Pattern of semantic errors in autism: a brief research report.
Autistic kids with mild ID skip broad-label shortcuts and zoom in on object parts when naming, a style difference you can use in therapy.
01Research in Context
What this study did
The team looked at how autistic kids with mild learning delays name pictures. They compared 20 autistic kids to 20 non-autistic kids who had the same IQ range. Each child named 50 common objects while the researchers wrote down every wrong answer.
Wrong answers got sorted into types like "under-extension" (calling a truck "car") or "part-name" (calling a shoe "lace"). The goal was to see if autism creates a special pattern of word mistakes.
What they found
Both groups made mostly the same kinds of errors. The big surprise: autistic kids never used under-extension. They never called a truck just "car" or a lion "cat."
Instead, they leaned on parts. When stuck, they said "wheel" for bike or "handle" for cup. Non-autistic kids with the same IQ used under-extension all the time. Error totals were equal, but the style was different.
How this fits with other research
Marsack-Topolewski et al. (2025) extends this work. They showed autistic listeners adapt to new speech sounds just like peers. Together, the two papers suggest autistic kids can learn and use words, they just take a different route when unsure.
McCauley et al. (2018) looked at kids with ADNP syndrome, a genetic form of autism plus ID. They also found social skills track closely with verbal ability. That matches I et al.'s finding that IQ, not autism itself, shapes error count.
Yin et al. (2026) used the same three-group design but studied play instead of words. Both studies find autistic kids show unique style, not just delay, compared to developmental-delay peers.
Why it matters
When you test vocabulary, watch for part-focused guesses instead of broad category guesses. If a child says "spout" for teapot, that may be typical autism pattern, not sign of worse language. Skip goals that push under-extension (learning broad labels first); target part-to-whole links instead. Use the child’s strength in detail to build word webs: wheel-bike-pedal-ride.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →During next vocabulary probe, score part-name errors separately and turn them into teaching trials ("Yes, that’s the wheel—what has wheels?").
02At a glance
03Original abstract
Semantic or vocabulary errors were measured among children with autism and mild learning disability and children with mild learning disability only (six children of 7.9-8.7 years in each group), testing the hypothesis that these were common in both groups. Different variables were investigated such as breadth of vocabulary and number of vocabulary errors, type of paraphasias, and mechanisms used to name the meanings that children were not aware of or could not recall, and the particular meanings that were difficult for each group. Preliminary findings showed that vocabulary errors were similar in both groups, except under-extension, which was not used by the autism group. Children with autism tended to use all mechanisms in order to name something they did not know and to focus on parts of the object in order to name it, while unknown words were similar in both groups.
Autism : the international journal of research and practice, 2003 · doi:10.1177/1362361303007002006