Accessibility testing 14 PDF creation tools
12 September 2009
There are many tools available for converting documents to PDF. We tested 14 of them for their ability to generate accessible output. The results were highly polarised. A few produced excellent results, but the rest performed poorly, if at all.
This article will give a brief overview of the products that work well. These include Acrobat Professional, LiveCycle Designer, Microsoft Word (2007 and 2003), OpenOffice Writer and Adobe InDesign. In addition to these we tested a further eight PDF creation tools, but none came close to producing accessible end-products.
What's required to produce an accessible PDF?
At a basic level a conversion tool must be able to generate a PDF containing tags that hold information about the document's various structural elements. These structural elements will include headings, lists, hyperlinks, tables, alternative text for images, and so on. There are many potential refinements of course, but an accurate representation of a document's structure in a PDF tag tree will go a long way to making the document accessible. At present only a handful of products are capable of meeting these basic requirements.
Microsoft Word (2007 and 2003)
With Microsoft Word 2007 you can download a free add-in that generates accurately tagged PDFs. Provided that the source document is structured properly results will be excellent. Word 2007 includes several important improvements over Word 2003 with respect to, for example, the elimination of certain alt text reading order bugs and the handling of footnotes by screen readers. These problems previously had to be corrected using Acrobat Professional.
For some editing tasks, such as ensuring that all abbreviations, acronyms and URLs are screen reader friendly, you still need Acrobat. However, such edits are the icing on the cake. In short, Word 2007, as an accessible PDF creation tool, is truly outstanding. Word 2003 works almost as well, but, as mentioned above, has a few more bugs that require fixing in Acrobat. It also doesn't have its own built in PDF conversion tool (unless you have Acrobat installed).
After Microsoft Word, the next best option is arguably OpenOffice's (free) word processor package, Writer, which is capable of producing tagged, well-structured and highly accessible PDFs.
As with Microsoft Word, structural elements created in Writer (headings, links, lists, tables, alt text etc) all appear exactly as they should when converted to PDF using Writer's built in PDF creation tool.
However, there are one or two problem areas. For example, unlike Microsoft Word, by default Writer creates tables of contents that don't include active links to a document's various sections. And the process by which you can add an active table of contents is neither obvious nor intuitive. This is a problem because an active table of contents in a PDF is much like the navigation system in a website: you don't need it for a one-pager but anything much longer and you really do need to be thinking about it.
Writer also handles footnotes poorly from an accessibility point of view. In Writer-generated documents footnotes appear at the bottom of the PDF's tag tree and hence, for screen reader users, at the end of the document's reading order, regardless of their actual position in the document. Also, the link from the footnote's reference point doesn't work, despite being announced as a link by screen readers. Together these problems make footnotes very difficult for screen reader users to use.
Acrobat Professional can be used to fix just about any PDF accessibility problem. With Acrobat you can quickly and easily add or amend tags to give the document its required structure or, where necessary, to fix reading order problems. It is particularly useful for fixing glossy brochures, newsletters and annual reports etc, which often arrive in web editors' in boxes with no accessibility features built in at all. It also has powerful Optical Character Recognition functionality (for dealing with scanned documents) and is excellent for creating accessible forms (see below). Overall, Acrobat Professional is likely to be an indispensable tool for anyone serious about editing PDFs for accessibility.
Adobe LiveCycle Designer
Adobe LiveCycle Designer is a programme that sits inside Acrobat Professional and is used solely for creating PDF forms. Many types of form can be made fully accessible using LiveCycle.
For some types of form LiveCycle has distinct advantages over HTML including, particularly, ease of use: you don't need HTML or programming skills to do surprisingly useful things with it. This is a piece of software with a big future – HR departments in particular take note – accessible job application forms are now easily attainable.
Adobe's flagship page layout program, InDesign, makes it easy to create tagged PDFs and to add many (but not all) of the required structural elements. It also allows the user to adjust the reading order for screen reader and other assistive technology (AT) users (independently of the visual appearance of the document), to add alt text to images and to remove artifacts such as page numbers and running footers which get in the way of AT users.
However, InDesign is not so good when it comes to lists and table headers which need touching up in Acrobat Professional. As of this writing it also struggles with accessible hyperlinks.
InDesign does have one distinct advantage over Word in that it contains a professional standard text justification engine. You can set letter and word spacing, resize characters and exercise fine control over hyphenation to avoid the accessibility problems typically associated with more crudely justified text. In Word you currently have a great deal less control.
In sum, InDesign has many useful features for authoring accessible source documents, but you will also need Acrobat Professional to complete the job.
Other Adobe products
Other Adobe products capable of creating tagged and structured PDFs include the desktop publishing programme FrameMaker (popular with technical writers) and GoLive (discontinued in 2008 in favour of Dreamweaver). These programmes were not tested for this article.
The rest of the field
As mentioned previously, we tested a further eight non-Adobe PDF creation tools. The best performer of this group did produce a tagged PDF (although not by default). However, the tag tree was in no way representative of the source document's actual structure. For example, the first
<p> in the tag tree it produced (see figure) was originally a heading 1 in the source document, the third
<p> was originally a heading 2 and the next 5
<p>s were originally links in a table of contents. In converting everything to
<p> tags nearly all of the document's structural information was lost, rendering it significantly less accessible than it would be if it had been tagged properly.
And this really was the best of the bunch: one other product created bookmarks and (inaccessible) links, but no tags. Some others produced PDFs containing just images of text rather than real text (made up from .tiff files). Such documents would need extensive correction in Acrobat Professional.
It's clear that, if the production of fully accessible documents is the objective, some care needs to be taken in choosing the right PDF creation tool. To date, despite the number of options available, only the established names (Adobe, Microsoft, OpenOffice) provide the means.