in interference movement before the

onset of each trial. After the practice trials, subjects were informed that testing will

begin. The first two trials were used as buffer trials (i.e., data were not recorded) without

subjects' knowledge. Thirty-two experimental trials followed. The order of the

interference conditions was counterbalanced across subjects. Data recorded during each

trail included response reaction time, response accuracy, and the number of interfering

movements produced (for calculating average frequency of movement). Time taken to

complete the Visual Match Test ranged between 15 to 25 minutes.


Phoneme Match


A Phoneme Match Test was implemented to control for the potential difference in

subjects' ability to determine if the end phoneme of word pairs matched. This was a







necessary control given that subjects' naming performance during the NAPM was

measured via their ability to determine matching phonemes. Subjects completed this task

without any interference. Stimuli were two sets of word pairs used during the NAPM.

These stimuli were presented aurally to subjects on a laptop computer via a program

written with PsychLab v.6.0.2. Each trial began with the word "listen," which stayed on

the screen until the end of the trial. A pair of words was presented by the computer 1500

msec after the onset of the word "listen." Subjects pressed one of two keys to indicate if

the two words' last phonemes matched. The computer recorded subjects' response and

reaction time from the onset of stimulus presentation. The screen then remained blank

until the examiner pressed one of two keys marking that trial as valid or invalid. No time

limit on response time was imposed.

The Phoneme Match Test included four practice trials followed by the two

stimulus sets, which totaled 64 trials. The order of the two stimulus sets was

counterbalanced across subjects. Data recorded during each trial included response

reaction time and response accuracy. Time taken to complete the Phoneme Match Test

ranged between 5 to 10 minutes.


Naming Test


A visual confrontation naming test was administered after the completion of the

NAPM, Visual Match, and the Phoneme Match Test. This Naming Test composed of

black and white line drawings from the NAPM. Each subject was asked to say the name

that they assigned the pictured item when they saw it during the NAPM task. For cases

where subjects stated an acceptable alternative response for an item (e.g., "bunny" for







48
"rabbit"), the name that the subject gave was used to determine if the two items from that

NAPM trial had names with matching end phonemes, and the subject's response

accuracy for that trial was determined accordingly. For cases where subjects were not

able to produce a response because of unfamiliarity with the object, the NAPM trial

including that object was deleted. Thus response to objects for which subjects were

unfamiliar was not included in data analysis.


Procedures


The examiner first interviewed the parent over the telephone to obtain each

subject's background information and to screen for ADHD. Based on this information,

subjects were assigned a subject number using the chart represented in Table 4. This

chart counterbalanced the order of task presentation (i.e., NAPM or Visual Match), the

order of stimulus sets used, and the order of interference conditions. Each subject's order

of test presentation, stimulus set used, and order of interference condition was determined

based on his/her assigned subject number.

Subjects completed the testing in either one or two settings, totaling 1.5 hour for

older normal readers to 3 hours for younger subjects with reading problems. Factors

influencing whether testing was completed in one or two settings included each subject's

time availability and their performance during the first hour. For those subjects who

experienced difficulty, testing was completed in two sessions to minimize their

frustration. Subjects were encouraged and praised for their effort rather than for their

accuracy. In no instance were subjects given feedback about the accuracy of their







49
Table 4. Sample of the chart for determining the order of task, interference, and stimulus
set for each subject.
Controls Phonologically Impaired

Order of NAPM Visual Match NAPM Visual Match

stimulus set first first first first

AB 1001 1002 2001 2002 Mouth

BA 1003 1004 2003 2004 Interference

AB 1005 1006 2005 2006 first

BA 1007 1008 2007 2008

AB 1009 1010 2009 2010
BA 1011 1012 2011 2012 Foot

AB 1013 1014 2013 2014 Interference

BA 1015 1016 2015 2016 first

AB 1017 1018 2017 2018

BA 1019 1020 2019 2020



response. The order in which testing components were completed is represented in Table

5.

The Test of Phonological Awareness (Torgesen & Bryant, 1994) was used to train

subjects to match end phonemes of words, thus familiarizing them to the task demand of

the NAPM. Subjects' performance on this task was not scored. The examiner remained

with subjects throughout testing. During the NAPM and Visual Match tasks, the

examiner monitored subjects' interference movements (mouth or foot) and reminded







Table 5. Order of test administration.
Telephone interview:

Demographic Questionnaire

ADHD Questionnaire

Testing session:

Test of Phonological Awareness

NAPM and Visual Match

Mouth and Foot Interference

Phoneme Match Test

Naming Test

Boston Naming Test

Articulatory Awareness Test

Lindamood Auditory Conceptualization Test

Woodcock Reading Mastery Test:

Word Identification

Word Attack

Passage Comprehension

Rapid Color Naming and Rapid Object Naming



subjects to engage in these movements during instances when they stopped. These parts

of the session were videotaped.













RESULTS


The statistical program, SPSS v7.5.2 for Windows, was used to analyze the

following data. Both means and medians were examined as central tendency statistics for

reaction time data (i.e., NAPM, Visual Match, Phoneme Match). Because reaction times

are subject to floor effects and have unlimited ceilings, distribution of scores may be

skewed and means may be overly influenced by extremely slow reaction times. Thus

extreme scores were trimmed in the following way in calculating means for each subject.

For each subject's performance in each condition (i.e., NAPM Mouth Interference,

NAPM Foot Interference, Visual Match Mouth Interference, Visual Match Foot

Interference, Phoneme Match Set A, Phoneme Match Set B), mean reaction time and

standard deviation were calculated. Extreme scores that lay outside of two standard

deviations from the mean were excluded, yielding a trimmed mean for each condition for

each subject. There was no difference in the pattern of results using trimmed means and

medians. Thus results reported here were based on trimmed means.


Articulatorv Knowledge


The Articulatory Feedback Hypothesis of Naming stated that articulatory

feedback facilitates naming. This hypothesis implied that better articulatory knowledge

should be correlated with faster reaction time on a naming test. This was tested using a

regression analysis with performance on the Articulatory Awareness Test as the






52
independent variable and reaction time on experimental measures as dependent variable.

The experimental measures considered here included the NAPM and Visual Match,

Mouth and Foot Interference conditions. The Visual Match was included as a non-verbal

control task. The hypothesis predicted a significant correlation between Articulatory

Awareness Test scores and NAPM, a name retrieval task, but not with Visual Match, a

nonverbal control task. The two interference conditions of the NAPM were predicted to

have different correlations with articulatory knowledge because subjects were able to use

appropriate articulatory feedback in one condition (i.e., Foot Interference) but not in the

other (i.e., Mouth Interference). In the condition where subjects were able to use

appropriate articulatory feedback (i.e., Foot Interference), better articulatory knowledge

was expected to be associated with faster naming time. In the condition where subjects

were not able to use articulatory feedback (i.e., Mouth Interference), reaction time was

expected to be slow; thus a non-significant correlation between articulatory knowledge

and naming time may be seen.

The Pearson correlations between reaction time on experimental measures and

scores on the Articulatory Awareness Test (10-item version) and the Articulatory

Awareness Test-Revised (AAT-R, 20-item version) were reported in Table 6. Reaction

time during the two interference conditions of the NAPM were either significantly

correlated with or approaching significance with the AAT (Mouth F = 4.52,p = .04; Foot

F = 3.48,p .07) and the AAT-R scores (Mouth F = 5.64,p = .02; Foot F = 3.74,p =

.06), whereas there was no relationship between reaction time on the Visual Match Test

and articulatory knowledge (AAT: Mouth F = 0.21,p = .65; Foot F = 0.92,p = .34; AAT-

R: Mouth F = 0.23,p = .64; Foot F = 0.37,p = .55). This pattern indicated that






53
increasing articulatory knowledge was associated with faster reaction time during name

retrieval but not during a nonverbal visual match task. However, the prediction that

articulatory knowledge would be more highly correlated with naming latency during the

Foot Interference than during the Mouth Interference condition was not supported.


Table 6. Pearson correlations between the Articulatory Awareness Test (AAT) score and
reaction time on experimental measures.
AAT AAT-R

NAPM

Mouth Interference -.32* -.36

Foot Interference -.29 -.30+

Visual Match

Mouth Interference .07 .08

Foot Interference .15 .10

Note: p < .05, +p <.10.


Phonologicallv Impaired vs. Controls


Montgomery (1981) found a difference in articulatory knowledge between

dyslexic children and normal readers. To examine if that finding can be replicated

among the present subjects, the difference in articulatory knowledge between normal

readers and phonologically impaired readers was examined. The Phonologically

Impaired (PI) group was composed of the DPD and ARPP subjects.







Articulatory Awareness Test


Scores obtained by PI and CTRL groups were reported in Table 7. A t-test of

independent samples was utilized to test for significant differences between the two

groups. Unlike Montgomery's (1981) finding, there was no difference between PI and

CTRL groups on articulatory knowledge as assessed by the 10-item version of the AAT (t

= 0.80,p = .43) or by the 20-item version (AAT-R, t = 0.64,p =.53). -


Table 7. AAT and AAT-R scores obtained by Phonologically Impaired (PI) and Control
(CTRL) groups.
PI CTRL

AAT 6.19(2.27) 6.70(1.75)

AAT-R 12.14 (3.55) 12.85 (3.53)



Although no difference was found between groups on articulatory knowledge

measures, it was possible that subjects with phonological impairment may demonstrate a

different pattern of relationship between articulatory knowledge and name retrieval

compared to controls. Thus Pearson correlations between the AAT score and reaction

time during each experimental condition (i.e., NAPM and Visual Match Mouth and Foot

Interference conditions) were examined for each group. Because the AAT and AAT-R

scores yielded similar pattern of results, only AAT scores were reported.

Table 8 showed Pearson correlations between the AAT score and reaction time

during experimental conditions for each group. The correlation between the articulatory

knowledge score and naming latency (i.e., NAPM) seen in the analysis with all subjects

combined was driven by the control subjects (Mouth, F = 5.31,p =.03; Foot, F = 2.53,p






55
= .13). The speed by which PI subjects retrieved names was not related to their level of

articulatory knowledge (Mouth, F = 0.65,p = .43; Foot, F = 0.84,p = .37). Neither

groups' AAT score was correlated with their performance on the visual match control.

The control group's correlation between articulatory knowledge score and naming latency

during Mouth Interference differed from their correlation between articulatory knowledge

and visual match latency during Mouth Interference (-0.48 vs. 0.19; Z(2o,20) = -2.07, p <

.05; subscripts denote the sample size of groups being compared). No other pairs of

correlation differed from each other.


Table 8. Pearson correlations between the AAT score and reaction time on experimental
measures for PI and CTRL groups._
PI CTRL

NAPM

Mouth Interference -.18 -.48*

Foot Interference -.21 -.35

Visual Match

Mouth Interference -.03 .19

Foot Interference .06 .27

Note: PI = Phonologically Impaired; CTRL = Controls; p < .05.


Descriptive Measures


Subjects' performance on descriptive measures were reported in Table 9. Scores

were compared by t-test of independent samples. On measures of naming, the PI group

performed more poorly on the BNT (t = 2.05, p = .05), but not on Rapid Color Naming (t

= 1.54,p = .13), Rapid Object Naming (t = 1.47, p =. 15), or on the experimental Naming
Is>







Test (t = 1.58,p = 12). The PI subjects scored significantly lower than the CTRL

subjects on the LAC, our measure of phonological awareness (t = 3.12,p = .00). The

TONI-2 and reading achievement scores were also reported to contrast between PI and

CTRL groups. The PI did not differ from the CTRL group on intellectual aptitude as

measured by the TONI-2 (t = -0.14,p = .89), but their reading achievement scores were

all worse than their age- and intelligence-matched controls (Word Attack, t = 9.28, p =

.00; Word Identification, t = 7.04,p = .00; Passage Comprehension, t = 4.87,p =.00).


Table 9. PI and CTRL groups' performance on descriptive measures.
PI CTRL
BNT -1.96 (2.07)' -0.85 (1.30)

Rapid Color Naming' -0.58 (1.57) 0.02 (0.68)

Rapid Object Naminga -0.88 (2.36) -0.05 (0.80)

Naming Test 87(6) 90(6)

LACc 55(17)e 75(23)'

TONI-2d 106(11) 106(8)

WRMT Word Attackd 75 (11)e 104 (8)

WRMT Word Identificationd 78 (14): 105 (10)'

WRMT Passage Comprehensiond 83 (16)' 103 (10)f

a b
Note: PI = Phonologically Impaired; CTRL = Controls; Z-scores. b Percentage correct.
' Raw score. d Age-corrected standard scores. Within each row, numbers with different
superscripts were significantly different from each other.







Experimental Measures


Phoneme match. Before group differences on the NAPM were further examined,

subjects' ability to match phonemes was evaluated first. Name retrieval on the NAPM

was assessed via subjects' ability to match the end phoneme of words. Thus it was

important to know if groups differed from each other on this ability. A multivariate

analysis of variance (MANOVA) was performed on the reaction time data from the

Phoneme Match Test with Group (PI vs. CTRL) as a between-subject variable and

Stimulus Set (A and B) as a within-subject variable. This analysis also allowed for

examination of differences between stimulus sets. Mean reaction time in milliseconds

and standard deviations were presented in Table 10. The Group X Stimulus Set

interaction was significant (F = 5.86, p = 0.02). Within each group, reaction times for the

two stimulus sets were compared using dependent samples t-test. The CTRL group's

reaction time on stimulus sets did not differ (t = 0.59, p = .56) while PI subjects were

faster in responding to Set A than to Set B (t = -2.51, p = .02). The group reaction time

for each stimulus set was compared using independent samples t-test. The two groups'

reactions times did not differ for Set A (t = -0.87,p = .39). The CTRL group was faster

than the PI group on Set B (t = -2.08, p = .04). A similar analysis was conducted on

response accuracy. This revealed a marginally significant Group effect (F = 3.57,p =

.07). The CTRL group tended to be more accurate than the PI group.

The two stimulus sets did not differ for the CTRL group, but for the PI group, Set

B was responded to more slowly and therefore it may have been harder. The most

important finding here was that the two groups did not differ in their ability to perform

the Phoneme Match Test, as measured by both reaction time (F = 2.85,p =. 10) and
1%







Table 10. Means and standard deviations of reaction time (RT in milliseconds) and
accuracy (% Correct) on the Phoneme Match Test.
PI CTRL
R_.T % Correct RT % Correct

Stimulus Set A 2899 (579)' 90(13) 2758(445) 95(5)

Stimulus Set B 3224 (966)b 89(11) 2711(543)a 94(7)

Note: PI = Phonologically Impaired; CTRL = Controls. Numbers with different
superscripts were significantly different from each other.


accuracy (F = 3.57,p = .07). Because there was no statistically significant group effect,

and because the counterbalance measures taken to pair Set A with Mouth Interference

approximately half of the time and with Foot Interference the other half of the time, the

difference in difficulty level between Set A and Set B was considered to be equally

dispersed among interference conditions. No further attempt to examine Stimulus Sets'

interaction with other variables in subsequent analyses was taken because the number of

subjects in this study limited the power available to detect such high level interactions.

NAPM and visual match. Subjects' performance on the NAPM and Visual Match

were reported in Table 11. Separate MANOVAs were conducted for reaction time and

accuracy, with Group as a between-subject factor (PI vs. CTRL), and Task (NAPM and

Visual Match) and Interference (Mouth and Foot) as within-subject factors. The

MANOVA for reaction time revealed a Group X Task interaction (F = 6.67, p= .01) and

a Task main effect (F = 27.23, p = .00). Subjects responded to NAPM faster than to

Visual Match. Because reaction time to each task was not important theoretically,

separate MANOVAs were conducted to further examine group differences and potential

interference effects within each task. The MANOVA for NAPM revealed a marginal







Group effect (F = 3.91, p = .06), with the CTRL group responding faster than the PI

group. The MANOVA for Visual Match revealed no significant interactions or effects.


Table 11. Reaction time and accuracy on the NAPM and Visual Match Tests for PI and
CTRL erouos.


PI CTRL

RT % Correct RT % Correct

NAPM

Mouth 4353(1547) 81(15) 3688(1551) 89(10)

Foot 4591(1411) 83(8) 3653(1135) 89(10)

Visual Match

Mouth 4946(1788) 77(13)a 5333(1694) 86(13)

Foot 5004(1333) 84 (9)b 4984(1719) 84(13)

Note: PI = Phonologically Impaired; CTRL = Controls. Numbers w
superscripts were statistically different from each other.


ith different


The MANOVA on accuracy data revealed a Group X Interference interaction (F =

4.11, p = .05) and a Group effect (F = 4.09, p = .05). Dependent samples t-test to

compare the accuracy difference between interference conditions indicated that with the

NAPM and Visual Match tasks combined, the CTRL group's accuracy on Mouth and

Foot Interference conditions was the same (t = -0.72,p = .48), whereas the PI group

achieved marginally less accuracy on Mouth Interference compared to Foot Interference

(t = 1.94,p = .07). Independent samples t-test to compare accuracy between groups

showed that the PI group's accuracy on Mouth Interference was statistically less than the

CTRL group's accuracy during this condition (t = 2.38, p = .02). Because there were a

priori reasons to examine the difference between interference conditions separately for






60
each task, and to examine if groups differed in this difference, dependent samples t-tests

were conducted separately for the NAPM and the Visual Match Test. The only

difference was found in the PI group's accuracy performance on the Visual Match Test.

They were less accurate during the Mouth Interference condition than during the Foot

Interference condition (t = 2.30, p = .03). Because their reaction time was not different

between these conditions (t = 0.28,p = .78), a speed-accuracy trade off was not a likely

explanation for their worse accuracy during the Mouth Interference condition.

Block effect. The present study included 12 individuals with ADHD. Children

with ADHD may have decreased sustained attention span and/or slowed reaction time.

Thus trials were divided into two blocks to examine if performance during the first half

of each condition differed from performance during the second half. A MANOVA on

reaction time data with Group (PI vs. CTRL) as a between-subject factor and Task

(NAPM and Visual Match), Interference (Mouth and Foot), and Block (1 and 2) as

within-subject factors revealed a significant Task X Block interaction (F = 16. 15,p = .00)

and Block effect (F = 5.06,p = .03), in addition to the Task X Group interaction and

Task effect already reported above. Follow up MANOVAs were conducted for each

Task. No significant effects were found for NAPM, but for Visual Match, a Block X

Group interaction (F = 4.98, p = .03) was found as well as a Block effect (F = 20.11 ,p =

.00). Table 12 showed the reaction time on the NAPM and Visual Match tasks broken

down by Block. Dependent samples t-test indicated that the CTRL group's reaction time

during Block 1 of Visual Match was much faster than their reaction time during Block 2

(t = -4.49, p = .00), but such a difference was not found for the PI group (t =-l.69,p =

.11).







Table 12. Means and standard deviations of reaction time for each block.
PI CTRL

NAPM

Block 1 4582(1476) 3784(1462)

Block 2 4374(1277) 3581(1206)

Visual Match

Block 1 4828(1451) 4768(1387)a

Block2 5095(1606) 5564(1908)b

Note: PI = Phonologically Impaired; CTRL = Controls. Numbers with different
superscripts were significantly different from each other.


Similar analyses were conducted for response accuracy. The MANOVA revealed

a significant Task X Interference X Block interaction (F = 5.16, p = .03), Task X Block

interaction (F = 7.42, p = .01), and Block main effect (F = 4.0,p = .05), as well as an

Interference X Group interaction and Group main effect. The latter two were discussed

already in the section on accuracy and so were not discussed here. Of the three findings

involving Block, only the highest level interaction was examined because lower level

interactions were represented in the higher level interaction. Table 13 showed the

response accuracy pattern reflected by the Task X Interference X Block interaction.

Dependent samples t-tests were conducted to compare all possible pairs of scores.

Subjects as one group became less accurate during the second block of Visual Match

Mouth Interference (Block 1 vs. Block 2, t = 3.27, p = .00; Block 2, Mouth vs. Foot

Interference, t = 3.15, p = .00). No other pair of scores was statistically different from

each other.








Table 13. Response accuracy (percentage) reflecting the Task X Interference X Block
interaction.
Mouth Foot

NAPM

Block 1 83 86

Block 2 86 85

Visual Match

Block 1 85a 85

Block 2 77 83a

Note: Numbers with different superscripts were significantly different from each other.


Separate MANOVAs were conducted for Block 1 and Block 2 to further explore

how time influenced data. Table 14 summarized and compared the overall findings to

Block 1 and Block 2 findings. The reaction time data were fairly consistent across the

overall analysis and the two time blocks. The overall Group X Task interaction reflected

faster response to the NAPM task by the CTRL group (see Table 11), but there was no

difference between tasks in the PI group's reaction time. The increase in the Group X

Task interaction from non-significance in Block 1 to significance in Block 2 could be

understood by comparing Table 15 with Table 16. The CTRL group became faster on the

NAPM with practice, but they slowed down significantly on the Visual Match with time.

While such a pattern was also evident with the PI group, their reaction time difference

between Block 1 and Block 2 was not as dramatic.

Table 14 also shows differences in response accuracy findings between blocks.

While the Group X Interference interaction was not significant in Block 2, the pattern of







Table 14. Comparison of overall findings with Block 1 and Block 2 findings.
Overall Block 1 Block 2
F p F p F

Reaction Time

Group X Task 6.67 .01 2.89 .10 8.86 .00

Task 27.23 .01 8.05 .01 40.70 .00

Accuracy

Group X Interference 4.11 .05 4.49 .04 2.93 .10

Group 4.09 .05 4.75 .04 2.89 .10

Task X Interference .24 .63 .92 .34 5.71 .02

Task 1.90 .18 .06 .81 5.31 .03



results in Block 2 was consistent with the pattern in Block 1 such that the overall Group

X Interference interaction was significant. This interaction showed that the PI group was

less accurate during Mouth Interference than Foot Interference, while the CTRL group's

accuracy during both interference conditions was commensurate (see Table 11). The

Task X Interference interaction significant in Block 2 was not significant in the overall

analysis, suggesting a great degree of variability during Block 1. The Block 2 Task X

Interference interaction reflected worse accuracy during the Visual Match Mouth

Interference condition in comparison to the Visual Match Foot Interference condition and

the two NAPM interference conditions. As can be seen in Table 16, this effect was

mainly driven by the PI group's performance.

The Block effects found were not anticipated a priori. Fatigue cannot completely

explain differences between blocks because while reaction time slowed down for Visual






64
Table 15. Block I reaction time and accuracy on the NAPM and Visual Match Test for
PI and CTRL groups.
PI CTRL

RT % Correct RT % Correct

NAPM

Mouth 4453(1757) 78(16) 3874(1944) 88(10)

Foot 4710(1553) 84(9) 3695(1121) 88(12)

Visual Match

Mouth 4797(1651) 81(16) 4910(1454) 89(9)

Foot 4859(1408) 85(11) 4625(1474) 85(14)

Note: PI = Phonologically Impaired; CTRL = Controls.


Table 16. Block 2 reaction time and accuracy on the NAPM and Visual Match Test for
PI and CTRL groups.
PI CTRL

RT % Correct RT % Correct

NAPM

Mouth 4264(1461) 83(17) 3543(1362) 89(12)

Foot 4484(1366) 82(12) 3619(1292) 88(12)

Visual Match

Mouth 5035(1931) 72(16) 5768(2024) 82(19)

Foot 5156(1407) 83(10) 5361(2027) 83(13)

Note: PI = Phonologically Impaired; CTRL = Controls.


Match, it speeded up for NAPM (Table 12). That accuracy was worse just during Visual

Match Mouth Interference but not during Foot Interference also argued against an overall







65
fatigue effect (Table 13). Because the role of time was unclear, Block effects were also

examined in subsequent analyses.

Excluding ADHD subjects. To examine how ADHD subjects' reaction time

differed from subjects without ADHD, their reaction time and accuracy data were

contrasted from non-ADHD subjects in Table 17. Overall, subjects with ADHD tended

to be slower and less accurate than their non-ADHD counterparts. Thus subjects with

ADHD were excluded in a MANOVA to examine if ADHD subjects contributed to the

findings involving Block. Similar to above, Group (PI vs. CTRL) was the between-

subject factor, and Task (NAPM and Visual Match), Interference (Mouth and Foot), and

Block (1 and 2) were within-subject factors. This analysis was similar to results reported

with the ADHD subjects included. Significant interactions from reaction time data

included Task X Block (F = 11.66,p = .00) and Group X Block (F = 5.02,p = .03).

Significant main effects included Task (F = 20.47, p = .00) and Block (F = 7.32, p = .01).

Dependent samples t-test to follow up the Task X Block interaction indicated that non-

ADHD subjects were slower during Block 2 of Visual Match (t = -4.36,p = .00), but

their reaction times did not differ between NAPM Block 1 and Block 2 (t = 0.8 1,p -=

.42). Dependent samples t-test to follow up the Group X Block interaction indicated that

the non-ADHD CTRL subjects were slower during Block 2 than Block 1 (t = -4.10,p =

.00), but the non-ADHD PI subjects did not show a difference in reaction time between

blocks (t = -0.28,p -= .78). Excluding ADHD subjects did not change theoretically

important pattern of findings (i.e., Group X Interference interaction). Repeating this

MANOVA with accuracy data revealed no significant effects or interactions. Thus

ADHD subjects were included in all subsequent analyses.







66
Table 17. Reaction time and accuracy of the Non-ADHD and ADHD subgroups on the
NAPM and Visual Match Tests.
PI CTRL


Non-

ADHD

(n= 12)

RT


ADHD

(n=9)


Correct
Correc__._t


Non-

ADHD

(n= 17)

RT


Correct


ADHD

(n=3)


Correct
Correct


Correct
Correct_


NAPM


Mouth

Foot


4109

(1181)

4481

(1692)


84

(12)

86

(7)


4679

(1964)

4739

(999)


76

(18)

79

(8)


3432

(1107)

3587

(1109)


5141

(3055)

4028

(1467)


75

(15)

77

(17)


Visual

Match


5002 82 4871 71 5253

Mouth (1532) (7) (2179) (17) (1733)

Foot 4926 82 5108 85 5020

(1225) (7) (1536) (12) (1804)

Note: PI = Phonologically Impaired; CTRL = Controls.


87

(14)

85

(13)


5786

(1688)

4780

(1404)


81

(8)

78

(12)


Interference movements. The number of interference movements (mouth or foot)

subjects completed during each trial was recorded. The average number of movements

completed per trial within each condition (i.e., NAPM Mouth and Foot Interference,

Visual Match Mouth and Foot Interference) was divided by the average reaction time per







67
trial for that condition to yield a movement frequency index (i.e., number of movements

per second). Post hoc analyses were conducted with these data to determine if subjects

engaged in mouth and foot interfering movements with equal facility. A MANOVA with

Group (PI vs. CTRL) as the between-subject factor and Task (NAPM and Visual Match),

Interference (Mouth and Foot), and Block (1 and 2) as within-subject factors was

conducted on the interfering movement frequency data. This revealed a significant

Group X Task X Interference X Block interaction (F = 4.35, p = .04), as well as

Interference (F = 4.51, p =.04) and Block (F =4.24, p =.05) main effects. Descriptive

statistics were reported in Table 18. Perusal of these descriptive statistics revealed the

following: There was no difference between PI and CTRL groups on Visual Match

Mouth or Foot Interference, even when interference conditions were broken down by

Block. There was also no difference between PI and CTRL groups on NAPM Foot

Interference, even when broken down by Block. However, on NAPM Mouth

Interference, the PI group produced interfering mouth movements more slowly than the

CTRL group. This difference was more salient during Block 1 than Block 2.

To examine how interference movement frequency related to subjects'

articulatory knowledge, Pearson correlations between the AAT score and interfering

movement frequency indices were calculated and reported in Table 19. Significant

correlations were found between the PI group's AAT score and their facility in producing

interfering mouth movements during NAPM. The better the PI group's articulatory

knowledge, the faster they were able to produce interfering mouth movements while

engaging in a name retrieval task. Unexpectedly, the PI group's AAT score was also

correlated with their facility in producing interfering foot movements during the Visual






68
Table 18. Interfering movement frequency index (i.e., number of movements per second)
for the PI and CTRL roups. --
PI CTRL

NAPM

Mouth .85 (.24) .97 (.32)

Block 1 .84 (.25) .98 (.35)

Block 2 .86(.26) .95(.31)

Foot 1.00 (.25) 1.01 (.26)

Block 1 .98 (.25) .97 (.26)

Block 2 1.03 (.26) 1.05 (.27)

Visual Match

Mouth 1.00 (.23) .99 (.23)

Block 1 .99 (.24) .98 (.25)

Block 2 .99 (.22) 1.01 (.24)

Foot 1.04 (.25) 1.04 (.30)

Block 1 1.01 (.26) 1.05 (.32)

Block 2 1.06 (.26) 1.03 (.30)

Note: PI = Phonologically Impaired; CTRL = Controls.


Match Test. The better their articulatory knowledge, the more slowly PI subjects

produced interfering foot movements during a nonverbal visual task.


Predictors of Articulatorv Knowledge

As a post hoc exploration of variables that predict performance on the AAT, age,

BNT Z-score, Rapid Color Naming Z-score, Rapid Object Naming Z-score, LAC raw






69
Table 19. Pearson correlations between the AAT score and interfering movement index
for PI and CTRL groups.--
PI CTRL

NAPM

Mouth Interference .49* -.14

Foot Interference -.03 -.24

Visual Match

Mouth Interference .02 -.14

Foot Interference -.44 -.13

Note: PI = Phonologically Impaired; CTRL = Controls; *p <.05.


score, TONI-2 Quotient, and the three reading achievement standard scores (Word

Attack, Word Identification, and Passage Comprehension) were entered into a stepwise

regression analysis. No variables were selected as statistically significant predictor of

AAT performance. This stepwise regression analysis was repeated for each group

separately to determine if predictors of AAT performance differed by group. Again, no

variables were selected as significant predictors of AAT score. Pearson correlation

between the AAT score and each of the variables entered were presented in Table 20.

The only variable from Table 20 that correlated with the AAT score was the LAC raw

score, and this correlation was significant only for the CTRL group.

Pearson correlations between the AAT score and the Phoneme Match response

time and accuracy were also calculated to examine the relationship between articulatory

knowledge and another task requiring phonological skills. The AAT score was positively

correlated with Phoneme Match accuracy (Pearson's r = .64, p = .00). Subjects with

better articulatory knowledge were more accurate on the Phoneme Match Test. Dividing








Table 20. Pearson correlations between the AAT score and variables entered into
stepwise regression analysis for PI and CTRL groups.
Groups Combined PI CTRL
.. ^ ..~~~.. .. - ,...-..- - ..-........-. ---^ -. - - -
Age .01 -.19 .23

BNT .03 -.16 .35

Rapid Color Naming .09 -.02 .31

Rapid Object Naming .00 -.11 .28

LAC .30* .18 .40*

TONI-2 -.24 -.25 -.23

WRMT Word Attack .18 .23 -.00

WRMT Word Identification .15 .16 -.04

WRMT Passage Comprehension .11 .15 -.18

Note: PI = Phonologically Impaired; CTRL = Controls; p <= .05.


subjects into groups showed that this correlation was significant only for the

phonologically impaired group (Pearson's r = .74,p = .00).


Predictors of Phonological Awareness


Predictors of the LAC score were also explored with stepwise regression analysis.

Age, BNT Z-score, Rapid Color Naming Z-score, Rapid Object Naming Z-score, TONI-2

quotient, the three reading achievement standard scores (Word Attack, Word

Identification, and Passage Comprehension), and AAT raw score were entered. Table 21

reported Pearson correlations between the LAC and each of these variables for the groups

combined and for each group individually. Both groups' reading achievement scores

were correlated with their LAC scores. However, because the inter-correlations between







the three achievement tests were high, only the most significant reading achievement

score was selected to predict LAC performance in the regression analysis. For the CTRL

group, Word Attack and age were selected as variables of predictive value. Word Attack

alone contributed 27% (adjusted R2) of the variance to the LAC score (F = 7.51 ,p = .01).

Word Attack and Age combined contributed 44% (adjusted R2) of the variance (F = 7.93,

p = .00). For the PI group, Passage Comprehension and Rapid Object Naming were

selected as variables of predictive value by the stepwise regression analysis.

Surprisingly, the unique variance contributed by the BNT score was not significant once

Passage Comprehension was entered. Instead, the variance contributed by Rapid Object

Naming was deemed significant. The reverse correlation between Rapid Object Naming


Table 21. Pearson correlations between the LAC score and variables entered into
stepwise regression analysis for PI and CTRL groups.
Groups Combined PI CTRL
Age .39 .29 .48'

BNT .34* .43* .07

Rapid Color Naming .18 .07 .16

Rapid Object Naming .04 -.11 .01

TONI-2 .01 .27 -.24

WRMT Word Attack .61* .50* .55*

WRMT Word Identification .59* .49* .44*

WRMT Passage Comprehension .58' .55* .39*

AAT .30* .18 .40*

Note: PI = Phonologically Impaired; CTRL = Controls; 'p <= .05.







72
and the LAC indicated that the better PI subjects performed on the LAC, the slower they

completed the Rapid Object Naming. Passage Comprehension alone contributed 27%

(adjusted R2) of variance to the LAC score (F = 8.31,p = .01). Passage Comprehension

and Rapid Object Naming together contributed 42% (adjusted R2) of variance (F = 8.33,

p =.00).

As a check of LAC's validity as a measure of phonological awareness, the

correlation between the LAC raw score and Phoneme Match Test performance was

calculated. With all subjects combined, the correlation between the LAC score and the

Phoneme Match performance was statistically significant (RT, Pearson's r = -.37,p = .02;

accuracy, Pearson's r = .48,p = .00). Dividing subjects into groups revealed that the LAC

score and Phoneme Match performance was correlated for CTRLs (RT, Pearson's r = -

.48, p = .03; accuracy, Pearson's r = .69,p = .00) but not for the PI group (RT, Pearson's r

=. 18,p = .44; accuracy, Pearson's r = .29, p = .20).


Developmental Phonological Dyslexics vs.

Adequate Readers with Poor Phonology vs. Controls


The PI group can be categorized into two distinct subgroups. As shown in Table

1, the DPD group was characterized by impaired phonological processing, single-word

reading, and comprehension (WRMT Word Attack, Word Identification, and Passage

Comprehension respectively) in comparison to their expected achievement level based on

their intellectual aptitude. The ARPP group, while demonstrating impaired phonological

skills, actually has single-word reading and comprehension skills commensurate to their

expected achievement level. This subtype of children with impaired phonological







73
processing but adequate reading ability has been described (Masutto & Cornoldi, 1992),

but little is known about them, such as whether these children represent a distinct subtype

of dyslexia or a milder form of the disorder. To explore if the DPD and ARPP groups

differed in their presentation on cognitive measures, differences between these groups

were examined in this section. The CTRL group was also included in order to compare

these two phonologically impaired groups with normal readers.


Articulatory Awareness Test


Articulatory Awareness Test scores obtained by DPD, ARPP, and CTRL groups

were provided in Table 22. An ANOVA conducted with Group as the between-subject

variable revealed that the three groups did not differ from each other on their AAT score

(F = 0.41,p = 0.66).


Table 22. Means and standard deviations of AAT scores obtained by DPD, ARPP, and
CTRL groups.
DPD ARPP CTRL

AAT 6.o (2.72) 6.4 (1.78) 6.7 (1.75)

Note: DPD = Developmental Phonological Dyslexia; ARPP = Adequate Reader with
Poor Phonology; CTRL = Controls.


To examine if DPD and ARPP groups demonstrate different patterns of

relationship between articulatory knowledge and name retrieval, Pearson correlations

between AAT and reaction time during experimental conditions (i.e., NAPM and Visual

Match interference conditions) were conducted for each group and reported in Table 23.

Although the DPD group's correlations between their AAT score and naming latency

appeared more similar to the CTRL group's and different from the ARPP group's, there







was no statistically significant difference between DPD and ARPP group's correlations

on the NAPM interference conditions (Mouth Interference, Z(I 1,1o) = -.83, p > .05; Foot

Interference, Z(i 1,10) = -.70,p > .05) or on the Visual Match interference conditions

(Mouth Interference, Z1 1,1o) = -1.51, p> .05; Foot Interference, Zij 1,1o) = -.52, p> .05).

Unlike the control group, whose correlation between naming latency and AAT score

differed significantly from its corresponding correlation on the Visual Match Test during

Mouth Interference (i.e., -.48 vs. .19, Z(20,0o) = -2.07, p < .05), no such corresponding

correlation pairs within the DPD and ARPP groups were statistically different (DPD:

NAPM Mouth vs. Visual Match Mouth, Z(H1,11) = -.06,p> .05; NAPM Foot vs. Visual

Match Foot, Z(l ,1) = -.60,p > .05; ARPP: NAPM Mouth vs. Visual Match Mouth,

Z(o10,io) = -.71,p > .05; NAPM Foot vs. Visual Match Foot, Z(,, .,,,i = -.39,p > .05).


Table 23. Pearson correlations between the AAT score and reaction time on
experimental measures for DPD, ARPP, and CTRL groups.
DPD ARPP CTRL
NAPM

Mouth Interference -.32 .10 -.48*

Foot Interference -.29 .06 -.35

Visual Match

Mouth Interference -.29 .45 .19

Foot Interference -.00 .26 .27

Note: DPD = Developmental Phonological Dyslexia; ARPP = Adequate Reader with
Poor Phonology; CTRL = Controls; p < .05.







Descriptive Measures


Table 24 showed each group's performance on descriptive measures. One-way

analyses of variance indicated a group difference on the BNT (F = 4.66, p= .02) but not

on any of the other naming measures (Rapid Color Naming, F = 2.89, p= .07; Rapid

Object Naming, F = l.54,p = .23; Naming Test, F = 2.30,p = .11). On the BNT,

independent samples t-test showed that the DPD group scored lower than the CTRL

group (t = 2.74,p = .01). The ARPP group's BNT score did not differ from either of the

other groups (ARPP vs. CTRL, t= 0.60, p= .55; ARPP vs. DPD, t = 1.85,p = .08). One-

way analysis of variance on the LAC scores also indicated difference between groups (F

= 7.71 ,p = .00). Independent samples t-test showed that the DPD group's phonological

awareness score was lower than both other groups (DPD vs. CTRL, t = 3.76, p = .00;

DPD vs. ARPP, t = 2.84,p = .01), while the ARPP group's LAC score did not differ from

that of the CTRL group's (t = l.23,p = .23).


Table 24. Means and standard deviations on descriptive measures for the DPD, ARPP,
and CTRL groups.
DPD ARPP CTRL
BNT 2.71(2.52)- 1.13 (1.02) -0.85 (1.30)e

Rapid Color Naminga -1.02 (1.85) -0.08 (1.07) 0.02(0.68)

Rapid Object Naminga -1.25 (2.80) -0.49 (1.81) -0.05 (0.80)

Naming Testb 85(7) 89(5) 90(6)

LACc 46(13)f 65 (16)g 75(23f

Note: DPD = Developmental Phonological Dyslexia; ARPP = Adequate Reader with
Poor Phonology; CTRL = Controls. Z-scores. b Percentage correct. C Raw score.
Numbers in each row with different superscripts were significantly different from each
other.









Experimental Measures


Phoneme match. A MANOVA was conducted to see if groups differed in their

reaction time on matching the end phoneme of words. Group (DPD, ARPP, and CTRL)

was entered as the between-subject factor and Stimulus Set (A and B) was entered as the

within-subject factor. The Group X Stimulus Set interaction was only marginally

significant (F = 2.85, p = .07), but there was a significant Stimulus Set effect (F = 6.09, p

= .02). The Stimulus Set effect was the same as that reported under the PI vs. CTRL

section, where it was shown that subjects were faster in responding to Set A than to Set

B. Each group's response time to the Sets A and B were presented in Table 25. A

similar analysis was conducted with response accuracy data. Only a marginal effect of

group was found (F = 2.89,p = .07). As there were no statistically significant Group

effects on reaction time and response accuracy, it was assumed that the measures taken to

counterbalance Stimulus Set with interference conditions dispersed differences between

stimulus sets among different interference conditions, and no further attempt to examine

Stimulus Set's interaction with other variables was taken.


Table 25. Reaction time and accuracy on the Phoneme Match Test for DPD, ARPP, and
CTRL groups.
DPD ARPP CTRL

RT % Correct RT % Correct RT % Correct

Set A 2958(633) 87(16) 2834(540) 93(6) 2758(445) 95(5)

SetB 3285(1136) 87(14) 3156(793) 92(6) 2711(543) 94(7)

Note: DPD = Developmental Phonological Dyslexia; ARPP = Adequate Reader with
Poor Phonology; CTRL = Controls.









NAPM and visual match. Descriptive statistics for the experimental measures,

NAPM and Visual Match, were presented in Table 26. Separate MANOVAs were

conducted for reaction time and response accuracy data, with Group as a between-subject

factor (DPD vs. ARPP vs. CTRL) and Task (NAPM and Visual Match), Interference

(Mouth and Foot) and Block (1 and 2) as within-subject factors. Because this analysis

was exactly the same as the analyses performed in the section on PI vs. CTRL, only

effects or interactions involving Group (DPD vs. ARPP vs. CTRL) were reported here to


Table 26. Reaction time and response accuracy on the NAPM and Visual Match Tests
for DPD, ARPP, and CTRL groups.
DPD ARPP CTRL

RT % Correct RT % Correct RT % Correct

NAPM

Mouth 4511 77 4179 84 3688 88

(1666) (17) (1472) (12) (1551) (10)

Foot 5117 82 4013 85 3653 89

(1510) (10) (1087) (5) (1135) (10)

Visual Match

Mouth 4936 76 4957 79 5333 86

(1902) (17) (1756) (7) (1694) (13)

Foot 5243 84 4742 83 4984 84

(1528) (11) (1100) (7) (1719) (13)

Note: DPD = Developmental Phonological Dyslexia; ARPP = Adequate Reader with
Poor Phonology; CTRL = Controls.







reduce redundancy. The MANOVA with reaction time data revealed a Group X Task

interaction (F = 3.76, p =.03). Pairwise comparisons using dependent samples t-test to

follow up on the Group X Task interaction revealed that both ARPP and CTRL groups

were faster in responding the to NAPM in comparison to the Visual Match (ARPP, t = -

2.33, p = .04; CTRL, t = -4.58,p = .00), but the DPD group's reaction time on these two

tasks did not differ (t = -0.83,p =.42).

The MANOVA on accuracy data did not reveal any Group main effects or

interactions. Other interactions and effects were exactly the same as those reported in the

PI vs. CTRL section and were not repeated here.

ADHD subjects. To see whether ADHD differentially affected the performance

of the two phonologically impaired groups, reaction time and accuracy data of ADHD

subjects were contrasted to those of non-ADHD subjects in Table 27. Dividing the DPD

and ARPP groups into ADHD vs. Non-ADHD subgroups dramatically reduced the

number of subjects in each subgroup, rendering multivariate analyses looking at

differences between groups unrealistic due to low power. Nevertheless, perusing Table

27 suggested that ADHD subjects tended to be less accurate than their non-ADHD

counterparts.

Interfering movements. To examine if the two phonologically impaired groups

engaged in mouth and foot interfering movements with equal facility, a MANOVA was

conducted with the interfering movement frequency data. The CTRL group was included

in this analysis as a comparison group. Group (DPD vs. ARPP vs. CTRL) was entered as

a between-subject factor and Task (NAPM and Visual Match), Interference (Mouth and

Foot), and Block (1 and 2) were entered as within-subject factors. Table 28 showed each









Table 27. Reaction time and accuracy of the phonologically impaired Non-ADHD and
ADHD subgroups.
DPD ARPP


Non-

ADHD

(n=5)

RT


Correct


ADHD

(n=6)


Non-

ADHD

(n =7)

RT


Correct


Correct


ADHD

(n=3)


Correct


NAPM


Mouth

Foot


4581

(356)

5551

(1951)


80

(13)

87

(9)


4454

(2333)

4756

(1081)


75

(21)

77

(9)


3772

(1468)

3716

(1036)


87

(11)

86

(6)


5130

(1164)

4705

(1035)


78

(14)

84

(1)


Visual

Match


5145 83 4761 69 4899

Mouth (1285) (8) (2415) (21) (1782)

Foot 5290 84 5204 84 4667

(1212) (8) (1868) (14) (1258)

Note: DPD = Developmental Phonological Dyslexia; ARPP =
Poor Phonology.


81 5092

(8) (2076) (

81 4917

(8) (797) (

Adequate Reader with


group's interference movement frequency index. The result of this analysis was similar

to the analysis with the two phonologically impaired groups combined, with the

exception that the four-way interaction (Group X Task X Interference X Block) was now
ii>






80
only marginally significant (F = 2.61, p = .09). Because breaking the PI group down into

the DPD and ARPP groups reduced the number of subjects in these subgroups, the power

to detect a four-way interaction in the present analysis if one existed was reduced.

Interestingly, an examination of Table 28 revealed that ARPP subjects appeared to

produce interference movements with greater facility than DPD subjects. The exception

Table 28. Interfering movement frequency index for the DPD, ARPP, and CTRL groups.
DPD ARPP CTRL
"NAPM~ ~ ---------- -
NAPM

Mouth .83 (.25) .87 (.25) .97 (.32)

Block 1 .83 (.27) .85 (.23) .98 (.35)

Block 2 .83 (.25) .89 (.28) .95 (.31)

Foot .90 (.19) 1.11 (.27) 1.01 (.26)

Block 1 .88 (.20) 1.08 (.27) .97 (.26)

Block2 .91 (.17) 1.15(.30) 1.05(.27)

Visual Match

Mouth .94 (.20) 1.04 (.24) .99 (.23)

Block 1 .93 (.18) 1.05 (.29) .98 (.25)

Block 2 .95 (.23) 1.04 (.20) 1.01 (.24)

Foot .96 (.17) 1.11 (.31) 1.04 (.30)

Block 1 .95 (.18) 1.07 (.33) 1.05 (.32)

Block2 .97(.18) 1.16 (.30) 1.03 (.30)

Note: DPD = Developmental Phonological Dyslexia; ARPP = Adequate Reader with
Poor Phonology; CTRL = Controls.







to this was during NAPM Mouth Interference, where both DPD and ARPP subjects

produced interfering movements with approximately equivalent facility, especially during

Block 1. Compared to the CTRL group, the ARPP subjects tended to produce interfering

movements with greater facility, except during the NAPM Mouth Interference condition,

whereas the DPD subjects tended to be slower in producing interfering movements than

the CTRL group. Comparing Mouth Interference to Foot Interference, the ARPP group

appeared to have the greatest discrepancy between these interference conditions during

NAPM, while neither of the other groups showed such discrepancy on either NAPM or

Visual Match.

To examine how interference movement frequency related to each group's

articulatory knowledge, Pearson correlations between the AAT score and interfering

movement frequency index for each group were calculated and reported in Table 29. The

positive correlation between the AAT score and the NAPM Mouth Interference

movement index was driven by the DPD group. The better the DPD group's articulatory

knowledge, the faster they were able to produce interfering mouth movements during a

name retrieval task. The negative correlation between the AAT score and the Visual

Match Foot Interference movement index was driven by the ARPP group. The better the

ARPP group's articulatory knowledge, the more slowly they produced foot interference

movements during a nonverbal visual task.


Predictors of Articulatory Knowledge


To explore if variables that predicted the AAT score differed for the DPD and

ARPP groups, a stepwise regression analysis was conducted with age, BNT Z-score,







Table 29. Pearson correlations between the AAT score and interfering movement
frequency index for DPD, ARPP, and CTRL groups.
DPD ARPP CTRL

NAPM

Mouth Interference .74* .06 -.14

Foot Interference .28 -.52 -.24

Visual Match

Mouth Interference .16 -.24 -.14

Foot Interference -.47 -.64 -.13

Note: DPD = Developmental Phonological Dyslexia; ARPP = Adequate Reader with
Poor Phonology; CTRL =Controls. p < .05.


Rapid Color Naming Z-score, Rapid Object Naming Z-score, LAC raw score, TONI-2

Quotient, and the three reading achievement standard scores (Word Attack, Word

Identification, Passage Comprehension) entered as independent factors. Pearson

correlations between the AAT score and each of these variables for DPD and ARPP

groups were presented in Table 30. The CTRL group's correlation between AAT and

each independent variable were also provided for comparison. For the DPD group, no

variable was correlated with the AAT. Consequently no variable was selected by the

stepwise regression as a predictor of the DPD group's AAT score. For the ARPP group,

the three reading achievement measures, which were highly correlated with each other,

were positively correlated with AAT score. A perusal of scatter plots of reading

achievement scores as a function of AAT scores indicated that these significant

correlations were valid and not due to the presence of extreme scores. The Passage

Comprehension standard score was selected by the stepwise regression analysis and







accounted for 52% (adjusted R2) of the variance to the ARPP group's AAT score (F =

10.59,p = .01). The correlation between the ARPP group's AAT score and reading

achievement measures indicated that the better their articulatory knowledge, the higher

reading attainment ARPP subjects were able to achieve.


Table 30. Pearson correlations between the AAT score and variables entered into
stepwise regression analysis for DPD and ARPP groups.
DPD ARPP CTRL
Age -.13 -.32 .23

BNT -.24 -.13 .35

Rapid Color Naming -.22 .45 .31

Rapid Object Naming -.28 .29 .28

LAC .20 .12 .40*

TONI-2 -.47 .25 -.23

WRMT Word Attack .02 .67* -.00

WRMT Word Identification -.04 .66* -.04

WRMT Passage Comprehension -.03 .76* -.18

Note: DPD = Developmental Phonological Dyslexia; ARPP = Adequate Reader with
Poor Phonology; CTRL = Controls; p <= .05.


Other than the variables examined in Table 30, Pearson correlations between the

AAT score and Phoneme Match performance were also examined. Both DPD and ARPP

groups' AAT scores were significantly correlated with their accuracy on the Phoneme

Match Test (DPD: r= .77,p= .01; ARPP: r= .76,p= .01). The better their articulatory

knowledge, the more accurate they were in deciding if phonemes matched.







Predictors of Phonological Awareness


Predictors of the LAC score for DPD and ARPP groups were also explored with

stepwise regression analysis. Age, BNT Z-score, Rapid Color Naming Z-score, Rapid

Object Naming Z-score, TONI-2 quotient, the three reading achievement standard scores

(Word Attack, Word Identification, Passage Comprehension), and AAT raw score were

entered as independent factors. Table 31 showed Pearson correlations between the LAC

score and each of these variables for DPD and ARPP groups. The CTRL group's

correlations were also listed for comparison. When the phonologically impaired group

was divided into DPD and ARPP groups, power to detect correlations was decreased such

that the previously significant correlations between the LAC score and reading

achievement measures were no longer significant for either the DPD or ARPP group.

The negative correlation between the LAC and Rapid Color Naming was significant for

the ARPP group only. The better their phonological awareness, the more slowly ARPP

subjects were able to complete rapid naming, especially of colors. No variables were

selected as variables of predictive value by stepwise regression analysis for either DPD or

ARPP groups.

Correlation between the LAC raw score and response time on the Phoneme Match

Test was calculated for the two phonologically impaired groups. Neither the DPD nor

ARPP group's LAC score was correlated with their performance on the Phoneme Match

Test (DPD: RT, Pearson's r = -.24, p =.49; accuracy, Pearson's r = .28, p = .41; ARPP:

RT, Pearson's r = -.08,p = .84; accuracy, Pearson's r = .08,p = .84).








Table 31. Pearson correlations between the LAC score and variables entered into
stepwise regression analysis for DPD and ARPP groups.
DPD ARPP CTRL
. ................Age -.07 .44 .48* .

BNT .24 .50 .07

Rapid Color Naming .18 -.61 .16

Rapid Object Naming -.08 -.50 .01

TONI-2 .14 .09 -.24

WRMT Word Attack .42 .18 .55*

WRMT Word Identification .26 .01 .44*

WRMT Passage Comprehension .24 .24 .39*

AAT .20 .12 .40*

Note: DPD = Developmental Phonological Dyslexia; ARPP = Adequate Reader with
Poor Phonology; CTRL = Controls; p <= .05.


Poor vs. Adequate Articulatory Knowledge


The articulatory feedback hypothesis of naming hypothesized that articulatory

feedback facilitates name retrieval. This hypothesis makes the assumption of the

presence of articulatory knowledge. There was theoretical interest in examining the

name retrieval process of those subjects with adequate articulatory knowledge and those

with inadequate articulatory knowledge. Presumably, the hypothesis may hold true for

those with adequate articulatory knowledge but not for those with poor articulatory

knowledge. To group subjects into poor vs. adequate articulatory knowledge groups, the

mean AAT score for the entire population of subjects was calculated, and one standard

deviation below the mean, which corresponded to a Z-score of-1, was selected as the
%*







86
cutoff score for grouping criterion. The mean AAT score of all 41 subjects was 6.44 with

a standard deviation of 2.03. Thus subjects with an AAT score of 4 or below were

grouped into the Poor Articulatory Knowledge group (PAK), and those with an AAT

score of 5 or above were grouped into the Adequate Articulatory Knowledge group

(AAK). Table 32 reported some descriptive statistics about each subject group.

Independent samples t-test indicated that the two groups did not differ in age (t = -0. 16,p

=.88) or intellectual aptitude as estimated by the TONI-2 (t= -1.46,p = .15). All of the

subjects in the PAK group were males, and the majority of this group was composed of

children with ADHD. Note there was unequal distribution of the number of subjects in

each group, with only seven subjects falling into the PAK group.


Table 32. Demographics of the Poor Articulatory Knowledge (PAK) and Adequate
Articulatory Knowledge (AAK) groups.
PAK AAK

(n=7) (n=34)

Age 9(2) 9(2)

TONI-2 111(7) 105(10)

M:F Ratio 7:0 20:14

ADHD 5 7



Descriptive Measures


The performance of PAK and AAK groups on descriptive measures was

examined for group differences. Table 33 summarized the groups' performance on these

measures. There was no difference between groups on any of the naming measures when






87
group differences were tested using independent samples t-test (BNT, t = -0.18,p = .86;

Rapid Color Naming, t = 0.84,p = .41; Rapid Object Naming, t = 0.23,p = .82; Naming

Test, t = 1.15, p = .26). Independent samples t-test indicated that the PAK group scored

more poorly on the LAC than the AAK group (t = 2.43, p = .02), not surprisingly as LAC

score was significantly correlated with AAT score (Table 20). Comparison of mean

standard scores between groups using independent samples t-test revealed that the PAK

group scored lower on Word Attack (t = 2.76,p = .01) and Word Identification (t = 2.27,

p = .03) compared to AAK, but not on Passage Comprehension (t = 1.77,p = .08).


Table 33. Means and standard deviations on descriptive measures for the PAK and AAK
groups.
PAK AAK
BNT -1.31 (1.83) -1.44(1.83)

Rapid Color Naming' -.65 (.72) -.21 (1.33)

Rapid Object Naminga -.63 (.87) -.46 (1.97)

Naming Testb 86 (5) 89 (6)

LACc 47(22)Y 68 (21

WRMT Word Attackd 74 (12)Y 92 (17)b

WRMT Word Identificationd 77 (13)8 94 (19)b

WRMT Passage Comprehensiond 83 (14) 95 (17)

Note: PAK = Poor Articulatory Knowledge; AAK = Adequate Articulatory Knowledge.
"Z-scores. b Percentage correct. 'Raw score. d Age-corrected standard scores. Within
each row, numbers with different superscripts were significantly different from each
other.







Experimental Measures


Phoneme match. Performance on the Phoneme Match Test was examined to see

if PAK and AAK groups differed in their ability to match phonemes. Separate ANOVAs

were conducted for reaction time and response accuracy with Group (PAK vs. AAK) as

the between-subject variable. Table 34 showed each group's performance on the

Phoneme Match Test. The PAK group was both slower in reaction time (F = 5.45, p=

.02) and less accurate (F = 66.91,p = .00) than the AAK group. The overall AAT score

and Phoneme Match reaction time was not correlated (i.e., all subjects combined;

Pearson's r = -.26, p =. 10). However, the AAT score did correlate positively with

Phoneme Match accuracy (Pearson's r= .64,p =.00). Because the PAK group's ability

to match phonemes was remarkably worse, their reaction time and accuracy on the

Phoneme Match Task were used as covariates in analyses involving NAPM because the

NAPM required phoneme match as an integral part of the task.


Table 34. Reaction time and accuracy on the Phoneme Match Test for the PAK and
AAK groups.
PAK AAK

Reaction time 3385 (1018) 2802 (490)

% Correct 76(11) 95(4)

Note: PAK = Poor Articulatory Knowledge; AAK = Adequate Articulatory Knowledge.


NAPM and visual match. Subjects' reaction time and accuracy on the NAPM

were reported in Table 35, and their Visual Match performance were reported in Table

36. Separate analyses were conducted with NAPM and Visual Match because Phoneme

Match performance was used as a covariate in analyzing NAPM data while the use of this
Ot







covariate would be inappropriate for Visual Match because phoneme match was not

required as part of the Visual Match Test. Table 35 represented subjects' scores without

the covariate extracted.


Table 35. Reaction time and accuracy on the NAPM for PAK and AAK groups.
Numbers represent data without the covariate extracted.
PAK AAK

RT % Correct RT % Correct

Mouth

Block 1 5795(2460) 70(17) 3836(1541) 86(12)

Block 2 5142(2113) 79(18) 3659(1151) 87(14)

Foot

Block 1 5838(1671) 82(12) 3881(1151) 87(10)

Block 2 4904(1687) 85(10) 3888(1275) 85(12)

Note: PAK = Poor Articulatory Knowledge; AAK = Adequate Articulatory Knowledge.


On the NAPM (Table 35), separate MANCOVAs were conducted for reaction

time and response accuracy data. Group (PAK vs. AAK) was the between-subject factor,

Interference (Mouth and Foot) and Block (1 and 2) were within-subject factors, and

Phoneme Test reaction time and response accuracy were covariates in MANCOVAs

analyzing reaction time and response accuracy, respectively. The reaction time

MANCOVA revealed a significant Group X Interference X Block interaction (F = 4.33,p

= .04), an Interference X Block X Covariate interaction (F = 6.65,p = .01), an

Interference X Block interaction (F = 6.57, p= .01), and a Group X Block interaction (F

= 4.43,p = .04). The Group X Interference X Block interaction was explored by