Abstract

Successful behavioral genetic studies require precise definition of a homogenous phenotype. This study searched for anatomical markers that might restrict variability in the reading disability phenotype. The subjects were 15 college students (8 male/7 female) diagnosed with a reading disability (RD) and 15 controls (8 males/7 females). All subjects completed a cognitive and reading battery. Only 11 of the RD subjects had a phonological deficit [phonological dyslexia (PD): pseudo word decoding scores < 90 (27th percentile)]. Thirteen RD (9 PD) and 15 controls received a volumetric MRI scan. Four anatomical measures differentiated the PD group from the remainder of the subjects: (i) marked rightward cerebral asymmetry, (ii) marked leftward asymmetry of the anterior lobe of the cerebellum, (ii) combined leftward asymmetry of the planum and posterior ascending ramus of the sylvian fissure, and (iv) a large duplication of Heschl's gyrus on the left. When these four measures were normalized and summed, the resulting variable predicted short- and long-term phonological memory. By contrast, oral and written comprehension skills were predicted by a different anatomical variable: low cerebral volume. These findings provide neurobiological support for an RD phenotype characterized by phonological deficits in the presence of normal or superior comprehension. The study of individual variation in cortical structure may provide a useful link between genotype and behavior.

Introduction

All normal humans learn to communicate with signs, but only some become fluent readers. Even in literate cultures, a substantial minority of adults read slowly, with considerable effort and rarely for pleasure. Some of these individuals may, however, complete high levels of professional education (Bruck, 1992). In the public school system, children are diagnosed as reading disabled (RD) when their reading achievement is discrepant with aptitude or measured intelligence (Fletcher et al., 1994). This criterion serves to distinguish RD from a more serious disorder, specific language impairment (SLI), where reading ability is not discrepant with other language abilities. In SLI, in spite of the term ‘specific’, measured intelligence is usually depressed and the long-term prognosis is poor (Bishop and Adams, 1990).

Phonological Processing

A large body of research has suggested that slow and inaccurate processing of the sound structure of language (phonology) is the core deficit in discrepant readers (Shankweiler et al., 1995) and they are frequently referred to as phonological dyslexics (PDs). A consensus is gradually building, however, that pure phonological deficits are rare and that most poor readers have deficits in additional linguistic areas such as semantics and syntax. As Bishop and Adams stated so eloquently:

Phonological factors are of particular theoretical interest because they seem to explain variation in reading acquisition that is not accounted for in terms of other, more general, verbal abilities. However, it should be emphasized that other language skills exert the major influence on reading progress. (Bishop and Adams, 1990, p. 1046)

Recent research has backed up this prescient statement. The contribution of oral language skills (Lombardino et al., 1997; Catts et al., 1999) and general intelligence (Stringer and Stanovich, 2000) to reading skill is not explained by a phonological deficit. Most poor readers, according to Manis and his colleagues (Manis et al., 1996), do not exhibit discrepant profiles, they are just ‘garden variety poor readers’ (Stanovich, 1988).

Poor readers with linguistic deficits share many cognitive deficits with SLI. Whether PDs are part of that continuum is not clear. Tallal and her colleagues have proposed that all reading and language disorders stem from nonlinguistic auditory and visual temporal processing deficits (Tallal et al., 1993; Albert et al., 1999). But the specificity of the link between nonlinguistic auditory discrimination and phonological ability is yet to be established (Mody et al., 1997).

Resolution of this issue is critical to the design of informative genetic and neurobiological studies. Heterogeneous phenotypes make such studies uninterpretable. In a recent article, Andreason and her colleagues suggested that identification of neuroanatomical variants might provide a link between genotype and behavioral phenotype in complex disorders such as schizophrenia (Wassink et al., 1999). In the present study we applied that strategy to RD phenotypes.

Anatomical Study of Language Disorders

Sagittal magnetic resonance images accurately represent individual cortical variation in the size and shape of the perisylvian region (Steinmetz et al., 1989; Leonard et al., 1998). Leftward asymmetry in these regions is associated with hemispheric dominance, handedness, reading ability in children, and verbal and musical ability in adults (Geschwind and Levitsky, 1968; Steinmetz and Seitz, 1991; Foundas et al., 1994; Schlaug et al., 1995; Leonard et al., 1996; Rumsey et al., 1997; Maron et al., 2000; Eckert et al., 2001). Many children with SLI have symmetry or rightward asymmetry (Jernigan et al., 1990; Gauger et al., 1997) although there is one negative report (Preis et al., 1998). Imaging studies of dyslexia, on the other hand, do not report symmetry (Schultz et al., 1994; Filipek, 1995; Rumsey et al., 1997; Best and Demb, 1999). In fact, in one early study, gyral duplications and marked leftward asymmetry were found (Leonard et al., 1993). This preliminary evidence that symmetry is characteristic of SLI but not dyslexia implies that the two conditions may not share a common etiology or neurobiological substrate.

The present study had three aims: (i) to replicate earlier findings of duplicated gyri and leftward asymmetry in a comprehensively tested sample of RD; (ii) to measure additional structures such as brain volume and the cerebellum, which might help characterize a consistent anatomical and behavioral phenotype; (iii) to determine if anatomical variation predicted cognitive deficits associated with RD.

Materials and Methods

Subjects

The main sample consisted of 30 college students, 28 of whom received magnetic resonance imaging (MRI) scans. Fifteen students (8 men/7 women) were certified as RD by the University of Florida Office of Student Services. These RD subjects were matched as a group with 15 controls on the fluid reasoning cluster of the Woodcock Johnson test of Cognitive Abilities–Revised (WJ-Cog) (Woodcock and Johnson, 1989), sex, and a quantitative measure of handedness (QHP) (Briggs and Nebes, 1974). All subjects signed an informed consent form and were paid $30 for participation. They completed the fluency portion of the Gray Oral Reading Test (GORT) (Wiederholt and Brayant, 1992), the Wide Range Achievement Tests of Spelling (WRAT-SP) and Reading (WRAT-RD) (Jastak and Wilkinson, 1993), and the Woodcock Reading Mastery tests [Word Attack (WA), a test of pseudo word reading; Word Identification (WI), a test of single word reading, and Passage Comprehension (PC), a test requiring a ‘fill in the blank’ response] (Woodcock, 1987).

The WJ-Cog was used to measure cognitive abilities (Woodcock and Johnson, 1989). Cognitive ability is multidimensional and cannot be reflected in a single intelligence quotient (IQ). On most IQ tests, only two or three factors can be extracted due to the limited number and diversity of their subtests (Carroll, 1993; Keith and Witta, 1997). The WJ-Cog, however, measures seven moderately independent factors: fluid reasoning, crystallized intelligence, visual and auditory processing, processing speed, and short- and long-term memory (McGrew and Murphy, 1995; Kranzler, 1997). The WJ-Cog is psychometrically sound, having been normed on a sample of more than 6000 individuals aged 5–90 (Reschly, 1990).

MRI

All control subjects and 13 of the original 15 RD subjects received a volumetric MRI scan in a GE 3 T scanner at the Veterans' Administration Medical Center. Parameters were TR = 10 ms, TE = 2.9 ms, flip angle = 20°, matrix = 256 × 128 × 128, image voxel size = 0.94 × 0.94 × 1.3 mm. Headers containing identifying characteristics were removed, images were concatenated into a series, assigned a randomly chosen blind numbers and reformatted into 1 mm thick slices in the Talairach planes. Two raters blind to subject characteristics made each measurement and differences were resolved by discussion.

Principles Guiding Choice of Anatomical Measures

Cerebral size was measured because a number of studies have found that cerebral size is related to measured intelligence (Willerman et al., 1991; Reiss et al., 1996), an ability that contributes to reading proficiency (Stringer and Stanovich 2000). Hemispheric volume asymmetries were measured in light of a recent report (Sisodiya et al., 1995) that marked (>4%) hemispheric asymmetries were associated with a radiological diagnosis of dysplasia, a pathological finding that has characterized the brains of a few dyslexics examined post mortem. Cerebellar size and asymmetry were measured because of Fawcett and Nicolson's proposal that automatization deficits in dyslexia are accompanied by cerebellar mediated deficits in coordination and balance (Fawcett and Nicolson, 1994). The planum temporale of the temporal lobe was measured because of its reported association with language ability (Leonard et al., 1996; Rumsey et al., 1997). The surface area of Heschl's gyri were measured because duplications of these gyri are heritable and were found in a previous study of RD (Leonard et al., 1993; Eckert and Leonard 1999).

Technical Details

Cerebral Hemispheres

The volume of each cerebral hemisphere was measured by tracing the area enclosed by the dura on every fourth sagittal image, and summing the averages of adjacent areas after multiplying by the width of the inter image gap. (Volumes calculated with this method correlate 0.98 with volumes calculated from measurements of every image.) The midsection was traced twice and half the slab volume added to each hemisphere. This measure includes intra-sulcal CSF and thus reflects original cerebral capacity before the onset of age-related cortical loss. Inter-rater reliability of this measure is >0.9 (intra-class correlation). The coefficient of asymmetry of the cerebral hemispheres (coacer) was calculated by dividing the left/right difference by the average volume of the two hemispheres.

Cerebellar Hemispheres

The posterior lobe in each hemisphere was outlined on every other 1 mm thick axial section. Volumes were calculated by summing these areas after multiplication by the section gap (2 mm). Reliability for these measurements was 0.9. (Volumes calculated with this method agree 0.96 with volumes calculated from measurements of every image). The anterior lobes were measured in sagittal images. Every 1 mm thick section on which the primary fissure could be seen was outlined and the areas added to calculate the volume in each hemisphere. As the primary fissure becomes indistinct laterally, the lateral boundary of the anterior lobe was defined as the image on which the superior cerebellar vessels disappeared. Reliability for this measurement was 0.87. Separate coefficients of asymmetry were calculated for the anterior (coaant) and posterior lobes (coapost).

Planum Temporale

The surface area of the temporal bank of the planum (PT) was measured between the posterior boundary of the first transverse gyrus of Heschl (Heschl's sulcus), and the termination of the Sylvian fissure, which in most cases was marked by a small elevation in the planum (see Fig. 1) and a bifurcation into a descending ramus and the posterior ascending ramus (PAR). The small posterior descending ramus which originated from a bifurcation was not included in the measure. In cases where the posterior ascending ramus originated proximally to the termination of the sylvian fissure (inverted formation) (Ide et al., 1996) the large extent of sylvian fissure posterior to PAR was included in the planum measurement. The posterior ascending ramus was measured from the bifurcation to its dorsal termination. In the small number of cases where no elevation or bifurcation marked the origin of PAR, the ‘knife cut’ method was used (Witelson and Kigar, 1992). In cases where the sylvian fissure merged with the superior temporal sulcus or other occipitoparietal sulci, the planar measurement was terminated at the point of the merge.

An index of surface area was calculated by averaging the length measured between standard Talairach positions (x = 46 and 56 mm) as reported previously (Leonard et al., 1993, 1996; Foundas et al., 1994, 1995). The medial and lateral coordinates were chosen to maximize lateral asymmetry as well as reliability. Medial to x = 46, the origin of the parietal bank is ambiguous, whereas Heschl's sulcus frequently becomes indistinct lateral to x = 56. A recent study by Best and Demb found that asymmetry measures using this index agreed well with asymmetry measures gained by measuring the whole planum (Best and Demb, 1999). Reliability for these measurements was 0.85. Figure 1a,b gives examples of typical PT asymmetry, while an example of extreme leftward PT asymmetry (due to the absence of a PT posterior to Heschl's sulcus on the right) is shown in Figure 1c,d.

The values for the temporal bank and PAR were summed (PT+) and a coefficient of asymmetry was calculated. PT+ is generally symmetrical because leftward asymmetry in PT is mitigated by rightward asymmetry of PAR. Summing the coefficients of asymmetry for PT and PAR provides an index of the relative strengths and direction of the asymmetries of the two banks (sumppar).

Heschl's Gyri

The surface areas of the first Heschl's gyrus (H1) and, when present, a second gyrus (H2) were traced between their limiting sulci on consecutive sagittal images between x = 34 and x = 48. Examples of H1 and H2 are shown in Figure 1. Two operators measured 10 H1 and all H2 to determine inter-rater reliability. Intraclass correlations were 0.9 for H1 and 0.85 for H2.

Data Analysis

All variables were entered into spreadsheets and analyzed with PC-SAS. Means and standard deviations were calculated and the t-statistic (with Bonferroni correction) was calculated to test for significance of differences in group means. (The scores for the WRMT-WA were converted to ranks because of the presence of one extremely high score in the RD group). A paired t-test was used to test the significance of structural asymmetries in each group separately.

For variables where the distributions in the RD and controls groups differed, both nonparametric and parametric analyses were used to define a quantitative threshold that separated diagnosed individuals from controls (anatomical risk factor). Nonparametric analyses were used to examine the frequency of combinations of extreme values. Parametric analyses were used to incorporate information from the entire range of measurements.

In the nonparametric analysis, the anatomical measures were converted to ranks and the individuals with the seven scores in the quartile containing the tail of the distribution were defined as having an anatomical risk factor. In the parametric analyses, the anatomical measures were first entered into a discriminant analysis and then converted to z-scores using the control means and standard deviations and adjusting the sign to indicate a consistent direction of risk. The four z-scores were summed and plotted against behavioral variables. Pearson rs were calculated to characterize the strength of the relationships between anatomical and cognitive variables.

Results

There were four major results: (i) four anatomical measures distinguished RD subjects with a phonological deficit (PD) from the remaining subjects; (ii) these variables had an additive effect — all subjects with more than one extreme measure had a phonological deficit; (iii) a cumulative risk factor score predicted short- and long-term memory deficits for arbitrary phonological symbol associations; (iv) the cumulative risk factor score did not predict comprehension deficits, which were predicted by low cerebral volume.

Identification of these findings required separation of subjects with PD (n = 11, 9 of whom returned to be scanned) from the remainder of the RD group. A phonological deficit was defined as a pseudo-word decoding score (WRMT-WA) <90 (<27th percentile). The other four RD subjects had WRMT-WA scores in the average (100, 99, 101) or superior range (141). Since each one had a somewhat different cognitive profile but scored in the 27th percentile on at least one reading or spelling test, they will be referred to as URD (unspecified reading deficit).

The results section is organized as follows: (i) characterization of the two original samples and the PD subgroup; (ii) description of the anatomical characteristics that distinguish the PD subgroup; (iii) nonparametric categorical analysis of combinations of risk factors; (iv) parametric analysis of normalized and cumulative measures; (v) identification of a separate risk factor for comprehension deficits.

Comparison of Original RD Sample with Controls

The means and standard deviations for the demographic and behavioral variables on which the groups were matched are given in Table 1. The control and RD groups did not differ in age, handedness, male/female composition or fluid reasoning ability (WJ-Cog Test of Analysis/Synthesis).

The means and standard deviations for the anatomical variables in the RD and control group are given in Table 2. The RD and the controls differed (P < 0.05) on only one measurement (asymmetry of the posterior lobe of the cerebellum), a finding that does not survive correction for multiple comparisons.

Planar Asymmetry

As reported previously for a clinically diagnosed group of dyslexic adults and children (Leonard et al., 1993), the RD group as a whole tended to have more marked leftward asymmetry of the PT, although the group difference was not significant. Both the RD and control groups had significant leftward asymmetry of the planum temporale and the summed coefficients of PT and PAR (Table 2). There were no group differences in the ratios of the planum temporale to PAR on either the left or the right. Unlike previous reports, there was no evidence of a transfer of tissue from the PT to the PAR and the coefficients of asymmetry of PT, PAR and PT+ did not predict scores on any reading or cognitive test (Leonard et al., 1996; Rumsey et al.,1997; Eckert et al., 1998).

Heschl's Gyri

Both groups had a significant leftward asymmetry of the first but not the second transverse gyrus of Heschl (see Table 2). All subjects, with the exception of one control, had a larger H1 on the left than the right. This asymmetry is even more consistent than reported by Penhune and her colleagues (Penhune et al., 1996) and agrees with a previous post mortem study (Musiek and Reeves, 1990).

Anatomical Risk Factors for PD

Most subjects had leftward asymmetry of the PT and rightward asymmetry of PAR, a consistent finding in previous studies (Steinmetz et al., 1990; Leonard et al., 1993; Witelson and Kigar, 1992). Three of 15 controls, 2/4 URD and 0/9 PD had rightward asymmetry or symmetrical PT. Rightward asymmetry of both PT and PAR was found in one URD subject. Leftward asymmetry of both PT and PAR was found in five PD subjects, a highly unusual finding (Jäncke et al., 1997).

The PD group had greater rightward asymmetry of the cerebral hemispheres, greater leftward asymmetry of the anterior lobe of the cerebellum, greater leftward asymmetry of the summed planum and PAR and a larger surface area of H2 (Table 3). The three asymmetries and LH2 surface area are plotted against pseudo-word decoding score in Figure 2. Note that all four risk factors discriminate the nine PD subjects from those with URD. The URD subjects always cluster near symmetry or at the opposite asymmetry from the PD group.

Nonparametric Categorical Analysis of PD Risk Factors

Scores on the four anatomical variables were ranked and a PD risk factor was defined as a value falling in the bottom quartile. Table 4 gives the distribution of risk factors in the three adult groups. Not one of the URD subjects has a risk factor, half the controls have one risk factor, while 8/9 PD subjects have two or more risk factors [χ2 (df = 6) = 25.9, P < 0.0002]. Table 5 gives the relative frequency of different risk factor combinations.

Parametric Analysis of PD Risk Factors

Two analyses were performed to determine if there was a cumulative quantitative effect of brain structure on the risk of PD. First, scores from the measures graphed in Figure 2 were converted to standard scores with the sign adjusted so that positive scores would predict a phonological deficit. The standard scores were then summed for each individual. This new variable completely separates the PD and URD groups (see Figure 3 top). All nine PD subjects have positive scores. Only two controls and no URD subjects have positive scores.

To further determine if multiple anatomical variables distinguish PD subjects from controls and URD, a discriminant function analysis was performed using the four risk factor measures as classification variables. A jackknife method was used to avoid having an observation influence its own classification. The nine PD subjects were separated from the four URD and 14/15 controls [F(4,23) = 22.3, P < 0.0001]. Each variable contributed significantly to the classification.

Separate Anatomical Risk Factors for Coding and Comprehension Deficits

Pearson correlation coefficients were calculated between all anatomical variables, the cumulative risk factor score and the WJ-Cog Cluster scores. When significant correlations were found at the Cluster level, the Subtest correlations were examined to determine specificity. A consistent pattern emerged, with cerebral volume demonstrating significant correlations with a number of related abilities such as reasoning, comprehension and oral language skill. The cumulative risk factor score, by contrast, predicted deficits in Short-Term Memory and Long-Term Retrieval of arbitrary phonological information (Table 6). The relationship between reversed digit span (Numbers Reversed) and the cumulative risk factor score is plotted in Figure 3 (upper right).

Discussion

The two most interesting findings in the present study were that marked anatomical asymmetries and gyral duplications posed a cumulative risk for a phonological deficit, and that a different anatomical variable, low cerebral volume, predicted oral language and comprehension deficits. Although these relationships were not predicted, they are intuitively satisfying and consistent with an emerging consensus on the separability of coding and comprehension mechanisms. Identification of separate neurobiological substrates, if confirmed in future prospective experiments, would have important implications for diagnosis, remediation and neurobiological investigations of these mechanisms. The study of individual variation in cortical structure may provide a useful link between genotype and behavior.

Cumulative Risk

Imaging studies of the relation between cognitive function or behavioral disease and brain structure are hard to replicate (Stevens, 1997; Zuffante et al., 2000). The relations are rarely strong and frequently disappear with doubling of the sample size (Chua and McKenna, 1995). However, a recent study of schizophrenia found that while measurements of individual structures did not distinguish patients and controls, there was a cumulative effect of extreme anatomy. When cerebral volume, third ventricle volume and measures of temporal and frontal sulcal continuity were put into a discriminant analysis, 75% of the schizophrenics were properly classified (Leonard et al., 1999). These anatomical measures also combined to produce a highly accurate estimate of full-scale IQ. Patients with higher numbers of extreme measurements tended to have lower measured intelligence. The patients in that study were not given the WJ-Cog, so it was not possible to determine if different anatomical variables were related to particular cognitive abilities. Nevertheless, the findings are consistent with the idea that schizophrenia and other complex behavioral disorders evolve from the random combination of many factors, with each one conferring a relatively low risk in isolation.

This is the third study from our laboratory that has identified an elevated risk associated with the combination of a Heschl duplication and an anomalous sylvian fissure. In the first study, performance on a phonological processing test decreased as the number of perisylvian anomalies increased (Leonard et al., 1993). In the second study, male subjects with a mutation in the thyroid receptor gene were more likely to have multiple perisylvian anomalies than family members without a mutation (Leonard et al., 1995). In the present study, PDs had a higher risk for combinations of perisylvian anomalies. Future work should determine whether the likelihood of a behavioral diagnosis is elevated due to either specific genetic or cognitive alterations associated with each separate risk factor or nonspecific effects of genetic load (Waddington, 1957; Markow, 1994).

Separate Risk Factors for Comprehension and Coding Deficits

Subjects who maintained good comprehension in the presence of severe phonological deficits had extreme scores on measures of brain asymmetry and gyral duplication, while subjects whose phonological ability was consistent with their comprehension level did not. The three subjects with RD whose comprehension and phonology were unexpectedly low for college students had low cerebral volume and symmetrical brain structures. It may be overly concrete to suggest that a restriction in processing sites associated with low cerebral volume restricts the number of associations that can be made and the amount of knowledge that can be accumulated. It should be noted, however, that reduction of the number of nodes in a computational model is associated with semantic (comprehension) not phonological deficits (Manis et al., 1996).

The data for the subtests in the Short-Term Memory Cluster were particularly informative about the relation between anatomy and the ability to use contextual cues to support phonological associations. As the cues provided by context increase through the three subtests (Numbers Reversed < Memory for Words < Memory for Sentences), the influence of cerebral volume increases, performance becomes less discrepant with measured intelligence, and the influence of the PD risk factors becomes insignificant. If further work supports this proposal that different anatomical variables predict the ability to use contextual cues (comprehension) and memory for arbitrary phonological symbol associations (coding), it would provide a neurobiological explanation for the ability of some RD subjects to comprehend without being able to decode.

These anatomical findings add neurological support to the emerging consensus that two independent factors contribute to reading ability: phonology and general intelligence or oral language ability (Manis et al., 1996; Lombardino et al., 1997; Catts et al., 1999; Stringer and Stanovich, 2000). Children with SLI resemble the URD subjects in the present study, in that their oral language deficits are associated with perisylvian symmetry and low cerebral volume (Jernigan et al., 1990; Gauger et al., 1997; Preis et al., 1998). Symmetry and marked asymmetry may be complementary signs of disordered interhemispheric regulation during neural development. Disorders of interhemispheric regulation that result in symmetry are apparently associated with different cognitive consequences than disorders that result in marked asymmetries. It is even possible that marked asymmetries mitigate some of the negative consequences of low cerebral volume. For the subjects in the bottom quartile of cerebral volume, the mean Passage Comprehension score for subjects with risk factors was 16 points higher than that for subjects with no risk factors.

Reading and Language Disorders

The present results do not support a common etiology for all reading and language disorders. These results, together with previous studies in SLI (Bishop and Adams, 1990; Jernigan et al., 1990; Plante et al., 1991; Gauger et al., 1997), suggest that the neuroanatomical correlates and prognoses of phonological and comprehension deficits are qualitatively, not just quantitatively different. The cognitive abilities that make up the Oral Language and Fluid Reasoning clusters are predicted by cerebral volume (Table 6). Cerebral volume does not predict the Short-Term Memory and Long-Term Retrieval deficits that characterize the group of PD subjects.

Only Sound Patterns performance showed a dependence on both cerebral volume (0.41, P < 0.01) and the anatomical risk factors (–0.34, P < 0.08). This dual relationship may explain why auditory temporal processing is disturbed in both PD and SLI. Mody et al. have argued that the temporal order of judgment tasks (of which the Sound Patterns Subtest is an example) is not a ‘pure’ test of auditory processing because of the demands on working memory (Mody et al., 1997). Perhaps the working memory component makes the task difficult for phonologically impaired subjects while the auditory component makes the task difficult for subjects with oral language problems. Superficially similar deficits could arise through different mechanisms.

Anatomical Specificity

Heschl's Gyrus

A previous structural imaging study of dyslexia reported an elevated incidence of gyral duplications and missing gyri in both dyslexics and their family members. A second Heschl's gyrus was found in either the left or the right hemispheres of 4/9 members of the dyslexic group (Leonard et al., 1993). Elevated frequencies of H2 were also found in a study of families with a genetic mutation of the thyroid receptor associated with attention deficit disorder (Hauser et al., 1993) but the presence of H2 was not associated with the behavioral diagnosis of ADHD. Although the mutation is also associated with language problems, reading and phonological decoding scores were, unfortunately, not available.

A measure of sensory processing efficiency (visual inspection time) is slower in both college students and children with H2 in the left hemisphere (Grudnik et al., 2000). In the visual inspection test, two lines of different length are exposed briefly and the subject must press a button below the longer line (Nettelbeck, 1982, 1987). Reaction time and movement time are not included in the measure. Future work could investigate whether a long inspection time impedes response automatization and increases the error rate in arbitrary phonological associations, such as those between graphemes and phonemes. Since the presence of an H2 runs in families (Leonard et al., 1995; Eckert and Leonard, 1999), investigation of its heritability has the potential to provide an informative phenotypic marker for genetic studies of dyslexia.

Marked Asymmetries

Several structural asymmetries were examined in this study, each justified by a different line of previous research. In a previous study of epilepsy, marked asymmetry was associated with cortical dysplasia but not hippocampal sclerosis Sisodyaet al., 1995). The fact that marked cerebral asymmetries were found in some PD subjects in the present study does not necessarily imply that they have cortical dysplasia. Marked asymmetries may represent one extreme of a developmental continuum. Such asymmetry could even result from compensatory neuroplasticity and signify that the developmental relationship between the two hemispheres is under flexible control.

Although most theoretical work in dyslexia has focused on auditory cortex in the left hemisphere, one group (Nicolson and Fawcett, 1990) has argued for a critical role of timing mechanisms in the cerebellum. The anterior lobe is associated with spinal and brainstem pathways controlling eye movements, balance and postural control (Bastian et al., 1999), while the posterior lobe, by contrast, acts as a modulator of voluntary cortical activity. The RD subjects differed from the controls in mean asymmetry of both the posterior and anterior lobes. Asymmetry of the posterior lobe was the only measure that distinguished reading disability on which the distributions of URD and PD overlapped. The group difference did not survive correction for multiple comparisons and needs verification in a prospective study. By contrast, asymmetry of the anterior lobe made a significant contribution to the discriminant function that differentiated PD from the remainder of the sample. The three PD subjects with the most extreme leftward asymmetries were ambidextrous males whose scores on short-term phonological memory were in the bottom quartile. Future work should investigate whether there is an interaction between ambidextrality, phonological automatization and marked anterior lobe asymmetry that is modulated by sex. Functional imaging studies that visualize anterior lobe activity during speeded testing in RD subjects might prove enlightening.

Sylvian Fissure Asymmetry

It is remarkable that the PD subjects had marked leftward asymmetry of the horizontal and vertical banks of the sylvian fissure. How can this finding be reconciled with reports of leftward planar asymmetry predicting verbal and reading skill (Leonard et al., 1996; Rumsey et al., 1997)? Data from studies in our laboratory suggest a possible answer (Maron et al., 2000; Eckert et al., 2001). Perhaps planar asymmetry is more closely associated with comprehension than pure phonological ability. Such a relationship would explain why reversed planar asymmetry is more frequently reported in studies of children with SLI than dyslexia (Eckert and Leonard, 2000). It is somewhat paradoxical that planar asymmetry is related to comprehension, as fMRI studies report equivalent activation in the planum during linguistic and nonlinguistic tasks (Binder et al., 1996). It is possible that good comprehension is associated with information processing abilities directed by the right parietal lobe and that a short right sylvian fissure allows increased right parietal dominance. Future fMRI studies could investigate possible information processing differences in individuals with large and small parietal asymmetries.

The summed asymmetry of the planum and PAR is a novel measure of hemispheric relationships and its usefulness needs prospective confirmation. We created this index after Jäncke et al.'s report of parallel leftward asymmetry of these structures in an individual with callosal agenesis who had normal intelligence and language ability (Jäncke et al., 1997). None of the 200 normal individuals in the Düsseldorf database had ‘substantial’ leftward asymmetry of these structures. Remarkably, five PD subjects in the present study had such parallel asymmetries. The fact that the acallosal patient had normal language function suggests the possibility that marked asymmetry might be associated with the ability to develop compensatory strategies.

Strengths and Limitations of the Study

The most serious limitation is the post hoc nature of this analysis. The subjects were separated into PD and URD groups after extreme anatomical measures were found to cluster in subjects with poor pseudo-word decoding (Word Attack). We gained confidence in the analysis upon finding that the anatomical measures that characterized the PD group also predicted deficits in short- and long-term phonological memory, skills that could reasonably be expected to facilitate the routine automatic encoding and retrieval of grapheme–phoneme associations. The emergence of a consistent story somewhat allayed concern over the post hoc initial step of the data analysis. The strength of the results, as well as the pracical implications for treatment and remediation, if the findings are confirmed, encouraged us to seek publication, so that other groups could examine their data for similar relationships.

The second limitation is that the study is of college students, an unrepresentative sample of the reading disabled, and it is questionable how the results will generalize to the school age population. Our own sample of 104 normal children provides limited support for the findings presented here: (i) the incidence of all the risk factors was low, with the exception of cerebellar asymmetry; (ii) there were very few cases of risk factor combinations and (iii) cerebral volume predicted text comprehension (C.M. Leonard, unpublished data).

Strengths of the study include the extensive battery of well-normed, theoretically motivated tests, the high-resolution volumetric scan and the well-matched groups. Future studies will demonstrate whether the separate anatomical risk factors identified here will prove useful for diagnosis and the design of remediation.

Notes

The authors are very grateful to Maria Saravanos and Harrison Kane for recruiting and testing the subjects, Anand Patel and Leila Mufdi for assistance with anatomical measurements, and John M. Kuldau and two anonymous reviewers for comments on an earlier version. The work was supported by an interdisciplinary grant to T.O. from the University of Florida Office of Research, Graduate Education and Technology and NIDCD R01 02922 to C.M.L. and L.J.L.

Address correspondence to Christiana M. Leonard, PO Box 100244, Department of Neuroscience, University of Florida Brain Institute, Gainesville, FL 32611, USA. Email: leonard@ufbi.ufl.edu.

Table 1

Means and standard deviations for matching and diagnostic variables for subjects receiving scans

 Control (8 M/7 F) RD (7 M/6 F) PD subgroup (5 M/4 F) 
aQHP: Handedness preference ratio varied from 0 (no preference) to 1.0 (extremely right handed) in the adults and from –1 to 1 in the children. All adults wrote with their right hand. 
bTests are identified in Materials and Methods. 
cMean significantly different from controls, P < 0.0001. 
Age (years)  22 ± 3  24 ± 3  24 ± 4 
QHPa 0.69 ± 0.26 0.59 ± 0.34 0.64 ± 0.24 
WJ-Cog fluid reasoning cluster    
Analysis/synthesis 108 ± 9 103 ± 12 105 ± 13 
Concept formation 110 ± 10 108 ± 15 110 ± 9 
Reading testsb    
Comprehension 126 ± 8 118 ± 19 122 ± 16 
Pseudo word 120 ± 7  91 ± 17c  83 ± 4c 
Single word 115 ± 8  92 ± 7c  91 ± 7c 
Fluency  15 ± 0.9 9 ± 2c 9 ± 2c 
Spelling 110 ± 6 89 ± 10c  87 ± 10c 
 Control (8 M/7 F) RD (7 M/6 F) PD subgroup (5 M/4 F) 
aQHP: Handedness preference ratio varied from 0 (no preference) to 1.0 (extremely right handed) in the adults and from –1 to 1 in the children. All adults wrote with their right hand. 
bTests are identified in Materials and Methods. 
cMean significantly different from controls, P < 0.0001. 
Age (years)  22 ± 3  24 ± 3  24 ± 4 
QHPa 0.69 ± 0.26 0.59 ± 0.34 0.64 ± 0.24 
WJ-Cog fluid reasoning cluster    
Analysis/synthesis 108 ± 9 103 ± 12 105 ± 13 
Concept formation 110 ± 10 108 ± 15 110 ± 9 
Reading testsb    
Comprehension 126 ± 8 118 ± 19 122 ± 16 
Pseudo word 120 ± 7  91 ± 17c  83 ± 4c 
Single word 115 ± 8  92 ± 7c  91 ± 7c 
Fluency  15 ± 0.9 9 ± 2c 9 ± 2c 
Spelling 110 ± 6 89 ± 10c  87 ± 10c 
Table 2

Means and standard deviations for brain structure size and asymmetry in the 13 RD and 15 controls (leftward asymmetries are positive)

Measure  Control (n = 15) RD (n = 13) 
aMean coefficient significantly different from 0.0, P < 0.0005 (paired t-test). 
bMean coefficient significantly different from 0.0, P < 0.005. 
cMean coefficient significantly different from 0.0, P < 0.05. 
d Mean significantly different from controls, P < 0.05. 
Abbreviations: Coaant, asymmetry – cerebellar anterior lobe; coacer, asymmetry – cerebral hemispheres; H1, H2, first and second tranverse gyrus of Heschl; PT, planum temporale; PAR, posterior ascending ramus; PT+, planum + PAR; sumppar, summed coefficients of asymmetry, PT, PAR. 
Cerebral memisphere (cm3580 ± 43 561 ± 59 
 583 ± 47 576 ± 64 
 coacer –0.005 ± 0.03 –0.026 ± 0.04a 
Cerebellum: anterior Lobe (cm36.2 ± 1.1 6.0 ± 1.1 
 6.8 ± 1.3 6.0 ± 1.8 
 coaant –0.08 ± 0.2a 0.02 ± 0.28 
Posterior lobe (cm380 ± 6 78 ± 8 
 82 ± 5 82 ± 9 
 coapost  –0.03 ± 0.04 c  –0.06 ± 0.04 a d 
Planum temporale (cm) 3.2 ± 1.0 3.2 ± 1.0 
 1.9 ± 1.0 1.6 ± 0.9 
 coap 0.52 ± 0.7c 0.72 ± 0.5a 
PAR (cm) 1.2 ± 0.8 1.5 ± 0.8 
 1.7 ± 0.7 1.9 ± 0.8 
 coapar  –0.47 ± 0.9  –0.19 ± 0.8 
Planum + (cm) 4.4 ± 0.6 4.6 ± 0.8 
 3.6 ± 1.1 3.4 ± 1.0 
 coap+ 0.22 ± 0.36c 0.32 ± 0.3b 
 sumppar 0.06 ± 0.55 0.53 ± 0.85c 
H1 (cm24.1 ± 0.8 4.0 ± 1.0 
 3.6 ± 0.5 3.2 ± 0.8 
 coah1 0.28 ± 0.22a 0.22 ± 0.16a 
H2 (cm20.98 ± 0.7 1.41 ± 0.9 
 1.05 ± 0.6 1.31 ± 0.7 
 coah2  –0.49 ± 0.9 0.01 ± 0.9 
Measure  Control (n = 15) RD (n = 13) 
aMean coefficient significantly different from 0.0, P < 0.0005 (paired t-test). 
bMean coefficient significantly different from 0.0, P < 0.005. 
cMean coefficient significantly different from 0.0, P < 0.05. 
d Mean significantly different from controls, P < 0.05. 
Abbreviations: Coaant, asymmetry – cerebellar anterior lobe; coacer, asymmetry – cerebral hemispheres; H1, H2, first and second tranverse gyrus of Heschl; PT, planum temporale; PAR, posterior ascending ramus; PT+, planum + PAR; sumppar, summed coefficients of asymmetry, PT, PAR. 
Cerebral memisphere (cm3580 ± 43 561 ± 59 
 583 ± 47 576 ± 64 
 coacer –0.005 ± 0.03 –0.026 ± 0.04a 
Cerebellum: anterior Lobe (cm36.2 ± 1.1 6.0 ± 1.1 
 6.8 ± 1.3 6.0 ± 1.8 
 coaant –0.08 ± 0.2a 0.02 ± 0.28 
Posterior lobe (cm380 ± 6 78 ± 8 
 82 ± 5 82 ± 9 
 coapost  –0.03 ± 0.04 c  –0.06 ± 0.04 a d 
Planum temporale (cm) 3.2 ± 1.0 3.2 ± 1.0 
 1.9 ± 1.0 1.6 ± 0.9 
 coap 0.52 ± 0.7c 0.72 ± 0.5a 
PAR (cm) 1.2 ± 0.8 1.5 ± 0.8 
 1.7 ± 0.7 1.9 ± 0.8 
 coapar  –0.47 ± 0.9  –0.19 ± 0.8 
Planum + (cm) 4.4 ± 0.6 4.6 ± 0.8 
 3.6 ± 1.1 3.4 ± 1.0 
 coap+ 0.22 ± 0.36c 0.32 ± 0.3b 
 sumppar 0.06 ± 0.55 0.53 ± 0.85c 
H1 (cm24.1 ± 0.8 4.0 ± 1.0 
 3.6 ± 0.5 3.2 ± 0.8 
 coah1 0.28 ± 0.22a 0.22 ± 0.16a 
H2 (cm20.98 ± 0.7 1.41 ± 0.9 
 1.05 ± 0.6 1.31 ± 0.7 
 coah2  –0.49 ± 0.9 0.01 ± 0.9 
Table 3

Variables on which PD and control groups differed significantly

Measure Control (n = 15) PD (n=9) Effect sizee 
aSignificantly different from 0.0, P < 0.0005 (paired t-test). 
bSignificantly different from 0.0, P < 0.005. 
cSignificantly different from 0.0, P < 0.05. 
dMean significantly different from controls, P < 0.05. 
eIf the four URD subjects are added to the control group, the effect sizes for all four comparisons increase to >1 (P < 0.01). 
Coacer –0.005 ± 0.03 –0.04 ± .04c,d –1 
Coaant  –0.08 ± 0.2a  0.11 ± .27d  0.81 
Sumppar 0.06 ± 0.55  0.85 ± 0.8c,d  1.17 
H2 (cm 20.98 ± 0.7  1.7± 0.9d  0.9 
Measure Control (n = 15) PD (n=9) Effect sizee 
aSignificantly different from 0.0, P < 0.0005 (paired t-test). 
bSignificantly different from 0.0, P < 0.005. 
cSignificantly different from 0.0, P < 0.05. 
dMean significantly different from controls, P < 0.05. 
eIf the four URD subjects are added to the control group, the effect sizes for all four comparisons increase to >1 (P < 0.01). 
Coacer –0.005 ± 0.03 –0.04 ± .04c,d –1 
Coaant  –0.08 ± 0.2a  0.11 ± .27d  0.81 
Sumppar 0.06 ± 0.55  0.85 ± 0.8c,d  1.17 
H2 (cm 20.98 ± 0.7  1.7± 0.9d  0.9 
Table 4

The distribution of individuals with different numbers of risk factors in PD, URD and control groups

Group No. of risk factorsa 
 n 
aIn this nonparametric analysis, a risk factor was defined as an anatomical measurement lying in the lowest quartile of the distribution. The presence of any of these four risk factors discriminates the PD from the URD group. The presence of combinations of risk factors discriminates the PD group from the controls. The probability that the distributions in the three groups differs is P < 0.0002 (χ2 = 25.9, df = 6). 
PD 
URD 
Control 15 
Group No. of risk factorsa 
 n 
aIn this nonparametric analysis, a risk factor was defined as an anatomical measurement lying in the lowest quartile of the distribution. The presence of any of these four risk factors discriminates the PD from the URD group. The presence of combinations of risk factors discriminates the PD group from the controls. The probability that the distributions in the three groups differs is P < 0.0002 (χ2 = 25.9, df = 6). 
PD 
URD 
Control 15 
Table 5

Proportions of each group with combinations of specific risk factors

Risk factor Coaant Sumppar LH2 
 PD Control PD Control PD Control 
Since no URD subject had any risk factors, their group is not included in table. 
Coacer 0.22 0.11 0.06 0.11 
Coaant   0.22 
Sumppar     0.44 
Risk factor Coaant Sumppar LH2 
 PD Control PD Control PD Control 
Since no URD subject had any risk factors, their group is not included in table. 
Coacer 0.22 0.11 0.06 0.11 
Coaant   0.22 
Sumppar     0.44 
Table 6

Pearson rs are given for significant relations between anatomical and behavioral variables

WJ-Cog cluster/subtest Pearson r 
 Cerebral volume Cumulative risk factors 
Significant correlation coefficients are shown in italics, P < 0.05. Coefficients >0.60 are significant, P < 0.001. 
Long-Term Retrieval Cluster –0.03 0.41 
Delayed Recall (Names) –0.02 0.47 
Memory for Names –0.14 0.38 
Short Term Memory Cluster 0.52 –0.06 
Numbers Reversed  0.17 0.64 
Memory for Words  0.34 0.38 
Memory for Sentences 0.40  0.22 
Fluid Reasoning Cluster 0.61  0.00 
Concept Formation 0.44  0.03 
Analysis/Synthesis 0.57 –0.10 
Cross Out (Processing Speed Cluster) 0.49 –0.18 
Sound Patterns (Auditory Processing Cluster) 0.44 –0.34 
Spatial Relations (Visual Processing Cluster) 0.57 –0.02 
Oral Language Cluster 0.58  0.07 
Listening Comprehension 0.51 –0.13 
Verbal Analogies (cont (contR 0.61 –0.14 
Reading Tests   
Fluency  0.07 0.56 
Reading  0.13 –0.34 
Spelling –0.18 0.43 
Pseudo word (rank)  0.07 0.70 
Single word  0.18 0.41 
Comprehension 0.50  0.02 
WJ-Cog cluster/subtest Pearson r 
 Cerebral volume Cumulative risk factors 
Significant correlation coefficients are shown in italics, P < 0.05. Coefficients >0.60 are significant, P < 0.001. 
Long-Term Retrieval Cluster –0.03 0.41 
Delayed Recall (Names) –0.02 0.47 
Memory for Names –0.14 0.38 
Short Term Memory Cluster 0.52 –0.06 
Numbers Reversed  0.17 0.64 
Memory for Words  0.34 0.38 
Memory for Sentences 0.40  0.22 
Fluid Reasoning Cluster 0.61  0.00 
Concept Formation 0.44  0.03 
Analysis/Synthesis 0.57 –0.10 
Cross Out (Processing Speed Cluster) 0.49 –0.18 
Sound Patterns (Auditory Processing Cluster) 0.44 –0.34 
Spatial Relations (Visual Processing Cluster) 0.57 –0.02 
Oral Language Cluster 0.58  0.07 
Listening Comprehension 0.51 –0.13 
Verbal Analogies (cont (contR 0.61 –0.14 
Reading Tests   
Fluency  0.07 0.56 
Reading  0.13 –0.34 
Spelling –0.18 0.43 
Pseudo word (rank)  0.07 0.70 
Single word  0.18 0.41 
Comprehension 0.50  0.02 
Figure 1.

Four images at Talairach x = 48. (Top) Control; (Bottom) PD. (a) Typical morphology of Heschl's gyrus and planum in the left hemisphere. Small arrowhead indicates border between H1 and planum temporale (PT). The elevation at the posterior end of PT (large arrowhead) is a consistent marker for the origin of the PAR which emerges on more lateral sections. (b) Typical right hemisphere morphology. PAR originates (large arrowhead) more medially and extends further superiorly than in left hemisphere. (c) PAR originates more medially and extends more superiorly than in control shown above. Boundaries of H2 are indicated by small arrowheads. (d)PAR originates at posterior boundary of H2, PT is absent, a conformation rarely found in the left hemisphere but present in 20% of right hemispheres (Witelson and Kigar, 1992). Note the large right-sided asymmetry in the size of the parietal lobe posterior to the PAR in the PD subject (bottom) compared to the control (top). Scale bar = 2 cm.

Figure 1.

Four images at Talairach x = 48. (Top) Control; (Bottom) PD. (a) Typical morphology of Heschl's gyrus and planum in the left hemisphere. Small arrowhead indicates border between H1 and planum temporale (PT). The elevation at the posterior end of PT (large arrowhead) is a consistent marker for the origin of the PAR which emerges on more lateral sections. (b) Typical right hemisphere morphology. PAR originates (large arrowhead) more medially and extends further superiorly than in left hemisphere. (c) PAR originates more medially and extends more superiorly than in control shown above. Boundaries of H2 are indicated by small arrowheads. (d)PAR originates at posterior boundary of H2, PT is absent, a conformation rarely found in the left hemisphere but present in 20% of right hemispheres (Witelson and Kigar, 1992). Note the large right-sided asymmetry in the size of the parietal lobe posterior to the PAR in the PD subject (bottom) compared to the control (top). Scale bar = 2 cm.

Figure 2.

Relation between nonsense word reading (WRMT-WA) and four quantitative anatomical measures in 28 college students. Circles, controls; open triangles, PD; filled triangles, URD. (Upper left) Coefficient of asymmetry for cerebral hemispheres: control distribution has moderate right sided asymmetry; URD cluster near symmetry, while PD distribution is skewed with 5/9 subjects falling outside the control and URD distribution. (Upper right) Coefficient of asymmetry, anterior lobe: control and URD distributions center around a moderate right sided asymmetry, while PD subjects are more dispersed and centered around left-sided asymmetry. (Lower left) summed coefficients of asymmetry for PT and PAR: controls and URD are symmetrical, PD have a significant left-sided asymmetry. (Lower right) PD tend to have larger H2 than controls or URD.

Figure 2.

Relation between nonsense word reading (WRMT-WA) and four quantitative anatomical measures in 28 college students. Circles, controls; open triangles, PD; filled triangles, URD. (Upper left) Coefficient of asymmetry for cerebral hemispheres: control distribution has moderate right sided asymmetry; URD cluster near symmetry, while PD distribution is skewed with 5/9 subjects falling outside the control and URD distribution. (Upper right) Coefficient of asymmetry, anterior lobe: control and URD distributions center around a moderate right sided asymmetry, while PD subjects are more dispersed and centered around left-sided asymmetry. (Lower left) summed coefficients of asymmetry for PT and PAR: controls and URD are symmetrical, PD have a significant left-sided asymmetry. (Lower right) PD tend to have larger H2 than controls or URD.

References

Albert M, Diamond A, Fitch R, Neville H, Rapp P, Tallal P (1999) Cognitive development. In: Fundamental neuroscience (Zigmond M, Bloom F, Landis S, Roberts J and Squire L, eds), pp. 1313–1338. New York: Academic Press.
Bastian A, Mugnaini E, Thach W (1999) Cerebellum. In: Fundamental neuroscience (Zigmond M, Bloom F, Landis S, Roberts J and Squire L, eds), pp. 973–992. New York: Academic Press.
Best M, Demb J (
1999
) Normal planum temporale asymmetry in dyslexics with a magnocellular deficit.
NeuroReport
 
10
:
607
–612.
Binder JR, Frost JA, Hammeke TA, Rao SM, Cox RW (
1996
) Function of the left planum temporale in auditory and linguistic processing.
Brain
 
119
:
1239
–1247.
Bishop D, Adams C (
1990
) A prospective study of the relationship between specific language impairment, phonological disorders and reading retardation.
J Child Psychol Psychiat
 
31
:
1027
–1050.
Briggs G, Nebes R (
1974
) Patterns of hand preference in a student population.
Cortex
 
11
:
230
–238.
Bruck M (
1992
) Persistence of dyslexics' phonological awareness deficits.
Dev Psychol
 
28
:
874
–886.
Carroll J (1993) Human cognitive abilities: a survey of factor, analytic studies. Cambridge: Cambridge University Press.
Catts H, Fey M, Zhang X, Toblin J (
1999
) Language basis of reading and reading disabilities: evidence from a longitudinal investigation.
Scient Stud Reading
 
3
:
331
–362.
Chua S, McKenna P (
1995
) Schizophrenia: a brain disease?
Br J Psychiat
 
166
:
563
–582.
Eckert M, Leonard C (
1999
) Heritability of Heschl's gyrus duplication: a neural risk factor for dyslexia.
J Cogn Neurosci
 
10
.
Eckert M, Leonard C (
2000
) Structural imaging in dyslexia: the planum temporale.
Ment Retard Disab Res Rev
 
6
:
198
–206.
Eckert M, Lombardino L, Leonard C (
1998
) Who is at risk for reading failure? Neuroanatomical and socioeconomic (SES) contributions to reading skill.
Soc Neurosci Abstr
 
24
:
923
.
Eckert M, Lombardino L, Leonard C (2001) Planar asymmetry tips the phonological playground and environment lifts the bar. Child Devel (in press).
Filipek P (
1995
) Neurobiological correlates of developmental dyslexia: how do dyslexic brains differ from those of normal readers?
J Child Neurol
 
10
:
62
–69.
Foundas A, Leonard C, Gilmore R, Fennell E, Heilman K (
1994
) Planum temporale asymmetry and language dominance.
Neuropsychologia
 
32
:
1225
–1231.
Foundas A, Leonard C, Heilman K (
1995
) Morphological cerebral asymmetries and handedness: the pars triangularis and planum temporale.
Arch Neurol
 
52
:
501
–508.
Gauger L, Lombardino L, Leonard C (
1997
) Brain morphology in children with specific language impairment.
J Speech Hear Lang Res
 
40
:
1272
–1284.
Geschwind N, Levitsky W (
1968
) Human brain: left–right asymmetries in temporal speech region.
Science
 
161
:
186
–187.
Grudnik J, Leonard C, Kranzler J, Oakland T, Eckert M, Lombardino L (
2000
) Auditory duplications predict slow inspection time in children and adults.
Soc Neurosci Abstr
 
26
:
1249
.
Hauser P, Zametkin A, Martinez P, Vitiello B, Matochik J, Mixson A, Weintraub B (
1993
) Attention-deficit hyperactivity disorder in people with generalized resistance to thyroid hormone.
New Engl J Med
 
328
:
997
–1001.
Ide A, Rodriguez E, Zaidel E, Aboitiz F (
1996
) Bifurcation patterns in the human sylvian fissure: hemispheric and sex differences.
Cereb Cortex
 
6
:
717
–725.
Jäncke L, Wunderlich G, Schlaug G, Steinmetz H (
1997
) A case of callosal agenesis with strong anatomical and functional asymmetries.
Neuropsychologia
 
35
:
1389
–1394.
Jastak S, Wilkinson G (1993) Wide range achievement test, 3rd edn. Wilmington, DE: Jastak.
Jensen A (1998) The g factor. London: Praeger.
Jernigan T, Hesselink J, Sowell E, Tallal P (
1990
) Cerebral morphology on MRI in language and learning-impaired children.
Arch Neurol
 
48
:
539
–545.
Keith T, Witta E (
1997
) Hierarchical and cross age confirmatory factor analysis of the WISC-III: what does it measure?
School Psychol Quart
 
12
:
89
–107.
Kranzler J (
1997
) What does the WISC-III measure? Comments on the relation between intelligence, working memory capacity, and information processing speed and efficiency.
School Psychol Quart
 
12
:
110
–116.
Leonard C, Voeller KS, Lombardino L, Morris MK, Alexander A, Andersen H, Garofalakis M, Hynd G, Honeyman J, Mao J, Agee O, Staab E (
1993
) Anomalous cerebral structure in dyslexia revealed with magnetic resonance imaging.
Arch Neurol
 
50
:
461
–469.
Leonard CM, Martinez P, Weintraub BD, Hauser P (
1995
) Magnetic resonance imaging of cerebral anomalies in subjects with resistance to thyroid hormone.
Am J Med Genet
 
60
:
238
–243.
Leonard C, Lombardino L, Mercado L, Browd S, Breier J, Agee O (
1996
) Cerebral asymmetry and cognitive development in children: a magnetic resonance imaging study.
Psychol Sci
 
7
:
79
–85.
Leonard C, Puranik C, Kuldau J, Lombardino L (
1998
) Normal variation in the frequency and location of human auditory cortex landmarks: Heschl's gyrus: where is it?
Cereb Cortex
 
8
:
397
–406.
Leonard C, Kuldau J, Breier J, Zuffante P, Gautier E, Heron D-C, Lavery E, Packing J, Williams S, DeBose C (
1999
) Cumulative effect of anatomical risk factors for schizophrenia: an MRI study.
Biol Psychiat
 
46
:
374
–382.
Lombardino L, Riccio C, Hynd G, Pinheiro S (
1997
) Linguistic deficits in children with reading disabilities.
Am J Speech Lang Pathol
 
6
:
71
–78.
Manis F, Seidenberg M, Doi L, McBride-Chang C, Petersen A (
1996
) On the bases of two subtypes of developmental dyslexia.
Cognition
 
58
:
157
–195.
Markow T (1994) Developmental instability: its origin and evolutionary implications. Dordrecht: Kluwer.
Maron L, Kuldau J, Ricciuti N, Eckert M, Bengtson M, Bauer R, Leonard C (
2000
) Planar asymmetry predicts reading comprehension in schizophrenia.
Soc Neurosci Abstr
 
25
:
1248
.
Mody M, Studdert-Kennedy M, Brady S (
1997
) Speech perception deficits in poor readers: Auditory processing or phonological code?
J Exp Child Psychol
 
64
:
199
–231.
Musiek F, Reeves A (
1990
) Asymmetries of the auditory areas of the cerebrum.
J Am Acad Audiol
 
1
:
240
–245.
Nettelbeck T (
1982
) Inspection time: an index for intelligence?
Quart J Exp Psychol A
 
34
:
299
–312.
Nettelbeck T (1987) Inspection time and intelligence. In: Speed of information processing and intelligence, pp. 295–346. Norwood, NJ: Ablex.
Nicolson R, Fawcett A (
1990
) Automaticity: a new framework for dyslexia?
Cognition
 
35
:
59
–182.
Nicolson R, Fawcett A (
1994
) Comparison of deficits in cognitive and motor skills among children with dyslexia.
Ann Dyslexia
 
44
:
147
–166.
Penhune VB, Zatorre RJ, MacDonald JD, Evans AC (
1996
) Interhemispheric anatomical differences in human primary auditory cortex; probabilistic mapping and volume measurement from magnetic resonance scans.
Cereb Cortex
 
6
:
661
–672.
Plante E, Swisher L, Vance R, Rapcsak S (
1991
) MRI findings in boys with specific language impairment.
Brain Lang
 
41
:
52
–66.
Preis S, Jäncke L, Schitter P, Huang Y, Steinmetz H (
1998
) Normal intrasylvian anatomical asymmetry in children with developmental language disorder.
Neuropsychologia
 
9
:
849
–855.
Reiss A, Abrams M, Singer H, Ross J, Denckla M (
1996
) Brain development, gender and IQ in children. A volumetric imaging study.
Brain
 
119
:
1763
–1774.
Reschly D (
1990
) Found: our intelligences. What do they mean?
J Psychoeduc Assess
 
8
:
259
–267.
Rumsey J, Donahue B, Brady D, Nace K, Giedd J, Andreason P (
1997
) A magnetic resonance imaging study of planum temporale asymmetry in men with developmental dyslexia.
Arch Neurol
 
54
:
1481
–1489.
Schlaug G, Jäncke L, Huang Y, Steinmetz H (
1995
) In vivo evidence of structural brain asymmetry in musicians.
Science
 
267
:
699
–701.
Schultz R, Cho N, Staib L, Kier L, Fletcher J, Shaywitz S, Shankweiler D, Katz L, Gore J, Duncan J, Shaywitz B (
1994
) Brain morphology in normal and dyslexic children: the influence of sex and age.
Ann Neurol
 
35
:
732
–742.
Shankweiler D, Crain S, Katz L, Fowler A, Liberman A, Brady S, Thornton R, Lundquist E, Dreyer L, Fletcher J, Stuebing K, Shaywitz S, Shaywitz B (
1995
) Cognitive profiles of reading-disabled children: comparison of language skills in phonology, morphology, and syntax.
Psychol Sci
 
6
:
149
–156.
Sisodiya S, Free S, Stevens J, Fish D, Shorvon S (
1995
) Widespread cerebral structural changes in patients with cortical dysgenesis and epilepsy.
Brain
 
118
:
1039
–1050.
Stanovich K (
1988
) Explaining the differences between the dyslexic and the garden-variety poor reader.
J Learn Disabil
 
21
:
590
–604.
Steinmetz H, Rademacher J, Huang Y (
1989
) Cerebral asymmetry: MR planimetry of the human planum temporale.
J Comput Assist Tomogr
 
13
:
996
–1005.
Steinmetz H, Seitz RJ (
1991
) Functional anatomy of language processing: neuroimaging and the problem of individual variability.
Neuropsychologia
 
29
:
1149
–1161.
Steinmetz H, Rademacher J, Jäncke L, Huang Y, Thron A, Zilles K (
1990
) Total surface of temporoparietal intrasylvian cortex: diverging left– right asymmetries.
Brain Lang
 
39
:
357
–372.
Stevens J (
1997
) Anatomy of schizophrenia revisited.
Schizophr Bull
 
23
:
373
–383.
Stringer R, Stanovich K (
2000
) The connection between reaction time and variation in reading ability: unravelling covariance relationships with cognitive ability and phonological sensitivity.
Scient Stud Reading
 
4
:
41
–54.
Tallal P, Miller S, Fitch RH (
1993
) Temporal processing in the nervous system: implications for the development of phonological systems.
Ann NYAcad Sci
 
682
:
27
–47.
Waddington C (1957) The strategy of the genes. London: George Allen & Unwin.
Wassink T, Nelson J, Crowe R, Andreason N (
1999
) Brain morphology in schizophrenia.
Am J Med Genet (Neuropsychiat Genet)
 
88
:
724
–728.
Wiederholt J, Brayant B (1992) Gray Oral Reading Tests, 3rd edn. Austin, TX: PRO-ED.
Willerman L, Schultz R, Rutledge J, Bigler E (
1991
) In vivo brain size and intelligence.
Intelligence
 
15
:
223
–228.
Witelson S, Kigar D (
1992
) Sylvian fissure morphology and asymmetry in men and women: bilateral differences in relation to handedness in men.
J Comp Neurol
 
323
:
326
–340.
Woodcock R (1987) Woodcock Reading Mastery Tests, Revised. Circle Pines, MN: American Guidance Service.
Woodcock R, Johnson M (1989) Woodcock–Johnson Psycho-Education Battery — Revised. Chicago, IL: Riverside.
Zuffante P, Leonard C, Kuldau J, Bauer R, Doty E, Bilder R (2000) Dissociation between working memory impairments and MRI measures of area 46 volumes in schizophrenia. Psychiat Imag (in press).