This paper looks at the web search needs of people with dyslexia, taking into consideration the difficulties they may face in query formulation, result triage, and information extraction. The authors specifically describe the differences in the behavior and “preferences between dyslexic and non-dyslexic [web] searchers,” and use the Bing search engine to perform various online and offline experiments (specifically, three different surveys).
First, the authors interviewed ten adults with dyslexia who “[use] web search engines at least several times per week.” The aim of these interviews was to understand the search needs, search processes, outcomes, and challenges of web searchers with dyslexia. Results suggest that people with dyslexia face challenges in all three aspects of web search, namely query formulation, search result triage, and information extraction. Among query formulation issues, searchers are not satisfied with the search engine’s spelling correction functionality. Many used the spelling suggestions from their preferred word processors to get more than one option for spell correction. A majority of the interviewees use image search results for query formulation and verification. Almost all of them reported an affinity with voice input, specifically on mobile devices. For search results triage, the interviewees prefer pages with multimedia along with text, pages avoiding long words, pages with few advertisements, and pages that are not visually dense. For extracting information from web pages, they prefer software that can read web pages. Similarly, they prefer consuming pages through highlighting and printing pages. Many people use the search engine as a linguistic tool, which accounts for three percent of search traffic.
Second, the authors conducted an online survey of 81 dyslexic and 80 non-dyslexic adults to validate the results of the interviews. Specifically, the questionnaire was developed to compare and contrast the preferences of people with and without dyslexia. The authors explain the demographic patterns of the survey participants in detail. Respondents used a five-point Likert scale (1 = strongly disagree, 5 = strongly agree) to rate their levels of agreement with several statements. Among dyslexic people, “78.8 percent agreed or strongly agreed that they prefer pages that make use of pictures or videos rather than relying only on text,” and “27.6 percent ... prefer to browse the mobile versions of web pages.” The authors also used Mann-Whitney U tests to compare dyslexic and non-dyslexic users. For example, “people with dyslexia ... were significantly more likely to prefer voice input.” On desktop devices, there was not much difference in using voice input between the two groups. People with dyslexia were also “significantly more likely to express a preference for” web pages containing pictures, videos, tables, and lists. But not much difference was found when it came to preference for web pages using easy vocabulary. Similarly, not much difference was recorded for using zoom for increased readability.
A third online study (346 participants, 174 of them with dyslexia) looks at relevance and readability. The authors presented ten informational queries along with moderated results to each respondent. Each search result included a set of questions based on web page accessibility, for example, whether the major points are clearly stated, how the design choices affect reading, distraction related to banners and advertisements, and so on. Each answer was given on a Likert scale. Scores of all these questions were aggregated to get combined readability and relevance metrics. The results show that readability and relevance are strongly correlated for both dyslexic and non-dyslexic respondents. Dyslexic respondents “consistently rated pages as less readable and as less relevant to the search task.”
Based on their experimental findings, the authors suggest a number of ways to improve web search for dyslexic people.
In summary, this well-written paper describes problems that people with dyslexia often face with respect to web search, along with some possible solutions. One can utilize these solutions to improve the search experience for 20 percent of the population. On the negative side, the paper suffers from a few shortcomings. The authors describe three different types of surveys and studies that appear to lead to similar findings. Moreover, due to the various demographics and other details presented, the paper is very dense to read. Another issue is the use of informational queries (rather than navigational ones) in the online study, with the authors failing to explain how they qualified a query as informational. Last, the authors could have written more about the implementation of the proposed search engine for dyslexic people.