Describing (pre)linguistic oral productions in 3- to 5-year-old autistic children: A cluster analysis.
Autistic preschoolers fall into five vocal types, so ditch the speaking/nonspeaking label and match goals to the child’s actual sound profile.
01Research in Context
What this study did
Maes et al. (2023) recorded natural play sounds from 3- to young learners autistic kids. They coded every squeal, babble, word, and echo. Then they ran a cluster analysis to see if the kids sorted into clear vocal types.
The sample was the preschoolers who had little or no spoken language. No tests or drills—just describing what their mouths already did.
What they found
Five vocal profiles popped out instead of the usual “speaks” vs “doesn’t speak” split. One group only cried or laughed. Another group babbled long strings but used no words. A third group mixed real words with copied phrases. The last two groups fell between these extremes.
In short, early autism language is a rainbow, not an on-off switch.
How this fits with other research
Jansen et al. (2013) also used cluster analysis on preschoolers with language delay. They found four groups based on joint-attention and symbol skills. Together, the two studies say: “Use data, not labels, to sort little kids.”
McGonigle-Chalmers et al. (2013) showed that nonverbal autistic kids can still understand grammar on a touch screen. Pauline’s work adds: even within “nonverbal,” kids sound different, so check both what they say and what they grasp.
Meng et al. (2026) tracked Mandarin-speaking autistic kids and found grammar unfolds in the same order as typical peers, just slower. Pauline’s vocal map gives you the starting point; Ziyan shows the road ahead.
Why it matters
Stop writing “nonverbal” in the goal box. Listen for 10 minutes, pick the child’s cluster, then aim your first target at the next sound that cluster usually adds—babble, single word, or phrase. You will write sharper goals and show faster progress notes.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Record five minutes of free play, note the child’s most advanced sound type, and set the next vocal step as your first goal.
02At a glance
03Original abstract
For most autistic children, spoken language emergence and development happen after the age of 3. Once they start developing and using spoken language, some eventually manage to reach typical levels of language abilities, while others remain minimally speaking into adulthood. It is therefore difficult to consider young autistic preschoolers as a homogeneous group in terms of spoken language levels. In our study, we breakdown a representative and inclusive group of children on the spectrum aged from 3 to 5 into five subgroups that correspond to different linguistic profiles. To do so, we qualitatively described children's (pre)verbal productions elicited during interactions with a parent and with an experimenter. We then used a type of statistical analysis called cluster analysis to group together the children that had a similar expressive (pre)linguistic behavior. Using this analysis, we were able to delineate five linguistic profiles with qualitatively different patterns of vocal production. Two of these profiles are composed of speaking children; the three others are composed of non- or minimally speaking children. Our findings show that traditional binary division of speaking versus nonspeaking autistic children is not precise enough to describe the heterogeneity of early spoken language in young autistic children. They also support the use of qualitative descriptions of vocal productions and speech to accurately document children's level of language, which could, in turn, help design very finely tailored language intervention specific to each child.
Autism : the international journal of research and practice, 2023 · doi:10.1177/13623613221122663