Video Abstract, Summary and FAQs

Eye-tracking reveals agency in assisted autistic communication



On May 12, 2020, our study on nonspeaking autistic people who communicate using a letterboard was published in Scientific Reports. A 2.5-minute animated video abstract is below, followed by a text overview of the study and why it is important. A webinar presented to the NJ Autism Center of Excellence in Fall 2020 follows that. Finally, some frequently asked questions (FAQs) follow that. You can access the complete paper for free here.


About one-third of autistic children and adults cannot communicate effectively using speech, and most are never provided an effective alternative. Not having a way to use language to convey their thoughts and feelings is arguably the most debilitating aspect of nonspeaking autistic people’s disability, severely limiting their educational, social, and employment opportunities.

However, some of these individuals have learned to communicate by pointing to letters printed on an alphabet board; an assistant holds the board in front of them while they point, but does not touch them. This method–spelling words and sentences, letter by letter–has enabled some nonspeaking people to share their perspectives at the United Nations, to go to college, and even to publish a best-selling memoir.

But some scientists and professionals question whether these or any nonspeaking autistic individual who communicates with assistance is expressing their own thoughts. They believe that the assistant somehow indicates which letters to point to, which would make the assistant rather than the autistic person the author of any text produced. This belief is based on experimental studies that use a method that may not be appropriate for assessing the abilities of nonspeaking autistic people. The belief is also based on problematic assumptions about the relationship between the ability to speak and act in normative ways and the ability to think.

In this study, we took a different approach, using eye-tracking technology to study letterboard users in a natural, everyday setting. Specifically, we investigated how quickly and accurately nine nonspeaking autistic young adults (who had at least 2.25 years of experience using a letterboard) looked at and pointed to letters as they responded to several questions. For example, one participant was asked to name something he had to wait for; he responded by spelling, “Waiting for my dream girl.” 

If the assistant directed participants to each letter–by, for example, moving the letterboard in a particular way to signal “W,” a different way to signal “A,” and so on–we expected that participants would spell slowly, misspell words frequently, and look at several letters before finding the one the assistant was trying to get them to point to. After all, for each letter in each word, they would have to rely on a subtle signal indicating which of 26 letters to select, a signal that could be easily missed or misinterpreted (especially given that autistic people have traditionally been thought to have difficulty reading subtle cues). 

We found the opposite. Participants spelled quickly and accurately: They pointed to about one letter each second, rarely made spelling errors, and usually began looking at the next letter in a word about half a second after pointing to the previous letter. It is unlikely they could have achieved this level of speed and accuracy–sustained in responses of over 20 letters in a row–if they had been relying on subtle signals from the assistant. 

We also identified two patterns in the data that are similar to those observed in non-autistic people as they type on a keyboard: Participants were faster to look at and point to letters within than between words, and they were faster to look at and point to the second letter in pairs of letters that are more frequent in English compared to pairs of letters that are less frequent (e.g., they were faster to look at and point to “e” following “h” than “e” following “j”). Together, these findings suggest that participants generated their own text, looking at and pointing to letters they selected themselves rather than letters the assistant directed them to. 

Our findings demonstrate that the blanket dismissal of methods of communication that rely on an assistant is unwarranted, as is the blanket dismissal of testimony composed by nonspeaking autistic people who use such methods. Our study shows that some of these individuals can convey their own thoughts by pointing to letters on a letterboard held by another person. 

The paper is freely available here, and the data on which the analyses were based are available at the Open Science Framework link here. Three videos showing participants in this study responding to questions were published with the paper and are also available for download here, here, and here.

NJ ACE Presentation (Nov. 30, 2020)


What is the most important message you hope people take from this study?

We should not be so confident in what we think we know, particularly when it concerns people whose experiences and neurologies are different from our own. As new approaches, perspectives, technologies, and data become available (and as autistic people are finally included in the research process), we have to be willing to reconsider some/many/most beliefs about autism that have become entrenched over the last 75 years.

Why did you do this study?

We had the opportunity to get to know some nonspeaking autistic people who use a letterboard, and we wanted to learn about the cognitive processes underlying this remarkable skill. We know that some scientists have dismissed the possibility that any nonspeaking autistic person who communicates with assistance is conveying their own thoughts, and some have suggested that it is not worth studying this phenomenon. But this blanket dismissal seemed to us inappropriate for at least two reasons. First, it is based on results from a particular kind of test that may not be suitable for assessing the abilities of nonspeaking autistic people (detailed in the next response). Second, we think that understanding something as complex as assisted communication requires a variety of methods and approaches.  

Head-mounted eye-tracking technology has been used previously to study complex skills like walking, driving, and piano playing. Knowing where someone looks and its relation to when and how they move can provide insight into the cognitive processes underlying the skill. This is the method and approach we took in the study here.

Why didn’t you use “message passing” tests?

Behavioral scientists have repeatedly demonstrated that tests that fail to take into account a group’s unique developmental history can underestimate or misrepresent the abilities of members of that group. One of the most compelling examples comes from seminal work by linguist William Labov in the 1960s and 1970s. He showed that inadequacies in the way language competence was assessed in African American children were responsible for scientists’ long-standing belief that African American children were linguistically deprived and cognitively incompetent. Tested in other, culturally appropriate ways, the very same children were shown to be highly competent. The scientists mistakenly assumed that it was possible to use the same “simple” tests to measure ability in everyone, regardless of their background. (An excellent treatment of this and similar issues is provided in this classic paper from 1971 by psychologists Michael Cole and Jerome Bruner.)

Message passing tests come in a few varieties, but in a prototypical case, the autistic person is shown one picture and the person assisting them is shown a different one. In studies using this test, the autistic person was usually (but not always) unable to report what they had seen (and sometimes they reported what the assistant had seen, demonstrating that the assistant can influence the person they are assisting). But message passing tests may not be an appropriate way to assess nonspeaking autistic people’s abilities to convey their own thoughts: Their experiences with communication and many other things have been very different from the experiences of people who have grown up being able to communicate using language and who do not face the same sensorimotor challenges. What seems like a simple test to one group of people may not be so simple when applied to a different group whose experiences (and neurologies) have been (and are) very different. This is why we used head-mounted eye-tracking, a method that allowed us to quantitatively analyze the behavior of participants as it occurred naturally.

Isn’t it possible that an assistant holding a letterboard could influence someone to point to particular letters, even if that wasn’t the case in your study?

Sure. The possibility of influence exists in all communication, whether it is between people who talk, sign, spell, or communicate in other ways. Sometimes that influence is perfectly appropriate, as when you help someone find a word or you complete their sentence. It is not appropriate if you put words in someone else’s mouth or hands that do not reflect their intent. 

It is important to acknowledge that someone who uses a letterboard could be influenced by the assistant and to take steps to minimize that possibility. One of the most important steps is to gradually reduce the involvement of the assistant until they are not holding the letterboard at all. All of the participants in our study are working on being able to communicate without anyone holding the letterboard, and some nonspeaking people have learned to do so (for some examples, see here). But learning to communicate independently takes time–think of the years of scaffolding and support most young children receive as they learn to speak in an intelligible manner. 

Aren’t there other interventions that can help nonspeaking people learn to use expressive language and which are not subject to concerns about influence?

Unfortunately, no. First, the most recent systematic review of communication interventions for “minimally verbal” autistic children (Brignell et al., 2018) identified only two randomized controlled trials: one involved a verbal intervention and the other involved training on a popular picture card-based system. The conclusion: “Neither of the interventions resulted in improvements in verbal or non-verbal communication that were maintained over time for most children” (p. 2).

Second, the methods used in these and other kinds of communication interventions rely on techniques in which the therapist explicitly attempts to get the client to say and do particular things–that is, influences them to do so. In speech therapy, for example, the therapist prompts clients to try to say specific things in specific ways; in applied behavior analysis (ABA), the therapist prompts individuals to do certain things in certain ways; and so on. Of course, the goal is to gradually reduce the involvement of the therapist, just as the goal for individuals who point to letters on a letterboard is to gradually reduce the involvement of the assistant.

Interestingly, for participants in our study, spending time learning how to communicate by pointing to letters on a letterboard did not preclude nor displace other interventions or therapies. All of the participants in this study received many years of speech therapy (with a focus on speech and other forms of augmentative and alternative communication), ABA, occupational therapy, and other services. Several participants continue to engage in these interventions. (See Supplementary Table 1 available in the Supplementary Materials here.) 

Are you saying that all nonspeaking autistic people can and should learn to communicate using a letterboard?

No. This was a study investigating the relation between where individuals who already had learned to use a letterboard looked and pointed. The results can be considered an “existence proof” that some individuals can convey their own thoughts using a letterboard. We do not know who can learn to communicate in this way, or what kinds of specific experiences lead to the skill documented in this study. These are important questions for future research, but given the heterogeneity of autism, it is unlikely that this (or any) particular method will be effective for everyone.

What we are confident in saying is that all people want to communicate and are capable of communicating. Individuals who do not yet express themselves effectively using language can and do express themselves effectively in a variety of other ways. But because being able to use language is essential in most educational, employment, medical, and social situations, nonspeaking people should be supported in learning methods that are effective for them. 

Why is there such resistance to communication with assistance?

Here are some possible reasons:

  1. People who don’t have a way to express themselves using language are a vulnerable lot. There are concerns (based on some tragic examples) that an assistant could, for example, harm a nonspeaking autistic person and later claim that the person gave consent. But unfortunately, we know that individuals take advantage of nonspeaking people (and people with other kinds of disabilities) even without making claims about their having given consent (see, for example, an NPR story on this issue here). Thus, we do not think this is a compelling reason to prevent people from learning to communicate with assistance. 
  2. Concerns have also been raised that the time spent learning to communicate with assistance could be better spent learning to communicate in other ways that do not involve an assistant. We are unaware of any data that validate this concern. Indeed, as detailed above, those other methods have not been shown to be effective for most nonspeaking people. And as also detailed above, for participants in the current study, learning how to communicate by pointing to letters on a letterboard did not displace other interventions or therapies. For example, they participated in speech therapy for, on average, about 14 years before being introduced to the letterboard.
  3. The notion that a nonspeaking autistic person could attend college or write poetry is not consistent with traditional beliefs about members of this group. Put plainly, the assumption has been that people who can’t speak and who act in non-normative ways can’t think. But this assumption is belied by recent research showing that the intelligence of “minimally verbal” autistic children using traditional ways of assessing IQ has been massively underestimated. 
  4. Communicating with assistance is complicated. Some individuals require a lot more assistance than others. Sometimes the same individual requires more support on one occasion and less on another. Practitioners and nonspeaking people say the assistant provides emotional, attentional, and regulatory support that has not yet been well quantified. The point is that it’s not as black-or-white as some people would like it to be. But we see this as an opportunity to learn more, not as a reason to dismiss it.

Can you provide details about your sample? Who were these participants?

You can find detailed descriptive characteristics of the sample in Supplementary Table 1 of the Supplementary Materials that accompany the paper, linked here. But in brief, the sample comprised eight men and one woman who were, on average, 20 years old (range: ~15-26 years old). They began learning to use a letterboard when they were, on average, 17 years old (range: ~12-24 years old), and they had been using it for, on average, 3 years (range: ~2-6 years) at the time they participated in the study. Before beginning to learn how to communicate using a letterboard, participants had received traditional speech therapy (in school and privately) for, on average, 14 years (range: ~10-22 years). At the time of the study, all participants had received (or continued to receive) applied behavior analysis for, on average, 11 years (range: ~5-18 years).

Did you study Rapid Prompting Method (RPM)?

No. RPM is one method designed to teach people how to communicate by pointing to letters on a letterboard. The focus of our study was on people who had already learned to communicate using a letterboard. We did not study the method by which they learned this skill. 

Did you study Facilitated Communication (FC)?

No. FC is a method that some nonspeaking people use to communicate by typing on a keyboard. Typically, an assistant provides gentle pressure at the person’s typing hand, wrist, arm, elbow, shoulder, or back as the person types. The participants in our study, in contrast, pointed to letters on a letterboard held by an assistant who did not touch them. 

How come some autistic people cannot communicate effectively using speech?

We don’t yet know. The nonspeaking autistic people we know are motivated to communicate, and most (but not all) of them express a desire to learn to be able to convey their thoughts using speech. Speech is the finest of fine motor skills, and so the most promising explanation to us involves the well-documented motor challenges that many autistic people who are not fluent speakers face (see, for example, this 2008 paper by Morton Gernsbacher and colleagues).