What Matters in Accessible Written Communication for Neurodivergent People? A Scoping Review

Selection of sources of evidence

The selection was done in two phases to assess if the study met the inclusion criteria. First, two independent reviewers selected them based on the title and abstract and exclusion/inclusion criteria with the support of the Rayyan automation tool. After this, the selected studies were retrieved and underwent a second screening, where two independent reviewers read the full text to see if they met the inclusion/exclusion criteria. If disagreements occurred, the reviewers got together and discussed the paper to reach an agreement. Overall, 242 studies were screened in the first phase and in the second phase, 93.

Synthesis of results

The studies were grouped according to the topics they covered. Additionally, the theories and approaches used, the population included, the participants’ involvement in the study, and the recruitment strategy related to the methodology were identified. The evidence will be presented both in a narrative format and through tables.

Understandability and readability

Eight topics emerged when discussing the understandability and readability of texts or information: lexical simplification system, complex word identification, natural language processing, medical information, defining complex words, literacy and ruling passions and tone indicators.

The works of Alarcon et al. (2023, 2021, 2020a) and Moreno et al. (2019) focus primarily on a lexical simplification system that uses complex word identification and natural language processing to identify and predict complex words to have the texts on webpages in compliance with several guidelines (Easy to Read, Plain Language and Web Content Accessibility Guidelines (WCAG)) to increase their readability and understandability. Overall, the results of these studies showed that machine learning approaches and linguistic databases can be helpful in developing a lexical simplification system that identifies complex words and proposes synonyms and definitions of those words that were identified by following accessibility guidelines.

In line with this, the study of Alarcon et al. (2020b) presents a WordSense Disambiguation system that provides a definition of a word, taking its context into consideration. Results show that the system performed positively in giving correct definitions of complex words to improve the cognitive accessibility of texts. However, coverage was low, and there were missing coincidences, as some definitions were in a different grammatical form than the words the system gave.

Another topic that emerged when understandability was mentioned was regarding the understanding of medical information. Here, studies were about the understanding that People with Intellectual Disabilities (PwID) had about their diagnosis and prognosis of cancer and how much they were told (Tuffrey-Wijne et al. 2010), and understanding medical information (Baumbusch et al. 2014; Mander 2016).

Regarding the understanding of the diagnosis and prognosis of cancer, although most of them were told they had cancer, they did not receive any help to understand the implications of the diagnosis or prognosis. Several other aspects were noted, such as: people with severe or profound ID were more likely to be protected from the truth; understanding was linked to the person’s cognitive ability, life experience and truth-telling. Furthermore, the authors state that the models used to give bad news are not suited for PwID (Tuffrey-Wijne et al. 2010). As for understanding medical information, results show that, sometimes, it was difficult for the participants to understand the information. In the study of Baumbusch et al. (2014), some participants indicated that one of the strategies the doctor used to explain to them was through plain language, while others said that they used alternative methods, such as asking the receptionist for more information or bringing someone else to accompany them. Complementary to this, the results from Mander (2016) imply that simple questions may lead to a lenient answer or misinterpretation, which will require the reformulation of the question and that the comprehension of the message will be influenced by the quality and accuracy of the accessible information resources, while also recognising that said resources might focus on what is easier to understand rather than on the topic/issue that is more challenging.

Regarding the autistic community, one strategy to understand texts is through tone indicators. According to the participants of the study, tone indicators are tools for (1) clarify the tone/context of the message, (2) emphasise the expression in the text to help express emotions and feelings clearly, as confusion regarding certain icons can appear (here they appear as a complement, e.g., the use of ‘/s’ to signify writing something sarcastic); and (3) minimise misinterpretation of messages, as when reading written messages there is always the risk of misinterpretation of the tone used (Christanti et al. 2022).

Communication strategies

Communication strategies were mostly identified when referring to accessible remote work, interviews and achieving effective communication.

In the study conducted by Das et al. (2021), when referring to collaboration and communication working remotely, neurodivergent professionals indicated having to adapt to the complexities of multimodal communication to ensure effective communication and understanding; for instance, typing a message instead of speaking, others use closed captions and lip reading to help with processing information via video/audio channels. Contrary to these strategies, those who have dyslexia prefer to communicate through speech rather than by writing, which proves a challenge when, in an online meeting, there is no dedicated time for questions and most of the communication is written. To help them with these, participants refer to using dictation tools to write, screen readers, or read-aloud features. Having access to meeting material before and afterwards was also mentioned as being helpful.

In interviews with PwID, Hollomotz (2018) identified that some of the communication strategies could be: asking questions in plain language accompanied by reference tools, such as picture cards and photo-story vignettes; and adapting the depth of the questions according to each participant’s willingness or capabilities.

Likewise, the strategies used to help contextualise the question asked should vary according to each person’s characteristics. For instance, in the study, the use of vignettes was helpful for some participants but not others. In these cases, a triangulation technique (i.e., what the participant was saying, a secondary confirmatory source, and participant observation) was adopted. In all, it is necessary to adapt the questions to the person, as keeping it too simple may interfere with the answers of those who have more to say and feel patronised. At the same time, if questions are too complex, it could generate interview errors.

Related to this, one of the communication strategies can be relating the written content to the ruling passions of neurodivergent people. According to Bailey (2023), if neurodivergent people have access from a young age to quality fiction that relates to their passionate interests, it could help them comprehend different personalities, motivations and relationships. To this end, the author recommends that they have access, contact and create multimodal texts to help increase their knowledge of their ruling passions. This way, the content is interesting for them, which in turn can also mean being more accessible to them, as it is about a topic of interest.

Accessible information

Accessible information is not only important for people with disabilities, in general, but also for people with different languages or educational backgrounds or in any other situation where oral and written communication are hindered. When mentioning accessible information, studies focused on the context of visual information and the specific needs of people with complex communication needs.

The studies of Niediek (2016) and Cameron and Matthews (2017) focused on visual information to complement written text. Niediek (2016) argues that illustrations and symbols have an important contribution to accessible information, mostly because through more concrete or detailed illustrations, it is possible to transmit more information. However, this can also represent a barrier for early-stage readers and be a bad facilitator option as they need to be familiar with the communication method. The same was maintained in the use of pictures and graphic signs, as they relate to a determined cultural knowledge acquired in childhood and interaction processes. As such, their use can also be limited if they follow the principle of ‘single-meaning picture’, as it cannot represent every concept or idea and can lead to misunderstandings. For this reason, it should be used in conjunction with plain or easy-to-read text.

From another perspective, Cameron and Matthews (2017) presented the process of developing an accessible communication resource, ‘Keeping Safe’, where the communication framework ‘Talking Mat’ was applied and found that it helped people with learning disabilities (PwLD) to reflect and express what was important to them at that moment.

In relation to text and print-based communications, Collier et al. (2012) denote that people with complex communication needs prefer to receive documents in plain language and have alternative formats available (e.g., with large fonts, without staplers, or send it to them in a format compatible with screen readers). Although service providers said that people with complex communication needs needed someone to help with reading and understanding documents, respondents with complex communication needs said to prefer accommodations that assist them to be more independent.

Digital accessibility

In the studies selected, those who mentioned digital accessibility were usually in the context of a voting interface (Harley et al. 2016; Summers et al. 2014), a web-based decision aid (Bogza et al. 2020), and the specific needs of individuals with CCN (Collier et al. 2012).

In all, the studies show similar results when it comes to digital accessibility. To ensure that the interface was accessible, participants with low literacy and mild ID mentioned that it was necessary to put instructions in fewer and simpler words, avoid the use of the word ‘choices’, increase the font size and spacing between lines, use a vertical layout, have additional information available if needed, and the possibility to deselect their vote (Summers et al. 2014). These results are similar to those obtained by Harley et al. (2016), as participants with language difficulties preferred the version in plain language and recommended the reduction of instructional text. Notwithstanding, participants also added that the text should be accompanied by illustrations to help explain the visual aspects of the ballot. Other interesting results obtained from Harley et al. (2016) were that dyslexic participants preferred text written in Helvetica font over the two fonts specifically designed for dyslexic people (Lexia Readable and Open Dyslexic).

These results are also consistent with those obtained by Bogza et al. (2020) in relation to simplifying the content. Concretely, the content should be simple but not simplistic, which means that the content should be simplified to a level that still retains enough detail to support informed and shared decision-making. It also means that the text should use clear and comprehensive words to include only the meaningful content, remove the unnecessary details and make the content easy to remember. For this, the results show that pictograms are a helpful strategy to assist in remembering the options available and to resume the information.

The same results were also obtained regarding digital accessibility for people with complex communication needs. According to Collier et al. (2012), this means having the option to: change font sizes and colour contrast, read out loud what is on the screen and access online plain versions of the documents. Moreover, other identified recommendations included having the option to change columns to a single one to facilitate navigation, a version with graphics to help with readability, and a graphic-free version to facilitate navigation and use of screen readers.

Employment

Regarding employment, communication accessibility was mentioned in terms of experiences in the workplace regarding the recruitment process, internships and in an image and data annotation platform.

Starting with the results of the identified needs of the communication process during the employment process to meet the needs of autistic individuals, Tomczak et al. (2021) found that in the recruitment phase, job advertisements should be modified, and job interviews should be less structured, as autistic people tend to find it difficult to brag about previous job experiences. During the onboarding stage, the results show that having a mentor, buddy or job trainer accompanying them would be beneficial to give confidence and reassurance during stressful situations. Lastly, to support job retention, employers should use clear messages, direct and sensitive feedback and limit emotions during the communication process. These last findings go in line with what Shute et al. (2012) found is necessary to have in an effective and inclusive communication. Adding that, to help have a stress-free communication process, organisations should use written and non-direct electronic mediated communication forms.

Still, regarding the communication preferences of autistic individuals and workplace experiences, the results from Remington et al. (2022) suggest that although autistic and non-autistic interns go through the same difficulties, some were specific to the autistic interns, such as: the report of prior negative workplace experiences – which has implications for their performance and self-confidence, mental health and well-being; and communication challenges that happen due to preferences in communication styles, such as preferring written communication via email instead of verbal communication, which helps reduce difficulties that can emerge in interpreting implicit social cues. Autistic participants also highlighted the necessity of having a tailored recruitment process and raising awareness with managers about their (individual) needs.

To understand the experiences that neurodivergent crowd workers had in an image and data annotation platform, Garrison et al. (2023) focussed on five topics: (1) responses regarding the receipt transcription, movie review and picture description; (2) work completion; (3) time per question; (4) effective hourly pay; and (5) participant experience where it was considered the participants’ task preferences, challenges faced with the tasks, and suggestions to have an inclusive and accessible digital work in data annotation for neurodivergent peers.

In brief, results showed that participants justified their answers in movie reviews and described pictures based on personal feelings and experiences; work completion was associated with how long the text was – if it was long, the participants skipped it or gave incomplete answers -, this also correlates with the time spent per question, the longer the text, the more time was needed; some of the participants would earn less than the minimum hourly wage due to incomplete and skipped tasks, and the time spent on the tasks; participants enjoyed all tasks because they were clear and straightforward, but they found challenging deciding annotations for subjective tasks, the long texts, reading the receipts and handling missing information.

Regarding the suggestions made, participants mentioned that clear instructions with visual examples should be provided in the request, as well as visual and auditory stimuli, accessible keyboard functions, word counters and screen readers.

Finally, the authors noted that participants had difficulties with vocal communication and that some preferred to write down their answers instead of saying them.

Guidelines

The guidelines identified across the studies pertained to the identification of complex words, interaction in plain language, alternative text and image descriptions, and communication of probabilistic information.

In essence, Alarcon et al. (2023) found that what constitutes complex words was:

• Words that are used in verbal communication but that can be unknown to PwID and/or older people;

• Long words that are difficult to read and pronounce;

• Technical jargon;

• Abbreviations or acronyms used without any explanation;

• Words in a different language than the one mainly used in the document;

• Roman numerals;

• Idioms with double meanings that are difficult to understand;

• Use of metaphorical expressions;

• Use of abstract terms that don’t have a physical form, which can be difficult to perceive or imagine;

• Multi-word terms (expressions with complex words, expressions with simple words whose more familiar meaning is different, and complex expressions with complex and simple words whose meaning has changed);

• Words with different meanings according to the context;

• Percentages and mathematical expressions;

• Adverbs ending in ‘-mente’ (-ly);

• Collective nouns;

• Obsolete words or in disuse.

For using plain language to interact with someone, Vermeiren (2018) established some considerations, of which:

• Preference for face-to-face interaction;

• Usage of words like ‘you’ and ‘me’ and well-structured words;

• Give information in small parts with the most important part coming first;

• Use a normal tone of voice;

• Speak clearly and briefly, with pauses from time to time, and be prepared to repeat and rephrase the sentences;

• The pitch should vary according to what is being said;

• The sentences should be in active voice, with strong verbs, and in the present tense, as well as in the positive form;

• It should be avoided the use of long strings of nouns and fillers (e.g., um, ah, okay, etc.);

• Words that may be new to the audience should be explained, as well as abbreviations or acronyms, the first time used;

• Use everyday words and avoid jargon, technical terms and foreign expressions;

• Reuse the same words instead of synonyms;

• Instead of using figures and percentages, it should use expressions such as ‘a few’ and ‘a lot’;

• Be concise and cut out unnecessary words;

• Feelings should be expressed to enhance speech acts and through emotion-expressing paralinguistic means.

Complementing this, Monroe and Morrison (2022) found that the best way to edit the alternative text to reduce the description’s cognitive load is to:

• Use clear and concise syntax, where the text should be edited for clarity;

• Organise the information in a predictable manner, starting from a general topic to a specific one, grouping and describing them by similarity and then differences.

• Describe the content in relation to each other and any trends or visually emphasised information;

• Give information in multiple modalities. For example, when using images, image descriptions and alternative texts are recommended, graphics should be transformed into tables and flow charts into lists;

• Avoid redundancies by ensuring that what is in the caption or surrounding text is not repeated.

While creating templates to be used to communicate probabilistic information, Heggli et al. (2023), incorporated guided principles of graphic design and social science in an attempt to produce effective and accessible probabilistic information visualisations in the templates.

In this sense, the authors developed guidelines for effective visual communication of probabilistic weather information that were grouped into six topics: accessibility, layout, content, colour, font and embellishments.