erp evidence of a dichotic left-ear deficit in some … averaged for each target condition within...

Abstract

Children with and without behavioral dichotic left-ear deficits participated inan event-related potential study with quasidichotic presentations of familiar fairytale segments. Electrical activity was recorded from the scalp while thechildren listened for semantically and/or syntactically anomalous words fromeither the right side or the left side while competing segments of the fairy talewere simultaneously presented from the opposite side. Latencies and amplitudeswere averaged for each target condition within the group with dichotic left-eardeficits (LED) and the group with normal dichotic listening performance (WNL).Individual global field power waveforms and topographic brain maps weregenerated for the average response in each of the two listening conditions,target right and target left. Cross-correlations were performed on the grandaveraged global field power waveforms to measure the degree of synchronybetween target right and target left responses in both groups. Integrationfunctions were performed to compare the accumulated sum of voltages duringtarget (right and left) and control (right and left) conditions. WNL childrenproduced typical ERP responses to the target words in both target right andtarget left conditions. Responses from LED children were at delayed latenciesin the target left condition and were at reduced amplitudes in both targetconditions. Topographic brain maps revealed more lateralized scalp distributionsand greater activation of frontal regions in LED children in the target leftcondition. Cross-correlational and integration function results demonstratedinteraural asymmetries in responses from the LED children. Overall resultssuggest that slowed neural conduction times, poor interhemispheric transferof neural activity, and a failure to suppress competing information arriving atthe right ear may be involved in poor left-sided processing in children withbehavioral left-ear dichotic deficits.

Key Words: Auditory processing disorder, children, dyslexia, event-relatedpotential, global field power, integration, topographic brain map

Abbreviations: ANOVA = analysis of variance; EEG = electroencephalic; ERP= event-related potential; GFP = global field power; HEOG = horizontal electro-oculographic; LED = left-ear deficit; SCAN-C = Screening Test for AuditoryProcessing Disorders—Children; VEOG = vertical electro-oculographic; WNL = within normal limits

Sumario

Niños con y sin deficiencias dicóticas conductuales del oído izquierdoparticiparon en un estudio de potenciales relacionados con el evento, conpresentaciones cuasi-dicóticas de segmentos familiares de cuentos de hadas.La actividad eléctrica se registró en el cráneo mientras los niños escuchabanbuscando palabras anómalas, semántica o sintácticamente, tanto del ladoderecho como del lado izquierdo, mientras que se les presentabansimultáneamente segmentos competitivos del cuento de hadas en el lado

ERP Evidence of a Dichotic Left-EarDeficit in Some Dyslexic Children

Deborah Moncrieff*James Jerger†Ilse Wambacq†Ralf Greenwald†Jeffrey Black‡

*University of Connecticut; †University of Texas at Dallas; ‡Texas Scottish Rite Hospital for Children

Reprint requests: Deborah W. Moncrieff, University of Connecticut, 850 Bolton Road, Unit 1085, Storrs, CT 06269; Phone: 860-486-3687; Fax: 860-486-5422; E-mail: [email protected]

518

J Am Acad Audiol 15:518–534 (2004)

EZ-PDF Style Info

This PDF file was created with EZ-PDF 1.5 and Acrobat Distiller PDF Style Information: ====================== On-demand 600dpi Laser Version: 1.1 Author: AKI Description: This PDF style will create a PDF file suitable for on demand laser printing. It creates PDF files optimized for the Xerox Docutech or other 600 dpi high speed laser printers. All fonts are embedded for maximum fidelity. Platform: Macintosh ======================

Children with language and readingdifficulties have been shown toperform poorly on tests of dichotic

listening. They often show a large interauralasymmetry with normal performance in theright ear and significantly reducedperformance in the left ear (Morton andSiegel, 1991; Lamm and Epstein, 1994;Moncrieff and Musiek, 2002). Because asimilar pattern of performance has beenobserved in patients with a split or disorderedcorpus callosum, it has been suggested thatchildren with a left-ear deficit during dichoticlistening may have poor interhemispherictransfer of information from the righthemisphere to the left hemisphere followingleft ear input of speech material (Jerger et al,1986; Musiek et al, 1984, 1988). Studies usingauditory event-related potentials to exploredifferences in the interhemispheric transferof information in children with reading

disorders have reported delayed latenciesand reduced amplitudes across a range ofresponses, beginning at approximately 100msec following stimulus input. A poorinterhemispheric transfer hypothesis hasbeen partially supported by evidence ofdelayed neural transmission in children withlanguage, learning, and reading difficulties instudies utilizing both nonspeech stimuli (Jirsaand Clontz, 1990; for review, see Lepannenand Lyytinen, 1997) and dichotic speechstimuli (Jerger, Thibodeau, et al, 2002).Deviant and nonfocal topographic brain mapshave also been reported for responses fromlanguage-impaired children (Tonnquist-Uhlen, 1996).

Many children with language andreading disorders are able to perform atnormal levels on dichotic listening tasks forinput to the right ear and only show asignificant deficit for input to their left ear.

ERP Evidence of a Dichotic/Moncrieff et al

519

opuesto. Se promediaron las latencias y las amplitudes para cada condicióndentro del grupo con deficiencias dicóticas en el oído izquierdo (LED) y delgrupo con desempeño auditivo dicótico normal (WNL). Se generaron ondasde poder de campo global individuales y mapas topográficos cerebrales,para la respuesta promedio en cada una de las dos condiciones auditivas:meta en el lado derecho y meta en el izquierdo. Se realizaron correlacionescruzadas con el gran promedio de las ondas de poder de campo global, paramedir el grado de sincronía entre las respuestas con meta en el lado derechoo meta en el izquierdo, en ambos grupos. Se realizaron funciones de integraciónpara comparar la suma acumulada de voltajes durante las condiciones meta(derecha e izquierda) y control (derecha e izquierda). Los niños WNL produjeronrespuestas ERP típicas ante las palabras meta, tanto en condiciones de metaderecha o meta izquierda. Las respuestas de los niños LED mostraron latenciasretardadas en las condiciones de meta izquierda y mostraron amplitudesreducidas en ambas condiciones meta. Los mapas topográficos revelarondistribuciones craneales más lateralizadas y mayor activación de las regionalesfrontales en los niños LED, en la condición meta izquierda. Las funciones deintegración y de correlación cruzada demostraron asimetrías interaurales en lasrespuestas del los niños LED. Los resultados globales sugieren que el hallazgode tiempos de conducción neuronal lentos, una pobre transferencia de actividadneuronal inter-hemisférica y una falla para suprimir información competitiva quearriba al oído derecho, puede estar involucrado en un pobre procesamiento dellado izquierdo, en niños con deficiencias conductuales dicóticas del oídoizquierdo.

Palabras Clave: Trastornos de procesamiento auditivo, niños, dislexia,potencial relacionado con el evento, poder de campo global, integración,mapa topográfico cerebral

Abreviaturas: ANOVA = análisis de variancia; EEG = electroencefalografía;ERP = potencial relacionado con el evento; GFP = poder de campo global;EGO = electro-oculografía horizontal; LED = deficiencia del oído izquierdo;SCAN-C = Prueba de Tamizaje para Trastornos de Procesamiento Auditivoen Niños; VEOG = electro-oculografía vertical; WNL = dentro de límitesnormales

Because the deficit may be specific to the leftear, electrophysiologic techniques thatexamine and contrast input to both the rightand left ear during competitive listeningtasks may prove useful in identifying specificweaknesses in these children. Standardbehavioral testing with dichotic listeningmaterial uses pairs of consonant-vowels,digits, or single-syllable words presentedthrough headphones to a listener seated ina sound-isolated booth. These methods areused to diagnose a left-ear deficit type ofauditory processing disorder in children, butthey may lack ecologic validity as a methodfor evaluating a child’s ability to processauditory information in real-world situationssuch as those encountered in a classroom. Inschool, children must be able to listen toongoing speech in order to extract theinformation necessary for learning. As such,they must be able to maintain auditoryvigilance while focusing attention on themost relevant stimuli and suppressingauditory information that is of lessimportance to the learning task.

A recently developed procedure forrecording the auditory event-related potential(ERP), patterned loosely after a vigilancetask, permits an evaluation of precisetemporal characteristics of the neuralresponse following the input of complexspeech more similar to the kind that childrenmust follow during routine tasks of listeningand learning in school (Jerger, Greenwald, etal, 2000). As previously described (Jerger,Moncrieff, et al, 2000), the procedure isquasidichotic, involving the presentation ofsegments from the fairy tales “Cinderella” and“Snow White” through loudspeakers situateddirectly to the right and to the left of thechild’s head. The child is asked to listen forsegments presented through one speaker ata time and to ignore the different segmentsbeing simultaneously presented from theother speaker. The task is to signal everytime a semantically and/or syntacticallyanomalous word is presented through themonitored speaker. By recording theresponses to anomalous words presentedboth to the right and left ears, it is possibleto analyze the ERP response not only foramplitude and latency characteristicsfollowing speech input per se, but also for theinteraural differences that may occur duringmonitoring of one ear while suppressing inputto the other ear, a task that directly relatesto behavioral dichotic listening tests. As

reported, the anomalous words produced anERP in normal children and young adultswith a robust positivity at a latency ofapproximately 800 to 900 msec with nosignificant interaural differences observedin the responses (Jerger, Moncrieff, et al,2000).

From individual average waveforms,both traditional and global field power (GFP)measures of latency and amplitude canprovide information about intra-aural andinteraural differences in the neural response.Traditional measures are derived fromelectrode locations where maximum activationoccurs, usually at centroparietal sites centeredat electrode CPz. This type of analysis canreflect whether an individual subject or groupof subjects demonstrates typical latency andamplitude characteristics at the region of thescalp where the ERP most often occurs. Theglobal field power (GFP) waveform (Skrandies,1989, 1990) takes into account the activity atall electrodes and resolves the dilemma overdetermining the appropriate electrode atwhich to measure response latency inindividual children.An advantage to the GFPmethod is that it can demonstrate instancesin which an ERP response occurs at electrodesites that differ from the standardcentroparietal locations. Topographic mapsderived from the maximum response latencyin the GFP waveform can then be used todisplay its scalp distribution in order todetermine if the response is shifted from theexpected centroparietal scalp locations. It hasbeen suggested that in studies whereneuromaturation factors may result indisordered responses, topographic mappingdifferences may prove more important thanstandard latency and amplitudecharacteristics from the ERP (Stauder, et al,1989).

The overall goal of this study was toinvestigate whether the ERP recorded duringa quasidichotic listening task would differin children who had previously demonstratedinteraural asymmetries during behavioraldichotic listening tasks. It was hypothesizedthat results following input to the left earwould differ in children with a dichotic left-ear deficit, both from the results in the samechildren following input to the right ear andfrom results obtained in the left ears ofchildren with normal dichotic listeningperformance. Several methods were used toexamine differences in the ERP from childrenwith behavioral left-ear deficits.

Journal of the American Academy of Audiology/Volume 15, Number 7, 2004

520

METHOD

Subjects

A total of 20 children were tested in apreliminary study to determine measures ofinteraural asymmetry during behavioraldichotic listening testing (Moncrieff andMusiek, 2002). As previously described, halfof the children had been previously diagnosedas dyslexic, and the other half wereperforming normally in school. Children withdyslexia were chosen for the study becausethey were predicted to be the most likelygroup of children to demonstrate a left-eardeficit on dichotic listening tests.All childrenwere 11 to 12 years old at the time of thestudy, and all were strongly right-handed asassessed by a handedness survey (Annett,1970). All children had normal (Type A)tympanograms and hearing sensitivity at 15dB or better across the frequencies from 500to 4000 Hz. None had been diagnosed withattention deficit disorder with or withouthyperactivity, psychiatric disorder, orneurological impairment. Only monolingual

native speakers of English were included inthe study. In each half, there were sevenmales and three females.

Behavioral dichotic listening tests wereused to measure interaural asymmetry inall of the children. A larger than normalinteraural asymmetry was defined assignificantly reduced performance in the leftear together with normal performance in theright ear, resulting in a significantly largedifference in performance between the twoears. The 20 children were divided into twogroups based on results from the CompetingWords subtest of the SCAN-C (Keith, 1986)as illustrated in Table 1. Asterisks are placedbeside the initials of each child whoseperformance indicated a left-ear deficit,defined as performance below 76% in the leftear (Moncrieff and Musiek, 2002), togetherwith an interaural difference greater than10%. As seen in Table 1, seven of the dyslexicchildren (five males and two females)demonstrated poor performance in the left earand a large interaural asymmetry on theCompeting Words test. This group is referredto as Left-Ear Deficit (LED). The remainingthree dyslexic children (two males and onefemale) and all of the control children


521

Table 1. Competing Words Test Results

Subject Dyslexic Right Ear Left Ear Ear Advantage Group

01* yes 94 66 28 LED

02* yes 74 56 18 LED

03* yes 82 60 22 LED

04 yes 86 76 10 WNL

05 yes 80 76 4 WNL

06 yes 86 80 6 WNL

07* yes 94 70 24 LED

08* yes 96 74 22 LED

09* yes 84 70 14 LED

10* yes 86 74 12 LED

11 no 90 82 8 WNL

12 no 76 70 6 WNL

13 no 92 78 14 WNL

14 no 86 80 6 WNL

15 no 96 86 10 WNL

16 no 92 88 4 WNL

17 no 86 84 2 WNL

18 no 92 88 4 WNL

19 no 80 76 4 WNL

20 no 90 88 2 WNL

demonstrated normal results on theCompeting Words subtest for both left earperformance and/or interaural asymmetryand are referred to as “Within Normal Limits”(WNL). One control child (12) demonstratedreduced performance in the left ear, butperformance in the right ear was also belownormal. Another control child (13)demonstrated an interaural asymmetrygreater than 10%, but left-ear performancewas above 76%. Since neither of thosechildren demonstrated both performancedeficits, they could not be characterized asleft-ear deficit and were therefore included inthe WNL group.

Test Stimuli

A complete description of the testparadigm has been previously described(Jerger, Greenwald, et al, 2000).A total of 200segments of the fairy tales “Cinderella” and“Snow White” were recorded by a male voice.Segments were approximately equivalent inlength. In half of the segments, chosenrandomly, one word was replaced with atarget anomalous word.The test stimuli werestored in digital form and delivered to thesubject via digital-to-analog conversion.Segments of the fairy tales were presented innormal sequence, with the sequence in onechannel delayed by about 80 seconds relativeto the presentations in the other channel.

Electrophysiologic Methods

For presentation through the twoloudspeakers, the sequence of 200 segmentswas divided into eight blocks of 25 segmentseach. Each child was directed to attend to fourblocks of segments presented through thespeaker directly opposite one ear whileignoring those presented through the speakerdirectly opposite the other ear. Thus, half ofall presentations were to be attended rightand half were to be attended left. Because halfof all presentations contained anomalouswords, there were 50 segments containinganomalous words (target) and 50 segmentsnot containing anomalous words (control)presented toward each ear throughout theexperiment. Because there is an advantageto the first hemispace toward which attentionis directed (Boliek et al, 1988), the first sideto be attended was randomly assigned acrossall subjects.

The child was seated comfortably in amedical examining chair. All sequences weredelivered from two loudspeakers at ear heightand one meter to the left and right of thechild’s ears. A 38 cm computer screen, placedat eye level 1.7 meters in front of the child,presented graphic directions pertaining tothe procedure. At the beginning of each blockof 25 segments, the child was directed eitherto listen right (with a red screen and anarrow pointing toward the right) or to listenleft (with a blue screen and an arrow pointingtoward the left). In order to reduce eyemovement throughout the procedure, thechild was further directed to focus attentionon a small, plastic character doll locateddirectly below the computer screen. Cottonballs were taped across the child’s eyelids tolimit the magnitude of excursions duringeyeblinks throughout the experiment.

Electrophysiologic RecordingApparatus

Brain electrical activity was recordedfrom 32 silver-silver chloride electrodes.Thirty electrodes were mounted in an elasticcap (Neuroscan QuickCap), and two weremounted above and below the left eye inorder to monitor vertical electro-oculographic(VEOG) activity. Electrical activity wasreferred to linked mastoid electrodes withFPZ as ground. Individual sweeps ofelectroencephalic (EEG) activity, time-lockedto the onset of target and nontarget words,were stored in a continuous file for offlineanalysis. Ongoing EEG activity was amplifiedand bandpass, analog filtered from 0.15 to 125Hz. Analog filter skirts were sloped at -12 dBper octave. The amplified and filtered EEGactivity was then digitized through anacquisition interface system (NeuroscanSynamp) and routed to a computer forindividual sweep storage, display, averagingand post hoc digital filtering (SCAN 4.0).Signal averaging was carried out after offlineartifact rejection and baseline correction.Individual epochs were examined and rejectedwhenever electrical activity in the eye channel(VEOG) exceeded ±30 microvolts. If morethan 30 of the 50 epochs of a given target ornontarget condition were rejected for anyreason, then the entire data set for thatsubject was rejected.Thus, each average ERPwas based on a minimum of 20 epochs.Successfully averaged evoked potentials were


522

then linearly detrended, digitally low-passfiltered at 20 Hz, and baseline corrected.Digital filter slope was -48 dB per octave.

Electrophysiologic Procedure

Prior to testing, the speech levels from thetwo loudspeakers were adjusted to achievemedian plane localization for each child aspreviously described (Jerger, Moncrieff, et al,2000).All stimuli were presented at the levelsdetermined through this localizationprocedure so that neither ear enjoyed anintensity advantage as the result of aninteraural asymmetry in sensitivity. Afteraffixing the electrode cap to the head, thechild was instructed to look at the computerscreen and to listen only to the fairy talepresentations arriving from whichever sidethe screen directed them to monitor. Theywere told to remain still and to blink normally,limiting eyeblinks as much as possible. Eachchild was told to press the button on thesame side of the response pad as the speakerbeing monitored whenever a word was heardfrom that speaker that did not fit in the fairytale story. In this way, while monitoring theright speaker, only the right-sided button onthe response pad should have been pressed,and while monitoring the left speaker, onlythe left-sided button on the response padshould have been pressed. Responses weremonitored throughout the experiment toassure that the child remained on task, butthe responses were not recorded for accuracy.

Data Analysis

In order to eliminate significantsinusoidal modulation seen in waveformsfiltered at 20 Hz and to facilitatemeasurement of peak responses, individualsubject waveforms were reanalyzed at digitallow-pass filter settings of 6 Hz (filter slope of-12 dB per octave). Epochs were separatelyaveraged for target right (anomalous wordspresented through the right sound fieldspeaker) and target left (anomalous wordspresented through the left sound fieldspeaker). Individual average waveforms ineach of the two conditions were analyzed forpeak amplitudes and latencies at thecentroparietal (CPz) electrode site. Absolutelatencies and amplitudes of the negativeearly response characterized as N400 andthe positive later response characterized as

P800, together with peak-to-peak amplitudesfrom the two responses, were measured foreach individual subject. Individual globalfield power (GFP) waveforms were alsocomputed in order to derive a single measureof peak latency and amplitude for each childfrom multiple electrodes across the scalp.GFP is defined by calculating, at each timesample of the epoch, the square root of themean of the squared derivations of eachvoltage from the common average referencevoltage. The resultant value is a GFPwaveform plotted as a function of time acrossthe entire epoch.

Individual latency and amplitudeinformation were measured both at electrodesite CPz and at the peak of the GFPwaveform. Mean latency and amplitudeinformation within each group, defined on thebasis of behavioral test performance (WNLand LED), were compared for the target rightand target left conditions. A mixed-designanalysis of variance (ANOVA) was used withone between-subjects factor (group) and onewithin-subjects factor (target condition). Posthoc comparisons were conducted byindependent samples t-tests constrained byLevene’s test for equality of variances.Statistical significance was evaluated at analpha level of 0.05.

Topographic brain maps of the GFPwaveforms were constructed by interpolationof voltages between adjacent electrodes, usinga 4-point linear-interpolation algorithm.Bivalent colors representing the range ofobserved voltages were then assigned to theresulting voltage matrix.

A cross-correlation function wasperformed on the grand averaged response foreach group. The cross-correlation functioncomputes a standard Pearson product-moment coefficient of correlation (r) from thevoltage levels obtained at each of 1000 1msec time intervals across the waveform.When one waveform (target right) iscompared to a different waveform (targetleft) from the same individual subject orgroup of subjects, the cross-correlationfunction measures the coefficient ofcorrelation as one waveform is displaced intime relative to the other waveform by aspecified time interval (1 msec). The finalplot of the coefficient, r, as a function of timedisplacement, tau, will reflect the extent towhich brain activation in response to onetarget condition correlated with brain


523

activation in response to the other targetcondition. If the two waveforms are identical,they will achieve perfect correlation (r = 1.0)at a tau value of 0. When the two waveformsare different, the displacement in the peak ofthe cross-correlation function quantifies theextent to which one waveform would need tobe delayed in time relative to the otherwaveform for the two waveforms to achievemaximum synchrony. In this way, a cross-correlation of the target right and target leftwaveforms can provide a measure of theasynchrony between the response obtainedwhen listening with the right ear and whenlistening with the left ear.

A final approach examined results whenthe individual averaged response at CPz toeach target condition (target right and targetleft) was integrated across the time intervaland compared to the individual averagedresponse at CPz to the corresponding controlconditions (control right and control left).The control condition is defined as the averageresponse when the child listened to the fairytale segments from one side (right or left) thatdid not contain anomalous words. Integrationfunctions are useful in oddball paradigmssuch as mismatch negativity, P300 and thistype of anomalous task because they clearlydifferentiate the waveform obtained duringthe control condition from the waveform

obtained when the listener has perceived atarget anomalous word. The integrationfunction computes an accumulation sum ofthe voltages across each sweep by multiplyingeach data point by an interpoint distance.Theresultant integrated sum of the waveformshould be small in the control condition anddemonstrate a significant increase inaccumulated voltages when listeners haveattended to oddball targets. A standardintegration function reflects accumulatedvoltages in both the negative and positivedirections. Alternatively, an integrationfunction performed on a rectified averagewaveform will convert all voltages intopositive values prior to summing them. In thisway, the rectified integration function reflectsthe overall increase in brain activationwithout regard for direction of the response.

ELECTROPHYSIOLOGIC RESULTS

Average Responses at CPz

Grand average waveforms for theresponse at electrode site CPz for each of thetwo groups are displayed in the left-handcolumn of Figure 1. In the WNL group,waveforms are similar for the two listeningconditions with a negative peak at a latency


524

Figure 1. Grand averaged waveforms of the ERP in the target right and target left conditions in both groups atelectrode site CPz (left-side waveforms) and grand averaged global field power waveforms in the target right andtarget left conditions in both groups (right-side waveforms).

of approximately 300 to 350 msec and apositive peak at a latency of approximately800 to 900 msec. Latencies are slightly earlierand amplitudes are slightly larger in thetarget left condition, but the differencesbetween the two conditions are slight. In theLED group, the waveforms are dissimilar. Inthe target right condition, there is a negativepeak at a latency of approximately 300 msecand a positive peak at a latency close to 800msec, but the amplitudes of these responsesare smaller than in the WNL group. In the

target left condition, there is no apparentnegative peak, and there are several positivepeaks with no clear maximum across thegrand averaged response. Overall, amplitudesfor the target left condition in the LED groupare indistinguishable from baseline acrossthe latency region typical for this ERP.

Group averages and standard errors forlatency and amplitude measures across thetwo listening conditions are summarized inthe top half of Table 2. With respect tolatencies of the P800 response, a mixed-design ANOVA showed a significant maineffect for group (p = .001) and for theinteraction between target and group (p =.009), but no significance for the main effectof target condition, right versus left (p =.120). Within each of the two groups, posthoc independent samples t-tests on the P800latencies showed a significant effect for earin the LED group (p = .035, equal variancesnot assumed), but no significance for theeffect of ear in the WNL group. Post hocindependent samples t-tests on the P800latencies in each target condition showed asignificant effect for group in the target leftcondition (p = .000, equal variances notassumed), but no significance for the effect ofgroup in the target right condition.

A plot of individual P800 latencies intarget right and target left conditions,separated by group membership, is displayedin Figure 2. As shown in Figure 2, the P800latencies in the LED children weredistributed differently from those in the WNLchildren, with longer latencies in the target


525

Table 2. Means and (Standard Errors) of Latencies and Peak-to-Peak Amplitudes of P800Response at CPz in Two Groups and Two Target Conditions

Target Right Target LeftGroup Latency P-P Amplitude Latency P-P Amplitude

WNL 886.0 (37.1) 14.2 (1.4) 823.4 (24.5) 13.4 (1.1)(n = 13)

LED 1000.0 (68.8) 9.1 (1.4) 1217.1 (59.8) 8.7 (0.8)(n = 7)

Means and (Standard Errors) of Latencies and Amplitudes of Peak in Individual GFPWaveforms in Two Groups and Two Target Conditions

Target Right Target LeftGroup Latency Amplitude Latency Amplitude

WNL 856.8 (73.7) 4.35 (0.36) 797.7 (26.9) 3.94 (0.35)(n = 13)

LED 830.9 (115.9) 3.57 (0.52) 823.4 (134.81) 2.81 (0.19)(n = 7)

Figure 2. Plot of the target right and target leftlatencies for individuals in each of the two groups(WNL and LED) for the P800 response at electrodesite CPz.

left condition relative to the target rightcondition. An enclosure around the latenciesfor children in the LED group demonstratesthat the results for the children in the twogroups can be separated along the measureof latency of the P800 response to the targetleft condition.

For peak-to-peak amplitude, measuredfrom maximum negativity of the N400response to maximum positivity of the P800response, a similar mixed-design ANOVAshowed no significance for the effect of targetcondition (p = .628) or for the interactionbetween target condition and group (p = .855),but did show a significant main effect of group(p = .005). As shown in Table 2, the peak-to-peak amplitude observed in LED children wassignificantly smaller than in WNL children.

GFP Waveforms

Grand averages of the individual globalfield power waveforms for both target rightand target left conditions for the two groupsof children are illustrated in the right-handcolumn of Figure 1. The morphology of theGFP waveforms for the two groups is similarto the morphology of the grand averagewaveforms from electrode site CPz.Again, theGFP waveforms for the WNL groupdemonstrate a similar response in the twotarget conditions, with peak responselatencies ranging between 600 and 800 msec.In the LED group, the GFP peak in the targetleft condition is also less distinct than thepeak for the target right condition or thepeaks in both target conditions observed inthe other group of children.

Topographic Brain Maps

A topographic brain map derived fromthe peak response in the GFP waveformindicates the region across the scalp wherethe greatest change in activity was recorded.As previously described for this task, themaximum positivity in each target conditionoccurred over the centroparietal region atapproximately 700 to 900 msec followingpresentation of the anomalous word (Jerger,Moncrieff, et al, 2000). An individual child’saverage GFP for a target condition shouldsubsequently reflect the contributions of allelectrodes across the scalp at that point intime. If the child’s response is normal,maximum positivity will reflect increases in

electrical activity centered near the CPzelectrode and distributed bilaterally acrossadjacent electrodes. There are several waysin which the topographic map may reflectabnormal responses in the GFP waveform.Slower neural conduction along typical neuralpathways would result in a delayed latencyfor the peak in the GFP waveform, but thetopographic map could still representmaximum positive activation at the normalcentroparietal recording site. Secondly, a shiftof activity to different regions of the brain thatare recruited to help participate in the taskcould result in a peak in the GFP waveformthat occurs across a different region of thescalp at either the same or a different latency.A third possibility is that neural activity isnonfocal and distributed generally across thescalp. In any of these cases, there may alsobe several peaks in the GFP waveform thatrepresent multiple latencies at whichactivation achieved similar levels. Multiplepeaks in the GFP waveform that occur atcomparable amplitudes might reflectincreased activation across different scalplocations at different points in time followingpresentation of the stimulus. Topographicbrain mapping methods can resolve thedilemma created by multiple peaks byidentifying the peak that represents positiveactivation at scalp locations that most closelyfit the anticipated response pattern.

Topographic brain maps were created atthe peak of the GFP waveform in target rightand target left conditions for each of thechildren in this study.Table 3 lists the patternof activation displayed in each topographicbrain map, together with the magnitude of theGFP peak for target right and target leftconditions. As shown in Table 3, bilateralresponses distributed over the centroparietalregion are indicated by the letter “B” andwere displayed for most of the WNL children.In a few cases, responses in the WNL groupwere unilateral (R or L) with scalpdistribution maximal over the side oppositethe ear toward which the targets were beingpresented.This pattern occurred only once inthe WNL group in the target right condition(18) and with three children in the targetleft condition (04, 06, and 11). Two of theseWNL children were among those initiallydiagnosed as dyslexic, but they had notdemonstrated the left-ear deficit on thebehavioral dichotic listening tasks and weretherefore placed in the WNL group.


526


527

Table 3. Scalp Distribution and Magnitude (µV) at Peak of the GFP Waveforms in Two Groups and Two Target Conditions

Subject Diagnosis Group Target Left Target Right

01 Dyslexic LED R 2.79 B/F 3.53

02 Dyslexic LED R/O 2.82 L/O 3.79

03 Dyslexic LED L/F 2.27 R/F 2.33

04 Dyslexic WNL B 4.58 L 3.68

05 Dyslexic WNL B 2.24 B/F 4.74

06 Dyslexic WNL B/O 4.00 L/F 3.98

07 Dyslexic LED B 2.08 B/F 1.79

08 Dyslexic LED L/F 3.31 L/F 4.87

09 Dyslexic LED R/F 2.97 R/F 2.99

10 Dyslexic LED B/F 3.44 B/F 5.70

11 Control WNL B/O 3.54 L 3.78

12 Control WNL B/F 2.49 B/F 4.07

13 Control WNL B 3.04 B 2.69




17 Control WNL B/F 3.20 B 4.88

18 Control WNL R 4.55 B 2.85

19 Control WNL B 4.30 B/F 4.72


B = bilateral, R = right, L = left, O = occipital, F = frontal

Figure 3. Topographic brain maps at the peak of the grand averaged GFP response in two groups and two lis-tening conditions, target right and target left.

Unilateral patterns of activation wereobserved in a majority of the LED children(5 of 7 in each target condition), usually withdistribution in the region of the scalpcontralateral to the ear being stimulated. Inone case (03), the pattern of activation wasipsilateral to the side toward which the targetwords were being presented. In another (08),both conditions produced maximumactivation only on the left hemisphere side ofthe scalp. Another notable difference in themaps of the LED children was a shift towardmaximum activation in the frontal region ofthe scalp, with much less positive activationin the centroparietal area.

Mean latency and amplitude informationfrom the individual GFP waveforms aredisplayed in the bottom half of Table 2. Amixed-design ANOVA on average latencies ofthe GFP showed no significance for the maineffects of target (p = .686) or group (p = .999)and no significant interaction between targetand group (p = .753). A similar ANOVA onaverage amplitudes of the GFP also showed nosignificance for the main effects of target (p =.100) or group (p = .065) or for the interactionbetween target and group (p = .520). Theseresults show a trend toward significance fortarget and group with respect to amplitude.Asshown in Table 2, the standard errors for theaverage latencies were much higher for theGFP analysis than for the analysis at CPz,especially in the LED group. The increasedvariability in the peak response may make itdifficult to quantify group differences on the

basis of the GFP peak values.Topographic brain maps created at the

average latency in the grand averaged GFPresponse within each of the two groups aredisplayed in Figure 3.All are scaled to a rangeof voltages from -5 µV to 5 µV. As shown inFigure 3, responses were bilateral in bothtarget conditions in the WNL children. Theresponses for the WNL children were also atgreater amplitudes than in the LED children.The distribution of the response for the LEDchildren is more lateralized in both targetconditions across the contralateral side of thescalp.The shift toward the left side of the scalpin the target right condition is likely due to theunilateral left-sided responses observed in theindividual maps for 4 of the 7 LED children.In the target left condition, activation is morefocal over the right hemisphere side of thescalp and fails to demonstrate a frontal spreadof activation over the left hemisphere portionof the scalp. Individual responses in the targetleft condition were highly variable for the LEDchildren. Three of them demonstrated right-sided unilateral responses; two of themdemonstrated left-sided unilateral responses;and two of them demonstrated bilateralresponses.With this much individual variabilityin the GFP peak response across the LEDgroup, the grand averaged GFP map displaysan overall weakness in the response, but itfails to demonstrate a typical pattern for thesechildren while performing this quasidichoticlistening task.


528

Figure 4. Cross-correlation functions at four electrode sites (CP3, CPz, CP4, and HEOG) measuring the syn-chrony between the target right and target left conditions within each of the two groups.

Cross-Correlation Functions

The cross-correlation functions derivedfrom the grand averaged target right andtarget left waveforms for the two groups ofchildren are displayed in Figure 4. Functionsare derived for three electrodes, the onecentered at CPz and the two adjacentelectrodes, one toward the left side of thescalp (CP3) and the other toward the rightside of the scalp (CP4). As shown in Figure4, the target right and target left waveformsin the WNL group were highly correlated atall three electrodes (CP3 r = .919, CPz r =.904, CP4 r = .850). The tau value is positiveat all three electrodes (CP3 tau = 18 msec,CPz tau = 48 msec, CP4 tau = 82 msec). Apositive tau value means that in order for thetarget left waveform to be synchronous withthe target right waveform, the target leftwaveform should be shifted positively (to alater latency) by those values at thoseelectrodes. As a measure of asymmetrybetween the two target conditions, the highdegree of correlation and the low values fortau suggest that the WNL groupdemonstrated synchronous responses withthe overall responses for the left ear occurringslightly ahead of the overall responses forthe right ear. Because the HEOG electrodewas not activated throughout the experiment,

background brain activity occurring duringthe acquisition of the ERP is displayed inthat electrode’s cross-correlation waveform.The highly correlated responses for the WNLgroup are strikingly dissimilar to theuncorrelated background noise evident inthe cross-correlation function for HEOG.

The target right and target left waveformsin the LED group failed to demonstrate a highdegree of correlation at any of the threeelectrodes (CP3 r = .459, CPz r = .440, CP4 r= .470). In general, the cross-correlationfunction within this group is much less distinctfrom the background brain activity displayedat HEOG.Tau values were also highly variablein the cross-correlation functions for this group.At electrode site CP3, the peak correlationoccurred at a tau of -480 msec, and at CPz, thetau value was -120 msec. This would suggestthat for the waveforms to be synchronous, thetarget right waveform would have to be shiftedlater in time by those amounts.

With respect to individual correlationvalues at each of the electrode sites, a mixed-design ANOVA showed a significant maineffect of group (p = .040).When each electrodesite was analyzed separately, the group effectwas significant only for electrode site CP3 (p= .019), although correlation at electrode siteCPz approached significance (p = .083). Aninterpretation of these results is that over the


529

Figure 5. Integration functions contrasting the accumulated sum of positive and negative voltages across therecording epoch of the ERP following input of target and control words in both right and left listening condi-tions and within each of the two groups.

left hemisphere side of the scalp in the LEDchildren, the response to the target rightcondition occurred at a much shorter latencythan the response to the target left condition.

At electrode site CP4, the tau value shiftsto 12 msec, suggesting that on the righthemisphere side of the scalp, the relationshipbetween the two target conditions is moresynchronous with the target left responseoccurring at a slightly earlier latency thanthe target right response.There are secondarypeaks for electrodes CPz and CP4, however,at much greater tau values of 436, and 428msec, respectively.This may be due to greatervariability in the individual results for childrenin the LED group, with some of the childrendemonstrating a more strongly lateralizedresponse over the contralateral portion of thescalp in both target conditions and othersdemonstrating a somewhat bilateral response.This cross-correlational evidence of multiplepeaks in the LED group suggests that childrenwith a left-ear deficit on behavioral dichoticlistening tests may not demonstrate consistentneural activation patterns in the ERP acquiredduring this quasidichotic listening task. It ispossible that in some of the LED children, afailure of the neural activity to synchronize atmidline following input to either of the two ears

may have resulted in a more contralaterallylateralized response. In others, the responsemay have synchronized and spread bilaterally,but at very delayed latencies and reducedamplitudes. A separate analysis of eachindividual LED child’s waveform andtopographic maps is needed to overcome thevariability involved in the grand averagedgroup response.

Integration Functions

Grand averaged responses within eachgroup were subjected to an integrationfunction to derive the accumulated sum ofvoltages at each electrode across the timeinterval following presentation of targetanomalous words. The same function wasperformed on the grand averaged responsesobtained following the presentation of normalwords during control segments of the fairytale. The purpose of the integration functionis to demonstrate that when listeners attendto target presentations, there is a substantialincrease in the accumulated sum of voltagesacross the acquisition period when comparedto the response when listeners are attendingto normal or control presentations that do notcontain oddball or anomalous stimuli. An


530

Figure 6. Integration functions contrasting the accumulated sum of rectified voltages across the recording epochof the ERP following input of target and control words in both right and left listening conditions and withineach of the two groups.

integration function performed on an averagewaveform will reflect the accumulated sumof voltages in both negative and positivedirections across the time interval of therecording epoch. The integration functionscomparing the grand averaged response for thetarget and control conditions within each sideof presentation (right and left) are displayedin Figure 5. Each function is derived for thechange in voltage at electrode site CPz.

As shown in Figure 5, the WNL groupdemonstrated a substantial increase inaccumulated voltage in the negative directionfor the first 500 to 600 msec followingpresentation of the target words, followed bya substantial increase in accumulated voltagein the positive direction for the next 900 msec.This change occurred regardless of the sidetoward which the anomalous words werepresented. When the same children werelistening to fairy tale segments that did notcontain anomalous words (the controlconditions), their responses reflected a reducedsummation of voltage for the right ear and anoverall decrease in voltage for the left ear.

Grand averaged responses for thechildren in the LED group are similar tothose for the WNL children for the comparisonbetween target right and control rightconditions. The functions reflect much lesschange in voltage in the negative direction,but demonstrate a substantial increase inthe accumulated sum of voltage in the positivedirection over the same time interval as seenin the WNL children’s responses. Theintegration functions for the target left versuscontrol left condition in the LED groupdemonstrate two dramatic differences. First,the integration function for the controlcondition appears similar to the function forthe target right condition.This would suggestthat when the LED children were listeningto normal fairy tale segments (the controlcondition) from the left side speaker, theirbrain activation was summating at thecentroparietal electrode in the same manneras during their response to target wordsarriving from the right side speaker.Throughout each interval that the childrenwere directed to attend for target wordsarriving from the left side speaker,acquisitions were being made both for targetwords and for control words. At the sametime, the children were instructed to ignoreall competing segments being presented fromthe right side speaker. During approximately

half of the control presentations on the leftside, a target word was simultaneouslypresented from the right side. The similarityin the shape of the integration functions forthe control left and target right conditions inthe LED group would suggest that some ofthese children may have failed to suppress theinput from the target words on the right sidewhile they were directed to attend topresentations from the left side.

A second difference in the integrationfunction for the target left condition is thatthe accumulated sum of voltages is lowerthan the sum of voltages seen in the controlleft condition. The integration functions forthe LED group display the typical pattern forthe contrast between target right and controlright with a progressively greateraccumulated sum of voltages in the targetcondition. The reversed pattern seen in thetarget left versus control left conditiondemonstrates that the neural response inthe LED children to the target words arrivingfrom the left side speaker was much weakerthan the same response to target wordsarriving from the right side speaker. It wasalso weaker than the response that may haveoccurred as the result of intrusions of targetwords arriving from the right side speakerwhen the children were supposed to beattending to control segments from the leftside speaker.

A final illustration of this difference canbe seen in Figure 6. In Figure 6, the sameintegration functions are provided for thegrand averaged response at electrode siteCPz in the two groups of children. In this case,the integration functions were performedafter rectifying the waveforms so that allvoltages assumed a positive value. In thisway, the functions demonstrate the totalchange from zero that occurred in the voltagein response to the different types of wordsbeing processed by the listeners. As seen inFigure 6, both the overall depression in thevoltage response in the target left conditionand the enhanced voltage response in thecontrol left condition are evident in thefunction for the LED children.

DISCUSSION

The purpose of this study was toinvestigate whether children with left-ear

deficits on behavioral dichotic listeningperformance would produce an ERP response


531

in a quasidichotic task that differed from thesame response in children without left-eardeficits and whether that response wouldreflect an interaural asymmetry. Severaldifferences in the ERP response wereobserved in the LED children.

1. The latency of the P800 response atelectrode site CPz was significantly delayedin children with a left-ear dichotic listeningdeficit when they were asked to listen foranomalous words presented from the leftsound field speaker. Average latencies toanomalous words presented from the rightsound-field speaker were also delayed relativeto the latencies from the WNL children, butthe differences failed to reach significance.This result is consistent with previousfindings of delayed latencies in corticalresponses in children with auditoryprocessing disorders (Jirsa and Clontz, 1990),language disorders (Tonnquist-Uhlen, 1996),and reading disorders (Mazzotta and Gallai,1992). Results from this study differ fromprevious studies in demonstrating that delaysin latency were more pronounced for inputto the left ear and were significantly differentfrom delays in latency for input to the rightear in children with dichotic left-ear deficits.This evidence of an interaural asymmetry inthe ERP is consistent with the behavioralevidence of more normal processing in theright ear and significantly reducedperformance in the left ear.

2. Children with dichotic left-ear deficitsproduced a response to target anomalouswords that was significantly smaller thanthe response observed in children without left-ear deficits.This result is consistent with reducedamplitudes reported in previous auditory ERPstudies (Holcomb et al, 1986; Pinkerton et al,1989;Erez and Pratt,1992;Mazzotta and Gallai,1992). In general, LED children producedsmaller amplitudes in the target left conditionthan in the target right condition,but the LEDchildren also demonstrated lower amplitudes inthe target right conditions. As a result, aninteraural asymmetry in the LED group withrespect to amplitudes of the response could notbe confirmed.

3. The topography of the grand averagedGFP responses did not reveal clear groupdifferences in the ERP response. Thevariability observed in the LED children’s

individual topographic maps suggests thatthere may be several factors involved in afailure to produce a normal P800 responsewhen listening in the target left condition.The bilateral and centroparietal scalpdistribution observed in the topographicmaps for two LED children suggests that aninterhemispheric transfer of informationthrough fibers of the corpus callosum mayhave been involved in their responses.Strongly lateralized features of thetopographic brain maps for three of the otherfive LED children’s responses, however, aremore consistent with the poorinterhemispheric transfer hypothesis. Inthose cases, the response in the target leftcondition was lateralized over the righthemisphere side of the scalp with littleevidence of crossover to the other side at thepeak of maximum activation. In theremaining two cases, the pattern of activationwas strongly lateralized over the lefthemisphere side of the scalp in the target leftcondition. Many of the responses alsoinvolved a significant amount of activity inthe frontal regions of the scalp. Increasedelectrical activity at frontal scalp locations inthe LED children is consistent with evidencethat in some dyslexics, frontal regions of thebrain, possibly those necessary for allocationof attentional resources, may be recruited toassist with difficult processing tasks(Georgiewa et al, 2002; Shaywitz et al, 2002).

4. Cross-correlational results in the LEDchildren provide additional support forsignificantly longer latencies in the target leftcondition for responses over the lefthemisphere portion of the scalp.The presenceof additional peaks in electrodes located overthe right hemisphere portion of the scalplend support to the possibility that in someof the LED children, results may have beenmore strongly lateralized in the target leftcondition than is evidenced by thetopographic brain map results.The possibilitythat activation was strongly lateralized inboth target conditions (over the left side inthe target right condition and over the rightside in the target left condition) would suggestthat poor interhemispheric transfer of neuralactivity was involved in some of the LEDchildren’s weaker responses.

5. The integration functions for the LEDchildren provide additional support for an


532

interaural asymmetry in this quasidichoticlistening task. Both the standard and rectifiedintegration functions clearly demonstratethat even though the response in the targetright condition was somewhat reduced, theaccumulated voltages in the target conditionsurpassed the voltages seen during thecontrol condition on that side.When the LEDchildren were listening on the left side, therelationship between the target and controlconditions was reversed with the sum ofvoltages in the target condition failing tosurpass those seen in the control condition.The similarity in the shapes of the targetright and control left functions suggests thatwhen the children were supposed to beattending with their left ear, the response thatoccurred during control conditions may havebeen stimulated by target words arriving atthe right ear. This suggests an increaseddominance of the right ear in these childrenthat may have resulted in a failure toappropriately suppress input from thenonattended side.

In summary, several of the methods usedin the analysis of the ERP produceddifferences between the WNL and LEDchildren. Longer latencies and reduced peak-to-peak amplitudes of the P800 response inLED children were apparent in the traditionalanalysis of the grand averaged waveforms atthe centroparietal electrode site. Analysis bycross-correlation of the grand averagedwaveforms also revealed evidence of delayedlatencies in the LED children’s responses.Despite the failure of the GFP analyses toreveal significant group differences, thetopographic brain maps derived from theGFP waveforms demonstrated smalleramplitudes in the LED children across bothlistening conditions. Generally reduced neuralactivation in the LED children in response tothe target conditions was also evident in theintegration functions derived from the grandaveraged ERP waveforms.

Interaural asymmetry in the LEDchildren was demonstrated by significantdifferences between the two ears for averagelatencies of the P800 response as measuredat CPz. Topographic brain maps from LEDchildren also differed between targetconditions. Maps created from the grandaveraged GFP response in the target rightcondition showed a more lateralized responsewith altered distribution toward more frontal

regions. Those contrasted with the diffuse,low-level, and bilateral response observed inthe target left condition. Overall, topographicbrain maps demonstrated that in the targetleft condition, responses from the LEDchildren could not be described on the basisof the grand averaged response. Instead,individual maps needed to be evaluated todetermine the pattern of response seen in eachof the LED children’s GFP waveform. Thecross-correlation of the target right and targetleft conditions revealed that LED children didnot demonstrate synchronous brain activityduring this listening task. Together with thevariability in the topographic brain maps, thepresence of multiple peaks in the cross-correlation results suggests that some LEDchildren may have demonstrated lateralizedactivation patterns whereas others may havedemonstrated more bilateral responses.

Even with factors related to age,intelligence, handedness, and attention deficitdisorder all controlled in this small group ofchildren with behavioral left-ear deficits,there was considerable variability in theelectrophysiologic responses observed duringthis listening task. Because the LED childrenin this study were all dyslexic, it is possiblethat some of the group differences may havestemmed from additional factors that werenot controlled. Dyslexia is a language-baseddisorder that may not have interfered withright-ear processing during standardbehavioral dichotic tests but may haveresulted in bilateral deficits in some of thechildren when listening to more complexspeech information during the quasidichoticlistening task utilized in this study. In someof the LED children, the electrophysiologicresponses indicated that performance in theirright ears was closer to normal as it hadbeen during the behavioral dichotic listeningtests. In others, the ERP response wasdifferent for both ears, suggesting thatdisordered language processes or otherunknown factors may have reduced thechildren’s ability to process the sentencescontaining the anomalous words. Resultsfrom this study suggest that efforts tocharacterize auditory processing abilitiesthrough electrophysiologic measures mayneed to examine both group results andindividual results if a greater understandingof the underlying neural activity involved indichotic listening deficits is to be achieved.


533

REFERENCES

Annett M. (1970) Classification of hand preferenceby association analysis. Br J Psychol 61:303–321.

Boliek CA, Obrzut JE, Shaw D. (1988) The effects ofhemispatial and asymmetrically focused attention ondichotic listening with normal and learning-disabledchildren. Neuropsychologia 26:417–433.

Erez A, Pratt H. (1992) Auditory event-related poten-tials among dyslexic and normal-reading children:3CLT and midline comparisons. Int J Neurosci63:247–264.

Georgiewa P, Rzanny R, Gaser C, Gerhard UJ,ViewegU, Freesmeyer D, Mentzel HJ, Kaiser WA, Blanz B.(2002) Phonological processing in dyslexic children:a study combining functional imaging and eventrelated potentials. Neurosci Lett 318:5–8.

Holcomb PJ, Ackerman PT, Dykman RA. (1986)Auditory event-related potentials in attention andreading disabled boys. Int J Psychophysiol 3:263–273.

Jerger J, Greenwald R,Wambacq I, Seipel A, MoncrieffD. (2000) Toward a more ecologically valid measureof speech understanding in background noise. J AmAcad Audiol 11:273–282.

Jerger J, Moncrieff D, Greenwald R,Wambacq I, SeipelA. (2000) Effects of age on interaural asymmetry ofevent-related potentials in a dichotic listening task.J Am Acad Audiol 11:383–389.

Jerger J, Oliver TA, Chmiel RA, Rivera VM. (1986)Patterns of auditory abnormality in multiple sclero-sis. Audiology 25:193–209.

Jerger J, Thibodeau L, Martin J, Mehta J, Tillman G,Greenwald R, Britt L, Scott J, Overson G. (2002)Behavioral and electrophysiological evidence of audi-tory processing disorder: a twin study. J Am AcadAudiol 13:438–460.

Jirsa RE, Clontz KB. (1990) Long latency auditoryevent-related potentials from children with auditoryprocessing disorders. Ear Hear 11:222–232.

Keith RW. (1986) Screening Test for AuditoryProcessing Disorders (SCAN). San Antonio, TX:Psychological Corporation and Harcourt BraceJovanovich.

Lamm O, Epstein R. (1994) Dichotic listening per-formance under high and low lexical work load insubtypes of developmental dyslexia. Neuropsychologia32:757–785.

Leppanen PH, Lyytinen H. (1997) Auditory event-related potentials in the study of developmentallanguage-related disorders. Audiol Neurootol2:308–340.

Mazzotta G, Gallai V. (1992) Study of the P300 event-related potential through brain mapping inphonological dyslexics. Acta Neurol 14:173–186.

Moncrieff DW, Musiek FE. (2002) Interaural asym-metries revealed by dichotic listening tests in normaland dyslexic children. J Am Acad Audiol 13:428–437.

Morton LL, Siegel LS. (1991) Left ear dichotic lis-tening performance on consonant-vowel combinations

and digits in subtypes of reading-disabled children.Brain Lang 40:162–180.

Musiek FE, Gollegly KM, Baran JA. (1984)Myelination of the corpus callosum and auditory pro-cessing problems in children: theoretical and clinicalcorrelates. Semin Hear 5:231–240.

Musiek FE, Verkest SB, Gollegly MA. (1988) Effectsof neuromaturation on auditory-evoked potentials.Semin Hear 9:1–15.

Pinkerton F, Watson DR, McClelland RJ. (1989) Aneurophysiological study of children with reading,writing and spelling difficulties. Dev Med Child Neurol31:569–581.

Shaywitz BA, Shaywitz SE, Pugh KR, Mencl WE,Fulbright RK, Skudlarski P, Constable RT, MarchioneKE, Fletcher JM, Lyon GR, Gore JC. (2002) Disruptionof posterior brain systems for reading in children withdevelopmental dyslexia. Biol Psychiatry 52:101–110.

Skrandies W. (1989) Data reduction of multichannelfields: global field power and principal componentsanalysis. Brain Topogr 2:73–80.

Skrandies W. (1990) Global field power and topo-graphic similarity. Brain Topogr 3:137–141.

Stauder JEA, Molenaar PCM, van der Molen MW.(1989) Equivalent dipole modeling of topographic ERPcomponents: a developmental case study. JPsychophysiol 3:361–368.

Tonnquist-Uhlen I. (1996) Topography of auditoryevoked cortical potentials in children with severe lan-guage impairment. Scand Audiol 25, Suppl. 44:1–40.


534

erp evidence of a dichotic left-ear deficit in some … averaged for each target condition within...

Documents