Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
The neural correlates of skilled reading: an MRI investigation of phonological processing
(USC Thesis Other)
The neural correlates of skilled reading: an MRI investigation of phonological processing
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
THE NEURAL CORRELATES OF SKILLED READING:
AN MRI INVESTIGATION OF PHONOLOGICAL PROCESSING
by
Allison Zumberge Orechwa
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(NEUROSCIENCE)
August 2009
Copyright 2009 Allison Zumberge Orechwa
ii
ACKNOWLEDGMENTS
I would like to acknowledge my wonderful advisor, Frank Manis, for his valuable
guidance and wisdom throughout this project. I am also grateful for the technical and
personal support from my other committee members, Zhong-Lin Lu, Elaine Andersen,
Toby Mintz, and Roumyana Pancheva. Also, JC Zhuang’s extensive knowledge of MRI
has been especially helpful. To my labmates, thanks for your suggestions during lab
meetings and overall support with the various technical aspects of my research: Jennifer
Bruno, Jason Goldman, Rachel Beattie, and numerous undergraduate assistants. Thank
you.
To my friends and family, thank you for your continued love and support
throughout my graduate career and the experiences that have led me here. I could not
have accomplished this without you!
iii
TABLE OF CONTENTS
Acknowledgments
List of Tables
List of Figures
Abbreviations
Abstract
Chapter 1: General Introduction
Chapter 2: The Relationship Between Behavioral and fMRI Measures
of Phonological Processing and Reading Ability in Adults
Chapter 3: Reading Ability and Functional Connectivity within the
Phonological Processing Network
Chapter 4: Reading Ability and White Matter Microstructure in Major
Reading Tracts
Chapter 5: General Discussion
References
ii
iv
v
vi
vii
1
30
70
87
102
111
iv
LIST OF TABLES
Table 1: Cognitive profile of sample
Table 2: Correlations between behavioral measures of reading,
verbal, and nonverbal abilities
Table 3: Properties of the functionally localized regions of interest
Table 4: Regions appearing in group-level contrast of Passage
Reading and Foreign Font conditions
Table 5: Correlations between behavioral measures and ROI activity
during easy and difficult passage reading
Table 6: Linear Regression Analyses
Table 7: Correlations between behavioral measures and ROI activity
during PLDT task
Table 8: Correlations between pairs of ROIs (average activation
during Passage Reading)
Table 9: Correlations between behavioral measures and strength of
interregion temporal correlations
Table 10: Correlations between behavioral measures and average FA
values of ROIs
45
46
51
52
55
58
61
79
81
95
v
LIST OF FIGURES
Figure 1a: Group contrast map of region-of-interest localizer:
Phonological Lexical Decision Task minus Line Judgment
Figure 1b: Group contrast map of experimental task: Passage
Reading minus Foreign Font
Figure 2: Scatter plots of select ability-activity relationships
Figure 3: Functional Connectivity Illustration; lines represent
significant correlations between average ROI activation during
Passage Reading
Figure 4: DTI regions of interest
Figure 5: Specific and nonspecific relationships in three white matter
regions
49
50
56
80
93
96
vi
ABBREVIATIONS
ACR – Anterior corona radiata
AG – Angular gyrus
aIFG – Anterior inferior frontal gyrus
BOLD – Blood-oxygenation-level-
dependent
CC – Corpus Callosum
CP – Cerebral peduncle
DTI – Diffusion tensor imaging
ERP – Event-related potential
FA – Fractional anisotropy
fMRI – Functional magnetic resonance
imaging
GORT – Gray Oral Reading Test
iCST – Inferior corticospinal tract
IFG – Inferior frontal gyrus
ILF – Inferior longitudinal fasciculus
MRI – Magnetic resonance imaging
ND – Nelson Denny
ND-Comp – Nelson Denny
comprehension
ND-RR – Nelson Denny reading rate
OT – Occipitotemporal region
PA – Phonemic awareness
PET – Positron emission tomography
PH – Pseudohomophone
pIFG – Posterior inferior frontal gyrus
PLDT – Phonological lexical decision
task
PLIC – Posterior limb of internal capsule
pSTG – Posterior superior temporal
gyrus
PW – Pseudoword
R IFG – Right inferior frontal gyrus
RD – Reading disability
ROI – Region of interest
ROQS – Reproducible Objective
Quantification Scheme
RPN – Rapid picture naming
SCR – Superior corona radiata
SLF – Superior longitudinal fasciculus
SMA – Supplementary motor area
STG – Superior temporal gyrus
TOWRE-PDE – Test of One Word
Reading Efficiency – Phonemic
Decoding Efficiency
TOWRE-W – Test of One Word
Reading Efficiency – Sight
Words
vii
ABSTRACT
Studies of reading development and disability have identified phonological skill
as a major contributor to reading ability in children. The relationship seems to hold across
a wide range of adult, skilled readers, as well, but its neural basis has yet to be identified.
In order to examine the predictive power of phonological ability on adult reading ability,
three studies were conducted in which both behavioral and neural indices of phonological
processing were correlated with measures of oral reading fluency and silent reading
comprehension. The first study investigated the influence of fMRI activity within the
phonological processing network during a self-paced passage reading task. Regions of
interest for phonological decoding were identified on an individual basis, using subjects’
performance on a phonological lexical decision task. Activity during the natural passage-
reading task in the left occipitotemporal region, posterior superior temporal gyrus,
supplementary motor area, and anterior inferior frontal gyrus was positively correlated
with reading measures, providing neural evidence for the relationship between
phonological processing and adult reading ability. The second study aimed to identify
interregion connectivity pathways that contribute to differences in reading ability. By
comparing timecourses of activation within the individually localized regions of interest,
it was discovered that superior readers showed more robust connections between two of
the regions. Connection strength between posterior superior temporal gyrus and anterior
inferior frontal gyrus was positively correlated with word recognition ability. In the third
study, diffusion tensor imaging was performed on a larger sample of adult subjects to
assess structural connectivity contributions to reading ability. Using a semi-automated
viii
tracing method, FA values of the corpus collosum and 7 bilateral tracts previously
reported to be associated with reading ability were recorded. Positive correlations
between reading ability and bilateral superior corona radiata, left inferior corticospinal
tract, and right superior and inferior longitudinal fasciculus support previous hypotheses
that higher integrity of white matter tracts connecting the reading network is one marker
of superior reading ability. Altogether, the results indicate that highly skilled readers
benefit from increased engagement of and stronger coherence within the phonological
processing network during natural, everyday reading.
1
CHAPTER 1: GENERAL INTRODUCTION
Of the multiple subcomponent processes that contribute to reading, the one that
has stood out as the best predictor of reading ability is phonological processing.
Converging lines of evidence have indicated that behavioral performance on tasks that
measure phonemic awareness and decoding skill, for example, has a strong positive
relationship with reading ability across age and skill level. Relatively fewer studies have
investigated the relationship at the level of brain mechanisms, especially in adult samples
that include highly skilled readers. If activity in and between brain regions involved in
phonological processing varies with phonological skill, then a relationship that is
established using behavioral techniques should also be observable using functional
neuroimaging techniques. The overarching aim of the current studies was to investigate
the relationship between phonological skill and reading ability in adults at both the
behavioral and the neural levels.
The first aim is to replicate previous findings that have identified phonological
ability as a contributor to variability in adult reading skill. The relationship between
phonological ability and reading in young children has been well established: findings
from longitudinal studies link early skill at phonological awareness (PA) with later
success at reading (Mann & Liberman, 1984; Wagner et al., 1997). Beyond development,
efficient use of phonological information in processing oral and written language, as
measured by tasks such as rhyming, nonword reading, and phoneme addition or deletion,
is the skill most often correlated with reading ability (Bruck, 1998; Ehri, 1998; Savage et
al., 2005). The core deficit in reading disability (RD) is phonological, and evidence is
2
mounting that portrays RD as one end of a continuum; phonological ability and reading
fluency are connected along the entire spectrum. Significant correlations between the
abilities have been found in a range of samples, from children (e.g., Mann, 1984) and RD
adolescents (e.g., Wagner et al., 1997), to college students with and without RD (e.g.,
Apthorp, 1995). It seems, therefore, that individual differences in phonological skill are
related to subsequent, as well as concurrent, individual differences in reading ability. The
current study aims to provide corroborating evidence that phonological skill, as measured
by nonword reading efficiency, is correlated with general reading ability in adults from a
wide range of skill levels.
The second aim is to link individual differences in reading skill to activity levels
in brain regions associated with phonological processing. The source of individual
differences in neural substrates of reading is an interesting and novel question to address.
Neuroimaging studies are beginning to discuss the existence of variability in anatomical
and functional localization, and metanalyses of reading studies usually blame the problem
on task effects and/or varied cognitive strategies, since studies often employ single-word
passive reading tasks that can be performed using more than one strategy. A task that
forces subjects into a strategy that relies on phonological decoding, nonword rhyming,
would tap into individual variation in phonological processing by disallowing slower
readers to rely on a compensatory orthographic strategy.
The large body of research on phonological processes in the brain has identified at
least two regions involved in the phonological aspect of reading, as revealed by tasks like
letter and word rhyming and nonword decoding: left inferior frontal gyrus (IFG) and left
posterior superior temporal gyrus (pSTG) (Poldrack et al., 2001; Heim, Opitz, Muller &
3
Friederici, 2003; Majerus et al., 2005; Joseph, Noble & Eden, 2001; Simos et al., 2001).
Whereas early imaging studies—and many still today—rely on group data to identify the
regions involved in a particular task, the recent trend in individual differences has
inspired some researchers to relate individual activity levels to skill (or degree of
disability or disease). Converging evidence from various populations indicates that
reading development and skill learning, alike, are associated with focalization,
lateralization, and integration of regions (Schlaggar & McCandliss, 2007; Booth et al.,
2001; Frost et al., 2009; Booth, 2007). Within the mature reading network, higher ability
is associated with activation in posterior loci, namely occipitotemporal and
temporoparietal regions (Shaywitz et al., 2002; Temple et al., 2001; Aylward et al., 2003)
and, less often, frontal regions (Maisog, Einbinder, Flowers, Turkeltaub & Eden, 2008).
The second aim of this study is to replicate these positive correlations with a specific
focus on the phonological processing network and in the context of a novel, natural
reading task.
The third aim is to relate reading ability to connectivity among brain regions
associated with phonological processing. By measuring the coupling of changes in
imaging signals, functional connectivity methods enable researchers to make inferences
about information relay among regions. Previous functional connectivity results have
established a connection between activation patterns in IFG and posterior regions (Bokde,
Tagametz, Friedman & Horwitz, 2001), as well as other sites that may act as a
convergence zone between the two (Bitan et al., 2005). Strength of connections between
regions can also be assessed within individuals and then related to individual skills.
Increasing coupling strength between IFG and a temporoparietal region, for example, has
4
been associated with increasing reading ability (Hampson et al., 2006). Structural
connectivity measures the white matter pathways that underlie the functional
connections. Using diffusion tensor imaging, or DTI, researchers have begun to
understand the major fiber tracts that make reading, and cognition, for that matter,
possible. Connectivity is a new and informative approach to studying the interactions of
reading regions, as well as the relationship between the interactions and behavior.
Phonological Processing in Reading – in acquisition, dyslexia, and skilled reading
At the beginning of the search for neural correlates of reading ability is the need
to link variation in reading ability to an underlying process, and the strongest candidate
seems to be phonological processing. Reading is not an innate cognitive process like
motion perception or tone detection, which do not vary widely across healthy individuals.
Variation exists in the ability to efficiently convert visual symbols into a phonological
code and meaning, a skill that continues to develop over many years of practice.
Individual differences are most clearly detected among young children, whose reading
acquisition may be at different stages, and among disabled readers, whose difficulties
may be more or less severe. Adult skilled readers, although universally literate, still show
variability in reading speed, comprehension, and reliance on different components, and
this variability is not rooted solely in differences in general cognitive ability (Stringer &
Stanovich, 2000). Phonological processing skill has been identified in studies of
development, disability, and skilled reading as an important contributor to variation in
reading ability (Shankweiler, Lundquist, Dreyer & Dickinson, 1996; Stanovich, Nathan
& Vala-Rossi, 1986; Snowling, 2001).
5
Phonological processes refers generally to the mental operations that use
knowledge of the sound structure of language. Skill at reading a written language requires
several component phonological processes: awareness of the phonological structure of
words (phonological awareness), knowledge of a particular language’s grapheme-
phoneme correspondences and ability to synthesize individual phonemes into a known
word (blending), skill at holding multiple phonemes or words in memory when sounding
out a new word or comprehending a sentence (phonological working memory), and the
ability to retrieve phonological information rapidly from memory (as in a rapid naming
task; Wagner & Torgesen, 1987). Word identification is a tough chore to beginning
readers with nascent knowledge of phonological structure and regularities, but to a skilled
reader, automatic application of the different types of phonological processing helps
make reading effortless.
The connection between phonological processes and reading is strongest in
reading development. When beginning to read, children must establish phonological
awareness (PA), which is explicit knowledge of and access to the sounds of one’s
language, or phonemes. Sufficient PA enables one to segment, and otherwise manipulate
phonemes. It is considered an enabling subskill in early reading (Stanovich, 1986;
Wagner & Torgesen, 1987). Understanding the alphabetic principle is a prerequisite of
the ability to attach pronunciations to letter strings, pronounce them, and blend them
together. Indeed, findings from longitudinal studies link early skill at PA with later
success at reading. Mann and Liberman (1984) found that the correlation between
syllable segmentation skill measured in kindergarten and reading achievement one year
later was .40, even after controlling for IQ. A stronger correlation of .75 was found
6
between kindergarten phoneme reversal and later reading achievement (Mann, 1984).
The predictive relationship extends beyond the first year of reading acquisition, at least
up to 4
th
grade (Wagner et al., 1997). In a study of children ages 6 to 10 that compared
contributions of lower-level processing skills to nonword reading ability, PA made the
largest independent contribution of 12-18% (Badian, 1993). PA also plays an important
facilitator role in reading development as a self-teaching mechanism for learning new
words (Share, 1995). In the process of learning to read, a beginning reader gains explicit
knowledge of the sound structure of language “that complements the largely tacit
knowledge acquired from experience at listening and speaking” (Wagner and Torgesen,
1987). Longitudinal and group comparison studies support the claim that the two skills
are reciprocally related (Perfetti, Beck, Bell & Hughes, 1987; Morais, Cary, Alegria &
Bartelson, 1979). The strong relationship in children between reading ability and PA may
be rooted in a shared etiology, as evidenced by twin studies; findings from Zumberge,
Baker & Manis (2007) indicated that common genetic and environmental influences on
reading ability and PA acted through a common factor that was independent of IQ.
Soon after acquiring the alphabetic principle, beginning readers learn to abstract
the relationship between orthographic and phonological representations and apply that
knowledge when encountering new words. Children with developmental dyslexia, a
biologically based reading disorder, experience difficulties in both the representation and
application of phonological information (e.g., Pennington, Van Orden, Smith, Green &
Haith, 1990; Ramus, 2001; Snowling, 2001). Behavioral markers of dyslexia or reading
disability (RD hereafter) include slow name retrieval, deficits in phonological awareness,
poor verbal short-term memory and, most noticeably, slow and nonfluent reading. The
7
cognitive explanation that has received the most empirical support is a core deficit in
phonological processing (Stanovich & Siegel, 1994). The experimental tasks aimed at
identifying the underlying cause tend to involve multiple cognitive processes and many
levels of phonological representation—from speech perception and short-term memory to
retrieval of lexical phonology and manipulation of sublexical phonology—and together,
the findings provide converging evidence that RD is rooted in inefficient and poorly
specified phonological representations (Dietrich & Brady, 2001).
One line of convincing evidence has found that a phonological deficit can explain
individual differences among RD children. Snowling, Goulandris & Stackhouse (1994)
found that more severe phonological problems are associated with poorer decoding skills
and ability to spell phonetically. Phonological processing is also predictive of decoding
skills; after age, IQ, and reading age were entered into one study’s regression analysis,
phonological processing, perceptual speed, and visual memory each accounted for unique
variance in nonword reading performance (Snowling, 2001). Phonological processing
skill has also been useful in establishing a distinction between subtypes of RD. Children
with RD in a study by Manis, Seidenberg, Doi, McBride-Chang, and Peterson (1996)
were either deficient in only phonological processing or both phonological and
orthographic processing; only a small number exhibited a deficit that was purely
orthographic.
The relationship between phonological skills and reading ability is not unique to
beginning readers and individuals with RD. Adult skilled readers, who are often
investigated along with children and/or adults with reading difficulties and who relatively
rarely comprise a population studied on its own, seem to continue the trend. Wagner and
8
colleagues (1997) studied children longitudinally from the beginning reading stage to the
more skilled stage, and the predictive relationship between phonological awareness and
word recognition held throughout the developmental window. The skills that facilitate
reading development, therefore, continue to facilitate skilled reading. Likewise, the skill
that is deficient in RD is also important for skilled reading in adults. Scarborough (1984)
reported “very similar relationships” both between and within skilled and RD groups of
adults between reading level and measures that included “phonic analysis.” The idea of a
continuum of reading ability did not become widely accepted until ten years later, when
Shaywitz, Escobar, Shaywitz, Fletcher, and Makuch (1992) demonstrated with RD and
nonimpaired children that any type of discrepancy score—a common diagnostic tool—
that could be calculated followed a univariate normal distribution. The inclusion of
severe and less severe reading difficulties at the lower end of this continuum, which also
included normal reading ability, suggests that reading ability has the same underlying
influences for all degrees of ability. Phonological processing abilities are critical to
fluent, skilled reading even among adults at the upper end of the continuum, who have
had a lifetime of practice.
That upper end has been relatively neglected, but studies that do compare highly
skilled to less-skilled nonimpaired readers have found significant variation in both word
recognition and phonological processing skills (Stanovich & West, 1989; Bell & Perfetti,
1994; Dietrich & Brady, 2001; Watson & Miller, 1993). Stanovich and West found that
these skills correlated .47 to .51, whereas performance on tasks that measured
orthographic skills correlated with word recognition only .24 to .29. A study that
classified readers according to comprehension scores reported a significant difference on
9
vocalization speed of high and low frequency words and nonwords, but not on digits
(Bell & Perfetti, 1994). Sabatini, 2002, also found slower voice onset time for nonwords
in poor readers, further confirming the relationship between phonological processing
skills and level of reading ability. Regression analyses have indicated that performance
on measures of phonological processing skills, such as nonword naming, phonological
awareness tasks, and verbal short-term memory tasks are most predictive of word
recognition ability, followed far behind by orthographic skills (Stanovich & West, 1989;
Olson, Wise, Conners, Rack & Fulker, 1989). Other “extraphonological” sources of
variance have been difficult to identify. Stringer and Stanovich (2000) tested processing
speed as a possible predictor, but after IQ and phonological awareness measures were
entered into a regression equation, reaction time measures failed to account for a
significant amount of additional variance in word reading. By contrast, phonological
awareness did contribute independently to variance in word (11.4%) and nonword
(24.5%) reading, after IQ was entered. From childhood to adulthood, phonological
processing remains the most robust predictor of reading ability.
Neural Bases of Phonological Processes – Localization
In order to compare the neural activity of skilled and less-skilled readers, the
network of regions that participate in phonological processing must first be characterized.
Advances in neuroimaging have enabled researchers to explore the components of
reading and other complex cognitive processes in healthy individuals and integrate the
findings with those from lesion studies. A clearer picture of the thinking brain is
emerging, with less of an emphasis on discrete functional modules than lesion studies
10
would have us believe. Component processes are not defined by distinct patches of
cortex; rather, the components of reading—phonology, orthography, and semantics—are
each subtended by distributed neural activity in multiple overlapping regions and the
connections between them. The identification of distributed regions that contribute to
phonological processing is a result of converging lines of evidence from the lesion
method and functional neuroimaging such as fMRI, PET, and TMS.
The lesion method of mapping functions to brain regions involves determining in
a patient of stroke or surgery the location of the lesion and the area of behavioral deficit.
This method has its limitations, including the scarcity of subjects and the lack of
specificity in location of the lesion. However, it has succeeded in identifying some
general regions that are necessary for phonological processing. Lesions in the left
temporoparietal region have resulted in selective deficits in phonological short-term
memory, but not rehearsal (Silveri & Cappa, 2003), and in vowel reading (Semenza et al.,
2006). Disruptions in one specific area within the temporoparietal junction, the posterior
superior temporal gyrus (pSTG), have been associated with phonological agraphia, the
inability to spell heard nonwords (Kim, Chu, Lee, Kim & Park, 2002). Electrocortical
stimulation to the same area in brains of healthy subjects resulted in complete inability to
read nonwords in three out of four subjects (Simos et al., 2000). The superior temporal
gyrus, however, does not carry out phonological decoding in isolation; nonword reading
and writing is also affected in patients with lesions in the left inferior frontal gyrus (Fiez,
Tranel, Seager-Frerichs & Damasio, 2006; Ferreres, Lopez & China, 2003; Marien,
Pickut, Engelborghs, Martin & De Deyn, 2001). These findings suggest that two regions
necessary for phonological processing are the IFG and pSTG.
11
Functional neuroimaging studies have confirmed these general findings,
localizing processes that rely on phonological skills to the IFG and pSTG (Poldrack et al.,
2001; Heim et al., 2003; Majerus et al., 2005; Joseph et al., 2001). A metanalysis
combined 45 functional imaging studies that reported activation peaks for tasks
considered to be phonological in nature (Vigneau et al., 2006). These tasks included
reading, rhyming, discriminating, articulating and repeating words and nonwords, and the
resulting clusters of peaks extended in the frontal region from the precentral gyrus to the
inferior frontal gyrus, and in the temporoparietal region from supramarginal gyrus to the
middle temporal gyrus. Contrasts aimed at localizing semantic and sentence processing
resulted in clusters in similar areas, many of which overlapped with phonological
clusters. A plausible interpretation of these results is that phonological processes do not
occur in highly specialized regions. Conceptual and methodological problems with this
meta-analysis, however, limit the usefulness of its findings. Combining functional data
across not only a population of subjects but a wide array of tasks and baselines introduces
huge variability into any functional-anatomical mapping. Due to substantial individual
variability in brain anatomy and function, imaging studies of vision processes almost
always employ a “retinotopy” scan to each subject. The occipital lobe handles the
processing of many components of the visual scene. Retinotopy scans functionally dissect
the occipital cortex into distinct regions responsible for distinct processes. A complex
task like reading activates a large network of regions because it entails countless
component processes, including not only phonology, orthography and semantics, but also
focused and sustained attention, maintenance of the task instructions, working memory in
some cases, and inhibition of movement and wandering thoughts. Localization of
12
phonological processes to general areas is a good first step, but it is not specific enough
to dissociate it from other, concurrent processes. For optimal mapping of areas selective
to phonological processing, two methodological steps seem necessary: careful design of
experimental and control conditions—equating mental processes as much as possible—
and subject-level localization.
Less noisy findings have suggested that phonological processing does, in fact,
occur in selective regions. In the following review of imaging phonology-specific
regions, different phonological processes will be considered independently, as is usually
the case in experiments. Speech perception, a basic phonological process that does not
require metalinguistic knowledge like PA, has most often been observed in pSTG (e.g.,
Benson et al., 2001; Buchsbaum, Hickok & Humphries, 2001; Burton, Small &
Blumstein, 2000; Heim et al., 2003). Burton, Small, and Blumstein, 2000, considered the
claim that IFG is also involved in perception and provided evidence that only complex
speech segmentation tasks recruit the IFG (i.e., discriminating the first sound in the pair
/dip/-/tip/ does not require complete segmentation and only elicits activity in STG,
whereas discriminating /dip/-/ten/ requires processing in both STG and IFG). Poldrack
and colleagues (2001) claim that IFG is involved in speech perception, but only the
posterior portion, the triangularis, is involved. The non-perception condition, nonword
rhyming, when compared to letter case judgment, activated both the triangularis and
opercularis portions of IFG. An ERP study of rhyming attempted to dissect the different
processes involved in rhyming, made possible by the high temporal resolution of the
method. In a paradigm that presented subjects with two consecutive items before eliciting
a behavioral response, an early frontal asymmetry supposedly was associated with
13
rehearsal of the first item. A later negative potential in right temporo-parietal region
represented phonological matching, according to the authors’ interpretation (Grossi,
Coch, Coffey-Corina, Holcomb & Neville, 2001). There is general agreement that
posterior systems are involved in rhyming decisions (Paulesu et al., 1996; Sergent, Zuck,
Levesque & MacDonald, 1992; Petersen, Fox, Posner, Mintun & Raichle, 1989). Rhyme
generation, on the other hand, has been shown to occur in frontal regions (Shaywitz et al.,
1995). Other language production tasks—counting syllables in nonwords and naming
objects, colors, words, and pictures—have consistently activated areas in inferior frontal
cortex (Poldrack et al., 2001; Price, Veltman, Ashburner, Josephs & Friston, 1999;
Murtha, Chertkow, Beauregard & Evans, 1999). Findings from Heim et al., 2003,
indicate that, although different types of tasks appear to differentially activate IFG and
pSTG, both production and comprehension processes occur in both IFG and pSTG.
Superior temporal gyrus has been most often associated with the auditory
modality of language processing. Extreme findings of selective activation by auditory
words in adults, compared with activation by both visual and auditory words in children,
suggest a developmental increase in specialization of language regions (Burton, Locasto,
Krebs-Noble & Gullapalli, 2005; Booth, et al., 2001). Phonological processing could be
considered high-level auditory processing, likening the relationship between STG and
nearby primary auditory cortex to that between areas that process high-level visual
properties and primary visual cortex. Whereas areas outside of V1, called extrastriate
cortex, are specialized for visuospatial processing, motion perception, and shape
recognition, pSTG seems to be specialized in adults for processes like phonological
encoding (Levelt, Praamstra, Meyer, Helenius & Salmelin, 1998; Graves, Grabowski,
14
Mehta & Gupta, 2008), phonological learning (Majerus et al., 2005), and subvocal object
naming (Hickok et al., 2000).
Posterior STG activation, however, is not limited to the auditory modality, calling
into question its specialized status. Activation by both visual letters and auditory speech
sounds was observed in ERP and fMRI studies, whose authors generally conclude that
pSTG is heteromodal and may underlie integration of visual and phonological forms
(vanAtteveldt, Formisano, Goebel & Blomert, 2004, Booth et al., 2001). From here, the
integrated information may be relayed to semantic regions for further integration
(McCrory, Mechelli, Frith & Price, 2005).
At some point in the processing stream, the phonological representations must be
converted into articulatory programs, and that seems to occur in the inferior frontal area,
near primary motor cortex (Fiez & Petersen, 1998, Hagoort et al., 1999). Inferior frontal
cortex has classically been referred to as Broca’s area. Pairing increased resolution with
carefully controlled task design has led to a functional parcellation of the area. Overt
semantic processing has been localized in the anterior, ventral aspect of IFG, while the
dorsal aspect is recruited during phonological tasks (Poldrack et al., 1999; Devlin,
Matthew & Rushworth, 2003). Gold, Balota, Kirchhoff and Buckner (2005) observed
through repetition priming that while both posterior and anterior aspects of inferior
prefrontal cortex are engaged in both phonological and semantic processing, the posterior
aspect displays preferentiality to phonological processing. This finding was replicated by
McDermott, Petersen, Watson, and Ojemann (2003) in a study that compared activation
during tasks that required attention to either semantic or phonological relations. In an
early interpretation of the semantic involvement of IFG, Fiez (1997) attributed IFG
15
processing, as a whole, to retrieval, maintenance, and control of semantic information,
while posterior regions are committed to long-term storage of conceptual semantics.
More recently, Gold and Buckner (2002) have claimed that both anterior and part of
posterior portions of IFG are involved in both semantic and phonological retrieval;
differing interpretations have been a result of differing task demands on retrieval
processes.
The right hemisphere cannot be ignored when considering phonological or any
language processing, although studies typically focus on the left hemisphere. The ERP
study by Grossi and colleagues found a N400 over the right temporo-parietal area. A
study that correlated functional timecourses of language-related regions—also termed
functional connectivity—dissociated phonological networks of dyslexic and good
readers, noting differences in connections with both the left IFG and the right (Stanberry
et al., 2006). Findings from an early fMRI study of nonimpaired adult readers discovered
a link between rightward lateralization in IFG and sensitivity to orthographic regularity;
readers who were more sensitive to regularity of spelling-sound correspondences in real
words, and therefore were presumed to rely more heavily on phonological processing
when identifying words, displayed more right IFG and right extrastriate activity than did
those who were not as sensitive (Pugh, et al., 1997). Investigations of phonological
processes should therefore not limit their search to the left hemisphere for fear of
excluding neural populations that may make a critical contribution.
16
Neural Bases of Phonological Processes – Individual Differences
Neuroimaging studies continue to localize subcomponents of reading to specific
networks, but the trend of the past decade has been to relate activity levels in those
networks to individual differences in age and skill. According to multiple accounts, the
development of reading is associated with increasingly lateralized, focalized, and
specialized neural activity (Schlaggar & McCandliss, 2007; Booth et al., 2001; Frost et
al., 2009). After many years of reading experience, activity levels are still susceptible to
change due to age, training, shifts in strategy, or skill effects. Previous reports
demonstrating that brain activity varies with reading ability have converged on a model
in which increasing skill is positively correlated with activity in posterior regions,
including those in the temporoparietal area and the fusiform gyrus. This is true across
tasks, age groups, and skill levels. The data from anterior portions of the reading network,
however, are less consistent and merit further investigation.
As part of the emerging field of developmental cognitive neuroscience, a
framework for reading development has been introduced. First, reading areas become
increasingly specialized with age. Regions that are heteromodal in children become
unimodal; adults activate STG in response to auditory words and more inferior regions in
response to printed words, whereas children’s activations to the different modalities
overlap substantially (Booth et al., 2001). Frost et al. (2009) attributes the relationship to
phonological awareness (PA). Both print and speech stimuli activated STG to varying
degrees in a sample of children ages 6 to 10, and the amount of overlap correlated
positively with PA. These data indicate that STG is important in integrating print and
speech, which lays the foundation for reading development.
17
Second, processing shifts occur. Pugh and colleagues (2001) describe a model of
reading development that seems to mirror the behavioral development of reading skill,
with the temporoparietal “analytic” zone dominating word processing at first, followed
by a shift to a ventral system underlying fluent word recognition. The temporoparietal
system is involved in learning mappings between orthographic and phonological forms,
but activity decreases as reading expertise is acquired, potentially reflecting adults’
decreased reliance on this slow pathway (Church, Coalson, Lugar, Petersen & Schlaggar,
2008). Meanwhile, activity increases in the ventral zone, which includes the
occipitotemporal (OT) region or “visual word form area,” to support fast identification of
printed words (Schlaggar & McCandliss, 2007). Processing also becomes more left-
lateralized with age (Turkeltaub, Gareau, Flowers, Zeffiro & Eden, 2003; Schlaggar &
McCandliss, 2007).
This fast, late-developing occipitotemporal region has been dubbed a “skill zone”
by Shaywitz and colleagues (2002) because of its positive association with reading skill
both during and beyond development. Shaywitz et al. (2002) reports a positive correlation
between children’s OT activity and performance on a standardized test of pseudoword
reading. The relationship also exists at the lowest end of the spectrum, in children with
RD (Temple et al., 2001; Aylward et al., 2003). These children show little or no
activation in this area, relative to both age- and reading level-matched controls, bringing
Hoeft and colleagues (2007a) to the conclusion that OT underactivation is a functional
atypicality that is specifically related to RD.
Another area marked by underactivation in RD children is the temporoparietal
area, including STG, angular gyrus (AG) and supramarginal gyrus (SMG; Meyler et al.,
18
2007; Hoeft et al., 2006, 2007a; Temple et al., 2001; Backes et al., 2002). Intervention
has the potential to bring at least some of this region back online. In a study on children
with RD, activity in the posterior STG increased following 80 hours of phonological
training (Simos et al., 2002). In another intervention study, activity increased in multiple
areas, including regions surrounding the temporoparietal area, after just 28 hours of
training (Aylward et al., 2003). Increases in activity are clearly associated with increases
in skill, as has been confirmed by positive correlations with phonological processing
tasks among both RD and nonimpaired children (Turkeltaub et al., 2003; Shaywitz et al.,
2002). At the opposite extreme, hyperlexia has been linked to abnormally high activation
in STG and IFG (Turkeltaub et al., 2004).
Activity in anterior regions, specifically IFG, also seems to increase with age
(Turkeltaub et al., 2003). This is a surprising finding, given that one of the earliest
neuroimaging studies on RD reported greater activation in frontal regions in poorer
readers (Brunswick, McCrory, Price, Frith & Frith, 1999). However, findings from Hoeft
et al., (2007a) make a strong case against IFG hyperactivation as a marker of RD. Group
differences between RD children and controls matched on age disappeared when the
control group was instead matched on reading level. By contrast, middle frontal gyrus
(MFG) does seem to be negatively correlated with reading ability in children (Hoeft et
al., 2007b), as well as adults (Prat, Keller, & Just, 2007). A whole-brain comparison
between groups of nonimpaired adults, separated by performance on a reading capacity
test, revealed two regions with significantly greater activity in the low capacity group, in
bilateral MFG. A third region, in the temporoparietal area, had a positive relationship
with reading capacity, but only appeared at a less stringent threshold.
19
When more extreme skill groups are compared, namely nonimpaired and RD
adults, significant underactivation in the temporoparietal area is a consistent finding.
Adults with RD in an early PET study, Rumsey et al. (1997), showed significantly less
activity in bilateral temporal and motor areas compared to nonimpaired adults. Drastic
differences between the groups’ in-scanner performance have been cited as limitations to
these findings (Price & McCrory, 2007), but other studies have since replicated them.
Paulesu et al. (2001) is frequently cited as the most comprehensive study on RD in adults
because it sampled adults across three countries and three languages. Underactivation in a
large portion of the posterior reading network, including STG and OT, was common to
all three languages and was hypothesized to be a “universal neurocognitive basis of
dyslexia.” Structural MRI findings of decreased gray matter in STG in RD children and
adults suggest that the underactivation could be related to underlying microstructure
differences in cortex (Brown et al., 2001; Eckert et al., 2003; Steinbrink et al., 2008).
We cannot be sure, however, whether the brain-behavior relationships can be
traced back to the phonological nature of the regions that display them because the
investigations have been performed separately. That is, studies that have localized a
network of regions based on their specific involvement in a phonological task have not
examined the effect of skill on those areas during a separate reading task. Rather, results
showing a significant effect of skill are interpreted based on previous functional claims
about the regions. This design is indirect and is vulnerable to variation in exact
anatomical location of functional regions across tasks and individuals.
Altogether, the evidence on brain-behavior relationships in children and adults
with RD points to a positive relationship between reading and related abilities and
20
activity in posterior regions of the reading network, STG and OT. The paucity of studies
designed to investigate the neural correlates of reading across a wider range of ability,
and the lack of direct evidence that it is phonologically driven regions that effect brain-
behavior relationships, have inspired the current studies.
Neural Bases of Phonological Processes – Connectivity
One characteristic of highly skilled readers may be improved communication
between regions working on the phonological task. Connectivity analyses make clear the
need to examine regions not as independent modules but as parts of a larger network.
Generally, functional connectivity methods employ correlational analyses to “segregate
brain regions exhibiting similar temporal behavior.” It provides information about which
regions are connected either directly or indirectly, without proposing direction of
connections.
Functional connectivity findings for skilled adult readers, who are often the
comparison group in RD studies, consistently show connections between widely
distributed classical language areas. Activity in left frontal regions is correlated with
activity in many regions, including occipito-temporal junction (Stanberry et al., 2006),
which is considered one of the first stages in visual word recognition (e.g., Cohen et al.,
2000); angular and supramarginal gyri in parietal cortex (Rumsey et al., 1997; Hampson
et al., 2006); and pSTG, both at rest and during tasks like auditory language
comprehension or pseudoword reading (Hampson, Peterson, Skudlarski, Gatenby &
Gore, 2002; Joubert et al., 2004). Effective connectivity studies of reading, which
indicate direction of influences among regions, have indicated that the left occipito-
21
temporal junction is the sensory input source and IFG and STG are receiving units
(Bullmore et al., 2000; Mechelli Penny, Price, Gitelman & Friston, 2002). Mechelli et al.
(2005) noted that different portions of OT may drive different regions, depending on the
task demands; whereas an increase in IFG-pars triangularis activation during exception
word reading was associated with influences from anterior OT, an increase in left dorsal
premotor cortex during pseudoword reading was associated with posterior OT. Networks
can therefore be distinguished by their assumed function, phonological or semantic.
One advantage of analyzing interregional connections within subjects across time,
as opposed to across subjects, is the ability to relate the connections to subjects’ task
performance. This technique, sometimes called “connectivity-behavior analysis,” is
gaining popularity as individual differences in brain activity and connectivity are
becoming more recognized. Impressively, functional correlation and connectivity studies
of reading have been successful in establishing a pattern of connectivity-behavior
relationships. Evidence comes from the domains of development, disability, and variation
in skilled reading. The coupling of IFG with other regions, for example, increases with
age, while STG coupling decreases (Bitan et al., 2007). RD children, whose functional
connectivity patterns are disrupted, may not experience the typical increase in coupling.
The disruptions may only be due to current reading level, according to Richards et al.
(2007); following intervention, improved reading scores were accompanied by functional
connectivity patterns that were indistinguishable from those in the nonimpaired group. In
adult readers, profiles of connectivity differ between RD and nonimpaired individuals.
Shaywitz et al. (2003) found that whereas nonimpaired readers have connections between
the occipito-temporal region and left IFG, poor readers do not show this pattern. Instead,
22
the activity in the occipito-temporal region is correlated with activity in right frontal
cortex. One possible interpretation is that while skilled readers benefit from automatic
orthographic-phonological conversion, RD individuals must rely on long-term memory.
Other evidence for disruptions that persist into adulthood include a lack of typical
connections between IFG and right middle and inferior occipital gyri (Stanberry et al.,
2006), as well as between AG and extrastriate occipital and temporal cortex (Horwitz,
Rumsey & Donohue, 1998). Even if poorer readers are activating the same regions, more
tenuous temporal connections between them may account for their deficits in fluency.
Relatively fewer studies have investigated normal variation within the typical
connectivity patterns. Hampson and colleagues (2006) hypothesized that correlations
between activity in language areas would be related to reading skill in healthy,
nonimpaired adults. Outside of the scanner, 19 subjects were assessed on four reading
skills, word reading, nonword reading, spelling, and comprehension. The scanner task
consisted of continuous sentence reading, a condition that was compared to rest. Voxels
in the IFG served as a reference area, because it was the region most consistently
activated in all subjects, and from these reference voxels a timecourse was extracted. This
temporal data was correlated with every other voxel in the brain, in search of a region
that was temporally and therefore functionally connected to IFG. Preliminary results
indicated that IFG was correlated with adjacent and homologous frontal regions, as well
as a region spanning the temporal and parietal cortices the authors referred to as AG.
Interestingly, correlational and group-differences analyses revealed that the strength of
connection between IFG and AG varied with the reading skill measures. Higher
correlations were related to higher reading ability. The authors propose that the
23
correlation could arise because “better readers directly access a better lexicon, while poor
readers rely more on orthographic to phonological decoding.” The finding can be
interpreted in many ways, since a limitation of the study was that the sentence reading
task requires many different levels of processing. A task that better isolates individual
processes like phonological analysis may result in different relevant regions (such as the
pSTG), as well as different connectivity-behavior relations.
Providing validity to this method, i.e., confirming that functional correlations
truly reflect anatomical connectivity, is evidence from diffusion tensor imaging (DTI)
studies that have found differences in white matter integrity between skilled and poor
readers (e.g., Klingberg et al., 2000; Niogi & McCandliss, 2006). DTI is a specific MRI
procedure designed to image white matter tracts by taking advantage of the diffusion
properties of water molecules. Water diffuses along the orientation of axonal fibers
because the myelin sheath obstructs diffusion in other directions, leading to uneven or
anisotropic diffusion. DTI detects the primary direction of diffusion in each voxel, as
well as the degree of directionality; higher fractional anisotropy (FA) values are observed
in brain tissue that contains intact, dense and uniform fiber bundles. The integrity of
white matter tracts is compromised in brain areas affected by neuropathological
conditions like Multiple Sclerosis or Alzheimer’s Disease, leading to lower FA values
(e.g., Edwards, Liu & Blumhardt, 2001; Rose et al., 2000). Variation in FA is not limited
to diseased populations. Myelination continues into adulthood and throughout the
lifespan (Ben Bashat et al., 2005; Hayakawa, Konishi, Kuriyama, Konishi & Matsuda,
1991; McGraw, Liang & Provenzale, 2002; Olesen, Nagy, Westerberg & Klingberg,
2003), and individual differences in cognitive functioning can be related to healthy
24
variation in FA in both pediatric and adult populations (Schmithorst, Wilke, Dardzinski
& Holland, 2005; Nagy, Westerberg & Klingberg, 2004).
Further evidence for individual differences in white matter integrity comes from a
growing body of research on the variability in tracts associated with reading. Functional
MRI findings of RD-related underactivation in posterior regions have been
complemented by DTI findings of decreased FA values in and around major tracts
connecting those regions. The inferior longitudinal fasciculus (ILF) is a major tract
carrying axons between anterior and posterior ends of the temporal lobe and is probably
involved in communication between pSTG and other temporal sites during reading and
language processing. The superior longitudinal fasciculus (SLF) overlaps with the arcuate
fasciculus, long considered the highway between Broca’s (IFG) and Wernicke’s (pSTG)
areas. Disruptions in these white matter tracts have been observed in RD children and
adults, leading some to the interpretation that RD is a “disconnection” syndrome (Silani
et al., 2005; Steinbrink et al., 2008; Carter et al., 2009). Another candidate region that has
appeared in numerous DTI studies on RD is the superior corona radiata (SCR), part of the
corticospinal tract—where axons from the spinal cord and lower sensory areas converge
and project superiorly and anteriorly to far reaches of the cortex, including visual,
auditory, and articulatory motor areas. Despite the problem for interpretation that arises
from this lack of functional specificity, many findings of decreased FA in the SCR
among poor readers have been reported.
Klingberg and colleagues (2000) demonstrated a correlation between white
matter integrity within a temporo-parietal region near the SCR and reading ability among
both a RD group and a control group. They interpreted white-matter measures to be
25
indicative of “the strength of communication between cortical areas involved in visual,
auditory, and language processing.” Niogi and McCandliss (2006) extended the range of
ability and included more children scoring at least 1SD below average on a standardized
word reading test. Their anatomically-based region of interest approach led to easy
identification of the SCR as the tract associated with reading ability. They demonstrated a
correlational double dissociation between two different tracts and two different cognitive
abilities. Positive correlations between word reading and FA values in SCR, and between
working memory and anterior corona radiata (ACR), suggested that specific cognitive
domains may be related to specific white matter tract circuitry. It is possible, therefore, to
characterize reading ability in terms of communication between regions associated with
phonological processing. More evidence is needed from a wider range of adult readers.
Summary
Evidence supporting the first aim—to correlate behavioral measures of
phonological skill with reading ability—has come from over two decades of research.
The development of phonological skill facilitates and is augmented by the development
of reading; the core deficit in reading disability is phonological; and phonological skill
continues to be predictive of reading ability in adulthood. Functional neuroimaging of the
language network has provided support for the remaining goals of the present study.
Activity levels within reading-related regions, as well as connections between the
regions, have been correlated with behavioral measures of reading ability. The existing
literature calls for a more direct, comprehensive investigation of the relationship between
phonological processing and reading ability in a wide range of adults—including
26
individual identification of phonological brain regions and observation of activity within
and between those regions during a state of natural reading. Both school-based reading
instruction and future neuroimaging studies of reading ability would benefit from
converging evidence—from brain and behavior—confirming the importance of
phonological processing in skilled reading.
Method and hypotheses
To approach the relationship between phonological and reading skills from two
different angles—behavioral and neural—both cognitive testing and MRI scanning were
conducted on a sample of 35 young adults. The cognitive testing battery consisted of the
following measures: word and pseudoword reading, oral and silent passage reading,
verbal and nonverbal processing speed, and verbal and spatial IQ estimates. All subjects
who scored at or above the 40
th
percentile on the spatial IQ estimate were invited to
return for an MRI session, during which both structural and functional volumes were
collected. The oral and silent passage reading measures placed subjects along a
continuum.
Correlational analyses were conducted between the passage reading measures and
two sets of measures of phonological processing, one set that represents skill at the
behavioral level and another set that collectively represents phonological processing
occurring in the brain during reading activities. We chose phonological decoding as the
phonological processing skill to be considered in the analyses for two reasons. First, of all
the different types of phonological processing—decoding, encoding, phonemic awareness
and verbal short term memory, among others—decoding is used most directly in reading
27
connected text. Second, adult performance on more basic phonological measures often
displays ceiling effects and little variability. Therefore, the pseudoword reading measures
were correlated with the passage reading measures to fulfull the first aim. The second two
aims, relating reading ability to level of neural activity within phonological brain regions
and the strength of connections between them, required identification of the phonological
decoding network. The task used to identify the network was designed to engage
phonological processing, specifically decoding. Once the regions of interest (ROIs) in the
network were localized in each subject, activity within those regions during reading of
passages was correlated with the behavioral passage reading measure. That is, ROI
activity served as a proxy to the phonological processing occurring during passage
reading, and variation in activity was related to variation in reading ability measured
outside of the scanner.
The use of a self-paced passage reading task is rare in neuroimaging research on
reading. Since the study was designed to explore the influence of decoding skill on
reading ability, and in order for the findings to possess ecological validity, it was
necessary to measure the neural correlates of the skill while individuals were engaged in
natural reading. Just as with reading single words, reading connected text is thought to
involve phonological processing to some extent. A neuroimaging study has identified a
“phonology network” that is common to a wide range of language tasks that vary in
modality and linguistic complexity, from single-word listening to sentence reading
(Jobard, Vigneau, Mazoyer & Tzourio-Mazoyer, 2007). This common network may
underlie many different types of phonological processing. One type, phonological
decoding, is especially important in the case of low frequency or unfamiliar words, and
28
phonological working memory is another process that plays a role in sounding out
unfamiliar words, not to mention text comprehension. Phonological recoding, or the
activation of the phonological form from a written word along the way to accessing
meaning, is thought to be an automatic consequence of reading single words and
connected text. For example, Lee (2008) manipulated the amount of phonological
information in one word within each line of a story. Deleting a nonsilent letter (‘p’ from
‘pasta’) compared to a silent letter (‘p’ from ‘psalm’) resulted in longer reaction times,
suggesting that efficient reading of connected text relies on complete phonological
information. Whether activity in the ROIs selected for their phonological involvement
reflects decoding, recoding, or working memory, or all of the above, skill-related
variability should be correlated with reading skill. To be able to view the brain’s activity
while in a state that approximates everyday reading is a valuable and novel affordance of
the passage task.
Beyond variations in activity level within phonological regions, variations in
connectivity between the regions may also contribute to variability in reading ability. To
this purpose, two types of connectivity were measured and correlated with the passage
reading measures: temporal correlations of activity fluctuations in pairs of regions, or
functional connectivity, and degree of integrity of large white matter tracts identified by
diffusion tensor imaging, or structural connectivity. The functional connectivity
technique used as input the ROI activity from the first study. Whereas the first study
focused on magnitude of ROI activity, the variables of interest in the current study were
correlations between ROI activity during passage reading. Higher correlations were
assumed to imply stronger synchronization within the network. Likewise, more intact
29
white matter structures were also taken to indicate more efficient communication
between brain regions. The structural connectivity method was based on anatomical
localization of white matter, so any implications cannot be directly related to functionally
defined phonological networks. But because the analyses focused on white matter
pathways that are known to connect or border on the classic reading regions, it is
reasonable to interpret skill-connectivity correlations in terms of reading.
The general design described above allowed the testing of the following
hypotheses: 1) phonological skill, as measured behaviorally, contributes to the variation
in reading ability; 2) level of neural activity during reading tasks in one or more regions
empirically associated with phonological processing contributes to variation in reading
ability; and 3) strength of interregion connections will vary with skill, with highly skilled
readers showing stronger coupling between regions involved in phonological processing.
The first two hypotheses were addressed in Study 1, and the third was tested in Studies 2
and 3, which investigated contributions of functional and structural connectivity,
respectively.
30
CHAPTER 2: THE RELATIONSHIP BETWEEN BEHAVIORAL AND FMRI MEASURES OF
PHONOLOGICAL PROCESSING AND READING ABILITY IN ADULTS
Efficient phonological processing is critical to fluent, skilled reading. The strong
behavioral relationship between phonological processing and reading ability has been
demonstrated in children learning to read, in children and adults with reading disability,
and in a wide range of nonimpaired adults. At the neural level, phonological processing
has been localized to distinct, distributed regions that have been separately identified as
markers of skilled reading. In the current study, I aimed to provide direct evidence that
the neural activity in the phonological processing network during natural passage reading
is predictive of adult reading ability. In addition, a broad range of reading measures was
investigated in order to provide descriptive evidence for roles played by individual brain
regions in reading.
Individual differences in reading ability are most clearly detected among young
children, whose reading acquisition may be at different stages, and among disabled
readers, whose difficulties may be more or less severe. In both cases, a large proportion
of the variance can be attributed to phonological processing skill. Phonemic awareness is
considered an enabling subskill in early reading (Stanovich et al., 1986; Wagner &
Torgesen, 1987), and it is predictive of later reading success. Significant correlations
have been found between kindergarten skills like phoneme reversal or syllable
segmentation and later reading achievement (Mann, 1984; Mann & Liberman, 1984).
Armed with knowledge of the sound structure of language, beginning readers learn to
abstract the relationship between orthographic and phonological representations and
31
apply that knowledge when encountering new words. Children with developmental
dyslexia, a biologically based reading disorder, experience difficulties in both the
representation and application of phonological information (e.g., Pennington et al., 1990;
Ramus, 2001; Snowling, 2001). Behavioral markers of dyslexia, or reading disability,
include slow word retrieval, nonword repetition difficulties, poor verbal short-term
memory and, most noticeably, slow and nonfluent reading. Converging evidence suggests
that RD is rooted in inefficient and poorly specified phonological representations
(Dietrich & Brady, 2001). In studies of RD and nonimpaired adults, regression analyses
have indicated that performance on measures of phonological processing skills, such as
nonword naming, phonological awareness tasks, and verbal short-term memory tasks are
most predictive of word recognition ability, followed far behind by orthographic skills
(Stanovich & West, 1989; Olson et al., 1989). The skill that is deficient in RD is also
important for skilled reading in adults.
This consistent relationship is not surprising, considering that reading ability
appears to differ along a continuum. Scarborough (1984) reported “very similar
relationships” across skilled and RD groups of adults between reading level and
measures that included “phonic analysis.” The idea of a continuum of reading ability did
not become widely accepted until nearly ten years later, when Shaywitz et al. (1992)
demonstrated with RD and nonimpaired children that any type of discrepancy score that
could be calculated followed a univariate normal distribution. The inclusion of reading
difficulties at the lower end of this continuum, which also included normal reading
ability, suggests that reading ability has the same underlying influences for all degrees of
ability.
32
Individual differences in reading ability are also detectable at the level of the
brain, but it is not yet clear what specific underlying influence is at work. Compared to
RD children or adults, skilled readers show higher activity in the occipitotemporal region,
which is thought to underlie rapid printed word recognition (Schlaggar & McCandliss,
2007; Kronbichler et al., 2007), as well as the temporoparietal area, where phonological
processes including orthographic-phonological mapping may take place (Meyler et al.,
2007; Hoeft et al., 2006, 2007a; Temple et al., 2001; Backes et al., 2002, Frost et al.,
2009). Interpretations of skill differences in terms of subcomponent processes are seldom
based on functional localization data collected from the same subjects; they are instead
based on assumptions that the areas are the same ones identified by other studies as
involving those processes. These localization studies collectively imply that inferior
frontal gyrus (IFG) and posterior superior temporal gyrus (pSTG), classically referred to
as Broca’s and Wernicke’s areas, respectively, are loci of phonological processing.
Posterior STG seems to be specialized in adults for processes like phonological encoding
(Levelt et al., 1998; Graves et al., 2008), phonological learning (Majerus et al., 2005),
and subvocal object naming (Hickok et al., 2000), among others. Frontal regions
including IFG seem to be involved in the conversion of phonological representations into
articulatory programs (Fiez & Petersen, 1998, Hagoort et al., 1999). While phonological
processing has been attributed to both anterior and posterior portions of IFG, there is
ample evidence that the anterior portion is also involved in semantic processes (Devlin et
al., 2003; Gold et al., 2005).
Given their roles in phonological processing, along with the behavioral link
between phonological processing and reading ability, one could hypothesize that
33
increasing activity in pSTG and IFG with increasing skill reflects the regions’ roles in
phonological processing. The present study was designed to test this hypothesis directly,
by identifying the phonological network on an individual basis and correlating reading
skill with reading-related activity within the network. Localization prior to hypothesis
testing is advantageous for methodological and theoretical reasons. First, separate
localizer tasks are becoming increasingly common in neuroimaging studies because they
increase statistical power, reduce statistical bias, and preserve individual variation in
terms of functional location (Saxe, Brett & Kanwisher, 2006). Behavior correlation
studies are especially vulnerable to statistical inflation if the localization of regions of
interest is not separate from the secondary data analysis. The design of the current study
avoids this “nonindependence error” by separating the selection process from the
correlation analysis on an individual subject basis (Vul, Harris, Winkielman & Pashler,
2009). Second, limiting the analysis of reading-related activity to the regions shown to be
involved in phonological processing in the same individual reduces the possibility that
other subcomponents are responsible for the resulting correlations. The localizer task
used in the current study was designed to engage subjects in phonological processing—
decoding, specifically—by asking if a wordlike stimulus sounded like a real word. The
use of only pseudohomophone (‘brane’) and pseudoword (‘brape’) conditions, equated
across trials for length, syllables, orthographic neighborhood and bigram frequency,
ensured that the decision could not be based on the orthographic form of the word. After
the regions were defined in each subject as the “decoding network,” activity within the
network during a silent passage reading task was measured and related to reading skill.
34
Reading skill was measured using a large battery of reading and verbal tests,
providing a rare opportunity to relate specific regions’ activation patterns to specific
subprocesses, in addition to more global reading measures. For example, the role of
pSTG in phonological decoding can be tested by comparing its activity to behavioral
decoding measures. Likewise, tests that tap into semantic processing might be more
closely associated with activity in aIFG than pIFG, which have been proposed to have
overlapping but dissociable roles in semantic and phonological processing (Devlin et al.,
2003; Gold et al., 2005). Using regression analyses, variables were entered in a certain
order to test for mediation of relationships between activity and global reading measures,
as well as overlapping contributions from multiple regions. I considered measures that,
theoretically and as illustrated by correlation results, are components of global measures
of reading ability. Because the study was designed with the primary aim in mind, these
analyses should be considered secondary and exploratory.
Another possible secondary analysis is the comparison of the localizer and
passage-reading tasks, which may also inform hypotheses about regions’ roles. The silent
passage reading task was designed to test the ecological validity of previous reports of
skill-modulated activity in the reading network. Rather than performing a single-word
reading task, as is the case in the majority of neuroimaging studies, our subjects were
engaged in a more natural, self-paced passage reading task. They were instructed to read
successive portions of easy and difficult stories at a natural pace, with the only overt
response being a button-press upon completion. Only a handful of other neuroimaging
tasks have examined the neural correlates of reading connected text. Kujala et al. (2007)
presented subjects with stories one word at a time, manipulating reading effort by altering
35
the presentation rate; Yarkoni, Speer, Balota, McAvoy & Zacks (2008) also controlled
the word-by-word presentation rate of story stimuli, for the purpose of frequency and
other word-level comparisons; and Prat et al. (2007) examined the effect of skill on
activity during task that included sentence reading followed immediately after by a
comprehension probe. While controlling presentation rate or inserting comprehension
questions between sentences does not approximate everyday reading, some studies have
employed more naturalistic passage reading tasks (Lindenberg & Scheef, 2007; Xu,
Kemeny, Park Frattali, & Braun, 2005). Altogether, findings from these studies indicate
that activation during sentence and text reading overlaps with single-word reading
regions but also extends to additional areas. Direct comparisons of sentence reading to
both verbal short-term memory and single-word reading tasks have identified these
additional areas in bilateral temporal and occipital lobes (Cutting et al., 2006). Region of
interest activity, specifically, during both the localizer task and the passage reading task
may differ, as well. Despite a growing awareness of task effects, no study has used a
separate localizer task in addition to a naturalistic reading task to isolate the influence of
specific subcomponents on reading skill. The current study’s use of a passage reading
task attempts to fill that void.
The use of separate localizer and experimental tasks, although advantageous for
many reasons, has a potential disadvantage. We cannot ignore the possibility that ROI
activity reflects different processes recruited by the different tasks. Activation maps for
reading are quite similar across different study designs, imaging parameters, populations,
and tasks. It is not surprising, therefore, that multiple processes have been associated with
the same region; for example, aIFG shows sensitivity to both semantic and phonological
36
parameters (see Devlin et al., 2003). Relationships between behavior and activation in the
so-called “decoding” network, therefore, is not necessarily due to involvement in
decoding processing during passage-reading. Fortunately, numerous behavioral measures
of more basic reading skills were available to test for mediation using regression
analyses, thereby addressing alternative interpretations.
Previous reports of skill and task effects on activity in the reading network, along
with the behavioral evidence for a relationship between phonological processing skill and
reading ability, led us to the following hypotheses. The network involved in decoding
was hypothesized to include pSTG and IFG; the same regions appear in passage reading
contrasts. ROI activity during passage reading was predicted to vary with skill. If the
difference between highly skilled and less skilled readers has the same underlying basis
as that between nonimpaired and disabled readers—as would be predicted by the
continuum hypothesis—then one would hypothesize increasing temporoparietal activity
with increasing skill.
METHOD
Participants
Thirty-five young adults (24 female; average age 20 years, 5 months) participated
in both a behavioral testing session and an MRI session. All participants were right-
handed, monolingual native-English-speakers with normal or corrected-to-normal vision
and a negative history of neurological abnormalities.
Behavioral testing
37
Estimates of verbal and spatial IQ were measured using Woodcock Cognitive
Abilities subtests Verbal Comprehension, which requires participants to produce names,
synonyms, antonyms, and analogy completions to increasingly difficult items; and Spatial
Relations, which requires participants to identify puzzle pieces that comprise complex
target shapes (Woodcock, McGrew & Mather, 2001). In order to be included in analyses,
participants had to reach a percentile ranking of 40 on Spatial Relations only; a score
below the 40
th
percentile on Verbal Comprehension was accepted, since verbal IQ is
moderately correlated with reading ability, and the sample necessarily included
participants with poor reading skills.
Single-word reading ability was assessed using the Word Identification and Word
Attack subtests of the Woodcock Johnson III Tests of Achievement (Woodcock, et al.,
2001), which measure participants’ ability to accurately read increasingly difficult words
and pronounceable nonwords, or pseudowords, respectively. Test of One-Word Reading
Efficiency (TOWRE; Torgesen, Wagner & Rashotte, 1999) – Sight Words (W) and
Phonemic Decoding Efficiency (PDE) subtests – also measured word and pseudoword
reading accuracy, respectively, but under time constraints. Each subtest consisted of 104
or 63 items (W and PDE, respectively) presented in columns, and the participant was
instructed to read the items as quickly as possible without making mistakes. Standardized
scores are only available for ages 7 to 24, and because some participants were older than
24, raw scores were used in all analyses. Raw scores were computed by averaging across
two trials the number of items read correctly in 45 seconds or less. A second timed
pseudoword reading test, Timed Word Attack, was also administered. Subjects were
instructed to read a list of pseudowords from a different form of the standardized Word
38
Attack test (Woodcock Johnson-Revised Tests of Achievement) as quickly as possible,
with no imposed time limit. Whereas the time limit in TOWRE-PDE does not allow for
slower readers to complete the most difficult items, this modified Word Attack measured
all subjects’ performance on all items. Because accurately pronouncing pseudowords
requires the phonological processing component of decoding, scores on both timed and
untimed Word Attack tests and TOWRE-PDE were used as measures of decoding ability.
Measures of passage reading ability were also included in the test battery. The
Gray Oral Reading Test (GORT; Wiederholt & Bryant, 1992) assesses oral passage
reading fluency by measuring accuracy (number of deviations from print), rate (duration
of reading), and comprehension (number of questions answered correctly). Due to the
poor construct validity of the original comprehension questions—Keenan and Betjemann
(2006) established that children and adults could correctly answer the questions
significantly above chance without having read the stories—they were replaced by new
original, open-ended questions. Because the new comprehension questions were not
standardized, only the Accuracy and Rate measures were analyzed. A second passage
reading measure was able to assess reading comprehension; Nelson Denny (ND) requires
participants to read passages silently and answer multiple-choice comprehension
questions until the 20-minute time limit is reached (Brown, Fischo & Hanna, 1993). Both
reading rate (ND-RR), corresponding to the line number the participant reaches after the
first minute, and comprehension (ND-Comp), the number of questions answered
correctly, were recorded.
Two additional behavioral measures were included in the battery to address
possible effects of speed on reading ability, Rapid Picture Naming and Cross Out. Rapid
39
Picture Naming measures automatic word retrieval speed by instructing participants to
quickly name successive images of concrete, high frequency objects. Cross Out, a
measure of processing speed, presents participants with rows of unfamiliar symbols,
within which they are instructed to cross out each symbol that is identical to a target
symbol. The time it takes to complete all rows was used in analyses.
MRI Procedure
Prior to scanning, participants were familiarized with the scanning procedure and
trained on the functional MRI tasks. The first task, a phonological lexical decision task
(PLDT), was designed to localize regions of interest (ROIs) that are involved in
decoding. Each trial in the PLDT condition consisted of a 2000 ms presentation of a
single word-like stimulus, to which the participant was instructed to respond to the
question “Does it sound like a real word?” Following each stimulus was a fixation cross
for 300 ms and a blank screen for 200 ms. Items were divided into two conditions,
pseudohomophones (PH; e.g., ‘rane’), where the correct response is ‘yes’, and
pseudowords (PW; e.g., ‘brap’), which deserve a ‘no’ response. By presenting only PH
and PW, and no familiar words, the task required that participants engage in a decoding
strategy throughout the entire task. This favored the interpretation that the ROIs localized
by this task were primarily involved in phonological decoding, although lexical and
semantic access were likely also occurring (e.g., for the word ‘rain”, a homophone of
‘rane’). To avoid faulty interpretations, a priori hypotheses about the regions likely to be
active during decoding were considered when defining ROIs. In the control condition,
which appeared in alternating blocks, participants judged whether individually presented
40
line patterns were symmetrical. Stimulus duration was identical to that in the PLDT
condition.
The second task consisted of successive portions of passages that the participants
read silently and at their own pace. The passages were culled from forms of GORT that
were not used in the behavioral assessment. Half of the passage blocks consisted of
“easy” passages (GORT Stories 1 through 6), and half consisted of “difficult” passages
(GORT Stories 10 through 13). All portions of passages were matched on number of
words. A button press delivered the next portion of the current passage or the first portion
of a new passage until the 30-second block was over, at which time a control-condition
block began. In the control condition, participants were presented with blocks of a
foreign font and instructed to scan the lines, as if reading, and press the button when they
saw a target stimulus, whose identity was consistent throughout the task and could appear
in any position. No subject was familiar with the font (Sorawin). The control condition
was designed to simulate the passage condition in terms of basic visual and attentional
properties, while containing no verbal or semantic processing. To create the control
stimuli, each passage stimulus was converted to the foreign font, and the target item was
inserted in a random position. A fixed number of stimuli were available to the
participants, and, since both parts of the task were self-paced, the number of items seen
per block varied with participants’ speed. If a participant finished reading or scanning all
items in a block, an instruction to wait for the next block and to refrain from pressing the
button was presented in either a verbal format (“Please wait for the next section to begin.
Do not press the button.”), or a pictorial format (a crossed-out index finger), depending
on the condition. Participants were instructed to read each passage at a natural pace and
41
remember the stories, because they would be given a comprehension quiz after the scan.
This ensured that the participants were reading and comprehending during the passage
condition. The nonstandardized, open-ended comprehension questions were based only
on the first, second, or third portion of the passage block, so slower readers were not at a
disadvantage. The quiz was taken after the entire scan after all the stories were read and
additional scanning was performed. It was only intended to ensure that subjects were
reading the passages for comprehension.
Items in the first task were presented in a block design, and the order of word-like
stimulus types was randomized within each of the 8 PLDT blocks. Each 40-second block
contained 16 trials, and each of 2 runs contained 8 blocks total. The second task was also
presented in a block design, with eight 30-second blocks per run, and was repeated twice.
A fixation cross was presented for 12 s at the beginning and end of each run. Throughout
the scan, participants view the stimuli projected on a screen behind their head through an
adjustable mirror mounted on the head coil. A fiber-optics button box was provided for
responses. Stimulus display was programmed in MATLAB (The MathWorks, Natick,
MA) and Psychtoolbox (Brainard, 1997). Following the functional MRI tasks, structural
MRI data was acquired for co-registration purposes.
Materials
The word-like stimuli in the PLDT were drawn from a previous study by the
authors that employed a phonological lexical decision task. The word condition consisted
of concrete nouns ranging in frequency from 0 (‘proxy’) to 8,034 (‘people’) occurrences
in a corpus of 5,088,721 printed words (American Heritage Word Frequency Book;
Carroll, Davies & Richman, 1971). Corresponding PH were created by altering the
42
spellings of the words without affecting pronunciation (e.g., ‘stoan’). Pseudowords (PW)
were generated from an online database (MCWord; Medler & Binder, 2005). Of the
existing set of stimuli, the only conditions included in the present study were PH and PW,
which were equated for number of letters (range 4-6; mean 5.3), syllables (1-2; 1.5),
bigram frequency (135-4005; 1298), and Coltheart’s orthographic neighborhood (0-19;
3.6; Coltheart, Davelaar, Jonasson & Besner, 1977).
The blocks in the passage reading task consisted of consecutive portions of stories
from GORT forms that were not used in the behavioral testing battery. Blocks alternated
between two levels of difficulty, with stories 1-6 comprising the “easy” blocks and stories
10 through 13 comprising the “difficult” blocks. The content of the easy and difficult
passages differed mainly in vocabulary and sentence structure. Each story was split up
into equivalent sized portions—average of 25 words and 142 characters per stimulus,
with no difference between the easy and difficult conditions in number of words or
characters (F(1, 75) = 1.01, 1.85; ns).
Image Acquisition
Functional and structural imaging was performed on a Siemens Magnetom Trio 3
Tesla MRI unit (Siemens Medical Solutions, Malvern, PA) using a 12 channel head coil.
Earplugs and sound dampening headphones were worn by the participants for protection
from noise. Foam padding placed between the participants’ necks and head cradle was
used to minimize head movement.
High resolution structural images were acquired using a T1-weighted MPRAGE
sequence (FoV 256mm x 256mm; TI 800ms; TR 2530ms; TE 3.09 ms; Flip angle 10; 208
coronal slices), resulting in 1mm
3
isotropic voxels that covered the entire brain and part
43
of the cerebellum. Functional image parameters during the T2*-weighted echo-planar
imaging sequence (FoV 224 mm; TR 2000 ms; TE 32 ms; 28 axial slices) yielded 3.5 x
3.5 x 4 mm voxels with no over-sampling.
MRI analysis
Data were subject to online 3D PACE motion correction during acquisition.
BrainVoyager QX 1.10.3 (Brain Innovation, Maastricht, the Netherlands) was used to
preprocess the data. The functional data images were aligned to the last run with a rigid
body transformation and subjected to additional motion correction using trilinear
interpolation. Data were spatially smoothed using a 4 mm FWHM Gaussian kernel.
Temporal filtering included both linear trend removal and a high pass filter with a cutoff
of three cycles per timecourse. Slice scan time correction was performed with sinc
interpolation. Data were then aligned, both automatically and manually, to unnormalized
isotropic structural images. Then the structural and coregistered functional data were
normalized into a standard stereotaxic space (Talairach & Tournoux, 1988).
ROI Localization
Individual participants’ decoding networks were defined on the basis of a group
analysis of the PLDT minus barcode contrast. The subsequent search for individual ROIs
was constrained based on the average center of gravity of each ROI found in the group
contrast map. Ten mm
3
bounding boxes were created around the relevant peak and
superimposed on the functional map of each participant. Activation clusters that fell
within the box at q(False Discovery Rate) < 0.05 were defined. In the case of multiple
clusters overlapping the box, the cluster that fell closest to the center was defined. In
order to maintain a consistent ROI size across participants, 125mm
3
cubes were created
44
around the peak of each ROI. This localization method was chosen because it preserves
the individual data, which is often blurred in group maps due to the intersubject
anatomical and functional variability (Saxe et al., 2006; Vul et al., 2009).
RESULTS
Behavioral relationships
The sample comprises a wide range of reading ability (see Table 1), with four
subjects scoring at or below one standard deviation below the mean on Word Attack.
Performance on the reading measures was moderately to strongly intercorrelated, as seen
in Table 2. Pseudoword reading measures correlated significantly with the GORT and
ND passage reading measures, as did the word reading measures. Verbal IQ was
significantly correlated with Word ID and two of the passage reading measures but no
other verbal or reading measures. The correlations indicate that the passage reading tests,
which measured silent and oral reading comprehension, accuracy and fluency, was
related to more foundational skills such as phonological decoding, word identification
and general verbal ability, a finding that is in keeping with a vast reading literature (e.g.,
Rayner, Foorman, Perfetti, Pesetsky & Seidenberg, 2001). The two nonverbal tests
showed no relationship with any of reading measures, as expected. The two speeded non-
reading measures, RPN and Cross Out, were correlated, possibly reflecting a common
underlying general processing speed component. Likewise, the two IQ estimates were
moderately correlated because they both tap into general cognitive ability.
45
Table 1: Cognitive profile of sample
Range Mean SD
Age 18y 7m – 27y 0m 20y 5m 22m
Handedness 55 - 100 79.57 14.12
Verbal IQ SS 81 - 128 103.51 10.99
Spatial IQ SS 98 - 134 112.40 9.67
Word ID SS 89 - 124 105.91 8.5
Word Attack SS 83 - 117 98.60 9.52
TOWRE-W raw 77.5 - 104 97.20 (out of 104) 6.85
TOWRE-PD raw 34 – 59.5 51.44 (out of 63) 6.58
Rapid Picture
Naming SS
72 - 130 103.66 11.38
Cross Out seconds 128 - 245 182.49 28.99
GORT passage raw 71 - 130 116.54 13.83
Nelson Denny
comprehension
Grade Equivalent
11.5 – 18.9 16.85 2.28
Reading
Composite Z-score
-2.56 – 1.05 0 0.90
46
Table 2: Correlations between behavioral measures of reading, verbal, and nonverbal abilities
GORT
Rate
GORT
Accuracy
ND
Compre-
hension Word ID
TOWRE-
W
Word
Attack
TOWRE-
PDE
Timed
Word
Attack RPN
Cross
Out
Verbal
IQ
Spatial
IQ
GORT Rate -- 0.72** 0.58** 0.38* 0.59** 0.42* 0.63** 0.49** 0.31 -0.17 0.32 0.04
GORT Accuracy 0.72** -- 0.55** 0.64** 0.59** 0.61** 0.59** 0.48** 0.30 -0.04 0.36* 0.18
ND Comprehension 0.58** 0.55** -- 0.48** 0.55** 0.46** 0.46** 0.48** 0.27 -0.19 0.42* 0.14
Word ID 0.38* 0.64** 0.48** -- 0.49** 0.65** 0.52** 0.33 0.48** -0.14 0.46** 0.06
TOWRE-W 0.59** 0.59** 0.55** 0.49** -- 0.50** 0.76** 0.52** 0.58** -0.04 0.06 -0.07
Word Attack 0.42* 0.61** 0.46** 0.65** 0.50** -- 0.52** 0.44** 0.34* -0.02 0.23 0.06
TOWRE-PDE 0.63** 0.59** 0.46** 0.52** 0.76** 0.52** -- 0.55** 0.42* -0.12 0.18 -0.08
Timed Word Attack 0.49** 0.48** 0.48** 0.33 0.52** 0.44** 0.55** -- 0.10 -0.19 0.30 -0.04
RPN 0.31 0.30 0.27 0.48** 0.58** 0.34* 0.42* 0.10 -- -0.40* -0.05 -0.18
Cross Out -0.17 -0.04 -0.19 -0.14 -0.04 -0.02 -0.12 -0.19 -0.40* -- -0.06 0.16
Verbal IQ 0.32 0.36* 0.42* 0.46** 0.06 0.23 0.18 0.30 -0.05 -0.06 -- 0.40*
Spatial IQ 0.04 0.18 0.14 0.06 -0.07 0.06 -0.08 -0.04 -0.18 0.16 0.40* --
* p < 0.05; ** p < 0.01
47
Functional MRI task performance
Performance on the localizer was consistently high across participants. Mean
(SD) accuracy rates and reaction time for the PLDT (91.1% [4.3]; 1088ms [128]) and line
conditions (95.5% [4.3]; 894ms [124]) were significantly different (t[34] = 4.36 and
-8.97, for accuracy and reaction time, respectively; p<0.001), indicating that performance
on the PLDT task was slower and more error-prone.
While the passage task did not require a decision to be made, and therefore has no
online accuracy to be reported, reading rate and post-scan comprehension were collected
to serve as evidence that subjects were engaged in the passive reading task. The median
reading time for the easy and difficult passage conditions were 4.70 seconds and 5.36
seconds, respectively, and the median reaction time for the foreign font condition was
3.38 seconds. Paired t-tests revealed significant differences between all three conditions.
The foreign font trials were performed fastest and difficult passages were read slowest
(easy-difficult t(34) = -5.22; difficult-font t(34) = 7.57; easy-font t(34) = 6.23; p<0.001
for all comparisons). Performance on the post-scan comprehension quiz ranged from 1 to
10 correct, out of 10 questions total. The mean(SD) score was 51% (21). The nature of
the comprehension quiz—questions were open-ended and difficult, and removed in time
from the reading of the passages—might account for the generally poor performance.
However, the subjects’ reaction time differences, as well as their tendency to answer
correctly more of the questions from the easy passages (t[33] = 2.23; p < 0.05), indicates
that they were engaged in the task.
48
Functional MRI task activation
Figure 1a displays the six cortical regions that appeared in the PLDT group
contrast map, as well as in the majority of the individual participants: left occipito-
temporal gyrus (OT), left posterior superior temporal gyrus (STG), left anterior and
posterior portions of the inferior frontal gyrus (aIFG and pIFG), left supplementary motor
area (SMA), and right inferior frontal gyrus (R IFG). Based on the method described
above, all of the 35 participants were included in the OT and aIFG analyses, but a few
were excluded from the other analyses due to nonexistent or subthreshold activation. ROI
statistics are summarized in Table 3.
The main analysis of the passage-reading task was constrained to these
individually localized ROIs, which were assumed to comprise the cortical network
involved in decoding. Whole-brain results from the passage task are also presented, in
Figure 1b and Table 4. Most ROIs overlapped the passage reading network, which also
recruited additional areas. One notable finding is the widely distributed activity along the
lateral surfaces of the bilateral temporal lobes, which is not commonly found in single-
word reading tasks. These regions have been attributed to processing prosody (e.g.,
Wildgruber, Ackermann, Kreifelts & Ethofer, 2006), maintaining inferences (Lehman-
Blake & Tompkins, 2001), and sentence comprehension (e.g., Cutting et al., 2006), all
important components of reading connected text.
49
Figure 1a: Group contrast map of region-of-interest localizer: Phonological Lexical
Decision Task minus Line Judgment
L
L medial
R
aIFG
pIFG
pSTG
OT
SMA
R IFG
50
Figure 1b: Group contrast map of experimental task: Passage Reading minus Foreign
Font
L
R
51
Table 3: Properties of the functionally localized regions of interest
Region Mean Coordinates (SD) N t
x y z
Occipitotemporal region -39 (4) -48 (7) -14 (4) 35 5.45
Posterior superior temporal gyrus -53 (5) -40 (9) 4 (4) 32 4.10
Anterior inferior frontal gyrus -44 (5) 28 (6) 9 (6) 35 3.80
Posterior inferior frontal gyrus -46 (7) 7 (8) 23 (7) 34 4.35
Supplementary motor area -7 (3) 3 (7) 54 (4) 34 3.90
Right inferior frontal gyrus 34 (6) 22 (6) 6 (6) 34 1.24
52
Table 4: Regions appearing in group-level contrast of Passage Reading and Foreign Font
conditions
Voxels Peak X Peak Y Peak Z t p
R superior temporal gyrus 9621 51 -7 -8 10.61 0.0000000
R precentral gyrus 1986 57 -13 43 6.13 0.0000000
R angular gyrus 2119 57 -58 10 7.12 0.0000000
R anterior inferior frontal gyrus 214 54 26 19 5.37 0.0000010
R cerebellum 2914 21 -67 -35 8.35 0.0000000
R white matter near posterior ventricles 73 27 -46 10 5.02 0.0000040
R hippocampus 312 21 -13 -8 5.39 0.0000010
Medial prefrontal 601 -3 44 -2 5.01 0.0000040
L dorsal posterior cingulate 61 -3 -58 19 4.14 0.0000970
L supplementary motor area 316 -6 -4 61 7.11 0.0000000
L medial superior frontal gyrus 2853 -9 47 46 7.10 0.0000000
L posterior cingulate 241 -6 -52 7 4.90 0.0000060
L hippocampus 781 -21 -13 -11 6.47 0.0000000
L superior/middle temporal gyrus 32409 -48 -37 1 17.81 0.0000000
L central sulcus 114 -36 -22 22 4.88 0.0000070
L inferior/middle frontal gyrus 7453 -51 20 19 8.52 0.0000000
L precentral gyrus 1719 -48 -13 49 6.41 0.0000000
53
Effect of difficulty
In order to test for effects of difficulty, paired t-tests were performed on the
activation values during easy and difficult passage blocks for each ROI. Significantly
higher activation for easy compared to difficult passages was found in OT (t[34] = -8.22;
p<0.001) and SMA (t[33] = -2.06; p<0.05), and the opposite was true for aIFG (t[34] =
2.13; p<0.05). It is difficult to attribute these effects to a single process, as the two levels
of difficulty varied on multiple parameters, including word frequency, syntactic
complexity, and conceptual complexity. There was no effect of difficulty in the other
ROIs. Although the means are statistically equivalent, the distributions and relation to
reading may differ. For the remainder of the analyses, therefore, I did not collapse across
difficulty level.
Brain-behavior relationships
Correlations between behavioral measures and ROI activity during the passage
reading task, compared to the control task, are reported in Table 5. Select correlations are
also displayed as scatter plots in Figure 2. Performance on Nelson Denny correlated
positively with STG activity during both difficulty levels, and with SMA activity during
difficult passages only. (Two outliers are evident in the STG-ND scatterplot; a significant
correlation remains after excluding them: r = 0.48; p < 0.01). The rate component of the
GORT fluency measure correlated positively with OT activity, while the accuracy
component correlated with STG and aIFG activity during the difficult passages. The
GORT and ND measures also correlated with activity in the other ROIs, although not
significantly. Activity in OT appears to be related to timed reading measures, as it
correlated significantly with GORT Rate, TOWRE-PDE and Timed Word Attack.
54
Activity in STG during the difficult passages also correlated with single word reading
rate (TOWRE-W and TOWRE-PDE) but not the passage reading rate (GORT Rate). For
both difficulty levels, activity in pIFG correlated with Timed Word Attack, suggesting
involvement in decoding. Finally, activity in R IFG correlated negatively with RPN.
Behavioral performance during the scanner passage reading task, reflected as average
time on each passage, did not correlate with ROI activity. Therefore, the relationship
between reading ability and ROI activity cannot be attributed to faster readers’ increased
exposure to passages.
55
Table 5: Correlations between behavioral measures and ROI activity during easy and difficult passage reading
GORT
Rate
GORT
Accuracy
ND
Compre-
hension Word ID
TOWRE-
W
Word
Attack
TOWRE-
PDE
Timed
Word
Attack RPN
Cross
Out
Verbal
IQ
Spatial
IQ
Passages
RT
OT easy 0.38* 0.18 0.28 0.08 0.32 0.10 0.44** 0.35* 0.28 -0.21 0.24 -0.10 -0.31
difficult 0.37* 0.18 0.28 0.07 0.29 0.14 0.45** 0.38* 0.20 -0.20 0.26 -0.12 -0.32
pSTG easy 0.09 0.27 0.49** 0.12 0.33 0.17 0.31 0.19 0.21 -0.29 -0.11 -0.11 -0.14
difficult 0.19 0.37* 0.53** 0.17 0.36* 0.24 0.35* 0.26 0.27 -0.28 -0.08 -0.09 -0.21
aIFG easy 0.00 0.27 0.20 0.25 0.11 0.25 0.03 0.07 0.07 -0.12 -0.02 0.02 0.03
difficult 0.05 0.34* 0.30 0.28 0.17 0.28 0.09 0.13 0.11 -0.13 0.03 -0.01 -0.03
pIFG easy 0.27 0.21 0.20 -0.06 0.13 -0.07 0.18 0.38* -0.26 0.10 0.11 -0.13 -0.03
difficult 0.30 0.26 0.24 -0.03 0.14 -0.03 0.19 0.42* -0.23 0.14 0.13 -0.11 -0.08
SMA easy 0.11 0.32 0.33 0.27 0.12 0.18 0.12 0.19 0.01 0.13 0.13 -0.09 -0.02
difficult 0.15 0.32 0.39* 0.28 0.11 0.22 0.14 0.24 0.05 0.13 0.13 -0.09 -0.14
R IFG easy 0.20 0.11 0.16 0.05 -0.15 -0.03 0.01 0.19 -0.35* 0.26 0.17 -0.01 -0.07
difficult 0.18 0.10 0.12 0.03 -0.16 -0.02 -0.03 0.14 -0.31 0.30 0.14 -0.02 -0.11
Passages
RT -0.45** -0.26 -0.48** -0.25 -0.14 -0.36* -0.23 -0.28 -0.19 0.18 -0.34* 0.00 1.00
* p < 0.05; ** p < 0.01
56
Figure 2: Scatter plots of select ability-activity relationships
57
Many of the brain-behavior correlations were overlapping, so the relationships
were further analyzed using regression analyses (Table 6). Both OT and pSTG were
related to multiple decoding measures but different components of passage reading.
When considered simultaneously, neither region contributed unique variance to TOWRE-
PDE, suggesting that they behaved fairly similarly in terms of phonological decoding.
The relationships between GORT Rate and OT activity and between GORT Accuracy
and pSTG activity were both mediated by TOWRE-PDE. However, The contribution of
pSTG activity to Nelson Denny did not diminish after additionally considering TOWRE-
PDE, or either of the word recognition measures. Although the role of pSTG in reading
accuracy was partially mediated by decoding skill, the region seemed to make an
additional contribution to reading comprehension that cannot be explained by decoding
or lexical access.
Activation in SMA during language tasks is generally associated with articulation
preparation (Blacker, Byrnes, Mastaglia & Thickbroom, 2006), although Assaf and
colleagues (2006) assigned it a higher-level role in linguistic memory. Regressions
results, which indicate that the relationship between reading ability and SMA activity is
mediated by word recognition but not speeded measures, are more in line with the latter
interpretation.
58
Table 6: Linear regression analyses
At time of entry Simultaneous
Step R
2
∆R
2
F p t p
Dependent Variable: GORT Rate
1 OT 0.14 -- 5.44 0.026 -- --
1 TOWRE-PDE 0.39 -- 21.19 0.000 3.74 0.001
2 OT 0.40 0.01 10.80 0.000 0.80 0.429
Dependent Variable: GORT Accuracy
1 Verbal IQ 0.10 -- 3.31 0.079 2.16 0.039
2 pSTG difficult 0.26 0.16 4.96 0.014 2.46 0.020
1 Verbal IQ 0.10 -- 3.30 0.079 1.70 0.100
2 TOWRE-PDE 0.47 0.37 12.58 0.000 3.68 0.001
3 pSTG difficult 0.50 0.03 9.24 0.000 1.36 0.186
1 Verbal IQ 0.13 -- 4.77 0.036 2.23 0.033
2 aIFG difficult 0.23 0.10 4.86 0.014 2.11 0.043
Dependent Variable: Nelson Denny
1 Verbal IQ 0.17 -- 6.12 0.019 3.47 0.002
2 pSTG 0.48 0.31 13.38 0.000 4.16 0.000
1 Verbal IQ 0.17 -- 6.12 0.019 3.11 0.004
2 TOWRE-PDE 0.32 0.15 6.80 0.004 1.52 0.140
3 pSTG 0.52 0.20 10.09 0.000 3.42 0.002
1 Verbal IQ 0.17 -- 6.12 0.019 3.52 0.001
2 TOWRE-W 0.44 0.27 11.31 0.000 2.85 0.008
3 pSTG 0.60 0.16 13.83 0.000 3.32 0.002
1 Verbal IQ 0.17 -- 6.12 0.019 2.48 0.020
2 Word ID 0.26 0.09 5.13 0.012 1.47 0.154
3 pSTG 0.52 0.26 9.99 0.000 3.85 0.001
59
Table 6 (cont.): Linear regression analyses
At time of entry Simultaneous
Step R
2
∆R
2
F p t p
Dependent Variable: Nelson Denny
1 Verbal IQ 0.19 -- 7.49 0.01 2.58 0.015
2 SMA difficult 0.30 0.11 6.65 0.004 2.22 0.034
1 Verbal IQ 0.19 -- 7.49 0.010 2.32 0.028
2 TOWRE-PDE 0.33 0.14 7.65 0.002 2.43 0.021
3 SMA difficult 0.42 0.09 7.10 0.001 2.09 0.046
1 Verbal IQ 0.19 -- 7.49 0.010 2.86 0.008
2 TOWRE-W 0.45 0.26 12.60 0.000 3.82 0.001
3 SMA difficult 0.53 0.08 11.26 0.000 2.28 0.030
1 Verbal IQ 0.19 7.49 0.010 1.65 0.109
2 Word ID 0.28 0.09 6.15 0.006 1.57 0.127
3 SMA difficult 0.35 0.07 5.47 0.004 1.79 0.083
Dependent Variable: TOWRE-PDE
1 OT 0.20 -- 8.15 0.007 -- --
1 pSTG difficult 0.12 -- 4.26 0.048 -- --
1 OT 0.17 -- 6.25 0.018 1.82 0.080
2 pSTG difficult 0.21 0.04 3.94 0.031 1.24 0.227
60
Correlations relevant to the phonological task are reported in Table 7. Activation
measures reflect activity differences between the PLDT and control conditions.
Significant, positive correlations were found between aIFG activity and a number of the
reading measures and between OT activity and TOWRE-W. The different correlation
patterns imply that choice of reading task has a large impact on the resulting brain-
behavior relationships. Overall, the passages task yielded more significant correlations
that would not have been detected if only the single-word localizer task had been
analyzed.
DISCUSSION
The aim of the current study was to investigate the relationship between adult
phonological processing skill and reading ability in behavior and neural activity. Positive
correlations between performance on pseudoword reading tasks and both silent and oral
passage reading measures indicated that, even beyond reading development, decoding
ability is a significant predictor of reading ability. The underlying neural basis of the
decoding process provides further support for this relationship; reading skill modulated
fMRI activity levels in the phonological network during natural passage reading. The
positive correlations between reading ability and activity in the posterior reading system
replicated and provided ecological validity to previous reports. Additional positive
correlations draw more attention to the less frequently cited, but equally important,
frontal areas.
61 Table 7: Correlations between behavioral measures and ROI activity during PLDT task
* p < 0.05; ** p < 0.01
GORT
Rate
GORT
Accuracy
ND
Compre-
hension Word ID
TOWRE-
W
Word
Attack
Timed
Word
Attack
TOWRE-
PDE RPN
Cross
Out
Verbal
IQ
Spatial
IQ
OT 0.22 0.17 0.11 0.06 0.37* 0.14 0.09 0.27 0.30 -0.13 0.08 -0.03
pSTG 0.10 0.22 0.24 0.11 0.21 -0.12 0.11 0.14 0.06 0.17 0.02 0.10
aIFG 0.25 0.42* 0.50** 0.34* 0.41* 0.48** 0.22 0.25 0.20 -0.19 0.06 0.17
pIFG 0.18 0.11 -0.01 -0.19 0.11 0.07 0.01 0.17 -0.07 0.00 0.06 -0.08
SMA -0.01 0.03 0.27 0.01 0.15 0.17 0.06 0.13 0.06 -0.15 0.00 0.01
R IFG -0.11 -0.09 0.13 -0.30 -0.15 0.01 -0.22 -0.26 -0.21 0.02 -0.01 0.25
62
The first set of findings replicated previous reports of a significant relationship
between behaviorally measured decoding ability and reading ability in adults (Stanovich
& West, 1989; Bell & Perfetti, 1994; Dietrich & Brady, 2001; Watson & Miller, 1993).
Significant positive correlations between decoding measures and the passage reading
measures within a sample of mostly average to above-average readers indicates that the
relationship extends beyond reading development and disability and into the skilled, adult
range. In addition, the reading measures were strongly intercorrelated; performance on
both timed and untimed single-word identification tests was highly related to GORT and
Nelson Denny, as was the verbal IQ estimate. Significant contributions from basic verbal
and reading abilities to more global reading ability are consistent with a vast literature on
reading development, disability, and adult variability (e.g., Rayner et al., 2001).
The second set of findings confirmed the relationship between adult phonological
processing skill and reading ability at the level of brain activity. Activity in regions of
interest, which were individually localized using a decoding task, was generally
positively associated with passage reading components. Significant correlations were
found between ND comprehension and STG and SMA, between GORT Accuracy and
aIFG and STG, and between GORT Rate and OT. This set of findings is a direct
indication that neural indices of phonological processing are positively related to reading
ability.
It is reasonable to assume that more than one aspect of phonological processing is
captured by activation across the phonological network, so it follows that different
regions of the network contribute to different components of reading ability. A broad
range of more basic reading and verbal tasks were included to explore these specific
63
contributions. The correlation and regression analyses revealed a particularly significant
role for pSTG in reading ability. Its relationships with GORT Accuracy, ND, and
TOWRE-PDE are strong and may be partly shared, indicating that activation in pSTG is
capturing some form of phonological processing that influences reading ability. The
absence of a pSTG-GORT Rate correlation could be due to the role of pSTG in decoding,
specifically, on which GORT Accuracy may rely more. The findings confirm previous
claims that pSTG activity reflects phonological processing (Démonet et al., 1992; Graves
et al., 2007; Levelt et al., 1998; Majerus et al., 2005; vanAtteveldt et al., 2004), as well as
our hypothesis that the variation in activity with skill is the neural basis for the
relationship between phonological processing skill and reading ability. However, its
contribution to reading comprehension does not fully overlap with that from TOWRE-
PDE. Its activity explains an additional portion of the variance, beyond decoding ability.
Possible candidates for the additional influential subcomponent served by STG include
phonological access at the whole-word (lexical) level (Graves, Grabowski, Mehta &
Gordon, 2008), or even semantic access (Ruff, Blumstein, Myers & Hutchison, 2008).
Entering into regression analyses the available measures of lexical access (Word ID and
TOWRE-W) did not diminish the contribution of pSTG, so the additional subcomponent
remains elusive. Future studies are needed to further clarify the role of pSTG beyond the
implications from the current descriptive results. For the present aims, the significant
correlation between reading ability and pSTG activity during a reading task is sufficient
evidence for the role of phonological processing in skilled reading at the neural level.
Although they are related to different components of passage reading, pSTG and
OT contribute overlapping variance to speeded decoding. The regression analyses
64
demonstrated that OT does not independently contribute to GORT Rate but is mediated
by TOWRE-PDE. A previous study which manipulated orthographic and phonological
familiarity of words found that OT was sensitive to orthographic, but not phonological,
familiarity (Bruno, Zumberge, Goldman, Lu & Manis, 2008). It is puzzling, then, that OT
not only correlated with decoding measures but also behaved similarly to pSTG, a
classically phonological region. Automatic activation to legal letter strings could be
responsible for the appearance of OT in the decoding task contrast—even subliminal
presentation of words has been shown to affect activation in OT (Dehaene, Le Clec’H,
Poline, Le Bihan & Cohen, 2002)—and it may relate to skill simply as a function of
reading rate. Its activity is positively associated with most of the measures that tap into
reading rate, even scanner reading rate, although not significantly. Modulation of OT
activation by reading rate fits with Shaywitz and colleagues’ claim that OT underlies
rapid word recognition. The involvement of reading rate and phonological decoding
might dissociate OT from pSTG but not be robustly different enough to be detected by
TOWRE-PDE.
The direction of the difficulty effects observed in OT was also surprising but can
possibly be understood in terms of reading rate. One could interpret the higher activity
during the easy passages, relative to the difficult passages, as evidence that OT is
specialized for highly familiar words. However, previous findings of an inverse
relationship between activation level and frequency makes this interpretation unlikely. A
different account of OT activity could explain both instances of higher activity. Easy
passages are read more quickly than difficult passages, leading to exposure to more
words. In single-word tasks, increased processing time on low-frequency words can be
65
considered more exposure; after a high-frequency word is fully processed, attention may
shift away from the stimulus. Event-related potential findings from Rosazza, Cai, Minati,
Paulignan & Nazir (2009) suggest that pseudowords require sustained serial attention.
Increased exposure, at a low-level of visual processing, may be a mediator of the
relationship between OT activity and reading rate.
The contributions of aIFG to the accuracy component of reading ability and of
pIFG to a speeded decoding measure are consistent with the hypothesis that Broca’s area
is responsible for sublexical processes like grapheme-phoneme conversion or phoneme
assembly (e.g., Joubert et al., 2004). This is not the first study reporting a positive
relationship between frontal activation and reading skill (see Maisog et al., 2008), but it
may be the first to show the relationship using a wide range of ability. Many previous
studies have demonstrated the opposite effect. In Brunswick et al. (1999), RD adults
showed increased activation in IFG; in Seghier, Lee, Shofield, Ellis & Price (2008), so-
called “slow lexical readers” relied more on anterior frontal regions and among others;
and in Prat et al. (2007), poor reading capacity was associated with great activity in
bilateral middle frontal gyrus. (The coordinates reported as middle frontal gyrus were
very close to those for pIFG in the current study; naming conventions vary, but
coordinates reported to be posterior or dorsal IFG and middle frontal gyrus often
overlap.) The authors interpreted this finding as evidence for skilled processing being
achieved by efficient use of neural resources. In other words, better performance is
associated with less activity in task-relevant brain regions. Inconsistent with this
“efficiency hypothesis” is the evidence that IFG activity increases with age. In a cross-
sectional study of subjects aged 6-22, one marker of reading development was an age-
66
related increase in left IFG activity (Turkeltaub et al., 2003). Some developmental
theories posit that age- and skill-related development follow parallel neural trajectories
(e.g., Johnson, 2001), so it is possible that the observed increase in aIFG activity with
skill is a continuation of the development of reading skill. The inconsistency of skill-
related findings in IFG could be due to task effects. The studies by Brunswick et al. and
Seghier et al. employed single-word reading tasks. Prat et al. used a sentence
comprehension task, yet the insertion of a comprehension probe after each sentence
imposed a limit on the task’s ecological validity. The findings could have been
confounded by the additional processing required to make a decision about the validity of
the sentence. The use of a more natural reading task in the present study makes for a
more ecologically valid interpretation.
Task effects were also evident in the present study, although the two tasks
produced correlations in the same direction. The study was not primarily designed to
speak to task effects; the decoding localizer was simply a means to an end, and the
different control tasks limit the validity of any interpretation about different processes
involved in word and passage reading. However, the results are reported for the sake of
completeness. Most of the brain-behavior correlations observed in the passage reading
context failed to generalize to the PLDT. Activity in aIFG was positively correlated with
multiple reading measures in the PLDT. Activaty in pSTG, on the other hand, showed no
relationship with skill. All the ROIs were active during PLDT, compared to the control
task, in the majority of the subjects (in keeping with its role as a localizer), but most of
the skill-related variation did not appear until they were engaged in a natural reading task.
67
The increased number of significant effects speaks to the value of using a natural, more
complex task to reveal skill patterns in skilled readers.
The conflicting correlation patterns can be reconciled by considering the
possibility that the ROIs serve different purposes in the face of different task demands.
For example, depending on the particular manipulation used, different studies have
reached drastically different conclusions concerning the role of aIFG in reading. Gold et
al. (2005) observed through repetition priming that both posterior and anterior aspects of
IFG are engaged in both phonological and semantic processing. Syntactic processing is
also on the list of functions subserved by aIFG. A recent review concluded that the
process that has received the cleanest empirical support as being localized to IFG is
syntactic movement (Grodzinsky and Santi, 2008). This process is engaged when reading
sentences in which elements have been displaced from their canonical positions, as in
questions. The difficulty effect observed in aIFG in the present study, in which difficult
passages elicited greater activation than easy passages, is actually consistent with this
functional interpretation. Greater activation might reflect enhanced syntactic processing
in the face of the difficult passages’ more complex sentences. However, the more often
cited phonological function of IFG is also consistent with the difficulty effect. Greater
activation during difficult passages might reflect increased stress on decoding processes
in the face of longer and less frequent and words. Considering the results from the
passage task, alone, it was not possible to claim that aIFG is more involved in semantic
processing, fitting with previous claims (e.g., Devlin et al., 2003). A comparison between
the correlation patterns from the passage task and the decoding localizer, however,
suggests that aIFG could be involved in both semantic and phonological processing.
68
An extension to the multi-functional ROIs account concerns strategy differences.
The efficiency hypothesis—that superior performance is achieved using fewer neuronal
resources in a region—is somewhat limited in its scope because it focuses on the
influence of individual areas on cognitive ability. Recently, cognitive neuroscientists are
rejecting hypotheses like this, which pin cognitive processes to individual regions, in
favor of a more system-centered view. The best predictor of skill could be the pattern of
activity across an entire network—in other words, strategy. This may be why the
relationship between pSTG activity and reading ability appears to depend on the specific
task demands. During the PLDT, which constrains all subjects into the same strategy—
decoding—activation does not depend on skill. In the unconstrained passage-reading
task, where readers of different skill levels may rely more or less heavily on different
processes to achieve the same task, different strategies may lead to different activation
patterns. According to this interpretation of system dynamics, our present results indicate
that highly skilled readers engage a phonological strategy to a greater degree to
accomplish the task of natural, everyday reading.
One of the study’s limitation is the poor average performance on the
comprehension quiz given at the end of the scan. The questions on the quiz were relevant
to the content presented within the first few stimuli of each block, so it is not likely that
subjects failed to answer them because they never read the relevant part of the story.
More likely is the possibility that the subjects simply forgot the content and were unable
to use specific recall to answer the open-ended questions. The delay between the passage
task and the quiz included 15-20 minutes of structural scanning, during which time many
subjects reported to have fallen asleep. Anecdotally, many subjects also complained of
69
forgetfulness while taking the quiz. This limitation points out an inherent disadvantage of
scanner tasks that are designed to be natural; eye-tracking data could have avoided this
issue. However, the significant effects of passage difficulty on reaction time and
comprehension question accuracy do provide some assurance that subjects were reading
the passages.
Another limitation is the specificity of the relationship that was investigated. It is
likely that the neural substrates of other forms of phonological processing also make
unique contributions to reading ability, but the current study can only speak to
phonological decoding. Future investigations should include both behavioral and fMRI
tasks designed to measure phonemic awareness, verbal short-term memory, and other
phonological processes.
Future directions also include further testing of the idea that it may be variations
in strategy, rather than skill, per se, that drive relationships between network activity and
high-level cognitive abilities. Functional and structural connectivity analyses could speak
to the coherence within networks.
70
CHAPTER 3: READING ABILITY AND FUNCTIONAL CONNECTIVITY WITHIN THE
PHONOLOGICAL PROCESSING NETWORK
Connectivity is a new and informative approach to studying the interactions of
reading regions, as well as the relationship between the interactions and behavior. By
measuring the coupling of changes in imaging signals, previous functional connectivity
studies of reading have established connections between reading-related regions,
particularly IFG and posterior regions (Bokde et al., 2001). Strengths of connections
between regions have also been related to individual reading skills. Increased
connectivity between IFG and a temporoparietal region, for example, has been associated
with increased reading ability (Hampson et al., 2006). The present study aimed to assess
the contribution of connectivity within one particular network, the phonological
processing network, on adult reading ability.
Functional connectivity findings for skilled adult readers consistently show
connections between widely distributed classical language areas. Activity in left frontal
regions is correlated with activity in many regions, including occipito-temporal junction
(Stanberry et al., 2006), which is considered one of the first stages in visual word
recognition (e.g., Cohen et al., 2000); angular and supramarginal gyri in parietal cortex
(Rumsey et al., 1997; Hampson et al., 2006); and pSTG, both at rest and during tasks like
auditory language comprehension or pseudoword reading (Hampson, Peterson,
Skudlarski, Gatenby & Gore, 2002; Joubert et al., 2004). Effective connectivity studies
of reading, which indicate direction of influences, have indicated that the left occipito-
temporal junction is the sensory input source and IFG and STG are receiving units
71
(Bullmore et al., 2000; Mechelli et al., 2002). Mechelli et al. (2005) noted that different
portions of OT may drive different regions, depending on the task demands; whereas an
increase in IFG-pars triangularis activation during exception word reading was associated
with influences from anterior OT, an increase in left dorsal premotor cortex during
pseudoword reading was associated with posterior OT. Different networks contribute to
different processes, depending on the task requirements.
Functionally connectivity also varies with reading skill. Connectivity-behavior
analysis is a technique that compares skill measures to interregional connections across
time. Evidence of connectivity-behavior relationships in reading comes from the domains
of development, disability, and variation in skilled reading. In a cross-sectional
developmental study, Bitan and colleagues (2007) compared connectivity across age
groups. The coupling of IFG with other regions increased with age, while STG coupling
decreased. In adult readers, profiles of connectivity differ between reading-disabled and
nonimpaired individuals. Shaywitz et al. (2003) found that while nonimpaired readers
were characterized by strong connections between the occipito-temporal region and left
IFG, poor readers did not show this pattern. Instead, the activity in the occipito-temporal
region was correlated with activity in right frontal cortex. Reading disability has also
been associated with weaker or missing connections between IFG and right middle and
inferior occipital gyri (Stanberry et al., 2006), as well as between AG and extrastriate
occipital and temporal cortex (Horwitz et al., 1998; Pugh et al., 2000). Findings from
Pugh et al. (2000) suggest that disruptions in connections with AG are only evident
during tasks requiring phonological assembly. Even if poorer readers are activating the
72
same regions, more tenuous temporal connections between them may account for their
deficits in fluency.
Relatively fewer studies have investigated normal variation within typical patterns
of connectivity. Hampson and colleagues (2006) hypothesized that correlations between
activity in language areas would be related to reading skill in healthy, nonimpaired
adults. Results from a sentence-reading fMRI task indicated that activity in a seed region,
IFG, was correlated with a region spanning the angular gyrus and the posterior aspect of
the middle temporal gyrus that the authors referred to as Brodmann Area (BA) 39.
Correlation and group-difference analyses revealed that the strength of connection
between IFG and BA39 varied with reading skill among the 19 subjects. Higher
correlations were related to higher performance on all four reading measures. The authors
propose that the correlation could arise because “better readers directly access a better
lexicon, while poor readers rely more on orthographic to phonological decoding.” The
finding can be interpreted in many ways, since a limitation of the study was that the
sentence reading task requires many different levels of processing, and it was unclear
what processes were supported by the specific regions of interest. Can weaker
connectivity at the low end of the nonimpaired scale be attributed to differences in
phonological processing, as would be predicted by Pugh et al. (2000) if reading disability
represents the lowest end of the same distribution?
Focusing on a wide range of skill, the present study aimed to characterize reading
ability in terms of communication between regions associated with phonological
processing. Network connectivity appears to play an important role in adult reading
ability, but more evidence is needed from a wider range of adult readers, whose ability is
73
determined by a broader battery of measures, and while they are engaged in a natural,
everyday reading task. Based on previous findings, I hypothesized that coherence
between regions within the decoding network would be positively associated with
reading ability.
METHOD
The participants, behavioral measures, and functional MRI procedure were
identical to those in Study 1. The same six regions of interest (ROIs) were localized in
each subject, and passage-control activation difference scores were computed from each
ROI. The methods up to here are briefly described again below. (For details, please refer
to Study 1.)
Participants
The same thirty-five young adults from Study 1 participated in both a behavioral
testing session and an MRI session. All participants were right-handed, monolingual
native-English-speakers with normal or corrected-to-normal vision and a negative history
of neurological abnormalities.
Behavioral Testing
Estimates of verbal and spatial IQ were measured using Woodcock Cognitive
Abilities subtests Verbal Comprehension and Spatial Relations. Single-word reading
ability was assessed using the Word Identification and Word Attack subtests of the
Woodcock Johnson III Tests of Achievement. The Test of One-Word Reading Efficiency
(TOWRE) – Sight Words (W) and Phonemic Decoding Efficiency (PDE) subtests – also
measured word and pseudoword reading accuracy, respectively, but under time
74
constraints. A second timed pseudoword reading test, Timed Word Attack, was also
administered. Two passage reading tests were administered to measure oral reading
fluency and silent reading comprehension, Gray Oral Reading Test (GORT) and Nelson
Denny (ND), respectively. Performance on each test was reported as two separate
measures: GORT Rate and Accuracy, and ND Reading Rate (ND-RR) and
Comprehension. Rapid Picture Naming and Cross Out were also included in the battery
to address possible effects of speed on reading ability.
Imaging Procedure
Subjects first performed a phonological lexical decision task (PLDT), which was
designed to localize regions of interest (ROIs) that are involved in decoding. Each trial in
the PLDT condition consisted of a 2000 ms presentation of a single word-like stimulus,
to which the participant was instructed to respond to the question “Does it sound like a
real word?” Following each stimulus was a fixation cross for 300 ms and a blank screen
for 200 ms. Items were divided into two conditions, pseudohomophones (PH; e.g.,
‘rane’), where the correct response is ‘yes’, and pseudowords (PW; e.g., ‘brap’), which
deserve a ‘no’ response. By presenting only PH and PW, and no familiar words, the task
required that participants engage in a decoding strategy throughout the entire task. In the
control condition, which appeared in alternating blocks, participants judged whether
individually presented line patterns were symmetrical.
The second task consisted of successive portions of passages that the participants
read silently and at their own pace. The passages were culled from forms of GORT that
were not used in the behavioral assessment. Participants were instructed to read each
passage at a natural pace and remember the stories, because they would be given a
75
comprehension quiz after the scan. A button press delivered the next portion of the
current passage or the first portion of a new passage until the 30-second block was over,
at which time a control-condition block began. In the control condition, participants were
presented with blocks of a foreign font and instructed to scan the lines, as if reading, and
press the button when they saw a target stimulus, whose identity was consistent
throughout the task and could appear in any position. No subject was familiar with the
font (Sorawin).
Items in the first task were presented in a block design, and the order of wordlike
stimulus types was randomized within each of the 8 PLDT blocks. Each 40-second block
contained 16 trials, and each of 2 runs contained 8 blocks total. The second task was also
presented in a block design, with eight 30-second blocks per run, and was repeated twice.
A fixation cross was presented for 12 s at the beginning and end of each run. Throughout
the scan, participants view the stimuli projected on a screen behind their head through an
adjustable mirror mounted on the head coil. A fiber-optics button box was provided for
responses. Stimulus display was programmed in MATLAB (The MathWorks, Natick,
MA) and Psychtoolbox (Brainard, 1997). Following the functional MRI tasks, structural
MRI data was acquired for co-registration purposes.
Image acquisition
Functional and structural imaging was performed on a Siemens Magnetom Trio 3
Tesla MRI unit (Siemens Medical Solutions, Malvern, PA) using a 12 channel head coil.
Earplugs and sound dampening headphones were worn by the participants for protection
from noise. Foam padding placed between the participants’ necks and head cradle was
used to minimize head movement.
76
High resolution structural images were acquired using a T1-weighted MPRAGE
sequence (FoV 256mm x 256mm; TI 800ms; TR 2530ms; TE 3.09 ms; Flip angle 10; 208
coronal slices), resulting in 1mm
3
isotropic voxels that covered the entire brain and part
of the cerebellum. Functional image parameters during the T2*-weighted echo-planar
imaging sequence (FoV 224 mm; TR 2000 ms; TE 32 ms; 28 axial slices) yielded 3.5 x
3.5 x 4 mm voxels with no over-sampling.
Image Analysis
Data were subject to online 3D PACE motion correction during acquisition.
BrainVoyager QX 1.10.3 (Brain Innovation, Maastricht, the Netherlands) was used to
preprocess the data. The functional data images were aligned to the last run with a rigid
body transformation and subjected to additional motion correction using trilinear
interpolation. Data were spatially smoothed using a 4 mm FWHM Gaussian kernel.
Temporal filtering included both linear trend removal and a high pass filter with a cutoff
of three cycles per timecourse. Slice scan time correction was performed with sinc
interpolation. Data were then aligned, both automatically and manually, to unnormalized
isotropic structural images. Then the structural and coregistered functional data were
normalized into a standard stereotaxic space (Talairach & Tournoux, 1988).
Individual participants’ decoding networks were defined on the basis of a group
analysis of the PLDT minus barcode contrast. The subsequent search for individual ROIs
was constrained based on the average center of gravity of each ROI found in the group
contrast map. In order to maintain a consistent ROI size across participants, 125mm
3
cubes were created around the peak of each ROI.
77
Activation during the passage task was analyzed within the individually defined
ROIs. Multistudy General Linear Modeling performed on each ROI in each subject
produced normalized “beta” values, representing average intensity for each condition.
The difference between passage and control condition “beta” values was computed,
resulting in a single value for each ROI in each subject. (At this point, the methods
diverge from those from Study 1.) The values for each ROI were then correlated with
values for all other ROIs across subjects. Regions-of-interest were drawn onto a brain
model, with lines connecting ROIs that were significantly correlated. The lines
represented pathways (direct or indirect) between regions within the decoding network
that demonstrated a functional relationship during passage reading. Only these pairs were
examined in the following analysis.
Preprocessed timecourse data was extracted from each ROI. The entire
timecourse was offset by 3 TRs (6 seconds) to correct for the delay in the hemodynamic
response function. Only time points corresponding to the passage condition were
selected, following the assumption that any correlations between regions’ activity during
the non-reading conditions are not specific to reading. After the control condition and
initial and final fixation conditions were excluded, the passage condition time points were
concatenated across the eight blocks and two runs, yielding 128 points per ROI. Within-
subject correlations were then computed between the BOLD signals of each pair of ROIs.
The resulting temporal correlation values were then transformed to z values and
correlated with the behavioral measures to test for connectivity-behavior relationships.
78
RESULTS
Interregion correlations across subjects
Many regions showed reliable relationships with other regions in the network,
indicating that average activation levels covary systematically across subjects (see Table
8). Activity in pSTG was positively correlated with activity in many regions: OT, aIFG
and SMA. In addition, activity in SMA was related to activity in R IFG. Figure 3
provides an illustration of the network, with lines drawn between regions that are
significantly functionally correlated.
Interregion correlations across time, related to reading skill
To determine whether the strength of interregion connections varied with skill,
temporal correlations were computed in each subject within each pair of regions
connected in Figure 1. These interregion correlation values were then correlated with the
behavioral measures (see Table 9). A significant positive relationship was found between
Word ID and the strength of connection between pSTG and aIFG, implying that better
word recognition ability is associated with stronger coherence between two regions in the
phonological network during passage reading.
79
Table 8: Correlations between pairs of ROIs (average activation during Passage Reading)
OT pSTG aIFG pIFG SMA R IFG
OT --
pSTG 0.39* --
aIFG 0.07 0.42* --
pIFG 0.20 0.11 0.22 --
SMA 0.00 0.38* 0.25 0.24 --
R IFG 0.04 -0.04 0.09 0.23 0.44* --
80
Figure 3: Functional Connectivity Illustration; lines represent significant correlations
between average ROI activation during Passage Reading
81 Table 9: Correlations between behavioral measures and strength of interregion temporal correlations
GORT
Rate
GORT
Accuracy
Nelson
Denny
Word ID
TOWRE
W
Word
Attack
TOWRE
PW
RPN
Cross
Out
Verbal IQ Spatial IQ
pSTG-OT 0.09 -0.08 -0.31 -0.05 -0.01 -0.08 -0.01 0.03 0.20 0.00 -0.01
pSTG-aIFG 0.24 0.25 0.14 0.37* 0.16 0.35 0.21 0.02 0.21 0.17 0.02
pSTG-SMA -0.01 -0.15 -0.33 -0.11 -0.35 0.01 -0.23 -0.20 0.28 0.00 0.14
R IFG-SMA 0.04 -0.12 -0.20 -0.29 -0.17 -0.11 -0.11 -0.15 0.04 -0.16 0.02
* p < 0.05; ** p < 0.01
82
DISCUSSION
Regions associated with phonological processing were functionally connected
during passage reading, with pSTG representing a central node in the network.
Furthermore, the coupling strength of one of the connections, that between pSTG and
aIFG, varied with performance on a word recognition task. The findings provide systems-
level evidence for the relationship between phonological processing and adult reading
ability.
The first set of findings, the correlations between average ROI activation levels,
are in line with previous literature. Hampson and colleagues (2002) established temporal
correlations between the fluctuations in IFG and pSTG activity during a linguistic task.
The regions were also linked during rest, although less strongly. The connection has also
been reported by Joubert et al. (2004) during a task that required single nonword and low-
frequency regular word reading. The authors interpreted the connection as contributing to
a phonologically-based form of reading. Their interpretation was based on the task that
was used, because pseudowords and low-frequency words generally require phonological
decoding, and also on the phonological roles assigned to the pSTG and IFG by previous
literature. The current results can make the same claim without calling on external
evidence; the regions’ connectivity can be attributed to phonological processing because
they are the same regions identified in the same individuals as involved in phonological
processing. In addition, they provide evidence that the regions are coactive not only
during single-word reading or story listening, but also during natural passage reading.
In addition to aIFG, pSTG was moderately linked to two other ROIs, OT and
SMA. The centrality of pSTG in the reading network is not often discussed; angular
83
gyrus (AG) is a more likely candidate because of its role in classic models of reading.
Lesion evidence led Geschwind (1965) to describe the AG as a mediator between
orthraphic processing units in visual areas and phonological processing units in temporal
sites (Geschwind, 1965). Recent neuroimaging has confirmed that role. Horwitz et al.
(1998) found correlational links between AG and extrastriate occipital and temporal
regions during single-word reading in nonimpaired men, but not in dyslexic men. Pugh et
al. (2000) replicated the group difference between AG and posterior regions, but only
during phonological tasks. This provided evidence for the specifically phonological
nature of functional disruptions in reading disability.
The present study was not able to extend the finding into the more skilled range
because AG did not appear in the localizer task. It follows that AG is not directly
involved in decoding but may have a mediating role between pSTG and OT, which
would conform to classic reading models, as well as to the correlation between pSTG and
OT average activation. (Functional connectivity methods leave open the possibility that
regions coactivate as a result of an indirect anatomical connection.) Alternatively, what
Pugh and colleagues referred to as AG might overlap with what the present study calls
pSTG, thereby providing converging evidence for the centrality of the “temporoparietal
area” in the reading network. The present findings nevertheless make a strong case for
the role of the phonological processing network in reading.
Another result that contributed to that interpretation was the correlation between
two additional phonological network ROIs, SMA and R IFG. This is the first study, of
which I am aware, reporting such a connection. In fact, discussion of right hemisphere
regions is relatively rare in the reading literature. The few studies that do devote time to
84
interpreting right-hemisphere involvement do not even agree on the role of R IFG. It may
be important in processing emotional prosodic intonation (Rota et al., 2009; Glasser &
Rilling, 2008), but it has also been linked to semantics (Taylor & Regard, 2003;
Il’yuchenok, Sysoeva & Ivanitskii, 2008), lexical memory access (Fay, Isingrini, Ragot &
Pouthas, 2005), and processing contextually relevant perceptual information (Lincoln,
Long, Swick, Larsen & Baynes, 2008). It should be noted that these processes may
overlap, and the functional relationship between R IFG and SMA may reflect the
contribution of any of these processes to silent reading passages. However, because R
IFG appeared in the localizer task, the present findings attribute its involvement to
phonological processing.
The second set of findings speaks to the influence of phonological network
connectivity on adult reading ability. Network activity can be assumed to reflect
phonological processing during passage reading, and covariance between at least two of
the network’s regions, pSTG and aIFG, is related to skill. One of the characteristics of
skilled readers, therefore, is improved communication between regions in the
phonological network. The connectivity-behavior relationship was only significant in the
context of Word ID, which measures single word recognition. It is unclear why the
connection was not significantly related to any of the decoding or passage reading
measures. Passage reading and nonword decoding may rely on the same connections as
single word reading, without similar variations with skill. Studies with larger sample
sizes and more reading measures may be able to address the different effects.
What mechanism mediates the relationship between strength of STG-aIFG
connectivity and word recognition skill? Stronger correlations indicate that the
85
timecourses of activity in the two regions were more tightly coupled. This might reflect
stronger synaptic connections either at the axons’ endpoints (the regions) or along the
way (via indirect routes). More frequent neuronal communication between pSTG and
aIFG may be a consequence of more time spent reading. According to the classic
Hebbian view, synapses are strengthened following regular coactivation of the two
component neurons. A history of Hebbian coactivation would account for the detection of
cognitive-task-specific functional correlations being detected during the resting-state
(Dosenbach, Fair, Cohen, Schlaggar & Petersen, 2008). Brain connectivity in superior
readers may be inherently stronger, as a result of a lifetime of experience.
The following technical issue should be addressed. In the temporal connectivity
analyses, only the passage condition was examined; timepoints during rest and control
were excluded because any interregion correlations during these times (with the
exception of the Hebbian-driven reading connections, which are always present) should
be interpreted as noise. It seems contradictory, then, that the variables in the first set of
correlations were passage-control difference scores. This was in fitting with the majority
of fMRI studies, which subtract control-related activity to focus solely on activity
associated with the cognitive process of interest. However, there is always a possibility
that it is the control condition activity that is driving the effects. The present study is
especially vulnerable to this confound because of the novel and complex control task. I
therefore reanalyzed the first set of correlations using absolute passage activity, rather
than the difference score. The pattern of correlations were slightly different. Many of the
same pairs still correlated (pSTG and aIFG, and pSTG and SMA), with the addition of a
few others. However, when considering these pairs during the second, temporal
86
correlation analysis, the results were identical to the first analysis. The only significant
correlation was between the pSTG-aIFG connection and Word ID. In this case, the choice
between contrasted and absolute activity did not impact the primary finding.
Some critiques of the functional connectivity approach have pointed out the
possibility that correlated BOLD signal is driven by the “blood” aspect of the fMRI
signal. Nearby cortical regions are often nourished by the same blood vessels. Some
signal fluctuations are a result of capillary blood flow that is not dependent on neural
activity. Many technical papers have addressed this possibility (e.g., Biswal, Van Kylen
& Hyde, 1997) and arrived at the conclusion that the coupling of BOLD signals more
closely matches actual functional connectivity maps, relative to coupling of blood flow
signals. Another point to the contrary is the lack of correlation between two nearby
regions in the current study, anterior and posterior IFG.
It is unclear from the results whether it is the forward or backward volley of
information between pSTG and aIFG that impacts reading skill. An inherent limitation of
functional connectivity is the inability to address directionality. Effective connectivity
methods, which do indicate direction of connection, should also be considered. In the
future, similar studies should be performed using techniques with higher temporal
resolution, such as MEG. Additionally, structural connectivity methods should also be
applied, in order to test for structural validation of the relationship between phonological
network connectivity and adult reading ability.
87
CHAPTER 4: READING ABILITY AND WHITE MATTER MICROSTRUCTURE IN MAJOR
READING TRACTS
Communication between brain regions is critical to achieving the complex task of
reading. The “highways” of communication are bundles of white matter tracts comprised
of thousands of myelinated axons, or fibers, projecting between discrete brain regions.
Early post-mortem dissection studies identified specific tracts as important to language
because they connected classical language regions. Today, diffusion tensor imaging
(DTI) allows for the virtual dissection of white matter in the living brain, enabling
investigations of individual differences in white matter tract microstructure. The current
study aims to relate phonological and reading abilities to variations in white matter
structures within the reading network.
Historically, the white matter tract that has received the most attention in
language research is the arcuate fasciculus, which projects from the superior temporal
gyrus to the frontal lobe as one of the fiber bundles in the superior longitudinal fasciculus
(SLF). It is thought to transfer language-related auditory information between Wernicke’s
and Broca’s areas (Dejerine, 1895). Lesions in the left arcuate often result in conduction
aphasia, or impaired repetition (Lichtheim, 1885). Spoken language is not always
affected by disruptions to this pathway, however; in a recent case study of a young girl
who underwent radiation treatment, neuroimaging revealed selective damage to left and
right arcuate fasciculus. As a result, she had severe reading difficulties but normal spoken
language. One process that is supported by the arcuate and possibly led to her reading
difficulties is phonological processing; the regions linked by the arcuate are involved in
88
phonological processing and have been shown to play an important role in reading ability
(Poldrack et al., 2001; Heim et al., 2003; Majerus et al., 2005; Joseph et al., 2001). This
and other case studies provide support for the hypothesis that loss of the arcuate can
cause pure alexia due to disconnection between dispersed reading regions (Rauschecker
et al., 2009).
DTI investigations using nondiseased populations suggest that the structure of
intact fiber tracts varies across individuals and can be related to specific cognitive skills.
The integrity of what matter tracts is most often measured by fractional anisotropy (FA),
or the nonrandom diffusion of water in the brain. A larger FA value within a specific
white matter region generally indicates that the fibers within that region are oriented in
the same direction. Increases in FA in high level association fibers like the arcuate
fasciculus throughout childhood and into early adulthood presumably reflect ongoing
maturation/myelination that may be related to skill development (e.g., Ashtari et al.,
2007).
Development of reading ability is associated most strongly with temporoparietal
white matter, including but not limited to the arcuate. Significant correlations have been
reported between FA in temporoparietal regions and children’s performance on word and
nonword reading, spelling, and rapid naming tests (Beaulieu et al., 2005; Deutsch et al.,
2005; Odegard, Farris, Ring, McColl & Black, 2009). The relationship has also been
established in adults varying in reading ability (Klingberg et al., 2000). Using a region of
interest technique, as opposed to the more common voxel-based analysis using
standardized brains, Niogi & McCandliss (2006) localized the skill-related white matter
differences to a specific temporoparietal tract, the left superior corona radiata (SCR). The
89
authors claimed that significant clusters identified by previous studies overlap with the
SCR. Unlike the nearby arcuate fasciculus, which travels along the anterior-posterior
axis, the SCR projects in an inferior-superior direction from the thalamus to sensory and
motor cortex. Despite the less anatomically obvious reasons for its important role in
reading, the left SCR appears to be associated with childhood and adult reading ability. A
number of other regions have received less consistent support for their role in reading
ability, including the anterior corona radiata (ACR; Qiu, Tan, Zhou & Khong, 2008), the
inferior longitudinal fasciculus (ILF; Steinbrink et al., 2008), the corpus collosum
(Dougherty et al., 2007) and various parts of the corticospinal tract (Beaulieu et al., 2005;
Bruno, in preparation).
The present study sought to replicate previous findings of skill-related variations
in the reading network’s white matter pathways within a relatively large sample of adults.
Correlations were computed between measures of reading ability and FA within regions
of interest that were segmented from individual brains using a semi-automated tracing
method. The Reproducible Objective Quantification Scheme (ROQS) allows for quick
and reliable analysis of individual white matter tracts and, unlike voxel-based analyses, is
not subject to gross morphological differences in brain anatomy (Niogi, Mukherjee &
McCandliss, 2007). It is particularly appropriate for studies focusing on specific,
hypothesis-driven regions. Based on the rapidly growing literature on regions implicated
in reading ability, I hypothesized positive correlations between reading ability and FA in
temporoparietal white matter regions, specifically SCR and the arcuate fasciculus.
90
METHODS
Participants
As part of two larger studies, 68 adults (43 female; ages 18 to 30) participated in a
behavioral testing session and an MRI session that included a DTI sequence. All
participants were right-handed, monolingual native-English-speakers with normal or
corrected-to-normal vision and a negative history of neurological abnormalities. The DTI
parameters used in the two different studies were identical. Only the tests that were
included in the behavioral testing batteries of both studies were analyzed, to preserve the
large sample size.
Behavioral Testing
Estimates of verbal and spatial IQ were measured using Woodcock Cognitive
Abilities subtests Verbal Comprehension and Spatial Relations. Single-word reading
ability was assessed using the Word ID and Word Attack subtests of the Woodcock-
Johnson Tests of Achievement-III, which measure participants’ ability to accurately read
increasingly difficult words and pseudowords, respectively. Test of One-Word Reading
Efficiency (TOWRE) Sight Words (SW) and Phonemic Decoding Efficiency (PDE)
subtests also measured word and pseudoword reading accuracy, respectively, but under
time constraints. A test of nonreading verbal skills, Rapid Picture Naming (RPN),
measured automatic word retrieval speed by instructing participants to quickly name
successive images of concrete, high frequency objects. A measure of passage reading
comprehension was also included in the test battery. Nelson Denny (ND) requires
participants to silently read passages and answer multiple-choice comprehension
questions until the 20-minute time limit is reached. Both reading rate (ND-RR),
91
corresponding to the line number the participant reaches after the first minute of reading
the first passage, and comprehension (ND-Comp), the total number of questions
answered correctly, were recorded.
Image Acquisition
MRI scanning was performed on a Siemens MAGNETOM Trio 3Tesla MRI unit
using a 12-channel head coil. Participants wore earplugs and sound dampening
headphones to minimize acoustic noise, and foam padding was used to minimize head
movement. DTI images were acquired from fifty 2.5mm-thick axial slices using a single-
shot EPI sequence (TE = 110 ms, TR = 7600 ms, resolution = 128 x 128, field of view =
230 x 230 mm, matrix size = 128 x 128 mm, pixel size = 1.8 mm). The diffusion tensor
acquired for each slice included one image without diffusion weighting (b = 0 s/mm2)
and diffusion weighted images along 64 directions (b = 1000 s/mm2).
Image Analysis
The images were preprocessed using FSL 4.0 (FMRIB Software Library, Center
for fMRI of the Brain, University of Oxford, UK). Each diffusion weighted image
underwent correction for effects of motion and eddy currents via an affine registration to
a reference volume (Behrens et al., 2003). The brain extraction tool was used to strip the
b-0 image of skull tissue to limit further analyses to only brain tissue (Smith, 2002). As a
final step, DTIFit was used to calculate diffusion tensors, from which principle diffusion
direction and fractional anisotropy were computed for each voxel (Behrens et al., 2003).
Average FA values were then extracted from anatomically guided regions of
interest in each subject’s nonstandardized brain using ROQS, a semi-automated
segmenting tool developed by Sumit Niogi (Niogi et al., 2007). The region of interest
92
analysis process requires the user to visually identify a region of interest on the diffusion
direction-encoded DTI color map and select a seed point within the boundaries of the
region. The program then initiates an algorithm that uses the direction information
encoded in the seed point to draw a boundary around all neighboring points that have the
same primary direction. This results in a data-defined, reproducible 2-D ROI that is
unique to each subject. Whereas manual tracings of ROIs by trained neuroradiologists are
time-consuming and vary across raters, this semiautomatic method is quicker and much
less variable in comparison, allowing for larger sample sizes and the use of non-expert
raters. In the current study, two raters who were blind to the identity of each brain
followed detailed instructions to identify and select ROIs. One rater analyzed 10 of the
brains twice, and a second rater analyzed 32 of the 68 once. Intraclass correlation
coefficients were then calculated using FA values from each analysis of each ROI to
assess interrater and intrarater reliability.
The 15 regions of interest are shown in Figure 4. Selection of the regions of
interest was guided by previous DTI studies that reported significant relationships
between specific tracts and reading ability. The regions are corpus collosum (CC),
superior corona radiata (SCR), anterior corona radiata (ACR), superior longitudinal
fasciculus (SLF; which includes the arcuate fasciculus), inferior longitudinal fasciculus
(ILF), posterior limb of the internal capsule (PLIC), inferior corticospinal tract (iCST),
and cerebral peduncle (CP). All but the CC were selected bilaterally.
93
Figure 4: DTI regions of interest
CC SCR ACR
SLF ILF
PLIC CP iCST
94
RESULTS
The highly significant intraclass correlation coefficients indicated excellent
agreement both within and across raters (intrarater r = 0.92; interrater r = 0.88),
confirming previous reports of the reliability of ROQS (Niogi et al., 2008).
Positive correlations were found between a number of the ROIs and behavioral
measures, as seen in Table 10 and Figure 5. Bilateral SCR were related to ND-Comp,
TOWRE-W and RPN, among others. Right SLF and right ILF were related to the two
TOWRE measures and ND-Comp, but no significant relationships were observed in their
left-hemisphere homologues. Left iCST and the two word reading measures showed a
positive association. Left ACR, right SLF, and left CP demonstrated positive
relationships with the nonverbal measure, Spatial IQ.
95 Table 10: Correlations between behavioral measures and average FA values of ROIs
* p < 0.05; ** p < 0.01
Word ID Word Attack RPN TOWRE-PDE TOWRE-W ND-RR ND-Comp Verbal IQ Spatial IQ
CC 0.10 0.10 0.07 -0.09 0.01 -0.17 0.00 0.01 0.08
L SCR 0.09 0.19 0.35** 0.15 0.34** 0.18 0.28* -0.03 -0.01
R SCR 0.24* 0.14 0.39** 0.28* 0.40** 0.11 0.31** 0.04 -0.08
L ACR 0.10 0.05 0.16 0.14 0.12 0.11 0.15 0.13 0.34**
R ACR 0.05 -0.04 0.14 -0.02 0.03 0.06 0.04 0.13 0.09
L SLF 0.04 -0.02 0.14 0.00 0.08 -0.04 0.10 0.16 0.05
R SLF 0.12 0.16 0.16 0.24* 0.27* 0.15 0.33** 0.19 0.23
L ILF 0.09 0.04 0.18 0.10 0.20 0.10 0.21 0.08 0.05
R ILF 0.29* 0.18 0.20 0.30* 0.35** 0.18 0.36** 0.24 0.25*
L PLIC 0.07 0.06 0.01 0.11 0.11 0.16 0.16 0.06 0.06
R PLIC -0.11 -0.15 -0.02 -0.13 -0.02 0.09 -0.01 -0.05 -0.07
L CP 0.12 0.05 0.03 0.02 0.00 0.01 -0.04 0.18 0.24*
R CP 0.19 0.14 0.14 0.21 0.20 0.12 0.11 0.07 0.05
L iCST 0.28* 0.17 0.17 0.19 0.24* 0.08 0.23 0.05 0.03
R iCST 0.11 -0.01 0.02 -0.06 0.04 -0.03 -0.06 -0.06 -0.09
96
Figure 5: Specific and nonspecific relationships in three white matter regions
97
DISCUSSION
The present study investigated the variation in white matter regions of interest
hypothesized to be associated with reading ability within a large sample of adults.
Positive correlations were found within bilateral SCR, left iCST, and right superior and
inferior longitudinal fasciculus, providing further support for the importance of cross-
cortical communication for reading achievement.
Consistent with all previous reports of relationships between reading ability and
white matter tract integrity (Klingberg et al., 2000; Niogi & McCandliss, 2006; Beaulieu
et al., 2005; Deutsch et al., 2005; Odegard et al., 2009), the direction of each correlation
was positive. In the sample of 68 adults covering a wide range of reading ability, higher
FA values in multiple regions were associated with superior reading performance.
Whether reflecting increased integrity, increased uniformity, or increased number of
axons, higher FA values seem to indicate more robust connections that allow for
increased synchrony between brain regions involved in reading.
The finding of reliable skill-related variation in SCR is consistent with previous
findings. Positive associations between reading ability and FA in SCR have been
demonstrated in smaller samples in both children and adults and both nonimpaired and
reading disabled samples (Niogi & McCandliss, 2006; Deutsch et al., 2005; Beaulieu et
al., 2005; Klingberg et al., 2000). A clear interpretation has yet to be offered. One author
states: “Because the corona radiata does not seem to connect commonly identified brain
regions involved when reading…integrating it into current theories of reading is not an
easy proposition” (Odegard et al., 2009). Considering its size and breadth across the
brain, the SCR could underlie any number of functions. White matter atlases and recent
98
imaging findings define the SCR as part of the corticospinal tract (CST), through which
almost every projection fiber in the brain passes. (The iCST is at the inferior end of the
same tract, and it is also positively related to reading skill.) Projection fibers carry
information to and from the cortex, whereas association fibers connect same-hemisphere
cortical regions. The CST includes thalamocortical fibers, some of which arise in the
visual cortex and project to other regions; motor corticospinal fibers, which arise in
primary motor cortex and travel down the spinal cord; and frontopontine fibers, which
also travel down the spinal cord but originate in the frontal lobe. Its fibers arise from
primary motor cortex, supplementary motor area, premotor cortex, and sensory cortex
(e.g., Parker, Wheeler-Kingshott & Barker, 2002). It has been suggested that lower FA in
SCR among reading disabled samples might be related to motor deficits sometimes
associated with reading disability (Bruce McCandliss, personal communication, February
2, 2008), but it is unlikely that motor skills (speed, coordination, and balance) are
systematically related to reading ability across the entire spectrum. A more likely
interpretation is that SCR is related to language abilities, in general. Some evidence for
this is its significant correlation with rapid naming, a nonreading measure. Post-hoc
regression analyses showed that both TOWRE-W and RPN mediate the relationship
between left SCR and Nelson Denny. These findings argue against the specificity of
white matter relationships proposed by Niogi and McCandliss (2006). Variation in SCR
may actually reflect abilities in the verbal domain, as opposed to reading, specifically.
The results pertaining to the other tracts were more surprising. Because of its
notable role in historical disconnection cases (Wernicke, 1874; Geschwind, 1965), its
frequent appearance in fiber tractography studies of language (Parker et al., 2005), and its
99
overlap with reading-related activation peaks (Glasser & Rilling, 2008), the left arcuate
fasciculus was expected to be related to reading ability. However, no significant
correlations were found in connection with left SLF, which includes the arcuate. It is
possible that while the connection between Broca’s and Wernicke’s areas exists for all
healthy individuals, its structure doesn’t vary with skill (or much across individuals at
all). Some evidence in support of this homogeneity comes from an analysis of the
distribution of avergae FA values in each ROI. The left SLF had the second smallest
range and standard deviation. Another, probably more valid, explanation for the negative
finding is that the technique is not sensitive enough to parse out the arcuate from the
other fibers in the large SLF bundle, which could have obscured a significant
relationship.
A significant relationship was found in the right SLF, however, as well as the
right ILF and right SCR. Right-hemisphere correlations are rarely reported, but at least
two reports agree that the right-hemisphere homologue of the left arcuate exists in the
majority of humans. Parker and colleagues (2005) report from tractography findings that
while the ventral route from Wernicke’s to Broca’s areas was restricted to the left
hemisphere, a second, dorsal route appeared bilaterally. Glasser and Rilling (2008)
reconstructed fibers between various regions of the temporal and frontal lobes, and
reported that the only consistently appearing right-hemisphere arcuate connection ran
between the right middle temporal gyrus and the right frontal lobe. Some have suggested
that right temporal regions are specialized to process suprasegmental phonological
processing, i.e., prosody and intonation (Wildgruber et al., 2006). Suprasegmental
phonological processing is understudied, owing partly to the field’s focus on single-word
100
reading performance. But given the strong link between reading ability and lower level
processes like decoding (Mann & Liberman, 1984; Wagner et al., 1997, Bruck, 1998),
it’s likely that this higher-level phonological process also contributes to variability in
reading ability, and part of the neural basis for the relationship is the right arcuate
fasiculus/SLF. It follows that performance on Nelson Denny, which may benefit from
skilled suprasegmental phonological processing, is significantly correlated with right
SLF.
The significant relationship between right ILF and reading is more difficult to
interpret. The occipitotemporal fibers that comprise the ILF link prestriate visual areas to
medial temporal structures, such as the hippocampus and amygdala. The right ILF may
therefore be important in processing of visual information with an emotional or memory
load (Catani, Jones, Donato & Ffytche et al., 2003). How this relates to reading ability is
unclear and deserves further investigation.
Not all of the significant correlations are reading-related. The one nonverbal
measure, Spatial IQ, is positively correlated with a number of tracts: left ACR and left CP
are only related to Spatial IQ, and right ILF is related to both Spatial IQ and the reading
measures. Comparison of the reading and nonverbal correlation patterns (see Figure 5)
provides partial support to the hypothesis that different cognitive domains are supported
by distinct white matter pathways. Niogi & McCandliss (2006) discovered a
“correlational double dissociation” between word reading and working memory and their
corresponding white matter tracts (left SCR and bilateral ACR, respectively). Here, some
of the tracts are specifically related to one cognitive domain, but there is also evidence
that at least one tract, the right ILF, captures cognitive ability more globally. The large
101
visual component of all the measures that correlate with right ILF suggests that it might
be broadly associated with cognitive ability in the visual domain.
The current study provides further support that the integrity of select white matter
pathways can predict adult reading ability. While the objectivity of the ROI tracing
technique is one of the study’s advantages, in addition to its relatively large sample size,
the technique is also responsible for some of the study’s limitations. Each DTI measure
does not truly reflect the properties of an entire white matter tract, as it is only produced
from one two-dimensional slice. The slice at which each was taken did contain the largest
cross-section, however, and the positive findings suggest that important information was
captured. Additionally, the method could not indicate whether the ROIs passed through
reading regions only and were therefore specifically relevant to the reading network.
Future DTI studies should use fiber tractography to identify projections between reading
regions that are defined on an individual basis using functional MRI.
102
CHAPTER 5: GENERAL DISCUSSION
This set of studies was motivated by the need to clarify the role of phonological
processing skill in adult reading ability. The skill—defined by one’s knowledge and
application of the sounds of his or her language system—has a significant, reciprocal
relationship with reading development in children, and it represents the core deficit in
reading disability at all stages. If the relationship extends into the adult, skilled range,
then both behavioral and neural indices of decoding ability, one type of phonological
processing, would be predictive of reading fluency and comprehension in a broad sample
of adults. To test this hypothesis, correlations were computed between silent and oral
reading ability and a) behavioral measures of decoding skill, b) fMRI activity in the
phonological network during a natural passage reading task, c) functional connectivity
within the network during a natural passage reading task, and d) structural integrity of
reading-related white matter pathways. The resulting positive correlations provide
converging evidence for the importance of phonological processing in a wide range of
adult reading ability.
The first study reported positive correlations between pseudoword reading
measures and both silent and oral passage reading measures. The indication that, even
beyond reading development, decoding ability is a significant predictor of reading ability,
replicated numerous previous demonstrations of the relationship (Stanovich & West,
1989; Bell & Perfetti, 1994; Dietrich & Brady, 2001; Watson & Miller, 1993, Rayner et
al., 2001). Another index of phonological processing was then examined using functional
MRI. Regions of interest were localized on an individual basis by a decoding task, and
103
activity during a passage reading task in those regions served as a proxy of phonological
processing use in the context of everyday reading. Activation levels in many of the
regions within the network showed positive relationships with the behavioral passage
reading measures. Better readers appear to engage their decoding network to a greater
degree when reading connected text, thereby confirming the hypothesis at the neural
level. Previous studies have linked some of these regions to phonological processing
(Demonet et al., 1992; Graves et al., 2007; Levelt et al., 1998; Majerus et al., 2005;
vanAtteveldt et al., 2004), and others have demonstrated that skill modulates their
activity (Brunswick et al., 1999; Shaywitz et al, 2002; Paulesu et al., 2001; Maisog et al.,
2008), but no single study has provided evidence for both in the same individuals.
Do the different regions within the network contribute in different ways? A
variety of more basic reading and verbal measures were next considered in correlation
and regression analyses to determine unique roles. The relationship between pSTG and
GORT accuracy was mediated by TOWRE-PDE, as well as the relationship between OT
and GORT rate, indicating that the two regions capture some aspect of the decoding
process. That’s not all pSTG is responsible for, it seems, because its contribution to ND
comprehension was still significant after considering TOWRE-PDE, and single word
reading measures. The classically phonological region could be capturing multiple
phonological components, but additional studies are needed to identify those processes.
Most of the speeded measures were related to activity in OT during passage
reading, and it was also the only place where a relationship with reading rate in the
scanner approached significance. Easy passages, compared to difficult, also elicited
greater activation in OT. Altogether, although it appears to overlap with that from pSTG
104
(to which OT is functionally connected), the contribution from OT to skilled passage
reading may have to do with reading rate. Previously, it has been associated with rapid
word recognition and the direct, ventral, skilled route (e.g., Shaywitz, 2002). While the
present findings lend support to the rapid word recognition claim, they argue that rapid
pseudoword decoding is also in its repertoire of functions.
Finally, the results in the inferior frontal area support previous claims that IFG is
involved in grapheme-phoneme conversion (Joubert et al., 2004), and that a double
dissociation exists between anterior and posterior portions (Devlin et al., 2003). Anterior
IFG was correlated with the accuracy component of passage fluency, as well as many
more reading measures, including decoding, when localizer activity was considered.
Posterior IFG, on the other hand, was related to Timed Word Attack. While the study
cannot directly confirm the semantic role of aIFG, its demonstration of a task effect on
the pattern of correlations with aIFG activity fits with the theory that aIFG is involved in
both semantic and phonological processing. Meanwhile, pIFG only displays evidence for
a phonological role during the passage task.
The interesting task effect brings up some interesting questions. First, why don’t
the findings of positive relationships between skill and activity across many regions of
the network generalize from passage reading to single-word lexical decision? Second,
why would a region which is clearly related to phonological processing fail to show skill
modulation during a phonological task? Answers to these questions lie in an
interpretation of the findings that considers strategy differences. Better readers may
achieve the same task using a different strategy, which is manifested as increased activity
in the regions relevant to that strategy. To illustrate, consider the localizer task, which
105
requires subjects to decode each PH or PW and decide if the phonological form has a
match in his or her lexicon. The subjects quickly learn that a strategy that focuses on the
orthographic properties of the word is not sufficient (because both stimulus types are
unfamiliar and matched on sublexical orthography). They are instead forced to rely on the
decoding process. With the exception of aIFG, activity in the neural network associated
with decoding does not vary along the skill spectrum, possibly because it is maximally
engaged. The passage reading task, on the other hand, allows for all possible reading
strategies, and only in this unconstrained context do skill differences appear. Anterior
IFG may consistently vary under difficult task demands, as it was also more active during
difficult passages. Regions which show negative correlations with skill would be
expected to represent strategies that less skilled readers employ. The lack of negative
correlations in Tables 4 and 6 might imply that less skilled readers employ a variety of
strategies; only highly skilled readers, as a group, reliably depend on a single, identifiable
strategy. Decoding might be the most efficient, dependable strategy to achieve the
everyday task of reading.
Another place where strategy differences may appear is in network connectivity.
Previous connectivity studies of reading have drawn links between temporoparietal
regions and both extrastriate and frontal sites (Joubert et al., 2004; Hampson et al., 2002;
Rumsey et al., 1997; Horwitz et al., 1998; Pugh et al., 2000). Study 2 examined the
connectivity within the phonological processing network, specifically, during passage
reading by correlating average activation levels between pairs of ROIs that were derived
from Study 1. The resulting correlations, the majority of which included pSTG as one
member of the pair, provided additional evidence that temporoparietal regions serve as
106
central nodes of the reading circuity. Since they were active in a decoding context, pSTG
and its correlated cousins seem to be specifically relevant to phonological processing as it
is involved in reading. One study found age-related decreases in STG connectivity (Bitan
et al., 2007), but it seems from these results that the posterior portion, at least, continues
to play an important role in adult reading.
Not only was the average activation in pSTG correlated with that in many other
ROIs (OT, aIFG, SMA), but its online fluctuations were tightly coupled with aIFG in the
more skilled readers. Out of all the pairs of ROIs that demonstrated functional
connectivity across subjects, this pair showed increasing temporal correlation values with
increasing word recognition skill. Integrating this finding with those from Study 1,
decoding contributes to passage reading accuracy at the neural level via the coordination
between and associated activation increases within pSTG and aIFG. Whether the coupled
increases in activity are a result of online information transfer or of more reading
experience, mediated by Hebbian strengthening of synapses, is a question for future
studies of skill learning.
Turning briefly to another node in the network, SMA likely serves a late-stage
role in the reading process (because of its proximity to primary motor cortex, which
sends input directly to the articulators). Its functional connections with pSTG and R IFG
may therefore be unidirectional; information from pSTG and R IFG converge in SMA
prior to subvocalization. If the phonological input from pSTG is sublexical, the input
from R IFG might be sentence-level prosody. As the technique used does not indicate
directionality, and because no previous literature is able to inform the interpretation of
SMA or R IFG in phonological decoding, this is only speculation. However, R IFG has
107
been associated with processing emotional prosody (Rota et al., 2009). Its activation in
response to pseudohomophones and pseudowords in Study 1, which are too short to
invoke sentence-level prosodic processing, may reflect rhythm or stress processing at the
syllable level. In any case, while there is no evidence that R IFG activity or connectivity
varies with reading skill, there is evidence for some involvement in decoding.
Studies 1 and 2 together provide substantial evidence that fMRI indices of
phonological processing during passage reading are predictive of adult reading ability.
Attempting to provide some degree of structural support for the functional connectivity
findings, Study 3 investigated white matter integrity in regions of interest that were
localized in each subject in an objective and reliable manner. Although the white matter
regions of interest were chosen based on their relevance in previous DTI studies of
reading, they do not necessarily represent specific anatomical connections between
cortical reading regions. Neverthless, a number of significant positive relationships were
identified between reading ability and measures of white matter integrity. Higher FA
values in bilateral SCR, left iCST, and right SLF and ILF were predictive of superior
performance on single word and passage reading tests.
No effects were observed in left SLF, a portion of which links Broca’s area to
Wernicke’s area. Although this seems surprising, considering the anatomy of the fiber
tract and many “disconnection syndrome” case studies that are a result of damage to it,
the negative result is actually in keeping with many other DTI studies of reading. It may
be the case that while SLF is intact in all neurologically healthy individuals, its individual
structure shows little variability. Another tract, the SCR, appears to be more relevant to
reading ability in both development and reading disability (Niogi & McCandliss, 2006;
108
Deutsch et al., 2005; Beaulieu et al., 2005; Klingberg et al., 2000). One of the brain’s
largest and most far-reaching fiber bundles, the SCR may carry projection fibers that
innervate or arise from multiple reading regions. Many plausible interpretations exist,
therefore, including one that relates SCR to language abilities, in general. Some evidence
for this is its significant correlation with rapid naming. In fact, post-hoc regression
analyses showed that both TOWRE-W and RPN mediate the relationship between left
SCR and Nelson Denny. These findings argue against the specificity of white matter
relationships proposed by Niogi and McCandliss (2006). Variation in SCR may actually
reflect abilities in the verbal domain, as opposed to reading, specifically. The overlapping
correlations between one region, right ILF, and both nonverbal and verbal measures
provide additional evidence against the specificity hypothesis. More fine-grained
separation of tracts may result in more domain-specific relationships, but at the level of
major fiber tracts, the results are not convincing.
The results also provide no evidence that reading-related differences in white
matter structures are specific to the left hemisphere, as has been suggested previously
(Niogi & McCandliss, 2006; Klingberg et al., 2000). In fact, right-hemisphere regions of
interest are responsible for the majority of the significant correlations. The source of the
inconsistencies may lie in the type of reading measure that was investigated by previous
studies. In Niogi & McCandliss (2006), childrens’ performance on single-word tasks was
correlated with left SCR only. In the current study, adults’ performance on both rapid
single-word reading and complex passage reading comprehension was correlated with
left and right SCR. Contributions from the right hemisphere may be more important in
sentence-level reading ability, due to the importance of right-lateralized processes like
109
prosody on reading beyond the single word level. It is also possible that the samples’ age
differences are producing the inconsistencies; right hemisphere components of the
language network may have more influence on adults’ reading ability.
Overall, the three studies have replicated and extended previous findings. The
relationship between phonological processing ability and reading ability is evident not
only in reading development and disability, but also in skilled adults. The relationship
can be detected not only in behavioral reading performance, but also in neural activity
and connectivity. The claim that effects of skill on neural activity are mediated by
phonological processing is not entirely novel. Past studies have extrapolated the
relationship, based on the functional roles assigned to the regions by separate studies. The
present study’s use of a decoding localizer allowed for a more direct test. Another
improvement on the design was the use of a self-paced passage reading scanner task,
which is a closer approximation to everyday reading than are the more common single-
word tasks. That the findings are in the same direction and in similar regions as those
from single-word studies of reading ability is comforting. The present findings can
luckily offer ecological validation, as opposed to refutation.
Although the passage reading task offered an ecological advantage, it was not an
ideal scanner task in terms of real-time performance tracking. Future studies might test
comprehension immediately after the task and/or use eyetracking to confirm subjects’
participation. Also, examining only decoding ability provides a narrow window into the
relationship between phonological processing and reading. Other aspects of phonological
processing should be studied in a similar way, using localizers to first identify the
relevant network. Further design limitations concern the connectivity methods.
110
Functional connectivity does not imply direction of regions’ influences, and the DTI
large-track tracing method did not allow for interpretations at the level of functional
specificity. Future DTI studies will use fiber tractography to identify projections between
reading regions that are defined on an individual basis using functional MRI. Future
directions also include further testing of the idea that it may be variations in strategy,
rather than skill, per se, that drive relationships between network activity and high-level
cognitive abilities.
111
REFERENCES
Apthorp, H. S. (1995). Phonetic coding and reading in college students with and without
learning disabilities. Journal of Learning Disabilities, 28, 342-352.
Ashtari, M., Cervellione, K. L., Hasan, K. M., Wu, J., McIlree, C., Kester, H., Ardekani,
B. A., Roofeh, D., Szeszko, P. R., & Kumra, S. (2007). White matter development
during late adolescence in healthy males: a cross-sectional diffusion tensor
imaging study. Neuroimage, 35, 501-510.
Assaf, M., Calhoun, V. D., Kuzu, C. H., Kraut, M. A., Rivkin, P. R., Hart, J. Jr., &
Pearlson, G. D. (2006). Neural correlates of the object-recall process in semantic
memory. Psychiatry Research, 147, 115-126.
Aylward, E. H., Richards, T. L., Berninger, V. W., Nagy, W. E., Field, K. M., Grimme,
A. C., Richards, A. L., Thomson, J. B., & Cramer, S. C. (2003). Instructional
treatment associated with changes in brain activation in children with dyslexia.
Neurology, 61, 212-219.
Backes, W., Vuurman, E., Wennekes, R., Spronk, P., Wuisman, M., van Engelshoven, J.,
& Jolles, J. (2002). Atypical brain activation of reading processes in children with
developmental dyslexia. Journal of Child Neurology, 17, 867-871.
Badian, N. A. (1993). Phonemic awareness, naming, visual symbol processing, and
reading. Reading and Writing, 5, 87-100.
Beaulieu, C., Plewes, C., Paulson, L. A., Roy, D., Snook, L., Concha, L., & Phillips, L.
(2005). Imaging brain connectivity in children with diverse reading ability.
Neuroimage, 25, 1266-1271.
Behrens, E. J., Woolrich, M. W., Jenkinson, M., Johansen-Berg, H., Nunes, R. G., Clare,
S., Matthews, P. M., Brady, J. M., & Smith, S. M. (2003). Characterization and
propagation of uncertainty in diffusion-weighted MR imaging. Magnetic
Resonance in Medicine, 50, 1077-1088.
Bell, L. C., & Perfetti, C. A. (1994). Reading skill: Some adult comparisons. Journal of
Educational Psychology, 86, 244-255.
Ben Bashat, D., Ben Sira, L., Graif, M., Pianka, P., Hendler, T., Cohen, Y., & Assaf, Y.
(2005). Normal white matter development from infancy to adulthood: comparing
diffusion tensor and high b value diffusion weighted MR images. Journal of
Magnetic Resonance Imaging, 21, 503-511.
112
Benson, R. R., Whalen, D. H., Richardson, M., Swainson, B. Clark, V. P., Lai, S., &
Liberman, A. M. (2001). Parametrically dissociating speech and non-speech
perception in the brain using fMRI. Brain Language, 78, 364-396.
Biswal, B. B., Van Kylen, J., & Hyde, J. S. (1997). Simultaneous assessment of flow and
BOLD signals in resting-state functional connectivity maps. NMR Biomedicine,
10, 165-170.
Bitan, T., Booth, J. R., Choy, J., Burman, D. D., Gitelman, D. R., & Mesulam, M.-M.
(2005). Shifts of effective connectivity within a language network during rhyming
and spelling. The Journal of Neuroscience, 25, 5397-5403.
Bitan, T., Cheon, J., Lu, D., Burman, D. D., Gitelman, D. R., Mesulam, M. M., & Booth,
J. R. (2007). Developmental changes in activation and effective connectivity in
phonological processing. Neuroimage, 38, 564-575.
Blacker, D., Byrnes, M. L., Mastaglia, F. L., & Thickbroom, G. W. (2006). Differential
activation of frontal lobe areas by lexical and semantic language tasks: a
functional magnetic resonance imaging study. Journal of Clinical Neuroscience,
13, 91-95.
Bokde, A. L., Tagamets, M. A., Friedman, R. B., & Horwitz, B. (2001). Functional
interactions of the inferior frontal cortex during the processing of words and
word-like stimuli. Neuron, 30, 609-617.
Booth, J. R. (2007). Brain bases of learning and development of language and reading.
In D Coch, G Dawson & KW Fischer (Eds), Human Behavior, Learning and the
Developing Brain, pp. 279-300. Guilford Press.
Booth, J. R., Burman, D. D., Meyer, J. R., Gitelman, D. R., Parrish, T. B., Mesulam, M.
M. (2004). Development of brain mechanisms for processing orthographic and
phonologic representations. Journal of Cognitive Neuroscience, 16, 1234-1249.
Booth, J. R., Burman, D. D., Van Santen, F. W., Harasaki, Y., Gitelman, D. R., Parrish,
T. B., & Marsel Mesulam, M. M. (2001). The development of specialized brain
systems in reading and oral language. Child Neuropsychology, 7, 119-141.
Brainard, D. H. (1997). The Psychophysics Toolbox. Spatial Vision, 10, 433-436.
Brebion, G., Amador, X., Smith, M. J., & Gorman, J. M. (1998). Memory impairment
and schizophrenia: the role of processing speed. Schizophrenia Research, 30, 31-
39.
Brown, W. E., Eliez, S., Menon, V., Rumsey, J. M., White, C. D., & Reiss, A. L. (2001).
Preliminary evidence of widespread morphological variations of the brain in
dyslexia. Neurology, 56, 781-783.
113
Brown, J., Fischo, V., & Hanna, G. (1993). Nelson-Denny Reading Test, Manual for
scoring and interpretation: Forms G & H.Chicago, IL: Riverside Press.
Bruck, M. (1998). Outcomes of adults with childhood histories of dyslexia. In C. Hulme
& R. M. Joshi (Eds.), Reading and spelling: Development and disorders (pp. 179-
200). Mahwah, NJ: Erlbaum.
Bruno, J. L., Lu, Z.-L., & Manis, F. R. (in preparation). Reading skill is related to
regional variation in white matter microstructure: A diffusion tensor imaging
study.
Bruno, J. L., Zumberge, A., Manis, F. R., Lu, Z.-L., & Goldman, J. G. (2008). Sensitivity
to orthographic familiarity in the occipito-temporal region. Neuroimage, 39,
1988-2001.
Brunswick, N., McCrory, E., Price, C. J., Frith, C. D., & Frith, U. (1999). Explicit and
implicit processing of words and pseudowords by adult developmental dyslexics:
A search for Wernicke’s Wortschatz. Brain, 122, 1901-1917.
Buchsbaum, B. R., Hickok, G., & Humphries, C. (2001). Role of left posterior superior
temporal gyrus in phonological processing for speech perception and production.
Cognitive Science, 25, 663-678.
Bullmore, E., Horwitz, B., Honey, G., Brammer, M., Williams, S., & Sharma, T. (2000).
How good is good enough in path analysis of fMRI data? Neuroimage, 11, 289-
301.
Burton, M. W., Locasto, P. C., Krebs-Noble, D., & Gullapalli, R. P. (2005). A systematic
investigation of the functional neuroanatomy of auditory and visual phonological
processing. Neuroimage, 26, 647-661.
Burton, M. W., Small, S. L., & Blumstein, S. E. (2000). The role of segmentation in
phonological processing: an fMRI investigation. Journal of Cognitive
Neuroscience, 12, 679-690.
Carroll, J. B., Davies, P., & Richman, B. (1971). The American Heritage Word
Frequency Book. American Heritage, New York.
Carter, J. C., Lanham, D. C., Cutting, L. E., Clements-Stephens, A. M., Chen, X.,
Hadzipasic, M., Kim, J., Denckla, M. B., & Kaufmann, W. E. (2009). A dual DTI
approach to analyzing white matter in children with dyslexia. Psyhiatry Research,
172, 215-219.
Catani, M., Jones, D. K., Donato, R., Ffytche, D. H. (2003). Occipito-temporal
connections in the human brain. Brain, 126, 2093-2107.
114
Cerella, J. (1991). Age effects may be global, not local: comment on Fisk and Rogers.
Journal of Experimental Psychology General, 2, 215-223.
Church, J. A., Coalson, R. S., Lugar, H. M., Petersen, S. E., & Schlaggar, B. L. (2008). A
developmental fMRI study of reading and repetition reveals changes in
phonological and visual mechanisms over age. Cerebral Cortex, 18, 2054-2065.
Cohen, L., Dehaene, S., Naccache, L., Lehericy, S., Dehaene-Lambertz, G., Henaff, M.
A., & Michel, F. (2000). The visual word form area: spatial and temporal
characterization of an initial stage of reading in normal subjects and posterior
split-brain patients. Brain, 123, 291-307.
Coltheart, M., Davelaar, E., Jonasson, J. T., & Besner, D. (1977). Access to the internal
lexicon. In S. Dornic (Ed.), Attention and Performance vI (pp. 535-555).
Hillsdale, NJ: Erlbaum.
Cutting, L. E., Clements, A. M., Courtney, S., Rimrodt, S. L., Schafer, J. G., Bisesi, J.,
Pekar, J. J., & Pugh, K. R. (2006). Differential components of sentence
comprehension: beyond single word reading and memory. Neuroimage, 29, 429-
438.
Dehaene, S., Le Clec’H, G., Poline, J. B., Le Bihan, D. & Cohen, L. (2002). The visual
word form area: a prelexical representation of visual words in the fusiform gyrus.
Neuroreport, 13, 321-325.
Dejerine, J. (1895). Anatomy of central nervous system. Paris: Masson.
Démonet, J. F., Chollet, F., Ramsay, S., Cardebat, D., Nespoulous, J. L., Wise, R.,
Rascol, A., & Frackowiak, R. (1992). The anatomy of phonological and semantic
processing in normal subjects. Brain, 115, 1753-1768.
Deutsch, G. K., Dougherty, R. F., Bammer, R., Siok, W. T., Gabrieli, J. D., & Wandell,
B. (2005). Children’s reading performance is correlated with white matter
structure measured by diffusion tensor imaging. Cortex, 41, 354-363.
Devlin, J. T., Matthew, P. M., & Rushworth, M. F. (2003). Semantic processing in the
left inferior prefrontal cortex: a combined functional magnetic resonance imaging
and transcranial magnetic stimulation study. Journal of Cognitive Neuroscience,
15, 71-84.
Dietrich, J. A., & Brady, S. A. (2001). Phonological representations of adult poor readers:
An investigation of specificity and stability. Applied Psycholinguistics, 22, 383-
418.
Dosenbach, N. U., Fair, D. A., Cohen, A. L., Schlaggar, B. L., & Petersen, S. E. (2008).
A dual-networks architecture of top-down control. Trends in Cognitive Sciences,
12, 99-105.
115
Dougherty, R. F., Ben-Shachar, M., Deutsch, G. K., Hernandez, A., Fox, G. R., &
Wandell, B. A. (2007). Temporal-callosal pathway diffusivity predicts
phonological skills in children. Proceedings of the National Academy of Sciences
USA, 104, 8556-8561.
Eckert, M. A., Leonard, C. M., Richards, T. L., Aylward, E. H., Thomson, J., Berninger,
V. W. (2003). Anatomical correlates of dyslexia: frontal and cerebellar findings.
Brain, 126, 482-494.
Edwards, S. G., Liu, C., & Blumhardt, L. D. (2001). Cognitive correlates of
supratentorial atrophy on MRI in multiple sclerosis. Acta Neurologica
Scandinavica, 104, 214-223.
Ehri, L. C. (1998). Grapheme-phoneme knowledge is essential for learning to read words
in English. In J Metsala & L. Ehri (Eds.), Word recognition in beginning literacy
(pp. 3-40). Hillsdale, NJ: Erlbaum.
Fay, S., Isingrini, M., Ragot, R., & Pouthas, V. (2005). The effect of encoding
manipulation on word-stem cued recall: an event-related potential study. Brain
Research: Cognitive Brain Research, 24, 615-626.
Ferreres, A. R., Lopez, C. V., & China, N. N. (2003). Phonological alexia with vowel-
consonant dissociation in nonword reading. Brain Language, 84, 399-413.
Fiez, J. A. (1997). Phonology, semantics, and the role of the left inferior prefrontal
cortex. Human Brain Mapping, 5, 79-83.
Fiez, J. A., & Petersen, S. E. (1998). Neuroimaging studies of word reading. Proceedings
of the National Academy of Sciences USA, 95, 914-921.
Fiez, J. A., Tranel, D., Seager-Frerichs, D., & Damasio, H. (2006). Specific reading and
phonological processing deficits are associated with damage to the left frontal
operculum. Cortex, 42, 624-643.
Frost, S. J., Landi, N., Mencl, W. E., Sandak, R., Fulbright, R. K., Tejada, E. T.,
Jacobsen, L., Grigorenko, E. L., Constable, R. T., & Pugh, K. R. (2009).
Phonological awareness predicts activation patterns for print and speech. Annals
of Dyslexia, 59, 78-97.
Geschwind, N. (1965). Disconnexion syndromes in animals and man. I&II. Brain, 88,
237-294 585-643.
Glasser, M. F., & Rilling, J. K. (2008). DTI tractography of the human brain’s language
pathways. Cerebral Cortex, 18, 2471-2482.
116
Gold, B. T., & Buckner, R. L. (2002). Common prefrontal regions coactivate with
dissociable posterior regions during controlled semantic and phonological tasks.
Neuron, 35, 803-812.
Gold, B. T., Balota, D. A., Kirchhoff, B. A., & Buckner, R. L. (2005). Common and
dissociable activation patterns associated with controlled semantic and
phonological processing: evidence from fMRI adaptation. Cerebral Cortex, 15,
1438-1450.
Gomez-Perez, E., Ostrosky-Solis, F., & Prospero-Garcia, O. (2003). Desarrollo de la
atencion, la memoria y los procesos inhibitorios: relacion temporal con
maduracion de la estructura y function cerebral. Revista de Neurologia, 37, 561-
567.
Grabner, R. H., Neubauer, A. C., Stern, E. (2006). Superior performance and neural
efficiency: the impact of intelligence and expertise. Brain Research Bulletin, 69,
422-439.
Graves, W. W., Grabowski, T. J., Mehta, S., & Gordon, J. K. (2007). A neural signature
of phonological access: distinguishing the effects of word frequency from
familiarity and length in overt picture naming. Journal of Cognitive Neuroscience,
19, 617-631.
Graves, W. W., Grabowski, T. J., Mehta, S., & Gupta, P. (2008). The left posterior
superior temporal gyrus participates specifically in accessing lexical phonology.
Journal of Cognitive Neuroscience, 20, 1698-1710.
Grodzinsky, Y., & Santi, A. (2008). The battle for Broca’s region. Trends in Cognitive
Sciences, 12, 474-480.
Grossi, G., Coch, D., Coffey-Corina, S., Holcomb, P. J., & Neville, H. J. (2001).
Phonological processing in visual rhyming: a developmental erp study. Journal of
Cognitive Neuroscience, 13, 610-625.
Hagoort, P., Indefrey, P., Brown, C., Herzog, H., Steinmetz, H. & Seitz, R. J. (1999). The
neural circuitry involved in the reading of German words and pseudowords: A
PET study. Journal of Cognitive Neuroscience, 11, 383-398.
Haier, R. J., Siegel, B. V., Tang, C., Abel, L., & Buchsbaum, M. S. (1992). Intelligence
and changes in regional cerebral glucose metabolic rate following learning.
Intelligence, 16, 415-426.
Hampson, M., Peterson, B. S., Skudlarski, P., Gatenby, J. C., & Gore, J. C. (2002).
Detection of functional connectivity using temporal correlations in MR images.
Human Brain Mapping, 15, 247-262.
117
Hampson, M., Tokoglu, F., Sun, Z., Schafer, R. J., Skudlarski, P., Gore, J. C. &
Constable, R. T. (2006). Connectivity-behavior analysis reveals that functional
connectivity between left BA39 and Broca’s area varies with reading ability.
Neuroimage, 31, 513-519.
Hayakawa, K., Konishi, Y., Kuriyama, M., Konishi, K., & Matsuda, T. (1991). Normal
brain maturation in MRI. European Journal of Radiology, 12, 208-215.
Heim, S., Opitz, B., Müller, K., & Friederici, A. D. (2003). Phonological processing
during language production: fMRI evidence for a shared production-
comprehension network. Brain Research: Cognitive Brain Research, 16, 285-296.
Hickok, G., Erhard, P., Kassubek, J., Helms-Tillery, A. K., Naeve-Valguth, S., Strupp, J.
P., Strick, P. L., & Ugurbil, K. (2000). A functional magnetic resonance imaging
study of the role of left posterior superior temporal gyrus in speech production:
implications for the explanation of conduction aphasia. Neuroscience Letters, 287,
156-160.
Hoeft, F. Hernandez, A., McMillon, G., Taylor-Hill, H., Martindale, J. L., Meyler, A.,
Keller, T. A., Siok, W. T., Deutsch, G. K., Just, M. A., Whitfield-Gabrieli, S., &
Gabrieli, J. D. (2006). Neural basis of dyslexia: a comparison between dyslexic
and nondyslexic children equated for reading ability. Journal of Neuroscience, 26,
10700-10708.
Hoeft, F., Meyler, A., Hernandez, A., Juel, C., Taylor-Hill, H., Martindale, J. L.,
McMillon, G., Kolchugina, G., Black, J. M., Faizi, A., Deutsch, G. K., Siok, W.
T., Reiss, A. L., Whitfield-Gabrieli, S., & Gabrieli, J. D. (2007). Functional and
morphometric brain dissociation between dyslexia and reading ability.
Proceedings of the National Academy of Sciences USA, 104, 4234-4239.
Hoeft, F., Ueno, T., Reiss, A. L., Meyler, A., Whitfield-Gabrieli, S., Glover, G. H.,
Keller, T. A., Kobayashi, N., Mazaika, P., Jo, B., Just, M. A., & Gabrieli, J. D.
(2007). Prediction of children’s reading skills using behavioral, functional, and
structural neuroimaging measures. Behavioral Neuroscience, 121, 602-613.
Horwitz, B., Rumsey, J. M., & Donohue, B. C. (1998). Functional connectivity of the
angular gyrus in normal reading and dyslexia. Proceedings of the National
Academy of Sciences USA, 95, 8939-8944.
Il’yuchenok, I. R., Sysoeva, O. V., & Ivanitskii, A. M. (2008). Two semantic systems in
the brain for rapid and slow differentiation of abstract and concrete words.
Neuroscience and Behavioral Physiology, 38, 963-970.
Johnson, M. H. (2001). Functional brain development in humans. Nature Reviews:
Neuroscience, 2, 475-483.
118
Joseph, J., Noble, K., & Eden, G. (2001). The neurobiological basis of reading. Journal of
Learning Disabilities, 34, 566-579.
Jobard, G., Vigneau, M., Mazoyer, B., & Tzourio-Mazoyer, N. (2007). Impact of
modality and linguistic complexity during reading and listening tasks.
Neuroimage, 34, 784-800.
Joubert, S., Beauregard, M., Walter, N., Bourgouin, P., Beaudoin, G., Leroux, J. M.,
Karama, S., & Lecours, A. R. (2004). Neural correlates of lexical and sublexical
processing in children. Brain Language, 89, 9-20.
Kail, R. (1986). Sources of age differences in speed of processing. Child Development,
57, 969-987.
Kail, R. (1988). Developmental fluctuations for speeds of cognitive processes. Journal of
Experimental Child Psychology, 45, 339-364.
Keating, D. P., & Bobbitt, B. L. (1978). Individual and developmental differences in
cognitive processing components of mental ability. Child Development, 49, 155-
169.
Keenan, J. M., & Betjemann, R. S. (2006). Comprehending the Gray Oral Reading Test
without reading it: Why comprehension tests should not include passage-
independent items. Scientific Studies of Reading, 10, 363-380.
Kim, H. J., Chu, K., Lee, K. M., Kim, D. W., & Park, S. H. (2002). Phonological
agraphia after superior temporal gyrus infarction. Archives of Neurology, 59,
1314-1316.
Klingberg, T., Hedehus, M., Temple, E., Salz, T., Gabrieli, J. D., Moseley, M. E., &
Poldrack, R. A. (2000). Microstructure of temporo-parietal white matter as a basis
for reading ability: evidence from diffusion tensor magnetic resonance imaging.
Neuron, 25, 493-500.
Kosslyn, S. M., Thompson, W. L., Kim, I. J., Rauch, S. L., & Alpert, N. M. (1996).
Individual differences in cerebral blood flow in Area 17 predicts the time to
evaluate visualized letters. Journal of Cognitive Neuroscience, 8, 78-82.
Kronbichler, M., Bergmann, J., Hutzler, F., Staffen, W., Mair, A., Ladurner, G., &
Wimmer, H. (2007). Taxi vs. taksi: on orthographic word recognition in the left
ventral occipitotemporal cortex. Journal of Cognitive Neuroscience, 19, 1584-
1594.
Kujala, J., Pammer, K., Cornelissen, P., Roebroeck, A., Formisano, E., & Salmelin, R.
(2007). Phase coupling in a cerebro-cerebellar network at 8-13 Hz during reading.
Cerebral Cortex, 17, 1476-1485.
119
Larson, G. E., Haier, R. J., LaCasse, L., & Hazen, K. (1995). Evaluation of a “mental
effort” hypothesis for correlations between cortical metabolism and intelligence.
Intelligence, 21, 267-278.
Lee, C. H. (2008). Testing the role of phonology in reading: Focus on semantic
processing. Journal of Psycholinguistic Research, Online First.
Lehman-Blake, M. T., & Tompkins, C. A. (2001). Predictive inferencing in adults with
right hemisphere brain damage. Journal of Speech, Language and Hearing
Research, 44, 639-654.
Levelt, W. J., Praamstra, P., Meyer, A. S., Helenius, P., & Salmelin, R. (1998). An MEG
study of picture naming. Journal of Cognitive Neuroscience, 10, 553-567.
Lichtheim, L. (1885). On aphasia. Brain, 7, 433-484.
Lincoln, A. E., Long, D. L., Swick, D., Larsen, J., & Baynes, K. (2008). Hemispheric
asymmetries in the perceptual representations of words. Brain Research, 1188,
112-121.
Lindenberg, R. & Scheef, L. (2007). Supramodal language comprehension: role of the
left temporal lobe for listening and reading. Neuropsychologia, 45, 2407-2415.
Luo, C. R., Johnson, R. A., & Gallo, D. A. (1998). Automatic activation of phonological
information in reading: Evidence from the semantic relatedness decision task.
Memory & Cognition, 26, 833-843.
Maisog, J. M., Einbinder, E. R., Flowers, D. L., Turkeltaub, P. E., & Eden, G. F. (2008).
A meta-analysis of functional neuroimaging studies of dyslexia. Annals of the
New York Academy of Sciences, 1145, 237-259.
Majerus, S. Van der Linden, M., Collette, F., Laureys, S., Poncelet, M., Degueldre, C.,
Delfiore, G., Luxen, A., & Salmon, E. (2005). Modulation of brain activity during
phonological familiarization. Brain Language, 92, 320-331.
Manis, F. R., Seidenberg, M. S., Doi, L. M., McBride-Chang, C., & Peterson, A. (1996).
On the basis of two subtypes of developmental dyslexia. Cognition, 58, 157-195.
Mann, V. A. (1984). Longitudinal prediction and prevention of early reading difficulty.
Annals of Dyslexia, 34, 117-136.
Mann, V. A., & Liberman, I. Y. (1984). Phonological awareness and verbal short-term
memory. Journal of Reading Disabilities, 17, 592-599.
Marien, P., Pickut, B. A., Engelborghs, S., Martin, J. J., & De Deyn, P. P. (2001).
Phonological agraphia following a focal anterior insulo-opercular infarction.
Neuropsychologia, 39, 845-855.
120
McCrory, E. J., Mechelli, A., Frith, U. & Price, C. J. (2005). More than words: a common
neural basis for reading and naming deficits in developmental dyslexia. Brain,
128, 261-267.
McDermott, K. B., Petersen, S. E., Watson, J. M., & Ojemann, J. G. (2003). A procedure
for identifying regions preferentially activated by attention to semantic and
phonological relations using functional magnetic resonance imaging.
Neuropsychologia, 41, 293-303.
McGraw, P., Liang, L., & Provenzale, J. M. (2002). Evaluation of normal age-related
changes in anisotropy during infancy and childhood as shown by diffusion tensor
imaging. American Journal of Roentgenology, 179, 1515-1522.
Mechelli, A., Crinion, J. T., Long, S., Friston, K. J., Lambon Ralph, M. A., Patterson, K.,
McClelland, J. L., & Price, C. J. (2005). Dissociating reading processes on the
basis of neuronal interactions. Journal of Cognitive Neuroscience, 17, 1753-1765.
Mechelli, A., Penny, W. D., Price, C. J., Gitelman, D. R., & Friston, K. J. (2002).
Effective connectivity and intersubject variability: using a multisubject network to
test differences and commonalities. Neuroimage, 17, 1459-1469.
Medler, D. A., & Binder, J. R. (2005). MCWord: an on-line orthographic database of the
English language. http://www.neuro.mcw.edu/mcword/.
Meyler, A., Keller, T. A., Cherkassky, V. L., Lee, D., Hoeft, F., Whitfield-Gabrieli, S.,
Gabrieli, J. D. & Just, M. A. (2007). Brain activation during sentence
comprehension among good and poor readers. Cerebral Cortex, 17, 2780-2787.
Morais, J., Cary, L., Alegria, J., & Bertelson, P. (1979). Does awareness of speech as a
sequence of phones arise spontaneously? Cognition, 7, 323-331.
Murtha, S., Chertkow, H., Beauregard, M., & Evans, A. (1999). The neural substrate of
picture naming. Journal of Cognitive Neuroscience, 11, 399-423.
Nagy, Z., Westerberg, H., & Klingberg, T. (2004). Maturation of white matter is
associated with the development of cognitive functions during childhood. Journal
of Cognitive Neuroscience, 16, 1227-1233.
Newman, S. D., Carpenter, P. A., Varma, S., & Just, M. A. (2003). Frontal and parietal
participation in problem solving in the Tower of London. Neuropsychology, 41,
1668.
Niogi, S. N., & McCandliss, B. D. (2006). Left lateralized white matter microstructure
accounts for individual differences in reading ability and disability.
Neuropsychologia, 44, 2178-2188.
121
Niogi, S. N., Mukherjee, P., Ghajar, J., Johnson, C. E., Kolster, R., Lee, H., Suh, M.,
Zimmerman, R. D., Manley, G. T., & McCandliss, B. D. (2008). Structural
dissociation of attentional control and memory in adults with and without mild
traumatic brain injury. Brain, 131, 3209-3221.
Niogi, S. N., Mukherjee, P., McCandliss, B. D. (2007). Diffusion tensor imaging
segmentation of white matter structures using a Reproducible Objective
Quantification Scheme (ROQS). Neuroimage, 35, 166-174.
Odegard, T. N., Farris, E. A., Ring, J., McColl, R., & Black, J. (2009). Brain connectivity
in non-reading impaired children and children diagnosed with developmental
dyslexia. Neuropsychologia, 47, 1972-1977.
Olesen, P. J., Nagy, Z., Westerberg, H. & Klingberg, T. (2003). Combined analysis of
DTI and fMRI data reveals a joint maturation of white and grey matter in a
fronto-parietal network. Brain Research: Cognitive Brain Research, 18, 48-57.
Olson, R., Wise, B., Conners, F., Rack, J., & Fulker, D. (1989). Specific deficits in
component reading and language skills: genetic and environmental influences.
Journal of Learning Disabilities, 22, 339-348.
Parker, G. J., Luzzi, S., Alexander, D. C., Wheeler-Kingshott, C. A., Ciccarelli, O., &
Lambon Ralph, M. A. (2005). Lateralization of ventral and dorsal auditory-
language pathways in the human brain. Neuroimage, 24, 656-666.
Parker, G. J. M., Wheeler-Kingshott, C. A. M., & Barker, G. J. (2002). Estimating
distributed anatomical connectivity using fast marching methods and diffusion
tensor imaging. IEEE Transactions on Medical Imaging, 21, 505-512.
Paulesu, E., Démonet, J. F., Fazio, F., McCrory, E. Chanoine, V., Brunswick, N., Cappa,
S. F., Cossu, G., Habib, M., Frith, C. D., & Frith, U. (2001). Dyslexia: cultural
diversity and biological unity. Science, 291, 2165-2167.
Paulesu, E., Frith, U., Snowling, M., Gallagher, A., Morton, J., Frackowiak, R. S., &
Frith, C. D. (1996). Is developmental dyslexia a disconnection syndrome?
Evidence from PET scanning. Brain, 119, 143-157.
Pennington, B. F., Van Orden, G. C., Smith, S. D., Green, P. A., & Haith, M. M. (1990).
Phonological processing skills and deficits in adult dyslexics. Child Development,
61, 1753-1778.
Perfetti, C. A., Beck, I., Bell, L. C., & Hughes, C. (1987). Phonemic knowledge and
learning to read are reciprocal: A longitudinal study of first grade children.
Merrill-Palmer Quarterly, 33, 283-319.
122
Petersen, S. E., Fox, P. T., Posner, M. I., Mintun, M., & Raichle, M. E. (1989). Positron
emission tomographic studies of the processing of single words. Journal of
Cognitive Neuroscience, 1, 153-170.
Poldrack, R. A., & Gabrieli, J. D. (2001). Characterizing the neural mechanisms of skill
learning and repetition priming: evidence from mirror reading. Brain, 124, 67-82.
Poldrack, R. A., Temple, E., Protopapas, A., Nagarajan, S., Tallal, P., Merzenich, M., &
Gabrieli, J. D. E. (2001). Relations between the neural bases of dynamic auditory
processing and phonological processing: evidence from fMRI. Journal of
Cognitive Neuroscience, 13, 686-697.
Poldrack, R. A., Wagner, A. D., Prull, M. W., Desmond, J. E., Glover, G. H., & Gabrieli,
J. D. (1999). Functional specialization for semantic and phonological processing
in the left inferior prefrontal cortex. Neuroimage, 10, 15-35.
Prat, C. S., Keller, T. A., Just, M. A. (2007). Individual differences in sentence
comprehension: a functional magnetic resonance imaging investigation of
syntactic and lexical processing demands. Journal of Cognitive Neuroscience, 19,
1950-1963.
Price, C. J., & McCrory, E. (2007) Functional Brain Imaging Studies of Reading and
Developmental Dyslexia. In M. J. Snowling & C. Hulme (Eds), The Science of
Reading: a Handbook,.pp. 473-496. Malden MA: Blackwell
Price, C. J., Veltman, D. J., Ashburner, J., Josephs, O., & Friston, K. J. (1999). The
critical relationship between the timing of stimulus presentation and data
acquisition in blocked designs with fMRI. Neuroimage, 10, 36-44.
Pugh, K. R., Mencl, W. E., Jenner, A. R., Katz, L., Frost, S. J., Lee, J. R., Shaywitz, S. E.,
& Shaywitz, B. A. (2001). Neurobiological studies of reading and reading
disability. Journal of Communication Disorders, 34, 479-492.
Pugh, K. R., Mencl, W. E., Shaywitz, B. A., Shaywitz, S. E., Fulbright, R. K., Constable,
R. T., Skudlarski, P., Marchione, K. E., Jenner, A. R., Fletcher, J. M., Liberman,
A. M., Shankweiler, D. P., Katz, L., Lacadie, C., & Gore, J. C. (2000). The
angular gyrus in developmental dyslexia: task-specific differences in functional
connectivity within posterior cortex. Psychological Science, 11, 51-56.
Pugh, K. R., Shaywitz, B. A., Shaywitz, S. E., Shankweiler, D. P., Katz, L., Fletcher, J.
M., Skudlarski, P., Fulbright, R. K., Constable, R. T., Bronen, R. A., Lacadie, C.,
& Gore, J. C. (1997). Predicting reading performance from neuroimaging profiles:
the cerebral basis of phonological effects in printed word identification. J
Experimental Psychology: Human Perception and Performance, 23, 299-318.
Qiu, D., Tan, L. H., Zhou, K., & Khong, P. L. (2008). Diffusion tensor imaging of normal
white matter maturation from late childhood to young adulthood: voxel-wise
123
evaluation of mean diffusivity, fractional anisotropy, radial and axial diffusivities,
and correlation with reading development. Neuroimage, 41, 223-232.
Ramus, F. (2001). Outstanding questions about phonological processing in dyslexia.
Dyslexia, 7, 197-216.
Rauschecker, A. M., Deutsch, G. K., Ben-Scachar, M., Schwartzman, A., Perry, L. M., &
Dougherty, R. F. (2009). Reading impairment in a patient with missing arcuate
fasciculus. Neuropsychologia, 47, 180-194.
Richards, T., Berninger, V., Winn, W., Stock, P., Wagner, R., Muse, A., & Maravilla, K.
(2007). Functional MRI activation in children with and without dyslexia during
pseudoword aural repeat and visual decode: before and after treatment.
Neuropsychology, 21, 732-741.
Rosazza, C., Cai, Q., Minati, L., Paulignan, Y., & Nazir, T. A. (2009). Early involvement
of dorsal and ventral pathways in visual word recognition: an ERP study. Brain
Research, 1272, 32-44.
Rose, S. E., Chen, F., Chalk, J. B., Zelaya, F. O., Strugnell, W. E., Benson, M., Semple,
J., & Doddrell, D. M. (2000). Loss of connectivity in Alzheimer’s disease: an
evaluation of white matter tract integrity with colour coded MR diffusion tensor
imaging. Journal of Neurology, Neurosurgery, and Psychiatry, 69, 528-530.
Rota, G., Sitaram, R., Veit, R., Erb, M, Weiskopf, N., Dogil, G., & Birbaumer, N. (2009).
Self-regulation of regional cortical activity using real-time fMRI: the right inferior
frontal gyrus and linguistic processing. Human Brain Mapping, 30, 1605-1614.
Ruff, I., Blumstein, S. E., Myers, E. B., & Hutchison, E. (2008). Recruitment of anterior
and posterior structures in lexical-semantic processing: an fMRI study comparing
implicit and explicit tasks. Brain and Language, 105, 41-49.
Rumsey, J. M., Horwitz, B., Donohue, B. C., Nace, K., Maisog, J. M., & Andreason, P.
(1997). Phonological and orthographic components of word recognition. A PET-
rCBF study. Brain, 120, 739-59.
Rypma, B., & D’Esposito, M. (1999). The roles of prefrontal brain regions in
components of working memory: effects of memory load and individual
differences. Proceedings of the National Academy of Sciences, 96, 6558.
Sabatini, J. P. (2002). Efficiency in word reading of adults: ability group comparisons.
Scientific Studies of Reading, 6, 267-298.
Salthouse, T. A. (1996). The processing speed theory of adult age differences in
cognition. Psychology Review, 103, 403.
124
Savage, R. S., Frederickson, N., Goodwin, R., Patni, U., Smith, N., & Tuersley, L.
(2005). Relationships among rapid digit naming, phonological processing, motor
automaticity, and speech perception in poor, average, and good readers and
spellers. Journal of Learning Disabilities, 38, 12-28.
Saxe, R., Brett, M., & Kanwisher, N. (2006). Divide and conquer: a defense of functional
localizers. Neuroimage, 30, 1088-1096.
Scarborough, H. S. (1984). Continuity between childhood dyslexia and adult reading.
British Journal of Psychology, 75, 329-348.
Schlaggar, B. L., & McCandliss, B. D. (2007). Development of neural systems for
reading. Annual Review of Neuroscience, 30, 475-503.
Schmithorst, V. J., Wilke, M., Dardzinski, B. J., & Holland, S. K. (2005). Cognitive
functions correlate with white matter architecture in a normal pediatric
population: a diffusion tensor MRI study. Human Brain Mapping, 26, 139-147.
Seghier, M. L., Lee, H. L., Schofield, T., Ellis, C. L., & Price, C. J. (2008). Inter-subject
variability in the use of two different neuronal networks for reading aloud familiar
words. Neuroimage, 42, 1226-1236.
Semenza, C., Bencini, G. M., Bertella, L., Mori, I., Pignatti, R., Ceriani, F., Cherrick, D.,
Caldognetto, E. M. (2007) A dedicated neural mechanism for vowel selection: A
case of relative vowel deficit sparing number lexicon. Neuropsychologia, 45, 425-
430.
Sergent, J., Zuck, E., Levesque, M., & MacDonald, B. (1992). Positron emission
tomography study of letter and object processing: empirical findings and
methodological considerations. Cerebral Cortex, 2, 68-80.
Shankweiler, D., Lundquist, E., Dreyer, L. G., & Dickinson, C. C. (1996). Reading and
spelling difficulties in high school students: Causes and consequences. Reading
and Writing, 8, 267-294.
Share, D. L. (1995). Phonological recoding and self-teaching: Sine qua non of reading
acquisition. Cognition, 55, 151-218.
Shaywitz, B. A., Shaywitz, S. E., Pugh, K. R., Constable, R. T., Skudlarski, P., Fulbright,
R. K., Bronen, R. A., Fletcher, J. M., Shakweiler, D. P., Katz, L., & Gore, J. C.
(1995). Sex differences in the functional organization of the brain for language.
Comment in Nature, 373, 607-609.
Shaywitz, B. A., Shaywitz, S. E., Pugh, K. R., Mencl, W. E., Fulbright, R. K., Skudlarski,
P., Constable, R. T., Marchione, K. E., Fletcher, J. M., Lyon, G. R. & Gore, J. C.
(2002). Disruption of posterior brain systems for reading in children with
developmental dyslexia. Biological Psychiatry, 52, 101-110.
125
Shaywitz, S. E., Escobar, M. D., Shaywitz, B. A., Fletcher, J. M., & Makuch, R. (1992).
Evidence that dyslexia may represent the lower tail of a normal distribution of
reading ability. New England Journal of Medicine, 326, 145-150.
Shaywitz, S. E., Shaywitz, B. A., Fulbright, R. K., Skudlarski, P., Mencl, W. E.,
Constable, R. T., Pugh, K. R., Holahan, J. M., Marchione, K. E., Fletcher, J. M.,
Lyon, G. R., & Gore, J. C. (2003). Neural systems for compensation and
persistence: young adult outcome of childhood reading disability. Biological
Psychiatry, 54, 25-33.
Silani, G., Frith, U., Demonet, J. F., Fazio, F., Perani, D., Price, C., Frith, C. D., &
Paulesu, E. (2005). Brain abnormalities underlying altered activation in dyslexia:
a voxel based morphometry study. Brain, 128, 2453-2461.
Silveri, M. C., & Cappa, A. (2003). Segregation of the neural correlates of language and
phonological short-term memory. Cortex, 39, 913-925.
Simos, P. G., Breier, J. I., Fletcher, J. M., Foorman, B. R., Mouzaki, A., & Papanicolaou,
A. C. (2001). Age-related changes in regional brain activation during
phonological decoding and printed word recognition. Developmental
Neuropsychology, 19, 191-210.
Simos, P. G., Breier, J. I., Wheless, J. W., Maggio, W. W., Fletcher, J. M., Castillo, E.
M., & Papanicolaou, A. C. (2000). Brain mechanisms for reading: the role of the
superior temporal gyrus in word and pseudoword naming. Neuroreport, 11, 2443-
2447.
Simos, P. G., Fletcher, J. M., Bergman, E., Breier, J. I., Foorman, B. R., Castillo, E. M.,
Davis, R. N., Fitzgerald, M., & Papanicolaou, A. C. (2002). Dyslexia-specific
brain activation profile becomes normal following successful remedial training.
Neurology, 58, 1203-1213.
Smith, S. M. (2002). Fast robust automated brain extraction. Human Brain Mapping, 17,
143-155.
Snowling, M. J., (2001). From language to reading and dyslexia. Dyslexia, 7, 37-46.
Snowling, M., Goulandris, N., & Stackhouse, J. (1994). Phonological constraints on
learning to read: Evidence from single-case studies of reading difficulty. In C.
Hulme and M. Snowling (Eds), Reading Development and Dyslexia, pp. 86-104.
Whurr: London.
Stanberry, L. I., Richards, T. L., Berninger, V. W., Nandy, R. R., Aylward, E. H.,
Maravilla, K. R., Stock, P. S., & Cordes, D. (2006). Low-frequency signal
changes reflect differences in functional connectivity between good readers and
dyslexics during continuous phoneme mapping. Magnetic Resonance Imaging,
24, 217-229.
126
Stanovich, K. E. (1986). Matthew effects in reading: some consequences of individual
differences in the acquisition of literacy. Reading Research Quarterly, 21, 360-
407.
Stanovich, K. E., & Siegel, L. S. (1994). Phenotypic performance profile of children with
reading disabilities: A regression-based test of the phonological-core cariable-
difference model. Journal of Educational Psychology, 86, 24-53.
Stanovich, K. E., & West, R. F. (1989). Exposure to print and orthographic processing.
Reading Research Quarterly, 24, 402-433.
Stanovich, K. E., Nathan, R. G., & Vala-Rossi, M. (1986). Developmental changes in the
cognitive correlates of reading ability and the developmental lag hypothesis.
Reading Research Quarterly, 21, 267-283.
Steinbrink, C. Vogt, K., Kastrup, A., Müller, H. P., Juengling, F. D., Kassubek, J., &
Riecker, A. (2008). The contribution of white and gray matter differences to
developmental dyslexia: insights from DTI and VBM at 3.0 T. Neuropsychologia,
46, 3170-3178.
Steinbrink, C., Vogt, K., Kastrup, A., Müller, H. P., Juengling, F. D., Kassubek, J., &
Riecker, A. (2008). The contribution of white and gray matter differences to
developmental dyslexia: insights from DTI and VBM at 3.0 T. Neuropsychologia,
46, 3170-3178.
Stringer, R., & Stanovich, K. E. (2000). The connection between reaction time and
variation in reading ability: Unravelling covariance relationships with cognitive
ability and phonological sensitivity. Scientific Studies of Reading, 4, 41-53.
Talairach, J., & Tournoux, P. (1988). A Co-planar Sterotaxic Atlas of a Human Brain.
Thieme, New York.
Taylor, K. I., & Regard, M. (2003). Language in the right cerebral hemisphere:
contributions from reading studies. News in Physiological Sciences, 18, 257-261.
Temple, E., Poldrack, R. A., Salidis, J., Deutsch, G. K., Tallal, P., Merzenich, M. M., &
Gabrieli, J. D. (2001). Disrupted neural responses to phonological and
orthographic processing in dyslexic children: an fMRI study. Neuroreport, 12,
299-307.
Torgesen, J.K., Wagner, R.K., & Rashotte, C.A. (1999). Test of Word Reading
Efficiency Austin, TX: Pro-Ed.
Turkeltaub, P. E., Flowers, D. L., Varbalis, A., Miranda, M., Gareau, L., & Eden, G. F.
(2004). The neural basis of hyperlexic reading: an fMRI case study. Neuron, 41,
11-25.
127
Turkeltaub, P. E., Gareau, L., Flowers, D. L., Zeffiro, T. A., & Eden, G. F. (2003).
Development of neural mechanisms for reading. Nature Neuroscience, 6, 767-
773.
vanAtteveldt, N., Formisano, E., Goebel, R., & Blomert, L. (2004). Integration of letters
and speech sounds in the human brain. Neuron, 43, 271-282.
Vigneau, M., Beaucousin, V., Herve, P. Y., Duffau, H., Crivello, F., Houde, O., Mazoyer,
B., & Tzourio-Mazoyer, N. (2006). Meta-analyzing left hemisphere language
areas: phonology, semantics, and sentence processing. Neuroimage, 30, 1414-
1432.
Vul, E., Harris, C., Winkielman, P., & Pashler, H. (2009). Puzzlingly high correlations in
fMRI studies of emotion, personality, and social cognition. Perspectives on
Psychological Science, 4, 274-290.
Wagner, R. K., & Torgesen, J. K. (1987). The nature of phonological processing and its
causal role in the acquisition of reading skills. Psychological Bulletin, 101, 192-
212.
Wagner, R. K., Torgesen, J. K., Rashotte, C. A., Hecht, S. A., Barker, T. A., Burgess, S.
R., Donahue, J., & Garon, T. (1997). Changing relations between phonological
processing abilities and word-level reading as children develop from beginning to
skilled readers: A 5-year longitudinal study. Developmental Psychology, 33, 468-
479.
Watson, B. U., & Miller, T. K. (1993). Auditory perception, phonological processing, and
reading ability/disability. Journal of Speech and Hearing Research, 36, 850-863.
Weiderholt, J.L., Bryant, G. (1992). Gray Oral Reading Tests Third Edition. Austin, TX:
Pro-Ed.
Wernicke, C. (1874). The aphasic symptom complex: a psychological study on a
neurological basis. Boston Studies in the Philosophy of Science, 4.
Wiederholt, J. L., & Bryant, B. R. (1992). Gray Oral Reading Tests, Third Edition.
Austin, TX: Pro-Ed.
Wildgruber, D., Ackermann, H., Kreifelts, B., & Ethofer, T. (2006). Cerebral processing
of linguistic and emotional prosody: fMRI studies. Progress in Brain Research,
156, 249-268.
Woodcock, R.W. (1998). Woodcock Reading Mastery Test-Revised. Circle Pines, MN:
American Guidance Service.
Woodcock, R.W., McGrew, K.S., & Mather, N. (2001). Woodcock-Johnson III. Itasca,
IL: Riverside Publishing.
128
Xu, J., Kemeny, S., Park, G., Frattali, C., & Braun, A. (2005). Language in context:
emergent features of word, sentence, and narrative comprehension. Neuroimage,
25, 1002-1015.
Yarkoni, T., Speer, N. K., Balota, D. A., McAvoy, M. P., & Zacks, J. M. (2008). Pictures
of a thousand words: investigating the neural mechanisms of reading with
extremely rapid event-related fMRI. Neuroimage, 42, 973-987.
Zumberge, A. C., Baker, L. A., & Manis, F. R. (2007). Focus on words: A twin study of
reading and inattention. Behavior Genetics, 37, 284-293.
Abstract (if available)
Abstract
Studies of reading development and disability have identified phonological skill as a major contributor to reading ability in children. The relationship seems to hold across a wide range of adult, skilled readers, as well, but its neural basis has yet to be identified. In order to examine the predictive power of phonological ability on adult reading ability, three studies were conducted in which both behavioral and neural indices of phonological processing were correlated with measures of oral reading fluency and silent reading comprehension. The first study investigated the influence of fMRI activity within the phonological processing network during a self-paced passage reading task. Regions of interest for phonological decoding were identified on an individual basis, using subjects’ performance on a phonological lexical decision task. Activity during the natural passage-reading task in the left occipitotemporal region, posterior superior temporal gyrus, supplementary motor area, and anterior inferior frontal gyrus was positively correlated with reading measures, providing neural evidence for the relationship between phonological processing and adult reading ability. The second study aimed to identify interregion connectivity pathways that contribute to differences in reading ability. By comparing timecourses of activation within the individually localized regions of interest, it was discovered that superior readers showed more robust connections between two of the regions. Connection strength between posterior superior temporal gyrus and anterior inferior frontal gyrus was positively correlated with word recognition ability. In the third study, diffusion tensor imaging was performed on a larger sample of adult subjects to assess structural connectivity contributions to reading ability. Using a semi-automated tracing method, FA values of the corpus collosum and 7 bilateral tracts previously reported to be associated with reading ability were recorded.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Neurobiological correlates of skilled and disabled reading
PDF
The angry brain: neural correlates of interpersonal provocation, directed rumination, trait direct aggression, and trait displaced aggression
PDF
Examining the neuroanatomical correlates of reading in developmental dyslexia
PDF
Cognitive-linguistic factors and brain morphology predict individual differences in form-sound association learning: two samples from English-speaking and Chinese-speaking university students
PDF
Brain and behavior correlates of intrinsic motivation and skill learning
PDF
Behavioral and neural influences of interoception and alexithymia on emotional empathy in autism spectrum disorder
PDF
Biological and behavioral correlates of emotional flexibility and associations with exposure to family aggression
PDF
Neural correlates of learning and reversal of approach versus withdraw responses to in- and out-group individuals
PDF
Functional models of fMRI BOLD signal in the visual cortex
PDF
Affective neuropsychiatric symptoms and neural connectivity in the early stages of Alzheimer’s disease
PDF
Heart, brain, and breath: studies on the neuromodulation of interoceptive systems
PDF
Contextual interference in motor skill learning: an investigation of the practice schedule effect using transcranial magnetic stimulation (TMS)
PDF
Behabioral and neural evidence of state-like variance in intertemporal decisions
PDF
Dedifferentiation of emotion regulation strategies in the aging brain: an MVPA investigation
PDF
The brain's virtuous cycle: an investigation of gratitude and good human conduct
PDF
Individual differences in heart rate response and expressive behavior during social emotions: effects of resting cardiac vagal tone and culture, and relation to the default mode network
Asset Metadata
Creator
Orechwa, Allison Zumberge
(author)
Core Title
The neural correlates of skilled reading: an MRI investigation of phonological processing
School
College of Letters, Arts and Sciences
Degree
Doctor of Philosophy
Degree Program
Neuroscience
Publication Date
07/28/2009
Defense Date
06/17/2009
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
cognitive psychology,fMRI,OAI-PMH Harvest,phonological processing,Reading
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Manis, Franklin R. (
committee chair
), Andersen, Elaine (
committee member
), Lu, Zhong-Lin (
committee member
), Mintz, Toben (
committee member
), Pancheva, Roumyana (
committee member
)
Creator Email
a.z.orechwa@gmail.com,azumberg@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-m2397
Unique identifier
UC1308855
Identifier
etd-Orechwa-3056 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-406639 (legacy record id),usctheses-m2397 (legacy record id)
Legacy Identifier
etd-Orechwa-3056.pdf
Dmrecord
406639
Document Type
Dissertation
Rights
Orechwa, Allison Zumberge
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Repository Name
Libraries, University of Southern California
Repository Location
Los Angeles, California
Repository Email
cisadmin@lib.usc.edu
Tags
cognitive psychology
fMRI
phonological processing