1 Introduction

Our interactions with the government are becoming increasingly digital. According to the UN E-Gov Survey of 2020 [2], over 84 % of countries offer at least one online transactional service, with the global average being 14, and an overall trend to increased digitalization. To speed up digitalization for government services, some countries are pursuing ambitious goals, partly also set by law, e. g., in Germany through the “Onlinezugangsgesetz” (literally translated: online-access-law), an ambitious plan to make 575 services available online until end of 2022 [3]. In the last year, the Covid-19 pandemic has painfully demonstrated the need for online services, as citizens and public employees were confined to their homes, but still had to request or provide government services.

E-government represents a comprehensive public sector reform concept based on the intensive use of information and communication systems [4]. E-government enables the redesign of the various interactions with the state. Processes can be reorganized as they no longer rely on presence and paper. The goals include improvements in quality, time, and costs. Unlike businesses, public administrations direct their actions towards the common good (not towards corporate profit). This public value orientation has implications for service delivery and interactions between government institutions and citizens.

While important for commercial websites, usability, i. e., “the extent to which a system, product or service can be used by specified users to achieve specified goals with effectiveness, efficiency and satisfaction in a specified context of use” [5], is crucial for e-government websites. E-government services have to be accessible to the whole population living in the country. Thus, the users of e-government websites are heterogeneous, encompassing people with diverse attributes (e. g., abilities, interests, prior knowledge, affinity for technology interaction [6]) and also with different disabilities or disorders. With more resources put on online services and first steps to switch to digital-only [7] and mandatory self-services [8] (still with possible exemptions), equal access is crucial as well. Accessibility enables social inclusion and equal opportunities [9].

Since no one should be left behind when it comes to public services, accessibility is frequently considered in the context of e-government. The focus is often on raising awareness about the extent and types of the most common accessibility problems based on the analysis of specific e-government websites (e. g., [10]). Further research focuses on tools that help developers implement accessibility standards (e. g., [11]). In this article, we focus on an area that might be easily overlooked when assessing government websites for usability problems.

Digital services are usually conducted via written language, which provides opportunities to make them accessible for people with disabilities, e. g., regarding perceptual abilities or motor skills. Accessibility features and apps on smartphones and personal computers can access and work with the written text, e. g., to provide different input or output modalities such as speech-to-text or text-to-speech. The Web Content Accessibility Guidelines (WCAG, at the time of writing version 2.1; [12]) have been developed to allow users with different impairments to access websites. The WCAG also informs other standards and laws, e. g., in Germany the “Barrierefreie-Informationstechnik-Verordnung” (BITV, [13], literally translated: “barrier-free-information-technology-ordinance”), which is the basis for accessibility in e-government services. Besides Germany, other countries refer to WCAG 2.0 or WCAG 2.1 in their individual guidelines within the public sector – for example Australia, China, India, Israel, New Zealand, Switzerland and the United Kingdom as well as the European Union [14].

However, the focus on written language can also make it more difficult for specific groups to use services, who would have had less problems in direct, face-to-face interaction with civil servants: users with dyslexia. They can see the written language but have difficulties reading and writing it. Thus, metaphorically, they are caught between two stools: They cannot work well with the website as it is, nor do they need a switch to another modality to completely replace the text.

As this period of increased digitization can be used to design these services in a way to make them accessible for all, we ask the questions:

R Q 1 : Do the Web Content Accessibility Guidelines sufficiently address people with dyslexia?

To look at a concrete example, we examine an adaptation of the WCAG for e-government websites, here the mentioned regulation (BITV) in Germany.

Based on this comparison, assisted by a literature review and interviews with people with dyslexia, we then focus on one area in which accessibility is a prerequisite for successful interaction with e-government websites: search within e-government websites. We ask the question:

R Q 2 : Which designs would increase usability of search functions for people with dyslexia, while not negatively affecting people without dyslexia?

To answer these questions, we start by providing background information regarding the Web Content Accessibility Guidelines (Section 2.1), dyslexia (Section 2.2), dyslexia and website use (Section 3) and the coverage of dyslexia by the WCAG (Section 4). This information is the basis for the proposed modifications to improve the search function for users with dyslexia (Section 5), which are then empirically evaluated by users with and without dyslexia (to also avoid negatively impacting the later, Section 6). We then provide answers to the research questions in the conclusion (Section 7). This paper expands a previously published version at the Mensch und Computer 2021 conference [1].

2 Theoretical Background

2.1 Web Content Accessibility Guidelines and Country Specific Implementations

On the Web, the goal of accessibility is to allow everybody to use websites. The whole diversity of human abilities should be covered and no-one, including users with disabilities, should be disadvantaged [14].

Referring to German law, disability can be seen as an interaction between physical or mental impairments of the individual (e. g., senses, motor functions, cognition) and requirements in the environment (e. g., written text is used, or access to a building is only possible by using the stairs). Requirements in the environment become barriers when impairments prevent their use without allowing for alternatives (e. g., no audio version, or no access ramp for wheelchairs). However, as disability depends on the match between environment and individual, it can be addressed by increasing the options of the individual with assistive technologies (e. g., by using a text-to-speech app for written text) and by the design of the environment itself (e. g., the website offering an audio version, or being compatible with text-to-speech apps).

Guidelines and toolkits exist to develop products that are accessible for all, e. g., the “Inclusive Design Toolkit” by the University of Cambridge [15], or Microsoft’s “Inclusive Design Toolkit” [16]). More generally, the World Wide Web Consortium (W3C) provides recommendations for standardization on the Web, including the Web Content Accessibility Guidelines (WCAGs; [12]).

The WCAGs are concerned mostly with vision, hearing, cognition, dexterity, speaking, and locomotion. They set conformity levels, inform the laws in different countries regarding accessibility on the web and are accepted as norm in the European Union.

To examine a specific adaptation of the WCAG, we focus on Germany. It is currently undergoing increased digitization of public services (see Section 1) and the authors are familiar with both language and public administration in the country. Germany’s laws protecting people with disabilities were strengthened multiple times, resulting – for e-government – in the “Barrierefreie-Informationstechnik-Verordnung” (see Figure 1) [13]. The BITV requires federal websites to be designed in an accessible manner, including sign language and “Leichte Sprache” (literally “easy language”, a simple version of the German language designed to improve accessibility and directed at people with low comprehension of German or low reading skills).

Figure 1

Legal development towards the new BITV 2.0 (own illustration).

As the BITV is referring to the European accessibility standard EN 301 549 V3.1.1 (of which WCAG 2.1 is a part), it is also focused on senses, cognition and motor control. However, the scope of the BITV is broader than these disabilities. It states as its goal to allow for and ensure a comprehensive and basically unrestricted barrier-free design of modern information and communication technology (“eine umfassend und grundsätzlich uneingeschränkt barrierefreie Gestaltung moderner Informations- und Kommunikationstechnik zu ermöglichen und zu gewährleisten”, § 1 Abs. 1 [13]).

This “comprehensive and basically unrestricted barrier-free design” also includes users with dyslexia. But are they covered by this WCAG-based ordinance?

3 Dyslexia and Website Use

To assess usability problems of users with dyslexia, a literature research and interviews with users with dyslexia were conducted. The literature research was conducted using Google Scholar and additionally identifying researchers in this area. Search terms were, among others, “dyslexia”, “accessibility”, “usability”, “usability + dyslexia”, “accessibility + dyslexia”, “BITV + dyslexia”, “usability + search”, and “accessibility + search”. Inclusion of papers were based on fit and overall scientific quality. In total, more than 100 papers were sighted. Table 1 and Table 2 show a summary of the relevant empirical and theoretical articles.

The interviews (semi-structured) were conducted with four users with dyslexia (ages 20–29 years, two male, two female) and one expert in dyslexia research (who is also dyslexic). They focused on the intersection of dyslexia and technology usage, specifically concerning difficulties in using smartphones and personal computers. We also assessed the use of assistive technologies, perception of public service websites, good and bad accessibility examples of websites, and evaluation of “Leichte Sprache” (see Section 2.1), navigation and search functions on websites. Using the strengths of online interviews, we also asked them to share their screen to observe their interactions with two online tasks (finding out when the day-care centers in a major city open after one of the Covid-19 lockdowns and finding out the office hours of the tax and revenue office of a specific city; the tasks were given, participants came up with search terms on their own). A qualitative analysis was conducted to identify topics (esp. usability problems).

Looking specifically at the current limitations and problems of website use of people with dyslexia, the main findings of the literature research and the interviews (see Table 5 for details) were that while speech to text software was seen as helpful, given that websites are frequently accessed on mobile devices in public, their use is not always possible due to environment conditions (noise, privacy). Text-to-speech was used selectively, not to have everything read aloud, but to have only specific words or passages read. “Leichte Sprache” (see Section 2.1) was not seen as helpful by any of the interviewees. It reduces the cognitive complexity of the content and uses a limited vocabulary to address readers with low educational background or limited language skills. However, people with dyslexia can understand sentences similar to people without dyslexia – if they are spoken to them [23]. They just cannot similarly process or unpack the text itself. As “Leichte Sprache” addresses the content, not the packaging, it is not helpful for users with dyslexia.

Both the literature research (e. g., [30]) and the interviews indicated problems with online search, which were also apparent in the observation of the two online tasks during the interviews. Three out of four made spelling errors during the search, which they found difficult to detect (a frequent problem, e. g., [26]).

To find information users often use the search function [40], [41]. In the e-government context, the search function is critical, as e-government websites are only visited infrequently and with a specific goal in mind (e. g., getting a new passport, to find information about public authorities, public services, or how to apply for benefits or permits). While navigation menus and reading the website content can be used to find the information, this process requires more effort for users with dyslexia, compared to using the search function. The later leads them directly to the relevant page, if the search is usable for them (e. g., interpreted correctly despite likely spelling errors).

4 Coverage of Dyslexia by Accessibility Guidelines

How well do the accessibility guidelines, here the country-specific implementation (BITV) of the WCAGs address the specific problems of users with dyslexia? To answer this question a checklist to meet the requirements of BITV was considered [42]. For each of the 60 criteria in this list, it was determined whether this would benefit people with dyslexia or not. If it did, each point was briefly discussed as to why the decision was made.

In general, all steps that improve the use of screen readers can also be helpful for people with dyslexia, when they are strongly affected and therefore use them. However, the use of screen readers does differ from that of people who cannot see (comment from one of the interviewees, see Section 3). They can see the content but just want to have information read out loud selectively. That is why the following Table 3 does not show the BITV items that only improve the use of screen readers, because these suggestions are designed to improve usability for blind and visually impaired user groups.

However, some of the other items were found to be helpful, for example, a meaningful order of the page content assists users in quickly visually assessing what is significant. Likewise, because of test step 1.4.4a, it must be possible to enlarge the text to 200 %. Given that text breaks into the next line when enlarged and fonts remain clear assists in readability. That texts are often greatly enlarged in order to be better able to read them was one of the points mentioned in the interview and observed in the user behavior.

Given the focus on senses, cognition and motor control, some elements of the BITV are already helpful for users with dyslexia, but other suggestions are not directly applicable to dyslexia, even though it might seem that they would help. BITV does not cover suggestions on type face and proposes the use of “Leichte Sprache” (not helpful for people with dyslexia, see Section 2.1 and Section 3). Suggestions regarding screen readers cannot be applied 1:1 from users who are visually impaired to people with dyslexia, as their usage differs (the latter use them more selectively, e. g., to understand specific words, see above). Also not covered are ways to improve the search, showing synonyms for complicated words, and avoidance of all-caps.

Given the difficulties with search and its relevance to get to the needed content, how could search be improved for users with dyslexia?

5 Proposed Modifications to Improve the Search Function for Users with Dyslexia

Based on the results of the literature research, the interviews, and the gaps of the WCAGs and BITV regarding the design of search functions for users with dyslexia, hypotheses regarding the improvement of the search function were developed (see Table 4). These had to be evaluated, to make sure their implementation really helps people with dyslexia, but also not negatively affect people without dyslexia.

Based on these hypotheses, suggestions to improve the search function for users with dyslexia were developed. We continue and empirically assess the research by [30] (see also Table 1), who give an overview of previous studies regarding web search and dyslexia, and who also found differences in behavior and preferences between dyslexic and non-dyslexic searchers.

In contrast to previous research, the focus of these suggestions is on the search field/results page of the search engine itself and on assessing the features empirically. We look specifically on solutions applicable to within-site search of e-government websites, as developers have the necessary control to change the design of the search function, compared to using general-purpose web search engines such as Google, DuckDuckGo or Bing. While large commercial search engines like Google already implement some of these suggestions (e. g., speech-to-text function, automatic recognition of typed terms, error tolerance or showing alternatives when spelling errors are made), given data privacy considerations, using Google Search as within-site search is not an option for many government websites (e. g., consider the search for financial assistance being tied to a person’s online profile). This situation leaves developers with the question of which search features to implement for their own website search given the limited resources, or, if they choose between frameworks with different features, which features should be rated higher. As the WCAG does not provide an answer, we assess the features empirically to determine empirically which features are helpful for users with dyslexia without impeding those without dyslexia.

Table 5 describes the proposed modifications with the underlying rationale (due to space reasons combined with the results). For images on how they were implemented, see the appendix.

6 Evaluation of the Proposed Modifications

To empirically assess whether the design proposals are beneficial for users with dyslexia, we compared each search feature to a default search field without this functionality. While modifications to improve readability for users with dyslexia usually also benefit those without it (e. g., [18], [23]), care was taken to empirically validate this finding for our modifications. Thus, we included a comparison group of users without dyslexia as a control group.

7 Conclusion

Policies can only be implemented effectively if the related administrative services reach the respective target groups. For example, children can often only benefit from social services or support measures if their parents apply for them. Destitute citizens can only get support if they make their plight known. This requires knowledge about these services as well as easily accessible application processes. The simplicity of the application process is determined, on the one hand, by the content requirements (which information do applicants have to provide) and, on the other hand, by the design of the information transfer, i. e., the usability of the tools used for the application.

Websites, especially for mandatory or literally life-saving interactions with the government, must be accessible for all. Using literature on dyslexia, interviews with people with dyslexia, and looking at accessibility guidelines (WCAG) and their country-specific implementation (BITV in Germany), we confirmed and identified gaps regarding the usability of websites for users with dyslexia, esp. regarding the search function on websites. Thus, we answer the first research question (RQ1) that the WCAG provides assistance for users with dyslexia, but does not sufficiently cover their specific needs, esp. regarding search functions.

Based on the literature and interviews, we empirically tested possible modifications to the search feature to find changes that are preferred by users with dyslexia, without negatively impacting those without. Based on the results, we answer the second research question (RQ2) that the usability of the search function can be improved for users with dyslexia by including autocompletion, a search that compensates for spelling errors, an indicator that the search was corrected, search term summary information, and an avoidance of all-caps. To avoid negatively impacting users without dyslexia, wider line spacing should be reserved for end-user customization (e. g., browser settings/plugins), as preferences of dyslexic and non-dyslexic users differed significantly for this feature.

While these features likely do improve the usability of e-government websites for those with dyslexia and those without, other measures should also be considered. Given that many people with dyslexia use a general-purpose search engine to find specific e-government web pages — as they had negative experiences with site-internal searches — government websites should allow general-purpose search engines (like Google, Bing, etc.) to index all their webpages. When it comes to the actual usage of eGovernment services (esp. transactions), forms should implement plausibility or sanity checks to formally validate for user input. Principles like “Once-Only” — while helpful for all users — will likely be a huge benefit to dyslexic users. Finally, while text is an important way for eGovernment websites to communicate to a large number of people, for those users for whom this proves to be a high bar, other contact methods (incl. telephone) should continue to be provided. In the future, powerful voice assistants could ask for relevant data in conversations. In general, the more alternatives exist for accessing digital administrative services, the better individual user requirements can be considered.

As the search feature is an important function of websites that can quickly cut through swaths of text to provide only a small amount of relevant text, and reading text takes a lot of effort for users with dyslexia, the search function should be especially accessible and usable for users with dyslexia. The proposed and empirically assessed modifications might help to improve accessibility for users with dyslexia, and – given that spelling errors do happen to everyone – also for everyone else.