Behavioural Brain Research 209 (2010) 179–182
Contents lists available at ScienceDirect
Behavioural Brain Research journal homepage: www.elsevier.com/locate/bbr
Short communication
Brain plasticity in developmental dyslexia after phonological treatment: A beta EEG band study Barbara Penolazzi a,b,1 , Chiara Spironelli b,2 , Claudio Vio c , Alessandro Angrilli b,d,∗ a
Department of Biomedical Sciences, University “G. d’Annunzio” of Chieti, Blocco A - Via Dei Vestini I-66013 Chieti, Italy Department of General Psychology, University of Padova, Via Venezia 8, 35131 Padova, Italy c Children’s Neuropsychiatric Medical Facility of San Donà di Piave, Venice, Italy d CNR Institute of Neuroscience, Padova, Italy b
a r t i c l e
i n f o
Article history: Received 15 December 2009 Received in revised form 18 January 2010 Accepted 20 January 2010 Available online 28 January 2010 Keywords: Dyslexia Training Beta band EEG Phonology Semantics Lateralization Cortical reorganization
a b s t r a c t Linguistic EEG hemispheric reorganization was investigated in 14 dyslexic children after a 6-month phonological training (10 min/day through PC software). Error rates from three linguistic tasks significantly decreased and reading speed improved after the training. A significant positive correlation (r12 = 0.536) was found at posterior sites for the phonological task only, showing that those children who had the greatest reading speed enhancement showed the largest left posterior EEG beta power increase. © 2010 Elsevier B.V. All rights reserved.
Developmental Dyslexia (DD) is a phenotypically heterogeneous clinic syndrome consisting in a pronounced and persistent difficulty in reading acquisition, despite normal intelligence, sensory acuity, motivation, and educational opportunities [21]. Although experimental evidence confirmed its genetic aetiology [3], individuals suffering from this disorder differ in their individual profiles for many characteristics [4,22] also depending on environmental factors. Among these, the intrinsic structure of the language to which the individual is exposed is decisive for the risk of developing the disorder and for the extent of the disability. Indeed, DD occurrence is much higher in deep orthographies (i.e., English) than in shallow ones (i.e., Italian), as the former involve more irregular grapheme–phoneme correspondences which make word decoding mechanism rather difficult [8]. DD is marked by incorrect and non-fluent written language decoding. In irregular orthographies, reading accuracy remains the greatest problem. Conversely, in regular orthographies, it can improve until it reaches a ceiling level
∗ Corresponding author at: Department of General Psychology, Via Venezia 8, 35131 Padova, Italy. Tel.: +39 049 8276692; fax: +39 049 8276600. E-mail address:
[email protected] (A. Angrilli). 1 Tel.: +39 0871 3554206; fax: +39 0871 3554163/049 8276600. 2 Tel.: +39 049 8276635/6914; fax: +39 049 8276600. 0166-4328/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.bbr.2010.01.029
[19], whereas reading speed in dyslexics follows a linear trend across years with an incremental speed approximately half-sized with respect to normal condition (i.e., in 7–13 year-old Italian children the mean increase per year is 0.3 vs. 0.5 syllable/s for impairedvs. normal-readers [19]). Since DD is often characterized by multiple interacting altered mechanisms, many theories on its nature still coexist, nevertheless, the strongest evidence converges in identifying the phonological deficit (i.e., a deficit in the representation and/or manipulation of the smallest speech sounds, the phonemes) as the core dysfunction of reading disabilities [9–11,13,14]. A number of functional brain imaging studies has localized this altered phonological processing in a large cortical network distributed mainly in the left hemisphere, and including the perisylvian and temporo-parieto-occipital areas [13]. In particular, past studies on DD pointed to an impairment of left posterior brain systems involved in the cross-modal integration of auditory and visual information which include the connections between occipitotemporal and parietotempotal circuits [13]. When performing phonological tasks, these posterior systems often exhibit reduced or absent activation in impaired compared with normal-readers. Similarly, also dyslexics’ left frontal system often revealed an altered languagerelated activity, likely secondary to the disruption of posterior linguistic networks which are selectively activated in the early phases of word reading [9,10,13,14]. An increasing body of data sug-
180
B. Penolazzi et al. / Behavioural Brain Research 209 (2010) 179–182
gested that the severity of reading difficulties can be diminished by different intensive trainings, aimed at improving reading speed and accuracy by stimulating cortical plastic reorganization [13,17,18]. Specifically, during developmental age, training-induced reading enhancement has been associated with increased activation of both left temporo-parietal and inferior frontal areas—usually involved in the phonological processing—and of several right hemisphere areas, which, after the achievement of automatic normal reading, are no more necessary for graphemic decoding [17]. However, most of research focussed on English orthographies and therefore on reading accuracy, whereas the development and improvement of fluent reading and its underlying neural systems still need to be clarified [13]. In the present study we aimed at investigating the improvement of reading fluency after a 6-month linguistic intervention based on phonological awareness, by using beta power band as a neurophysiological marker of cortical cognitive activation [15]. Evidence of dysfunctional beta band alterations in DD has been recently demonstrated in several studies [5,7], and, using the same well-validated linguistic paradigm of the present study a sample of impaired-readers showed, differently from controls, a dysfunctional anterior right lateralization in all linguistic tasks, and left posterior lateralization during both phonological and orthographic tasks [14]. The altered recruitment of left posterior regions, specialized in the encoding and integration of the phonological components of words, was interpreted as due to dysfunctional mechanisms of DD. However, without data on training-induced plastic changes, we could not resolve if dyslexics’ beta activity patterns reflected a primary impairment or a cortical reorganization aimed at compensating their deficit. With the present research we specifically addressed this issue. In addition, to prove the efficacy of phonological intervention, we expected to find, together with reading speed improvement, a hemispheric reorganization of beta activity in posterior sites and selectively to the phonological task. Specifically, according to Bakker’s theory on dyslexia as depending on an imbalance in the contribution of hemispheres to reading [1], we expected the EEG beta activity reorganization in the left hemisphere to be associated with greater reading speed enhancement. After parents’ informed consent signature according to the Declaration of Helsinki, 14 native Italian-speaking children (11 boys; mean age: 118.6 months ± 21.6) participated in the study. Children were recruited from the Children’s Neuropsychiatric Medical Facility of San Dona’ di Piave: they all had a diagnosis of phonological dyslexia, assessed by means of the standard tests for reading evaluation [2]. Additional inclusion criteria were normal IQ [20] and no attention deficit disorder comorbidity. They had an average handedness of 92.9%, and normal or corrected-to normal vision. Dyslexics participated in the experimental session before and after a 6-months phonological training, occurring at home, 5 times a week, for 10 min/day. The training was carried out by means of a standardized rehabilitative software, the WinABC program (5.0 version) based on the improvement of phonological awareness through timed passage reading (additional information is available on-line at: http://www.impararegiocando.it/WinABC50.htm). In each experimental session carried out before and after the training phase, children were asked to judge (by providing a response with the left hand) visually presented word pairs on the basis of different linguistic criteria (each corresponding to an experimental task): the type similarity in the control orthographic task, the rhyme relationship in the phonological task and the semantic classification in the semantic task. The same set of 80 words was used as first stimulus of each pair throughout the tasks, to exclude interfering effects related to word uncontrolled features. Stimulus pairs were presented word by word, the first word (W1) lasting 1500 ms, the second (W2) until the response (but no longer than 5000 ms), with a 2000 ms interstimulus interval (ISI), and a 3000 ms intertrial interval. In both pre- and post-training sessions, passage
reading speed (measured in syllable/s) was assessed and correct response times (RTs), error rates (ERs) and EEG data were collected during task execution. One-tailed t-test comparisons were used to contrast the pre- and post-training reading speed measures, whereas for both RTs and ERs an analysis of variance (ANOVA) was performed, with Session (pre-training vs. post-training) and Task (orthographic vs. phonological vs. semantic) as within group factors. Newman-Keuls post hoc tests further specified significant effects and when necessary the Greenhouse–Geisser correction was applied (and corrected probabilities were reported). EEG was continuously recorded in DC mode from 38 sites: 31 placed according to the International 10–20 system, and the remaining positioned around each eye (Io1, Io2, F9, F10), on the mastoids (M1, M2) and on the nasion. Impedance was kept below 5 k and EEG was amplified with a SynAmp system (NeuroScan Labs, Sterling, USA), using DC-100 Hz bandpass filter, and sampling rate of 500 Hz. Analyses, performed through Brain Vision Analyzer (Brain Products GmbH, Germany), started with eye movement and blink corrections, by applying the Independent Component Analysis correction method. Data were segmented with respect to W1 onset, then artifact rejection was performed (maximum increment voltage step: 150 V; minimum/maximum amplitude: ±75 V), leading to a percentage of rejected trials (34.75%) not statistically different between sessions. Each segment was divided into four 1024-ms time intervals representing different processing phases required by the task: 1024 ms before W1 onset (baseline interval); 1024 ms after W1 onset (W1 interval); 1500 ms to 2524 ms after W1 onset (initial ISI: iISI); and 2476 ms to 3500 ms after W1 onset (terminal ISI: tISI). Whereas W1 interval was clearly related to word reading, iISI mainly referred to stimulus encoding in the verbal working memory, and tISI concerned the late processing of W1 features to be compared with those of W2 [12]. The FFT was performed using a Hamming window and including 512 samples (0.98 Hz resolution). Epochs were averaged within each task for both pretraining and post-training sessions. Beta band power (∼13–20 Hz) was normalized for all locations by computing its percentage in the 0–100 Hz spectral range, in order to measure its relative contribution with respect to the whole spectrum. Electrodes were clustered into 4 regions of interest: anterior left (Fp1, F7, F3, He1, Ve1), anterior right (Fp2, F8, F4, He2, Ve2), posterior left (T7, P3, P7, O1, M1), and posterior right (T8, P4, P8, O2, M2). In order to test whether there was a hemispheric asymmetry depending on the training, we performed a preliminary ANOVA with Session (prevs. post-training), Task (orthographic vs. phonological vs. semantic), Caudality (anterior vs. posterior electrodes) and Lateralization (left vs. right electrodes) as within group factors. Then, to evaluate the relative contribution of one hemisphere with respect to the other, at both anterior and posterior clusters, we computed the hemispheric lateralization index as ratio between the difference of left and right clusters and the sum of left and right clusters (i.e., left − right/left + right). This universally accepted index offers the possibility of normalizing a cluster with respect to the activity of both left and right sites, and its value ranges from +1 to −1, corresponding to 100% left or right lateralization, respectively. In order to test our hypothesis, for each task and interval, Pearson’s correlation analyses were performed by including the differences between the post- and the pre-training values for the two correlated variables: the reading speed measure and the beta laterality index (computed separately for anterior and posterior locations). Thus, a positive correlation indicates that those children who had greater reading skills improvement, had also an increase of left beta activity in the post-training session. In correlational analyses it is advised that outliers (one in our sample, detected because the corresponding value was more than 2.5 standard deviations below the mean reading improvement of the sample) are excluded to avoid the influence of extreme values, but in small samples this is rarely
B. Penolazzi et al. / Behavioural Brain Research 209 (2010) 179–182
181
Fig. 1. Positive Pearson’s correlations between reading speed improvement and beta band laterality index in posterior sites during the Phonological task for (A) iISI and (B) tISI time windows. Confidence intervals and regression lines are computed by including the outlier (black circles).
accomplished, for this reason we decided to present correlations with and without the outlier [16]. Analysis revealed that linguistic training was effective, since t-test comparisons showed a significant improvement of reading speed (t(13) = −5.48, p < 0.001), with a mean increase of 0.37 syllable/s (pre-training:1.18 ± 0.56 SD; post-training: 1.55 ± 0.59 SD). Concerning RTs, ANOVA revealed a significant effect of Task (F(2,26) = 25.68, p < 0.001), RTs being slower for the semantic with respect to both phonological (p < 0.01) and orthographic tasks (p < 0.001), and for the phonological compared with the orthographic one (p < 0.001). Analysis of Error Rates (ERs) revealed a similar significant main effect of Task (F(2,26) = 7.57, p < 0.01), with the semantic task inducing higher percentages of errors (24.87%) compared with phonological (16.61%, p < 0.01) and orthographic (15.27%, p < 0.01) tasks. Interestingly, the factor Session was also significant (F(1,13) = 5.12, p < 0.05), ERs decreased in post- (17.2%) compared with pre-training session (20.63%). As regard to EEG data, ANOVA revealed a significant interaction between the factors Training, Caudality and Laterality (F1,13 = 8.2, p < 0.05). Post hoc tests showed in the pre-training session dyslexic children having bilateral distribution of beta band activity over anterior regions, whereas over posterior areas they had a higher level of beta in the right than in the left sites (p < 0.05). Instead, in the post-training session, beta level increased in the right anterior sites with respect to the left ones (p < 0.05), whereas, over posterior areas, beta percentage increased significantly over left hemisphere with respect to the pre-training session (p < 0.01). By moving onto correlational analyses, positive correlations were found between reading speed and beta laterality index (computed as post- minus pre-training changes) in posterior areas, exclusively for the phonological task. In detail, including the outlier (n = 14), positive correlations were found at posterior sites in both iISI (r12 = 0.615, p < 0.05, Fig. 1A) and tISI (r12 = 0.536, p < 0.05, Fig. 1B). By discarding the outlier (n = 13), the correlation in iISI did not reach the significance (r11 = 0.273, p = n.s.), but that in posterior areas during tISI remained significant (r11 = 0.584, p < 0.05). Hence, even without the outlier, correlations revealed that reading improvement following the linguistic training was significantly associated with greater left beta increments in posterior sites only in the last interval of the phonological task. The correlations for the semantic and orthographic tasks were all low and not significant (r12 ≤ 0.329 and r12 ≤ 0.383, respectively).1
1 In order to avoid the problem of biased p values, due to the fact that the actual underlying distribution of our small sample might depart from normality, we also performed the not-parametrical Spearman’s rank correlational analyses. Results of
An important result of the present research was the proven efficacy of the linguistic training aimed at improving the phonological awareness. Indeed, in addition to a post-training overall decrease of ERs during task execution, the passage reading speed of our sample showed a 6-months mean increase of 0.37 syllable/s, corresponding to an yearly improvement of 0.74 syllable/s, which, compared with normative data from Italian children [19], is higher than normal-readers’ mean annual spontaneous improvement (i.e., 0.5 syllable/s) and more than twice with respect to impaired-readers’ mean annual improvement (i.e., 0.3 syllable/s). This demonstrates that reading amelioration cannot be merely ascribed to maturational factors. In particular, unlike other functional studies which correlated the neural reorganization with variables indirectly associated to reading speed (like oral language skills [17]), we found a positive correlation which directly links the training-induced hemispheric reorganization and the reading performance. This result was mainly due to a significant increase of beta percentage of left posterior sites after the training rather than to a right posterior contribution (or a combination of left and right sites contribution) as further specified by ANOVA. Since the phonological training was directly associated to the fluency improvement, and, at the same time, beta plastic changes were related to this reading enhancement only for the phonological task (i.e., improvements did not correlate with beta lateralization in all tasks, but only in the phonological one, during the last interval tISI which required the typical phonemic manipulation needed for rhyming judgement), the present experiment provides further support to the key role of the phonological awareness in determining a fluent reading. Results point on the effectiveness of phonological interventions in DD, but also on the validity of the phonological deficit theory in accounting this learning disability [11]. In addition, unlike a past electrophysiological study on DD treatment which did not investigate the reorganized linguistic regions [6], we succeeded in finding beta plastic reorganization specifically in left posterior areas. The present results further proved that traininginduced reading automaticity involves neural systems close to those recruited in normal-readers, and therefore, pointed out normalizing, rather than compensatory, effects of remediation [13]. At the same time, we can now re-interpret past results, achieved with the same paradigm on a different sample of dyslexic children, which
Spearman’s rank correlations nearly replicated Pearson’s results. Indeed, positive correlations between reading improvement and beta laterality index were found, in posterior sites, exclusively for the phonological task, in both W1 (r12 = 0.55, p < 0.05, Fig. 1A) and tISI (r12 = 0.63, p < 0.05). Therefore, this further analysis confirmed the strong relationship between reading improvement and post-training increase of beta band over posterior left areas during phonological processing.
182
B. Penolazzi et al. / Behavioural Brain Research 209 (2010) 179–182
showed an EEG beta left posterior lateralization during word decoding [14]. In fact, with the present experiment we demonstrated that this EEG effect is functionally correlated with faster reading speed, therefore, it marks behavioural ameliorations, rather than a fundamental deficit of DD. In conclusion, beta EEG band proved to represent a valid tool for measuring language lateralization and its reorganization in children affected by reading impairment. Left posterior regions, considered important for the early stages of word reading (especially for grapheme–phoneme conversion) were functionally activated with respect to right homologous sites in children who underwent an effective phonological treatment, which was followed by a substantial reading speed improvement. The adopted training is especially suited for the treatment of dyslexic children as it requires only 10 min per day and can be administered at home through a PC, without a continuous and engaging intervention of a speech therapist. Such a daily short-lasting training is especially advised as, following their frustrating reading difficulties, dyslexic children often refuse intensive logopedic trainings.
[7] [8] [9] [10]
[11] [12] [13] [14]
[15]
[16] [17]
References [1] Bakker DJ. Neuropsychological classification and treatment of dyslexia. J Learn Disabil 1992;25:102–9. [2] Cornoldi C, Colpo M, Gruppo MT. Battery for Reading Evaluation. Florence: Special Organizations; 1981. [3] Francks C, Fisher SE, Marlow AJ, MacPhie IL, Taylor KE, Richardson AJ, et al. Familial and genetic effects on motor coordination, laterality, and readingrelated cognition. Am J Psychiatry 2003;160:1970–7. [4] Heim S, Tschierse J, Amunts K, Wilms M, Vossel S, Willmes K, et al. Cognitive subtypes of dyslexia. Acta Neurobiol Exp 2008;68:73–82. [5] Klimesch W, Doppelmayr M, Wimmer H, Gruber W, Röhm D, Schwaiger J, et al. Alpha and beta band power changes in normal and dyslexic children. Clin Neurophysiol 2001;112:1186–95. [6] Kujala T, Karma J, Ceponiené R, Belitz S, Turkkila P, Tervaniemi M, et al. Plastic neural changes and reading improvement caused by audiovi-
[18]
[19] [20] [21]
[22]
sual training in reading-impaired children. Proc Natl Acad Sci USA 2001;98: 10509–14. Milne DR, Hamm JP, Kirk IJ, Corballis MC. Anterior-posterior beta asymmetries in dyslexia during lexical decisions. Brain Lang 2003;84:309–17. Paulesu E, Démonet JF, Fazio F, McCrory E, Chanoine V, Brunswick N, et al. Dyslexia: cultural diversity and biological unity. Science 2001;291:2165–7. Penolazzi B, Spironelli C, Angrilli A. EEG delta activity as index of linguistic impairment in dyslexic children. Psychophysiology 2008;45:1025–33. Penolazzi B, Spironelli C, Vio C, Angrilli A. Altered hemispheric asymmetry during word processing in dyslexic children: an event related potential study. Neuroreport 2006;17:429–33. Ramus F. Neurobiology of dyslexia: a reinterpretation of the data. Trends Neurosci 2004;27:720–6. Ruchkin DS, Johnson R, Maheffey D, Sutton S. Toward a functional categorization of slow waves. Psychophysiology 1988;25:339–53. Shaywitz SE, Shaywitz BA. Paying attention to reading: the neurobiology of reading and dyslexia. Dev Psychopathol 2008;20:1329–49. Spironelli C, Penolazzi B, Angrilli A. Dysfunctional hemispheric asymmetry of theta and beta EEG activity during linguistic tasks in developmental dyslexia. Biol Psychol 2008;77:123–31. Steriade M. Cellular substrates of brain rhythms. In: Niedermeyer E, Lopes da Silva F, editors. Electroencephalography: Basic Principles, Clinical Applications, and Related Fields. Baltimore: Williams & Wilkins; 1993. p. 27–62. Stevens J. Intermediate Statistics: A Modern Approach. Hillsdale: NJ: Lawrence Erlbaum Associates; 1990, 303. Temple E, Deutsch G, Poldrack R, Miller S, Tallal P, Merzenich M, et al. Neural deficits in children with dyslexia ameliorated by behavioral remediation: Evidence from fMRI. Proc Natl Acad Sci USA 2003;100:2860–5. Tressoldi PE, Lonciari I, Vio C. Treatment of specific developmental reading disorders, derived from single- and dual-route models. J Learn Disabil 2000;33:278–85. Tressoldi PE, Stella G, Faggella M. The development of reading speed in Italians with dyslexia: a longitudinal study. J Learn Disabil 2001;34:414–7. Wechsler D. Wechsler Intelligence Scale for Children-Revised. Florence: Special Organizations; 1986. World Health Organization. The International Classification of Diseases. Classification of Mental and Behavioural Disorders, vol. 10. Geneva: World Health Organization; 1993. Ziegler JC, Castel C, Pech-Georgel C, George F, Alario FX, Perry C. Developmental dyslexia and the dual route model of reading: simulating individual differences and subtypes. Cognition 2008;107:151–78.
