Leppänen et al DN 22 1 2002

Full text

DEVELOPMENTAL NEUROPSYCHOLOGY, 22(1), 407–422
Copyright © 2002, Lawrence Erlbaum Associates, Inc.
Brain Responses to Changes in Speech
Sound Durations Differ Between Infants
With and Without Familial Risk
for Dyslexia
Paavo H. T. Leppänen
Department of Psychology
University of Jyväskylä
Ulla Richardson
Institute of Cognitive Neuroscience
University College London
Department of Finnish
University of Jyväskylä
Elina Pihko
Department of Psychology
University of Jyväskylä
BioMag Laboratory, Engineering Centre
Helsinki University Central Hospital
Kenneth M. Eklund, Tomi K. Guttorm,
Mikko Aro, and Heikki Lyytinen
Department of Psychology
University of Jyväskylä
A specific learning disability, developmental dyslexia, is a language-based disor-
der that is shown to be strongly familial. Therefore, infants born to families with a
history of the disorder are at an elevated risk for the disorder. However, little is
Requests for reprints should be sent to Paavo H. T. Leppänen, Department of Psychology, University
of Jyväskylä, P.O. Box 35, FIN-40351 Jyväskylä, Finland. E-mail: paavo.leppanen@psyka.jyu.fi

known of the potential early markers of dyslexia. Here we report differences be-
tween 6-month-old infants with and without high risk of familial dyslexia in brain
electrical activation generated by changes in the temporal structure of speech
sounds, a critical cueing feature in speech. We measured event-related brain re-
sponses to consonant duration changes embedded in ata pseudowords applying an
oddball paradigm, in which pseudoword tokens with varying /t/ duration were pre-
sented as frequent standard (80%) or as rare deviant stimuli (each 10%) with an
interval of 610 msec between the stimuli. The infants at risk differ from control
infants in both their initial responsiveness to sounds per se and in their change-
detection responses dependent on the stimulus context. These results show that in-
fants at risk due to a familial background of reading problems process auditory
temporal cues of speech sounds differently from infants without such a risk even
before they learn to speak.
It is widely accepted that developmental dyslexia is often genetically transmitted
(Gilger, Pennington, & DeFries, 1991; Pennington, 1995). As a consequence, in-
fants born to families with affected parent(s) are at increased risk of the disorder.
Therefore, studying these infants makes easier the identification of those who will
eventually become dyslexic. Also, in this way, finding early markers of the disor-
der becomes possible. In this study, we compared the brain electrical activation of
6-month-old infants from affected families with that of infants without dyslexia in
their families.
Dyslexia is usually related to insufficient phonological processing abilities and
is characterized by difficulties in single-word decoding (Bradley & Bryant, 1978,
1983; Lyon, 1995; Wagner & Torgesen, 1987). Word decoding and phonological
processing problems may be assumed, in turn, to result at least partly from more
basic lower level processing or speech perception deficits (Manis, 1997; Reed,
1989; Tallal, 1980). Individuals with dyslexia are reported to differ, for example, in
their electrocortical activation to auditory stimuli (Brunswick & Rippon, 1994;
Kujala, Myllyviita, Tervaniemi, Alho, Kallio, & Näätänen, 2000; Leppänen &
Lyytinen, 1997; Wood, Flowers, Buchsbaum, & Tallal, 1991) and to have problems
in processing temporal sound features, such as briefly or rapidly presented sequen-
tial, and amplitude- or frequency-modulated auditory information (Farmer &
Klein, 1995; Hari & Kiesilä, 1996; Helenius, Uutela, & Hari, 1999; McAnally &
Stein, 1997; Tallal, 1980; Tallal, Miller, & Fitch, 1993; Witton et al., 1998). Recent
research suggests that they may also have a deficit in fast motor–sensory and visual
processing (Lovegrove, 1993; Stein & Fowler, 1985; Stein & Walsh, 1997).
Adult dyslexics with a familial background of the disorder have been found to
deviate in the perception of a silence gap duration that represents stop consonant
length within a syllable; they require a longer silence duration than controls to
identify a stimulus as /sta/ in a /sta/–/sa/ continuum (Steffens, Eilers, Gross-
Glenn, & Jallad, 1992). This feature of variations in speech sound duration is a
408 LEPPÄNEN ET AL.

highly distinctive phenomenon, for example, in the Finnish language. The ability
to discriminate between short and long consonants and vowels forms an essential
basis for a child to learn the language, because these duration variations are criti-
cal in cueing opposites and semantic differences. For example, perceived changes
in durational patterns, by themselves alone, can contribute to the identification of
a word; for instance, the word /mato/ (worm), with a short /t/-sound duration, has
a different meaning than the word /matto/ (carpet), which has a long /t/-sound du-
ration. Finnish dyslexic readers have been shown to have problems in perceiving
such durational differences (Richardson, 1998; Richardson, Leppänen, Leiwo,
& Lyytinen, 2002), and they also make disproportionally more mistakes in differ-
entiating vowel or consonant durations when reading pseudowords as compared to
normal readers (Lyytinen, Leinonen, Nikula, Aro, & Leiwo, 1995). Using this
speech feature applied to pseudowords, Richardson (Richardson, 1998; Richardson
et al., 2002) found that 6-month-old Finnish infants at risk for familial dyslexia
require a longer consonant duration than control infants to respond to a long con-
sonant in a categorical fashion in a behavioral study employing a conditioned head-
turn paradigm.
In this study, we recorded brain event-related potentials (ERPs) from
6-month-old infants to determine whether brain responses elicited by changes in
speech sound durations, requiring accurate durational discrimination, are related
to an at-risk status of dyslexia. Half of the infants were at risk because of their
family background of dyslexia, and the other half were controls without a
dyslexic background. We presented sequences consisting of ata pseudowords
with variable /t/ durations in two separate conditions. In the first, a short ata was
presented as a frequent standard stimulus and was occasionally replaced by one of
two deviant pseudowords, differing only in terms of /t/ duration (either an ata with
an intermediate /t/ duration or a long atta with a long /t/, marked with a double t).
In the second condition, the presentation probabilities of the short ata and long
atta were reversed.
METHODS
Participants
Seventy-six healthy infants, with a mean age of 6 months and 5 days and without
hearing deficit or neurological disorder, participated in this study. All the infants
were from families which were recruited, according to institutional informed con-
sent procedures, for the Jyväskylä Longitudinal Study of Dyslexia. Of these, 37
(20 boys) were from families with a familial background of dyslexia and belonged
to the at-risk group. Thirty-nine of the infants (21 boys) were from matched con-
trol families without any signs of dyslexia, and they belonged to the control group
BRAIN RESPONSES AND RISK FOR DYSLEXIA 409

(for details on the participant characteristics and selection criteria, see Leinonen,
Muller, Leppänen, Aro, Ahonen, & Lyytinen, 2001; Lyytinen et al., 1995). The
parents in both groups reported no hearing problems, nor any sensory or neuro-
logical abnormalities.
The inclusion criteria for the at-risk group were either parent’s report of his or her
own reading disorder, a comparable report concerning at least one close relative, and
multiple diagnostic test results indicative of dyslexia. To be diagnosed with
dyslexia, the parent had to have a score of at least 1 SD below the norm in accuracy
or speed of oral text reading, or in accuracy of written spelling, and also in at least
two separate single-word measures (either accuracy or response latency of word
recognition, pseudoword decoding or lexical decision). Their IQ had also to be 85 or
above (assessed with the Raven B, C, and D matrices; Raven, Court, & Raven,
1992). Some of the parents who had several relatives with a reading disability, de-
spite self-reported school-age and present reading problems, did not score signifi-
cantly below the norm on all the required diagnostic measures. However, because of
their strong family history of reading difficulties, they were included as compen-
sated dyslexics (it should be noted that inclusion of the families with compensated
dyslexics was a “conservative” decision, in that their infants could only be expected
to differ to a lesser degree than other at-risk infants from control infants). The moth-
ers’ educational status, as determined on the basis of the length or level of education,
or their IQ, did not differ between the groups.
Stimuli and Procedure
Three naturally produced pseudowords were used as stimuli. They comprised
the short ata, which had a voiceless stop with a silent period of 95 msec in the mid-
dle of the sound (with a total duration of 300 msec; the duration of the first part of
the stimulus, the initial glottal stop and /a/ vowel together, was 72 msec, and that of
the second part, including the explosion of /t/ and the final /a/ vowel, was 133 msec),
an intermediate ata, and the long atta. The latter two stimuli were produced by
lengthening the silent gap to 195 and 255 msec, respectively. All other acoustical as-
pects, such as fundamental frequency and intensity, were held constant (for details;
see Pihko, Leppäsaari, Leppänen, Richardson, & Lyytinen, 1997; Richardson, 1998;
Richardson et al., 2002). ERP data reported here are for the short ata and long atta.
Stimuli were presented via a loudspeaker in short sequences with an intensity of
75 dB sound pressure level in two different conditions: In the long atta deviant con-
dition (data from 25 at-risk and 27 control infants), the short ata was presented as
the frequent “standard” stimulus, with an 80% probability of occurring. The inter-
mediate and the long atta stimuli were presented as rare deviant stimuli on 10% of
trials for each. The fixed offset-to-onset interstimulus interval (ISI) was 610 msec.
The stimulus sequence in the short ata deviant condition (data from 12 infants from
410 LEPPÄNEN ET AL.

each group) was similar to the first one, except that the long atta occurred on 80%
of trials (as the standard) and short ata on 10% (as the other deviant). In this con-
dition, two ISIs of 450 msec and 610 msec were used, and as no statistical differ-
ence between them was found in the reported ERP measure in the control group,
the data were pooled across the two ISI conditions.
ERP Recording
Infants were seated on their parent’s lap in an electroencephalogram (EEG) labora-
tory room during the ERP recordings. ERPs were recorded from eight scalp sites
(F3, F4, T3, T4, C3, C4, P3, P4) according to the international 10 to 20 electrode
system with Ag/AgCl electrodes attached in an EEG-cap and referred to the ipsilat-
eral mastoid electrodes. Electrooculograms (EOGs), referenced to the left mastoid,
were recorded, with one electrode lateral to and slightly above the left eye, and an-
other lateral to and below the right eye. The online filtered (passband 0.5–35 Hz,
sampling rate 200 Hz) EEG epochs of –50 to 840 msec (with a 50-msec prestimu-
lus baseline) were averaged offline separately for each stimulus type (for the stan-
dard stimuli only the last three epochs before the deviants were included). Epochs
with artifacts, deflections exceeding ±200 µV at EOG (excessive eye movements)
and EEG channels (muscle activity or other extra cerebral artefacts) were excluded
from the averaging. The mean number of acceptable EEG epochs in the long atta de-
viant condition was 105 (range = 73–137) for the deviant, 383 (range = 270–512)
for the standard stimulus, and in the short ata deviant condition 102 (range =
77–136) for the deviant and 377 (range = 258–504) for the standard stimulus. In-
fants with less than 70 acceptable EEG epochs were excluded from this report.
Data Analysis
In the long atta deviant condition, individual mean amplitudes were calculated over
the 30-msec period centered at each major peak in the waveform (P190, N340,
P470, N600, P840 deflections named according to polarity and approximate peak
latencies; see Figure 1b). Group differences in the mean peak amplitude measures
were statistically analyzed using a Deflections ×Scalp Sites ×Group multiple
analysis of variance (MANOVA) for repeated measures (details in Results section).
Further comparisons were carried out for different deflections separately for each
hemisphere using Scalp Sites ×Group Repeated Measures analyses of variance
(ANOVAs). To investigate how the brain responds differently to the stimuli with
the short and long silence gaps, difference wave ERPs, typically used with oddball
paradigms, were calculated for each participant by subtracting the response to short
ata from that to the long atta. The difference waves were then compared between
BRAIN RESPONSES AND RISK FOR DYSLEXIA 411

the groups in a data-point by data-point analysis with two-tailed ttests. Five con-
secutive data points were required to differ between the groups at an alpha level of
0.01 to show a group difference.
In the short ata deviant condition, individual mean amplitudes were calculated
over the latency period of 475 to 675 msec, centered at the second negative peak in
the grand averaged ERP to the short ata in the control group. The shorter 30-msec
period was not appropriate because a clear peak was missing in the at-risk infants.
Group differences in this measure were analyzed by a Scalp Sites ×Group
Repeated Measures MANOVA.
RESULTS
Long atta Deviant Condition
The ERP waveforms for the long atta deviant condition are shown in Figure 1.
Both the standard short ata and the deviant long atta elicited an initial small neg-
ative deflection (N80), followed by a widespread positive-negative-positive wave-
form pattern (P190, N340, P470; Figure 1b). However, the response to the long
atta was markedly different from that to the short ata in both groups. For the long
atta, the major negative deflection at about 330 to 340 msec (N340) was shorter in
duration, and it was followed by another negative peak at 600 msec (N600). These
ERP differences indicate that in both groups of infants, the brain differentiated the
two stimuli varying in stop consonant duration.
The standard and deviant stimuli were exactly the same up to 167 msec. After
this point, the difference between the two stimuli, the prolonged silence of deviant
stimuli, commenced. Thus, the early N80 and the following positive P190 deflec-
tion, with an onset before this time point, could be regarded as reflecting the brain’s
response to the initial part of the pseudoword. Correspondingly, the brain’s activa-
tion caused by the change in the temporal structure of the stimulus is reflected in
the waveform pattern starting from the N340 deflection (Figure 1b).
A 4 Deflections (N340, P470, N600, P840) ×8 scalp sites ×Group MANOVA
for repeated measures of the response to the deviant long atta showed a group
main effect, F(1, 50) = 5.04, p< .03. This indicated that the general change-
driven brain response pattern was different between the groups (Figure 1b).
The difference wave, used to investigate how the brain activation patterns dif-
fer for short and long silence gaps, had a significantly smaller amplitude in the at-
risk group at the left hemisphere (at C3) at the time windows of 590 to 625 msec
and 715 to 755 msec as compared to the control group (two-tailed ts > 2.7,
ps < .01; see Figure 1). This revealed that the responses between the standard and
deviant stimuli differed less in the at-risk infants. No such group differences were
found over the right hemisphere (ps > .05).
412 LEPPÄNEN ET AL.

That the smaller negative difference wave in the at-risk infants is related to
differential responding to deviancy, rather than to the response to the repetitive
standard stimulus, was confirmed in further analyses for the deviant separately
for each deflection and each hemisphere (4 Sites ×Group Repeated Measures
ANOVAs). These revealed that in the left hemisphere, across the F3, C3, T3, and
P3 electrode sites, the amplitude of the deflection at 470 msec was more posi-
tive in the at-risk than in the control group, F(1, 50) = 5.19, p< .03. Conversely,
the amplitude of the negative deflection at about 600 msec was larger in the
control group, F(1, 50) = 7.15, p< .02. No group effects were found for any
deflection in the right hemisphere. That group differences in this negative de-
flection are not explained by a general shift in the earlier more positive P190
BRAIN RESPONSES AND RISK FOR DYSLEXIA 413
FIGURE 1 (A) Difference waves (event-related potential [ERPs] to the standard short ata sub-
tracted from ERPs to the deviant long atta) averaged across 27 control (thick line) and 25 at-risk
(thin line) infants. The difference waves indicate that the standard- and deviant-stimulus re-
sponse patterns clearly differ from each other in both groups. For the late response at 600 msec,
these patterns were less differentiated in the at-risk group at the left hemisphere, most markedly
at the central scalp area (at the C3 electrode location). (B) ERPs, averaged across the same con-
trol and at-risk infants, for the standard ata (thin line) and the deviant atta (thick line) at C3. A
similar waveform pattern was observed at all electrode sites. However, amplitudes were much
smaller at parietal and occipital sites. The stimuli are marked with the two boxes, with an empty
space in between, above (short ata) and below the baseline (long atta). Peaks are labeled ac-
cording to their polarity and approximate peak latency—negativity up.

peak in the at-risk group (cf. Figure 1b) was shown in a separate ANOVA analy-
sis using P190 amplitude for the deviant stimulus as a covariate. The group dif-
ference for N600 at the left hemispheric C3 site remained despite P190 as a
covariate, F(1, 49) = 5.77, p< .03).
Discussion for the long atta deviant condition. These results show that
the groups differ in their response generated by a change of temporal properties
in the stimulus sequence; that is, by the prolonged silence gap generating a per-
ception of consonant duration change. This is reflected in the larger positivity at
about 470 msec in the at-risk group, and in the larger negativity at 600 msec in
the control group, both in the difference wave and in the response to the deviant
atta. These group differences are not explained by a more negative overall pattern
of the waveform in the control group (see Figure 1). The covariance analysis
showed that the more negative response at 600 msec in the at-risk group to the
prolonged tduration is not due to the earlier larger obligatory positive response.
Further, as evident in Figure 1b, the responses to the deviant and standard stim-
uli initially have very similar patterns in each group. This suggests that the
observed group effects in the difference wave are not caused by any general am-
plitude level difference, but rather reflect differential processing of consonant pro-
longation. The larger positive deflection at about 450 to 475 msec in the at-risk
group partly reflects differences in responsiveness to stimulus onset or stimuli
per se because this positive deflection seems to be generated by the commence-
ment of the second part of the long atta stimulus, and seems to correspond to the
earlier P190 positive deflection for the onset of the first stimulus part.
Short ata Deviant Condition
In the short ata deviant condition, the probabilities of the short ata and long atta
were reversed (now 10% and 80%, respectively) to further test whether the stim-
uli as such or their difference determined ERP and related group differences. If
the stimulus context had no effect, then the brain’s response to the short ata as the
deviant stimulus should be similar to that elicited by the same ata as the standard
stimulus (in the long atta deviant condition). However, this was not the case. The
short ata as the deviant stimulus, elicited an additional negative peak at 550 msec
(at 380 msec from the deviant and standard stimulus difference) which followed
the N340 deflection at the frontal, central, and parietal scalp areas in the control
infants (Figure 2a). Remarkably, it was completely absent in the at-risk group at
both hemispheres (Figure 2a, right panel; 2b). A 6 Scalp Sites (F, C, and P sites at
both hemispheres) ×Group Repeated Measures MANOVA showed a group main
effect for the mean response amplitude at the latency range of this late negative
deflection, F(1, 22) = 7.66, p< .02 (Figure 2c). It should be noted that the
414 LEPPÄNEN ET AL.

415
FIGURE 2. Event-related potentials ([ERPs] averaged across infants) at C3 (stimuli marked
with boxes on the calibration line). (A) ERPs to the standard short ata in the long atta deviant
condition (thin line, p= 80%) and to the same ata as the deviant stimulus in the short ata de-
viant condition (thick line, 10%). Left panel; the control group (n= 27 and 12 for long atta and
short ata deviant conditions, respectively). Right panel; the at-risk group (n= 25 and 12 for
long atta and short ata deviant conditions, respectively). In the control group, the deviant ata
elicits an additional negative deflection at about 550 msec, reflecting change detection. (B) The
ERPs to the deviant ata in the short ata deviant condition in the control (thick line, n= 12) and
at-risk (thin line, n= 12) groups. The second negative deflection is missing from the at-risk
group, indicating the lack of change detection response. (C) The mean amplitude over the 475
to 625 msec range for the short ata as the standard and as the deviant stimulus in the control
(thick line) and at-risk groups (thin line). Bars indicate SEM.

bilateral appearance of this group effect is not in contradiction with the left hemi-
spheric group effect in the long atta deviant condition, because the additional
response in this condition was absent altogether in the at-risk infants.
Discussion for the short ata deviant condition. The waveform of the ad-
ditional negative deflection resembles a typical adult change-detection response,
the mismatch negativity ([MMN] Näätänen, 1992; Näätänen & Alho, 1997;
Näätänen, Gaillard, & Mäntysalo, 1978), and the latency of the negativity is in
line with the reported infant MMN (Cheour et al., 1998). The additional deflection
could thus be interpreted as an MMN. This suggests deficient passive discrimina-
tion of temporal sound features (duration) in children with risk for dyslexia.
DISCUSSION
Infants at risk for developmental dyslexia differ in passive discriminative pro-
cessing of sound changes, as well as in responsiveness to sounds per se. The
brain responses were measured in conditions typically eliciting, both in children
and adults, the MMN component that is thought to index a preattentive change-
detection process in the auditory cortex (Giard, Perrin, Pernier, & Bouchet, 1990;
Näätänen & Alho, 1997; Näätänen, 1992). MMN is elicited by incoming sounds
that violate some previously invariant characteristic of an auditory sequence en-
coded in the short-duration sensory memory (Escera, Alho, Schröger, & Winkler,
2000). In this study, the additional MMN-like negative deflection to the short ata
(when presented as the deviant stimulus) also suggests that the group differences
in the late negativity at 600 msec for the deviant long atta could be at least partly
accounted for by differences in MMN-like response. Both of these deflections
occur at about the same latency from the point at which the deviant differs from
the standard stimulus (i.e., at 380 msec and 430 msec for the short ata and long
atta, respectively). Thus, at-risk infants seem to differ in their change-detection
responses, which depend on the formation of memory traces for repetitive stim-
uli (Winkler & Czigler, 1998). Such short duration memory traces can be thought
to form a basis for long-term speech sound representations. In this view, impre-
cisely formed memory traces would lead to “fuzzy”, or inexact, representations.
In addition to differences in the passive change-detection response, the initial re-
sponsiveness to sounds per se seems to differ in infants at risk for dyslexia, as
suggested by their more positive responses to sound onsets preceding the MMN-
like response (as reflected in P190 and P470). In line with this, we have found
more positive responses in at-risk infants at birth to vowel duration change
(Leppänen, Pihko, Eklund, & Lyytinen, 1999) and to consonant–vowel (CV)
syllables (varying in their CV transitions; Guttorm, Leppänen, Richardson, &
Lyytinen, 2001; Guttorm, Leppänen, Tolvanen, & Lyytinen, 2002).
416 LEPPÄNEN ET AL.

Less precise sound representations of temporal features in at-risk
infants. Taken together, these group differences suggest that at-risk children
may be less efficient at laying down representations of the temporal envelope of
speech signals resulting in poorer prototypes of speech sounds (Kuhl, 1991). This
weakness might help to explain the poor categorical perception found later in de-
velopmental dyslexia (Steffens et al., 1992; Werker & Tees, 1987). Deficits in
forming speech sound representations can have cascading effects, leading to a vi-
cious circle. It is known that long-term memory traces can, in turn, facilitate
preattentive change detection processing in a top-down fashion (Näätänen, 1992;
Näätänen, 1999; Winkler et al., 1999). For example, Näätänen et al., (1997) have
shown that MMN is smaller to a deviation in non-native speech sounds as com-
pared to that in native speech sounds even when the deviation magnitude is simi-
lar in both cases. The idea of poorer speech sound representations related to
dyslexia gets support from findings showing deviant MMN in learning disabled
children (Kraus, McGee, Carrell, Zecker, Nicol, & Koch, 1996) and adults with
dyslexia (Kujala et al., 2000). It should be noted that our study relates to the
underlying mechanism necessary for speech representations; that is, change de-
tection, rather than to the long-term representations themselves. Therefore, it re-
mains the task of future studies to investigate whether the at-risk infants already
differ from controls in their prototypes for different speech sound durations at this
age. This could be best done by studying responses to speech sounds presented
with equal probability and with long interstimulus intervals (to avoid the interfer-
ence of the fast-stimulus rate effect usually present in MMN paradigms), as well as
by using cross-linguistic comparisons to determine the age at which such repre-
sentations should be formed. Cheour et al. (1998) have shown, for example, that
Finnish childrens’ brain responses to Estonian vowels are different from that of
Estonian children by the age of 1 year.
The lack of an MMN-like response in the at-risk group for the deviant short
ata could be at least partly explained by differences in temporal integration
(Tervaniemi, Saarinen, Paavilainen, Danilova, & Näätänen, 1994; Yabe,
Tervaniemi, Reinikainen, & Näätänen, 1997). At-risk infants may have slower
temporal integration (or a longer temporal integration window), which results in
a more “merged,” or integrated, sensory event, as shown by their negative de-
flection having only one peak. These kinds of differences in the temporal inte-
gration window could also be connected to reported deficits in the processing of
rapidly changing information in dyslexics (Farmer & Klein, 1995; Hari &
Kiesilä, 1996; Tallal, 1980; Tallal et al., 1993).
Auditory/language areas of the brain may be differently recruited for
speech sound processing in at-risk infants. A general notion is that there is
left hemispheric preponderance for processing of speech (Carr & Posner, 1995;
Neville, Kutas, & Schmidt, 1982). In line with this, when an index of hemispheric
BRAIN RESPONSES AND RISK FOR DYSLEXIA 417

preponderance of ERP (N600 amplitude to the deviant long atta in the right hemi-
sphere subtracted from that in the left) was compared to performance in the behav-
ioral head-turn task partly in the same infants (16 at-risk and 15 control infants
together; for details, see Richardson, 1998; Richardson et al., 2002), a negative as-
sociation (Pearson r= –.40, two-tailed p< .03) was found between behavioral per-
formance and the hemispheric index at the central scalp sites. This indicated that
the more negative the amplitude at the left central hemisphere (C3) in relation to
the right hemispheric (C4) amplitude, the better the behavioral performance (more
head turns for the stimulus that should have been perceived as long in a short
/ata/–long /atta/ continuum with eight variations, and on which the performance
also differed between the at-risk and control groups; Richardson et al., 2002). Thus,
the left-hemispheric preponderance of N600 deflection for the deviant long atta
seems to be associated with better categorization skills for consonant duration.
It is interesting that when the relationship of this late negative deflection to be-
havioral performance was estimated separately at each hemisphere and in each
group, a differential pattern emerged in the groups. This correlation at the left
hemisphere (at C3) differed significantly between the groups (p< .05). In the at-
risk group, the smaller negative amplitudes bilaterally were associated with better
categorical performance (at C3, r= .52, two-tailed p< .05, and at C4, r= .66,
two-tailed p< .006), whereas in the control group there was a tendency for greater
negative amplitudes only at the left hemisphere to be related to better performance
(at C3, r= –.26, p> .05). That the lower negative deflection at the left hemisphere
is associated with better categorical perception in the at-risk infants suggests that
their auditory and language areas of the brain are differently recruited for speech
sound processing. The more pronounced correlations between the ERPs and be-
havioral responses for at-risk infants could be expected because within this fam-
ily-history group, a subgroup (i.e., those who will later be affected by reading
problems) could be expected to deviate from others and create more interindivid-
ual variance. These correlation findings are also compatible with previous brain-
imaging findings of left hemispheric functioning differences in individuals with
dyslexia (Hiscock & Kinsbourne, 1995; Hynd, Semrud-Clikeman, & Lyytinen,
1991; Salmelin, Service, Kiesilä, Uutela, & Salonen, 1996).
CONCLUSIONS
This study shows that by 6 months of age, infants with high familial risk for devel-
opmental dyslexia process differences in consonant duration in a different way from
those not at risk. These findings are particularly convincing, because two different
paradigms, behavioral (Richardson, 1998; Richardson et al., 2002) and psychophys-
iological, using ERPs as an index of brain electrical activation, and applied partly to
the same individuals, yield converging evidence of similar speech sound processing
418 LEPPÄNEN ET AL.

differences between at-risk and control infants. These data suggest that those who go
on to become dyslexic may have a deficit related to timing and perception of tempo-
ral cues in speech, seen already at the preattentive level of auditory processing at a
very young age, which may explain the development of their difficulties with phono-
logical processing (McBride-Chang, 1995). Such problems may result from both
genetic and environmental factors, such as early speech input, as well as their com-
plicated interaction. Our previous findings for differences in electrocortical responses
to vowel duration change (Leppänen et al., 1999) and to CV syllables (varying in
their CV-transitions; Guttorm et al., 2001; Guttorm et al., 2002) between infants with
a risk for dyslexia and control infants already at birth suggests the presence of a ge-
netic component in line with other studies (Gilger et al., 1991; Pennington, 1995).
Taken together, our results provide evidence of the existence of an early precursor for
developmental dyslexia and that developmental dyslexia may be predicted at a very
early age. Early identification would facilitate well-directed remedial means even be-
fore language problems are typically diagnosed. It must be noted, though, that the risk
of dyslexia is possibly not more than 50:50, meaning that about half of the at-risk
children will not have any problem with reading in the future.
ACKNOWLEDGMENTS
We thank the families who participated in this study, S. Leinonen, and the mater-
nity clinics of Central Finland for their help in enrolling the families, J. Erskine
and J. Thomas for polishing the language, A. Tolvanen for assistance in statistics,
and R. Näätänen and J. Stein for valuable comments.
This article was prepared as a part of the project (No. 44858) “Human
Development and Its Risk Factors,” financed by the Academy of Finland
(Finnish Centre of Excellence Program, 2000–2002). This work was also sup-
ported by the University of Jyväskylä and the Niilo Mäki Foundation.
REFERENCES
Bradley, L., & Bryant, P. (1978). Difficulties in auditory organization as a possible cause of reading
backwardness. Nature, 271, 746–747.
Bradley, L., & Bryant, P. (1983). Categorizing sounds and learning to read: A causal connection.
Nature, 301, 419–421.
Brunswick, N., & Rippon, G. (1994). Auditory event-related potentials, dichotic listening performance
and handedness as indices of lateralisation in dyslexic and normal readers. International Journal
of Psychophysiology, 18, 265–275.
Carr, T. H., & Posner, M. I. (1995). The impact of learning to read on the functional anatomy of lan-
guage processing. In B.Gelder (Ed.), Speech and reading: A comparative approach (pp. 267–301).
Hove, England: Lawrence Erlbaum Associates, Inc.
BRAIN RESPONSES AND RISK FOR DYSLEXIA 419

Cheour, M., Ceponiene, R., Lehtokoski, A., Luuk, A., Allik, J., Alho, K., et al. (1998). Development
of language-specific phoneme representations in the infant brain. Nature Neuroscience, 1,
351–353.
Escera, C., Alho, K., Schröger, E., & Winkler, I. (2000). Involuntary attention and distractibility as
evaluated with event-related brain potentials. Audiology & Neuro-Otology, 5, 151–166.
Farmer, M. E., & Klein, R. M. (1995). The evidence for a temporal processing deficit linked to
dyslexia: A review. Psychonomic Bulletin & Review, 2, 460–493.
Giard, M. H., Perrin, F., Pernier, J., & Bouchet, P. (1990). Brain generators implicated in the process-
ing of auditory stimulus deviance: A topographic event-related potential study. Psychophysiology,
27, 627–640.
Gilger, J. W., Pennington, B. F., & DeFries, J. C. (1991). Risk for reading disability as a function of
parental history in three family studies. Reading and Writing, 3, 205–217.
Guttorm, T. K., Leppänen, P. H. T., Richardson, U., & Lyytinen, H. (2001). Event-related potentials and
consonant differentiation in newborns with familial risk for dyslexia. Journal of Learning Disabil-
ities, 34, 534-544.
Guttorm, T. K., Leppänen, P. H. T., Tolvanen, A., & Lyytinen, H. (2002). Event-related potentials in
newborns with and without familial risk for dyslexia: Principal component analysis reveals differ-
ences between the groups. Manuscript submitted for publication.
Hari, R., & Kiesilä, P. (1996). Deficit of temporal auditory processing in dyslexic adults. Neuroscience
Letters, 205, 138–140.
Helenius, P., Uutela, K., & Hari, R. (1999). Auditory stream segregation in dyslexic adults. Brain, 122,
907–913.
Hiscock, M., & Kinsbourne, M. (1995). Progress in the measurement of laterility and implications for
dyslexia research. Annals of Dyslexia, 45, 249–268.
Hynd, G. W., Semrud-Clikeman, M., & Lyytinen, H. (1991). Brain imaging in learning disabilities. In
J. E. Obrzut (Ed.), Neuropsychological foundations of learning disabilities: A handbook of issues,
methods, and practice (pp. 475–511). San Diego, CA: Academic.
Kraus, N., McGee, T. J., Carrell, T. D., Zecker, S. G., Nicol, T. G., & Koch, D. B. (1996). Auditory neu-
rophysiologic responses and discrimination deficits in children with learning problems. Science,
273, 971–973.
Kuhl, P. K. (1991). Perception, cognition, and the ontogenetic and phylogenetic emergence of human
speech. In S. E. Brauth, W. S. Hall, & R. J. Dooling (Eds.), Plasticity of development (pp. 73–106).
Cambridge, MA: MIT Press.
Kujala, T., Myllyviita, K., Tervaniemi, M., Alho, K., Kallio, J., & Näätänen, R. (2000). Basic auditory
dysfunction in dyslexia as demonstrated by brain activity measures. Psychophysiology, 37,
262–266.
Leinonen, S., Muller, K., Leppänen, P. H. T., Aro, M., Ahonen, T., & Lyytinen, H. (2001). Hetero-
geneity in adult dyslexic readers: Relating processing skills to the speed and accuracy of oral text
reading. Reading and Writing, 14, 265–296.
Leppänen, P. H. T., & Lyytinen, H. (1997). Auditory event-related potentials in the study of develop-
mental language-related disorders. Audiology & Neuro-Otology, 2, 308–340.
Leppänen, P. H. T., Pihko, E., Eklund, K. M., & Lyytinen, H. (1999). Cortical responses of infants with
and without a genetic risk for dyslexia: II. Group effects. NeuroReport, 10, 969–973.
Lovegrove, W. (1993). Weakness in the transient visual system: A causal factor in dyslexia? Annals of
the New York Academy of Sciences, 682, 57–69.
Lyon, G. R. (1995). Toward a definition of dyslexia. Annals of Dyslexia, 45, 3–27.
Lyytinen, H., Leinonen, S., Nikula, M., Aro, M., & Leiwo, M. (1995). In search of the core features of
dyslexia: Observations concerning dyslexia in the highly orthographically regular Finnish lan-
guage. In V. W. Berninger (Ed.), The varieties of orthographic knowledge II: Relationships to
phonology, reading, and writing (pp. 177–204). Dordrecht, The Netherlands: Kluver.
420 LEPPÄNEN ET AL.

Manis, F. R. (1997). Are speech perception deficits associated ith developmental dyslexia? Journal of
Experimental Child Psychology, 66, 211–235.
McAnally, K. I., & Stein, J. F. (1997). Scalp potentials by amplitude-modulated tones in dyslexia.
Journal of Speech, Language, and Hearing Research, 40, 939–945.
McBride-Chang, C. (1995). Phonological processing, speech perception, and reading disability:
Integrative review. Educational Psychologist, 30, 109–121.
Näätänen, R. (1992). Attention and brain function. Hillsdale, NJ: Lawrence Erlbaum Associates Inc.
Näätänen, R. (1999). Phoneme representatios of the human brain as reflected by event-related poten-
tials. Functional Neuroscience: Evoked Potentials and Magnetic Fields, 49, 170–173.
Näätänen, R., & Alho, K. (1997). Mismatch negativity—The measure for central sound representation
accuracy. Audiology & Neuro-Otology, 2, 341–353.
Näätänen, R., Gaillard, A. W. K., & Mäntysalo, S. (1978). Early selective-attention effect on evoked
potential reinterpreted. Acta Psychologica, 42, 313–329.
Näätänen, R., Lehtokoski, A., Lennes, M., Cheour, M., Huotilainen, M., Iivonen, A., et al. (1997).
Language-specific phoneme representations revealed by electric and magnetic brain responses.
Nature, 385, 432–433.
Neville, H. J., Kutas, M., & Schmidt, A. (1982). Event-related potential studies of cerebral specializa-
tion during reading: I. Studies of normal adults. Brain and Language, 16, 300–315.
Pennington, B. F. (1995). Genetics of learning disabilities. Journal of Child Neurology, 10, 69–77.
Pihko, E., Leppäsaari, T., Leppänen, P., Richardson, U., & Lyytinen, H. (1997). Auditory event-related
potentials (ERP) reflect temporal changes in speech stimuli. NeuroReport, 8, 911–914.
Raven, J. C., Court, J. H., & Raven, J. (1992). Standard progressive matrices. Oxford, England: Oxford
Psychologists Press.
Reed, M. A. (1989). Speech perception and the discrimination of brief auditory cues in reading
disabled children. Journal of Experimental Child Psychology, 48, 270–292.
Richardson, U. (1998). Familial dyslexia and sound duration in the quantity distinctions of Finnish in-
fants and adults. Unpublished doctoral dissertation. Studia Philologica Jyväskyläensia, University
of Jyväskylä, Jyväskylä.
Richardson, U., Leppänen, P. H. T., Leiwo, M., & Lyytinen, H. (2002). Speech perception differs
in infants at familial risk for dyslexia as early as six months of age. Manuscript submitted for
publication.
Salmelin, R., Service, E., Kiesilä, P., Uutela, K., & Salonen, O. (1996). Impaired visual word process-
ing in dyslexia revealed with magnetoencephalography. Annals of Neurology, 40, 157–162.
Steffens, M. L., Eilers, R. E., Gross-Glenn, K., & Jallad, B. (1992). Speech perception in adult subjects
with familial dyslexia. Journal of Speech and Hearing Research, 35, 192–200.
Stein, J., & Fowler, S. (1985). Effect of monocular occlusion on visuomotor perception and reading in
dyslexic children. Lancet, 8446, 69–73.
Stein, J., & Walsh, V. (1997). To see but not to read: The magnocellular theory of dyslexia. Trends in
Neurosciences, 20, 147–152.
Tallal, P. (1980). Auditory temporal perception, phonics, and reading disabilities in children. Brain and
Language, 9, 182–198.
Tallal, P., Miller, S., & Fitch, R. H. (1993). Neurobiological basis of speech: A case for the preemi-
nence of temporal processing. Annals of the New York Academy of Sciences, 682, 27–47.
Tervaniemi, M., Saarinen, J., Paavilainen, P., Danilova, N., & Näätänen, R. (1994). Temporal integration
of auditory information in sensory memory as reflected by the mismatch negativity. Biological
Psychology, 38, 157–167.
Wagner, R. K., & Torgesen, J. K. (1987). The nature of phonological processing and its causal role in
the acquisition of reading skills. Psychological Bulletin, 101, 192–212.
Werker, J. F., & Tees, R. C. (1987). Speech perception in severely disabled and average reading
children. Canadian Journal of Psychology, 41, 48–61.
BRAIN RESPONSES AND RISK FOR DYSLEXIA 421

Winkler, I., & Czigler, I. (1998). Mismatch negativity: deviance detection or the maintenance of the
“standard.” NeuroReport, 9, 3809–3813.
Winkler, I., Kujala, T., Tiitinen, H., Sivonen, P., Alku, P., Lehtokoski, A., et al. (1999). Brain responses
reveal the learning of foreign language phonemes. Psychophysiology, 36, 638–642.
Witton, C., Talcott, J. B., Hansen, P. C., Richardson, A. J., Griffiths, T. D., Rees, A., et al. (1998). Sen-
sitivity to dynamic auditory and visual stimuli predicts nonword reading ability in both dyslexic
and normal readers. Current Biology, 8, 791–797.
Wood, F., Flowers, L., Buchsbaum, M., & Tallal, P. (1991). Investigation of abnormal left temporal
functioning in dyslexia through rCBF, auditory evoked potentials, and positron emission tomogra-
phy. Reading and Writing, 3, 379–393.
Yabe, H., Tervaniemi, M., Reinikainen, K., & Näätänen, R. (1997). Temporal window of integration
revealed by MMN to sound omission. NeuroReport, 8, 1971–1974.
422 LEPPÄNEN ET AL.