[accessibility] Problem: screen reader + ligatures

[Jason White]

Does the Unicode map in PDF allow this problem to be solved? Surely this 
would be an important use case, but I haven't looked it up in the PDF 
specifications.

Also, my pragmatic solution would be to generate an accessibility 
standards-conforming  HTML version of the document, which is more 
accessible to more screen reader users anyway.


On 18/6/21 7:12 am, Jonathan Fine wrote:
> Hi
>
> This arose from a post from Sanchit Ghule, a blind user of LaTeX who's 
> reported that his screen reader skips ligatures, such as "fi" in 
> "significant".
>
> My first step is a web search.  I searched for 'screen reader 
> ligatures'. I've found the following interesting links (followed by my 
> brief comments).
>
> 1. https://www.mobileread.com/forums/showthread.php?t=296719 
> <https://www.mobileread.com/forums/showthread.php?t=296719>
> A discussion in 2018 among contributors to the MobileRead library. 
> They mention Microsoft's TTS (Text To Speech). The participants are 
> well informed and cooperative. The conclusion (on page 3 of the 
> thread) is that two separate versions (sighted and blind) would be best.
>
> 2. https://miktex-users.narkive.com/t5tuU541/suppressing-ligatures 
> <https://miktex-users.narkive.com/t5tuU541/suppressing-ligatures>
> A discussion in 2002 in MikTeX users forum. Notes that ligatures cause 
> search to fail, as well as screen readers. Discussion focuses on 
> preventing ligatures, mainly via active characters and virtual fonts 
> (which I consider impractical).
>
> 3. https://github.com/google/material-design-icons/issues/732 
> <https://github.com/google/material-design-icons/issues/732>
> Ligatures causing accessibility issues
> Issue raised in 2018. Text in icons is read by screen readers, and a 
> "ligature-method" is used. "The point is, the default solution with 
> ligatures, albeit elegant, does come with a whole package of 
> accessibility issues that needs to be addressed."  No responses, and 
> the issue is still open.
>
> 4. https://paizo.com/threads/rzs2u5ra?The-ff-typographic-ligature 
> <https://paizo.com/threads/rzs2u5ra?The-ff-typographic-ligature>
> Paizo is a role-play game publisher, eg Pathfinder. This 2017 post is 
> about the failure of search in Paizo PDFs for words such as 
> "janderhoff". The problem was reported to the same forum in 2010. The 
> thread mentions tagged PDF.
>
> 5. 
> https://accessibility.oit.ncsu.edu/generating-accessible-pdf-documents/ 
> <https://accessibility.oit.ncsu.edu/generating-accessible-pdf-documents/>
> IT Accessibility page. States "Fonts are mapped into standard Unicode 
> equivalents, ensuring reliable translation of all text and correct 
> interpretation of ligatures and hyphens. This allows the screen reader 
> to correctly read all characters and words." BINGO. An accurate and 
> succinct statement of the problem.
>
> I've also searched for 'screen reader ligatures latex'. This gives 
> additional interesting results. Mostly, I've not had time to add comments.
>
> 6. 
> https://tex.stackexchange.com/questions/262009/historic-font-and-missing-ligatures 
> <https://tex.stackexchange.com/questions/262009/historic-font-and-missing-ligatures>
>
> 7. 
> https://subscription.packtpub.com/book/hardware_and_creative/9781784395148/3/ch03lvl1sec37/enabling-the-searching-and-copying-of-ligatures 
> <https://subscription.packtpub.com/book/hardware_and_creative/9781784395148/3/ch03lvl1sec37/enabling-the-searching-and-copying-of-ligatures>
>
> 8. https://tug.org/pipermail/pdftex/2018-September/009143.html 
> <https://tug.org/pipermail/pdftex/2018-September/009143.html>
>
> 9. 
> https://www.homepages.ucl.ac.uk/~ucahmto/elearning/latex/2019/05/06/accessibility-regulations.html 
> <https://www.homepages.ucl.ac.uk/~ucahmto/elearning/latex/2019/05/06/accessibility-regulations.html>
> Interesting page, but doesn't contain "ligatures".
>
> 10. https://latex.org/forum/viewtopic.php?t=21808 
> <https://latex.org/forum/viewtopic.php?t=21808>
> 2012 post: Searchable ligatures in PDF docs from XeTeX
>
> 11. 
> https://ctan.math.washington.edu/tex-archive/macros/latex/contrib/tagpdf/tagpdf.pdf 
> <https://ctan.math.washington.edu/tex-archive/macros/latex/contrib/tagpdf/tagpdf.pdf>
> 2021: This package is not meant for normal document production. It is 
> mainly a tool to research tagging
> It's interesting that in this PDF document "difficult" has the "fi" 
> ligature AND "difficult" works for search AND for copy-and-paste.
>
> That's all for now.
> --
> Jonathan
>
>
>
>
>