Since @lrhodes just recently suggested that there should be a feature to hide inaccessible posts, and @gemlog reminded me today that folks might not even realize their names and toots might be hard for some people to read unless they are told, I guess it is my duty now to write another round of educational posts.
Hmm, this isn't really my area of expertise ... maybe I should just go find some old one and reboost?
Nah, let's do this.
In this thread, I'll list a couple suggestions for adjustments you can do to help out people who are , , or those who for other reasons use a . Boosts appreciated!

But before I start:
If you can't do these things, that's fine. It is not my intention to bash other disabilities. The ways our lives suck are different, but we still should get along. You all are, after all, awesome people!
And if you just don't feel like doing them, that's OK. We are used to it. :)

Show thread

1. Images
Some people on the Fedi can't see your memes, doggos, flowers, art. That doesn't mean we wouldn't want to enjoy them.
There are also those who have to browse on data, and turned loading images off to save it. Those things eat up quite a bit of bandwidth!
And there are those who have trouble figuring out what they are looking at. Maybe because of the way their brain works, maybe because your image isn't clear to everyone.
You can help all of those by writing a caption for your image.
It doesn't need to be an essay. Even just a few words will do. Enough to explain the joke, or the cute pose your kitten is making. Don't worry about it; just write something!
If you have trouble remembering to do so, and would like a reminder, follow @PleaseCaption

Show thread

2. Emojis
Emojis are a wonderful invention! They help us humans, emotional as we normally are, express those emotions in places where voice tone and facial expressions aren't available. Much better than trying to put those feelings into words, right?
Well, people using a screen reader (a software that allows people, primarily visually impaired ones, to read what is on the screen using text to speech) still hear those emojis as words.
For example, πŸ˜‰ is read as "graphic Winking Face", with graphic of course explaining what kind of element it is.
And by default for most screen readers, each graphic is on its own line.
So, what this means is that if you put 3 emojis in your name, your name will be 4 lines long to us. I know some of you like having emojis in your name, but if you could at least limit them to <=3, that'd be lovely.
And if you put a clapping emoji after every word, they'll be on separate lines, too. So maybe don't do that either? πŸ€·β€β™€οΈ

Β· Β· 5 Β· 17 Β· 16

3. "Unusual" Symbols
Unusual is in quotation marks because, with us coming from all around the world, of course not everyone will use the same letters, and they are all equally valid. Hell, Slovenian has 3: č, ő and ž.
There are also fancier characters, for when someone wants to look cooler by bending Unicode to their will.
Just keep in mind that screen readers are not well-equipped to deal with those sorts of characters. It also depends on the TTS engine and language; for example, English eSpeak reads all Chinese letters as "Chinese letter", which is not ideal.
More examples:
ὴ =Letter 1 F 7 4
ΕΎ =z hatchek, though many other TTS voices read it as just z, or not at all.
𝚊 =Letter 1 6 8 A
And so on.
So, while it is not a must, if you could use less of such characters in your nickname when possible, and CW posts consisting only of those, that would be great, too!

Show thread

Oh, right! Almost forgot.
If you have any questions, ask them, and I'll do my best.
If you are curious how a particular emoji or symbol is read, send it, and I'll tel you. I can check with eSpeak, which is the default TTS engine for NVDA, as well as Microsoft's voices and a few others.
And also: Thank you. Both for reading this far, and if it makes you change anything. πŸ’™

Show thread

@Mayana oh, I had no idea an emoji was its own line in screen readers! In practical terms does that mean there is a pause between lines?

@ljwrites Not really. When reading by line, with up and down arrows, it means you have to press down arrow to get to the next line. When, for example, reading by paragraph, there aren't any pauses.
Screen reader users of differing experience navigate differently. I tend to move around by line most of the time, because I prefer reading smaller chunks of text. Others might not mind it; there are some people who use the Say All function to read the entire text at once!
There are also shortcuts for jumping to the next element of a particular type. For example, the letter "b" for the next button. And there is the tab key. So it isn't nearly as tedious as I might've made it sound like.

it also sounds like screen reader implementations are lagging behind on internet culture.
How often do software screen readers get updates, and can most hardware implementations be updated at all?
Are you aware of open source implementations where users can make feature requests that other volunteer devs can try to implement?
What features would you for instance like to see implemented to the one(s) you use?

@FiXato @ljwrites Oh, that they definitely are.
JAWS, a closed-source Windows screen reader that you have to pay about 800 fucking euros for, gets updated once a year, and still has terrible emoji support, in 2020.
NVDA, an open source screen reader that is completely free, gets updates about 3-4 times a year, and has been reading emojis properly for several years.
Which is ... no surprise, really.

@Mayana that sounds like a problem that could be mitigated much more consistently by changing the way screen readers deal with emojis

@ternarypulsar What would you suggest?
Emojis being read and described to us is a good thing. Screen readers that don't do so yet (like old versions of JAWS) are generally seen as bad.
And in most cases, graphics, links, buttons and the like each being on their own line is good, too. It makes things more readable, and pressable things easier to press.
NVDA, the screen reader I use on Windows, does also have an option to use screen layout, which displays things closer to how you see them. I turn that on sometimes, but it can be a pain when, say, some badly made lists become just a long string of text. Most people keep it turned off, because they're used to doing things in a particular way at this point.

@Mayana assuming (!) that emojis are encoded as utf8 or something else that makes them recognizable, the reader could detect repetition or interspersing, as a naive proposition.
I'd say that it's up to you to decide what would be a good way to handle them, and for authors of screen readers to implement those suggestions

@ternarypulsar Hmm ... it does, in some cases. If you, say, put 5 🐘 emojis in a row in a word document, at least NVDA would read that as "5 Elephant".
But since they often show as graphics around the internet and in messaging apps, that's less possible there.
Yes, perhaps screen readers could constantly read ahead, do text prefetching, and use machine learning to guess how to best read it before you even get to it. But what someone is going to read next isn't always predictable.
And besides, I actually had a discussion about this with a friend yesterday, and we agreed that the biggest disadvantage would be the lag that might result. Screen readers, above all, have to be fast.

@Mayana did you experience such a lag? it seems unlikely to me that that would occur

@ternarypulsar Well, the feature does not exist yet, so I can't answer that question. :)
But considering how laggy screen readers can be in other areas, how often they (still!) crash, and how badly some features are implemented ...
I can't be as optimistic as you. :)

@Mayana fair enough, but it still seems more efficient to attempt to get those performance issues fixed than to convince the entire world to not use certain patterns.
FTR, I find those also uncomfortable.

@ternarypulsar I am not trying to do that. See toot 2 in my thread.
I know most people will not change, and that's fine. It is not a huge problem, merely a inconvenience that I'd rather be without. The image description part is the only one that imo, people *really* should adapt on (but yes, yes, I know, image-describing AI).
Look, in the end, I am not a developer, so cannot help you much. If you are, and know Python, NVDA is open source:
As is, of course, Orca on Linux:

@Mayana my intention was not to suggest that you did, rather to emphasize that it might be very simple to implement (you wouldn't need AI for what I described)! and you can help, by opening a feature request on one of those project sites!

@Mayana so you mentioned "graphic Winking Face", that's something that could be string matched pretty reliably for most pages, I'd assume

@Mayana If I can add?

This isn't JUST for social media. If you have a website/webpages/write books that are available in e-format, etc this all applies.

I have a kid who frequently uses screen readers for school reading assignments and assignments with lots of graphics in them tend to be an absolute pain bc the text is ALWAYS referencing the map/picture/chart/thing, but the screen reader has no reference/info for the graphic and it makes the reading impossible.

@jessmahler Yes, definitely! Accessibility doesn't apply just to Mastodon, or just to social media. Image descriptions especially should be provided wherever that is possible. We need more of those!
Thank you for your thoughts. I am just one person and only thought of a few things, so additions from others are more than welcome. My university is actually going through the process of making all our pdfs and websites accessible and my is it a THING sometimes.

@aldersprig @jessmahler That is awesome! But yeah, I can imagine it can be quite a lot of work. Especially with PDFs, which are just ... inaccessible by nature.

Yes. It's sad how the pdf format has become so popular. It's an obfuscated, compiled, page description language. It is very good at what it does, but does not lend itself to indexing for screen readers.
Tex/latex style things should be the future from our past.
This seems to be like what @pizza_pal is attempting. @aldersprig @jessmahler

@gemlog @Mayana @aldersprig @jessmahler PDF is kind of the devil. About 50% of the computing that I've gotten paid for has, in some form, involved extracting structured data from PDF files, and it is not good.

It should be a lot better. One thing that is pretty interesting is TEI (Text Encoding Initiative). It is kind of like LaTeX, but more about the semantics of document layout.

@gemlog @Mayana @aldersprig @jessmahler Things like TEI aren't end-user solutions, but if it was implented in existing software, it could help to produce better documents.

Right. That was my point yesterday: bury the stuff from high level usage.
@Mayana @aldersprig @jessmahler

I took mayana off this thread as she hates this stuff, but karl has an interest I know @kmj
@aldersprig @jessmahler

@gemlog @kmj @aldersprig @jessmahler Yeah, I kind of destroyed my dream job working in the semiotics archive and developing software for working with TEI by being an idiot. But them's the knocks.

It's really a great tool for describing textual documents.

Well, let's explore the depths of your idiocy bud! Lay on McDuff! And damned be him who first cries Latex is enough! :-)
Give some links and ideas on it please.

@kmj @aldersprig @jessmahler

@Mayana this may be a silly question, but how do you tell posts apart? i. e. where one post ends and the other begins. a while ago i tried the built-in iOS screen reader, it announced "landmarks" to me. but it only announced them after reading the text. i found it very inconvenient, as you hear the text first and only then get told it isn't related to previous paragraphs. how do other screen readers do it?

@Mayana oh, and another one if you don't mind. does the capital letter in the beginning of a sentence change the way it is read? is it the full stop that terminates the sentence, or the full stop, space, upper case letter sequence?

@leip4Ier Hmm. That's a hard question to answer actually, because there's a lot of text to speech engines out there, and their rules for this differ.
For example, Eloquence, eSpeak and RHVoice seem to ignore the full stop if the next word after it is not capitalized, where as Microsoft's voices and DECTalk separate the sentences the same way regardless of capitalization.
But of course, I only tested this with a couple sentences. Since I use eSpeak regularly, I can tell you that sometimes, it ignores capitalization, too. There is no logic to be found here! :)

@leip4Ier Both approaches make sense; some languages finish their abbreviations with a period, and pausing isn't desirable there. But people are lazy, and don't like capitalizing things. :) So I can't say either side is doing it wrong.
I'd say that, whenever possible, try using correct grammar. It'll be more easily readable for everyone. Not just blind people; dyslexic people, those for whom English is not the first language, and grammar nazis, too!

@Mayana i see, thanks! that's unfortunate, always using lowercase letters is somewhat personal to me (and many other people too, as far as i can tell)...

@leip4Ier Personal how, if I may ask? Just a way to rebel against the rules of the English language, or ...?

@Mayana more against the smartphone keyboards capitalizing everything automatically, even when it isn't necessary :)

for me there's also a personal story, but i don't wanna talk about it on fediverse

@leip4Ier Isn't there a way to disable that though? There definitely is on mine. But, being a grammar nazi, I obviously keep it turned on for convenience.
It is a pain when writing poetry, though. That format does deserve some more freedon.
And of course. I didn't mean to pry.

@leip4Ier Can't speak for iOS, as I don't use it, but I assume that's something that can be changed in the settings? It definitely can be in Talkback; by default, the name of the element is read first, and the type and state after that, but I was able to change it to state, type, name.
On Windows, the type of the element is read first. So for example, "button Reply to thread"

@leip4Ier On Android, at least with Tusky and Fedilab, each toot is its own object, and is read all at once. I actually don't bother searching for them on the screen most of the time, just swiping left and right to get to the next element.
As mentioned, on Windows I generally move with arrows. Figuring out where one toot ends and another begins isn't a problem at all; there's an username before each one, and several buttons after! :)

@Mayana i see, thanks for the answer! i couldn't figure out how to separate posts on my website: there're visual borders, but not anything for screen readers (except for the post dates, which also act as links). so i decided to add those landmark labels. now after your reply i'm worried those landmarks are just annoying and the date would be enough..

i think i'll try installing NVDA myself and looking at how it's works in mastodon and pinafore :)

@leip4Ier The best way to separate them would probably be headings. At least on desktop, that's one of the main things I use when trying to skip to content that interests me (h key).
Navigation by links (k, or u for unvisited and v for visited) is also possible, but if you also have links inside your posts, slower.
As is, of course, navigation by landmarks using the d key, though those new to screen readers might not remember to use that.
If you want, you can send me the link to your website and I can take a look. There are also tools for checking its accessibility. I'll see if I can find them.

@Mayana it displays mastodon posts, so headings aren't an option...
I would be very grateful if you could do that! The link is On the main page there's the "see demo" link, you can click it and it should say "showing your toots" in a couple seconds. That's the page with the posts.

@leip4Ier That does show up fine. Although, there is no username before the date. Is that intentional?
No, I guess using landmarks (or complimentary landmarks, as they're read here) isn't really necessary in this case. The post privacy and date make the start of the next post pretty clear. Though it does make it easier to skip to the next post, as I've mentioned.
Hmm. Maybe you could put these in a list? Then people could use the letter i to move to the next/previous list item.
But if not, then link navigation works well enough.
(Sorry for the late reply; I had it mostly written, but then something came up.)

@Mayana yeah, it is intentional, bc these all are posts by the same person (supposedly the user). Boost, favorite and bookmark pages have usernames at the top of each post, as those may be from different people.
I see! I think I'll keep the landmarks though if they aren't annoying.
I'm not sure about a list, I'll look into it. It's an interesting idea, usually only menus and lists are done using lists, so I didn't even think of that as an option.
Thanks a lot! Good to know it is usable :)

@leip4Ier Interesting. I don't recall seeing any posts with an username at all; maybe there were no boosts/favorites, but maybe I was just not paying enough attention when reading. I'll check again in a bit.
Well, to be fair, I don't know how distracting a list would be visually. It was just a suggestion, and I am no web dev.
The landmarks are slightly annoying in the sense that they are read for each post, and "complimentary landmark" is two words, which is a bit distracting even when the speech is at a high speed. But they are the sort of thing people could get used to.
I'll have to check the guides I sent you. The one from Thomasorus will surely have some better alternatives than lists.

@Mayana boosts and favorites are on separate pages, there should be links at the bottom of the page.
Lists can be styled very differently, I think most (sighted) users usually don't know when the element they're interacting with is implemented as a list. So that probably won't be an issue. And it would make sense logically, as the page is basically a list of toots.
Oh, that's no good!

@Mayana I think I'll read the links tomorrow, I only skimmed through them for now. Previously I tried to follow WCAG guidelines, but those don't say much about such cases. Or maybe I missed the place where they did...

@leip4Ier It's entirely possible they don't; this is a pretty specific case.
No rush! And feel free to ask if you ever want more opinions, and I'll make sure to boost it. I'm far from the only blind person on here, and there are some blind webdevs around.

@leip4Ier Oh, you're right! I'm sorry, missed those links. Everything shows up fine, then!
I'm glad to hear that wouldn't be a problem. :)

@Mayana I installed NVDA and am testing a list-based implementation. Turns out the complimentary landmark was not the toot itself, but the "aside" element with the date and post visibility. So I think NVDA doesn't even see toot containers as landmarks, even though iOS' VoiceOver does. This also explains why it doesn't read the post number (I thought it would do that).

@Mayana in my test implementation it reads list items in one go, which seems inconvenient to me. But you say you read toots that way on your phone, so maybe it isn't actually bad. Other than that it works fine. I think I need to do more research before deciding on a solution, HTML accessibility turned out less straightforward than I expected it to be.

@leip4Ier There should be options for reading by word, line, paragraph and such. There definitely are on Android, by swiping up or down. I think on iOS you access the settings menu VoiceOver has by making a sort of half-circle shape? Not sure though.

@leip4Ier That's interesting, since reporting landmarks and regions is turned on by default. But I've definitely been able to access some landmarks before that hadn't been reported, so I guess that makes sense.
For example, on Mastodon's web interface (at least in advanced view) the headings for Notifications, Home, Local Timeline and such are also landmarks apparently (they can be navigated to using the D key), but aren't reported as such.

@Mayana sorry to bother you again, but I thought you might be interested in my findings. It turns out that NVDA doesn't report article tags (I use them for toots) by default, but that's configurable! Reporting can be enabled in settings in the Document Formatting category and hotkeys for article navigation can be set by opening the gestures dialog from a browse mode document.

@Mayana I guess I'll keep it the way it is on my website, as other screen readers support article navigation out if the box and I think link navigation is a reasonable compromise for NVDA users. I'll remove the complimentary landmark though, it's annoying.

@leip4Ier@infosec.exchangeOh yeah, didn't even think of that!
I've definitely seen that option sometime, but after experimenting chose to keep it disabled because of how many articles there were. It's been a while though, I need to test again.
Thank you for doing so much research into this! Pretty impressive, honestly. :)

Sign in to participate in the conversation

Hometown is adapted from Mastodon, a decentralized social network with no ads, no corporate surveillance, and ethical design.