Neuropsychologia 122 (2019) 105–115

Contents lists available at ScienceDirect

Neuropsychologia
journal homepage: www.elsevier.com/locate/neuropsychologia

Diminished brain responses to second-language words are linked with T

native-language literacy skills in dyslexia
Sari Ylinena, , Katja Junttilaa, Marja Laasonenb,c,f, Paul Iversond, Lauri Ahonene, Teija Kujalaa
⁎

a
Cognitive Brain Research Unit, Department of Psychology and Logopedics, Faculty of Medicine, University of Helsinki, Finland
b
Department of Psychology and Logopedics, University of Helsinki, Finland
c
Department of Otorhinolaryngology and Phoniatrics, Head and Neck Surgery, Helsinki University Hospital and University of Helsinki, Finland
d
Division of Psychology and Language Sciences, Department of Speech, Hearing and Phonetic Sciences, University College London, UK
e
Finnish Institute of Occupational Health, Finland
f
Department of Psychology and Speech-Language Pathology, University of Turku, Finland

ARTICLE INFO ABSTRACT

Keywords: Dyslexia is characterized by poor reading skills, yet often also difficulties in second-language learning. The
Dyslexia differences between native- and second-language speech processing and the establishment of new brain re-
Reading skills presentations for spoken second language in dyslexia are not, however, well understood. We used recordings of
Speech the mismatch negativity component of event-related potential to determine possible differences between the
Language learning
activation of long-term memory representations for spoken native- and second-language word forms in Finnish-
Second language
speaking 9–11-year-old children with or without dyslexia, studying English as their second language in school. In
Mismatch negativity
addition, we sought to investigate whether the bottleneck of dyslexic readers' second-language learning lies at
the level of word representations or smaller units and whether the amplitude of mismatch negativity is corre-
lated with native-language literacy and related skills. We found that the activation of brain representations for
familiar second-language words, but not for second-language speech sounds or native-language words, was
weaker in children with dyslexia than in typical readers. Source localization revealed that dyslexia was asso-
ciated with weak activation of the right temporal cortex, which has been previously linked with word-form
learning. Importantly, the amplitude of the mismatch negativity for familiar second-language words correlated
with native-language literacy and rapid naming scores, suggesting a close link between second-language pro-
cessing and these skills.

1. Introduction Boets et al. (2013) found that speech sounds induced similar activation
of auditory cortices in dyslexic and typical readers but the structural
Developmental dyslexia refers to reading impairment despite suffi- and functional connectivity of the left arcuate fasciculus, connecting
cient tuition and normal intelligence, hearing, and vision. It is the most left superior temporal and frontal brain regions, was hampered in
common learning difficulty with prevalence of 3–10% (Snowling, dyslexic as compared with typical readers. Also other studies have
2000). Despite of a complex etiology, dyslexia is typically characterized linked arcuate fasciculus with reading skills (Hoeft et al., 2011;
by a phonological deficit, including deficient phonological awareness, Yeatman et al., 2012; Myers et al., 2014; Gullick and Booth, 2015),
phonological working memory, and rapid automatized naming (RAN) supporting previous notions that cerebral disconnection might con-
(Snowling, 2000; Ramus et al., 2003; for a recent review, see Ramus tribute to reading difficulties in developmental dyslexia (Paulesu et al.,
and Ahissar, 2012). The phonological deficit may be linked with in- 1996).
efficient auditory processing since newborn infants at risk for dyslexia In dyslexia, impaired reading in the native language is not the only
show atypical pattern of auditory event-related potentials (ERP) outcome of underlying phonological processing difficulties. In practice,
(Leppänen et al., 2010). Ramus and Szenkovits (2008) have, however, dyslexic readers often struggle with learning a foreign or a second
suggested that phonological representations may be intact in dyslexia, language, resulting in poor proficiency in this academically and socially
whereas access to phonological units is compromised. Recent neuroi- important skill. Second-language learning in dyslexia has been ad-
maging results have been interpreted to support this view. For example, dressed by Di Betta and Romani (2006), who compared the learning of

⁎
Corresponding author. Current address: CICERO Learning, Faculty of Educational Sciences, University of Helsinki, P.O. Box 9, FIN-00014 Helsinki, Finland.
E-mail address: [email protected] (S. Ylinen).

https://doi.org/10.1016/j.neuropsychologia.2018.11.005
Received 13 March 2018; Received in revised form 2 November 2018; Accepted 7 November 2018
Available online 08 November 2018
0028-3932/ © 2018 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/BY/4.0/).
S. Ylinen et al. Neuropsychologia 122 (2019) 105–115

native-language pseudowords and non-native words associated with and second-language words.
pictures between adults with or without dyslexia. Those with dyslexia To this end, we examined the activation of long-term memory re-
were found to have a specific deficit in the learning of word forms presentations for spoken native-language and second-language word
across languages, although the processing of word meanings was intact. forms in Finnish-speaking 9–11-year-old children with or without
Paired associate learning tasks with pseudowords have resulted in si- dyslexia, studying English as a second language in school. To tap these
milar conclusions (Litt and Nation, 2014). Soroli et al. (2010), in turn, brain representations, we used recordings of auditory ERP and its
explored non-native speech perception and production in adults with or mismatch negativity (MMN) component (Näätänen et al., 1978; for a
without dyslexia. They used discrimination and repetition tasks in- review, see Näätänen et al., 2007). The MMN is typically elicited in an
cluding native and foreign phonetic segments and word-stress cues oddball paradigm, which refers to a sequence of sounds with a high
embedded in pseudowords. Only the discrimination of foreign word probability of repeating standard sounds and a low probability of de-
stress was impaired in dyslexic readers, when short-term memory was viant sounds that differ acoustically from the standard. In the brain, the
challenged. standards automatically generate a predictive model about following
Recent findings showing deficient functional connectivity through sounds (Winkler, 2007), and deviants violating these predictions elicit
the arcuate fasciculus in dyslexia (Hoeft et al., 2011; Boets et al., 2013; the MMN. Its amplitude varies as a function of acoustic deviance, but it
Gullick and Booth, 2015) may be critical for explaining dyslexic is also enhanced by the activation of long-term memory representations
readers’ difficulties in second-language learning. The left frontal areas for speech sounds and words (e.g., Näätänen et al., 1997; Pulvermüller
have been suggested to participate in the categorization of speech et al., 2001; for reviews, see Näätänen, 2001; Pulvermüller and Shtyrov,
sounds and the establishment of non-native phonetic categories (Myers, 2006). Importantly, the MMN is elicited even when attention is directed
2014). However, the participation of these areas in the categorization of away from stimulation (Näätänen et al., 2007), which may be parti-
speech sounds depends on their connections with auditory cortex in the cularly important when studying developmental disorders (Ramus and
temporal lobe, a candidate tract being the arcuate fasciculus (Saur Ahissar, 2012).
et al., 2008). In addition to the processing of speech sounds, the left By comparing MMN responses between children with or without
arcuate fasciculus has been suggested to mediate word learning (López- dyslexia, we aimed, firstly, to determine whether the processing of
Barroso et al., 2013). Thus, the functional connectivity between the left second-language words is more compromised than that of native-lan-
temporal and frontal brain areas may be particularly important for guage words in dyslexia, and whether this is due to sublexical or lexical
establishing both phonetic and word representations in the brain, and (word-form) representations. Secondly, to account for potential pro-
both kinds of representations are required for second-language cessing differences between the groups of children with or without
learning. As a result, weak connectivity between these brain areas in dyslexia, we used source modeling to investigate the activation of
dyslexia may be expected to result in poor second-language learning. temporal auditory areas and the left frontal cortex during the early
However, several questions about second-language learning and its word processing stages (< 200 ms from the recognition point) that are
link with literacy in dyslexia remain unanswered. Neuroimaging studies most relevant for word recognition (Pulvermüller and Shtyrov, 2006).
have suggested that the arcuate fasciculus is deficient in dyslexia (Hoeft Thirdly, we sought to investigate whether the activation of brain re-
et al., 2011; Boets et al., 2013; Gullick and Booth, 2015), yet its effect presentations for native and second language is linked with literacy and
on different levels of speech processing and their temporal dynamics, aspects of phonological processing, such as rapid naming, which are
which is out of the scope of functional magnetic resonance imaging typically impaired in dyslexia. Since dyslexia is often accompanied by
(fMRI) and diffusion tensor imaging (DTI) studies, remains unclear. The difficulties in second-language learning, we hypothesized that children
connectivity between the left frontal and temporal brain areas might with dyslexia would show particularly pronounced difficulties in
affect the learning of both non-native phonetic categories (Myers, 2014) second language processing. Further, we expected that if the processing
and words or word forms (López-Barroso et al., 2013), raising a ques- of unfamiliar second-language items (lacking lexical representations)
tion whether the 'bottleneck' of spoken second-language learning in was inefficient but native-language processing was intact in dyslexia,
dyslexia is the establishment of sublexical (phonetic or phonemic) re- then second-language phonetic representations would be of poor
presentations or lexical representations that consist of word forms quality or poorly activated. If this was the case, there would be a deficit
linked with meaning. On the basis of findings suggesting intact pro- at least at the sublexical level. However, inefficient processing of fa-
cessing of native and foreign phonetic segments (Soroli et al., 2010) but miliar second-language words that are represented in long-term
impaired learning of native or non-native word forms in dyslexia (Di memory but intact processing of unfamiliar second-language items in
Betta and Romani, 2006; Litt and Nation, 2014), we hypothesize that children with dyslexia would suggest that the learning of word forms is
the bottleneck is the establishment of new word representations, spe- more impaired than their auditory processing. In addition, we hy-
cifically that of word forms required for full lexical representations, pothesized that second-language processing as reflected by the MMN
which may be linked with more general serial processing deficits in responses might correlate with literacy and phonological skills due to
dyslexia (Hari and Renvall, 2001; Ben-Yehudah et al., 2004; Laasonen the role of the same brain structures in literacy (Hoeft et al., 2011; Boets
et al., 2012; Ramus and Ahissar, 2012; Romani et al., 2015; Majerus et al., 2013; Gullick and Booth, 2015) and word learning (Hickok and
and Cowan, 2016). In this case, however, the question rises whether Poeppel, 2007; López-Barroso et al., 2013). Because of potentially de-
dyslexia exerts a different influence on the processing of spoken native- ficient functional connectivity between temporal and frontal areas,
language and second-language words in the brain, since the establish- children with dyslexia were also hypothesized to show weaker activa-
ment of representations for word forms in either language requires tion of the left frontal cortex for second-language words.
serial processing. On the one hand, the learning of new word forms has
been suggested to be similarly impaired for native and second language
in adults with dyslexia (Di Betta and Romani, 2006), and delays in 2. Materials and methods
children's native-language development are sometimes associated with
reading difficulties (Lyytinen et al., 2005). On the other hand, however, 2.1. Ethics statement
dyslexia cannot be detected on the basis of native-language spoken
language skills only and dyslexic readers typically struggle more with The study was approved by the Ethical Committee of Helsinki
second-language learning than with native-language learning, speaking University's Faculty of Behavioural Sciences. Participation was volun-
against a general word learning difficulty in dyslexia. Thus, it is not tary. Participants and their caregivers gave informed oral or written
clear whether children with or without dyslexia show a different pat- consent, respectively.
tern of activation of neural representations for familiar native-language

106
S. Ylinen et al. Neuropsychologia 122 (2019) 105–115

2.2. Participants Finnish-speaking environment except one child with dyslexia who had
been exposed to English between 2 and 4 years of age while living
Participants were 9–11-year-old native speakers of Finnish, who abroad. According to parental reports, the children had no other neu-
studied English as a second language in school. Based on pre-tests and rological problems, had normal hearing, and were right-handed, with
background screening, participants were assigned into two groups, the exception of two control children being left-handed. Parents re-
namely, children with or without dyslexia. The latter served as con- ported that outside school, children with or without dyslexia were ex-
trols.1 The dyslexia group included 19 children (7 girls, 12 boys; mean posed to English TV programs for 3.5 and 3 h and English games for 2
age 10 years 9 months, SD 8 months). On average, they had studied and 1.9 h per week, respectively.
English for 16 months (SD 7 months). The control group consisted of 21 As shown by Table 1, no differences were found between the groups
children (9 girls, 12 boys; mean age 10 years 6 months, SD 10 months) in WISC block design, vocabulary, and coding tasks, whereas in digit
who had studied English for 15 months on average (SD 7 months). span tapping phonological short-term memory controls outperformed
Participants were screened with pre-tests that measured literacy, the children with dyslexia. RAN was significantly faster in controls than
intelligence, and phonological skills. All of them were conducted in in dyslexic readers for both colors and letters. Controls had also better
Finnish. Literacy was assessed with standardized Finnish tests ALLU phonological awareness than the children with dyslexia.
(Lindeman, 1998) and LukiLasse (Häyrinen et al., 1999). ALLU requires Since 8 out of 19 children were included in the dyslexia group on
the segmentation of letter strings lacking spaces into words within the the basis of their pre-test performance, we also compared z-scores of
time limit of 3.5 min. This test enables to divide readers into weak children with or without a dyslexia diagnosis to see whether these
(23%), average (54%) and above average (23%) groups, which are subgroups’ literacy and phonological skills differed from each other. No
further divided into three skill levels each (i.e., altogether nine skill differences were found in word reading, technical reading, phonolo-
levels). LukiLasse tests the accuracy and speed of reading single words gical awareness or RAN colors (see Table 2). Regardless of diagnosis, all
aloud within the time limit of 2 min and writing dictated words or children in the dyslexia group had scores falling to the lowest 10th
sentences without time limit. Intelligence was examined with Wechsler percentile in word reading. However, the children without a diagnosis
Intelligence Scale for Children IV (WISC; Wechsler, 2010), including outperformed those with a diagnosis in dictation, whereas children with
block design, digit span, vocabulary, and coding. Naming was de- a diagnosis named letters faster than those without a diagnosis (see
termined with Rapid Alternating Naming test (RAN; Ahonen et al., Table 2). Since reading skills were very similar between the subgroups
2003), including color and letter tasks. Finally, phonological awareness (if anything, scores were slightly lower in those without a diagnosis), it
was assessed with Common unit test, where the task is to name a seems likely that the observed differences were mainly due to some
common phoneme in two auditorily presented words. Specifically, the other factors than the severity of dyslexia. A possible factor affecting
child is presented with 15 spoken pseudoword pairs (e.g., lauhkua – the results is time: the children without a diagnosis were particularly
terike, aamestus – hilpialli) via headphones. The pseudowords consist of slow as indicated by RAN letters task, but they could compensate their
Finnish phonemes and their length varies between five and 12 pho- deficit in the dictation task lacking time constraints. It is also note-
nemes. The common phoneme can be at any position in the pseudo- worthy that in line with previous observations that males are more
word. Before the assessment trials, the child is presented with three likely to be referred for evaluation than females with equivalent
practice trials (e.g., mirki – lukke). In the practice trials, if the child's reading problems (Quinn and Wagner, 2015), among diagnosed chil-
answer is correct, the researcher praises the child and repeats the cor- dren there were ten boys and one girl, whereas among children without
rect answer. If the child does not give the correct answer, the researcher a diagnosis there were six girls and two boys. Such ascertainment bias
repeats the pseudowords and tells the child the correct answer. After (Quinn and Wagner, 2015) may explain why most of our female par-
practice trials, each of the 15 assessment trial pairs is presented only ticipants did not have diagnoses despite equivalent deficits in reading.
once and the child is asked to name aloud which sound they heard to be
similar in both pseudowords. No feedback is given during the assess-
ment trials. The test score is the sum of correct answers in the assess- 2.3. Experimental design
ment trials. In the pre-tests, children sat in a quiet room with an ex-
perimenter who presented the tests. Testing took less than one hour. 2.3.1. Stimuli
To be included in the study, all children had to reach the criterion of Stimuli consisted of spoken English and Finnish words and pseu-
not having scores below one standard deviation in any WISC subtest dowords (marked with an asterisk from hereon) [ʃu:] shoe, [ʃi:] she,
included in the study (i.e., minimum standard score was 7 in block [ʃɑɪ] shy, and [ʃᴐɪ]* shoy*, and [su:] suu ‘mouth’, [si:]* sii*, [sɑi] sai ‘got’
design, digit span, vocabulary, and coding). In addition, the criteria for and [soi] soi ‘plays/rings/played/ringed/allowed’ (for simplicity, the sti-
inclusion in the group of dyslexic readers were a dyslexia diagnosis or muli will be hereafter referred to with their orthographic form).
performance in the literacy-skill pre-test as follows: 1) the standard Because Finnish phonology does not include /ʃ/, the English items
score in the single word reading task was at least one standard devia- could be recognized as foreign from their very beginning. The English
tion below the average and either 2a) the standard score in the dictation items were expected to differ as a function of the frequency of use: she
task was at least one standard deviation below the average or 2b) scores was familiar and frequently used on English lessons, shy less frequently
in the technical reading task belonged to the weakest 23%. Out of 19 used, and pseudoword shoy* not used at all. According to a frequency
children with dyslexia, 8 were included on the basis of their perfor- ranking, she is among the most frequent English words, occurring 3801
mance and 11 had a dyslexia diagnosis, yet all of them fulfilled the pre- times per million tokens, whereas shy and shoe are less frequent, each
test criteria. The criteria for inclusion in the control group were as occurring 11 times per million tokens (Leech et al., 2016). The fre-
follows: no dyslexia suspicions, standard scores in the single word quencies of the Finnish words were 355 per million tokens for sai, 32 for
reading and dictation tasks being no more than one standard deviation soi and 32 for suu (Huovilainen, 2018). The Finnish word forms were
below the average, and scores of the technical reading task not be- chosen to match the English word forms phonetically rather than by
longing to the weakest 23%. All participants had lived their life in a their frequency of occurrence.2 The frequent English word she has a

2
Unfortunately, all aspects could not be matched at the same time between
1
Also a group of younger typically developing children matched with the the Finnish and English stimuli due to the limited selection of monosyllabic
dyslexia group according to reading level would have been useful as a control. words in Finnish. Since it is well-established that the MMN is very sensitive to
However, it would have been very difficult to find younger children with acoustic changes, the comparison between MMNs to Finnish and English items
comparable exposure to English. would have been difficult or impossible, if the stimuli had been matched

107
S. Ylinen et al. Neuropsychologia 122 (2019) 105–115

Participants’ literacy and cognitive skill scores. The two columns on the left show the mean ( ± SD) and range [min max] of standard scores for standardized tests (WISC and LukiLasse), skill level (1–9) for ALLU, raw
response times (in seconds) for RAN, and raw score for Common unit. The next two columns show the mean and range [min max] of z-scores for all tests. LukiLasse and ALLU were used as the criteria of inclusion to the higher frequency value than any of the Finnish words, yet it is note-

t-test comparing the groups

worthy that due to their limited exposure to English, the Finnish par-

t = −15.79, p < 0.001

t = −6.92, p < 0.001
t = −7.06, p < 0.001

t = −2.76, p = 0.009

t = −2.33, p = 0.025
t = −3.17, p = 0.003
t = −3.76, p = 0.001
ticipants must have been exposed to any of the native-language words
used here more than any of the English words.

t = −0.20, n.s.

t = −1.06, n.s.
t = −1.63, n.s.
The stimuli were modified from natural recordings. During the re-
cording in a sound-shielded room, an early bilingual male native
speaker of Finnish and English pronounced the words and pseudowords
several times. The recordings were screened by native speakers of
Finnish and English, and then processed so that they were equalized in
terms of segmental durations, pitch contours, and amplitude envelopes
within Praat (Boersma and Weenink, 2010). Specifically, each re-
− 0.16 [− 2.69 1.06]
− 0.08 [− 2.33 1.98]
cording was segmented into fricatives and vowels; an overlap-add
0.57 [− 0.33 1.33]

0.60 [− 0.67 1.67]

0.05 [− 1.47 1.65] method was used to adjust the duration of each fricative and vowel to
Controls: z-score

0.30 [− 0.67 1]

0.40 [− 1 2.33]
0.16 [− 1 1.33]
0 [− 0.7 1.9]

their respective average duration; an envelope-add method was used to

0.68 [− 1 3]

impose the pitch contour for an example syllable (shoy*) onto each
recording; then the root mean square intensity envelope for an example
syllable (shy) was imposed on each of the recordings (see Fig. 1).

2.3.2. EEG recordings and analysis

Children with dyslexia: z-score

During electroencephalography (EEG) recording, the children sat

comfortably in a chair and watched a soundless video without subtitles,
− 1.26; [− 2.6 − 0.1]

− 0.96 [− 4.12 0.78]

− 1.27 [− 3.96 1.26]
− 1.96 [− 3 − 1.33]

while auditory stimuli were delivered through headphones on a 50 dB

− 1.07 [− 2.61 0.8]
− 1.46 [− 3 0.33]

− 0.42 [− 1 0.67]

hearing level. Before the measurement, they were instructed to ignore

0.35 [− 1 1.33]
0.18 [− 1 2.67]

the auditory stimuli and to concentrate on the video. EEG was recorded
0.33 [− 1 2]

with Biosemi ActiView 6.05 system and a 64-channel cap and addi-
tional electrodes on the nose, two mastoids, and an extra electro-
oculogram (EOG) below the left eye. Reference electrode was the de-
fault reference of the Biosemi system (approximately at PO1). Testing
took less than two hours including preparation and breaks.
Finnish and English items were presented in separate sequences in
Controls: standard score/skill

an oddball paradigm. The order of stimuli in the sequences was pseu-

dorandom (there were always at least two standards between deviant
11.71 ( ± 1.31) [9 14]
10.90 ( ± 1.55) [8 13]

11.19 ( ± 3.19) [7 17]

10.48 ( ± 2.20) [7 14]
11.81 ( ± 2.34) [8 15]
12.05 ( ± 3.04) [7 19]

9.38 ( ± 2.94) [4 15]

5.10 ( ± 1.37) [4 9]

stimuli). In the English sequence, shoe served as the standard stimulus

44 ( ± 7) [33 58]
26 ( ± 4) [18 37]

(p = 0.80) and she, shy and shoy* as the deviant stimuli (p = 0.066
level/raw score

each). In the Finnish sequence, suu served as the standard (p = 0.80)

and sii* , sai and soi as the deviants (p = 0.066 each). Stimulus onset
asynchrony was 800 ms. Both sequences were divided into six blocks,
groups of children with dyslexia and their controls and WISC subtests as participation criteria.

which were presented to the participants in sequences of three blocks

for each language, after which the language changed. The order of
languages was counterbalanced. Altogether, each deviant stimulus was
presented 156 times and each standard stimulus 1902 times.
Children with dyslexia: standard score/

After the recording, EEG data were re-referenced to nose to check

whether responses had typical MMN topography, bad EEG channels, if
any, were interpolated using the other EEG channels, the data were
filtered with a pass-band of 1–30 Hz (roll-off 24 dB/octave), epoched
11.00 ( ± 2.83) [7 16]

11.05 ( ± 2.15) [7 14]

10.53 ( ± 2.84) [7 18]

− 100–800 ms, epochs with artifacts exceeding ± 100 µV at any

5.63 ( ± 3.09) [1 11]

8.74 ( ± 1.73) [7 12]

5.42 ( ± 3.72) [0 12]

2.42 ( ± 0.96), [1 5]
4.11 ( ± 1.73) [1 6]
skill level/raw score

49 ( ± 11) [33 78]

channel were rejected, and the ERP responses were separately averaged
31 ( ± 7) [18 45]

for each stimulus type with BESA Research 6.0. Using in-house Matlab
scripts, the baseline of the waveforms was corrected to zero at a 100 ms
pre-stimulus window and standard ERPs were subtracted from the de-
viant ERPs to determine their difference and to identify the MMN re-
sponses. Before quantification, the data were re-referenced to the
average of the mastoids to improve signal-to-noise ratio. To quantify
MMN amplitudes, peak latencies were first determined from grand-
Common unit (phonological awareness)

average waveforms for each stimulus type and group at FCz showing
maximal amplitudes, and then the peak latencies were averaged across
the groups for each stimulus type. If responses to word forms with
diphthongs had two peaks, the latency of the later peak was selected.
LukiLasse, word reading

ALLU, technical reading

LukiLasse, dictation

WISC block design

RAN letters, speed

RAN colors, speed

(footnote continued)
WISC vocabulary
WISC digit span

according to the word frequency rather than the phonetic structure. We cannot
WISC coding

exclude the possibility that word frequency caused some MMN amplitude dif-
ferences between stimulus types (see Shtyrov et al., 2011), yet the frequency
Table 1

does not solely explain any group differences which were of exclusive interest
here (see Discussion for details).

108
S. Ylinen et al. Neuropsychologia 122 (2019) 105–115

Table 2
Literacy and phonological skill z-scores and range [min max] in children with or without a dyslexia diagnosis within the dyslexia group.
Children with a diagnosis Children without a diagnosis t-test comparing the subgroups

LukiLasse, word reading −1.91 [−3 −1.33] −2.04 [−3 −1.33] t = 0.49, n.s.
LukiLasse, dictation −1.88 [−3 0] −0.88 [−1.67 0.33] t = −2.34, p = 0.032
ALLU, technical reading −1.13 [−2.6 −0.1] −1.3 [−1.9 −0.9] t = 0.65, n.s.
RAN colors, speed −0.85 [−4.12 0.78] −1.11 [−2.83 0.5] t = 0.43, n.s.
RAN letters, speed −0.69 [−2.93 1.26] −2.07 [−3.96 −0.61] t = 2.49, p = 0.024
Common unit (phonological awareness) −1.27 [−2.61 0.8] −0.8 [−2.32 0.51] t = −0.95, n.s.

Fig. 1. Spectrograms of the stimuli. An asterisk denotes a pseudoword.

This is because especially in the native language with larger lexicon addition to the second vowel change. A 40 ms time window, centered at
and, consequently, many lexical competitors, the words could not be the average latency, was then used to measure mean amplitudes from
recognized with certainty before the last vowel. Therefore, it is likely individual difference waveforms at fronto-central scalp sites F3, Fz, F4,
that the first peak was elicited by the first vowel change, whereas the FC3, FCz, FC4, C3, Cz, and C4. For both groups, the time windows were
second peak reflected word recognition (Pulvermüller et al., 2001), in 218–258 ms for she, 210–250 ms for shy, 224–264 for shoy*,

109
S. Ylinen et al. Neuropsychologia 122 (2019) 105–115

199–239 ms for sii*, 258–298 ms for sai, and 295–335 for soi. and 12 controls). We were also interested in differences between the
To localize the sources of MMN responses differing between the groups in the activation of the left frontal source because it was hy-
groups, equivalent current dipole (ECD) methods were applied to de- pothesized to be deficient in dyslexia. However, we could not compare
viant-minus-standard difference waveforms in BESA Research 6.0. statistically the groups’ dipole strengths because the left frontal source
Source localization was conducted only for the difference waveforms of was found in too few children.
those participants who showed an MMN, that is, who showed a fronto- To study whether the activation of brain representations is linked
central negative peak between 200 and 300 ms (as determined by the with cognitive skills, we conducted correlation analysis (Pearson's r)
grand-average difference waveform). Since individual MRIs were not between MMN mean amplitude for the frequent familiar second-lan-
available, the head model of Besa Research 6.0 for 10–12 year-old guage word she, averaged across fronto-central electrode sites, and
children was used. Principal component analysis (PCA) was used to scores in literacy and RAN. Literacy and RAN scores used in compar-
estimate the contribution of sources in Besa Research 6.0. We hy- isons consisted of the average z-score across the three literacy tasks
pothesized that the MMN would have bilateral temporal sources and (LukiLasse reading, LukiLasse dictation, ALLU) and the average z-score
possibly a left frontal source, if speech production areas contributed to across RAN speed in color and letter tasks, respectively, where positive
the processing. To capture the activation of bilateral auditory areas in values denote above-average performance and negative values denote
the temporal lobes, two dipoles were fitted in 40 ms time windows below-average performance. In addition, since previous work by
centered at individual MMN peak latencies between 200 and 300 ms. Kimppa et al. (2018) found different effects between children with
Activation was expected to spread to frontal areas slightly later (Opitz dyslexia and typical readers when correlating ERP and LukiLasse dic-
et al., 2002), and therefore a third dipole was fitted either in the same tation, we run correlation analysis of MMN and LukiLasse dictation z-
time window or in a time window that started 10 ms later, depending scores separately for the two groups. Finally, correlation analysis was
on the timing of residual activity that was not explained by the first and run for the MMN amplitude for she and the strength of its right temporal
second dipoles. In few cases, fitting of a fourth dipole was attempted to source to see whether the activity of this source is linked with the MMN
see whether there was bilateral frontal activity, if the third dipole was differences. All statistical comparisons were conducted with SPSS and
located in the frontal area between the hemispheres. Since we aimed to alpha level 0.05 was applied.
study source strengths of specific, pre-determined sources (left and
right temporal and left frontal) and averaging source waveforms across 3. Results
different brain areas was not feasible, the source waveforms of ECDs
falling outside the regions of interest were not further analyzed. For ANOVA for the MMN revealed a significant interaction of Group,
example, the strength of ECDs at parietal, parieto-occipital, right frontal Language, and Word form [F(2,76) = 3.63, p = 0.031]. According to
or subcortical areas was not analyzed. In addition, if there were two pairwise comparisons, the groups did not differ significantly from each
temporal sources in the same hemisphere, only the waveform of the other in any of the native-language word forms. For the second-lan-
source closest to auditory cortex was chosen for source strength ana- guage word forms, the dyslexic readers' MMN was significantly weaker
lysis. for the second-language word she compared with controls (p = 0.004),
Individual participants were included in the statistical analysis of whereas no significant difference between the groups was found for the
sources if they had a minimum of 75% goodness of fit of the model and other second-language word forms (see Figs. 2 and 3 and Table 3).
at least one source in the cortex or its vicinity. As a result, 17 children Source modeling typically (but not always) resulted in bilateral
with dyslexia and 17 controls were included in the statistical analysis of temporal dipoles and a third dipole in frontal, temporal or parietal
sources, yet not all participants showed all sources that were sought. On areas, yet only the strength of the bilateral temporal and left frontal
average, the model with three (or four) dipoles reached a goodness of fit sources were further analyzed. No significant differences were found
of 87%. To quantify the activation of the left and right temporal and left between the strength of ECDs in the left temporal source [25.15 vs.
frontal sources, we measured peak latencies of grand-average source 20.98 nAm for controls and children with dyslexia; t(20) = −0.57,
waveforms for each source and each group and averaged the latencies n.s.]. However, the right temporal source was more strongly activated
across the groups. Then individual dipole strengths were measured for in controls than in children with dyslexia [23.23 nAm vs. 13.06 nAm,
each dipolar source in 40 ms time windows centered at the averaged respectively; t(23) = −2.14, p = 0.043; see Fig. 4]. The strength of the
latencies, which were 228–268 ms for the left temporal source, right temporal source correlated with the MMN amplitude for she,
217–257 ms for the right temporal source, and 220–260 ms for the left r = −0.41, p = 0.043. The left frontal source was found only in nine
frontal source. controls (7.09 nAm) and four children with dyslexia (12.13 nAm).
Overall, the contribution of the frontal source to the MMN was small:
2.4. Statistical analysis according to PCA, the contribution of the left temporal source was up to
9% in individuals with bilateral temporal and left frontal sources.
Statistical comparisons of cognitive test scores between the groups Correlation analysis conducted across groups indicated a significant
were conducted with independent samples t-tests and the statistical negative correlation between literacy scores and the MMN amplitude
analysis of the MMN responses with repeated-measures ANOVA in- for the familiar second-language word she, r = −0.49, p = 0.001,
cluding factors Group (children with dyslexia vs. controls), Language suggesting that larger (more negative) MMN was associated with higher
(native vs. second language), Word form (she/sii* vs. shy/sai vs. shoy*/ literacy scores (see Fig. 5). When correlations were tested within each
soi), Coronal scalp site (frontal, fronto-central, central), and Sagittal group, they did not reach significance (children with dyslexia:
scalp site (left, midline, right). Significant effects were followed up by r = −0.30, n.s.; controls r = −0.15, n.s.). However, it is noteworthy
Bonferroni-corrected pairwise comparisons if needed. Only main effects that the use of literacy scores as an inclusion criterion decreased var-
or interactions involving the factor Group are reported, because any iation of these scores within the groups, as participants not reaching the
conclusions about dyslexia are based on differences between the criterion were excluded. This affects especially the scores of LukiLasse
groups, whereas the other factors are not per se relevant for the re- word reading where −1 SD cut-off was applied to all children. Previous
search questions. work by Kimppa et al. (2018) has suggested that LukiLasse dictation
The statistical analysis of the dipole strengths of temporal sources task correlates differently with ERP between children with dyslexia and
was conducted with independent samples t-test comparing the groups. typical readers. Therefore, correlations with MMN and LukiLasse dic-
Bilateral temporal sources, or at least one temporal source, were found tation z-scores were run for the two groups separately. A significant
in most participants (in the left hemisphere in 12 children with dyslexia negative correlation was found in children with dyslexia only,
and 10 controls; in the right hemisphere in 13 children with dyslexia r = −0.46, p = 0.046 (controls: r = 0.22, n.s.). Across groups, a

110
S. Ylinen et al. Neuropsychologia 122 (2019) 105–115

4. Discussion

The present study aimed to determine whether long-term memory

representations for spoken second-language word forms are compro-
mised in dyslexia, whether this is due to sublexical (phonetic) or lexical
(word-form) processing, and which cortical sources underlie their de-
ficient activation. In addition, we investigated how the activation of
brain representations is linked with cognitive skills that are impaired in
dyslexia, such as literacy and rapid naming. Compared with typical
readers, in children with dyslexia the speech processing as indexed by
the MMN was found deficient for the most familiar second-language
word only, whereas no significant group differences were found in the
processing of the other native-language and second-language word
forms. Source localization suggested that for the most familiar second-
language word, the right auditory cortex was weakly activated in
children with dyslexia. Children's MMN amplitude for the most familiar
second-language word correlated significantly with literacy scores and
RAN speed, particularly in children with dyslexia.
Diminished MMN for the most frequent second-language word she
in children with dyslexia with the absence of MMN group differences
for the other second-language or native-language word forms suggests
that the dyslexic readers' auditory or speech processing was not gen-
erally impaired - at least when processing speech stimuli lacking rapid
cues of recognition (cf. Schwartz and Tallal, 1980). This is in line with
Froyen's et al. (2011) previous findings, suggesting similar MMNs to
vowels in children with or without dyslexia in an auditory condition
(yet not in an audiovisual condition). Poor sublexical representations
do not account for the pattern of results in the current dyslexic sample
either, since poor phonetic representations should have affected the
processing of the other word forms as well. This is because the initial
consonant /ʃ/ was included not only in she but also in the other second-
Fig. 2. MMN responses to foreign and native items in children with or without
language stimuli which did not elicit diminished responses. Similarly,
dyslexia at FCz. An asterisk denotes a pseudoword.
the vowel /i/, which is quite similar in English and Finnish (Wiik, 1965;
see also Fig. 1 for spectrograms), was not only included in she but also
in the native-language word form sii*, and the response to the latter was
not diminished in children with dyslexia. If only the sublexical level
determined the processing, the categorization of /i/ should have been
possible via the intact native-language /i/ representation. In addition,
sublexical deficits in second language should have been observed in the
MMN for shoy* which is an unfamiliar second-language item and does
not have a lexical representation in the long-term memory, yet no dif-
Fig. 3. Scalp distributions of MMN responses to foreign and native items in ference was observed between the groups for shoy*. Thus, in line with
children with or without dyslexia (average reference). An asterisk denotes a Soroli et al. (2010), we found no evidence of deficits in the processing
pseudoword. of second-language speech sounds in dyslexia.
Group differences were observed for the second-language word she
that has a high frequency. Since the MMN amplitude may be modulated
significant negative correlation was also found between the MMN am-
by word frequency (Shtyrov et al., 2011), we cannot exclude the pos-
plitude for the second-language word she and RAN speed, r = −0.57,
sibility that word frequency may have affected the amplitudes in the
p < 0.001. When tested within each group, this correlation persisted
present data as well. However, the high word frequency of she in
being significant in children with dyslexia (r = −0.48, p = 0.04) and
English cannot solely account for the group difference, since no such
marginally significant in controls (r = −0.43, p = 0.054).
difference was observed for the most frequent native word sai and the
children must have been exposed to native sai considerably more than
to second-language she. In addition, it has been shown that reading
promotes vocabulary growth (Duff et al., 2015), which raises a

Table 3
MMN amplitudes ( ± SD) in children with dyslexia and controls, averaged across fronto-central channels that were included in statistical analysis. An
asterisk denotes a pseudoword.
MMN amplitude (µV) in children with dyslexia MMN amplitude (µV) in controls

Second-language, she − 1.27 ( ± 1.65) − 2.91 ( ± 1.74)

Second-language, shy − 1.08 ( ± 1.61) − 1.63 ( ± 1.64)
Second-language, shoy* − 0.62 ( ± 1.49) − 0.76 ( ± 1.64)
Native-language, sii* − 1.99 ( ± 2.05) − 1.56 ( ± 1.73)
Native-language, sai − 1.09 ( ± 1.97) − 1.00 ( ± 1.69)
Native-language, soi − 0.27 ( ± 1.42) − 0.70 ( ± 1.91)

111
S. Ylinen et al. Neuropsychologia 122 (2019) 105–115

Fig. 4. Top: Source waveforms obtained from ECD analysis for left temporal, right temporal and left frontal sources in children with or without dyslexia. Bottom:
ECDs in left temporal, right temporal and left frontal areas in a child with dyslexia and a control child (placed on MRI template, not individual MRIs).

possibility that deficits with auditory word form representations could word has enabled the control children to establish a robust re-
be mediated by the reduced exposure to print in dyslexic individuals. presentation for it, whereas in children with dyslexia the representation
However, according to WISC vocabulary results, the current groups did is weaker, as indicated by a smaller MMN. The reason for not seeing a
not differ from each other significantly in this respect in their native significant group difference in the other second-language word shy may
language. In addition, the amount of exposure to the second language in be that, on average, the representation for this less frequent non-native
spoken or written form could affect the activation of auditory second- word is still quite weak in both groups due to limited exposure. In line
language word forms. Nevertheless, the present groups did not differ with this, no difference was observed between the groups for the un-
from each other in the length of their English studies in school, and familiar pseudoword shoy* that cannot have a word-form representa-
according to parental reports, children with dyslexia had not been ex- tion in the long-term memory at all. Taken together, this data pattern
posed to English less than controls outside the school (average exposure suggests that dyslexia is associated with an impairment in forming ro-
to English TV programs 3.5 h vs. 3 h per week and to English games 2 h bust brain representations for new second-language word forms, which
vs. 1.9 h per week in children with dyslexia and controls, respectively). manifests when the representations should have emerged to the long-
Thus, exposure being similar between the groups, differences between term memory. This impairment may form a bottleneck for second-lan-
them suggest that children with dyslexia may not benefit from the ex- guage learning in dyslexia. Related to long-term memory, recent re-
posure as much as controls. search has shown associations between reading skills and overnight
Since acoustic or sublexical processing deficits, word frequency ef- vocabulary consolidation effects at the cortex (Landi et al., 2018).
fects, vocabulary differences, or the length of exposure do not account Furthermore, poor learning of novel words has been linked with the
for the findings, the observed data pattern seems to be best explained by consolidation of vocabulary during sleep in children with dyslexia
dyslexic readers' compromised long-term memory representations for (Smith et al., 2018). Therefore, the consolidation of word forms may
second-language word forms. The difference between the groups that contribute to the impairment in forming novel word representations.
reached significance was found for the most familiar second-language The finding of compromised neural representations for second-lan-
word she. According to our interpretation, repeated exposure to this guage words is compatible with previous behavioral findings by

Fig. 5. Correlations between MMN amplitudes for the familiar second-language word she and literacy scores (left) and rapid naming (RAN, right).

112
S. Ylinen et al. Neuropsychologia 122 (2019) 105–115

Di Betta and Romani (2006) and Litt and Nation (2014), who suggested short-term memory, and naming (Ramus and Ahissar, 2012; Vellutino
a specific deficit in the learning of word forms in dyslexia. In general, et al., 2004; Wagner and Torgesen, 1987). Interestingly, literacy and
difficulties in word-form learning could be linked with the deficit in naming skills were found to correlate with the MMN amplitude for the
serial processing in dyslexia (Hari and Renvall, 2001; Ben-Yehudah second-language familiar word (Fig. 5), suggesting that better perfor-
et al., 2004; Laasonen et al., 2012; Ramus and Ahissar, 2012; Romani mance in these cognitive tasks is linked with stronger activation of
et al., 2015; Majerus and Cowan, 2016), although with short words second-language representations (as indicated by larger MMN re-
used here the serial processing requirements should be minimal. It is sponses). Previous studies have suggested that reading and naming
noteworthy, however, that according to the present data, children with share the same neural networks in the brain (McCrory et al., 2005),
dyslexia did not show any word-form processing deficit in their native albeit with a different degree of activation (Price et al., 2006). Corre-
language, suggesting that their long-term memory representations for spondingly, a plausible account for the link between literacy skills,
native words were likely as robust as those of the controls (see naming and brain representations for spoken second-language words is
Pulvermüller et al., 2001). This implies that difficulty to represent word the involvement of the same brain structures in all of them, most likely
forms that is observed in learning experiments (Di Betta and Romani, including the dorsal stream of speech processing (see Scott and
2006; Litt and Nation, 2014) might be overcome by sufficient exposure Johnsrude, 2003; Hickok and Poeppel, 2007). In their dual-stream
or practice, which results in automatization. That is, dyslexic in- model, Hickok and Poeppel (2007) suggest that the dorsal stream serves
dividuals might need more exposure to establish robust word re- auditory-motor integration by connecting the auditory analysis areas of
presentations than typical readers, but for native language this need the temporal cortex and the articulatory network in the frontal cortex
may not become evident outside laboratory since native-language ex- via a sensory-motor interface. Furthermore, auditory-motor interaction
posure is readily available. is proposed to occur at two levels, where a segmental level is involved
According to ECD analysis for the MMN response to the most fa- in articulatory-phonetic processing and learning, whereas a sequence
miliar second-language word she, the only source differing significantly level is involved in learning new words by coding the sequences of
between the groups was the right temporal one at auditory cortex or its sounds and syllables (Hickok and Poeppel, 2007). The processing of
vicinity. The strength of the right temporal source was also significantly serial order of sound sequences required for word learning (Majerus
correlated with the MMN amplitude for she. Thus, although differences and Boukebza, 2013) has been proposed to be impaired in dyslexia
in additional sources of activation cannot be excluded, the observed (Szmalec et al., 2011; see, however, Staels and Van den Broeck, 2015).
group differences in MMN may have at least partly been due to the In the light of these and present results, we suggest that second-lan-
weak activation of the right auditory cortex in children with dyslexia guage learning is hampered in dyslexic readers because the sequence-
(for right-hemispheric sources of activity for words, see MacGregor level processing of the dorsal stream is inefficient in dyslexia, affecting
et al., 2012). Although we found no differences in MMNs between the specifically word-form learning. This could be caused by deficient
groups to native monosyllabic words, the weak activation of the right functional connectivity through the arcuate fasciculus and especially its
auditory cortex for second-language words in dyslexia is compatible direct pathway (Hoeft et al., 2011; Boets et al., 2013; Gullick and Booth,
with previous findings showing abnormal processing of speech in the 2015), which has been found to mediate word learning (López-Barroso
right auditory cortex in dyslexia (Abrams et al., 2009; Molinaro et al., et al., 2013).
2016). In line with Goswami's (2011) “temporal sampling” framework, The fact that we found differences between the groups in the acti-
suggesting atypical entrainment at delta and theta rates, poor readers vation of brain representations for second language, but not for native
have been demonstrated to have abnormal asymmetry pattern for syl- language (cf. Froyen et al., 2011), highlights a difference in native-
lable-rate processing (i.e., no right-hemispheric dominance; Abrams language and second-language speech processing. At sublexical level,
et al., 2009) and impaired speech entrainment in the delta band native-language speech sounds activate mostly brain areas involved in
(0.5–1 Hz) which is accompanied by reduced delta synchronization auditory-phonetic processing, whereas second-language speech sounds
between the right auditory cortex and the left inferior frontal cortex have been found to activate also brain areas involved in articulatory-
(Molinaro et al., 2016). The right hemisphere may be specifically in- motor processing in frontal cortex (Callan et al., 2004). Beyond sub-
volved in the processing of foreign words: Nora et al. (2017) have re- lexical level, there is cumulative evidence that native-language pro-
cently reported that children's learning effects were right-lateralized for cessing that aims at comprehension takes place in the ventral stream of
foreign word forms. It is also noteworthy that according to the current speech processing in typical listening conditions (Scott and Johnsrude,
ECD analysis, the right temporal source had the shortest peak latency, 2003; Hickok and Poeppel, 2007; DeWitt and Rauschecker, 2012). In
that is, it was typically activated first for the familiar second-language contrast, reading, naming, discrimination, working memory, and
word, implying that in children the right hemisphere plays an im- second-language learning have been suggested to activate the dorsal
portant role in the recognition of second-language words. stream of speech processing (Hickok and Poeppel, 2007). Interestingly,
In our data, frontal sources were found in very few children with these involve more complex tasks with explicit processing, mental
dyslexia. Together with weak right temporal activation in these chil- manipulations, and phonological working-memory load that often re-
dren, this may point to the same direction as Molinaro's et al. (2016) sult in impaired performance in dyslexia (for a review, see Ramus and
previous finding which suggested that synchronized activity between Ahissar, 2012). Taken together, these findings linking simple native-
the right auditory cortex and the left inferior frontal cortex is reduced in language speech listening predominantly with the ventral stream but
poor readers. However, in the current data the ECD strengths of ob- second-language processing and complex native-language processing
served frontal sources neither seemed to differ between the groups (see with the dorsal stream may explain the more pronounced deficits in
Fig. 4), nor could be statistically tested because frontal ECDs were phonological skills and second-language learning than in native-lan-
found in too few children. Therefore, the hypothesis of deficient acti- guage speech processing in dyslexia.
vation of the left frontal cortex in second-language processing in dys- In school environment, second language is learned not only by lis-
lexia requires further research. tening and speaking, but also from textbooks. Children are exposed to
Besides significant differences in literacy scores that were used to both auditory and orthographic forms of words and it is difficult to
assign the participants into groups, cognitive tests indicated that as disentangle their contribution to learning. Since our word of interest she
compared with controls, the children with dyslexia had a shorter digit is very frequent, the participants must have had a plenty of chances to
span (i.e., poorer phonological short-term memory) and they were learn the word from speech, and therefore it is unlikely that dyslexic
slower in rapid naming. Taken together, as a group, the dyslexic readers readers’ difficulties to encode the orthographic form of the word she
appeared to show typical symptoms of dyslexia that have been asso- could solely account for the current pattern of findings. Nevertheless,
ciated with a phonological deficit, including poor reading, phonological the orthographic form is likely to affect more the learning of less

113
S. Ylinen et al. Neuropsychologia 122 (2019) 105–115

frequent words that are not heard and produced so often. In this case, case for a Matthew effect. J. Speech Lang. Hear. Res. 58 (3), 853–864. https://doi.
the difficulty to learn second-language word forms in dyslexia might be org/10.1044/2015_JSLHR-L-13-0310.
Froyen, D., Willems, G., Blomert, L., 2011. Evidence for a specific cross-modal association
modulated by the orthographic properties of the target language. For deficit in dyslexia: an electrophysiological study of letter-speech sound processing.
example, children with dyslexia might have more difficulties with the Dev. Sci. 14 (4), 635–648. https://doi.org/10.1111/j.1467-7687.2010.01007.x.
orthographic forms and their integration with the auditory word forms Goswami, U., 2011. A temporal sampling framework for developmental dyslexia. Trends
Cogn. Sci. 15 (1), 3–10. https://doi.org/10.1016/j.tics.2010.10.001.
in opaque languages (such as English) than in transparent languages Gullick, M.M., Booth, J.R., 2015. The direct segment of the arcuate fasciculus is predictive
(such as Finnish). These difficulties may be especially pronounced if of longitudinal reading change. Dev. Cogn. Neurosci. 13, 68–74. https://doi.org/10.
native language is transparent and second language is opaque. 1016/j.dcn.2015.05.002.
Hari, R., Renvall, H., 2001. Impaired processing of rapid stimulus sequences in dyslexia.
To the best of our knowledge, this is the first ERP study focusing on Trends Cogn. Sci. 5 (12), 525–532. https://doi.org/10.1016/S1364-6613(00)
the early stages of second-language word processing in dyslexia. We 01801-5.
show that during the early processing stages (< 200 ms from the de- Hickok, G., Poeppel, D., 2007. The cortical organization of speech processing. Nat. Rev.
Neurosci. 8 (5), 393–402. https://doi.org/10.1038/nrn2113.
viance or recognition point), the activation of brain representations for
Hoeft, F., McCandliss, B.D., Black, J.M., Gantman, A., Zakerani, N., Hulme, C., Gabrieli,
familiar spoken second-language words is weaker in children with J.D., 2011. Neural systems predicting long-term outcome in dyslexia. Proc. Natl.
dyslexia than in typical readers. Since source localization suggested Acad. Sci. 108 (1), 361–366. https://doi.org/10.1073/pnas.1008950108.
significant differences between the groups in the right but not in the left Huovilainen, T.M., 2018. Psycholinguistic Descriptives. The Language Bank of Finland.
Retrieved from: http://urn.fi/urn:nbn:fi:lb-2018081601.
hemisphere and the strength of the right temporal source correlated Häyrinen, T., Serenius-Sirve, S., Korkman, M., 1999. LukiLasse. Lukemisen, kirjoittamisen
with the MMN amplitude, weak activation of the right temporal cortex ja laskemisen seulontatesti ala-asteen luokille 1–6. Psykologien Kustannus Oy,
in children with dyslexia may at least partly drive this effect. The Helsinki, Finland.
Kimppa, L., Shtyrov, Y., Partanen, E., Kujala, T., 2018. Impaired neural mechanism for
current data suggest that dyslexia is associated with a deficit in re- online novel word acquisition in dyslexic children. Sci. Rep. 8 (1), 12779. https://doi.
presenting second-language word forms, which may hamper language org/10.1038/s41598-018-31211-0.
learning. This interpretation is supported by correlations observed be- Laasonen, M., Virsu, V., Oinonen, S., Sandbacka, M., Salakari, A., 2012. Phonological and
sensory short-term memory are correlates and both affected in developmental dys-
tween second-language word processing and literacy and naming lexia. Read. Writ. 25 (9), 2247–2273. https://doi.org/10.1007/s11145-011-9356-1.
scores. Although further research is needed, the present results clarify Landi, N., Malins, J.G., Frost, S.J., Magnuson, J.S., Molfese, P., Ryherd, K., et al., 2018.
the links between reading and speech as well as the bottlenecks of Neural representations for newly learned words are modulated by overnight con-
solidation, reading skill, and age. Neuropsychologia 111, 133–144. https://doi.org/
language learning in dyslexia. Since word learning tasks can be con- 10.1016/j.neuropsychologia.2018.01.011.
ducted at younger age than reading tasks, the results showing that Leech, G., Rayson, P., Wilson, A., 2016. Companion website for: Word frequencies in
reading skills are more closely linked with second-language word written and spoken English: based on the British National Corpus [Internet]. Cited
2016 July 1. Available from 〈http://ucrel.lancs.ac.uk/bncfreq/〉.
learning than native-language processing may also have applicability to
Leppänen, P.H., Hämäläinen, J.A., Salminen, H.K., Eklund, K.M., Guttorm, T.K.,
early diagnostics of dyslexia. Lohvansuu, K., Lyytinen, H., 2010. Newborn brain event-related potentials revealing
atypical processing of sound frequency and the subsequent association with later
Acknowledgements literacy skills in children with familial dyslexia. Cortex 46 (10), 1362–1376. https://
doi.org/10.1016/j.cortex.2010.06.003.
Lindeman, J., 1998. Ala-asteen Lukutesti: käyttäjän käsikirja. University of Turku,
The authors thank the participants and their families and schools, Oppimistutkimuksen keskus, Turku, Finland.
Education Departments of cities of Helsinki and Espoo for the possibi- Litt, R.A., Nation, K., 2014. The nature and specificity of paired associate learning deficits
in children with dyslexia. J. Mem. Lang. 71, 71–88.
lity to recruit participants through their schools, the Finnish Diverse López-Barroso, D., Catani, M., Ripollés, P., Dell'Acqua, F., Rodríguez-Fornells, A., de
Learners’ Association for help in recruitment of children with dyslexia, Diego-Balaguer, R., 2013. Word learning is mediated by the left arcuate fasciculus.
Professor Heikki Lyytinen for his advice on inclusion criteria and Proc. Natl. Acad. Sci. 110 (32), 13168–13173. https://doi.org/10.1073/pnas.
1301696110.
comments on an earlier version of the manuscript, Tatu Huovilainen for Lyytinen, P., Eklund, K., Lyytinen, H., 2005. Language development and literacy skills in
the Finnish word frequency data, and Miika Leminen and Tommi late-talking toddlers with and without familial risk for dyslexia. Ann. Dyslexia 55 (2),
Makkonen for technical assistance. This work was supported by the 166–192. https://doi.org/10.1007/s11881-005-0010-y.
MacGregor, L.J., Pulvermuller, F., van Casteren, M., Shtyrov, Y., 2012. Ultra-rapid access
Academy of Finland, Finland (grant numbers 131963, 274058, and
to words in the brain. Nat. Commun. 3, 711. https://doi.org/10.1038/ncomms1715.
276414) and Jane and Aatos Erkko Foundation, Finland. Majerus, S., Boukebza, C., 2013. Short-term memory for serial order supports vocabulary
development: new evidence from a novel word learning paradigm. J. Exp. Child
Psychol. 116 (4), 811–828. https://doi.org/10.1016/j.jecp.2013.07.014.
References
Majerus, S., Cowan, N., 2016. The nature of verbal short-term impairment in dyslexia: the
importance of serial order. Front. Psychol. 7. https://doi.org/10.3389/fpsyg.2016.
Abrams, D.A., Nicol, T., Zecker, S., Kraus, N., 2009. Abnormal cortical processing of the 01522. (article number 1522).
syllable rate of speech in poor readers. J. Neurosci. 29 (24), 7686–7693. https://doi. McCrory, E.J., Mechelli, A., Frith, U., Price, C.J., 2005. More than words: a common
org/10.1523/JNEUROSCI.5242-08.2009. neural basis for reading and naming deficits in developmental dyslexia? Brain 128
Ahonen, T., Tuovinen, S., Leppäsaari, T., 2003. Nopean Sarjallisen Nimeämisen Testi (2), 261–267. https://doi.org/10.1093/brain/awh340.
[Rapid Alternating Naming Test]. Niilo Mäki Instituutti & Haukkarannan koulu, Molinaro, N., Lizarazu, M., Lallier, M., Bourguignon, M., Carreiras, M., 2016. Out-of-
Jyväskylä, Finland. synchrony speech entrainment in developmental dyslexia. Hum. Brain Mapp. 37 (8),
Ben-Yehudah, G., Banai, K., Ahissar, M., 2004. Patterns of deficit in auditory temporal 2767–2783. https://doi.org/10.1002/hbm.23206.
processing among dyslexic adults. Neuroreport 15 (4), 627–631. https://doi.org/10. Myers, C.A., Vandermosten, M., Farris, E.A., Hancock, R., Gimenez, P., Black, J.M.,
1097/00001756-200403220-00011. Hulme, C., 2014. White matter morphometric changes uniquely predict children's
Boersma, P., Weenink, D., 2010. Praat: doing phonetics by computer. Version 5.1.31 reading acquisition. Psychol. Sci. 25 (10), 1870–1883. https://doi.org/10.1177/
[software]. Downloaded April 4. Available from: 〈http://www.praat.org/〉. 0956797614544511.
Boets, B., de Beeck, H.P.O., Vandermosten, M., Scott, S.K., Gillebert, C.R., Mantini, D., Myers, E.B., 2014. Emergence of category-level sensitivities in non-native speech sound
Ghesquière, P., 2013. Intact but less accessible phonetic representations in adults learning. Front. Neurosci. 8, 238. https://doi.org/10.3389/fnins.2014.00238.
with dyslexia. Science 342 (6163), 1251–1254. https://doi.org/10.1126/science. Nora, A., Karvonen, L., Renvall, H., Parviainen, T., Kim, J.Y., Service, E., Salmelin, R.,
1244333. 2017. Children show right-lateralized effects of spoken word-form learning. PLoS One
Callan, D.E., Jones, J.A., Callan, A.M., Akahane-Yamada, R., 2004. Phonetic perceptual 12 (2), e0171034. https://doi.org/10.1371/journal.pone.0171034.
identification by native-and second-language speakers differentially activates brain Näätänen, R., Gaillard, A.W., Mäntysalo, S., 1978. Early selective-attention effect on
regions involved with acoustic phonetic processing and those involved with articu- evoked potential reinterpreted. Acta Psychol. 42 (4), 313–329.
latory–auditory/orosensory internal models. NeuroImage 22 (3), 1182–1194. Näätänen, R., Lehtokoski, A., Lennes, M., Cheour, M., Huotilainen, M., Iivonen, A.,
https://doi.org/10.1016/j.neuroimage.2004.03.006. Vainio, M., Alku, P., Ilmoniemi, R.J., Luuk, A., Allik, J., Sinkkonen, J., Alho, K., 1997.
DeWitt, I., Rauschecker, J.P., 2012. Phoneme and word recognition in the auditory Language-specific phoneme representations revealed by electric and magnetic brain
ventral stream. Proc. Natl. Acad. Sci. 109 (8), E505–E514. https://doi.org/10.1073/ responses. Nature 385, 432–434.
pnas.1113427109. Näätänen, R., 2001. The perception of speech sounds by the human brain as reflected by
Di Betta, A.M., Romani, C., 2006. Lexical learning and dysgraphia in a group of adults the mismatch negativity (MMN) and its magnetic equivalent (MMNm).
with developmental dyslexia. Cogn. Neuropsychol. 23 (3), 376–400. https://doi.org/ Psychophysiology 38 (1), 1–21. https://doi.org/10.1111/1469-8986.3810001.
10.1080/02643290442000545. Näätänen, R., Paavilainen, P., Rinne, T., Alho, K., 2007. The mismatch negativity (MMN)
Duff, D., Tomblin, J.B., Catts, H., 2015. The influence of reading on vocabulary growth: a in basic research of central auditory processing: a review. Clin. Neurophysiol. 118

114
S. Ylinen et al. Neuropsychologia 122 (2019) 105–115

(12), 2544–2590. https://doi.org/10.1016/j.clinph.2007.04.026. Scott, S.K., Johnsrude, I.S., 2003. The neuroanatomical and functional organization of
Opitz, B., Rinne, T., Mecklinger, A., Von Cramon, D.Y., Schröger, E., 2002. Differential speech perception. Trends Neurosci. 26, 100–107. https://doi.org/10.1016/S0166-
contribution of frontal and temporal cortices to auditory change detection: fMRI and 2236(02)00037-1.
ERP results. Neuroimage 15 (1), 167–174. https://doi.org/10.1006/nimg.2001.0970. Shtyrov, Y., Kimppa, L., Pulvermuller, F., Kujala, T., 2011. Event-related potentials re-
Paulesu, E., Frith, U., Snowling, M., Gallagher, A., Morton, J., Frackowiak, R.S., Frith, flecting the frequency of unattended spoken words: a neuronal index of connection
C.D., 1996. Is developmental dyslexia a disconnection syndrome? Brain 119 (1), strength in lexical memory circuits? Neuroimage 55 (2), 658–668. https://doi.org/
143–157. 10.1016/j.neuroimage.2010.12.002.
Price, C.J., McCrory, E., Noppeney, U., Mechelli, A., Moore, C.J., Biggio, N., Devlin, J.T., Smith, F.R.H., Gaskell, M.G., Weighall, A.R., Warmington, M., Reid, A.M., Henderson,
2006. How reading differs from object naming at the neuronal level. Neuroimage 29 L.M., 2018. Consolidation of vocabulary is associated with sleep in typically devel-
(2), 643–648. https://doi.org/10.1016/j.neuroimage.2005.07.044. oping children, but not in children with dyslexia. Dev. Sci. 21 (5), e12639. https://
Pulvermüller, F., Kujala, T., Shtyrov, Y., Simola, J., Tiitinen, H., Alku, P., Näätänen, R., doi.org/10.1111/desc.12639.
2001. Memory traces for words as revealed by the mismatch negativity. Neuroimage Snowling, M.J., 2000. Dyslexia. Blackwell, Oxford, UK.
14 (3), 607–616. https://doi.org/10.1006/nimg.2001.0864. Soroli, E., Szenkovits, G., Ramus, F., 2010. Exploring dyslexics' phonological deficit III:
Pulvermüller, F., Shtyrov, Y., 2006. Language outside the focus of attention: the mismatch foreign speech perception and production. Dyslexia 16 (4), 318–340. https://doi.org/
negativity as a tool for studying higher cognitive processes. Prog. Neurobiol. 79 (1), 10.1002/dys.415.
49–71. https://doi.org/10.1016/j.pneurobio.2006.04.004. Staels, E., Van den Broeck, W., 2015. No solid empirical evidence for the solid (serial
Quinn, J.M., Wagner, R.K., 2015. Gender differences in reading impairment and in the order learning impairment) hypothesis of dyslexia. J. Exp. Psychol.: Learn. Mem.
identification of impaired readers: results from a large-scale study of at-risk readers. Cogn. 41 (3), 650. https://doi.org/10.1037/xlm0000054.
J. Learn. Disabil. 48 (4), 433–445. https://doi.org/10.1177/0022219413508323. Szmalec, A., Loncke, M., Page, M., Duyck, W., 2011. Order or disorder? Impaired Hebb
Ramus, F., Ahissar, M., 2012. Developmental dyslexia: the difficulties of interpreting poor learning in dyslexia. J. Exp. Psychol.: Learn. Mem. Cogn. 37 (5), 1270. https://doi.
performance, and the importance of normal performance. Cogn. Neuropsychol. 29 org/10.1037/a0023820.
(1–2), 104–122. https://doi.org/10.1080/02643294.2012.677420. Vellutino, F.R., Fletcher, J.M., Snowling, M.J., Scanlon, D.M., 2004. Specific reading
Ramus, F., Rosen, S., Dakin, S.C., Day, B.L., Castellote, J.M., White, S., Frith, U., 2003. disability (dyslexia): what have we learned in the past four decades? J. Child Psychol.
Theories of developmental dyslexia: insights from a multiple case study of dyslexic Psychiatry 45 (1), 2–40. https://doi.org/10.1046/j.0021-9630.2003.00305.x.
adults. Brain 126 (4), 841–865. https://doi.org/10.1093/brain/awg076. Wagner, R.K., Torgesen, J.K., 1987. The nature of phonological processing and its causal
Ramus, F., Szenkovits, G., 2008. What phonological deficit? Q. J. Exp. Psychol. 61 (1), role in the acquisition of reading skills. Psychol. Bull. 101 (2), 192. https://doi.org/
129–141. https://doi.org/10.1080/17470210701508822. 10.1037/0033-2909.101.2.192.
Romani, C., Tsouknida, E., Olson, A., 2015. Encoding order and developmental dyslexia: a Wechsler, D., 2010. WISC-IV – Wechsler Intelligence Scale for Children – IV. Psykologien
family of skills predicting different orthographic components. Q. J. Exp. Psychol. 68 Kustannus Oy, Helsinki, Finland.
(1), 99–128. https://doi.org/10.1080/17470218.2014.938666. Wiik, K., 1965. Finnish and English Vowels. University of Turku, Turku, Finland.
Saur, D., Kreher, B.W., Schnell, S., Kümmerer, D., Kellmeyer, P., Vry, M.S., Huber, W., Winkler, I., 2007. Interpreting the mismatch negativity. J. Psychophysiol. 21 (3–4),
2008. Ventral and dorsal pathways for language. Proc. Natl. Acad. Sci. 105 (46), 147–163. https://doi.org/10.1027/0269-8803.21.34.147.
18035–18040. https://doi.org/10.1073/pnas.0805234105. Yeatman, J.D., Dougherty, R.F., Ben-Shachar, M., Wandell, B.A., 2012. Development of
Schwartz, J., Tallal, P., 1980. Rate of acoustic change may underlie hemispheric spe- white matter and reading skills. Proc. Natl. Acad. Sci. 109 (44), E3045–E3053.
cialization for speech perception. Science 207 (4437), 1380–1381. https://doi.org/ https://doi.org/10.1073/pnas.1206792109.
10.1126/science.7355297.

115