The Harder Task Was Making Metaphors, Not Reading Them

What the research found

Most work on autism and figurative language asks one question: can autistic children understand a metaphor? This study asked two. It looked at comprehension, but it also looked at production, the far less studied skill of coming up with a fresh metaphor yourself. That second question is what makes the paper worth reading closely.

The sample was small and specific. Forty-nine Greek-speaking children took part: 18 with a clinical autism diagnosis and 31 typically developing peers, matched for age, with a mean of around nine years. Every autistic child was verbally able, meaning they could produce flexible sentences and talk about things outside the here and now. Crucially, the two groups did not differ significantly on non-verbal intelligence (Raven’s matrices) or on expressive vocabulary. That matters, because it lets the authors argue that any gap in metaphor skill is not simply a gap in general language or IQ.

Comprehension was tested with a deliberately low-verbal task. Children heard a short metaphor in context (“Turn off the tap. Water is running”) and chose, from four pictures, the one that matched its non-literal meaning. Out of ten items, the autistic children scored 7.8 on average and their peers scored 9.3. So the autistic children understood most metaphors, but reliably fewer of them. This is a real gap, though not a dramatic one, and it echoes earlier low-verbal studies.

Production is where the two groups pulled apart. Here children heard a starter such as “Love is a rainbow” and were invited to complete a new one their own way (“Love is...”). Out of ten, the autistic children produced 3.2 apt metaphors on average; their peers produced 6.0, nearly double. Inventing figurative language was hard for both groups, but markedly harder for the children with autism. An error analysis showed why: the autistic children gave significantly more inappropriate answers (responses that were neither metaphorical nor a real definition) and significantly more non-responses. They were not, however, more literal-minded, the two groups produced literal answers at similar rates, which quietly complicates the old stereotype of the autistic child who takes everything literally.

The study also probed what underpinned performance. For the autistic children, expressive vocabulary predicted both comprehension and production, and working memory (measured by recalling digits backwards) predicted comprehension. For the typically developing children, neither factor predicted much of anything. In other words, the two groups appeared to lean on different mental machinery to do the same job. The authors read this as a sign that autistic children may be recruiting vocabulary and memory to solve a task their peers solve by other, perhaps more intuitive, routes.

What it means for your family

Read plainly, this study says something narrow and something useful. The narrow part: in one group of 18 verbally able autistic children, understanding conventional metaphors was slightly harder, and inventing metaphors was substantially harder, than for matched peers. The useful part: the production gap was the larger one, and it is the part almost no one measures. If your child understands “time is flying” but freezes when asked to make up a comparison of their own, that pattern is now documented rather than dismissed.

It is worth being clear about what this cannot tell you. Eighteen children is a small sample, drawn from two regions of Greece, tested in one language. The authors themselves flag the limitations honestly: females were badly underrepresented (16 boys, 2 girls), the age range was wide (7 to 12), the groups were matched on age rather than on non-verbal IQ, and only children without intellectual disability took part. They also tested only one kind of executive function, working memory, and did not measure theory of mind. So this is a careful snapshot, not a definitive map. It describes a correlation between skills; it does not prove that weak vocabulary causes weak metaphor use, or the reverse.

There is also a genuinely hopeful reading here. The finding that vocabulary predicted metaphor skill for the autistic children, but not for their peers, suggests that for these children, figurative language may be more teachable through vocabulary than we assume. If metaphor understanding rides on word knowledge and working memory in autism, then building both, patiently and explicitly, is a plausible lever. That is a hypothesis this study raises, not one it confirms, but it points practice in a sensible direction.

For parents and tutors, the practical takeaway is one of emphasis. Comprehension often gets all the attention, because it is easy to test: does the child pick the right picture? Production is harder to notice and harder to score, yet it may be where a verbally fluent autistic child struggles most, and where the social cost is highest, since inventing a comparison is how we joke, comfort, and connect. Gently inviting a child to finish “happiness is like...” in their own words, and treating an odd or blank answer as information rather than failure, is well aligned with what this study observed.

Finally, hold the numbers lightly. A nine-year-old who scored 7.8 out of 10 on metaphor comprehension is understanding a great deal. The story here is not deficit but difference: autistic children in this sample reached figurative meaning by a different path, leaning more on the words and memory they had built. Knowing the path can help you support the journey.