Ensuring accessibility of PDFs in UW courses

Recommendations for March 2025

Prepared by the Course Content Working Group on PDF Accessibility:

• El Schofield (Lead), Service Manager, UW-IT Teaching & Learning Systems
• Shannon Garcia, Program Manager, Disability Resources for Students
• Dr. Larry Goldman, Associate Teaching Professor, Department of Chemistry
• Mary-Colleen Jenkins, Instruction Accessibility Specialist, UW-IT Accessible Technology Services
• Gaby de Jongh, Senior Computer Specialist, UW-IT Accessible Technology Services
• Dr. Jennifer Taggart, Teaching Professor, Department of Mathematics
• Ana Thompson, Academic & Access Technologist, UW Bothell IT

Special thanks to Marcus Hirsch and Priya Keefe for your assistance throughout this process.

Table of Contents

• Summary of Recommendations
• Background
  • Objective of PDF Accessibility Subgroup
  • Explanation of Issues
• Current State of Course PDF Accessibility
  • Current Individual and Unit-Level Support
  • Current Central Support
  • Current Training and Education
  • Current Policy and Guidance
  • Current Technology/Tools
  • Current Data on the Issues
• Approaches to Ensure PDF Accessibility in UW Courses
  • Options for Requirements and Guidelines
  • Options for Faculty Training and Support
  • Options for Central Support
  • Options for Technology/Tools
  • Options for Tracking and Reporting
• Conclusions

Summary of Recommendations

As the working group discussed ways to expand our current approach to handling course PDFs, we identified some core principles that we believe are essential to a successful approach:

• Prioritize actionable steps toward substantial compliance
• Minimize burden on faculty
• Empower student-centered teaching practices
• Invest in scalable, long-term solutions

Based on these principles, we believe that a successful model for creating accessible PDFs would combine three elements: Minimum requirements, Faculty support, and Central services.

1. Minimum requirements: Set clear requirements for what faculty are expected to do for their own course content.
   • a. We recommend sharing these requirements as a path to relieve the stress and uncertainty of this work: “if you meet these minimum requirements, we’ll help you do everything else you need.”
     • i. Indemnity is an important incentive. Faculty are legitimately concerned about legal liability, with some considering taking out insurance to address potential litigation for inaccessible course content.
   • b. The focus for faculty should be on resolving basic accessibility issues using user-friendly tools, with a minimum of required expertise in the specifics of WCAG 2.1 AA requirements. 100% compliance is unrealistic for most faculty to achieve without advanced support.
   • c. These requirements could include:
     • i. Host all courses in Canvas.
     • ii. Post syllabus (including required course materials and an accessibility statement) in advance.
     • iii. Meet or exceed a minimum course Ally score (e.g., 80%)
       1. Eliminate unnecessary files, particularly PDFs or Excel files.
     • iv. Use alternatives to PDFs where appropriate, particularly for content created in Microsoft Office (which will be more accessible in its original format).
     • v. Abide by fair use policies for course texts.
     • vi. Follow up in a timely manner for DRS requests
     • vii. Complete training on accessibility basics

2. Faculty support: Provide training, tools, and technical support for faculty to meet basic requirements on most of their course content
   • a. Cover basic practices for creating accessible text-based content
   • b. Point to accessibility checkers in Canvas, Word, and PowerPoint
   • c. Advise on fair use and alternatives to manual scans or screenshots of course texts
   • d. Supply tools like TidyUP to identify and either eliminate or replace unnecessary course content
   • e. Provide instructions for how to connect with ATS, DRS, and UW Libraries for additional support meeting accessibility requirements

3. Central Services: Invest in expanding central staffing and support models for addressing “difficult” PDFs to replace, remediate or recreate
   • a. Staffing, including specialists in specific disciplines (e.g. STEM, languages, visual arts)
     • i. Student staff can help with basic content at scale
     • ii. To resolve sophisticated issues that faculty cannot address independently, we’ll need professional staff with subject matter expertise who can invest time in this work and develop advanced accessibility skills.
   • b. AI and automation tools, including auto-tagging
   • c. Outsourced remediation services
   • d. Database storage and management systems to archive existing remediated materials for faculty to reuse

Background

In 2023, the Department of Justice (DOJ) issued new standards that require the University’s web content, including academic course content, to be accessible by April 24, 2026. To address this need, the University created a task force and multiple action teams. The Course Content Accessibility Action Team met to identify and prioritize key areas to address in course content. One of those areas is PDF files.

Remediating inaccessible PDFs is time-consuming and requires both technical expertise and human intervention. Automated accessibility checks provide useful guidance but are insufficient to ensure full compliance. Additionally, many instructors are unaware of best practices for creating accessible documents or lack the time and training to address these issues effectively. For this reason, the action team convened a working group of faculty and staff specialists to tackle this specific file format directly.

Objective of PDF Accessibility Subgroup

This working group is charged with the following:

1. Review and elaborate on the Action Team’s documentation of the current state of PDF accessibility in UW courses
2. Incorporate data to illustrate the extent of the issue and effectiveness of existing solutions
3. Brainstorm multiple options for ways UW can ensure substantial WCAG 2.1 compliance of course PDFs, both by adapting existing support models and by creating new structures and workflows

The group was not required to recommend specific solutions among those generated through brainstorming. However, the working group has included some recommendations in this report for general approaches to tackling these issues and communicating about solutions, particularly with the faculty who have curated or created these course materials.

Explanation of Issues

Compared to other common file types found in a course, inaccessible PDFs are the most complicated file type to fix and comprise the most commonly used inaccessible files.

How are PDFs generated?

PDFs are generated through a variety of sources:

1. Files exported from standard applications for word processing or slide presentations (e.g., Microsoft Word, Microsoft PowerPoint, Google Docs)
2. Digital “prints” of web pages or screenshots
3. Documents exported from discipline-specific software for creating images or reports
4. Documents generated using LaTeX or other typesetting software
5. Scans or photographs of course texts or other materials
6. Handwritten notes from whiteboards or tablet applications (e.g., OneNote): image PDF

Many of these formats, particularly large image-based PDFs, do not have the structure or content of the text encoded so that assistive technology can “read” them. Even with applications that are designed with accessibility in mind, variations in software and user practices result in inconsistent quality, particularly for content that heavily relies on symbols, graphs, or diagrams.

What are the specific accessibility issues that PDFs present?

In order for a PDF to be accessible, it has to be tagged. Tags define the structure of a PDF, such as headings and the order in which text should be read, and help assistive technology identify and navigate the document’s content. Untagged PDFs are unnavigable by assistive technology.

Some tools for creating documents (e.g., Microsoft Word, Adobe InDesign) can automatically generate tagged PDFs when exporting to PDF. However, the quality and completeness of the tagging can vary significantly depending on both the tool and the steps taken by the user to create and export the document.

Image PDFs are created by scanning documents, taking photographs, converting images into a PDF format, or exporting files from software without tagging capabilities. Regardless of whether the content is originally hand-written or typed, digital or physical, the resulting PDF is made up of a snapshot-like image of the pages without machine-readable text. This means that the text in an image-only PDF cannot be read, searched, modified, or marked up by assistive technology, rendering the documents unreadable to many users.

Unfortunately, even applications designed to generate properly structured documents can still produce untagged PDFs. For example, LaTeX (or other versions of TeX) is a common standard for both communication and publication of mathematical content, converting source files of typesetting information into polished-looking PDF files. While highly legible for purposes of Optical Character Recognition (OCR), which can recognize and translate the text from these files, the resulting PDFs are not automatically tagged, meaning that they pose the same challenges as scanned documents for users of assistive technology.

Current State of Course PDF Accessibility

UW’s strategy for digital accessibility is primarily reactive, relying on individual remediation requests rather than proactive content development. Support services, such as Disability Resources for Students (DRS) and Accessible Technology Services (ATS), provide limited remediation for high-priority cases and students with documented accommodations. However, the scale of inaccessible content exceeds current support capacity, with over one million accessibility issues flagged in PDFs during the 2023-2024 academic year.

While various accessibility tools are available – including Adobe Acrobat Pro, Ally, Equidox, and Document Conversion Service – many require training or additional investment. Existing faculty training opportunities are offered through workshops, self-paced courses, and specialized consultations, yet participation is voluntary, and systematic enforcement of accessibility standards is lacking.

What are the obstacles to fixing these issues?

1. Original document quality. Scanning quality greatly affects both the initial accessibility of the PDF and the ability to efficiently remediate. Getting clean scans or seeking alternative digital options for course materials can make a huge difference. Libraries can assist with this provided that it falls within fair use or existing copyright permissions for the text.
2. Limited knowledge and skills. Creating accessible documents requires an understanding of multiple accessibility principles. Incomplete or inaccurate understanding of these principles can result in inadvertently creating less accessible documents that pose additional barriers to learners.
3. Limited tools. Built-in accessibility checkers are helpful, but they may miss issues and cannot fully replace human knowledge and intervention. For example, the accessibility checker in Microsoft Word can automatically generate alternative text to describe image content, but this automatically generated text is not reliably accurate or of sufficient quality to meet accessibility requirements.
4. Limited time and confidence. Some faculty and instructional staff do not address issues identified by accessibility checkers because they don’t have the time to learn new accessibility skills or to remediate documents on top of their existing responsibilities.

Current Individual and Unit-Level Support

Instructors and units can improve document accessibility by avoiding the use of PDFs when possible, opting instead for formats like Microsoft Office applications or Canvas pages. These formats are generally easier to make accessible. They should also take advantage of built-in accessibility checkers, such as those in Microsoft Office, Canvas, or Ally, which can help identify and fix accessibility issues.

Unfortunately, many document authors are unaware of the built-in accessibility checkers or how to use them effectively. Additionally, many are not familiar with the techniques necessary for creating fully accessible content, which highlights the need for more widespread education on these tools and practices. For example, when converting a Word document to PDF, it’s important to use the “Save As” function to preserve accessibility tags, rather than printing the document to PDF, as this method can strip the document of important accessibility features.

If the original source document is unavailable, instructors may use tools like Adobe Acrobat Pro (if available) to create necessary tags and ensure accessibility. Acrobat Pro licenses must be purchased individually for each user, and it is generally up to the user to complete any necessary training to begin using the software. Some faculty and staff already have licenses and are comfortable using this software for completing paperwork or marking up documents. Still, the Acrobat Pro accessibility checker does require a unique set of skills beyond the basics of document editing, so we cannot presume that even experienced Acrobat Pro users will be able to correctly implement the accessibility checker without additional training.

In the case of LaTeX-based PDFs, document authors can install certain TeX packages to help render PDFs with structural tags and accessible mathematical notation. Currently, the available packages boast dense documentation and a steep learning curve. For users who are motivated enough to adopt one of these packages, they still must integrate the relevant packages directly into source files in order to re-render accessible versions of existing documents. In the case of older documents, these source files may not be available, so the author would need to either fully recreate the source file from scratch or manually tag the PDF using a tool like Adobe Acrobat Pro.

When using journal articles from library databases, it’s best to provide a permalink, which may offer access to an HTML version of the article —including an audio version in some cases—, as opposed to a downloaded PDF file.

Current Central Support

Instructors and units can utilize the centrally-funded Document Conversion Service to transform PDF documents into more accessible formats, such as Word documents (recommended) or editable, searchable PDFs through optical character recognition (OCR).

Disability Resources for Students (DRS) offers specialized remediation services for students with documented accommodations to ensure access to course materials. Instructor reuse of previously-assigned materials is a strategic advantage for accessibility; DRS retains ALL previously remediated materials and can distribute these archived accessible materials to students as needed. One major challenge is finding a practical way to share a list of this previously-remediated content with instructors. DRS is restricted from providing remediated content to anyone except the student.

Analysis of DRS (ATT) remediation workflows in 2023-2024 identifies three primary categories of PDF accessibility barriers, in order of impact:

1. Instructor-scanned content with quality issues:
   • a. Uncited scans uploaded to Canvas (prevents source verification and fair use assessment, requiring extensive staff time to locate accessible versions)
   • b. Legacy course packs compiled from multiple sources
   • c. Extensive collections of chapter scans from single textbooks (may exceed copyright guidelines, so cannot be provided by Course Instruction Scanning Service)
2. Materials without accessible retail/publisher versions, including but not limited to:
   • a. Specialized materials (foreign language instructional materials, certain legal texts)
   • b. Out-of-print texts
   • c. Small press publications
3. Content that passes Ally accessibility checks, but still requires manual remediation:
   • a. Scanned PDFs that have been processed with automated tools like SensusAccess or Acrobat Pro, but the text layer has significant errors/inaccuracies
   • b. Complex academic layouts (tables, footnotes)
   • c. Publisher PDFs with inadequate automated tagging
   • d. STEM content where text is searchable but equations are presented as images or read incorrectly.

Accessible Technology Services (ATS) also focuses on remediating a limited number of high-priority, high-visibility, and strategically important PDF documents that are frequently used or widely distributed. For example, ATS has a set number of Student Assistants allocated for document remediation and can handle a specific volume of documents each quarter.

For further guidance, the IT Accessibility Liaisons are available to consult on accessibility matters and connect to central support resources.

Current Training and Education

The University of Washington offers several resources and initiatives aimed at improving accessibility in teaching and learning. These include, but may not be limited to:

• The Accessibility Self-Paced Training Course developed by Ana Thompson at UW Bothell has served as a model for other accessibility resources. This training covers key accessibility principles and practices, many of which have been adapted into other available resources. An updated version of this course will be available in Canvas summer 2025.
• The tri-campus Teaching@UW team has hosted workshops on accessibility topics for instructors, including Building an ADA compliant syllabus and Preparing to meet new federal digital accessibility rules,
• The 2020 Teaching with UW Technologies Canvas course from UW Learning Technologies has included the module “Making Content Accessible”, which provides instructors with foundational knowledge and practical strategies for ensuring their course materials are accessible to all students.
• The Deque University Enterprise Subscription provides UW with access to a suite of tools and training resources aimed at improving digital accessibility across campus.
• Other units offer faculty-focused workshops that incorporate accessibility basics:
  • Accessible Technology Services (ATS) offers workshops, presentations, and consultations on request for creating accessible course materials, including PDFs. They also offer monthly meet-ups to discuss a variety of accessibility topics and content formats, often with guest speakers who can speak to recommended practices.
  • Other instructional support units collaborate on occasional programming related to accessibility and universal design, including the tri-campus teams who contribute to Teaching@UW, UW Learning Technologies, and college/department-level instructional consultants.
  • These workshops typically serve smaller audiences and cover only basic accessibility strategies. Teams are not staffed to scale up the programs substantially beyond current demand.

While there are several asynchronous training modules and occasional synchronous sessions available, the current messaging is that “Training will be available” as the university continues to develop a more comprehensive training ecosystem. This ideal ecosystem would include elements like community support, structured instruction, scheduling, accountability, and clear requirements, which are currently in progress.

Current Policy and Guidance

The University of Washington is committed to ensuring equal access for individuals with disabilities to the same services and content that are available to people without disabilities, including services and content made available through the use of information technology. This commitment extends to all IT systems that the university develops, procures, uses, and maintains, ensuring they offer the same functionality, user experience, and information access to students, staff, and faculty with disabilities as it provides to others. UW’s IT Accessibility Policy applies to a wide range of IT, including websites, software, electronic documents, videos, and devices such as kiosks, telephones, and digital signage.

Additionally, under the Student Governance and Policies, specifically Chapter 208, the university establishes guidelines for providing reasonable accommodations to students with disabilities. These policies are designed to ensure that students with disabilities have equitable access to educational resources and can fully participate in university life, with necessary adjustments or support made to accommodate their needs.

Current Technology/Tools

Some tools to support accessible PDF creation are provided centrally by the University, while others require an additional purchase. All of these tools typically require some minimal training in order to familiarize users with the accessibility guidance they provide and the steps that can be taken to address issues.

Current Data on the Issues

Between Autumn 2023 and Spring 2024, 55,939 PDFs were Image PDFs (flagged by Ally as Scanned PDFs). These files did not have optical character recognition (OCR) and were therefore not searchable.

According to Ally reports, the total occurrence of issues with PDFs for Autumn 2023 and Spring 2024 was over one million. Often, multiple issues may be counted from a single document.

• Around 6% of these issues occurring in PDFs are considered severe by Ally metrics. 58% are major. 37% are minor.
• The top 3 severe issues identified by Ally over Autumn 2023 and Spring 2024 point to general aspects of how the file is created and saved rather than specifics of the file structure or content.
  • Document does not have OCR to allow it to be screen-readable
  • Document is malformed or corrupted and therefore unusable to most users
  • Document is encrypted with security features that prevent it from being opened

Disability Resources for Students (DRS) Maintains a consistent workload of approximately 1,500 requests per quarter, originating from around 400 students across upwards of 1,000 courses. During the 2023-2024 academic year, DRS Access Text & Technology (ATT):

• Served 521 students requiring accessible course materials.
• Processed over 4,500 accommodation requests across 3,450 courses
• Remediated over 41,000 pages of content, including approximately 3,100 pages of mathematical notation (includes graduate-level math/chem/engineering content).
• Charged the budget $2 million for accommodation services they provided (not specific to PDFs).

Approaches to Ensure PDF Accessibility in UW Courses

Due to the complexity of the issues presented, the committee recommends considering a combination of solutions to address the following needs:

1. Set clear expectations for faculty and instructional staff
   Actionable and realistic goals for faculty and instructional staff to bring their course content close to compliance
   Tools: Policies/guidelines, metrics, incentives
2. Support faculty work on basic content
   Provide necessary skills and minimize effort required to meet requirements for content that is easier to address
   Tools: Training/education, Canvas integrations
3. Increase capacity of central services to address complex content
   Hub-and-spoke model to expand capacity of central support to address difficult content while managing relationships with departments and colleges
   Tools: Staffing, technology licenses, automation

Options for Requirements and Guidelines

Our advisory group agrees that policy solutions should be aimed to encourage best accessibility practices rather than discouraging effective teaching practices for the sake of accessibility compliance. Given that faculty have strong pedagogical reasons to share additional resources, adjust assignment expectations, or even reschedule curriculum components during the term, the working group does not recommend prohibiting particular types of PDF content.

Ideally, university policies provide faculty with a list of clear and realistic requirements, preferably with the added incentive that the institution will indemnify faculty who meet those requirements and provide the resources necessary to address remaining content. Here are some examples of requirements that we believe could meaningfully improve PDF accessibility while being realistic for most faculty to implement:

Requiring Canvas use for all courses – Less than 70% of UW courses currently publish their Canvas course sites each quarter, with lower adoption in Seattle than at the other campuses. Requiring that course content be hosted in the LMS, rather than in external sites or storage systems, provides several benefits:

• Faculty can leverage built-in accessibility checkers for files and embedded content
• Students can generate Ally alternative formats available for most files
• DRS can readily access and remediate course content
• Administrators can use Ally and Canvas reports to track accessibility work across departments so that additional resources can be targeted to areas of greatest need
• Staff who support learning technology within and across departments can better scale their support services if faculty are all working within the same platform

Setting a target course Ally score – while Ally does not capture all elements of WCAG 2.1 compliance with perfect accuracy, it is a good bellwether for course content accessibility as a whole, particularly around pages and files. Setting a high but not overwhelming score to aim for in every course – e.g., 80% – would provide a specific target for faculty, which our working group believes would help relieve some of the anxiety UW faculty feel at the seemingly overwhelming task of achieving full compliance.

The University of Tampa implemented this strategy with an 80% minimum requirement, and found that this score was fairly achievable for most courses using basic accessibility principles, and that the specificity of the goal helped motivate faculty to start taking action on files. Winter 2025 Ally scores in UW Canvas courses already averaged around 65%, implying that a requirement around 75%-80% might not be that difficult to achieve by simply eliminating unneeded content or finding more accessible alternatives to key course materials.

Reinforcing use of alternatives to PDFs when possible. When a more accessible file format is easily available and used by students, we should strongly advise against converting to PDF. For example, all Word documents, PowerPoint files, and other standard Microsoft document types should be shared with students in their original file format, not exported to PDF, since these file types are easier to make accessible and all UW students have full access to the Microsoft suite. However, PDFs are still a first habitual response for many instructors who wish to publish content to a course. This is in part due to an outdated idea that PDFs are the preferred file format for sharing with students, since they can be opened by anyone with a browser or mobile device.

To help dispel this belief, we recommend communicating guidelines (not mandates) for which file types should be shared in their original format (.docx, .pptx, etc.) rather than exported to PDF. The guidelines should point to documentation for how students can use their UW licenses for Office 365 on both computers and mobile devices.

Requiring Syllabi/list of course texts be shared with DRS by the given deadline – DRS staff have indicated that timely submission of course material information greatly improves their ability to remediate these files in a timely manner for students with accommodation requests. Clarifying that these deadlines are policy, not guidelines, would enable the DRS team to source or create accessible files in advance, reducing the risk of delayed or unmet accommodations.

Abiding by Fair Use policies – According to DRS staff, many of the larger and more difficult PDFs to remediate are large-scale scans of course texts. These scans often include content well in excess of what is permitted through fair use, and as a result are intentionally created outside of standard workflows through UW Libraries or other services. The working group observed that many faculty have intentionally chosen this route with the admirable intention of reducing the cost of course materials for students, effectively eliminating the need to buy a course textbook.

We believe it would be a mistake to reinforce fair use policies without addressing the underlying positive intent of many faculty who have worked around these policies in the past. However, it would still be worthwhile to reinforce the specifics of these policies as they apply to course texts, so long as any guidance is accompanied by legitimate options for faculty who wish to reduce costs for their students.

Policies to avoid

We do NOT recommend the following policies, which would be difficult to implement and may result in backlash that slows down other accessibility efforts.

Banning PDFs entirely for course materials – This policy idea is ill-advised, if not impossible, in an educational environment as large and diverse as the University of Washington. Many disciplines rely on software outside of Microsoft Office or Google Drive to generate visuals or reports that cannot easily be shared with students without using a universal format like the PDF. If PDFs are partly or fully banned, there would need to be truly massive exceptions to maintain faculty buy-in. If not, faculty are likely to seek even less accessible work-arounds to ensure that they can still share helpful visuals and popular industry tools with their students.

A key example of this is LaTeX. This typesetting format is a disciplinary norm in mathematics, statistics, computer science, and physics, and is a commonly used tool in typesetting academic publications in many other fields that rely on mathematical or scientific notation (health sciences, chemistry, engineering, economics, etc.). Given that many students in these fields will be expected to know LaTeX basics as part of their field of study, banning its use would effectively eliminate an essential professional skill from the curriculum of these programs.

Beyond typesetting; many disciplines rely on specialized software to create visuals, such as circuit diagrams, molecular structures, prototypes, maps, and sheet music. The software itself may not be accessible, affordable, or practical to share with introductory students. While difficult to remediate, PDFs offer more options for tagging and structuring content than exported images, and are extremely handy for sharing specialized content like the examples above. Banning PDFs would make it extremely difficult for faculty to share the content created in these platforms for instructional purposes, discouraging them from creating helpful visuals or teaching students how to use these tools in their work.

Restricting file types that can be used in Canvas – Similar to above. If the University issues a blanket limitation on any file type without substantial exceptions, a subset of instructors will have to either rewrite or simply remove sections of existing course content. Some faculty will certainly be frustrated by this, and those who are most impacted may be more motivated to simply ignore the policy or work around UW systems entirely to avoid having these materials flagged by administrators. For this reason, we recommend using guidelines rather than requirements to minimize excess use of PDFs without needing to explicitly identify all exceptional cases.

Requiring course packs in advance of the term – The biggest challenge for this policy concept is that it requires instructors to have all course materials finalized prior to the beginning of the quarter, not just course texts and a planned schedule currently requested by DRS. For faculty whose contracts start only a couple weeks before the start of instruction, this is an extremely difficult standard to meet. Further, the Center for Teaching and Learning advocates for faculty to seek and respond to student feedback while their course is running as part of a reflective teaching practice. Requiring course packs limits how much faculty can modify content and pacing in response to midterm feedback.

Options for Faculty Training and Support

Training and educational support for this topic, and connected topics, is necessarily multimodal. There are aspects that are best carried out at scale, but some elements really require personalized feedback and resources based on discipline, level, and teaching style.

Ideally, after completing basic training, participants should be able to:

1. Identify federal requirements and UW policies around accessible course content
2. Apply basic accessibility principles when creating new documents or course pages:
   • Heading structure
   • Color contrast
   • Descriptive links
   • Image alt text
   • Appropriate use and structure of tables
3. Use accessibility checkers and built-in accessibility tools in Canvas and Microsoft Office
4. Select more accessible alternatives to PDFs when possible
5. Access UW central services for additional support, including DRS, ATS, DSO, and UW Libraries

Faculty and staff might be incentivized to complete additional training through badges, stipends, or awards/recognition. However, it should be noted that only a subset of faculty will have the capacity to do this based on the other expectations and constraints of their role.

• Contract and part-time instructors are only compensated for contact hours, and may have only a few days between gaining access to their Canvas course and the start of the quarter.
• Junior faculty who contribute to accessibility work may not have it recognized as service in their promotion packet.

To minimize the burden placed on faculty, we recommend keeping required training short – ideally one hour or less – and asynchronous. Additional learning opportunities to supplement this required training can be made available to those with time and motivation, ideally with some systems for recognizing and rewarding this extra work.

Asynchronous sources of training content

Several existing training resources are already available to UW faculty and staff:

Asynchronous UW-hosted training

• Some training has already been built and disseminated in Canvas by various teaching and learning specialists across the three campuses, most notably the Accessibility 101 online course authored by Ana Thompson at UW Bothell.
• Generally, hosting training courses in Canvas is prohibited due to violations of the use policy and data retention risks. However, there is an exception for training on online teaching and learning content, including accessible teaching practices. UW-IT policy permits the use of Canvas courses to support training related to these topics.
• Canvas does not work well to host large numbers of students in a single course, especially if the course includes multiple assessments or interactive components. UW-IT recommends limiting any Canvas training courses to 1,500 students, which may necessitate some additional work to offer training at scale on this platform.
• Any in-house training program would need to have some staff to maintain materials, host sessions, and answer questions. Based on past work for Title IX and Title VI training, this could require as much as one full-time staff member to build and maintain, at least for the first year of implementation and launch.

UW enterprise subscription to Deque University – create an account by following ATS instructions for UW users

• 1 Basic course on creating accessible PDFs, including tagging, remediation, and how to create docs to be accessible when exported
• 1 Advanced course on PDF remediation (presumes existing expertise)
• Courses are likely too detailed/advanced for users with no experience with these tools or other accessibility workflows
• If we rely on Deque, we’ll want to provide recommendations for who completes which trainings based on what they want to accomplish
• Deque tracking has to happen through their platform (completion) – relies on completing entire course

Linkedin Learning paths

• Typically very lightweight without much in the way of assessments.
  Example: Creating Accessible PDFs Online Class
• UW users can create and share custom paths using different courses or even segments of courses
• Grants digital badges for completion of individual courses, which could offer a small incentive to complete training
• Unclear from LinkedIn Learning documentation how effectively admin reports would help track progress of UW users in a specific learning path

Ecosystem Approach to Training

Differentiation is key. Some faculty know more than many technical specialists do about different aspects of digital accessibility. At the same time, many faculty will be true beginners who will know nothing about accessibility at all. There will be no one-size-fits-all solution to learning the skills to make PDFs accessible.

While we recommend keeping minimum required training simple and concise, it’s unlikely that a single training mode will address all learning objectives for learners with diverse existing knowledge and motivation. A robust training model should include asynchronous, synchronous, and personalized solutions to support faculty and staff who need additional instruction to implement these principles, as well as those who are interested in learning more.

We encourage faculty to use Universal Design for Learning (UDL) principles in their teaching and UW’s approach to accessibility training, whether it’s PDFs or anything else, should also follow this approach – multiple means for engagement, representation, and action and expression. Documentation and asynchronous online learning will work for a lot of people, but there will be many faculty, students, and staff who would benefit from instructor-guided synchronous and hybrid options.

Learning communities and peer-to-peer support

In addition to formal required training, informal learning communities could empower continuous discussion and improvement of accessibility practices across departments. Several tri-campus communities of practice have already formed in the past around digital teaching and learning topics, most notably the Evidence-Based Teaching program hosted by the Center for Teaching & Learning. These past programs could easily offer a model for faculty-focused communities of practice (CoPs) to discuss principles and practices for digital accessibility in teaching.

UW-IT also hosts CoPs for staff to discuss shared practices in specific disciplinary areas, including an existing Accessible Web group. These groups could be expanded or bolstered by funding to encourage staff specialists in accessibility to collaborate and share professional development opportunities.

Models from other Institutions:

• Northwestern Model for remediating Canvas Courses – central accessibility support teams remediate all PDFs for faculty, so faculty can be trained and expected to do the other accessibility remediation of their Canvas courses themselves.
• Georgia Tech’s AccessCorps model – students enroll in courses to learn and practice remediation. It’s Pass/Fail, and their work is to actually remediate course content for faculty.

Options for Central Support

If faculty are able to make meaningful progress on simpler accessibility issues through the strategies already described, accessibility specialists will have more capacity to focus on files that require advanced expertise and tools. Rather than distributing tools university-wide, investing in staffing, software licenses, and professional development for accessibility staff ensures that these resources are used efficiently to address the most difficult content to remediate or replace.

One challenge of current support for accessibility work is that expertise is distributed unevenly across colleges and departments. Unit-based instructional support staff are highly motivated to learn accessibility skills, but often do not have the capacity to become true experts in this space. Central teams are small and heavily staffed by students, whose high turnover rate and limited hours can constrain their ability to support advanced remediation. In essence, there are many accessibility generalists in different corners of the university, but very few full-time staff who can afford to specialize in particular tools or discipline areas.

For this reason, we recommend adopting a Shared Services model for centralized accessibility support: offer central support for accessibility, but with staff and liaisons whose specific role is to maintain relationships and offer personalized services to a subset of schools or departments.

• This hub-and-spoke model allows central staff to divide work efficiently while providing units with a single point of contact who is informed about their unique needs. We encourage hiring staff who can specialize in particular content types, such as LaTeX documents, multilingual content, and data visualizations.
• The model can be further bolstered by connecting with faculty ambassadors within departments, who can help facilitate communication between their academic unit and their contact within central services. To make this role a reasonable ask, it should not exceed a couple of hours of work per month, and should be recognized as service in promotion and tenure materials.
• The larger network of ambassadors and staff involved in accessibility work can also weigh in on ongoing issues or ideas that come up around course content accessibility, serving as an ad-hoc advisory council.
• Faculty who take on a more specific or time-consuming role, such as providing direct accessibility support to colleagues or facilitating cross-unit collaboration, could also be recognized through release time or additional compensation.
• All schools/depts have subject-matter librarians assigned to them. With strong partnerships between faculty and their SME librarians, faculty could get help locating permalinks, digitally scanned historic documents, properly OCR’d scans, or cleaner hardcopy texts to scan if there’s no other option.

Automation

To help small teams handle difficult PDFs, we should try to empower these staff with tools that simplify their work and automate simple tasks.

Auto-tagging for PDFs is a first obvious resource to help build efficiency. Modifying tag structure is vastly simpler than creating these tags from scratch, making it easier for specialists to focus their time on the more challenging pieces of a larger document. Adobe Acrobat’s cloud-based auto-tagging is one example of such a tool.

Ohio State has been piloting olmOCR, “an open-source tool designed for high-throughput conversion of PDFs and other documents into plain text while preserving natural reading order. It supports tables, equations, handwriting, and more.” Ohio State staff estimate the cost of running olmOCR at approximately 6 cents per page. This could be used as an alternative to Acrobat auto-tagging, or as a complement to it, with Acrobat as an option.

Options for Technology/Tools

Compared to policies and instructional models, the working group did not arrive at nearly so strong a consensus on the best tools available to support accessibility work at scale. Below are some of the options we explored, categorized by their intended purpose.

Tools to simplify content

• CidiLabs TidyUP – identifies unused course content in Canvas that may be ready for archiving or deletion. This can drastically reduce the cost of remediating course content by eliminating unnecessary files. The iSchool already has purchased and implemented this tool in their Canvas subaccount, and has seen upwards of 50% reduction in files where it was applied.
• Simple Syllabus – Canvas integration to make it easier to create, modify, and maintain syllabi in Canvas directly. The tool includes a lot of improvements over the built-in Canvas Syllabus tool, making it more customizable for faculty, more readable for students, and more easily searched and updated by administrators. Accessibility is a main focus for the vendor, with an aim for WCAG 2.2 AA compliance.

Core Remediation

• Adobe Acrobat Pro/Adobe Creative Cloud – This software is the most widely used for making PDFs accessible, and includes a powerful accessibility checker capable of fixing the vast majority of PDF accessibility issues. However, its popularity does not necessarily distinguish it as the best available software for this purpose: there are both more powerful tools (ABBY FineReader) and more user-friendly options (Cidilabs, Equidox). Committee members advise that Acrobat is not easy for a new user to adopt; even tagging a single PDF would likely require substantial time and training – precisely the burden on faculty that this group has been charged to avoid.
  • A site-wide license would cost between $800,000 and $1.3 million. Given this cost and the challenges of adoption, we recommend that purchasing a smaller pool of licenses to share among motivated faculty and staff could accomplish the same goal at a lower cost.
  • Caveat: the committee does not have access to information about existing Acrobat Pro purchases or other drivers for a site-wide Creative Cloud license outside of PDF remediation. We do not inherently advise against adopting this suite of tools, but do not believe it is a practical solution
• Equidox – Has a reputation for being more user-friendly than Acrobat Pro, with informative tooltips and a gentler learning curve. In a recent pilot, ATS determined that Equidox may not be practical for teams like DRS that handle large volumes of files, but it might be a better fit than Acrobat Pro for individual faculty and employees with less accessibility expertise to tackle remediation of PDFs if necessary.
• UDOIT (Cidilabs) – Similar to Ally, this tool automatically identifies accessibility problems, helps users fix issues, offers alternative formats, and provides reporting to instructors and administrators. It provides more useful tools for handling PDFs than Ally, including a feature that can convert a PDF into an editable Canvas page.
• Canvas LMS Accessibility solution (Pope Tech) – Similar to Ally. Automatically identifies accessibility problems, helps users fix issues, and provides reporting to instructors and administrators. No clear benefit over other platforms.
• YuJa Structural Remediation Max – AI-powered LMS integration to make it easier for users to convert image PDFs into tagged PDFs with correct structural tags and reading order. The vendor also offers other accessibility-related products and LMS integrations that are more similar in functionality to Ally.
• ABBYY FineReader Server – DRS (ATT) is evaluating a more automated workflow based on this (natural upgrade/evolution of current ABBYY setup)

Mathematical content

• Equatio – This would significantly improve DRS efficiency and timeliness with STEM content, but production use is not allowed without a sitewide license (we’re specifically prohibited from using individual licenses for our volume and category of work).
• MathType – Another similar tool being used by an untracked number faculty to create accessible mathematical expressions with slightly fewer features than Equatio. Unfortunately, it cannot be used to remediate PDFs.
• Canvas Equation Editor – A way to create new or remediate content to produce accessible math directly on Canvas without the need of additional tools by either using LaTeX (faster and more straight forward), the interface buttons (slower), or both.

Note: The University of Central Florida (UCF) shared a list of Accessible Math Languages & Tools for Content Presentation with a list of popular tools and their accessibility capabilities. Making mathematical content accessible requires specific expertise – having specialists, or even a committee on this topic, could greatly contribute to this effort.

Remediation Services

• PREP remediation software – AI tool to auto-tag PDFs and other types of documents, then allow user to modify and correct tagging.
  • Continual Engine PDF Remediation service – submit documents for external remediation by Continual Engine. Fixed rate pricing model; turnaround time ranges from days to weeks depending on content. Uses PREP remediation software.
• Crawford Technologies – provides document remediation services at prices and turnaround times that seem to be a good match for higher education institutions. Currently recommended on ATS website for outsourcing PDF remediation.
• SensusAccess – available through either self-service tool or LMS integration.
  • Automates document conversion to alternate formats including digital Braille, MP3 audio files, DAISY structured audio books, e-books and BeeLine Reader.
  • Remediates otherwise inaccessible documents such as image-only PDF files, scanned documents, pictures of text, LaTeX documents and Microsoft PowerPoint presentations into more accessible formats.
  • Currently listed as free on DRS site.

Options for Tracking and Reporting

The committee identified four primary analytics resources that could be used to track institutional progress on accessibility goals and related work. The details of how these could be used would depend on which technology tools the university chooses to adopt.

• The Canvas Adoption Dashboard (hosted in the UW BI portal) tracks college-level rates of Canvas courses getting published, which would help us track trends in overall Canvas usage by instructors.
• Built-in Ally dashboards show both accessibility scores and rates of Ally usage across courses, departments, colleges, and campuses. These can be tracked over time, with both online dashboards and granular data exports.
• ATS, DRS, and other central support teams already use surveys and service metrics to track their work and make improvements. These existing reporting methods will contextualize patterns observed in Canvas, as well as changes in demand over time as more content is brought into compliance.
• If UW adopts TidyUP and SimpleSyllabus, the usage statistics offered from these tools will also illustrate progress on the specific issues they are designed to address. Short-term usage data can currently be drawn from Canvas-generated reports. UW-IT is already working on a project to expand these reporting capabilities.

Conclusions

There is no single elegant solution to ensuring course PDFs are accessible. All the policies, tools, and approaches explored by this committee came with tradeoffs and caveats, and none of them addressed every single case where a PDF might be used in a course. Part of this stems from the sheer popularity of this file type, and part of it comes from the variety of justifications for its use, some of which are legitimate and practical.

As the action team considers how to synthesize the elements mentioned, we advise leveraging the same focus that we drew on for our own analysis: investing in actionable, efficient approaches that minimize the burden on faculty and don’t prohibit effective teaching strategies. Whatever combination of solutions the university implements, we recommend communicating explicitly about this intent and how each piece of this solution will fit with the whole.

While this report may not present easy answers, the working group came away optimistic about the potential for meaningful improvement in the next year. We believe that with a strategic combination of tools, policies, support structures, and compassionate communication, the university will be able to dramatically improve the accessibility of course materials and documents.