Billet publié sur Bluesky le 27/05/2026 12:25

Résumé

Background and aims Autism research has traditionally been shaped by the idea that autistic people have “deficits” compared to neurotypical people. This approach often involves directly comparing the two groups, with the expectation that autistic people will perform worse. However, neurodiversity advocates argue that autism should be seen as a valid way of being, and should not be defined in contrast to norm-centred ideals. This review aimed to assess to what extent autism research in the field of language and communication is structured around such norm comparisons. Methods Following PRISMA protocols, a cross-sectional systematic review methodology was used to gather a sample of 249 relevant articles published in 2023. Articles were assessed as to whether they included a neurotypical comparison group and how the autistic group was expected to perform in contrast to neurotypicals. Additionally, we coded whether the study was an intervention, and whether non/minimally speaking (NMS) participants or those with intellectual disability (ID) were included, as we expected patterns to differ in these cases. Main contribution We found that almost half (49%) of the studies compared autistic and neurotypical participants, and this rose to 75% when interventions and studies involving ID or NMS participants were excluded. When predictions were made about which group would perform better, 60% of studies expected autistic people to do worse, while only 4% expected them to do better. Intervention studies were less likely to make direct comparisons, and studies involving ID or NMS participants also tended not to use neurotypical comparison groups. Conclusions Overall, our findings show that autism research is still largely built on comparisons with neurotypical norms, reflecting an assumption of deficit. Implications For research to align with the neurodiversity perspective, future studies will need to move away from these deficit-based traditions. Suggestions for alternative, less norm-based approaches are discussed.