DANSSL 2007

Platform Abstracts

Converging Methodologies Structure, Function and Behavior

April A. Benasich, Ph.D. Center for Molecular & Behavioral Neuroscience Rutgers, The State University of New Jersey
Surprisingly few studies have examined the intersection between brain and behavior over the first years of life. Even fewer have done so in the context of a particular research question. Differences in the rate and extent of brain maturation, both structural and functional, are likely to have an effect on behavioral performance even within the normal range. Thus, it is important to examine individual variation within groups at high risk for atypical language development as well as in the typically developing population. Ongoing research in our laboratory provides evidence that the ability to perform fine-grained analyses in the tens of millisecond range during infancy appears to be one of the most powerful and significant predictors of subsequent language development and disorders. We are beginning to examine further relations among behavioral performance, function (using EEG/ERPs) and brain structure (MRIs) in the context of language delays. In this presentation, a brief summary will be given of studies that demonstrate difficulties in discriminating rapidly successive sensory events for a subset of children early in infancy, well before verbal language is in place. Moreover, it will be shown that these differences in infant processing thresholds are predictive of later language outcome. Preliminary data from converging paradigms, specifically EEG/ERPs, structural MRIs and behavioral assessments of information processing, will be presented and the difficulties of analysis and comparison of data sets will be discussed. Analyses of the differences in the pattern and density of power spectra in resting EEG will also be shown for children with a family history of languagebased learning disorders as compared with controls. Finally, the need for developing converging assessment tools that can be applied reliably in individual infants at higher risk for developing a language disorder will be discussed.

Defining maturity of information processing based on electrophysiological manifestations: how it looks vs. what it does
Rita Ceponiene, M.D., Ph.D. Asst. Project Scientist, Project in Cognitive and Neural Development, Center for Research in Language, University of California, San Diego
Magnitude, timing, oscillatory content, and the contributing source componentry of electrophysiological brain activity (EEG, ERP) change with age. In order to appreciate significance of these changes, the underlying processes must be understood and their relative contributions to the recorded brain activity differentiated. I will conjecture that there are two major closely inter-dependent, but also in important ways distinct types of processes underlying maturational changes. The first type of change provides quantitative maturation of electrophysiological brain response and behavior (e.g., synaptic pruning = more accurate sensory encoding = smaller response amplitudes). The second type of change builds on the first one but in addition it offers novel, more efficient and more integrative, ways of information processing. This permits integration of the existing processing mechanisms into the expanding hierarchical as well as parallel associative mentation. Unfortunately for us, changes in the recorded brain activity do not map linearly onto the underlying brain processes. Some of the recorded changes may index simply more (or less) of the same process, while other EEG/ERP changes may index newly emergent information processing algorithms. In addition, certain persisting brain processes may not be equally identifiable in the activity recordings across different ages, while some predominantly quantitative changes may masquerade as newly emerging phenomena. I will present sensory and lexical processing data demonstrating such dissociations, discuss implications of these findings to understanding of systems supporting language, and offer several approaches that may assist in differentiating among the phenomena underlying maturational changes in scalp-recorded brain activity.

Analysis and Interpretation of Infant ERP and Behavioral Data: Findings from a 4Year Longitudinal Study.
Naseem Choudhury, Ph.D. Center for Molecular & Behavioral Neuroscience Rutgers, The State University of New Jersey
The challenges of understanding the development of basic brain mechanisms that may lead to the expression of complex learning and language disorders, such as Specific Language Impairment (SLI), are particularly difficult when examined in the end-state as a number of factors may have lead to either the exacerbation or amelioration of the phenotype. Early studies of language disorders have mainly focused on older children and adults who may have spent a lifetime developing strategies to cope with their disability. Prospective longitudinal studies with preverbal infants, however, provide an efficient and powerful way in which to disentangle the causes of language disorders from its covariates. Studies begun in infancy, when designed with care, enable researchers to address the question of which basic mechanisms may predict to later difficulties in language acquisition. With this as our primary aim, we present ERP data from one such longitudinal study. Infants with and without a family history of SLI were recruited at 6 months and are being followed to 8-years of age. The data presented here is restricted to a sub-sample who have completed their 4 year visit (FH+ = 15, FH-= 25). We explore age and risk related differences in ERPs that may be associated with perturbations in auditory detection and/or discrimination processes and show how these differences are related to concurrent and predictive behavioral performance.

Insights on neurodevelopment from animal studies and APD

Jos J. Eggermont, Ph.D. Department of Physiology and Biophysics; Department of Psychology University of Calgary
Animal central nervous system development and maturation proceeds always faster than its human equivalent. Translating the time scales of these two developmental processes into each other is difficult but not impossible. I will present some general findings on the maturation of central auditory processing in the cat, and the effects of hereditary or induced neonatal deafness on these maturational processes. I will try to fuse these findings with those from human studies. For that purpose the effects of conductive and sensorineural hearing loss on spectral and temporal cortical processing in animals will be reviewed. I will then discuss the role of bottom-up and top-down processing on auditory development and maturation that suggest an intricate dependence and could explain some of the puzzling findings such as the missing N100 in late implanted children. Differing effects of exposure to enriched acoustic environments in neonates and adults become understandable as well. Our recent findings of evoked potential studies in children with dyslexia and language delay suggests that the temporal processing deficits that these children show in behavioral tests are not reflected in cortical auditory evoked potentials. However, there are strong indications that these children exhibit delayed auditory maturation.

The importance of building developmental trajectories: Electrophysiological studies of language processing in Williams syndrome.
Manuela Friedrich, Ph.D. Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany
Williams Syndrome (WS) is a neurodevelopmental disorder caused by a microdeletion of approximately 25 genes on chromosome 7. WS is associated with relative proficiencies in language, face, and affective processing with deficits in spatial abilities as well as concomitant brain dysmorphology in dorsal relative to ventral stream areas. Much of what is known about abnormal brain structure and function in this population comes from studies of adults. However, developmental behavioral studies suggest that infants and children do not necessarily display the adult cognitive phenotype in WS. These studies suggest that it is important to understand the developmental process in order to fully understand the neurocognitive phenotype of the disorder. The present study examines developmental trajectories of brain activity linked to language processing in infants, children, young and middle aged adults with WS and from typically developing populations. By characterizing variability in the WS electrophysiological phenotypes of abnormal language processing, we can examine individual variability in genetic, neurocognitive, and behavioral profiles. More specifically, this methodological approach allows us to examine the extent to which individuals with partial deletions of the WS area map onto the WS or typical distributions, variability in brain function due to factors such as IQ, genetic imprinting, sex, and behavioral phenotypes.

Issues in the application of source analysis to developmental data

Gwen Alexandra Frishkoff, Ph.D. Research Fellow, Learning Research and Development Center, University of Pittsburgh Research Scientist, NeuroInformatics Center, University of Oregon
This talk will review basic principles of ERP source localization and issues in the application of source analyses to development (children's ERP) data. The talk will begin with a review of forward and inverse models, which are common to all source analysis methods, and then consider differences between discrete or sparse (i.e., equivalent dipole) and dense or distributed (e.g., linear inverse) approaches. While experts disagree about which approach is best (most accurate, most informative), I will suggest that distributed and discrete methods may have different strengths and weaknesses and may be optimal for addressing different types of questions. Beyond these general considerations, several issues are particularly important to consider in source analysis of child ERPs. These include: choice of head model (representation of cortical geometry and tissue conductivity estimates), selection of inverse model, and identification and removal of systematic (e.g., biological) and random noise. Examples will be drawn from analyses of simulated ERP data and from real data acquired from adults and children during lexical semantic processing tasks.

Comparing adult and child ERP components indexing syntactic and morphosyntactic processing
Arild Hestvik, Ph.D. Department of Linguistics and Cognitive Science, University of Delaware Speech and Hearing Sciences, CUNY Graduate Center
In two studies conducted jointly by Richard Schwartzs and Valerie Shafers labs at CUNY, we examined ERPs to both short-distance (tense) and long-distance (filler-gap) dependency violations in adults, TD children and SLI children (8-12 years range). The purpose was to determine whether children with SLI perceived ungrammaticality in the same domains where they make errors during production. For the shortdistance violations (as in Yesterday, I walk to school), adults and TD children both responded with a Left Anterior Negativity (starting 200ms after the offset of the verb). SLI children did not exhibit a clear ERP response, suggesting a lack of grammatical control of inflection. In the second study we violated filler-gap expectancies (as in *The zebra that the hippo kissed the camel on the nose ran far away); again both adults and TD children exhibited a similar early LAN response (about 115ms after the camel). The SLI children did not exhibit an eLAN. However, using spatial Principal Component Analysis, we were able to uncover another sub-component of the overall ERP response which did contain a condition effect. The onset latency of this response was later than in control groups (400900ms after the camel) and had a different topography (anterior positivity and posterior negativity). We interpret this as showing that SLI children did detect the ungrammaticality but did so later in time (and possibly using a different strategy). This illustrates how PCA can be used to aid in observing atypical ERPs in atypical populations.

Analyzing multichannel child ERP data: comparing different approaches

Paavo H.T. Leppnen, Ph.D. Academy Research Fellow, Docent Department of Psychology, University of Jyvskyl, Finland
High density event-related potential (ERP) data poses new challenges for analyses. Traditional analyses with peak or area measures are sufficient and providing enough information for many purposes with EEGdata measured with a small number of electrodes, such as telling at what time windows or, at a gross level, at what scalp areas or hemisphere certain brain processes take place. These analyses are, however, less satisfying when using multi-channel recordings, which provide rich spatial information not captured using peak or area measures for pre-selected electrode locations. Principal component analysis (PCA) and independent component analysis (ICA) analysis techniques have been developed and successfully used to summarize or encapsulate the multi-channel data both in time and spatial domains. These analyses allow one to obtain a limited number of temporal and/or spatial measures (components) which can be applied in further statistical analyses. One problem also faced by analysis techniques like these is the spreading of cortical currents due to volume conductance producing voltage maps covering large scalp areas. Current source density conversion of the voltage maps (CSD) provides more restricted maps, and for example in the case of a source in the auditory cortex, with well defined and more focused negative-positive maps on both sides of the Sylvian fissure. In the present study, I illustrate the use of PCA and ICA using examples from both adult data and data from school-aged children applying these analyses for both voltage maps and CSD-maps of averaged ERPs. I also discuss concerns and problems with different approaches, for example application of PCA/ICA to controls and clinical samples in the same model, when we expect different brain activity organization in different groups.

The paradox of interpreting longitudinal data in the context of dynamic systems

Dennis Molfese, Ph.D. Birth Defects Center, University of Louisville
One major frontier in developmental neuroscience that has received relatively little direct attention is the link between structure-function relationships as they relate to emerging cognitive functions in the developing human from birth into the childhood years. Much of the current speculation is based on extrapolations from adult populations and animal studies, with relatively little consideration of the dynamic nature of the developing neural and cognitive systems involved at all levels. However, advances in histological and neural imaging approaches provide some exciting glimpses into this domain. An overview of selected work from fMRI, MRI, histology, and electrophysiology reviews recent progress, suggestions are made for key methodological issues to address, and a model is presented hypothesizing neural organizational features that support emerging brain-cognitive relationships during early normal and abnormal development.

Single trial analysis of EEG

Lucas Parra, Ph.D. Department of Biomedical Engineering, City College of New York, CUNY
Electro-encephalography (EEG) is a readily available tool to study human perception and cognition. It has good temporal resolution compared to fMRI or PET, costs less than MEG, but lacks the specificity of those other modalities chiefly due to its limited spatial resolution. Over the last few years we have developed a couple of algorithms to extract information from event related activity in EEG. The main idea is to look at activity on a single trial basis while combining many electrodes to obtain sufficient signal-to-noise ratio. The main benefit of single-trial analysis is to quantify variability across trials and relate this to external observables, such as behavioral detection performance, reaction time, or stimulus attributes. The talk will give examples on what the experimenter may expect from such an analysis.

Neurodevelopment of auditory processing

Curtis Ponton, Ph.D. Chief Scientist, Compumedics Neuroscan Adjunct Professor, University of Texas at El Paso
Typically, the most critical manifestation of auditory neurodevelopment is the presence or absence of specific landmark behaviors within a certain age range. However, to understand the neurodevelopment of auditory processing, it is important to take into account all aspects of the human maturational landscape. Ultimately, the presence or absence of any auditory behavior is closely tied first, to anatomical development, and second, to the maturation of physiological processes within that anatomical substrate. Therefore, whenever possible, it is critical that we try to relate behavioral, anatomical, and physiological processes together in order to best understand the development of normal or disrupted development of auditory processing skills. The first objective of this presentation will be to describe normal auditory maturation starting with research on the auditory periphery and brainstem, through to cortex and place these developmental changes in the context of physiological and behavioral maturation in the auditory system. The second objective of this presentation will be to examine the impact of deafness on maturation of the central auditory system based on electrophysiological, evoked potential measures of brain activity recorded from profoundly deaf children who use a cochlear implant. Finally, approaches for future clinicallydirected research will be suggested.

Relevance of electroencephalographic (EEG) responses for assessing auditory cortical function in humans and monkeys

M. Steinschneider1, Y. Fishman1, J.C. Arezzo1, H. Kawasaki2, H. Oya2, and M. Howard2 Albert Einstein College of Medicine1, Bronx, NY University of Iowa College of Medicine2, Iowa City, IA.
While evoked potentials are capable of probing cortical physiology at high temporal resolution, it is uncertain whether they are optimal indices of auditory cortical organization. Averaging procedures maximize phase-locked activity within lower frequency bands at the expense of non-phase-locked activity within higher frequency gamma bands (> 30 Hz). Thus, a central goal of this study was to compare the relative sensitivity and specificity of EEG responses within different frequency bands in primate A1 and to illustrate translational relevance of these responses for assessing human auditory cortical function. Presentation of best frequency tones increased EEG power across the range of frequencies examined (4290 Hz) in monkey A1. The largest relative increases in power occurred at very high gamma bands (130210 Hz). Power increases at frequencies from 30-210 Hz were more reliable than power increases in lower frequency bands, and were better correlated with the tonotopic organization than power increases in lower frequency bands. Intracranial recordings from human auditory cortex revealed large amplitude increases in high gamma power similar to that seen in the monkey. These increases persisted for a more prolonged period than increases in low frequencies. Later increases in high gamma occurred in surrounding cortical areas, suggesting that they may be a valuable tool for examining informational flow in auditory cortex. These findings highlight the value of examining high frequency EEG components in exploring the functional organization of auditory cortex. Full details of the primate study are available (Steinschneider et al., Cerebral Cortex, 2007). Supported by DC00657 and DC042890.

DANSSL 2007

Poster Abstracts

Functional significance of the mismatch response in infants

Hia Datta, Karen Garrido-Nag, Valerie L. Shafer Speech and Hearing Sciences, CUNY Graduate Center
The Mismatch Negativity (MMN) is an event related potential attributed to the detection of a change in auditory input from an established sensory memory. MMN has been reliably elicited to various auditory inputs (Naatanen, 1990; Aaltonen et al. 1992; Ritter et al. 1995) in the adult population. The findings for infants, however, have been controversial. Experimenters have reported a negative as well as a positive component (PC) as the mismatch response to change in auditory stimuli for infants (Morr et al. 2002; Cheour et al. 1998). Others (e.g. Kushnerenko E., 2003) suggest that the PC might be functionally comparable the P3a (an index of the orienting response) found in adults. We examined the amplitude of the PC in infants from ages 4 to 8 months across the time course of an experiment. Infants listened to 1000 vs. 1200 Hz tones in an oddball paradigm. Data were collected from seven sites (Fz, Cz, F3, F4, left and right mastoids, Oz) with Oz as the reference. The electroencephalogram was bandpass filtered between .05 and 100 Hz and digitized at 256 Hz. The data were then segmented and averaged to the standard and deviant stimuli with a prestimulus baseline of 75 milliseconds. A clear PC was elicited to the deviant stimulus. Comparing the amplitude of the positivity for the first 100 versus the latter 100 trials showed a decline in positivity across the timecourse of the study. We will discuss possible explanations for this decline with reference to the functional significance of the PC.

Comparison groups in studies of developmental disorders

Hilary Gomes1,2, Martin Duff1, Miguel Ramos1, Sophia Barrett1, and Alice Brandwein2 1 Cognitive Neuroscience, CUNY Graduate Center and City College 2 Neuropsychology, CUNY Graduate Center and Queens College
In a study of auditory selective attention in children with language impairments (LI), we initially decided to recruit a comparison group of clinically referred children who did not meet criteria for LI. Referred control groups are frequently used in clinical research as they are thought to better control for miscellaneous participant variables. However, unexpected findings in our comparison group prompted us to recruit an additional group of typically developing children. This poster presents the data from the two control groups. Sixteen children (5 females; mean age 10.4 years (1.6 sd)) referred for academic and/or behavioral difficulties in school but who failed to meet criteria for LI or AD/HD form the clinically referred group and 9 children (5 females, mean age 10.8 years (2.1 sd)) recruited from a middle school for gifted and talented students in math, science, and technology form the typically developing group. The task required that participants attend to one of two stimulus channels, differentiated by frequency and location, and respond to infrequent targets of shorter duration or lower intensity. Nds, which reflect the differential processing of attended and unattended stimuli, were elicited from the typical control group in both conditions but only from the clinically referred controls when they were listening for duration targets. Clearly, the composition of the comparison group can impact on the conclusions of a study. This is potentially extremely problematic for electrophysiological research in children with developmental disorders as there is only limited normative data for many of the components of interest.

Processing of rise time in paired tones by children with dyslexia

Jarmo A. Hmlinen, Paavo H. T. Leppnen, Tomi K. Guttorm, Heikki Lyytinen Department of Psychology, University of Jyvskyl, Finland
Processing of two different rise times (10 and 130 ms) was investigated in paired tones with two different intervals between the tones in the pair (10 and 255 ms) in children with dyslexia and typically reading children. Temporal principal component analyses (PCA) were used to identify event-related potential (ERP) components and factor scores for the components of interest were chosen for further analyses. In an equal probability paradigm with 1 5 second inter-stimulus intervals typically reading children showed smaller N1 response to a longer rise time compared to a shorter rise time in the second tone of the pair when the tones in the pair were close to each other. In contrast, children with RD had equally large N1 to both short and long rise times. They also had larger N1 to the longer rise time than control children. When the same stimuli were presented in a passive oddball paradigm, the children with RD showed larger response at the mismatch negativity (MMN)/N1 time window and smaller late discriminative negativity (LDN) response compared to control children in their responses to the rise time change when the tones in the pair were far apart. These findings support earlier behavioral findings of deficient rise time discrimination in children with dyslexia. The atypical rise time processing was seen already in N1 the exogenous response suggesting problems in onset detection mechanisms.

How song learning experience affects auditory responses in the adult zebra finch
K.K. Maul1,2, H.U. Voss3, L. C. Parra4, D. Salgado-Commisariat5, O. Tcherniechovski1,2, *S.A. Helekar5 1 Biol. Dept. City Col. of New York, New York, NY; 2Speech and Hearing, CUNY Grad Ctr., New York, NY; 3Weill Med. Col. of Cornell, Citigroup Biomed. Imaging Ctr., New York, NY; 4Biomed. Engineering, City Col. of New York, New York, NY; 5Dept Neurol., Methodist Neurological Inst., Houston, TX
Songbirds and humans are among the few species that exhibit vocal learning modifying their vocalizations over development to match same-species sounds heard in their environment. Like humans, many songbirds also have a sensitive period for typical vocal development, which closes once the bird has learned to imitate a song. Using BOLD-fMRI, we show that auditory responses to sounds and songs differ significantly in birds that were deprived of external auditory stimulation (isolates) and box tutored, livetutored, and female control birds. Isolates show, in contrast to trained birds, strong activation of premotor brain areas that are required for song learning. In trained birds, but less in isolates, activation of auditory areas showed sound selectivity. Our results suggest that the song system becomes less sensitive to sounds with song learning, whereas responses of auditory areas become more selective. These findings are corroborated by Local Field Potential recordings in the primary song system and auditory nuclei. We conclude that early vocal experience could shape auditory perception in songbirds as it does in human, and that BOLD fMRI is a suitable method to study auditory developmental learning in songbirds.

Electrophysiological Evidence of Lexical Access Disruptions

Yael Neumann, Valerie Shafer, Loraine K. Obler, and Hilary Gomes Speech and Hearing Sciences, CUNY Graduate Center
This study investigated in healthy younger and older adults: 1) the effects of age on the phonological retrieval process, and 2) the time-course of the phonological substages, segments vs. syllables. The N200, P300 and Visual Evoked Potential (VEP) components were investigated using a Go/Nogo paradigm with implicit picture naming. Results support the Transmission Deficit Hypothesis as N200 latencies on both phonological tasks were later (100 ms) in the older, as compared to the younger, group. In particular, within the older group, retrieval of syllabic information was significantly later (50 ms), as compared to retrieval of segmental information. These findings further suggest that the phonological breakdown is greater at the syllabic, than segmental, level. Additionally, P300 and VEPs revealed no significant latency delays between groups and on both tasks. This suggests that phonological processing delays in the older group were due to stage-specific, not generalized, phonological processing deficits. However, group differences in P300 amplitude on the segment task, and VEPs (at 201-250 ms) on both tasks, were seen. These findings suggest that with age greater cognitive effort was needed to perform the segment task, and greater visual attention was expended on both tasks. Lastly, results support the parallel view of processing for segmental and syllabic phonological substages, although in the older adults the data showed that syllabic access is affected more than segmental access. Implications of the study are that healthy older adults might benefit from practice with phonological tasks, mainly of syllabic information, to improve retrieval.

Finding the Forest and the Trees: Symbolic encoding treatments of language related ERP data.
Douglas Saddy and Peter beim Graben, School of Psychology and Clinical Language Sciences, University of Reading
Neural events as measured from individual neurons up to large scale anatomical regions are seen to be electro-chemical oscillations. Nonlinear complex system approaches to data analysis provide both new analytic measures to apply to the signals we record as well as a new set of computational modeling approaches for characterizing and emulating cortical oscillations. An important advance for this approach is that the same tools are used for analytic and computational modeling analysis. This contrasts with the more traditional approach in which data analysis was essentially orthogonal to the modeling techniques. However it needs to be stressed that these are models of the physical observations. Models of cognitive events are also needed and indeed the interpretation of the relation between the physical systems and their interactions and cognitive phenomena cannot be approached without clear models on both sides. Symbolic Encoding is a complex systems tool that is effective in discriminating oscillatory behaviors. It provides an effective general approach to signal analysis for dense time series like those collected in EEG/ERP paradigms. Using data from a series of ERP paradigms that are aimed at discriminating between structural and semantic variables during real time sentence processing, we will demonstrate how symbolic encoding approaches can be used to augment and refine conventional, averaging based, data analysis methods, and how they can reveal systematic processing variations that cannot be detected using the conventional techniques.

ERPs reveal atypical pitch-change processing in newborns at-risk for dyslexia who later become dyslexic
Hanne K. Salminen, Jarmo A. Hmlinen, Tomi K. Guttorm, Kenneth Eklund, Annika Tanskanen, Heikki Lyytinen and Paavo H.T. Leppnen Department of Psychology, University of Jyvskyl, Finland
It has been widely accepted that problems in phonological processing are causally related to dyslexia. In contrast, the role of auditory processing deficit in dyslexia has been debated for several decades. Some studies have found only speech processing deficits in dyslexic readers, whilst several recent studies have found deficits for example in processing of pitch differences in tones. We sought to ascertain whether the pitch processing deficit can be seen already in newborns who later became dyslexics. We measured auditory event-related potentials (ERPs) to pitch change from 31 newborns, 8 of whom had familial risk for dyslexia and later became dyslexic, and 23 who had no family history of dyslexia and who later acquired normal reading skills. Using time window identified with temporal principal component analysis we found differences between dyslexic readers from the at-risk group and normal readers from the control group in their newborn ERP responses to pitch change. Even at birth normally reading controls showed clear differentiation of pitch in tones, whereas dyslexic readers failed to show any differentiation. Dyslexic readers and controls differed from each other in their responses to the deviant stimulus at the right fronto-central recording sites. The results of our study strongly suggest that dyslexic readers are affected by atypical basic auditory processing that is already present at birth and connected to later reading outcome. This very early nature of the deficit could indicate that basic auditory processing problems could underlie speech perception deficits, affecting in turn the phonological representations, and ultimately dyslexia.

Child-Adult Differences in the Working Memory LAN

Baila Tropper Speech and Hearing Sciences, CUNY Graduate Center
Several studies support the use of the left anterior negativity (LAN) as an index of the storage of a syntactically displaced constituent in working memory. These investigations have demonstrated that sentences with higher relative to lower working memory demands are associated with greater negativity over left frontal regions in adults (e.g., Mecklinger, Schriefers, Steinhauer, & Friederici, 1998; Phillips, Kazanina, Wong, & Ellis, 2001). We used the working memory LAN to examine the processing of subject and object wh-questions in children with typical development (TD), children with specific language impairment (SLI), and in adults with normal language. We hypothesized that the storage of syntactic information over the filler-gap distance would be demonstrated by a sustained left anterior negativity for object relative to subject questions. Mean voltages of multi-word ERPs for the adult group confirmed this expectation, consistent with prior evidence. The ERP data for the child groups revealed a working memory effect that markedly differed from the adults: a sustained anterior positivity for object as compared with subject questions in the left and right anterior regions in TD children and in a small section of the right anterior region in SLI children. These findings demonstrate that ERP components tied to high-level language processing may differ in polarity and topography across child and adult populations. Possible explanations for these differences will be discussed.

ERP Indices of Speech Processing in 6-month-old Bilinguals and Monolinguals: Topographic differences?
Yan H. Yu1 & Valerie L. Shafer1 1 Speech and Hearing Sciences, CUNY Graduate Center
Studying speech sound processing abilities in infants can provide important information about the association between speech processing deficits and language impairment in young children (e.g., Trehub & Henderson, 1996). Discrimination and categorization of speech sounds are shaped by early language experience (Werker & Tees, 1984), and speech discrimination in bilingual versus monolingual infants develops differently (Bosch & Sebastian-Galles, 1997; Burns, Werker & McVee, 2002; Sundra & Polka, 2004). The purpose of current study is 1) to investigate whether bilingual exposure to Spanish and English affects processing of speech stimuli that are phonemic only in English at six months of age; 2) to investigate appropriateness of implementing three different methods (GFP, TANOVA, & selected sites ANOVA) of data analysis in current study. Nineteen 6-month-old infants listened to 250ms-long, phonetically similar vowel contrasts (I vs. E) presented in an oddball paradigm while event-related potentials (ERPs) were collected from 65 scalp sites. Global field power (GFP) was used to decide time range (160-360 msec) for topographic analysis of variance (TANOVA) and also time range for ANOVAs on selected sites. Topographic surface potential voltage maps were made using BESA 5.1 to assist selecting clusters of sites of interests for traditional ANOVA analyses. GFP peaks in the time range of 260-290msec for both monolingual and bilingual groups, with monolinguals slightly earlier. Between group comparison from TANOVA obtained p-values that approach significance in the time range of 260-300msec. Traditional ANOVA suggests that there are topographical differences at the mastoids. Strengths and weaknesses of each method are discussed.