Supporting effective alternative access for individuals with physical disabilities: state of the science, emerging technologies, and future research directions

Abstract

Alternative access methods accommodate an individual’s specific physical disability, by adjusting the typical keyboard/mouse/touchscreen interface or using an alternative interface such as head controls, eye trackers, or switches. Ideally, they allow for efficient and comfortable use of any computing device, such as laptops, tablets, and smartphones. This paper reviews the literature to address two main questions: how well do existing access methods meet users’ needs? And what is the best way to choose the “right” access method(s) for a given individual? We report on typing speeds for various access methods across 57 studies, as well as experimental multimodal access methods across 34 papers. Results support the intuitive understanding that the access method affects the end user’s experience and productivity. We also reviewed 33 papers and resources related to alternative access assessment and service provision. Existing evidence remains emerging and suggests that use of a systematic, user-centered approach to alternative access assessment may reap significant benefits, including better team performance and improved user satisfaction. Yet although many alternative access assessment resources are available, providers may not always use them. While today’s alternative access methods can be an impactful means of enabling participation for people with physical disabilities, there is room for improvement, both in the access methods themselves, as well as their associated service provision and support. To identify and implement these improvements, the field needs research and development that is fully inclusive of people with disabilities in all phases of the work. We list some additional priorities for future work.

Computing devices, such as laptops, tablets, and smartphones, generally require fine motor control of the hands and fingers to operate them efficiently and easily. The typical control interfaces of keyboard, mouse, or touchscreen may be difficult or impossible to use for people who have physical disabilities. This may include individuals with acquired conditions (e.g., spinal cord injury, traumatic brain injury), degenerative conditions (e.g., amyotrophic lateral sclerosis [ALS], multiple sclerosis), and congenital conditions (e.g., cerebral palsy), comprising over 2 million people in the United States alone (Centers for Disease Control and Prevention, 2018a, 2018b; Hersh & Fox, 2018; Mehta et al., 2018; National Spinal Cord Injury Statistical Center, 2018). And for those who cannot use speech as their primary means of being understood, difficulties with physical access can also make it harder to use speech-generating devices and apps effectively for augmentative and alternative communication (AAC).

Alternative access methods ideally accommodate an individual’s specific physical needs and allow for efficient and comfortable use of any computing device (see Figure 1). Alternative access may involve adjustments to the typical keyboard/mouse/touchscreen interfaces, such as changing settings or positions. It can also involve using a different control interface, such as alternative keyboard and mouse options, adapted stylus solutions, speech input, head controls, switch input devices, and eye tracking technology (Beukelman & Light, 2020). These control interfaces can be operated by the body sites(s) that work well for the individual, such as hand, single finger, head, eyes, foot, etc.

Figure 1.

Alternative access to computing devices.

Note. user controls an access technology interface to use a computing device for desired activities.

Although a variety of alternative access methods already exist, unfortunately not everyone gets what they need. Without an accurate and efficient access method, individuals can struggle to perform basic activities such as email, social media, internet use, and communication, which can negatively impact participation in education, employment, and the community (Cook & Polgar, 2013; Fager et al., 2012a, 2012b, 2019; Johnson et al., 2006).

Effectiveness of an access interface depends on many factors, including an individual’s physical, perceptual, cognitive, and language skills; the suitability of the interface for those skills; the suitability of the interface for the individual’s desired tasks; and the amount of support required and available for initial setup, positioning, training, and maintenance (Cook & Polgar, 2013; Simpson, 2013). At least partly because of this complexity, it is not always easy to know which alternative access methods will be the best fit for a particular individual.

This paper reviews the state of the science related to two key questions in the area of alternative access:

1. To what extent does it matter what access method one uses? That is, what is known about how well existing interfaces have performed?

2. What is the best way to choose the “right” access method to provide the best fit for an individual with physical disabilities?

For each question, we review the available literature, describe key takeaways from current research, and outline priorities for future research. The author team includes an interdisciplinary group of researchers with the RERC on AAC (https://rerc-aac.psu.edu/), with backgrounds in speech-language pathology, engineering, and user interface design, as well as an individual who has been using alternative access for the past 15years (author KJ). This paper includes her insights as well as those gleaned from the Future of AAC Research Summit held in May 2024 (Jin 2024). Our overall purpose is to summarize the existing evidence of current alternative access methods and identify key future directions to improve alternative access—the technology itself as well as the practice of providing it—so that more people get what they need to effectively participate in desired activities.

Part 1. Does the access method matter?

We start by addressing the overall question: Does the access method matter? That is, how much difference does it make to the person using it? This question is too broad to address in every aspect, given the variety of populations, interventions, and outcomes that could be considered. But we can examine three related sub-questions within the scope of this paper:

1. For people with physical disabilities, what typing speeds are achieved when using commonly available alternative access methods?

2. Do the configuration settings for an access method make a difference?

3. Are there advantages to multimodal approaches that use access methods in combination with each other?

Typing speed review: performance with current access methods

Many indicators of success with alternative access methods exist such as speed, accuracy, ergonomic comfort, satisfaction, psychosocial impact, goal attainment, and AAC usage. Perhaps the most common indicator cited in the research literature is typing speed, also referred to as text entry rate or communication rate. This metric gives insight into how efficient a user is when entering text with their alternative access method and provides an engineering benchmark as well. To address the question of typing speeds, we expanded the systematic review completed by Koester and Arthanat (2018), which determined the typing speeds reported for various access methods, to support more informed decision-making when choosing an interface.

Methods for typing speed review

We updated the data from Koester and Arthanat (2018) review to add all studies performed since its publication, using identical search strategies and data extraction methods, and continuing its focus on exclusively individuals with physical disabilities. Briefly, we searched 10 databases and included studies in which: typing speed was reported in words per minute (WPM) or equivalent; the access interface was available for public use; and individuals with physical disabilities were in the study population. For each study, we extracted the words per minute data reported for participants with disabilities, then combined those across studies for each basic type of access method.

The results reported here reflect all 57 papers meeting the criteria for quantitative synthesis: 18 new articles were identified from the recent search and were combined with the 39 papers identified in the original review (Koester & Arthanat, 2018). The data represent typing speed results from 454 individuals with physical disabilities, spanning a time period of 1986–2023. (See this public sheet for a table of all the data. Data are also browsable in AT-node (2024) at https://kpronline.com/atnode.)

Results for typing speed review

Table 1 summarizes the overall averages for each type of access method, in descending order of average typing speed. For the 7 types of access methods in the original 2017 review, the average typing speeds hardly changed at all after adding the new data. The update includes data for two new access methods, touchscreen typing and eye typing, which were not represented in the original review.

Table 1.

Overall typing statistics for each type of access method.

Access Method	N	Avg WPM	SD	min	max
Speech recognition	50	15.42	9.99	3.50	32.20
Standard keyboard	179	12.63	8.66	1.20	48.00
2-switch morse	1	12.39	0.00	12.39	12.39
Touchscreen typing	15	6.79	1.07	4.40	8.30
1-switch morse	1	4.88	0.00	4.88	4.88
Direct oSK	69	4.77	3.26	0.60	11.82
Eye typing	20	4.72	4.63	0.16	15.00
Scanning oSK	54	2.09	1.93	0.15	12.94
BCi (P300)	60	0.66	0.08	0.30	1.74
other	5			1	26
total	454			0.15	48

Note. N: the number of individuals in the dataset for this access method. All other columns reflect typing speed in words per minute. SD: standard deviation; oSK: on-screen keyboard; BCi: Brain-computer interface.

The original Koester and Arthanat (2018) review provides extensive detail about sub-groups within each access method, such as variation with diagnoses and body sites, and the new data remain consistent with those findings. For the purposes of this paper, we present five main results from this dataset that relate to the question of whether the access method matters, specifically the impact of access method on typing performance.

Standard vs. touchscreen keyboards.

A “standard” keyboard is the typical physical keyboard found on laptops or plugged into tablets. It remains a viable access method for many with physical disabilities. For example, people with cervical spinal cord injuries type an average of around 13 wpm on a standard keyboard, typically with one or two typing sticks held in the hand with splints, and mouth stick typists average almost 10 wpm (Koester & Arthanat, 2018). In contrast, typing on a touchscreen, like the on-screen keyboard built-in to an iPad tablet, comes in at only 6.8 wpm for this dataset, about half the speed reported for standard keyboard. These touchscreen data are from two studies, involving people with essential hand tremor, arthritis, and similar difficulties with typing (Kar et al., 2021; Sarcar, 2021). While more data with a wider range of individuals and experience is needed for a firm conclusion, these results suggest that it is unwise to assume that touchscreen keyboards and standard keyboards will yield the same performance.

Cursor selection on-screen keyboards (OSK).

A “cursor selection on-screen keyboard” presents a virtual keyboard on the screen. To enter text, the user moves an on-screen cursor to the desired target and either clicks or dwells there to select it. The cursor may be moved with a pointing device such as a mouse, trackball, or head-controlled mouse (listed as “Direct OSK” in Table 1). Or it may be moved using an eye tracking camera that follows where the user’s eyes are looking (listed as “Eye typing” in Table 1). The data in this review suggest that typing speeds are similar for both of these basic types, averaging about 4.7 wpm for both, with ranges of 0.6 to 11.8 for Direct OSK and 0.16 to 15 wpm for eye typing.

Switch-based methods.

Interface types that use one or two switches for input include scanning on-screen keyboards and Morse code. Scanning uses one or two switches to select items from a selection set presented on the display (see “Switch access scanning” (Wikipedia, 2022) for a good introduction). It is generally regarded as the slowest commonly available access method, and the data support that, with an average of 2.1 wpm. However, it is interesting to note the reported maximum for scanning of almost 13 wpm. It is also intriguing to note the relatively fast typing speeds reported for Morse code, albeit representing only 2 individuals. The 12.4 wpm rate was achieved by an individual with C3 spinal cord injury, using sip-puff for the 2 switches (Levine et al., 1986). Two-switch Morse is one of the earliest alternative access methods, yet it may be under-utilized or dismissed as too difficult. Recently, Google has revived Morse to some extent, with its Morse keyboard and training tool (Google, n.d.); however, we have not yet found any data on people using Google’s Morse for alternative access.

Brain-computer interface (BCI).

This updated review provides substantially more data on performance with brain-computer interfaces, as performance from 60 individuals is now available, compared to only four individuals in the original review. Yet the average of the reported rates remains unchanged compared to 2017, at 0.66 wpm. These data are all from one basic type of BCI—P300 matrix spellers with surface electrodes (Brumberg et al., 2018; Peters et al., 2022)—given our criterion that the method be available to the public, as opposed to solely within a research study. While it is difficult to predict the impact of future developments and innovations, the lack of improvement reported in studies with P300 BCI suggests that dramatic increases may be unlikely. This isn’t all that surprising, considering that the achievable rate with P300 is largely determined by the implementation, including the number of flashing sequences per letter, and how long each sequence takes. Author KJ has used BCI with surface electrodes as a research participant, and notes: “BCI is a very cool idea with LOTS of room for improvement. I think the top typing speed [0.66 wpm] is generous, at least for me! It was frustrating to get so close to the right selection and not be able to get it, and not know what to do differently”. It may be time to look at surface-electrode BCI systems in a different way, which is a motivator for our Smart Select project, described in more detail below.

In contrast, BCIs that use implanted electrodes have achieved some intriguingly fast results in the last few years: including 5.6 correct wpm for cursor control on an on-screen keyboard (Pandarinath et al., 2017), 18 wpm decoding attempted handwriting (Willett et al., 2021), and 62 wpm (with 25% error rate) decoding attempted speech (Willett et al., 2023). These BCI variations can currently be used only within research studies, and because they require neurosurgery to implant the electrodes within the brain, they will not be a solution for everyone.

Case examples for individuals.

The studies in the systematic review also provide some insight into how the access method matters for a given individual. Here we present a few case examples from those studies.

In Lancioni et al. (2011) case study of 3 individuals with cerebral palsy, subject S1 had been using an on-screen keyboard with a hand joystick. They set up single-switch scanning for him, and speed improved by 30%. So, in that case, the keyboard he was using was actually slower for him than the single-switch scanning method. This illustrates the importance of taking performance measurements, instead of assuming that general trends based on interface type will hold for every individual.

The setup for an interface (its “configuration”) can make a big difference, too. Body site and positioning is one important aspect of setup. Pouplin et al. (2015) study measured typing speed for 15 people with C4–5 spinal cord injury who had used a cursor on-screen keyboard for years. Those who used a head-controlled mouse typed about twice as fast as those who used a trackball with their chin. Software settings for the interface also make a difference. With switch scanning, for example, adjusting the scan time, scan pattern, item layout, and other factors can readily double typing speed (Koester & Simpson, 2014, 2019). This can occur even when performance “looks good”, as in the case of S004 (Koester & Simpson, 2014). He had excellent switch control, excellent command of the layout, very few errors, and high satisfaction with his baseline system. Yet initial performance was only 1.2 wpm. Careful adjustments yielded a 136% increase in typing speed (Koester & Simpson, 2014). With eye typing, a small change in dwell time can have a big impact; in Pi et al. (2020) study, typing speed for one user improved from 7.5 to 11 wpm by changing the dwell time from 500 to 400 ms. Author KJ also notes configuration issues with her eye gaze system: “Before ALS, I could regularly type 100 wpm with few errors. I knew it would be slower with eye gaze, but there is room for improvement. Often it can’t keep up with my eyes so the accuracy goes way down. I can adjust the settings to increase the speed but then I have more accuracy issues”.

Multimodal review: combining access methods in new ways

While the studies reviewed above focused on performance when using one primary access method, the field has also been developing and exploring creative combinations of access methods, or “multimodal” methods. Much of the research exploring multimodal access methods has been motivated by challenges experienced with single modality methods. For example, challenges associated with eye tracking—including Midas touch or inadvertent activations, inconsistent accuracy due to lighting conditions in the environment, etc.—have been described in the literature as the rationale for exploring combinations of modalities to mitigate these challenges. Here we review some key findings to address the multimodal aspect of whether access methods matter. Specifically, are there advantages to using access methods in combination with each other?

Methods for multimodal review

The method for this narrative review involved serially searching 5 databases (Google Scholar, PubMed, CINAHL, ACM digital library, and PsycINFO) using a series of search terms including multimodal + AAC, multimodal + alternative access, eye tracking + switch scanning, eye tracking + head tracking, eye tracking + gesture recognition, eye tracking + voice recognition, head tracking + voice recognition, and BCI + eye tracking. Resulting abstracts were screened to include papers that appeared to investigate a multimodal access method for computing and/or AAC purposes. After full text review, papers were retained if the purpose of the work was to help individuals with physical disabilities use AAC or computerized technologies. This process yielded 34 papers for this review.

Results for multimodal review

Our review of the literature revealed a range of combinations of access solutions including technology such as eye tracking, head tracking, switches, joysticks, regular keyboard and mouse options, surface electromyography (sEMG), and brain computer interfaces (BCIs). The physical movements for controlling these technologies include eye movements and blinks, head movement, facial muscle movements including mouth and tongue, hand movement, and feet movement. For example, head gestures with voice (Mosquera-DeLaCruz et al., 2021), and blinks with head tracking (Krapic et al., 2015) have both shown improved performance in rate and accuracy compared to single-modality options. Some investigated the inclusion of physiologic measures (e.g., photoplethysmography or blood volume changes) to enhance detection of movements that can be difficult to detect (e.g., tongue movements) (Gemicioglu et al., 2023). Additionally, software interface adaptations have been described, such as an algorithm that enhanced target selection with devices such as mouse, trackball, touchscreen, and stylus (Biswas & Langdon, 2012). In some multimodal studies, the component access methods are used sequentially; in others, they are used simultaneously, in parallel.

Eye tracking has been described in the literature as a technology that could benefit from a multimodal approach due to a common phenomenon called the “Midas touch”. When dwell is used as the primary selection modality (e.g., looking at a target for a predetermined amount of time), users may accidentally activate or select targets by gazing upon them. We found eleven studies that explored alternative selection methods for eye tracking, including switch, joystick, voice, head gesture, breathing, foot movement, sEMG, and photoplethysmography (Bissoli et al., 2016; Biswas & Langdon, 2012, 2015; Feng et al., 2021; Kumar et al., 2020; Meena et al., 2016; Onishi et al., 2022; Rajanna et al., 2022; Singh & Prasad, 2015; Špakov et al., 2014; Zander et al., 2010). In 8 of these studies, the additional selection modality resulted in increased accuracy. The impact on the rate of target acquisition varied. For example, Onishi et al. (2022) found that breath detection (e.g., using inhalations and exhalations as a switch method) for target selection was slower but had less errors than dwell selection with eye tracking. Feng and colleagues’ (2021) Hgaze system, which combined eye tracking with head gestures, improved overall rate and accuracy compared to eye gaze with dwell selection. These examples indicate that novel alternatives to dwell selection with eye tracking may improve accuracy and selection rate, but not universally.

Eye tracking has also been combined with switch scanning, for both on-screen keyboard (Fager et al., 2023) and visual scene display layouts (Mandak et al., 2021). In this approach, the individual first uses eye tracking to indicate a gross region on the display, then uses switch scanning to select a specific target within that region. This approach yielded mixed results across subjects and interfaces and appeared to provide the most benefit with a larger number of targets (e.g., keyboard interface versus visual scenes displays with a limited number of targets). Fager et al. (2023) found that for individuals who struggle with using eye tracking, the combined approach may help them learn or eventually improve performance using eye tracking over time by providing a method to practice using their eyes as an access method without requiring a high level of resolution. For some, this multimodal approach helped when their eye tracking was inconsistent (e.g., dry eyes, positioning constraints, etc.).

Noninvasive BCI is another area that may benefit from a multimodal approach. Studies of noninvasive BCI and AAC have indicated challenges with selection, training, and overall inconsistent acquisition of signals to control AAC technology (O’Doherty et al., 2014). To address these challenges, the RERC on AAC is investigating a multimodal approach called Smart Select, that combines motor imagery BCI and sEMG (Brumberg et al., 2024). In this approach, EEG brain signals based on motor imagery (imagining a physical movement) are potentially bolstered by clinical and subclinical sEMG signals (elicited when the individual attempts to move in addition to imagining the movement). For this early stage feasibility project, the combined EEG and sEMG signals will control a scanning interface through switch access.

Many investigations into the combination of EEG signals with eye tracking or eye movements have been described in the literature. The EEG in these studies included SSvEP, P300, or motor imagery signal types; see Brumberg et al. (2018) for a tutorial description of these BCI types. In our search, 18 papers explored the development and feasibility testing of hybrid BCI systems that incorporated EEG (SSvEP, P300, or motor imagery) with eye tracking or eye movement signals (Barreto et al., 2000; Biswas & Langdon, 2011; Brennan et al., 2020; Cheng et al., 2020; Huang et al., 2013; Kalaganis et al., 2018; Kalika et al., 2017; Lee et al., 2013; Lin et al., 2019a, 2019b; Liu et al., 2024; Mannan et al., 2020; O’Doherty et al., 2014; Saravanakumar & Ramasubba Reddy, 2019; Stawicki et al., 2017; vilimek & Zander, 2009; Wang et al., 2023; Yong et al., 2011). The studies found benefits related to increased accuracy, rate, and information transfer rates on a range of tasks from simple target acquisition to spelling. While these results are encouraging, only one study included individuals with disabilities (Brennan et al., 2020), and all were conducted in laboratory settings. There is still much to learn about the benefit of these hybrid systems on the AAC and computer access needs of individuals with disabilities in real-life settings.

Limitations and future directions for multimodal research

The results show some potential for multimodal methods, but the experiences of individuals who may actually use these solutions are largely unaccounted for. Out of the 34 studies considered here, only 5 collected data from those with physical disabilities (Biswas & Langdon, 2012; Brennan et al., 2020; Fager et al., 2023; Mandak et al., 2021; Mosquera-DeLaCruz et al., 2021). While some benefits to multimodal approaches have been demonstrated in individuals without neurologic impairment (typically students associated with research labs), little is understood about the true impact of these technologies for the individuals who rely on alternative access methods to AAC and computers.

Investigations into multimodal access reviewed above have largely consisted of feasibility testing and have primarily occurred in laboratory settings. Adding modalities to access has yielded some benefits such as increased accuracy and at times increases in the rate of target selection. However, there is little information available to guide clinical decision-making for multimodal access methods. In some cases, technology has not yet been translated to market, limiting the availability of these options for individuals who use AAC and practitioners. Guidance as to whether access methods should be combined simultaneously or sequentially is not well understood. Much of the existing literature focuses on examining performance outcomes when combining modalities but limited information is provided on development of technology that allows individuals to switch modalities to accommodate specific needs (e.g., when becoming fatigued). Rationale as to when to use multimodal access does not currently exist, and details on the kinds of options available and the long-term outcomes of these methods remains unclear. Additionally, in some areas, including the United States, funding of multiple access options is not yet available. While some multimodal access methods have proven feasible, much is left to be understood before these methods are proven to be beneficial and become widely available.

Discussion: does the access method matter?

The simple answer to our first question—does the access method matter?—is Yes. The results of the typing speed review support the intuitive understanding that different access methods tend to yield different performance, on average. Perhaps more importantly, for a given individual, changes to the access method can significantly influence performance and productivity; these changes can include adjusting the interface to be a better fit for an individual, by changing configuration settings or positioning, or moving to a completely different type of method. New multimodal access methods under development have potential but have limited and mixed results.

The typing speed review includes averages and ranges for 9 different types of access methods. These provide a starting point, but since everyone is unique, it can be misleading to assume that group means will translate to a specific person’s performance. Note, too, the wide ranges are observed for all the interface types. For example, the lowest speed for standard keyboard typing is 1.20 wpm, much lower than the 12.6 wpm average. Similarly, the lowest speed reported for eye typing is only 0.16 wpm, much lower than the 4.7 wpm average for that type of interface.

This highlights the need to monitor performance for individuals, as part of helping an individual find the access method(s) that are the best fit for their needs. A low-performance situation may not be obvious (see example of individual S004 above). Thus, we recommend taking measurements during and after assessments. This helps identify people who may be struggling and to identify accommodations that may improve the situation. It can also confirm when things are going well.

Limitations of part 1

The studies reviewed above provide examples of how access methods can matter, in a variety of ways and for a variety of people. However, the limited scope of this paper cannot provide a comprehensive answer that covers all possible situations. We’ve focused heavily on performance metrics like typing speed and accuracy thus far (the reported typing speeds tend to be in correct WPM, so they reflect accurate typing). Speed and accuracy can be very important metrics for many individuals, but it is still just one relevant outcome. Comfort, ease-of-use, satisfaction, and the like are other important indicators of success. Different end tasks may necessitate different measures.

Service providers also need to consider the challenging reality that there are numerous factors that can influence an individual’s performance with any access method, including the interface type and its setup, as discussed at length above, as well as client experience, motivation, health status, training, and support. These results also do not address the important issue of equity—availability of resources and skill-building opportunities for alternative access use for everyone who needs it. People need resources throughout the process: to work through an assessment process with a service provider if needed, to try a variety of potential access methods, to purchase the access methods that are the best fit (not just the most readily available or least expensive), and to get support and training if needed as well. Author KJ notes that follow-up support is an area where services can fall short: “Users need someone who can check in periodically and provide support. When I got each of my three eye gaze systems, they set up the physical stuff but I was left to figure out 85% or more of the software stuff on my own. I can see how people give up”.

Part 2. How to choose the “right” access method?

In Part 1, we explored why an alternative access method matters: the right choice can increase the speed, accuracy, and effectiveness of performing tasks, including communication using AAC. Given that a range of access methods exist and there are many ways to configure each one, what can be done to ensure that a person receives an access method that fits their unique needs? We address that question in Part 2.

The typical recommended approach is to conduct an alternative access assessment, a process involving a team of service providers and other stakeholders, especially including the individual with physical disabilities at the center of the team. A thorough assessment involves gathering information, identifying candidate solutions, trying and selecting from those candidates, implementing specific solutions, and providing training and follow-up. Many resources, tools, and approaches are available to aid the alternative access assessment process. Yet up to 70% of users abandon their assistive technology (AT), at least partially due to limitations in service provision and the assessment process (Howard et al., 2022; Scherer et al., 2005). To understand how to maximize positive outcomes for people who use alternative access methods, what does the current evidence-base recommend for teams to optimize the alternative access assessment process?

Methods for assessment review

We reviewed the literature centered on the following questions:

1. What protocols, models, and tools are available to guide alternative access assessments?

2. What is the impact of the use of these protocols, models, and tools on the person using alternative access methods and their team?

Databases were searched to identify studies that describe assistive technology and/or AAC assessment models, resources, or tools, and their impact on people who use alternative access, their communication partners, and assessment team members. The search yielded 33 papers and resources for this review.

Results for assessment review

Before presenting the findings of this review, it is important to acknowledge a few key considerations. First, the topic of alternative access assessment is vastly understudied. Thus, the results should not be viewed as definitive or prescriptive but, instead, as emerging and as a call to action to fuel more clinical research questions and exploration. Second, across the globe, people with disabilities, their families, their care teams, and their communities are using alternative access methods, assessing the merit of these methods, and adapting to meet their needs with the resources available to them. For several reasons, a substantial part of this work is not documented in scholarly journals or accessible through online searches, resulting in an incomplete understanding of the field. For instance, alternative access tools such as eye tracking and switch access have become more widely available within commercial products that support face-to-face communication, computer access, and more. With the rapid democratization of alternative access tools and practical constraints, people with disabilities, practitioners, and teams must also rapidly assess the impact of these tools on daily life, often without time to fully document the process or methods for future analysis. Consequently, the findings of the review offer only a partial representation of the broader landscape, limited to what is available in the research literature and published resources. Nevertheless, these findings may serve as a call to action for the AAC and assistive technology community to identify, document, and elevate the daily clinical evidence that remains underrepresented in the existing literature.

Available models, protocols, and tools for assessment

A variety of empirical and clinical models exist to help practitioners structure the assessment process and gather relevant information necessary to make informed access decisions. Examples of commonly referenced models include HAAT (Human Activity Assistive Technology model) and SETT (Student Environments Tasks Tools framework) (Cook & Polgar, 2013; Zabala, 2005). A recent scoping review (Burnham et al., 2024) identified 14 empirical models to guide the AAC assessment process, 10 of which are general AT models that apply to alternative access assessment. Burnham et al. (2024) found several similarities across models for alternative access. For example, half of the models included in the review explicitly use the International Classification of Functioning, Disability Health (World Health Organization, 2001) examining body functions and structures, activity and participation of the individual, and contextual factors (environmental, personal), suggesting an increased adoption of a biopsychosocial approach to assessment and movement away from a medical model of disability (Bernd et al., 2009). Further, most models framed the assessment process as ongoing or iterative, and considered both personal factors and environmental factors for a more holistic assessment. Most models advocated for a collaborative approach involving the person who uses AAC, their families, and practitioners from multiple disciplines; however, less than half of these detailed how such collaboration should occur. Collectively, the essential features of the assessment process have been described as “team collaboration [including the person who uses AAC], family involvement, systematic and comprehensive assessment procedures, goal setting, and detailed planning and evaluation” (Copley & Ziviani, 2007, p. 110).

A variety of practical resources also exist to help identify an access solution that meets a person’s unique needs (Anson, 1996; Arthanat, 2007; Bauer et al., 2014; Cook & Polgar, 2013; Hoppestad, 2006; Scherer et al., 2007; Simpson, 2013). These resources may fit into an existing alternative access assessment model and take many forms including informational resources such as website articles (Wisconsin Assistive Technology Initiative, 2018), online forums and questionnaires (QIAT, 2012; OCALI, n.d.), tutorials (Sowers & Wilkinson, 2023), and statewide assistive technology programs (Indiana PATINS Project, 2019; PaTTAN, 2024). Additionally, there are software tools and procedures, including Compass software (Koester Performance Research, 2019), to gather specific performance data when trialing different candidate alternative access methods (Dumont et al., 2002; Hoppenbrouwers et al., 2014; Koester Performance Research, 2017, 2019; Koester et al., 2006, 2007; Nguyen et al., 2023). Recently, screening and assessment tools such as those developed by Pitt and Brumberg (2018a, 2018b) have expanded systematic feature matching processes to guide selection of newer alternative access methods such as BCI-controlled AAC systems.

Impact of using the available models and tools

Do we as a field have evidence that using the available models and tools for alternative access assessment leads to better outcomes? Intuitively, it makes sense that using a systematic, quality alternative access assessment process would benefit both the person who uses alternative access and the practitioner conducting the assessment. A well-defined, systematic process has the benefits of being repeatable, teachable, and transferable to aid in consistent and comprehensive service delivery. For service providers, following a clear procedure can build confidence, reduce worry, and minimize the opportunity for important assessment components to be missed. For all stakeholders, including people with physical disabilities, a shared process that clearly defines roles of team members can enhance transparency and effective collaboration.

Although we searched the literature extensively to identify studies that evaluated the outcomes of systematic alternative access assessment, much of the existing research describes the assessment process, rather than evaluates the process. For example, Sowers and Wilkinson (2024) used a collective case design to understand the decision-making process of SLPs completing alternative access assessment with children with cerebral palsy. Although important insights were gleaned from this study, it is unknown how the decision-making and assessment processes impacted clinician or user outcomes. The limited outcome data presently available in the literature may not be surprising as many AAC assessment models do not specify or measure explicit outcomes (Burnham et al., 2024). For those models that do include outcome specifications, outcomes take many forms, including (a) performance of the person who uses AAC, (b) quality of life ratings of the person who uses AAC, (c) AAC usage, (d) AAC usability (i.e., quality and quantity of functioning), and (e) abandonment rate. If the assessment models used do not include outcome measures, it may be difficult to understand their impact.

However, there are a few studies that demonstrate a positive effect of using a systematic assessment method. Karnezos (2018) found that after using the AAC Assessment Checklist and the SETT framework, AAC assessment teams in schools included more components in their AAC assessment plans, both in case example situations and with real students, compared to only using the SETT framework. When using both the checklist and SETT framework, teams were also quicker developing their plan in the real-world school context (Karnezos, 2018). In a different study, after using the Lifespace Access Profile (LAP) and Lifespace Access Profile Upper Extension (LAPUE), teams formulated more educational plan goals and those goals more clearly included assistive technology (Copley & Ziviani, 2007). In the same study, 100% of providers and 83% of caregivers reported that the assessment process was beneficial to the team. Additionally, school staff and caregivers of children who use AAC reported increased involvement in the assessment process, increased AT awareness, knowledge, and skills, and increased confidence in using AT (Copley & Ziviani, 2007).

Evidence also suggests that people who receive assistive technology due to a systematic assessment process report less technology abandonment (Bernd et al., 2009) and greater satisfaction (Ranada & Lidström, 2019). A main reason for abandonment can be poor assessment of the recipient’s needs and preferences (Scherer et al., 2005; verza et al., 2006). When a team-based process with structured assessment procedures is conducted, it is more effective especially when family and caregivers are highly involved in the process (Ranada & Lidström, 2019; verza et al., 2006). Components that appear to be especially essential include the involvement of the recipient of alternative access, completion of a needs assessment, provision of training, and ongoing follow-up (Ranada & Lidström, 2019; verza et al., 2006).

Discussion: how to choose the right access method?

Figure 2 summarizes the major findings from our review of alternative access assessment. Existing evidence suggests that use of a systematic approach to alternative access assessment reaps benefits for everyone involved, including the user with physical disabilities and the service providers on the team. Specific benefits reported in the literature include: (a) improved team performance during the assessment process, (b) improved knowledge, skills, confidence, and involvement of team members, (c) less technology abandonment, and (d) greater satisfaction. Elements of a quality assessment process include:

• a team that involves skilled providers and centers the user with physical disabilities and their family

• systematic procedures to gather information, identify candidate solutions, trial and select from those candidates, implement selected solutions, and provide training and follow-up.

Figure 2.

Infographic on alternative access assessments.

Another key component is to recognize the need for low- or no-tech access methods, as a complement to methods that require specific hardware and software technology for their use (Beukelman & Light, 2020). While higher tech methods can offer powerful functionality, they come with challenges as well. They might break, or be difficult to set up, or not work well in every situation and environment. Author KJ relates the experience of an eye gaze user with ALS, who also used a ventilator and could not rely on speech for his communication: “His nursing home had staffing changes, and the new staff didn’t know how to place the eye gaze computer. Eventually they turned it off and pushed it into a corner. He had no backup method of communication and ended up with serious bedsores and other complications. Finally, a friend brought him a no-tech communication board, and that started a process of healing and recovery”. There are many things wrong with this picture, but a critical one is the lack of a communication method that could be used quickly and reliably.

Although numerous benefits exist when AAC assessment teams involve the technology recipients and use a systematic assessment process, gaps exist limiting the field’s understanding of how to optimize and translate these processes in real-world and diverse contexts. First, although many alternative access assessment resources are available, it is unclear how or if these processes and tools have been successfully adopted in clinical practice. The research suggests that practitioners often describe using either no assessment model or a reliance on a highly personal, internal framework when completing alternative access assessments (e.g., Burnham et al., 2024; Friederich et al., 2010; Koester et al., 2023; Lynch et al., 2019; Ranada & Lidström, 2019; Sowers & Wilkinson, 2024). One contributing factor may be a result of limited pre-service opportunities to learn how to conduct systematic alternative access assessments. Costigan and Light (2010) reported that most practitioners receive very little preservice training in technology access and others cite practitioners’ lack of confidence in providing AAC services due to their inexperience (Lindsay, 2010; Sutherland et al., 2005). Ensuring that future occupational therapists and physical therapists receive high quality pre-service AAC training and future practitioners such as SLPs and special educators receive high-quality training in motor function, AAC, and alternative access remains paramount to build competent and effective interprofessional teams that can support people with physical disabilities who require alternative access systems.

Second, more research on the effectiveness of alternative access assessment models and tools is warranted. Specifically, future studies should measure the impact of the use of these models/tools on (a) people who AAC, (b) their partners, and (c) the practitioners who lead alternative access assessments. Limited information exists regarding the comparative effects of different alternative assessment models within clinical practice (Burnham et al., 2024). It is currently unknown which model is most effective within different contexts, how to best adopt alternative assessment models into real-world settings, or how use of these models differentially affects people who use AAC and their partners. Studies designed to compare the effectiveness of alternative access assessment models on both short-term and long-term outcomes would help guide clinicians and increase adoption of relevant, evidence-based techniques.

Existing tools, models, and systematic assessment processes may also fall short of clinical translation because they are “comprehensive and overly complicated” (Bernd et al., 2009, p. 156). This can be a barrier to adoption into rehabilitative and educational practices, especially to teams who face practical time constraints related to their workplace and reimbursement structures. For instance, although the Switch Access Measure (Nguyen et al., 2023) is a highly reliable and comprehensive tool to assess switch access techniques, it requires at least 13 hours of training for a practitioner to use in a clinical setting. This time dedication for training may not be feasible for practitioners. A need exists for evidence-based alternative access tools and processes that fit within the workflow and lives of people who use AAC and their team. Thoughtful, intentional implementation and dissemination strategies are warranted to translate the use of effective, systematic alternative access processes to real world settings and including people who use alternative access as well as service providers will be vital to the success of such efforts.

As one piece toward addressing these needs, the RERC on AAC is developing an app called Access Navigator (Koester et al., 2023). This is a web-based tool that supports the quality assessment process defined at the start of this section and helps individuals with physical disabilities and their providers identify efficient, accurate, and desired access methods. An early version of this app has been completed and is currently being evaluated with people who use alternative access and their clinicians.

Part 3. Future research directions

Here we try to summarize priorities for future research directions in alternative access for people with physical disabilities. These priorities emerged from the literature reviews and insights of this author team as well as those identified by participants at the Future of AAC Research Summit. Our literature reviews revealed a paucity of research and development focused on alternative access; there are opportunities for impactful work in all areas of the field, and it needs to be viewed as a priority with AAC and assistive technology.

Inclusive research & development

Essential to all future work is the involvement of individuals with disabilities in all aspects of research and development. This goes beyond simply including end users in evaluation studies (although as noted above, progress there is definitely needed as well) to include full inclusion in requirements, design, and implementation. A fully inclusive approach to research and development will help ensure that research teams are asking the most important questions and facilitate the translation of project results out of the lab and into real-world use.

Understanding performance of existing access methods

There is a significant need for more data on how well all types of alternative access methods work for those who use them, especially the following: touchscreen typing (such as on iPads), eye typing, speech recognition built-in to computing platforms, recognition of dysarthric speech, and new implementations of Morse code.

Methodologically, the field needs a more comprehensive way of evaluating meaningful success. This includes better methods to quickly and easily measure performance, in real-world settings, and to help people use that information for better decision-making. There is a need to include individuals who use alternative access in developing performance metrics that they find useful and important.

Improved design and function

There is a significant opportunity to design, evaluate, and deliver new and improved alternative access methods. This includes improvements in existing access methods as well as development of entirely new types of access methods.

Improvements in existing access methods

Some aspects of current access methods that are ripe for improvements include the following:

• Simpler and more robust eye gaze systems. As author KJ describes it, “Eye gaze is finicky. While writing this paragraph, I have calibrated my system three times and still have lots of frustrating errors”.

• Better integration across devices and applications, so users can use the same alternative access technology for their laptop, phone, wheelchair, etc., and use it for games, virtual gatherings, or any application.

• Better performance and adaptation across different conditions, such as varied environments, variations in user’s position or condition.

• Lower cognitive demands for using complex alternative access methods, such as speech recognition and eye gaze typing.

Application of artificial intelligence for new types of access methods

The potential for artificial intelligence to enhance alternative access is tremendous. Here again the inclusion of end users throughout all phases of a project will be crucial to avoid “solutions” that miss that mark. One example is the integration of large language models such as ChatGPT to generate text and other content more quickly and accurately. Bridging voice, a nonprofit serving individuals with ALS, is one organization that is already integrating ChatGPT into AAC apps. Their Conversation Enhancer helps transform simple sentences into longer, more robust conversations (see video demo at https://www.youtube.com/watch?v=fLwL7lvGups).

A second example is intuitive technology that can adapt to users based on performance or other indicators of success. This idea has been described in the past (Trewin, 2004; Wobbrock et al., 2011) and implemented in fairly simple ways to date, such as automatically changing Windows keyboard settings based on the user’s moment-by-moment typing data (Koester & Mankowski, 2015). With advances in AI and dynamic modeling, the field may now be able to implement this concept much more fully and successfully. Development of intuitive technology that adapts to the user could benefit individuals as they learn new access methods, or require changes to their access over time as they recover (e.g., recovering movement abilities after brainstem stroke) or decline in their physical abilities (e.g., losing movement abilities with a progressive condition such as ALS). This technology should incorporate the metrics of importance to the individual user and offer a way to control whether changes are made to the access method based upon performance.

Enhancing service provision

Enhancing service provision is one important way of more effectively leveraging the alternative access technology that is already available. This includes conducting systematic, effective assessments, that lead to sustainable, robust interventions. The priorities for future research and development here include:

• Investigate effectiveness of procedures for assessment and provision of alternative access technology.

• Build awareness for all stakeholders of reasonable expectations and best practices, including a moving past a so-so solution to seek something more optimal.

• Make it easier for end users to advocate for access methods that they are interested in trying and for providers to know what is available and where they can go to try it out.

• Make it easier for practitioners/teams to use the processes that have been shown to be effective, especially processes that center on the user and their needs.

• Make it easier for users to receive ongoing support as needed.

• Develop tools to support effective decision-making, such as the RERC Access Navigator project (Koester et al., 2023).

Other priorities

It should be noted that these priorities most directly address issues of technology design and function, as well as service and support provision, as those are most directly within the scope of this paper. These are certainly important priorities, but there are also barriers that relate to psychological attitudes, societal attitudes, and public policy (Howard et al., 2022). These barriers, too, need to be reduced through continued inclusive research, development, and advocacy.

Conclusion

Today’s alternative access methods can be an impactful means of enabling participation for people with physical disabilities. Yet there is room for improvement, both in the design and function of the access methods, as well as their associated service provision and support. Fully inclusive research and development is needed to identify and implement these improvements. The ultimate goal is that more people receive effective access methods to meet their needs, allowing them to use any type of computing device easily and efficiently, for all types of tasks, in all types of environments.

Data availability statement

The data from the review of typing speeds are publicly available at: https://docs.google.com/spreadsheets/d/1y19Prix8Hc8bDLWifp1veCxdLQvSmnDLwQNETiY6OxQ/edit?usp=sharing

Supplemental data for this article can be accessed online at https://doi.org/10.1080/07434618.2025.2499676.