Electrophysiological studies of language and language impairment

Electrophysiological studies of language and language impairment

Electrophysiological Studies of Language and Language Impairment Debra L. Mills and Helen J. Neville The organization of language-relevant brain syste...

862KB Sizes 0 Downloads 17 Views

Recommend Documents

Phonology and children with specific language impairment
Two studies are reported in which the phonological characteristics of preschool-age children with specific language impa

Executive functions and language in children with different subtypes of specific language impairment
The marked heterogeneity among children diagnosed with specific language impairment (SLI) highlights the importance of s

Language effects in second-language learners: A longitudinal electrophysiological study of spanish classroom learning
•New learners of an L2 show increasing N400 ERP amplitudes as a function of learning.•These changes occur over a single

Relation between language, audio–vocal psycholinguistic abilities and P300 in children having specific language impairment
Specific language impairment is a relatively common developmental condition in which a child fails to develop language a

Structural alterations of the language connectome in children with specific language impairment
•Diffuse deficits of language connectome were found in children with SLI.•Ventral language stream structural alterations

The Neural Basis of Language Development and Its Impairment
The neural correlates of early language development and language impairment are described, with the adult language-relat

Severe Impairment Battery Language scale: A language-assessment tool for Alzheimer's disease patients
Communication problems are common in Alzheimer's disease (AD) patients, but instruments to assess these symptoms ar

Association between audiological profile and primary language impairment in children
The aim of the present study was to characterize the Brazilian Portuguese speaking children seen at the Speech-Language

Electrophysiological Studies of Language and Language Impairment Debra L. Mills and Helen J. Neville The organization of language-relevant brain systems was examined in normally developing and language-impaired children. Atypical patterns of brain activity were observed in subsets of children with specific language impairment (SLI) for both sensory (auditory and visual) and language processing. However, it was not the same groups of children who displayed abnormalities across the different tasks. The results supported a multiple-factors and multiple-subtypes framework for interpreting the neurobiology of SLI. The roots of SLI were also considered in normal infants and late talkers in studies of primary language acquisition. These studies suggest that the organization of neural systems important in language acquisition display dramatic changes during this time. Some of these are linked to the attainment of language milestones and appear to be independent of chronological age. Moreover, abnormalities in the lateral organization of electrophysiological activity may help predict which late talkers will catch up and who will later display SLI. More generally, the event-related potential technique is a powerful tool in studying the neurobiology of language and language impairment. Copyright 9 1997 by W.B. Saunders Company

VER THE PAST century, several different experimental methods and techniques have been used to characterize the structural and functional organization of the brain in the adult human. These studies show a highly differentiated mosaic of systems and subsystems that are specialized to support different and specific types of sensory and cognitive activity, including language. At present, we know very little about the development of this collection of cerebral systems, the degree to which different aspects are maturationally constrained, and the role of input from the environment in specifying their functional architecture. Basic research describing both the cerebral systems important for language in the adult and in normally developing infants and children is essential to an adequate characterization of developmental language disorders. It is also central to an understanding of the mechanisms underlying developmental language impairment. Over the past several years, we have used the event-related brain potential (ERP) technique to characterize languagerelevant systems in normal adults, alterations in these systems in adults with different language experience (eg, deaf signers, bilinguals), the development of the language systems of the brain in infants and children, and to characterize these systems in language-impaired children and in language-delayed infants. The results from these studies of children and infants are summarized later. It is clear from this perspective that the neural systems important in the acquisition of language are constantly shifting and undergo massive shifts in their organization during the first decade of life at least (Fig 1). This undoubtedly accounts for the markedly different effects of cerebral lesions in

O

Seminars in Pediatric Neurology, Vol 4, No 2 (June), 1997: pp 125-134

adults and children (see Chapter X; see also 1). It is against this dynamically and constantly shifting neural substrate that we must access sensory, attentional, mnemonic and linguistic processing in children with language disorders. THE ERP TECHNIQUE

ERPs are recorded from electrodes on the scalp, as in the EEG, but with a bandpass that includes lower frequencies (usually 0.01 or 0.1 Hz) that are important in cognitive processing. ERPs are electrical events synchronized to the occurrence of stimulus presentations of interest, for example, all the nouns in a set of sentences. They are averages over several stimulus presentationsithe number required varies depending on the amplitude of the signal (40 is often adequate for language stimuli). ERPs reflect the earliest stages of sensory processing as well as later effects of cognitive processing including arousal, attention, mnemonic factors, meaning, and linguistic form. Therefore, it is necessary to assess or control for these factors before attributing differences in ERPs to language processing. Large individual differences in structural anatomy and cortical physiology make the ERP ill suited for assessment of an individual at From the University of California at San Diego, San Diego, CA; and the University of Oregon, Eugene, OR. Supported in part by Grants DC00481, DC00128, DC01289, and NS22343 from the National Institutes of Health, Bethesda, MD. Address reprint requests to Debra L. Mills, PhD, Center for Research and Language, University of California, San Diego, LaJolla, CA 92093-0113. Copyright 9 1997 by W.B. Saunders Company 1071-9091/97/0402-000955.00/0 125

126

MILLS AND NEVILLE

!nfemts//~ ,.'~, ~,/~, j , , , ,

6 mo

5 - 6 yr (N=11)

(.=10)

10 m o

,.~, ~, ,~,~,,,, ,

(.=11)

"~'V'

8-10 ~

' ' ' ' ' '

18-17 mo

(N=:~)

11-12

(N=S8)

(N'-16)

20 m o

13-14, yr (.-21)

(.-24)

28-30 mo

~

1 5 - 1 8 yr

(,=14)

(.:~)

36-42, m o (r-SO)

Ch4]d.ren

1 7 - 1 8 yr lVl|l|

"1/

I

I

I

I

I

I

I

I

I

(,..12) 19-22 ~'

(,-22) 1 9 - 2 2 yr

(.-22)

2.0 / ~ V - I ' , , , I , , , I , ,

0

400

800 ms

present. However, within a group differences in experimental condition are often very robustly superimposed on individual differences in the morphology and size of response. SPECIFICALLY LANGUAGE-IMPAIRED CHILDREN

Behavioral studies of specifically languageimpaired (SLI) children have raised several different hypotheses about the factors that may give rise to their language deficit, including basic problems in auditory or visual sensory processing, specific

.... famflll. WOr~

Fig 1. ERPs to familiar auditory words over left anterior temporal regions from infants 6 to 42 months of age (left side) and from school-aged children and adults (right side).

problems in grammatical or phonological processing, attentional or mnemonic problems. In view of this large literature, it is important that the same group of SLI children be evaluated on as many of these dimensions as possible. ERPs to simple auditory and visual sensory stimuli have provided input on the integrity of aspects of sensory processing in normal adults and children and those with altered sensory experience or cortical damage. 2-4 Using these same paradigms, we have observed that a subset of SLI children (ages 8 to 10 years) display evidence of early

ELECTROPHYSIOLOGICAL STUDIES OF LANGUAGE

sensory processing deficits within both the auditory and visual modalities. The earliest responses (within 250 msec of stimuli onset) display timing delays and reduced amplitudes in some, but not all, SLI children. In addition, we have compared aspects of language processing in these same SLI children and normally developing controls. We reported that ERPs to words that provide grammatical information in English (eg, function or "closed class" words including "the, but, if, and") elicit highly focal and asymmetrical ERP responses over anterior regions of the left hemisphere in normal adults (the left anterior negativity [LAN]). 5 By contrast, in adults who learned English late and imperfectly, this response is absent or aberrantly distributed. 5,6 Normal children aged 8 to 10 years display a pattern of results that is similar to, but less focal than that observed in native-speaking adults. We have observed that a subset of SLI children, specifically those who score poorly on tests of English grammar, display a highly aberrant response to closed class words that lacks the left hemisphere asymmetry. However, this subset of SLI children does not consist of the same children who display sensory processing deficits. In addition, we observed that many of these same SL1 children tend to display larger than normal N400 responses to less predictable words and sentence final words that render a sentence nonsensical. The N400 response indexes the degree to which a word is expected or primed in a given situation. 7 Thus this effect is likely to arise from SLI children's greater reliance on contextual cues secondary to their auditory or visual processing deficits or to their syntactic deficits. Figure 2 displays ERPs in response to visually presented closed class words presented in sentences 8 from SLI and control children. This displays the diminished sensory responses P150 and P350 over occipital regions, the reduced asymmetry over temporal regions, and the enhanced N400 response in SLI children compared with matched controls. These results, showing multiple aspects of processing affected heterogeneously across the same LI sample, are consistent with a multiple factors and multiple subtypes framework for considering the mechanisms that give rise to SLI. Future research will determine the contribution of neural systems important in attentional and mnemonic processing in SLI and will further specify the

127

nature of the auditory, visual, and linguistic processing deficits noted here. PRIMARY LANGUAGE ACQUISITION IN NORMAL INFANTS AND INFANTS AT RISK FOR SLI

Another important line of research is to determine the roots of SLI in young infants as they acquire language for the first time. This necessarily involved characterizing the neurobiology of language acquisition in normally developing infants and children and the variability in this process. Only then were we able to search for links to SLI. ERPs AND PRIMARY LANGUAGE ACQUISITION

The course of primary language acquisition is characterized by the attainment of different language milestones at several key ages. 9 Typically, children understand a few simple words by 10 months, say their first words around 13 months, begin to put two words together at 20 months, and speak in short phrases by 28 to 30 months. Most children speak in full sentences by the time they are 3 years and have acquired most aspects of English grammar by 4 years of age. However, there is considerable individual variability in the precise age at which normal children reach these milestones. Why do some children speak in sentences at 18 months while other normal children do not say their first words until after they are 2 years old? This research examines patterns of brain activity that may differentiate precocious language learners from children who lag behind the norm. Additionally, this technique may provide a promising tool for predicting which "late talkers" will catch up from those who may remain language delayed. The use of event-related potentials is one of the few noninvasive techniques available to study developmental changes in the organization of brain activity. The ERP technique is also especially suited for studying infants and toddlers because it does not require an overt response. Performance on behavioral language assessment tasks may be influenced by a variety of factors unrelated to the child's language abilities, that is, willingness to cooperate, shyness, memory span, and coordination skills needed to execute the task. Although ERPs bypass some of these limitations, behavioral assessment of language abilities are necessary to determine which ERP components are linked to language abilities. Therefore, we use a combined behavioral-electrophysiological approach.

128

MILLS AND NEVILLE

Right Hemisphere

Left Hemisphere

+"-L " ~ l '

....

Frontal

§

+VL" ~ j

....

Anterior Tonperal

"

Toiperal

"

+71 " V

" ~i

,/,,~...

+~-J- " T t l

. . . .

V .//x~2:.

,

.... +~-/

-

....

.~.

_T..j~

" V'-'v

"~

Parietal

9

N230 rI ~t

s I I~

h : 9 ".'.~.Y'O

Occipital

"_T

N400

li. , ',~,,.

Fig 2.

PtB0

"~o: ,~o' r~o' Bill:

The research described here examines developmental changes in neural systems linked to the attainment of specific language milestones during the first 4 years of life. To this end we study changes in the organization of language-relevant brain systems in children as they pass through different ages and contrast these with changes linked to vocabulary size when age is held constant. Additionally, we examine ERP patterns from 28- to 30-month-old late talkers that may be predictive of language performance at a later age.

"-"

Language Impaired Controls

ERPs to closed class

words in the middle of sentences from left and right frontal, anterior temporal, temporal, parietal, and occipital sites of control and U children. (Reprinted with permission. 8)

SUBJECTSAND PROCEDURE Procedures and subject recruitment were the same for all studies described as follows.

Subjects The normal infants were recruited through advertisements in local parent magazines and consisted of healthy full-term infants with monolingual exposure to English. Approximately half of the subjects at each age group were girls. Late talkers were

ELECTROPHYSIOLOGICAL STUDIES OF LANGUAGE

recruited through the Project in Communication and Neural Development (PICND) at the University of California, San Diego, and scored below the 10th percentile for children in their age range on the Mac Arthur Communicative Development Inventory (CDI) at the time of testing.

Procedure More detailed information about the procedures has been described elsewhere. ~0.11Briefly, EEG was recorded over frontal, temporal, parietal, and occipital regions of the left and right hemispheres, and from central and parietal midline sites. Additionally, we recorded from over and under the eye to reject trials on which blinks and vertical eye movements occurred and compared EEG from the left and right frontal electrodes to reject trials containing horizontal eye movement. All recordings were referenced to linked mastoids with a bandpass of 0.1 to 100 Hz. Impedances were kept at or below 5 Khoms. Before participating in the electrophysiological portion of the study, each child was seen for language testing. The behavioral measures were used to determine the list of comprehended words. The tests included a language comprehension task (ie, pointing to pictures and objects), and two parental report measures: (1) The Mac Arthur CDI, which was designed to provide an estimate of the size of a child's vocabulary for both language comprehension and production; and (2) a rating scale for a subset of the words from the CDI, which provided a measure of how confident the parents were that their child understood or said each word in a variety of different contexts and with a variety of exemplars. During ERP testing, the children sat on their parent's lap and watched a moving puppet as they listened to a series of words presented through a speaker located 40 inches in front of them. The stimuli included 10 words whose meanings were comprehended by the child, 10 English words whose meanings the child did not understand, and 10 backward words. Each word was repeated 6 times for a total of 180 trials. Approximately 60% of the trials were artifact-free and included in the analyses. All effects reported here were significant at or below P = .05 unless otherwise specified. CHANGES IN ERPs FROM 6 TO 42 MONTHS There are dramatic changes in the morphology, latencies, and distribution of ERP components to

129

auditory words during the first 4 years of life and beyond (Fig 1). Figure 1 shows ERPs to words from children at several key ages from 6 months to 3 years of age and adults. ERPs to words are characterized by a positive component at approximately 100 msec (P100), two negative components at approximately 200 and 350 msec (N200) and (N350), respectively, followed by a broad negative component from 600 to 900 msec (N600 to 900). The P100 observed here in infants is probably the functional equivalent of the auditory P1 (30 to 100 msec) observed in adults. The amplitude of the auditory P1 can be modulated by changes in the physical parameters of the stimuli (ie, loudness, duration) and shows little variation in amplitude and latency across subjects who are similar in age. 12-14 In the infant data, the amplitude and latency of the P100 decreased with increasing age but showed considerable reliability across individual subjects within a given age group in its latency and distribution. The P100 displayed an anterior distribution and a left greater than right asymmetry. This pattern was consistent with other reports of left hemisphere asymmetries to speech versus nonspeech stimuli in this age range. J5 The distributions of the N200 and N350, which have been linked to word meaning, 1~ changed markedly with age. Consistent with previous studies, the latencies of these components decreased with increasing age? 3,~7 Initially, from 6 to 17 months, the amplitudes of these components were larger over posterior than anterior regions. By 20 months, the amplitudes of the N200 and N350 were most prominent over temporoparietal regions and were attenuated over frontal and occipital areas. Changes in the distribution of these components may reflect differentiation of the auditory system. The functional significance of these components is discussed in more detail later. The N600 to 900 displayed a frontal distribution, and consistent with other research, became less prominent with increasing age. Large frontal negativities have been observed in infants and children to a variety of both auditory and visual stimuli and the amplitude of this response is modulated by stimulus probability. ~8-21 In summary, developmental changes in ERPs to auditory words are characterized by decreases in the latencies and increases in amplitudes of the components linked to language processing over temporoparietal regions. The right side of Figure 1 shows continued development in ERPs to auditory

130

MILLS AND NEVILLE

words, including changing morphology, through the late teens. These developmental changes in ERPs, from childhood through adolescence, are described in detail elsewhere.13

cant only over temporal and parietal regions of the left hemisphere. Moreover, there are no significant amplitude differences over frontal or occipital regions of either hemisphere. 10,1~ In summary, the data from these two age groups show that there are differential brain responses to comprehended and unknown words by 200 msec after the onset of the word. At 13 to 17 months, these differences are widely distributed over anterior and posterior regions of both hemispheres. In contrast, by 20 months of age, these differences are limited to temporal and parietal regions of the left hemisphere. Thus, these data suggest increasing lateral and anterior-posterior cerebral specialization for language comprehension from 13 to 20 months. Although there are marked changes in the distribution of this effect, from bilateral to left lateralized, this is not the mature pattern. The left hemisphere asymmetry in the ERP differences to known and unknown words is driven by a standing right hemisphere asymmetry to unknown words. The more mature left hemisphere asymmetry to words is not observed until later in development.

ERP INDICES OF LANGUAGE COMPREHENSION

Changes in ERPs to Known and Unknown Words From 13 to 20 Months Of particular interest was to observe brain responses to words whose meaning was understood by the child versus words whose meaning was not understood by the child. 13 to 17 Months. ERPs to comprehended and unknown words are directly compared for the 13 to 17-month-olds in the top half of Figure 3. The solid lines represent ERPs to comprehended words and the dashed lines show ERPs to unknown words. The shaded area shows that the amplitudes of the N200 and N350 are larger to comprehended than unknown words. This effect is present in both hemispheres and is significant at all sites anterior to the occipital regions. These results show that at 13 to 17 months, ERPs to words whose meanings are understood by the child are differentiated from ERPs to words whose meanings are not understood by 200 msec after word onset. Additionally, these differences are bilateral and widely distributed over areas anterior to the occipital regions. 20 Months. ERPs to comprehended and unknown words from the 20-month-olds are shown in the bottom half of Figure 2. The shaded area shows that, like the 13- to 17-month-olds, the difference between comprehended and unknown words is present at 200 and 350 msec. But in contrast to the 13- to 17-month-olds, these differences are signifi-

A

Changes Linked to Level of Ability Independent of Chronological Age To further assessthe extent to which the changes

in the organization of"brain activity were caused by changes in lexical abilities independent of chronological age, we examined ERPs to comprehended and unknown words in two groups of 13- to 17-month-olds.H The subjects described previously were divided into two groups who differed in the size of their receptive vocabularies. The high comprehenders scored above the 50th percentile, and the low comprehenders scored below the 50th

Right Hemisphere ~o ~6oo-ooo

Left H e m i s p h e r e

PIO0,->

temporal -I~ . . . . . . .

5.o,,v[, 0

,,, 400

,,, 1000

(.~,o)

I. 1600

C o m p r e h e n d e d Words

' U,,,~,,,,,qrn'lords

Fig 3. ERPs to known versus unknown words over left and right temporal regions in infants 13 to 17 months (A) and 20 months of age (B).

ELECTROPHYSIOLOGICAL STUDIES OF LANGUAGE

percentile, on the number of words they understood according to the MacArthur CDI. The low comprehenders, like the data described above, showed broadly distributed ERP differences to comprehended and unknown words over frontal, temporal, parietal, and occipital regions of both the left and right hemispheres. In contrast, the high comprehenders, like the 20-month-olds, showed a more limited distribution of ERP responsiveness to the different types of words. The high comprehenders, like the 20-month-olds, showed ERP differences only at sites anterior to the occipital regions and over the left but not right parietal regions. Unlike the 20-month-olds, the 13- to 17-month-old high comprehenders also showed ERP differences over both the left and right frontal and temporal regions. These results are consistent with the hypothesis that increased specialization of brain activity is linked to increased language abilities and that the observed changes are, at least in part, linked to increasing vocabulary size and are independent of chronological age.

Changes in ERPs to Semantic Versus Grammatical Function Words From 20 to 42 Months Previously, we reported evidence that different aspects of language, that is, semantic and grammatical processes, are mediated by nonidentical neural systems and mature at different rates. 5,8,~3,2z In this section, we study the development of these specialized systems during the course of primary language acquisition. We chose to study open and closed class words because of the different roles they play in English sentences. Open class words convey semantic information and make reference to specific events or attributes and include nouns, verbs, and adjectives. Closed class words typically convey information about grammatical relations and include such categories as prepositions, articles, and auxiliaries. The acquisition of open and closed words have different developmental trajectories. Closed class words are typically acquired later and are more vulnerable in abnormal development than open class words. In our studies of normal hearing adults, ERPs to open and closed class words presented in written English sentences show marked differences in the latencies and distributions. As described above, closed class words elicit a negative response that displays a left anterior distribution (LAN) that is absent or attenuated to open class words. In con-

131

trast, open class words elicit a negative response that is later (around 350 or 400 msec) and displays a posterior distribution and tends to be larger over the right than the left hemisphere. The findings from adult native and non-native speakers as well as studies of normal and language impaired children 8 and children and adults with Williams Syndrome 22,23 suggest that the LAN is linked to grammatical processing. When do these systems become specialized? Are the responses to open and closed class words different from the onset of speech? Or do they become progressively differentiated as children become more sophisticated in their language use? The top of Figure 4 directly compares ERPs to open and closed class words in 3-year-old normal children. The shaded areas represent ERP differences between open and closed class words that are statistically reliable. 24 Open and closed class words are different by 150 msec after the onset of the word. The negativity at 200 msec (N200) to open class words is larger than to closed class words over frontal, temporal, parietal, and occipital regions of both the left and right hemispheres. The second negative component at 450 msec (N450) is larger to open class than closed class words over the right hemisphere. Differences over the left hemisphere were not significant. More importantly, open and closed class words display markedly different distributions. The N200 and N450 to closed class are larger from the left than the right hemisphere over temporoparietal regions. In contrast, ERPs to open class words elicit a symmetrical N200. Over posterior regions, the N450 to open class words is larger over the right than the left hemisphere. These data suggest that, like the adults, 3-year-old normal children show a left hemisphere asymmetry to closed class words and display symmetrical or a right hemisphere asymmetry to open class words. These data suggest that nonidentical neural systems mediate processing of these two-word classes. These differences are not surprising given that most 3-year-old children are quite proficient with English g r a m m a r - they speak in sentences and use closed class words to specify relations among the open class words in those sentences. At 28 to 30 months of age when children are typically just beginning to speak in multiword phrases, ERP differences to open and closed class words are also observed (middle of Fig 3). However, the distribution of the differences to open and

132

MILLS AND NEVILLE

closed class words vary markedly from the pattern displayed by the 3-year-olds. Like the 3-year-olds, the NZ00 to open class words is larger than to closed class words, but only over the left hemisphere. In contrast to the left hemisphere asymmetry observed in the 3-year-olds, ERPs to closed class words are larger from the fight than the left hemisphere. In summary, the data suggest that as children begin to acquire grammar, different neural systems mediate the processing of open and closed class words; however, the left hemisphere asymmetry to closed class words is acquired later than 28 to 30 months. What happens earlier when children are speaking in single words or just starting to put words together? At 20 months of age, there were no reliable ERP differences to open and closed class words (bottom of Fig 4). All of these children understood and produced both the open and closed class words. Moreover, the ERPs to both open and closed class words differed from ERPs to unknown words. However, there was significant variability across children. To determine if there were differences in language abilities that might account for the variability in ERP responsiveness, we divided the children into two groups based on the size of their productive vocabulary. At 20 months, vocabulary size is the best predictor of later grammatical abilities. 25 The results showed that the 20-month-olds who

scored above the 50th percentile on the MacArthur Inventory displayed ERP differences to open and closed class words that were similar to the 28- to 30-month-old children, z4 In contrast, the 20-montholds whose vocabulary size fell below the 50th percentile did not show ERP differences between the word classes. In addition, we also studied 28- to 30-month-old late talkers. These children had a productive vocabulary that was equivalent to the 20-month-old low producers. Like the languagematched 20-month-olds, the 28- to 30-month-old late talkers did not show ERP differences to open and closed class words. 1~ These data lend further support to the hypothesis that the functional organization of specific subsystems within language is linked to levels of language abilities. ERP PREDICTORS OF LANGUAGE ABILITIES

In our previous studies of 13- to 42-month-old normal children and children with focal brain injury, the N200 and N350 have been shown to index language comprehension.l~ Moreover, these studies have shown that the responsiveness of different ERP components to different types of words is linked to vocabulary size independent of chronological age. In this section, we explore the hypothesized link between auditory sensory processing and language skills (see above) from children who were the same chronological age but who

A ~

Le~t

Right Temporal

Temporal v

r.::::'~t .:~ ~.::~,, ,.

B

~:. I.~',~

.........

C

C~mmd 5-0/~V[

,

o

I

,

4o0(

I

I

,

p

0

I

~

L

1600

---

open c:m

Worda wor

Fig 4. ERPs to open versus closed class words over left and right temporal regions in three groups of infants: 36 to 42 months N -- 17, (A); 28 to 30 months N = 14, (B); and 20 months N = 17, (C).

ELECTROPHYSIOLOGICAL STUDIES OF LANGUAGE

varied in language abilities. Here, the focus is on the organization of the P100 response, which is thought to be an index of early sensory processing. If acoustic deficits underlie, or in part contribute to some language deficits as predicted by behavioral research 26 and by electrophysiological evidence with 9-year-old language-impaired children, 8,26 then abnormalities in auditory sensory processing may also be apparent in some of most children who are delayed in the initial stages of language acquisition. In this study, children at three age groups--13 to 17 months, 20 months, and 28 to 30 months--were each divided into two categories, high and low producers, based on the size of their productive vocabulary relative to the norm for their age group57 Children who scored below the 30th percentile for their age on the MacArthur CDI were categorized as low producers, whereas children who scored above the 50th percentile were categorized as high producers. At 28 to 30 months, most of these children were classified as "late talkers," that is, they scored below the 10th percentile for their age, and were recruited through the PICND at UCSD as noted earlier. Figure 5 shows that for the high producers at each age group, the amplitude of the P100 response to speech was larger over the left than the fight hemisphere. In contrast, for the low producers, including the late talkers, the amplitude of the P100 was symmetrical from over the two hemispheres. High Producers

Low Producers

10

9 7

0~ 0

Fig 5. PIO0 peak amplitudes for high and low producers at each of three age groups: 13 to 17 months, 20 months, and 28 to 30 months, solid bar = left hemisphere; striped bar = right hemisphere.

133

Unlike the results from our previous studies (above) in which the distribution of the N200 and N350 are linked to specific language milestones, the P100 asymmetries to words appear to be linked to the child's percentile ranking, for example, how well the child is doing relative to other children of the same age. A comparison of the 13-month-old high producers and the 20-month-old low producers further illustrates this point. These two groups were matched on productive vocabulary size. However, only the 13-month-old high producers showed the asymmetry. We have also recorded ERPs to pure tones in the 28- to 30-month and 36- to 42-month old age groups for controls and late talkers. For the high producers at each age, the P100 to tones is larger over the right than the left hemisphere. In contrast, for the late talkers, the P100 response to tones was symmetrical. Thus, to the extent that P100 is indexing activation within early stages of auditory processing, and perhaps material specific priming within those systems, these aberrant patterns of activation within language delayed infants may be useful in identifying infants at risk for further delay. Indeed, preliminary data suggest that the presence or absence of the P100 asymmetry may be predictive of later language performance. We have follow-up language data from the Reynell test of language comprehension on nine of the late talkers who participated in this study. 27At 28 to 30 months old, three of the late talkers showed a left greater than right asymmetry in the P100 to words. Follow-up data for these children showed that by 42 months these children had comprehension and production scores within the normal range on the Reynell. In contrast, five of the other six late talkers still showed significant delays on the Reynell for comprehension, production, or both. Much further research is needed to explore the functional significance of ERP responses to speech in this age range and to access their use in clarifying and predicting language delays. In summary, it is clear the ERP is a powerful tool in fractionating language into its functional subsystems and that it can be used to index changes in language-relevant neural systems following different types of early sensory and language experience. There is a dearth of information on the neurobiology of specific language impairment, but the avail-

134

MILLS AND NEVILLE

able e v i d e n c e is c o n s i s t e n t w i t h the w o r k i n g h y p o t h -

ACKNOWLEDGMENTS

esis that l a n g u a g e i m p a i r m e n t is a final c o m m o n

We are grateful to our numerous colleagues who have participated in the several studies described here, to Linda Heidenreich for help with manuscript preparation, to the NIH for grant support, and to the parents and children who helped make these studies possible.

p a t h a r i s i n g f r o m m u l t i p l e different causes. F u t u r e r e s e a r c h will b e c e n t r a l i n i n d e x i n g a n d c l a r i f y i n g t h e s e s e v e r a l different etiologies.

REFERENCES

1. Bates E, Thai D, Trauner D, et al: From first words to grammar in children with focal brain injury. Developmental Neuropsychology (Special issue on origins of communication disorders). (in press) 2. Neville HJ, Schmidt A, Kutas M: Altered visual evoked potentials in congenitally deaf adults. Brain Res 266:127-132, 1983 3. Neville HJ, Mills DL, Bellugi U: Effects of altered auditory sensitivity and age of language acquisition on the development of language-relevant neural systems: Preliminary studies of Williams syndrome, in Broman S, Grafman J (eds): Atypical Cognitive Deficits in Developmental Disorders: Implications for Brain Function. Hillsdale, NJ, Lawrence Erlbaum Associates, 1994, pp 67-83 4. Knight R, Hillyard SA, Woods DL, et ah The effects of frontal and temporal-parietal lesions on the auditory evoked response in man. Electroencephalography & Clinical Neurophysiology 50:112-124, 1980 5. Neville HJ, Mills DL, Lawson DS: Fractionating language: Different neural subsystems with different sensitive periods. Cerebral Cortex 2:244-258, 1992 6. Weber-Fox CM, Neville HJ: Maturational constraints on functional specializations for language processing: ERP and behavioral evidence in bilingual speakers. J Cogn Neurosci 8:231-256, 1996 7. Kutas M, Hillyard SA: Brain potentials during reading reflect word expectancy and semantic association. Nature (Loud) 307:161-163, 1984 8. Neville HJ, Coffey SA, Holcomb PJ, et ah The neurobiology of sensory and language processing in language impaired children. J Cogn Neurosci 5:235-253, 1993 9. Barrett M: Early lexical development, in Fletcher P, McWhinney B (eds): Handbook of Child Language. Oxford, Basil Blackwell, 1995, pp 362-392 10. Mills DL, Coffey-Corina SA, Neville HJ: Language acquisition and cerebral specialization in 20-mouth-old infants. J Cogn Neurosci 5:317-334, 1993 11. Mills DL, Coffey-Corina SA, Neville HJ: Language comprehension and cerebral specialization from 13 to 20 months, in Thai D, Reilly J (eds): Special Issue on Origins of Language Disorders. Developmental Neuropsychology (in press) 12. Hillyard SA, Picton TW: Electrophysiology of the Brain, in Plumb F (ed): Handbook of Psychology: The Nervous System. Bethesda, MD, 1987, pp 519-584 13. Holcomb PJ, Coffey SA, Neville HJ: Visual and auditory sentence processing: A developmental analysis using eventrelated brain potentials. Developmental Neuropsychology 8:203241, 1992 14. Wood CC, Wolpaw JR: Scalp distribution of human auditory evoked potentials. II. Evidence for overlapping sources and involvement of auditory cortex. Electroencephalography and Clinical Neurophysiology 54:25-38, 1982 15. Molfese D, Molfese VJ: Short-term and long-term developmental outcomes: The use of behavioral and electrophysiologi-

cal measures in early infancy as predictors, in Dawson G, Fischer KW (eds): Human behavior and the developing brain. New York, NY, Guilford Press, 1994, pp 493-517 16. Mills DL, Coffey SA, Neville H J: Variability in cerebra/ organization in infancy during primary language acquisition, in Dawson G, Fischer K (eds): Human Behavior and the Developing Brain. New York, NY, Guilford Press, 1994, pp 427-455 17. Kurtzberg D, Stone C, Vaughn H: Cortical responses to speech sounds in the infant, in Cracco RQ, Bodis Wollner I (eds): Evoked Potentials. New York, NY, Liss, 1986, pp 513-520 18. Courchesne D, Ganz L, Norcia A: Event-related potentials to human faces in infants. Child Dev 52:804-811, 1981 19. Karrer R, Ackles P: Visual event-related potentials of infants during a modified oddball procedure: Current trends in event-related potential research. EEG Supplement 40:603-608, 1987 20. Kurtzberg D: Late auditory evoked potentials and speech sound discrimination by infants, in Karrer R (chair): Eventrelated potentials of the brain and perceptual/cognitive processing of infants. Symposium presented at the meeting of the Society for Research in Child Development. Toronto, Canada, 1985 21. Kurtzberg G, Vaughn H: Electrophysiologic assessment of auditory and visual function in the newborn. Clin Perinatol 12:277-299, 1985 22. Neville HJ, Corina D, Bavelier D, et at: Biological constraints and effects of experience on cortical organization for language: An fMRI study of sentence processing in English and American Sign Language (ASL) by deaf and hearing subjects. Society for Neuroscience Abstracts 20, 1994 23. Neville HJ: Developmental specificity in neurocognitive development in humans, in Gazzaniga MS (ed): The cognitive neurosciences. Cambridge, MIT Press, 1995, pp 219-231 24. Mills DL, Coffey SA, Dilulio L, et al: Development of cerebral specialization for different lexical items in normal infants and infants with focal brain lesions (Tech. Rep #CND9507). LaJolla CA, University of California, San Diego, Center for Research in Language, Project in Cognitive & Neural Development, 1995 25. Bates E, Bretherton t, Snyder L: From first words to grammar: Individual differences and dissociable mechanisms. Cambridge, MA, Cambridge University Press, 1988 26. Tallal P: An experimental investigation of the role of auditory temporal processing in normal and disordered language development, in Caramazza A, Zurif D (eds): Language Acquisition and Language Breakdown: Parallels and Divergences. Baltimore, MD, The John Hopkins University Press, 1978, pp 25-61 27. Mills DL, Thai D, Dilulio L, et ah Auditory sensory processing and language abilities in late talkers: An ERP study (Tech. Rep. #CND-9508). LaJolla, CA, University of California, San Diego, Center for Research in Language, Project in Cognitive & Neural Development, 1995