Creating accessible PDFs for conference proceedings
Erin Brady, Yu Zhong, Jeffrey P. Bigham · 2015 · Proceedings of the 12th International Web for All Conference (W4A) · doi:10.1145/2745555.2746665
Summary
This paper investigates the accessibility of PDF research papers published at major ACM conferences related to accessibility and human-computer interaction. The authors conducted a two-part analysis: first, an automated accessibility check of 1,811 papers from four years of CHI, ASSETS, and W4A proceedings (2011-2014), examining whether documents were tagged, contained structural heading tags, and specified document language. Second, they performed a manual accessibility evaluation of 26 tagged papers from ASSETS 2014 and W4A 2014, checking for proper tag structure, alternative text on images, and correct tab order. The paper also documents the authors' experience volunteering to make CHI 2015 camera-ready submissions accessible, processing 25 PDFs over two weeks. The research highlights a fundamental tension in academic publishing: PDF format, while ensuring visual consistency across platforms, creates significant barriers for screen reader users when documents lack proper structural tagging and metadata. The authors trace the history of PDF accessibility from the format's introduction in 1993 through the development of PDF/UA (ISO 14289), and review existing guidelines from Adobe, WCAG, and WebAIM, noting that these resources, while comprehensive, can be overwhelming for authors unfamiliar with accessibility — Adobe's guidelines alone total 188 pages.
Key findings
The automated analysis revealed stark differences across conferences. By 2014, only 26.8% of CHI papers were tagged, compared to 71.4% for ASSETS and 100% for W4A technical papers. W4A showed the most dramatic improvement, going from 0% tagged documents in 2011 to 100% by 2014, coinciding with the introduction of accessibility guidelines on the conference website. ASSETS showed a slight decline from 92% to 71% tagging rates, possibly due to community growth bringing in researchers less familiar with accessible authoring. The manual check of 26 tagged papers found that only 61.5% passed Adobe Acrobat's full accessibility check. While 73.1% had alternative text for all figures and 84.6% had proper tab order, only 11.5% had the title correctly tagged with an H1 heading — demonstrating that the presence of tags does not guarantee correct application. The CHI 2015 remediation effort showed that making a 10-page paper accessible took approximately 15-20 minutes, suggesting the entire CHI technical program (~500 papers) could be made accessible for under $5,000 USD, and smaller conferences like W4A for under $250.
Relevance
This paper makes a compelling economic and practical case for treating document accessibility as a publishing responsibility rather than solely an author obligation. The finding that even accessibility-focused conferences produce largely inaccessible papers underscores that the problem lies not with author indifference but with inadequate tools and workflows. The cost estimates — as low as $250 for a small conference — challenge the assumption that accessibility is prohibitively expensive. For practitioners, the paper highlights that automated accessibility checks alone are insufficient; documents may appear tagged yet still fail manual review. The authors' suggestion to consider HTML as an alternative publication format anticipated a shift that ACM and other publishers have since begun adopting. This research remains relevant as organizations grapple with making document repositories accessible and weighing the trade-offs between PDF and web-based formats.
Tags: PDF accessibility · document accessibility · tagged PDF · PDF/UA · screen readers · alternative text · automated testing · conference proceedings
Standards referenced: PDF/UA · ISO 14289 · WCAG 2.0