PDF is essentially a graphical format, but developments in the last few years have made it more intelligent. In particular, since version 1.6 (2004), it has been possible to embed the document’s logical structure using tags, which make it possible to perform a basic reflow for small screen devices or aid copying text out of the PDF by removing linebreaks and ensuring that the reading order is correct (e.g. it not jump across columns on a multi-column page).
Many programs, such as Microsoft Word, now have export functions that generate correctly tagged PDFs if the content is correctly marked up (i.e. the styles are used correctly). Note that the logical structure is lost if the document is simply “printed” to PDF.
Accessible PDF, or PDF/UA, is an extension of this concept to ensure that a document’s tagging is sufficiently rich to enable the content to be accessed by assistive technology such as screenreaders. This means that headings, paragraphs, lists, tables and hyperlinks all need to be fully tagged. Alternative (“alt”) tags need to be included for all images intended to convey meaning (the textual description can differ from the caption) and blank alt tags should be used for decorative images. Because of this, in order to generate a PDF that is compliant with the Web Content Accessibility Guidelines (WCAG 2.0), some manual remediation is always necessary, although the output of Microsoft Word or InDesign + plugin is often a good starting point.
I am available for both remediation work (i.e. making existing PDFs accessible) and consulting on this topic, both for manually created and automatically generated PDF documents. Please contact me for further information.