New evidence for melodic speech in Autism Spectrum Disorder

Abstract

Since the very beginnings of research into Autism Spectrum Disorder (ASD), there have been contradicting descriptions of speech in ASD as being “singsongy” or melodic on the one hand and “robotic” or monotonous on the other. We highlight some issues regarding the terminology and methodologies used in previous studies as well as their comparability, concluding that previous accounts, particularly of monotonous speech in ASD, may have been misleading. We expand on a previous pilot study in using the same method of quantifying the spaciousness and liveliness of speech along two dimensions in order to analyse an extended data set (~ 5 hours) of semi-spontaneous conversations. We compare 14 German adults diagnosed with ASD and 14 matched control speakers (CTR), recorded in disposition-matched dyads (ASD-ASD; CTR-CTR). Using Bayesian modelling, we present evidence that most (but not all) ASD speakers in our corpus produced a more melodic intonation style than non-autistic CTR speakers, while, crucially, none produced a more monotonous intonation style. We emphasise the importance of inter-individual variability in groups of autistic speakers and point out that our results align with a clear tendency in recent studies to report more melodic speech in ASD.