Language and labels

Identity-first language puts the identity first ("autistic person"); person-first language puts the person first ("person with autism"). A 2025 systematic review of 19 online studies with 6,350 participants found that in 10 of the 14 studies that asked for a preferred term, more autistic participants chose identity-first language. Between 4 and 39% still preferred person-first, and the two studies that found the reverse were both Dutch. The authors concluded there is no consensus, across languages or always within them.¹

The best-known survey did not single out "autistic". Among 3,470 people in the UK autism community, the terms endorsed most across groups were "autism" and "on the autism spectrum". "Autistic" was endorsed by many autistic adults and parents but considerably fewer professionals, and "person with autism" by almost half of professionals.² Later surveys of autistic adults leaned further toward identity-first terms. Of 654 English-speaking autistic adults, diagnosed or self-identified, from about 30 countries but mostly the UK, US, Canada, Australia, Ireland and New Zealand, 79.5% endorsed "autistic person" and 23.9% "person with autism" or its variants.³ In the US, 299 autistic adults overwhelmingly preferred identity-first terms, though 13% preferred person-first, while 207 professionals leaned person-first.⁴ Results ran the same way in 784 North American adults, most of them autistic, white and speaking,⁵ and in 541 French-speaking autistic adults, a survey that shares authors with the international one.⁶ In 198 Australian autistic adults, "person on the autism spectrum" ranked first overall, and "autistic" and "autistic person" were the other two most preferred terms.⁷

Two Dutch-language surveys found the opposite: 68.3% of 1,026 Dutch autistic adults aged 16 and over preferred person-first language,⁸ and of 414 Dutch-speaking autistic adults in Belgium and the Netherlands, 54% preferred person-first, 27% had no preference and 14% preferred identity-first.⁹ Samples drawn in different ways have given different answers, though no study has compared recruitment methods directly. In a US study of 2,098 autistic people and 3,181 non-autistic parents drawn from a large research registry, almost half fell into an "open" group that liked or was neutral about most terms. Autistic people were overrepresented in the identity-first group and non-autistic parents in the person-first group, though that group also included many autistic people.¹⁰

Some style guides have moved away from requiring person-first language, while others still default to it. The sixth edition of the APA Publication Manual (2010) told writers to "put the person first".¹¹ The APA's current guidance says that both approaches "are fine choices overall", gives "autistic person" as an identity-first example, calls "high functioning" and "low functioning" "both problematic and ineffective", and lists "special needs" among terms to avoid.¹² The NIH style guide says that "person on the autism spectrum" is generally "the least polarizing term", notes that autistic communities often prefer identity-first language, tells writers to default to person-first language when a preference is unknown, and recommends "nonspeaking" over "nonverbal".^13,14 Journals differ. Autism in Adulthood requires identity-first language or neutral terms such as "person on the autism spectrum", and anti-ableist language throughout.¹⁵ The journal Autism's guidance says it does not expect terms such as "suffer from" in its articles, while calling itself "not a prescriptive checklist".¹⁶

Research writing has changed more slowly. In 12,962 abstracts from 11 autism journals, 65% used person-first language, 16% identity-first and 20% both, with identity-first use rising in recent years.¹⁷ In 2,322 autism articles published in 2021, traditional medical language still dominated. Affirming language was more common in journals with identity-first guidelines and in studies of autistic adults.¹⁸ One author's count across article titles, abstracts and full texts found person-first phrasing used most for children with the most stigmatized conditions: in about 75% of references to children with autism and 93% for intellectual disability, against under 1% for gifted children.¹¹ That count shows how the wording is used; it does not test whether the wording adds to stigma.

In 2021 five researchers proposed replacing wording they described as ableist in autism research: "co-occurring" for "comorbid", "increased likelihood" for "risk", "characteristics" or "traits" for "symptoms", "comparison group" or "nonautistic" for "healthy controls", and a description of specific strengths and needs in place of functioning or severity labels.¹⁹ The paper is a commentary drawing on autistic scholarship, not a study. In a 2023 commentary, four authors replied that the push to limit terms came from "only a subset" of the autism community and that scientists should be free to use any scientifically accurate term.²⁰ A response with many autistic and non-autistic co-authors argued that "likelihood" and "co-occurring" are accurate equivalents, and that terms such as "profound autism" and "severe" are inconsistently defined.²¹

A few experiments have tested whether a label changes how readers judge people. In 530 US adults recruited online, people described in short scenarios as having "special needs" were viewed more negatively than people described as having a disability or a named disability.²² In a registered experiment with 522 UK adults, "people with a physical disability" led to fewer positive stereotypes and more dehumanization than "the physically disabled", with no effect on negative stereotypes. The same wording had different effects for another group, and autism was not tested.²³ No experiment found as of October 2026 has compared identity-first and person-first wording about autistic people.

"High-functioning" usually meant an autistic person without intellectual disability. Studies of adaptive behavior show why the label fits poorly. In 2,225 children and adolescents aged 1 to 18 on a Western Australian diagnostic register, adaptive scores in those without intellectual disability fell significantly below IQ, while in those with intellectual disability they were close to it. The authors found IQ a weak predictor of adaptive scores.²⁴ In 417 North American adolescents with an IQ of 85 or above, 90% of them boys, about half had daily living skills well below what their IQ would suggest, and about a quarter scored below 70.²⁶ In 417 autistic people aged 6 to 31 in Europe, mostly male, the average gap between IQ and adaptive scores was about 28 points in children and 33 in young adults, a difference the authors call modest.²⁵

How much IQ predicts is disputed. An analysis of 1,089 verbal children and adolescents from the same North American collection as the adolescents above found IQ a strong predictor, explaining 55% of the variation in adaptive skills, though the gap between the two was largest at higher IQ and in older participants.²⁷ Even where IQ predicts well on average, a high IQ is no guarantee of few support needs.

Most autistic adults surveyed reject functioning labels. In a survey of 516 autistic people, mostly from the UK and US and mostly women or non-binary, 74% of the 510 who answered the question said the labels should not be used and 13% said they should. Of those against them, 86% preferred stating a person's individual needs, alone or alongside a diagnosis. In that survey, views on language, including functioning labels, were associated with how connected respondents felt to the autistic community, in a direction the authors could not establish; samples recruited through community channels may therefore show more rejection.²⁸ In open-ended answers, most of 784 North American adults, 611 of them autistic, described the labels negatively.⁵

DSM-5 introduced three levels, "requiring support", "requiring substantial support" and "requiring very substantial support", rated separately for social communication and for restricted, repetitive behaviors. The 2022 revision, DSM-5-TR, left them unchanged.²⁹ In US surveillance records of 15,450 autistic children aged 4 and 8, only 40.4% had any level recorded, from 4.8% to 73.2% depending on the site. Among children with a level recorded, those with intellectual disability were rated Level 3 for social communication in 46.6% of cases, against 17.8% of those without. Levels were more often missing for Black children than for white children, and among those with a level, Black children were more often rated Level 3 for social communication, after allowing for intellectual disability.²⁹ In 248 children at six Autism Treatment Network clinics, ratings were higher in younger children and those with lower IQ, and the authors suggested that clinicians may be folding cognitive ability into the levels.³⁰ When the same 49 toddlers aged 16 to 31 months were assessed once by telehealth and once in person, the levels agreed only fairly for social communication (kappa 0.32) and slightly for repetitive behaviors (kappa 0.20), with wide uncertainty, even though agreement on the diagnosis itself had been high.³¹ Critics argue that severity in the two core areas is not the same as disability, which also depends on intellectual ability, language and conditions such as anxiety.³²

A Lancet Commission proposed "profound autism" in late 2021, in a report printed in 2022, for autistic people with the highest support needs: those who need 24-hour access to an adult who can care for them, cannot be left alone in a residence and cannot manage basic everyday needs. It is usually associated with an IQ below 50 or very limited language, and was suggested only from early school age, for example 8 or older.³³ A 2026 Delphi exercise turned this into a research definition: adaptive functioning well below age level, a need for adult supervision to protect the person's health, safety and well-being, an age of at least 8, an autism diagnosis, and either an IQ below 50 or speech limited to single words or fixed phrases used mainly for basic needs. The exercise drew on researchers, caregivers, autistic people and clinicians, and 76% of the final round agreed. That round had 78 respondents, a 58% response rate, and the authors say the sample was overwhelmingly from the United States and may underrepresent parts of the stakeholder community.⁵⁰ The term is in use in research: a 2026 article in Biological Psychiatry, by Matthew State and Robert Malenka, calls for clinical and translational research on it.⁵¹

How many people it covers depends on the definition, and on which definition a service uses to count. Counting children with an IQ below 50 or little or no speech, 26.7% of 20,135 autistic 8-year-olds in US surveillance from 2000 to 2016 qualified; these children were more likely to be girls, Black, Hispanic or Asian or Pacific Islander, and of low socioeconomic status.³⁴ With an added criterion for low adaptive scores, an estimated 16.5% of 7,084 did in 2022, after statistical filling-in of missing data: the assessments needed to decide were missing for almost half of the children.³⁵ Both estimates come from the same CDC surveillance network, and across other samples the share has ranged from 11% to 48%.⁵² The choice reaches beyond research: in a 2025 analysis of US survey data, four definitions of developmental disability, autism among them, gave prevalence estimates from about 4% to 16% in children and young adults, and the authors say the narrower definitions may better capture the people most likely to use publicly funded services.⁵³

The label has critics, and the criticism is about the label, not the people it describes. In the largest survey of autism terminology, with 5,279 US respondents, about three-fifths of them non-autistic parents, "profound autism" was the most disliked of the subgroup terms, and opposition to those terms was strongest among participants who favored identity-first language.¹⁰ A 2026 preprint, not peer reviewed, looked at genetic, gene-activity and regulatory data in a large cohort and found no evidence that "profound autism" is a distinct biological group; the differences it did find were in inferred gene regulatory networks separating children who do not speak from those who do.⁵⁴ The people the term describes, meanwhile, have large needs that often go unmet. Among 238 autistic adolescents, a third met at least one criterion for "profound autism", most often intellectual disability, and their caregivers were more than eight times as likely as other caregivers to expect that the adolescent would need supervised housing, a marker for long-term care planning.⁵⁵ Caregivers reported an unmet need for 28 of the adolescents (11.8%), most often dental care, and 174 of the 238 caregivers (73.1%) reported an unmet need for themselves.⁵⁵ The criteria are often missing from service records, which the authors say contributes to missed chances at service planning.⁵⁵ In one study of 20 caregivers of adults who met the criteria, 85% described at least one experience of stigma.⁵²

DSM-5 (2013) merged autistic disorder, Asperger's disorder, childhood disintegrative disorder and pervasive developmental disorder not otherwise specified (PDD-NOS) into one diagnosis, autism spectrum disorder.³⁸ The WHO's ICD-11, adopted in 2019 and in effect from January 2022, also has a single autism spectrum diagnosis, divided only by intellectual development and functional language, with no Asperger's category.^37,38

Clinicians had not applied the subtypes consistently. In 2,102 children, mostly boys, at 12 US university clinics, whose scores on standard measures were similar across sites, the share given an Asperger's diagnosis ranged from 0 to nearly 21% depending on the site, and in statistical models of each site's diagnoses, the verbal IQ cut-off that best separated Asperger's from the other subtypes ranged from 62 to 127.³⁹ Whether Asperger's is a distinct group remains disputed. A review of studies up to 2002 found few qualitative differences between Asperger's and autism without intellectual disability,⁴⁰ while a meta-analysis of 52 studies found that groups with Asperger's had higher IQ scores, with verbal IQ above nonverbal IQ, than groups with "high-functioning autism", and its authors read this as support for separate subtypes.⁴¹ Since verbal IQ predicted who received the Asperger's label in many clinics, some of that difference may reflect how the groups were chosen.³⁹

DSM-5 states that people with a well-established DSM-IV diagnosis of autistic disorder, Asperger's disorder or PDD-NOS should be given the new diagnosis.⁴² Studies that re-applied the DSM-5 criteria to earlier samples estimated, in a meta-analysis of 33 studies, that 20.8% of people with any earlier autism diagnosis would not meet them, and 23.3% of those with Asperger's, with a wide range of uncertainty (12.9 to 38.5%) for Asperger's. Overall, the authors found smaller drops in autism diagnoses than earlier reviews had; an earlier systematic review reported that most studies expected 50 to 75% to keep a diagnosis.^43,36 That review of 25 studies found the largest drops in people with an IQ above 70 and earlier diagnoses of PDD-NOS or Asperger's, though the later meta-analysis found no significant drop for PDD-NOS.^36,43 These are re-scorings of existing records, not counts of people who lost a diagnosis.

When the change was proposed, members of an online Asperger's forum, in 19 discussion threads, reacted in two ways: some welcomed the wider spectrum, while others feared losing services and an "aspie" identity.⁴⁴ Of 547 health and education professionals surveyed by an Australian team, about half opposed the merger, and they saw "autism" as the more stigmatizing label.⁴⁵ For members of the public the label made less difference: in 465 US adults reading a short description of a child, labeling the child with autism spectrum disorder rather than Asperger's did not change stigma.⁴⁶ In interviews with 12 Australian adults diagnosed with Asperger's before DSM-5, views of the change tracked how far each had integrated an autistic or disabled identity.⁴⁷

In recent samples of autistic adults, a minority use the old term. Of 654 English-speaking autistic adults, 23.2% were happy to use "Asperger's syndrome" and 18.0% "Aspie". Some kept it because it was their original diagnosis; others felt it set some autistic people above others.³ In the survey of 516 autistic people described above, about a quarter of them self-identified, 22.2% of the 501 who rated the term found "Asperger's" acceptable and 61.9% did not.²⁸ In the registry study that included non-autistic parents, "Asperger's syndrome" was among the better-liked subgroup terms, with a median rating of neither like nor dislike.¹⁰

The name itself carries a history. Archival research documents that Hans Asperger publicly supported Nazi-era race hygiene and on several occasions cooperated with the child "euthanasia" program in Vienna.⁴⁸ A later study of his referrals to the Spiegelgrund clinic found no evidence that he knew about the killing program when he referred two girls who died there in 1941; the historian whose work is cited above disputes this reading.⁴⁹

Many autistic writers reject functioning labels and explain why from their own lives. Megan Anna Neff asks readers to "Imagine if neurotypical people were asked their 'functioning' levels". She writes that her own functioning shifts with her sensory surroundings, burnout and the energy spent masking, and suggests talking about support needs instead.^c1 Devon Price describes a nurse who is excellent at her job yet some days cannot speak, and the writer Ido Kedar, who could not speak and was treated as "low functioning" for years until he learned to type. For Price these cases show "how superficial functioning labels can be."^c2 Eric Garcia writes that he will not use functioning labels and instead describes people as having "higher support needs" or "lower support needs".^c3 Steve Silberman's history records that the founders of Autism Network International had all been labeled low-functioning as children and understood that "functioning levels change not only in the course of the life span but also day to day."^c4

Not everyone drops the labels. Price stands by a nonspeaking friend, Angel, who calls himself low functioning and severely autistic, and writes that such labels can mark the "social privileges" of autistic people who can talk, dress themselves or hide their meltdowns.^c5

The label at the other end has its own cost. Daniel Burford, writing for adults after a late diagnosis, writes: "'High-functioning' means you're in pain and the pain costs nobody else anything."^c10

On identity-first language, Silberman quotes Jim Sinclair, a founder of Autism Network International: "It is only when someone has decided that the characteristic being referred to is negative that suddenly people want to separate it from the person."^c4 Price writes that some people still identify as having Asperger's, though the label "no longer exists", and that people who had it forced on them "may feel an attachment to it, or a desire to reclaim it."^c5 Sarah Nannery begins her book with "I have Asperger's syndrome" and then explains that it is now part of autism spectrum disorder.^c6

Writers who prefer identity-first language often defend other people's choices as well. Jules Edwards writes, "I choose to use identity-first language for myself, I'm an autistic person," and that someone who prefers "person with autism" should be supported too; Edwards objects to parents correcting autistic people's own choice.^c11 The editors of Sincerely, Your Autistic Child adopt identity-first language as an organization but write that "people have the right to choose how they identify."^c12 Pete Wharmby writes that "people with autism" makes autism sound "like a piece of baggage that we carry around".^c14

Writers also take on other words. Garcia dislikes "special needs" because it "implies that disabled people are getting preferential treatment rather than the assistance that is their right."^c7 Price writes that nonspeaking autistic people are "fully formed people who are typically capable of communicating" through devices, letter boards, sign language, texting or behavior.^c8 Price also separates "allistic", meaning simply non-autistic, from "neurotypical", which they use for someone with no mental illness or cognitive disability.^c5 Wharmby prefers "non-speaking" to "non-verbal", because "'Non-verbal' suggests a lack of any kind of language use", when many non-speaking autistic people are "fully verbal internally".^c14 Some writers replace "neurotypical" with "predominant neurotype", or PNT. Luke Beardon writes that the term came from an email sent by an autistic adult, and Beardon uses it for the non-autistic majority.^c15 Nick Chown, who credits the term to Beardon, prefers it because it separates autistic and non-autistic people "on the basis of numbers only" and "makes no judgement as to whether one cognitive experience is better than another."^c16 Research on these words is in the sections above; Double empathy covers "allistic" and Neurodiversity covers "neurotypical".

In a 2024 explainer, Katharine Annear, who teaches disability and community inclusion at Flinders University, describes "neurospicy" as one of several newer, informal coinages from online neurodivergent communities, not from the 1990s neurodiversity movement or from clinicians. It can describe a person with more than one form of neurodivergence, or a group with several neurodivergent members.^c17 For some it avoids disclosing a diagnosis; for others it is a creative way of pushing back against clinical wording such as "mild autism". Others dislike it: the neurodivergent podcaster Danielle Sullivan asks whether it is just a cute, quirky word or a way to avoid saying disabled, and some argue for dropping such words and saying disabled plainly.^c17 The pedia does not use the word in its own voice.

Nonspeaking writers have their own views on these words. Amy Sequenzia, who is nonspeaking and grew up called "very low-functioning", writes: "There are big implications and great damage caused by functioning labels."^c9 Alan B. Alexander, who learned to communicate at 40 using the rapid prompting method, chooses person-first language, writing of being "a person with autism" and that "All of us are people first."^c13