The Effect of Semantic Context On Prosodic Processing Across the Autism Spectrum

Background: Children with autism spectrum conditions (ASC) display difficulties in interpreting contrastive stress (Paul et al., 2005; Peppé et al., 2007). This may be due to the reduced use of contextual information that is typically used to guide meaning (Happé & Frith, 2006).

Objectives: To assess the influence of semantic context on prosodic processing in ASC and typically-developed adults with different degrees of autistic character traits.

Methods: The study was divided into two parts. Part one compared 19 young adults with High-Functioning Autism or Asperger’s syndrome (HFA/AS) with 19 IQ-matched typically-developed young adults. Part two assessed the relationship between autistic traits in 68 typically-developed young adults. The individuals with HFA/AS were given the Autism Diagnostic Observation Schedule (ADOS; Lord et al., 1999). All participants completed the Raven’s (Raven, Raven, & Court, 1998), the Millhill, the Autism-Spectrum Quotient (Baron-Cohen et al., 2001a), and two tests of prosodic processing. These measured the accuracy and response times of the listeners’ ability to identify contrastive stress. In Test A, the placement of contrastive stress was contextually congruent for some items but incongruent for others. Test B measured discrimination of stress positions in laryngeal recordings with no discernible semantic or lexical information.

Results: Part One: The Raven’s-matched HFA/AS sample showed a significant main effect of Congruency (accuracy, F(1,36)=6.12, p<0.05; RT,  F(1,36)=20.14, p<0.001);  performance in congruent trials was significantly better than in incongruent trials (accuracy, t(37)=2.51, p<0.05; RT, t(37)=-4.56, p<0.001). There was no significant main effect of Group or Group x Congruency interaction. However, there was a significant positive correlation between the Millhill and the Raven’s with accuracy in the congruent trials in Test A (r=0.53, r=0.50; p<0.05, respectively) and between the Millhill and accuracy in Test B (r=0.34, p<.05). Part Two: The typically-developed sample performed significantly better in congruent trials than in incongruent trials (accuracy, t(67)=3.36, p<0.01; RT, t(67)=-4.05, p<0.001). Performance was not related to autistic traits. However, there was a significant positive correlation between the Millhill and accuracy (r=0.25, p<0.05) and a significant negative correlation between the Millhill and RT (r=-0.25, p<0.05) in the congruent trials in Test A. There were no relationships between autistic traits and performance of prosodic processing in Test B.

Conclusions: The identification of contrastive stress is influenced by congruency with semantic context, but this effect is not reduced as a function of increasing autistic traits in those with and without HFA/AS. Difficulties with contrastive stress were not found in the HFA/AS sample nor was performance related to autistic traits. The current sample was an adult one; there may be fewer difficulties with contrastive stress in this age group as prosodic skills develop gradually into adulthood (Wells, Peppé,  & Goulandris, 2004). Further research is needed (a) to assess whether reduced processing of contrastive stress in semantic context is not captured by the current task or sample and (b) to understand the roles of verbal and nonverbal IQ on identification of contrastive stress.