Atypical Cortical Processing of Pitch Contours Predicts Social Communication Deficit in Tonal-Language-Speaking Children with Autism

Background: Linguistic relevance shapes the neural coding of critical acoustic features pertinent to one’s native phonology. Pitch contour is a defining feature of lexical tones in tonal languages such as Mandarin Chinese, and native Chinese listeners’ cortical responses have been found to be more sensitive to tones with a curvilinear frequency trajectory than native English listeners (Krishnan, Gandour, & Suresh, 2014). It remains unknown to what extent whether such a language-specific neural marker of pitch encoding is developed in tonal language users with autism.

Objectives: The current electrophysiological study employed nonspeech stimuli similar to those in Krishnan et al. (2014) varying in pitch trajectory to examine closely the cortical processing of linguistically relevant pitch information in Chinese children with autism. Clinical significance was explored by examining the correlations between the neural measures and autism core symptoms.

Methods: EEG was recorded from 22 native Chinese-speaking children with autism (9~13 years) and 27 age-matched TD controls. The participants were instructed to watch a self-chosen muted cartoon while ignoring any sounds. The stimuli were trains of nonspeech sounds containing either a prototypical Chinese rising tone (Tone 2 or T2) with a naturalistic curvilinear frequency trajectory or a linearly accelerated T2 that is linguistically unnatural. The linear T2 shared the same onset and offset frequencies of the prototypical T2, thus coarsely approximated the frequency trajectory of T2. The purpose of using nonspeech was to eliminate the influence from phonological and semantic contexts and thereby examine more precisely the acoustically driven cortical encoding of the pitch features. The ERP (event-related potential) responses in selected time windows were examined to see possible between-group differences. Scores from Social Communication Questionnaire (SCQ; Rutter, Bailey, & Lord, 2003) were obtained from the autism group to see if language-dependent auditory encoding could autism symptom measure.

Results: Both groups demonstrated significantly larger N250 amplitude in the processing of the prototypical curvilinear T2 relative to the linear T2. However, a group*stimulus condition interaction was observed in an earlier window of 180-230 ms. Specifically, the prototypical contour of T2 but not the linear contour produced ERP positivity in the TD group. In contrast, the ERPs for the two tones in the autism group did not differ, indicating a lack of distinction. Moreover, there was a significant correlation between ERP in this window at Pz for the linear T2 and SCQ measure. Greater ERP positivity for the linear T2 (towards reduced prototypical vs. linear distinction) was predictive of more severe autism symptoms.

Conclusions: Chinese school-age children with autism and TD controls were both sensitive to the linguistic pitch features represented by the prototypical and linear Tone 2 trajectories in later cortical processing stage. However, unlike the TD group, the autism group failed to show an early ERP distinction of the tones. The correlation between SCQ and language-dependent cortical pitch encoding suggests a potential link between low-level auditory perception and the clinical symptomology of autism. The results call for further developmental studies of atypical language-dependent perceptual reorganization and underlying mechanisms in autism.