Analysing Disability Descriptions and Student Suggestions as a Foundation to Overcome Barriers to Learning

Describing and understanding disabilities

The diversity of views on definition and language, and relationships between how disability is understood and managed, are the subject of decades of research and debate. Before approaching technologies and solutions, it is important to recognise that disability can be described in very different ways, that these communications form an essential part of overcoming barriers to study, and that they reflect underlying conceptions which impact individuals, organisations, and society.

Descriptions of disability can focus on explaining how functioning in life activities is somehow limited or difficult (Leonardi et al. 2006). Or on the barriers experienced due to the design of environments or other people’s actions (Oliver 2013). In many cases, however, disability is expressed in terms of medical diagnoses and conditions (Dirth & Branscombe 2017). Disability is an umbrella term for a diverse range of challenges, including mental health and cognitive differences which can require quite different understanding of physical disabilities or long-term health conditions. Many individuals can be characterised as having multiple categories of disabilities, impacting further their chances of success in their study (Richardson 2010).

While there are commonalities in the experiences of diverse disabled people, negative perspectives towards disability often discourage any desire to identify in this way (Finkelstein 1993). Many people who do not consider themselves disabled or feel uncomfortable or angry about being labelled as such, are however bound up with this term if they wish to gain support that they are legally entitled to (Thomas & Woods 2003). Several studies provide evidence that disabled people have varied preferences as to how to describe themselves. What is meant by disability is interpreted and expressed differently between individuals, and their preferences around this can change according to the context of the communication (Lister, Coughlan & Owen 2020).

Different conceptions of disability impact the understanding of support needs. A focus on identifying barriers, rather than on individuals and their medical conditions, is argued to shift the emphasis towards institutional and social responsibility for removing barriers, rather than viewing disability support only in terms of individual diagnosis and response (Oliver 2013). Despite the increased adoption of approaches that aim to enhance the general accessibility of web content and learning, most prominently through the Web Content Accessibility Guidelines (WCAG) and Universal Design for Learning (UDL), there remains the recognition of the need to understand, support and adjust according to individuals, and technology may make this easier (Iniesto, Rodrigo & Hillaire 2023).

A functional perspective that focuses on the difficulty or inability to achieve specific tasks promotes the consideration that people with very different conditions may benefit from similar tools or support. For example, difficulties in reading text, related to either dyslexia or visual impairment could be overcome by spoken versions of learning material. The functional perspective also helps to explain how mainstream technologies can be repurposed as assistive tools, as they support a task to be done in a way that is relevant to all people, but particularly valuable in overcoming barriers. For example a voice recorder or smart pen is a useful memory aid for many people, but more essential for people who find memorising information in the moment more challenging (Bouck et al. 2012). This applies regardless of any medical reason for their memory challenges.

While medical diagnoses remain central to disability support processes in education and everyday life, approaches have been established which aim to standardise a functional perspective, such as The World Health Organisation Disability Assessment Schedule (WHODAS) survey instrument (WHO 2012), and the development of defined functional needs to underpin further development of WCAG (W3C 2021). Other work in healthcare has focused specifically on structuring the process of matching persons to suitable assistive technology (e.g. Scherer & Craddock 2002), and guidance for conducting assistive technology assessments has been developed in specific education contexts such as US schools (e.g. ASNAT). This highlights multiple elements to understand before identifying solutions, including a detailed analysis of the student and their past and present assistive technology use, the current barriers faced, and aspects of the school environment and learning tasks (Gierach 2009).

While there have been efforts to standardise functional needs as a basis to support online and digital learning (explored further in later sections), this is not widespread. Also, while assistive technology is an important element, there is a wider space of human support, adjustments and strategies that can be just as important to overcoming the barriers student face. Processes to identify appropriate support for students, such as the needs assessment conducted as part of the Disabled Students’ Allowance in the UK, still require a medical diagnosis of disabilities, but consider the functional needs of the student and the potential support, technology or adjustments that could meet this.

Current challenges

While there appear reasons to be optimistic about the potential to overcome barriers to learning for disabled people, there remain persistent gaps in access to education and employment. Worldwide, disabled people get an average of 2.2 years less schooling. Large gaps in adult literacy exist in almost every country with underrepresentation in tertiary education and employment even in richer nations. Disabilities compound other inequalities, with disabled women less likely to complete education (UNESCO 2018). However, as noted already, treating disabled students as a homogeneous group is problematic, and more detailed analyses within The Open University highlight how completion and pass rates vary significantly according to the categories of disability a student discloses (e.g. Richardson 2015a, 2015b, 2015c; Open University 2020). Analysis of feedback survey data according to these categories also evidences that their experiences and the barriers they face are different (Coughlan, Ullmann & Lister 2017). Therefore, there is a need to increase awareness of the diversity of barriers and take approaches that help to address this diversity.

The diversity of disabilities, the wide range of interactions expected in learning activities, and the changing technological landscape, all present a messy and multifaceted space. Worldwide, there is a lack of support for the use of assistive technology, inhibiting the potential of this to overcome barriers (WHO & UNICEF 2022). Despite the wider appreciation of the social model in legislation and in institutional rhetoric, responsibilities are often unclear. Educators, technologists and institutional decision-makers often lack an understanding of disability and accessibility (Bong & Chen 2021; Lister et al. 2022) and with other competing demands, create barriers which disabled students then need to overcome.

Further challenge arises from the wide range of learning tasks that students could face barriers with. Erdem (2017) uses a literature review to summarise the range of uses of assistive technologies, finding categories of communication; reading; writing; mathematics; support for seeing and hearing; seating, positioning, and mobility; social skills and leisure; daily living; organisation; and computer access. Features of the learning and the environment also change the need for technologies. For example, Fichten et al. (2009) found that diverse disabled students made use of scanners and Optical Character Recognition (OCR) tools, which could enable text from books or images to be read digitally. Such technologies can be a lifeline where materials are inaccessible but these tools could be redundant if materials are all provided in accessible web pages or digital documents.

A common approach is therefore for persons with knowledge of disability, assistive technology and study to assess the needs of students on an individual basis. Depending on the context of these conversations, personalised adjustments may be made to teaching and assessment, or technology, training and mentoring activities could be prescribed, with any costs of these covered either by the institution, third parties, or the individual. But the required expertise is limited and access to support is not evenly distributed. There are calls and initiatives to further develop the profession of Assistive Technologist, to create a larger, more effective workforce to work with disabled people in this capacity (Norrie, Waller & Hannah 2021; Slaughter & Waller 2022).

The responsibilities and capacity of educational institutions, governments and third-party organisations to support disabled students vary around the world and are changing. In the UK for example, changes to the Disabled Students’ Allowance aim to move more responsibility away from government-funded support towards the institution and individual. For the individual, there is a substantial administrative burden around explaining their disabilities and gaining support (Disabled Students UK 2022; Policy Connect & Higher Education Commission 2020). This burden produces negative impacts on students, including on their mental health and wellbeing, and their time and ability to study (Coughlan & Lister 2018).

While it is challenging for any single intervention to untangle the complex socio-technical phenomena and bureaucracy that lead to administrative burden, conversational agents and AI do present some compelling potential to address these challenges. For example, administrative burden is commonly increased by having only limited, or inaccessible, channels for completing a process (e.g. a paper-based form or phone call). Conversational user interfaces can offer a more flexible and accessible means of communication, particularly where systems can interpret input and bridge the gap between how the person wants to communicate and the often-rigid requirements of the administrative process. These interfaces also offer an alternative to the negative affective responses which arise from repeatedly filling in complex forms and repeatedly explaining sensitive personal information to other people. With further development and integration, such systems could complete multiple administrative processes and requests through one conversation, and could offer guidance and answer questions at the same time (Iniesto et al. 2023).

In a varied and shifting landscape of responsibilities, one aspect that is unlikely to change is that disabled students who are able to explain their disabilities and are knowledgeable about adjustments, strategies and tools that can help them study should be more likely to succeed. This is particularly important when we recognise that personal development involves transitions, such as between schools, universities and jobs, or at a lower granularity, between courses, teachers or environments. Disabled people benefit from a greater ability to self-advocate effectively and solve problems they face due to these changes (Bremer, Kachgal & Schoeller 2003). A further finding across multiple studies and contexts is that simply providing students with assistive technology is not sufficient – issues such as inadequate training or support for them to use these technologies present further barriers to study (McNicholl et al. 2021; Seale et al. 2015).

Providing relevant and useful information and supporting disabled students to develop in this way is not trivial. Research has highlighted their concerns, such as that recommendations for technology are made based on ‘labels’ rather than individual needs and preferences, and that support needs to be timely and built on existing digital literacies (Seale, Draffan & Wald 2008). As such, further research and development needs to be underpinned by greater understanding of how disability, barriers and solutions are communicated, as well as how innovations offer ways to overcome these barriers.

Opportunities for innovation to overcome barriers

There is a history of research that has explored the potential of new technologies to represent and overcome barriers caused by disability. The breadth of work done to develop assistive technologies and design for accessibility is beyond the scope of this article, but a particular thread of work has focused on the processes by which barriers are understood and support is offered. This has highlighted possibilities such as adapting learning content according to identified needs and preferences, crowdsourcing and connecting people to overcome barriers regardless of distance and using artificial intelligence (AI) to understand the person or assist in a specific situation (Iniesto & Rodrigo 2016).

Boticario et al. (2012) describe how an adaptive learning approach underpinned the EU4ALL project and framework. While following standards and guidelines to support access to systems and content is important, they highlight that it is not sufficient, as this does not consider how diverse learners get to a position of having suitable individualised study experiences. To achieve this, elements such as a user model, representing goals, needs and preferences, a recommender system, highlighting appropriate actions, and a metadata repository, containing information about the accessibility and nature of a learning resource, were devised and evaluated. While successful in illustrating the potential, the sustainability of such an approach is noted to be reliant on continued interoperability and following of standards by all actors involved and is challenged by the breadth and scope of learning management systems and courses.

Making recommendations based on profiling learners and capturing the accessibility features of Open Educational Resources (OERs) was more recently explored by Elias et al. (2020). Their system asks users about aspects such as which assistive technology they use, and whether they have requirements for alternative media, such as text versions of audio content, alongside other questions such as their goals and educational level. Alongside this, they assess and score OERs based on their metadata. A small-scale evaluation with accessibility experts suggested value in this approach.

While this literature suggests promise for the idea of automatically adjusting learning based on created profiles of accessibility needs, McAndrew, Farrow and Cooper (2012) note the difficulty of students providing adequate self-description, and consider that such systems should support information flow, keeping humans in the loop, rather than complete automation. Neither would making a Virtual Learning Environment (VLE) adaptive and accessible remove all the barriers students can face. Learning occurs across a number of platforms, and students use a changing plethora of technologies across multiple spaces to support their learning, including mainstream web and mobile apps as well as assistive technologies (Seale et al. 2021).

Outside of education, there are various examples where researchers have developed socio-technical approaches to overcoming disability-related barriers. These include crowdsourcing approaches, for example, Liu et al. (2017) present a project where they engaged wheelchair users to identify the types of barriers they faced in navigating routes through urban environments, and then provide a mobile application where such barriers can be captured by members of the public and then included in map-based route planning. Lasecki et al. (2013) describe the development and enhancement of Remote Sighted Assistance systems that provide crowdsourced support at a distance to blind people, in their case by sharing a video stream from the person’s device and supporting sequences of questions to be asked and answered to share visual understanding from sighted to blind persons. While the potential of artificial intelligence in this space is recognised, human involvement is considered essential, not least because AI is not considered reliable in the visual recognition of objects in complex environments. Exploring this further, Lee et al. (2022) speculate that challenges in remote-sighted assistance could be best resolved by building new models of human-AI collaboration.

Other research looks specifically at the potential of AI in understanding the person and providing guidance and support. Again, there is diversity in the contexts and populations this aims to support, from systems using games and assessment techniques to guide people with cognitive disabilities towards assistive technologies appropriate for them (Heumader, Murillo-Morales & Miesenberger 2022), to a virtual assistant that supports disclosure and profiling for disabled students in higher education. Evidence from a trial of one virtual assistant ‘Taylor’, suggests that disabled students preferred a conversation with a virtual assistant to complete forms when disclosing their disabilities and that they would like to be able to use such a system to receive personalised guidance and to complete other support-related processes (Iniesto et al. 2023). Researchers highlight that offering appropriate recommendations based on individuals and their contexts, simplifying language and supporting alternative communications, are areas of particular opportunity for AI in assistive technology (Draffan & Heumader 2020).

Bringing these threads together, in order to address challenges of educational access and equity, there is a need to improve access to guidance and information sharing. AI and crowdsourcing technologies present opportunities in this space, but would benefit from a clearer understanding of the information needed to understand disability in the context of learning, and the types of guidance that disabled students could find useful and relevant to them.

Research questions and method

Placing the focus on students, the research questions are as follows:

• RQ1: How do students describe their disabilities?

• RQ2: What types of suggestions do students share about the support that works for them?

The objective was to obtain information that is not currently gathered by The Open University on suggestions to inform an AI virtual assistant with knowledge base. The limitations of university processes included the inability to use existing institutional data without considering potential data protection barriers and ethical aspects related to personal information in students’ profile. Therefore, to explore the research questions an open-ended online survey was designed. The survey was designed by the research team to capture how students at The Open University describe and understand their disabilities, as well as to assimilate different types of suggestions for student support. The questionnaire is available in the Appendix. To reduce concerns about privacy and the handling of sensitive personal information, the survey was completely anonymous, with students not asked for any identifiable details in the questions. Ethical clearance for this research was approved by the University’s Human Research Ethics Committee (HREC/4455/Coughlan) and Student Research Project Panel (SRPP/2022/2235).

This paper focuses on a thematic analysis of the free text disability descriptions and the suggestions made by students. Following the approach described by Braun and Clarke (2019), familiarisation with data led to an initial set of codes, which were then organised and developed under themes, then defined and summarised. Two members of the research team were involved in the coding, one taking a primary role with the descriptive analysis and the other with the analysis of the suggestions, with each of these then being reviewed with the other. NVIVO software was used to support the analysis.

Sample

An email invitation to the survey was sent to 3,000 current students who had disclosed disabilities to The Open University. For that original sample and following Sapsford & Jupp (2006) we used a quota sampling approach, which is a non-probability sampling method that divides the population into non-overlapping profiles to keep the sample balanced among different types of declared disabilities. 143 responses in total were received, where consent was fully given in 138 and so these were taken forward to the analysis stage. Of these, all 138 had answered the free text disability description question, and 108 provided one or more free text suggestions of what supported them in their studies. The response rate was lower than those found in literature in similar cases (Bessaha et al. 2020; Murphy, 2021). However, even though the response rate was lower than expected and we could not control the balance between types of disabilities in the final sample, the class size can be considered valid and reliable following Nulty (2008) and with the lens of the constructivist perspective acknowledged in this research (Bada & Olusegun 2015).

The respondent population comprised 99 females (72%), 38 males (28%) and 1 who preferred not to say. Most participants were between 26-45 years (38.1%), and the more representative subject areas were Psychology (34, 24.8%), Arts and Humanities (30, 21.9%), Social Sciences (16, 11.7%) and Computing (14, 10.2%). The numbers who stated that they fitted with each of the formal university categories of disability are shown in Table 1. Other disabilities reported included fibromyalgia, brain injury and heart disease. Note that participants can choose more than one of these categories if they have multiple disabilities.

RQ1: How do students describe their disabilities?

The thematic analysis of question 5 (see the Appendix) led to the themes and codes in the thematic map included in Figure 1 having two levels of depth.

Figure 1

The analysis identified five different ways in which disability was discussed in the responses. Medical and health related ways of describing disability were most frequent, but descriptions of functional barriers and limitations, and of forms of support to overcome barriers, were also commonly provided. Experiential elements, such as explaining past events, and administrative aspects of disability such as diagnosis or disclosure processes, were also described. In the remainder of this section, we describe these five themes with examples of the codes found within them.

RQ2: What types of suggestions do students share about the support that works for them?

The thematic analysis of questions 8-20 (see the Appendix) led to the themes and codes in the thematic map included in Figure 2 having in this case three levels of depth. The analysis produced three main themes.

Figure 2