Assessing And Training Affective Prosody Perception And Production In Children With Autism Spectrum Disorder

Aims: This thesis comprised a series of studies that examined the assessment and training of affective prosody perception and production in children with high-functioning Autism Spectrum Disorder (HF-ASD), using a range of methodologies. The studies are motivated by evidence that individuals with ASD experience significant auditory processing difficulties, which hinders their perception of the subtle cues in speech that govern emotion expression (affective prosody). This is thought to also affect the production of affective prosodic cues. A number of studies have reported successful behavioural training of emotion perception in children and adults with ASD, especially when delivered via a computer- or technologybased platform. One aim of this research was to investigate whether enhancing the incoming auditory signal by using remote microphone hearing aids (RMHAs) would further improve the effects of computer-based emotion training. Another focus of this research was to explore the neurophysiological patterns underlying affective prosody perception, to determine whether this is different between typically developing (TD) and HF-ASD individuals, and whether these are influenced by emotion training and RMHA use. Methods: The first study implemented a 3-week computer-based emotion training intervention with 24 children with HF-ASD. Twelve of those children underwent computerbased training only. The other 12 children wore RMHAs during the computer-based training sessions, and in addition trialled the RMHAs at school for the duration of the intervention period. The HF-ASD participants were assessed on prosody perception and production, twice before and twice after the intervention. Fourteen TD children made up the control group, and were assessed at one time point. Prosody production was analysed objectively by examining acoustic parameters of recorded emotional speech productions, and subjectively via third party listeners rating the conveyed emotion and intensity of emotion. The listeners were blinded to the intended emotion and the recording session (pre- versus post-intervention). Pre- and post-intervention effects were examined for both HF-ASD groups, and results from the computer training only group were compared to those from children who received the combined intervention. Data from the HF-ASD children were also compared to those from their TD peers. Furthermore, HF-ASD children from the RMHA intervention group, along with their teachers, completed questionnaires to evaluate whether wearing the RMHAs influenced classroom listening experiences and behaviours. The second study involved the recording of cortical auditory evoked potentials (CAEPs) from TD adult participants. Natural speech stimuli with angry, happy, sad, and neutral emotional tones were presented in a passive oddball paradigm. Angry, happy, and sad served as deviant stimuli, occurring less frequently in the sequence of stimuli mainly made up of the neutralstandard. Subtracting the CAEP response of the standard from that of a deviant, results in a derived mismatch response (MMR) representing pre-attentive auditory change detection of affective prosody. CAEP and MMR morphology was examined for the adult participants, and investigated for differences between emotions, with the purpose of evaluating the feasibility of the novel natural speech stimuli introduced in this study. The third study involved the additional recording of CAEPs and deriving of MMRs from TD children participants (on one occasion), as well as HF-ASD participants who were recorded before and after they underwent computer-based emotion training combined with RMHAs. CAEP and MMR morphology was compared between 1) TD adults and children, 2) TD children and HF-ASD children, and 3) HF-ASD children before and after intervention. Results: Outcomes from the first study showed that children with HF-ASD were able to engage in the computer-based emotion training intervention, and that repeated exposure to the activities improved their abilities in both facial expression recognition and affective prosody perception. There were no large differences between children who went through the intervention with the RMHAs, and those who completed the training without the RMHAs. However, only the RMHA intervention group maintained a stable, high performance on affective prosody perception during the 2-week follow up period post-intervention, suggesting better maintenance of training effects when training is combined with RMHAs (effectively increasing the 'auditory dose'). The objective acoustic analyses and the subjective raters' judgements of affective speech productions did not reveal substantial intervention effects for either group of HF-ASD children. However, correlation results indicated that several acoustic factors - namely mean fundamental frequency (pitch), various measures of pitch range, mean intensity (loudness), and various measures of intensity range - were correlated with raters' emotion perceptions indicating that these acoustic features could be considered to contribute to, or even drive, the subjective perception of emotion in speech. Children with HF-ASD who trialled the RMHA at school reported significantly improved listening experiences in the classroom, in particular in a noisy environment. Their teachers also reported significantly improved listening behaviours, and evaluated the effectiveness of the RMHA positively. Joint outcomes from the second and third studies showed that the auditory stimuli, implemented for the first time in a neurophysiology experiment, were successful in evoking typical adult and children CAEP responses that corroborated the existing literature, although latencies were generally later than what is typically reported, which could reflect the complexity of the stimuli. In addition, all participants showed a late N400 response in their standard and deviant waveforms. In previous studies N400 has been associated with the cognitive appraisal of semantic and emotional information, thus the presence of N400 suggests that the auditory stimuli successfully engaged neural processes associated with emotion processing. There were significant differences in CAEP and MMR morphology between TD children and children with HF-ASD. Recordings from the children with HF-ASD before they underwent training did not show any significant separation between mismatch difference waves for angry, happy and sad emotions. This suggests that, unlike their TD peers and the TD adults, children with HF-ASD are to detect changes in the ongoing stimulus stream when a deviant was presented, but did not differentiate the emotional sounds based on either their acoustic features or emotional content. Recordings from the same HF-ASD children after intervention revealed significant differentiation between emotions. However, unlike their TD peers, both late and early MMRs showed differences between emotions. It is hypothesised that this may reflect an increase in perceptual effort or attention required for participants with HF-ASD to perceive changes in the auditory stimulus sequence compared to their TD peers who had access to more pre-attentive automatic processes for differentiating the emotional deviants from the neutral standard stimuli. Conclusion: It is the hope that this series of studies will lead to further research investigating atypical affective prosody and other aspects of auditory processing in children with ASD, using both behavioural and neurophysiological methods. Results suggest the potential benefits to social communication of developing interventions targeting social perception of audio-visual stimuli for children with ASD. The recording and analysis of CAEPs can potentially be considered alongside traditional diagnostic practices of clinical observation and behavioural assessments to screen for early markers of affective prosody and auditory processing dysfunction associated with ASD.