Making published research more accessibleMichael Upshall
Accessibility has been a consideration of web design for almost as long as the web has existed. Accessibility guidelines are arguably well understood, and relatively straightforward to implement when it comes to websites and apps. Academic books and articles, however, predate the internet by many years, and it may seem that academic articles are the most unlikely starting points for accessible text. At first sight, articles can look very daunting – full of technical abbreviations, and demanding considerable prior knowledge, presenting challenges for the general reader and students alike. The vocabulary and syntax of research papers frequently seem at odds with the concept of accessibility. For example, most spelling and grammar checkers recommend active rather than passive verbs; but academic articles use the passive voice by default. Here is a typical medical research paper:
A total of 206 patients, from both abdominal and orthopedic surgery, were randomly assigned in 1:1 ratio to receive 800 mg IV-ibuprofen or placebo …
Of course, all scholarly articles include an abstract, which is designed to provide a summary of the contents for non-specialists. Yet even scholarly abstracts aren’t always intelligible without some understanding of the subject. Academic researchers aren’t necessarily best placed to summarize their own work – being close to a subject doesn’t mean you’re able to express its ideas in simple terms. Writing a concise, easy-to-read abstract is a considerable skill, which is why there are initiatives (from Kudos and others) to help researchers write clear outlines of their articles for non-specialists.
Why accessibility is so important
Making content accessible is more than just a nice to have. Academics such as Pierre Bourdieu, with the concept of the “hidden curriculum”, have argued that the more assumptions lie hidden behind a learning text, the less democratic the educational opportunity: if readers don’t have an equal chance of understanding a paper, it shouldn’t be surprising that there is a disparity in achievement. A hundred years ago, for example, scholars writing an article could expect a general understanding of Latin phrases; today, the use of Latin (or other untranslated languages) provides an obstacle to all readers. And although English is the most widely used language for academic writing, the majority of users of English are non-native speakers.
The Web Accessibility Guidelines and usability experts have plenty of advice on how to make text on the screen more readable, and there are many books on the subject. Jeffrey Johnson, in his widely regarded Designing with the Mind in Mind (3rd edition, 2020) includes several excellent principles : Content should have subheadings, and text broken up by bullet points and paragraph breaks. Sentences should not be too long. But that’s not enough.
We can ask authors, and even train them, to structure their articles better, and to write their text more clearly. We can ask them to provide simple, non-technical summaries. But with over 5,000 new scholarly articles being published every day, it is all but impossible to train all authors to write more clearly. Nor can we transform the millions of articles and books that have already been written.
Publisher roles have changed
And the trend over the last ten or twenty years has been for scholarly publishers to provide less, rather than more, assistance to authors. Publishers are under pressure to reduce the time (and cost) of processing submissions. At one end of the spectrum, a highly regarded humanities journal, the William and Mary Quarterly, estimates they provide 130 hours of editorial time for every submission to reach publication. At the other extreme, a preprint article receives precisely zero publisher time, since the submission does not pass through the hands of a publisher. But that doesn’t remove the need for the article to be intelligible.
Whereas formerly a scholarly article would be thoroughly read and commented on by an editor employed by the publisher to check for structure and language, today, with the vast increase in the number of articles published, and the move to open access, less and less attention is given to checking the article for these aspects. The content continues to be reviewed, but often the only content review is by peer reviewers, who are already familiar with the subject matter and are masters of the “hidden curriculum”. All too often, nobody in the peer review process takes responsibility for making the content clearer. With Preprints, the situation is even starker. Nobody except the author sees the preprint before it appears on the public site. There is nobody to tell the author to use fewer passive sentences, or to expand their abbreviations, or to add more subheads.
Most academic content is not written to be accessible. Do we, then, simply accept the situation for the foreseeable future?
Practical tools available
There is a way of improving accessibility of scholarly content. We may not be able to do much about the way the content is written, but today there is a growing number of tools designed to make the reading and understanding of scholarly literature easier both for lay readers and academics.
Some would claim there is no need to create tools to facilitate comprehension. But, after all, scholarly articles are written, and read, for one purpose: to communicate information. If there is any way in which we can assist users towards that goal, we should make use of it. Can we use tools to make articles more accessible? Talking to researchers reveals several methods they use to scan an article in outline before deciding to read the content in detail. Most will read the abstract, of course, but other methods used by academics include:
- Reading the conclusion before the rest of the paper
- Looking at the tables and figures to see what (if anything) they reveal
- Looking at the references – which articles are cited by this paper?
- Scanning the major headings, although for academic articles, the headings tend to be similar, at least for scientific content, so this may not be very revealing.
Even these methods may take several minutes per paper. Can machine-based tools be used to make the initial scan more effective? Here are some tools that do just that:
- Discovery: Have I found the relevant papers for my topic? Try Semantic Scholar, Elicit, R Discovery, or iris.ai.
- Identifying confirmation and refutation statements. Typically, an academic article will be of the format “in the past, people believed X. Previous studies, such as Y, confirmed this. In this paper we identify a new idea, Z. From our study, we argue differently to earlier theories A and B”. An example of how this can be done is using ai or Scholarcy. Here is an example of Scholarcy’s Comparative Analysis:
- Can I identify agreements and disagreements within the papers, confirmations or refutations of a statement? Try ai, or Obsidian, to create your own links between papers, Try connectedpapers.com if you prefer to see a graph display showing related papers.
The trend towards open access has transformed access for readers. But are the published articles more accessible? For maximum impact, researchers today have to take responsibility for ensuring their articles are readable and accessible. But the good news is that there are plenty of tools available to assist them, both for the discovery and the writing stage: using Scholarcy is an excellent way for authors to make their research more thorough, and their content clearer. There is no excuse for not making academic content accessible.