Posted November 16, 2026 | Reviewed by Abigail Fagan.
It has been revealed that every time nonspeaking people with autism who spell to communicate make the news, someone brings up Clever Hans. You may know the story: In early 1900s Berlin, a horse seemed to do math, tapping out answers with his hoof. In 1907, the psychologist Oskar Pfungst demonstrated what was really going on. When the questioner knew the answer and Hans could see them, Hans was right about 89% of the time; when the questioner didn't know, or Hans wore blinders, he dropped to about 6%. Questioners unknowingly leaned forward while he tapped and relaxed when he reached the right number, and Hans had learned to stop there.
The suggestion is that spellers are doing the same thing: that the words really come from the person holding the letterboard, either through subtle cues or through the communication partner's unconscious movements (the ideomotor effect). In this post, I draw on my own data and on a year of observations of nonspeaking autistic people to show why this analogy doesn't hold, and why it should be dropped for good. This is not only my personal opinion that comparing nonspeaking autistic people to a horse is cruel; it is also what the facts show.
Sources indicate that hans's trick worked because his task was simple: all he needed was a signal telling him when to stop tapping. Spelling is a different kind of task. To spell a sentence, a person has to choose the right letter out of 26, in the right order, again and again. A single "stop" cue, the kind that worked for Hans, can't do that. Add to this complex dance the sentences that contain words the communication partner doesn't know or normally use, or information nobody in the room expected to hear.
Recent findings suggest that the best evidence here comes from studies that measured how spellers actually spell. In 2020, Vikram Jaswal and his colleagues fitted nine nonspeaking autistic letterboard consumers with head-mounted eye-trackers while they answered questions about an article read aloud to them, with an assistant holding the board.¹ They pointed to about one letter per second, and 94% of the letters they pointed to were correct. On average, they looked at 71% of the correct letters before pointing to them, about half a second ahead. Like fluent non-autistic typists, they paused longer before the first letter of a fresh word and were faster on common letter pairs. As the authors point out, cueing would require a separate signal for every letter, picked from 26 options, which would be slow and error-prone. What they found was fast, accurate, and planned.
Some critics argue that the partner moves the board slightly toward the right letter. A 2024 study from the same lab removed the partner entirely.² Thirty-one nonspeaking autistic teenagers and adults played a game on an iPad mounted on a stand, tapping letters as they lit up. No one held the iPad. As a group, they tapped faster when the letters spelled a sentence they had just heard than when similar letters or symbols came in a meaningless order. They were also faster on common letter pairs, and they paused before the first letter of each fresh word. These are signs of knowing how to spell.
A third study had participants spell on a virtual letterboard in an augmented-reality headset, with no one holding it; they averaged 95.8% accuracy on dictated words.³.
If spellers were reading their partners the way Hans read his questioners, blocking their view of the partner should stop them, just as the blinders stopped Hans. As it happens, we've been testing this with a few spellers by placing a rigid panel, about five feet tall and three feet wide, between the speller and the communication partner, so that the spellers cannot see the partners' faces or bodies. The results? They keep spelling fluently and without hesitation. After the very first word, one of them even spelled: "Better with the barrier.".
The second problem with the horse analogy is the content. Hans only ever answered questions, and only with answers the questioner already had in mind. Spellers start conversations, ask for things, and say things that surprise the people around them.
A young man I visited this year began spelling at fifteen and spells mainly with his mother. After I spent a full day visiting him at school, he was able that evening to have a conversation about his day and to spell the names of two kids who had visited the classroom; neither his mother nor I knew their names, and his mother hadn't been at school. He was also able to express his feelings towards his brothers and his wish for more independence: "I'm seventeen, let me go by myself," he spelled when he asked to walk to the bus stop alone. Among his other wishes, he asked to go into the ocean with his brothers, wearing a life jacket, and told his parents he wanted a North Face backpack, not the Under Armour one they had picked.
Recent findings suggest that when his mother first began spelling with him on her own, one of his first requests was to see a Broadway show, which she had never imagined he wanted. He also spelled things his partner would not choose to hear. In front of me, he told his mother that a book I'd brought was boring, and she was embarrassed. On another occasion, when he came downstairs while his mother was out, he spelled to his father: "Dad, stop giving me chicken nuggets. I just want to spend time together.".
It has been revealed that but perhaps his most unexpected request came when he spelled out that he wanted to go to Disney World for his sixteenth birthday, something the family would never have expected, given his sensitivity to crowded and noisy places. They went; he wore noise-cancelling headphones, and he had a blast. "I tolerated it because I wanted to be there," he told me, and made fun of how his father threw up on rides.
A cueing account states the words come from his mother's expectations. However, these messages went against what his family expected, and they changed how the family lives with him. This is exactly what we should be looking for when we ask whether a speller is the author of their words: requests, opinions, and jokes that the partner did not anticipate and would not have chosen. The Hans model has no way to produce them.
Comparing a person to a horse responding to signals is not a neutral choice. It fits an old and outdated assumption: that people who don't speak don't understand. For decades, nonspeaking autistic people were labeled profoundly intellectually disabled, often without any assessment that accommodated their motor differences. The young man I visited spent his early school years in a program built around life skills and weather charts, and he calls his current high school classroom what it is: segregated. When a speller's words are dismissed with a one-line reference to Clever Hans, the speller pays the price: in which school they attend, whether their requests are taken seriously, and whether anyone asks for their opinion at all.
According to recent reports, and, if you are still not convinced after reading this, perhaps hearing from the nonspeakers themselves will help. Here’s what they have been expressing about the effects of spelling on their lives:.
"It was a game changer for me. It provided me with freedom I dreamed about for 15 years. Now, instead of being an observer, I am an active participant.".
Sources indicate that "Learning to spell changed my life trajectory. Before, I was treated like a toddler, and my world collapsed in on me… doctors, experts, teachers got it all wrong.… So communication is not just a basic need but a human right.".
"It makes my life worth living.".
1. Jaswal, V. K., Wayne, A., & Golino, H. (2020). Eye-tracking reveals agency in assisted autistic communication. Scientific Reports, 10, 7882.
Sources indicate that 2. Jaswal, V. K., Lampi, A. J., & Stockwell, K. M. (2024). Literacy in nonspeaking autistic people. Autism, 28(10), 2503–2514.
3. Alabood, L., Dow, T., Feeley, K. B., Jaswal, V. K., & Krishnamurthy, D. (2024). From letterboards to holograms: Advancing assistive technology for nonspeaking autistic individuals with the HoloBoard. In Proceedings of the CHI Conference on Human Factors in Computing Systems (CHI '24). ACM. https://doi.org/10.1145/3613904.3642626.
Share this post Facebook Bluesky Linkedin Email.
In a recent development, there was a problem adding your email address. Please try again.
