Skip to main content


When I write alt text, my approach is to ask "what information does this image convey to me, a sighted person?". If the post is a joke, the information is the context needed to get the joke. If the post is art, the information's what the reader must know to guess what feeling the art would convey.

I don't mention detail if it's detail a sighted person would overlook; if the post contains text, I only caption text I would read (IE don't caption the NYT header, say "an NYT page saying…").

reshared this

in reply to mcc

Minor rules I follow:

- Sometimes a joke is a sort of secret. Like, the "joke" hinges on you knowing without being told that the man in the photo is David Hasselhoff. If captioning helpfully would "ruin the joke" for sighted readers, I let the joke ruin.

- I was once told that CAPITAL LETTERS are often read by screen readers as acronyms. So, because alt text more than anything else is for the benefit of screen readers, I never use capitals as EMPHASIS, not even when captioning uppercase text.

bituur esztreym reshared this.

in reply to mcc

The correct way to caption/alt-text images is, of course, whatever it is that blind users say would be most useful. However I have read several statements by blind users on this subject, and they contradicted each other a little bit. So the above is the best synthesis position I've been able to find.

One thing I'd very much like to know is, when a screen reader encounters emoji, whether it is more likely to say the name of the emoji ("Face with Party Horn and Party Hat") or just skip it.

in reply to mcc

Oh, one other alt-text rule I follow. Sometimes an image contains no additional information at all. When this is the case, I actually say so, as briefly as possible. If the post already describes the image fully, I just caption "As described". If multiple photos contain the same thing, I just say "ibid." (or maybe "a burning dumpster", "same dumpster from another angle", "ibid.", "ibid."). If I *say* there's no useful information, a screen reader user won't have to *wonder* why it's blank.

bituur esztreym reshared this.

in reply to mcc

i specifically wanna know if it's more accessible to give that info in the post itself or under the pictures.. (i'm imagining there's an extra interaction step to have the picture description read out)
in reply to mcc

my understanding is that they say the name of the emoji; that's certainly the behaviour for macOS's TTS system anyway. i read a post once (attached) complaining about how irritating clapposting is on a screenreader and i did a deep dive.

my personal take is announcing the name of the emoji makes every message fucking hysterical, and I'd be totally down for a font that renders each emoji as its description, but I can see how it would get irritating if you lack the facility to skim

in reply to mcc

just stop using emoji.

Specifically for this topic, emoji are visual and you're writing alt-text for blind people, so they absolutely have no place there.

This entry was edited (15 hours ago)
in reply to mcc

if it's a meme image I call out the image or template, but then just provide the text (if any)

Ex:

"Guy in hotdog suit meme"

"Drake meme (anime):
Disapprove: ...
Approve: ..."

"Butterfly meme:
Butterfly: ...
Caption: Is this a ... ?"

Basically, assume they have the same grasp of pop culture as the rest of my audience and will "fill in" the rest of the details based on that.

This entry was edited (18 hours ago)
in reply to mcc

Bullet point 1 is extremely reasonable, especially because for sighted readers on every user-agent I know that most sighted readers use, the alt-text won't show anyway unless they ask for it.

If it "ruins the joke," it's acting as an explainer someone doesn't have to go search for, and that's gravy. I think we can class this as an example of "accessibility benefits the abled too."

in reply to mcc

this is sensible and is the approach I go for as well. I haven't done it in a while but I like writing alt text for aesthetic photos I've taken like I'm translating the photo to a poem.
in reply to mcc

NYT Headline, should be included; just NYT Headline: Headline