Electrophysiological correlates of syntactic processing in developmental dyslexia: preliminary results
Chiara Cantiani¹², Maria Teresa Guasti¹, Maria Luisa Lorusso², Paolo Perego²
c.cantiani@campus.unimib.it
¹ University of Milano - Bicocca ² Scientific Institute “E. Medea”, Bosisio Parini, Lecco
INTRODUCTION
In this study, the syntactic processing of subject-verb agreement violations was investigated through event-related potentials (ERPs) in 10 adults with developmental dyslexia (aged 20-28 years) and 10 unimpaired controls matched on age and sex. The rationale for this study stems from the frequent observed overlap between specific language impairment and developmental dyslexia (e.g. Bishop & Snowling, 2004). Several studies pointed out vocabulary and morphosyntactic difficulties in pre-school children at genetic risk of dyslexia and in dyslexic children (van Alphen et al., 2004; McArthur et al., 2000). These difficulties often resolve over time. However, as Scarborough (1990) noted, this recovery may be an artefact of the way language skill are assessed, using insufficiently sensitive measures. According to this line of research, anomalies in language processing are expected in dyslexic subjects, even if linguistic difficulties did not emerge from standardized test of language comprehension. ERP may be useful to reveal subtle differences between groups in the processing of linguistic information. In previous studies with unimpaired populations, subject-verb agreement violations were associated with two ERP components: a left anterior negativity (LAN) indicating the detection of the morphosyntactic error, and a late, centro-parietal positivity (P600) reflecting processes of reanalysis (in Italian, De Vincenzi et al., 2003).
RESULTS
BEHAVIOURAL TASK: No differences were found in the behavioural task, considering both accuracy and latency (ps>.2)
CONTROL SUBJECTS
F7 F3 FZ F4 F8 F7 F3
DYSLEXIC SUBJECTS
FZ F4 F8
C3 T3
CZ
C4 T3 T3
C3
CZ
C4 T4
P3
PZ
P4
P3
PZ
P4
01
02
01
02
METHOD
SUBJECTS
Two groups were considered: a group of 10 Italian adults with Developmental Dyslexia (diagnosis made on the DSM IV criteria) (M=22,6 years old) and a control group of 10 control subjects matched for chronological age and sex. Only dyslexic subjects without any language disorders, assessed by adequate scores on the Bilingual Aphasia Test (BAT) (Paradis, 1987), were included.
P600 ANALYSES:
STATISTICAL ANALYSES: repeated measures analysis of variance for mean amplitude values Within-subject factors: Grammaticality (correct vs incorrect); Laterality (left, central, right); Longitude (frontal, central, parietal). Between-subject factor: Group (dyslexic vs control subjects).
T Ws WHOLE SAMPLE 450 900 CONTROL SUBJECTS DYSLEXIC SUBJECTS EFFECT grammaticality gramm.*caudality grammaticality grammaticality grammaticality grammaticality F 6,593 16,279 5,353 2,154 5,830 0,575 p-value .016 <.001 .046 >.176 .039 >.4
LAN ANALYSES:
STATISTICAL ANALYSES: repeated measures analysis of variance for mean amplitude values Within-subject factors: Grammaticality (correct vs incorrect); Laterality (left, central, right); Longitude (frontal, central). Between-subject factor: Group (dyslexic vs control subjects). Time window: 50 – 350 ms
EFFECT grammaticality WHOLE SAMPLE gramm.*laterality gramm.*caudality* laterality grammaticality grammaticality F 6,263 3,450 3,362 1,545 4,892 p-value .022 .055 .032 >.2 .054
PZ
NUMBER MANIPULATION
CONTROL SUBJECTS DYSLEXIC SUBJECTS
PZ
TASK
Event-related potentials were recorded while subjects listened to 168 sentences with and without subject-verb agreement violations (42 sentences for each experimental condition). As shown in table 1, two variables were manipulated: grammaticality (correct or incorrect) and subject number (singular or plural). In addition, subjects listened to 32 fillers with different syntactic structure and type of violation. A behavioural task was also performed, requiring a grammaticality judgement on every sentence.
STIMULI
Grammaticality Correct Incorrect NP subject number Singular
La bimba bionda (S) gioca (S) con la palla the blond child (S) plays (S) with the ball *La bimba bionda (S) giocano (P) con la palla *the blond child (S) play (P) with the ball
Plural
Le bimbe bionde (P) giocano (P) con la palla the blond children (P) play (P) with the ball *Le bimbe bionde (P) gioca (S) con la palla *the blond children (P) plays (S) with the ball
650 900
CONTROL SUBJECTS DYSLEXIC SUBJECTS
CONTROL SUBJECTS DYSLEXIC SUBJECTS
STATISTICAL ANALYSES: repeated measures analysis of variance for mean amplitude values EFFECT:Number *Grammaticality Within-subject factors: Number (singular vs F p-value plural); Grammaticality (correct vs incorrect); WHOLE 5,271 .034 Laterality (left, central, right); Caudality SAMPLE (central, parietal). CONTROL 1,424 >.2 Between-subject factor: Group (dyslexic vs SUBJECTS control subjects). DYSLEXIC 4,198 .071 Time window: 650 – 900 ms SUBJECTS
All the sentences were recorded by two Italian native female speakers. Sentences were balanced for frequency, concreteness and length. A plausibility pre-test was performed, in order to value the semantic plausibility of the sentences.
DISCUSSION
In line with previous research, agreement violations evoked in the whole sample a negative wave in the 50-350 ms time-window (LAN), even if not completely left lateralized, and a broad positive wave (P600) between 450-950 ms. However, further analysis performed separately for control and dyslexic subjects showed subtle differences between the two groups. These results support the evidence of two functionally distinct phases involving syntactic processing, revealed by two different ERP components (Friederici, 2002). According to previous results (Rispens et al., 2006), control subjects showed a broader P600 component with respect to dyslexic subjects. The difference is particularly strong in the second part of the wave (in the 650-900 ms time window), reflecting the reanalysis process and the attempt of repairing the violation. On the contrary, dyslexic subjects seem to present a bigger LAN, reflecting the detection of the morphosyntactic error. This latter result could be an artefact due to the large variability we observed among subjects. In fact, previous evidence obtained with different types of syntactic violations in dyslexic children (Sabisch et al., 2006) or G-SLI children (Van der Lely & Fontaneau, 2008) was at odd with this finding. We expect this difference to disappear by enlarging the sample, while P600 differences should be reinforced. Unexpected results emerged from the number manipulation: in general sentences with singular and plural subject NPs seem to be processed in different manners. This trend is stronger in the dyslexic group. Similar results were obtained by Rispens et al. (2006), and could be explained by the bigger complexity in terms of semantic and discourse operations of plural sentences (Kaan, 2002). These processing demands of plural NPs could affect the dyslexic group more than the control group. These results might support the hypothesis of different language processing modalities in pure dyslexic subjects whose language comprehension system was supposed to be intact. A comparison with dyslexic subjects with previous language impairments might further highlight these findings.
REFERENCES van Alphen, P., de Bree, E., Gerrits, E., de Jong, J., Wilsenach, C., Wijnen, F. (2004) “Early language development in children with a genetic risk for dyslexia”, Dyslexia, 10: 1-24. Bishop D.V.M., Snowling M. (2004), “Development dyslexia and specific language impairment: same or different?”, Psychological Bulletin, 6: 858-886; De Vincenzi M., Job R., Di Matteo R., Angrilli A., Pelonazzi A., Ciccarelli L., Vespignani F. (2003), “Differences in the perception of time corse of syntactic and semantic violations”, Brain and Languages, 85: 280-296.2003; Fontaneau E. & Van der Lely H.K.J (2008), “Electrical brain responces in language-impaired children revealed grammar specific deficits”, Plos one, 3; Friederici A.D. (2002), “Toward a neuronal basis of auditory sentence processing”, TRENDS in Cognitive Sciences, 6: 78-84; Kaan, (2002), “Investigating the effects of distence and number interference in processing subject-verb dependencies: an ERP studies”, Journal of Psycholinguistic research, 31: 165-193; McArthur, G.M., Hogben, J.H., Edwards, V.T., Health, S.M. Mengler, E.D. (2000), “On the “specifics” of specific reading disability and specific language impairment”, Journal of Child Psychology and Psychiatry, 41, 869-874. Paradis, M. (1987), “The assessment of bilingual aphasia”, Hillsdale, NJ: Erlbaum; Rispens J.E., Been P.H., Zwarts F. (2006), “Brain responces to subject-verb agreement violations in spoken language in developmental dyslexia: an ERP study”, Dyslexia, 12: 134-149; Sabisch B., Hahne A., Glass E., von Suchodoletz W., Friederici A.D. (2006), “Auditory language comprehension in children with developmental dyslexia: evidence from evented-related potentials”, Journal of cognitive neuroscience, 18(10): 1676-1695.
EXPERIMENTAL PARADIGM
EEG RECORDING
Brain electrical activity was continuously recorded with a SynAmps amplifier (Neuroscan) from 15 Ag/Agcl electrodes placed according to the 10– 20 system. The electro-oculogram (EOG) was monitored by electrodes placed below and lateral to the left eye. Electrode impedances were kept below 5 kΩ. The EEG was amplified, digitized (1 kHz per channel), pre-filtered (band pass, 0.01–40Hz), and digitally stored in a PC for off-line analysis. Off-line low-pass digital filtering 15Hz (zero-phase filter) was applied to the averaged data. Epochs of 100 ms pre-stimulus and 1200 ms post stimuli from the second trigger placed on the verbs (at the confirmation or at the violation of the rule) were separately averaged for the standard and for deviant stimuli. The mean voltage of the 100 ms pre-stimulus period served as a baseline for amplitude measurement. During the averaging procedure, epochs where the voltage rise up 60 µV from the baseline were eliminated, classifying them as contaminated by EOG artefacts.
Subjects were comfortably seated in a chair in a dimly lit and sound attenuated chamber. The auditory stimuli were presented via headphones, using STIM2 software.
MANY THANKS TO EVERYONE WHO VOLUNTARILY PARTECIPATED IN THIS STUDY!!