Search Abstracts | Symposia | Slide Sessions | Poster Sessions
A cross-linguistic comparison of neural encoding of speech in autistic school-age children
There is a Poster PDF for this presentation, but you must be a current member or registered to attend SNL 2024 to view it. Please go to your Account Home page to register.
Poster B43 in Poster Session B, Friday, October 25, 10:00 - 11:30 am, Great Hall 4
Nikolay Novitskiy1, Hoyee Hirai1, Janna Guilfoyle2, Joseph C.Y. Lau2, Kritika Nayar2, Trent Nicol2, Jennifer Krizman2, Nina Kraus2, Molly Losh2, Patrick C. M. Wong1; 1The Chinese University of Hong Kong, 2Northwestern University
Pitch serves as a fundamental aspect of language and communication, signaling both lexical meaning and intonational patterns in speech. Atypical intonation is widely observed in autism, reflecting a component of a core symptom domain (i.e., social-communicative differences). Difficulties in effectively perceiving and utilizing pitch may lead to breakdowns in communication, as emotional nuances, emphasis, and the intended meaning of the message are conveyed within the dynamic patterns of intonation. Although differences in pitch perception have been found in some studies of autistic children, the extent to which language background (e.g., tone vs. non-tone languages) may further modulate the autism-related effects is not clear. In tone languages, pitch can alter word meaning, making perception and processing of pitch even more crucial for semantics. While differences in neural pitch encoding have been consistently found in autistic children speaking a tone language (e.g., Cantonese), mixed results have been observed in their non-tone language-speaking (e.g., English) counterparts. The current cross-linguistic study examined the extent to which neural encoding of speech, particularly pitch features, may be modulated not only by autism-related characteristics but also by language background (tone vs non-tone). Specifically, we hypothesized that larger differences would be observed between Cantonese-speaking autistic and typically developing (TD) children, relative to their non-tone autistic and TD peers. 129 children between 8 to 15 years old were enrolled, encompassing 49 autistic and 50 TD native Cantonese-speaking children in Hong Kong and 15 autistic and 15 TD native English-speaking children in Chicago. Autistic children met ADOS-2 autism classification criterion in addition to having a formal autism diagnosis. All children listened to a /da/ syllable via insert earphones while electroencephalography (EEG) was recorded. Specific to the Hong Kong site, children also listened to lexical pitch patterns embedded in the syllable /ji/. 19 early-latency response metrics (seven of which were pitch-related) were extracted, consistent with prior studies of neural encoding in autistic children with different language backgrounds. A series of ANCOVA with Language (Cantonese vs English), Diagnosis (autistic vs TD), Sex, and Age was conducted to test the potential of an interactive effect of Language x Diagnosis on neural speech encoding. Main effects of Diagnosis were found in two EEG features: Pre-stimulus Noise (p=0.0062) and Response Consistency (p=0.0177), supporting the notion of neural hyperexcitability in autism. Significant Language x Diagnosis interaction was found only in Response Consistency (p=0.0471) and Middle-Frequency (260-750 Hz) Spectral Amplitude (p=0.0476). In general, these results did not support the hypothesis that differences in neural encoding of speech in autistic children were driven by language background. In addition, machine learning (ML) was used to conduct binary classification of diagnostic status within each Language group. Taking neural responses to /da/ as input, ML revealed more accurate classification of diagnostic status in the English groups (AUC=0.83) than the Cantonese groups (AUC=0.64), indicating cross-language differences. Curiously, classification results were much higher for the /ji/ responses (AUC=0.83), which were available only for the Cantonese group. A larger and more linguistically balanced sample size is needed to for conclusive results.
Topic Areas: Disorders: Developmental, Phonology