Are There Phonological Consequences of Auditory Processing Differences in Autism Spectrum Disorder? Evidence from Phonological Categorization Tasks

Background: Speech comprehension requires rapid and accurate phonological categorization of temporal and spectral features of the acoustic signal. One well-studied example of phonological categorization is the contrast between voiced and voiceless stop consonants (e.g. /b/ vs. /p/) using voice onset time (VOT), the length of time between stop release and onset of vocal fold vibration. While VOT is continuous, perception of these consonants seems to be categorical. Stop consonants produced along a range of VOTs are perceived as /b/, but with a VOT increase of just a few milliseconds, listeners perceive that sound as /p/. However, underneath this seemingly discrete behavior, the underlying cognitive representations for most listeners are highly gradient and continuous, which can be measured in tasks that accommodate a gradient response format. There appear to be reliable individual differences in the degree of gradiency or categoricalness; one study showed individual differences relate to the degree to which typically developing (TD) listeners integrate multiple cues, such as the primary temporal cue (VOT) and a secondary spectral cue (fundamental frequency [f0]). Although individuals with autism spectrum disorder (ASD) have largely been thought to possess intact phonological abilities, phonological impairments have been reported. Studies of low-level auditory processing, likely relevant for phonological development, have reported impaired temporal processing in ASD and heightenedspectral processing. Additionally, individuals with ASD show more veridical perception, with less influence of top-down information, across perceptual domains.

Objectives: The current study examines whether auditory processing differences in ASD result in differences in phonological categorization or in cue weighting of temporal and spectral cues to the voicing contrast. We hypothesized that heightened spectral processing might be associated with greater use of f0. 

Methods: Participants with and without ASD completed a classic two-alternative forced choice (2AFC) task and a visual analog scale (VAS) task. Adolescents with ASD (n=12) and TD (n=12) categorized bilabial (/b/, /p/) and alveolar (/d/, /t/) stop consonants with VOTs of 1- 45 ms and f0 of 90 or 125 Hz as either bull/pull or den/ten.

Results: Analyses indicated no group differences in boundary placement or categorization, and no group differences in use of f0 as a cue on the 2AFC task (Figure 1a). Additionally, there were no group differences in how continuously the voicing contrast was perceived using a VAS task (Figures 1b-d). Both groups showed high fidelity to VOT in the VAS task.

Conclusions: These data suggest that, in the absence of informative top-down lexical information, individuals with ASD surprisingly show no disadvantage in perceiving temporal information relevant for phonological categorization. Also surprising is the lack of difference in use of a spectral cue (f0) in individuals with ASD. While previous research has found that individuals with ASD show poor auditory discrimination of temporal differences and heightened spectral processing, these auditory processing differences appear not to affect phonological processing, at least in adolescence.