SEARCH

SEARCH BY CITATION

Keywords:

  • Language lateralization;
  • fMRI;
  • Rolandic epilepsy

Summary

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

Purpose: Benign childhood epilepsy with centrotemporal spikes (BECTS) is the most common epilepsy syndrome of childhood and can be associated with language difficulties. The exact profile of these difficulties and their neurofunctional underpinnings, however, are not yet clear.

Methods: To further understand the impact of the BECTS syndrome on language, we assessed language performance using standard neuropsychological measures, and patterns of language lateralization using functional magnetic resonance imaging (fMRI) in children with typical BECTS (n = 20) and healthy controls (n = 20).

Results: The fMRI analyses revealed that language-related activation was less lateralized to the left hemisphere in anterior brain regions in the patients relative to the control group. This finding was consistent with decreased performance in the BECTS group compared to the control group on the neuropsychological measure most dependent on the integrity of anterior aspects of the language axis, namely, sentence production.

Discussion: The converging lines of evidence from the neuropsychological and activation methodologies support the view that BECTS is associated with language difficulties that are regional, and anterior, in nature.

Benign childhood epilepsy with centrotemporal spikes (BECTS) is the most common epilepsy syndrome of childhood (Cavazzuti et al., 1980; Engel, 2001). The syndrome has a characteristic electroencephalography (EEG) signature of frequent focal interictal epileptiform discharges arising from the perirolandic region, and in its typical form occurs without underlying structural pathology or frequent seizures. The syndrome resolves spontaneously by adolescence and is defined as occurring in children of normal intellect without neuropsychological deficits (Loiseau & Duche, 1989). Recent studies, however, indicate that the syndrome can be associated with mild cognitive and learning difficulties. Poor performance is reported in a number of functional domains including language (Staden et al., 1998; Baglietto et al., 2001; Monjauze et al., 2005; Wolff et al., 2005; Riva et al., 2007), attention (D’alessandro et al., 1990; Piccirilli et al., 1994), spatial perception (Volkl-Kernstock et al., 2006), memory (Croona et al., 1999), executive function (Metz-Lutz et al., 1999; Baglietto et al., 2001), and academic achievement (Vinayan et al., 2005).

Given the proximity of the seizure focus to classical language areas in BECTS, several studies have attempted to clarify the nature of difficulties within the language sphere (Staden et al., 1998; Baglietto et al., 2001; Monjauze et al., 2005; Wolff et al., 2005; Riva et al., 2007), but a clear account of the effects of BECTS on language, and its neurofunctional underpinnings, is yet to emerge. Some studies indicate that language skills are broadly affected in children with BECTS (Baglietto et al., 2001; Wolff et al., 2005). Studies conducted within a psycholinguistic framework, on the other hand, describe a selective profile of language difficulties involving sentence formulation (Staden et al., 1998; Monjauze et al., 2005), or fluency (Riva et al., 2007), whereas comprehension (Deonna et al., 1993; Monjauze et al., 2005) and naming (Monjauze et al., 2005; Riva et al., 2007) are spared. This pattern suggests selective dysfunction within the language system in children with BECTS.

Lateral specialization for the processing of language is detectable at an early stage in healthy children (Kinsbourne & Hiscock, 1977). Behavioral studies suggest that basic systems for language perception and control are well lateralized by approximately 3 years of age, and that the degree of lateralization is stable until adulthood (Kimura, 1967; Hiscock & Kinsbourne, 1980; Hiscock et al., 1985). Functional neuroimaging studies further support this view: Activation associated with the processing of meaningful speech is left-hemisphere dominant, encompassing frontal and temporoparietal cortices in infants as young as 3 months of age (Dehaene-Lambertz et al., 2002). A similar pattern is seen in children aged 5–7 years (Ahmad et al., 2003). In middle to late childhood, patterns of activation associated with covert language production are also left lateralized (Holland et al., 2001; Wood et al., 2004; Szaflarski et al., 2006), and patterns of lateralization can be related to neuropsychological measures of language performance (Wood et al., 2004).

Childhood epilepsies with refractory focal seizures and structural lesions can be associated with alterations in the pattern of language lateralization, manifesting as bilateral recruitment of homologous regions, or right hemisphere dominance (Rasmussen & Milner, 1977; Liegeois et al., 2004; Anderson et al., 2006; Yuan et al., 2006). It is not clear, however, if idiopathic partial epilepsies such as BECTS exert similar effects.

To gain a better understanding of the impact of the BECTS syndrome on language, we characterized the language profile and assessed language lateralization by means of functional magnetic resonance imaging (fMRI) in children with BECTS and healthy controls. The relationship between performance on language tests and the indices of language lateralization was then explored.

Methods

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

Participants

One hundred forty children found to have centrotemporal spikes on EEG recordings obtained at one of three major teaching hospitals in Melbourne, Australia—Austin Health, the Royal Children’s Hospital, and Monash Medical Centre—were screened for inclusion in the study. To obtain a pure cohort of children with typical BECTS that was uncomplicated by any additional neurologic condition, the digital EEG recordings as well as the medical, epilepsy, and developmental histories for each patient were reviewed by a pediatric epileptologist (ASH). Our inclusion criteria were (1) the presence of classical centrotemporal spikes arising from a normal background on EEG and a clinical history including at least one seizure that was consistent with a diagnosis of typical BECTS, and (2) no additional clinically diagnosed neurologic condition. To maximize compliance with the fMRI procedure we also restricted our sample to children of at least 6 years of age and native English speakers.

Eighty-four cases failed to meet these inclusion criteria. Forty-six children with centrotemporal spikes were excluded because their clinical history was inconsistent with a clinical diagnosis of typical BECTS [e.g., centrotemporal spikes but no history of a typical BECTS seizure (N = 40), EEG or clinical features indicating atypical BECTS (N = 6)]; six children had typical BECTS and an additional clinical diagnosis [e.g., autism (N = 2), Down’s syndrome (N = 1), developmental language delay (N = 1), nonverbal learning disability (N = 1), child was deaf (N = 1)]; 28 children were younger than 6 years of age, and English was a second language in 4 children.

Of the remaining 56 children eligible for the study, 8 children could not be contacted and 24 children declined to participate. The remaining 24 children completed the study. In this group, four children had unsuccessful fMRI language studies [excessive movement (N = 1), technical issues with data collection (N = 3)]. We report on the 20 children who completed the study and had successful fMRI language studies (boys = 12; right-handed = 16; age range 6.5–11.8 years). Eight children had left-sided spikes, seven children had right-sided spikes, and five children had independent bilateral spikes recorded on EEG. Consistent with reports of incidental structural abnormalities in children with BECTS (Gelisse et al., 2003), routine MRI collected as part of the scanning protocol revealed structural abnormalities in 2 of our 20 patients. These were an incidental nodule of gray matter along the right lateral ventricle in an 8-year-old boy and a type 1 Chiari malformation in an 11-year-old girl. Ten patients were treated with antiepileptic medication at the time of study (carbamazepine = 6; sodium valproate = 4).

The control group consisted of 20 healthy children (boys = 12; right-handed = 18; age range 7.4–12.8 years) recruited via local schools and departmental staff. All control children had acquired English as their first language and were screened for a history of medical and developmental problems. Structural MRI in the control children was unremarkable.

The patient and control groups were well matched on age at assessment (patients: M = 9.24 ± 1.5 years; controls: M = 9.90 ± 1.5 years; t = −1.41, p = 0.17) and nonverbal reasoning as indexed using the Matrix Reasoning subtest from the Wechsler Abbreviated Scale of Intelligence (Wechsler, 1999) (patients: M = 49.63 ± 9.4; controls: M = 50.87 ± 8.9; t = −0.39, p = 0.70). The project received approval from the human research ethics committee at each hospital. Parents or guardians gave written informed consent for their child to participate in the study.

MRI scanning

Language activation paradigm

While in the MRI scanner the participants completed a verb-generation task. Verb-generation is a robust paradigm for assessing the lateralization of fundamental language function in children (Holland et al., 2001; Wood et al., 2004). In healthy subjects the task is associated with a consistent pattern of activation that encompasses classical language territory (e.g., inferior frontal, superior and middle temporal, angular and supramarginal gyri) as well as other language related sites (e.g., middle frontal gyrus and medial frontal cortex). Task-related activation is typically left hemisphere dominant (Holland et al., 2001), particularly in anterior language regions (Wood et al., 2004). Activation can be quantified using a standard pixel counting approach to provide a measure of lateralization that is consistent with information obtained by invasive methods such as the Wada technique (Benson et al., 1999; Szaflarski et al., 2008).

The paradigm consisted of eight 36-s blocks of alternating task and rest in a standard block design. During the task blocks, participants covertly generated verbs (“think of an action or doing word”) to a series of visually presented nouns. Nouns were presented every 4 s, yielding a total of nine nouns per task block. During the rest blocks, subjects viewed a cross hair in the center of the screen and were instructed to rest. Stimuli were developed as presentation slides and projected onto a screen that was placed at the foot of the scanner bed.

Subject preparation

To facilitate cooperation and alleviate anxiety, children were prepared in a mock MRI unit immediately prior to scanning. During this session children were familiarized with the bore, head coil, scanner noises, and mirror arrangement to view the screen. Children also practiced the task using different stimuli. All children mastered the task easily.

Image acquisition

fMRI data covering the whole-brain were collected on a 3-Tesla GE Signa LX scanner (GE Medical Systems, Milwaukee, WI, U.S.A.) using a standard birdcage quadrature head coil and a gradient-recalled, echo-planar imaging (EPI) sequence (TR = 3600 ms, TE = 40 ms, flip angle = 60 degrees, matrix = 128 × 128 mm, 4-mm thick slices with 1-mm gap). To minimize susceptibility artifact the slices were acquired in an axial plane tilted 30 degrees toward coronal. A total of 90 images were acquired for each of the 22-slice positions. To aid in the interpretation of activation maps a set of T1-weighted and MR angiographic images was also collected in the same plane as the fMRI data.

Data processing and analyses

fMRI data were processed and language activation maps were generated using the Statistical Parametric Mapping software (SPM2: Wellcome Department of Imaging Neuroscience, London, UK) with the assistance of iBrain (Abbott & Jackson, 2001). In brief, preprocessing included slice–time correction before images were realigned using SPM2 to a single optimum target image and were then spatially normalized to an existing site-specific EPI template approximating the SPM standard space. The EPI template had been constructed from 30 healthy controls scanned at our site, as previously described (Waites et al., 2005). Images were inspected to verify the homogeneity of the raw signal across the left and right hemispheres as well as the success of the normalization procedure. Images were then smoothed to twice the original in-plane voxel size. Within-brain voxels were determined in iBrain using an automated iterative procedure. A statistical parametric map was obtained for each child by performing a one-tailed Student’s t-test at each pixel location, comparing signal intensity between task and rest states. Additional regressors were included in the general linear model to account for variability in the fMRI signal due to subject motion; these were modeled by the six rigid-body transformation parameters estimated during the image realignment preprocessing. In addition, we used an automated motion rejection scheme to effectively discard volumes with excessive movement relative to the previous volume by including a delta function regressor centered on the volume to be discarded (Lemieux et al., 2007). We used a threshold of 1-mm displacement between volume acquisitions for rejection. Following the rejection of a volume for motion, the following three volumes were also discarded in a similar manner to allow for the MR signal to return to a steady state. All motion regressors described previously were included in our SPM2 design matrix specified as effects of no interest. Resultant individual activation maps thresholded at 0.001 uncorrected for multiple comparisons were visually inspected and also used to calculate laterality indices as described in subsequent text.

Calculation of laterality indices

Laterality indices (LIs) were calculated from the individual activation maps thresholded at p < 0.001 (uncorrected for multiple comparisons) in four language-related regions of interest (ROIs); the dorsolateral prefrontal region encompassing the middle frontal gyrus (MFG); the inferior frontal gyrus (IFG) encompassing Broca’s area; the angular and supramarginal gyri (ANG); and posterior temporal lobe (Post. TEMP). For the purposes of this study a template defining the regions was constructed and applied to each map in standard space using iBrainTM. The number of significantly activated voxels within each ROI was computed in the left and right hemisphere. The LIs were then obtained from the ratio: LI = (Left − Right)/(Left + Right), where “Left” and “Right” were the number of activated voxels within the left- and right-sided ROI, respectively. Using this approach LIs were expressed as a continuous variable from −1 to 1, where a positive LI indicated dominant left hemisphere activation, whereas a negative number indicated dominant right hemisphere activation.

Language performance

Three aspects of language function were examined out of the scanner using standardized cognitive tests. The tasks were selected for their ability to interrogate different neuroanatomically based aspects of the language system according to current clinicopathologic and functional neuroimaging evidence. Conceptualizing the language system within this framework provided a strong evidence base for the interpretation of results and minimized the number of neurocognitive variables required to test our major hypothesis.

Sentence production, a function highly dependent on the anterior language cortex including Broca’s area (Mohr et al., 1978), was assessed using the Formulated Sentences subtest from the Clinical Evaluation of Language Fundamentals Fourth Edition (CELF – IV) (Semel et al., 2003).

Sentence repetition, a function dependent on anterior and posterior aspects of the perisylvian language axis including the insula and arcuate fasciculus (Damasio & Damasio, 1980), was assessed using the Recalling Sentences subtest from the CELF – IV.

Reading accuracy, a task highly dependent on the posterior language zone (Henderson, 1986), was assessed using the Single Word Reading subtest from the Wide Range Achievement Test – Third Edition (Wilkenson, 1993).

Each test was administered and scored in accordance with the test manual, and statistical analyses were conducted on the age-standardized data.

Statistics

Performances on language measures and LIs were compared between groups using unpaired t-tests. The degree of linear correlation between the out-of-scanner measures and LIs was determined. A significance threshold of p = 0.05 was set for all analyses.

Results

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

fMRI

Cortical activation

Visual inspection of the individual statistical parametric maps revealed typical patterns of language-related activation in all subjects. There were, however, subtle differences in the distribution of activation across homologous language regions between patients and controls (Fig. 1).

image

Figure 1.   Functional language maps in eight individual controls (A) and patients (B). Activation maps show areas of increased blood-oxygen-level dependent (BOLD) signal during the noun–verb task relative to rest. Maps are overlaid on the subjects’ own normalized data and displayed in radiologic convention (Left side of the image = Right hemisphere) at a threshold of p < 0.005 (uncorrected for multiple comparisons).

Download figure to PowerPoint

In the control subjects, activation was observed in standard language sites including the middle (100%) and inferior (100%) frontal gyri; the angular/supramarginal gyri (100%); and the superior/middle temporal gyri (85%). Consistent with previous reports in healthy children the number of activated pixels in each region was generally greater in the left hemisphere than the right hemisphere (Holland et al., 2001; Wood et al., 2004; Szaflarski et al., 2006). As expected from a noun–verb task, task-related activation was clearly lateralized to the left hemisphere in anterior language regions.

In the patients, task-related activation was also observed in typical language sites; middle (95%) and inferior (95%) frontal gyri; the angular/supramarginal gyri (90%); and the superior/middle temporal gyri (90%). The striking left-sided lateralization of activation in anterior regions observed in controls, however, was less apparent in this group; activation frequently extended to homologous areas of the right hemisphere.

Comparison of laterality indices

A quantitative comparison of the LIs generated from the four predefined ROIs confirmed that language-related activation in anterior ROIs was less lateralized to the left hemisphere in the patient group relative to controls (Fig. 2). Although the LIs were broadly similar between groups in posterior language regions of interest [ANG Patient: mean (M) = 0.43 ± 0.13, range −0.61 to 1.00; Control: M = 0.57 ± 0.14 range −1.00 to 1.00; t(36) = −0.71, p = 0.48] [Post. TEMP Patient: M = 0.49 ± 0.16, range −1.00 to 1.00; Control: M = 0.44 ± 0.18, range −1.00 to 1.00; t(33) = 0.22, p = 0.83], there was a statistically significant difference in the lateralization of activation across the middle frontal region [Patient: M = 0.42 ± 0.14, range −1.00 to 1.00; Control: M = 0.88 ± 0.03, range 0.45–1.00; t(37) = −3.10, p = 0.006]. A similar trend was observed in the inferior frontal ROI, but it failed to reach statistical significance in this study [Patient: M = 0.41 ± 0.14, range −1.00 to 1.00; Control: M = 0.66 ± 0.09, range −0.09 to 1.00; t(37) = −1.57, p = 0.13].

image

Figure 2.   Mean lateral indices (LIs) [± standard error of the mean (SEM)] for the patient and control group in each region of interest (top panel). The regions of interest encompassed left and right middle frontal gyri (MFG), inferior frontal gyri (IFG), angular and supramarginal gyri (ANG), and posterior temporal lobe (Post. TEMP.) (bottom panel).

Download figure to PowerPoint

Age at assessment was not significantly correlated with the LIs in the patient or control groups. Within the patient group, LIs were not significantly different between patients with left-(N = 8) or right- (N = 7) sided unilateral spikes or between patients treated with (N = 10) or without antiepileptic medications (N = 10).

Language performances

Mean performances in both groups on the out-of-scanner language measures fell comfortably within normal limits (Table 1). Against this background, mean language performances selectively differed between the patient and control groups. Specifically, performance on the sentence production test was significantly poorer in the patients relative to control subjects. Poor performance in the patient group reflected a relatively high number of grammatical errors (Table 2).

Table 1.   Language performance in patient and control groups
 Patients M (SEM)RangeControls M (SEM)Rangetp
  1. M, mean; SEM, standard error of the mean.

Sentence production9.05 (0.67)2–1310.89 (0.52)5–14−2.180.04
Sentence repetition9.30 (0.69)4–1410.79 (0.54)5–14−1.680.10
Reading accuracy102.60 (2.92)76–122106.32 (2.34)89–126−0.990.33
Table 2.   Examples of the grammatical errors made by three patients on the sentence production test
PatientSexAgeTarget word Child’s response
1Boy6 years 6 monthsNeverNever the man don’t cross the road when the red button is on.
2Boy9 years 0 monthNeverNever to cross the road why there is a green light on.
3Boy10 years 11 monthUntilThe man with until his friend.

There were no statistically significant differences between the patient and control groups on the sentence repetition or reading tests.

The relationship between performance measures and laterality indices

Within the patient group, better performance on the sentence production test was significantly correlated with increasing left-sided lateralization in the inferior frontal region (Table 3). By contrast, higher reading scores were significantly correlated with increasing left-sided lateralization in the angular ROI. Performance on the repetition test was not significantly correlated with the laterality index in any of the ROIs sampled in the study.

Table 3.   Degree of correlation between language measures and LIs in the patient group
 LI MFGLI IFGLI ANGLI Post. TEMP
  1. LI, Laterality index; MFG, middle frontal gyrus; IFG, inferior frontal gyrus; ANG, angular gyrus; Post. TEMP., posterior temporal lobe.

  2. Probability values under the null hypothesis are indicated in parentheses. Significant results highlighted in bold (p < 0.05).

Sentence production0.14 (0.57)0.50 (0.03)0.18 (0.50)0.17 (0.53)
Sentence repetition−0.10 (0.67)0.20 (0.42)0.21 (0.39)−0.06 (0.82)
Reading accuracy0.40 (0.10)0.29 (0.23)0.49 (0.04)0.41 (0.10)

There were no significant correlations between performances on the language measures and LIs in any ROI in the control group.

Discussion

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

This study provides significant insights into the nature of language dysfunction in children with BECTS. We have shown that language organization is subtly, but significantly, altered in brain regions responsible for completion of the language tasks in which BECTS children show poor performance. The converging lines of evidence from the neuropsychological and activation methodologies support the view that BECTS is associated with language difficulties that are regional rather than global.

A less-lateralized pattern of activation

The fMRI results show a difference in lateralization of language networks in children with newly diagnosed BECTS relative to healthy controls. In the BECTS children, language was less lateralized to the left hemisphere in anterior language ROIs. This finding is consistent with the only other report of language activation in BECTS, a single case included in a mixed epilepsy cohort (Yuan et al., 2006). Language activation with a verb-generation task in a 12-year-old right-handed male with BECTS showed a left–right shift relative to an age-matched control. The relative contributions of anterior and posterior language regions, however, were not explored.

Our findings also extend a handful of studies that have investigated language lateralization in BECTS using behavioral techniques (Piccirilli et al., 1988; Metz-Lutz et al., 1999; Lundberg et al., 2005; Bulgheroni et al., 2008). A verbal–manual time-sharing paradigm in which children were asked to vocalize a sequence of animal names while performing a finger-tapping task raised the possibility that BECTS is associated with an alteration in the interhemispheric representation of language (Piccirilli et al., 1988). The verbal task showed the expected effect of more interference on right-hand than on left-hand tapping rates in healthy children. In children with BECTS and a left-sided seizure focus, the interference effect was similar for both hands, indicating possible bilateral representation of language. Subsequent studies using a dichotic listening approach have also highlighted an alteration in functional representation in BECTS. Dichotic listening performance in children with BECTS is characterized by increased error rates relative to controls (Metz-Lutz et al., 1999; Lundberg et al., 2005) and may be associated with a loss of an established marker of left hemisphere contributions to linguistic processing, the right ear advantage (Bulgheroni et al., 2008). Our findings support and refine these observations by showing that alterations in the lateral representation of language in BECTS are regionally specific, and consistent with the documented language profile.

Poor performance on specific language tasks

In our study, detailed language testing revealed a regionally specific functional profile in the patients, consistent with the fMRI findings. Performance in the BECTS group was significantly below that of the control group on the task most dependent on the integrity of anterior aspects of the language axis, namely, sentence production. Within the patient group, performance on the sentence production test was significantly correlated with the laterality index in the inferior frontal region, showing reduced left-sided activation in children who struggled most with sentence production. These data show that differences in activation may reflect differences in task performance.

The specific profile of difficulties in the BECTS children on detailed language testing is consistent with previous reports (Staden et al., 1998; Monjauze et al., 2005). We found that BECTS children performed significantly worse than controls on the sentence production task but not the sentence repetition or word reading tasks. In comparison to our study, Staden et al. (1998) reported that poor sentence production was accompanied by reduced reading performance. The discrepancy between these findings is likely to reflect a difference in selection criteria between the two studies. Our study specifically selected children who did not have a history of developmental delay and learning problems. In the Staden et al. (1998) study, parents were concerned about language and learning problems in 11 of 20 children studied, with 3 children also having a history of speech therapy, suggesting underlying learning problems.

To our knowledge only one other study has examined sentence repetition in BECTS (Northcott et al., 2005). Consistent with our findings, Northcott et al. (2005) reported no significant reduction in the performance of BECTS children relative to normative data on the repetition task. In psycholinguistic terms, the relative sparing of sentence repetition with respect to sentence production raises the possibility that BECTS is associated with a high-level language difficulty. Age-appropriate performance on the sentence-repetition task suggests that basic processes for phonologic encoding, monitoring of output, and articulation at a sentential level are intact in our patient group. By contrast, the discrepancy between performance on the sentence production and repetition tasks highlights a potential difficulty with syntactic aspects of sentence formulation. In repetition, the syntactic structure of the sentence is predefined. The production task, on the other hand, demands self-initiated construction of sentential frames according to the rules of syntax (Levelt, 1989). Further evidence of high-level language dysfunction is also shown by the high incidence of grammatical errors observed at a qualitative level on the sentence formulation task in our patient group as well as reports of grammatical errors in the spontaneous language of children with BECTS (Deonna et al., 1993; Staden et al., 1998).

Anterior language network is affected

A key component of the neural network–mediating syntactic aspects of sentence formulation includes left inferolateral frontal cortex. Overt production of syntactically correct sentences relative to utterances of syntactically unrelated words is associated with a significant increase in cerebral blood flow in the left frontal operculum caudally adjacent to Broca’s area (BA 44) (Indefrey et al., 2001). Activation in the inferior frontal region (BA 45) also shows a graded response depending on the complexity of the syntactic demand (Indefrey et al., 2001; Haller et al., 2005). A role for the left inferior frontal region in syntax processing is also found in lesion-based evidence (Mohr et al., 1978). These data highlight the importance of interactions between Broca’s area and surrounding brain regions, including left middle frontal gyri for efficient syntax processing. Although the functional contributions of inferior and middle frontal regions are not yet understood, it has been proposed that the left inferior frontal region reflects syntactic encoding or a syntax-specific short-term memory subsystem that supports the generation of syntactic structures (Caplan & Waters, 1999). The middle frontal region appears to contribute verbal working memory resources to syntactic processing (Just & Carpenter, 1992), bringing language output in line with increased cognitive demand (Fu et al., 2006), such as that involved in the formulation of propositional language (Blank et al., 2002). Less-lateralized activation in these areas, therefore, might reflect poor connection or altered specialization of anterior language systems.

Interpretation

Although establishing a causal relationship is beyond the scope of this study, our findings are consistent with several possibilities. First, a recurrent focal functional disturbance, namely, epileptic spiking, might affect language organization. Epilepsy can be associated with atypical language lateralization, even when the lesion is remote from language cortex (Liegeois et al., 2004), and the frequency of epileptiform discharges correlates with the degree of left–right shift in language lateralization (Janszky et al., 2006). Alternatively, a common underlying mechanism, possibly maturational, might cause both the epilepsy and the language difficulty (Doose & Baier, 1989). Consistent with this view, the language dysfunction demonstrated in our study is in one of the brain areas undergoing rapid development at the time of investigation, for example, the frontal lobes. Finally, the possibilities might interact such that the epileptic spikes further exacerbate an underlying language problem. Although it is likely that the spikes will resolve over time in these children, it remains to be determined if the language difficulties also recover.

Acknowledgments

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

We would like to thank the children and their families for participating in our research as well as Janet Barchett and the neurophysiology teams at Austin Health, the Royal Children’s Hospital, and Monash Medical Centre for assistance with the identification of potential subjects. We also thank the radiographers involved in the project, Todd Little, Heather Ducie, and Shawna Farquharson, as well as Esther Hutchinson for testing some of the participants. The authors gratefully acknowledge the support of the National Health and Medical Research Council, Neurosciences Victoria, and the Brain Imaging Research Foundation, Australia. We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.

Disclosure: None of the authors has any conflict of interest to disclose.

References

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References