Publications in Google Sites
A publication list has to do two jobs at once: let people navigate to the work, and let people cite it.

Faculty and research sites almost always include a list of publications, and almost always attach a link to each entry so readers can reach the paper, the preprint, or the code. The most common practice is to put that link at the end of the citation as a single word:

  • link
  • pdf
  • paper
  • code
  • arXiv

This reads cleanly for sighted users, but it creates problems for people navigating with assistive technology such as a screen reader.

The problem

It's rare for someone to read a thirty-plus list of entries from top to bottom. They jump around — and assistive technology jumps by link. Strip away the citations, and here is the entire navigable page:

        pdf


        pdf


        link


        pdf


        code


        arXiv

How it impacts people who use assistive tech

  • Screen reader users pull up a list of every link on the page and skim it. Entries reading "PDF" identify nothing, so they have to go back into the page body, find their place, and read around each link to work out where it goes.
  • Voice control users say "click PDF" and the software can't tell which one they meant, so it numbers every match and waits for them to count down the page.
  • Keyboard users tab through identical stops with nothing to tell one from the next

Google Sites offers no hidden-text or ARIA options, so the only fix is to change the visible text itself.

The fix

Most accessibility guidance addresses only navigation, replacing "PDF" with descriptive link text, This usually means deleting the URL and breaking copy-and-paste for anyone building a reference list.

This hybrid approach splits the jobs between two elements instead of forcing one to do both: the publication title becomes the link, so a screen reader's links list reads as a list of papers rather than thirty identical "PDF" entries, and the DOI or URL stays at the end as plain text, so the whole entry can still be selected and pasted as a complete reference. 

Neither audience gets served at the other's expense, and the change costs nothing but a few minutes per entry.

In short, to make your publications list more accessible:

  1. Link the publication title to the DOI, arXiv record, or publisher page.
  2. Put that DOI or URL at the end of the entry as plain, unlinked text.

Examples

Example 1 - Before

Duckworth, A. L., Quirk, A., Gallop, R., Hoyle, R. H., Kelly, D. R., & Matthews, M. D. (2019). Cognitive and noncognitive predictors of success. Proceedings of the National Academy of Sciences, USA, 116(47), 23499–23504. [link]

Example 1 - After

Duckworth, A. L., Quirk, A., Gallop, R., Hoyle, R. H., Kelly, D. R., & Matthews, M. D. (2019). Cognitive and noncognitive predictors of success. Proceedings of the National Academy of Sciences, USA, 116(47), 23499–23504. https://doi.org/10.1073/pnas.1910510116

Example 2 - Before

In this example, the publication listing has several destinations.

Yun, J., Alçalar, Y. U., & Akçakaya, M. (2025). No alignment needed for generation: Learning linearly separable representations in diffusion models. arXiv.
Link PDF Code

Example 2 - After

Option 1: Listing the additional links as plain text

Yun, J., Alçalar, Y. U., & Akçakaya, M. (2025). No alignment needed for generation: Learning linearly separable representations in diffusion models. arXiv. https://arxiv.org/abs/2509.21565
PDF: https://arxiv.org/pdf/2509.21565
Code: https://github.com/JN-Yun/LSEP

Option 2: Linking to the other destinations

Yun, J., Alçalar, Y. U., & Akçakaya, M. (2025). No alignment needed for generation: Learning linearly separable representations in diffusion models. arXiv. https://arxiv.org/abs/2509.21565
PDF for No alignment needed for generation
Code repository for No alignment needed for generation

Option 3: Selectively including links

In this example, the PDF-specific link is omitted because it can be accessed from the standard publication page.

Yun, J., Alçalar, Y. U., & Akçakaya, M. (2025). No alignment needed for generation: Learning linearly separable representations in diffusion models. arXiv. https://arxiv.org/abs/2509.21565
Code repository for No alignment needed for generation