Abnormal Cry in Toddlers with Autism Spectrum Disorder: A Comparative Analysis of Acoustic Features of Cry in Clinical and Non-Clinical Populations.

Background: Preverbal vocalizations, e.g. crying and babbling, have been found to reflect characteristics of a child’s target language, especially with regard to a language’s prosodic (melodic) patterns (Boysson-Bardies et al. 1984; Mampe et al. 2009). Differences in prosody are one of the most common clinical features and among the earliest characteristics of Autism Spectrum Disorder (ASD) to be identified (Kanner, 1943; Dieh & Paul 2012). However, few studies have examined prosodic differences in vocalizations of toddlers with and without ASD (Sheinkopf et al., 2012; Venuti et al., 2012).

Objectives: This study was designed to investigate whether the cries of toddlers with ASD differ from the cries of their typically developing peers across four acoustic features of prosodic contours (duration, fundamental frequency (f₀), standard deviation of fundamental frequency, and peak frequency). It was hypothesized that the acoustic correlates of cry melodies of toddlers diagnosed with ASD would differ significantly from those of toddlers with typical development (TD) with the same native language.

Methods: As part of a larger study, seven toddlers with ASD (M­_age=24.4 mo, SD=3.5) from monolingual English-speaking households were administered the ADOS-2 (Mod T) by research-reliable assessors, and the sessions were recorded and coded for crying. Previously published audio data for the TD group (n=13, M­_age= 31 mo (SD=6)) were collected at home during tantrums with a LENA device (Green et al., 2012). Each cry was analyzed for four acoustic features (ASD, n=506 samples; TD, n=250); duration, fundamental frequency (f₀), standard deviation of fundamental frequency (SD f₀), and peak frequency.

Results: Independent sample T-Tests indicated that each of the four basic acoustic measures of cries were significantly different between groups (Table 1). Average duration was found to be shorter in the ASD group (p=0.003). Mean F₀ (p<0.0001) and SD f₀ (p=0.01) were found to be higher in the ASD group. Mean peak frequency was found to be lower for the ASD group (p<0.0001).

Conclusions: Each of the acoustic measures differed between the cries of children with ASD and the cries of children with TD in meaningful, predictable ways (Sheinkopf et al., 2012). Cries of children with ASD showed a higher mean fundamental, but also a lower peak frequency, meaning that the cries were higher in pitch overall, but were showed less modulation, which describes a flatter, less prosodically diverse cry. Higher average SD f₀ was found in the ASD group, indicating that the cries were more variable, thereby suggesting less structure. Overall, these results suggest that the ASD group did not follow the basic prosodic pattern of their native language, which can be analyzed by constructing and comparing average melody contours of cries from each group. Planned analyses include construction and comparison of average melody contours of cries from each group, which would graphically describe differences in the melodic structures of the cries.