Autistic Traits and Lexical Effects On Speech Perception in Children

Background: Recent research on linguistic aspects of Autism Spectrum Disorder (ASD) shows that individuals with ASD tend not to employ top-down processing to resolve ambiguities in linguistic structures such as homographs, homophones and sentences with multiple interpretations. Stewart and Ota (2008) examined the use of lexical information to resolve phonetic ambiguity, and showed that typically developed adults with high scores on the Autism-Spectrum Quotient (AQ; Baron-Cohen et al., 2001) were less likely than low AQ scorers to interpret acoustically ambiguous auditory stimuli to be part of a real word (e.g., kiss) than a non-word (e.g., giss). In other words, autistic tendencies in neurotypical adults are related to an attenuated lexical effect on speech perception. There was no evidence that this effect was due to increased acoustic sensitivity or poor lexical access of individuals with high AQ scores.

Objectives: Our main objective of the study was to explore such a connection between autistic traits and lexical effects on speech perception among neurotypical children. Given that signs of ASD manifest themselves during childhood, it was hypothesized that the same relationship should be found in young individuals.

Methods: Participants were 4-6 year old typically developing children. None of them had been diagnosed with ASD. Their parents completed the Social Responsiveness Scale (SRS; Constantino et al., 2003), as a measure of the children’s autistic traits. The main experimental task was a phoneme identification task with three pairs of /k/-/g/ non- to real word 7-step acoustic continua: kiss-giss, kift-gift; keep-geep, keese-geese; kept-gept, kess-guess. Stimuli were created by digitally cross-splicing endpoint items (e.g., kiss and giss) to interpolate the intermediate items. In addition, as measures of their acoustic sensitivity and lexical access ability, participants were given an auditory discrimination task on the same voice onset time continuum as well as a auditory lexical decision task involving non- vs. real word pairs on voicing contrasts (e.g., cake, gake).

Results: Preliminary data collected from 12 participants show a significant negative correlation between the SRS and the mean identification shift between the two VOT continua (r = -.75, p < 0.01). There was also a significant positive correlation between the SRS and the reaction time in the lexical decision task (r = .617, p < 0.05).

Conclusions: Like adults, children who have more behavioral traits typically associated with ASD tend not to be influenced by lexical information in their phonemic identification. There are also indications that this effect may be caused by slower lexical access, although it remains to be seen whether this correlation will be upheld when more data are added.