| Problem: It has been previously shown that, unlike typically developing children, children with autism do not use category information in free-recall tasks. This suggests that children with autism may process or organize semantic information differently than typically developing individuals. This hypothesis was tested using a semantic priming paradigm. This study investigated whether high-functioning children with autism show semantic priming effects similar to those for typically developing children. Procedure: Thus far, 6 children with high-functioning autism (10-16 years), 7 typically-developing children (8-14 years), and 6 adults (18-20 years) have participated in this study. Participants sat at a laptop computer and were shown words presented on the screen. They judged whether a real word or a pretend word was presented in uppercase. Half the words were common real words; the other half were words whose vowels had been changed and were therefore nonsensical. On each trial a prime word was presented in lowercase letters for 200 milliseconds. Primes were always real words, and were semantically related to targets on half the trials. Participants were told not to respond to the lowercase-letter word, but to wait for the uppercase-letter word. Two stimulus-onset-asynchronies (SOAs) were used in the 4 experimental blocks of 48 trials each. In two of the blocks, primes appeared 300 ms before target onset. In the other two blocks, primes appeared 2000 ms before target onset. Four lists of words were used. The order of the blocks and lists was randomized across participants. Results: No differences were found across SOAs, so results for the short and long SOA were combined. Table 1 shows descriptive statistics for each group. Also, the data were combined for the typically-developing children and adults because they showed similar effect sizes [F(1, 11)=.68, p=.43]. Typically-developing individuals showed semantic priming [t(12)=2.15, p=.05]. Children with autism did not [t(5)=.12, p=.91]. The difference between groups was not significant [F(1, 17)=1.55, p=.23], but more data are being collected to boost statistical power. There was no difference in overall reaction times [t(17)=.19, p=.85] or response accuracy [t(17)=.87, p=.40] across the groups. Thus, the lack of semantic priming in children with autism cannot be attributed to their inability to do the task. Discussion: No published research has studied semantic priming in children with autism using a lexical decision task. These results apparently show that children with autism do not show typical semantic priming effects. This supports the hypothesis that the way semantic information is stored or processed in children with autism may differ from typically-developing individuals. These results are consistent with the claim that persons with autism are less likely to spontaneously elaborate upon ideas semantically. Table 1: Medians, standard deviations, priming effects, and standard errors, in milliseconds. |
| This poster was presented at the Biennial Meeting of the Society for Research in Child Development in Tampa, Florida, in 2003. |
 |
 |