Some (but not all) Pragmatic Inferences are Difficult for Autistic Children.

Petit et al. (2025) tested 48 autistic and 48 neurotypical kids . All spoke in full sentences and had average IQ. The team gave two kinds of pragmatic tasks: scalar implicatures (what "some" really means) and metaphor comprehension ("he is a rock").

Kids pointed to pictures or answered yes/no. The study was quasi-experimental: no random assignment, but groups were matched on age and language level.

Autistic children scored lower on scalar implicatures. When they heard "some elephants have trunks," many missed that the speaker meant "not all."

On metaphors, the two groups looked the same. Autistic kids could say that "her voice was velvet" meant soft and pleasant. The gap showed up only on the spontaneous, in-the-moment inference.

Vierck et al. (2015) seems to disagree. Their Mandarin-speaking autistic kids passed scalar implicatures. The clash is real: same task, opposite result. The likely fix is language context. Mandarin grammar and schooling may cue the "not all" reading more strongly than English.

Richman et al. (2001) reported broad pragmatic trouble, including metaphors. Petit et al. (2025) narrows the problem: metaphors can be intact. The older study used harder, open-ended questions; the newer one used simple picture choice. Method change, clearer picture.

Song et al. (2024) also found autistic kids struggle with irony, another high-level pragmatic task. Together, the papers show that tasks needing rapid mind-reading (scalar, irony) are tough, while stored, figurative meanings (metaphor) can be learned.

Check your language targets. If your lesson banks on quick inferences like "some but not all," add extra cues or explicit rules. Metaphor goals can stay on the plan, but monitor context. And if you work with bilingual children, test both languages—pragmatic difficulty can hide in one tongue and show up in another.