Assistive tech can't handle the unusual fonts people use on Twitter to be cute


No. It should serve the needs of those with visual impairments first and foremost.

If doing that means accounting for symbols that are normally seen on the net - the app should be able to do so in a way that is useful to those people it’s supposed to support.


And if those symbols are being used in the expected way? How much discernment between the two cases are you expecting from mere algorithms?


I’m expecting that it can work in the real world in a fashion that works for the users. If it doesn’t- it’s a hindrance- not an accommodation.


You are right. The problem is that developers haven’t figured that out yet. Hopefully, developers will get it down soon. In the meantime, people who rely on the software that exists are just stuck with it.

Edit: you ruining the software developers lives isn’t helping the situation, @KathyPadilla!


And that’s an ongoing issue. It’ll get fixed - things change- it’ll need a new fix.


Ah! We spur innovation through our own lack of foresight. Job security! Capitalism!!


Whaddaya mean we- I just reads the webs :innocent:


Just spurring them to be their best selves!


Touché, @KathyPadilla


Of course, VoiceOver (and every other piece of accessibility software) seems to mostly ignore math accessibility. There are lots more reasonable things to say (that respect the semantic meaning) than defaulting to the name of “𝖍”'s unicode codepoint’s formal name, “Mathematical Bold Fraktur Small H”. For example:

  • math frak h
  • frak h
  • the Lie algebra h
  • h


Thankfully twitter isn’t my problem; but based on what the voiceover (things like the ‘mathematical’ prefix being used character by character) it looks like people are using visually similar, but otherwise unrelated, chunks of unicode to do a font’s job. Text to speech software that decided to start going letter-by-letter just because you italicize something would, indeed, be rather pitiful; but distinguishing an English sentence with chunks of domain specific characters being used because they are twee from, say, a sentence from your average math textbook is a deeply non-obvious problem.

The architecturally sane approach that gives everyone what they want would just be to use the correct glyphs, modified by whatever font and style tags are desired, for a result that meets both visual standards and accessibility needs; but since self-imposed limitations are at least as essential to twitter’s identity as racist trolls and inscrutable feuds that option is presumably off the table here.


I suspect that it’s more complex; but that anyone whose ‘analysis’ isn’t 100% snake oil probably has a toolset that would be amenable to drawing inferences concerning which symbols are commonly treated as substitutes for which letters; and potentially even what the use of a given class of substitutions for part of a tweet suggests.



Hey, Facebook search can’t handle either.

So if your friend uses some screwball font to look cute, you won’t be able to search for them with normal search words.


I believe that’s what’s known in typography as The Raygun Quandary.


Leave fancy text where it belongs, as headings for your primary school poster projects.


This topic was automatically closed after 5 days. New replies are no longer allowed.