Electrocortical Dynamics in Children with a Language-Learning Impairment Before and After Audiovisual Training

Abstract

Detecting and discriminating subtle and rapid sound changes in the speech environment is a fundamental prerequisite of language processing, and deficits in this ability have frequently been observed in individuals with language-learning impairments (LLI). One approach to studying associations between dysfunctional auditory dynamics and LLI, is to implement a training protocol tapping into this potential while quantifying pre- and post-intervention status. Event-related potentials (ERPs) are highly sensitive to the brain correlates of these dynamic changes and are therefore ideally suited for examining hypotheses regarding dysfunctional auditory processes. In this study, ERP measurements to rapid tone sequences (standard and deviant tone pairs) along with behavioral language testing were performed in 6- to 9-year-old LLI children (n = 21) before and after audiovisual training. A non-treatment group of children with typical language development (n = 12) was also assessed twice at a comparable time interval. The results indicated that the LLI group exhibited considerable gains on standardized measures of language. In terms of ERPs, we found evidence of changes in the LLI group specifically at the level of the P2 component, later than 250 ms after the onset of the second stimulus in the deviant tone pair. These changes suggested enhanced discrimination of deviant from standard tone sequences in widespread cortices, in LLI children after training.

Keywords: Auditory sequential processing; Computerized training; Electroencephalography (EEG); Event-related potential (ERP); Specific language impairment (SLI).

Fig. 1

Layout of the sensor array. Frontal electrodes are shown at the top of the figure. Sites roughly corresponding to locations of the international 10–20 system are also depicted (green). Different groups of electrodes were formed for each CSD-based ERP component, and voltages averaged within each participant for statistical analyses, as described in the “Materials and Methods” section. Adapted from Net Station Acquisition—Technical Manual by Electrical Geodesics, Inc., 2003 (Color figure online)

Fig. 2

Grand mean CSD waveforms over a representative group of fronto-central sensors (Cz and their nearest anterior neighbors 5 and 55, Fcz and their nearest posterior neighbors 9 and 58) at each visit for the two groups in the study, 12 children with TLD (top plot) and 21 children with LLI (bottom plot). Waveforms are shown at the latencies (see “Materials and Methods” section for details) of the P1-N1-P2 peaks interspersed by a negative-going deflection in the MMN latency range (following the N1) in response to standard (gray lines) and deviant (black lines) tone pairs. The inner abscissa in each plot indicates the time scale with respect to the first tone in a doublet, the outer abscissa the time scale with respect to the second tone. At both Visit 1 (solid lines) and Visit 2 (dashed lines), waveform morphology was similar across study groups. Note the superposition of the deflections evoked by the two subsequent stimuli of each tone-pair type

Fig. 3

Grand mean spline-interpolated CSD distribution of the difference wave (deviant–standard) during the MMN time range (see “Materials and Methods” section) at each visit for the two groups in the study, 12 children with TLD (left) and 21 children with LLI (right). Note the consistent topography of the electrocortical response across visits and groups, illustrated in this example (Color figure online)

Fig. 4

Mean amplitude of the negative-going deflection in the MMN latency range averaged across a subset of fronto-lateral sensors (F7, 16, 20, C3, 25 and F8, 57, 56, C6, 50 on the left and right, respectively) for deviant and standard tone pairs at Visits 1 and 2. Values show means of 12 children with TLD (open circles) and 21 children with LLI (filled circles). Vertical bars reflect standard errors of mean. Typical for a mismatch response, the deviant tone pairs elicited an overall larger negativity than standard pairs in both groups of children. The magnitude of the deviant-related negativity was even more pronounced at Visit 2, compared to Visit 1, and did not vary as a function of group membership

Fig. 5

The grand mean (n = 33) CSD deviant–standard difference waveform, for the entire sample, at Visits 1 (solid line) and 2 (dashed line), averaged across a subset of fronto-lateral sensors (F7, 16, 20, C3, 25 and F8, 57, 56, C6, 50 on the left and right, respectively). The bottom abscissa indicates the time scale with respect to the first tone in a doublet, the top abscissa the time scale with respect to the second tone. Note the pronounced negative deflection in the time range between 160 and 220 ms after onset of the second tone

Fig. 6

Mean amplitude of the P2 downward slope averaged across a subset of fronto-central electrode sites (Cz, 5, and 55) for deviant and standard tone pairs in the two groups of children at Visits 1 and 2. Values represent means of 12 children with TLD (open circles) and 21 children with LLI (filled circles). Vertical bars indicate standard errors of mean. There were no systematic variations in the P2 amplitude across visits in the TLD group, but the LLI group showed a pronounced amplitude increment for the deviant tone doublet from Visit 1 to Visit 2. This induced a significant group difference at Visit 2 indicating stronger deviant evoked responses in LLI than TLD children

Fig. 7

Grand mean CSD difference waveforms [∆ (deviant − standard)] over a representative group of fronto-central sensors, including Fcz and their posterior neighbors 5 and 55, at Visit 1 (solid lines) and Visit 2 (dashed lines) for the two groups in the study, 12 children with TLD (gray lines) and 21 children with LLI (black lines). The bottom abscissa indicates the time scale with respect to the first tone in a doublet, the top abscissa the time scale with respect to the second tone. Children with LLI showed a delayed difference wave in the downward slope of the P2 post-intervention, with latencies at Visit 2 and Visit 1 amounting to 284 and 224 ms, respectively, following onset of the second tone in a pair. No systematic latency change evinced in the TLD group