Wikipedia talk:Manual of Style/Accessibility/Alternative text for images/Archive 6

This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.

Archive 1

←

Archive 4

Archive 5

Archive 6

I removed the note: "The screen reader emulator Fangs confirms this." Behaviours of screen readers changes, and are also made to produce as good results as it can despite a website's mistakes and poor quality. And there are other kinds accessibility hardware that behaves differently. An accessibility expert thaugh me testing with a screen reader is not sufficient and not reliable. We should comply to W3C'WCAG 2.0 techniques, and try to meet the AA accessibility level. That is all. Yours, Dodoïste (talk) 20:03, 30 July 2010 (UTC)

{{

) 00:36, 7 September 2010 (UTC)

I guess the page got changed. It used to have good examples. Now the examples are to few and not super helpful. - Peregrine Fisher (talk) 22:57, 10 September 2010 (UTC)

The page got changed, yes. It used to have long alt text and examples that were not very effective for screen readers users and difficult to apply by editors.

Now if you can give a few examples and give us hints to improve this page it would be appreciated. ;-) Regards, Dodoïste (talk) 02:40, 11 September 2010 (UTC)

The page is titled "Alternative text for images". The focus should be on alt text examples and how to properly write alt text, not on what alt text sounds like in screen readers, and I disagree that they were difficult to apply, the more examples you have to learn from the easier it is, at least for me. -- œ^™ 22:26, 31 December 2010 (UTC)

The page got massively changed I noticed. With all due respect to SlimVirgin, Eubulides' version, with all of its informative examples, is much better. If the page needed improvement then improve what's already there, the page didn't need a complete overhaul and reduction, that's going backwards. I propose that all the old examples be moved to a separate page of examples linked from this one. -- œ^™ 22:19, 31 December 2010 (UTC)

This suggestion is never appropriate as alt text, and is in no way supported by the reference attached to it. Pretty troubling that it's been part of this page for so long. Unless I'm missing something critical, this should be removed. Chris Cunningham (user:thumperward: not at work) - talk 09:18, 9 October 2010 (UTC)

What is the matter? Could you please explain how you came to those two conclusions? This is an official WCAG 2.0 technique: G74: Providing a long description in text near the non-text content, with a reference to the location of the long description in the short description.

However, you may have some really awesome arguments, which makes you a better accessibility expert than all the guys who wrote WCAG 2.0 and the members of WebAIM. ;-) So please provide details. Dodoïste (talk) 10:50, 9 October 2010 (UTC)

Providing the location of the fuller explanation in the alt text is fine. Making that the sole content of the alt text certainly isn't. Chris Cunningham (user:thumperward: not at work) - talk 03:02, 11 October 2010 (UTC)

You have a point there. :D The guidelines for this technique are indeed vague, and I would not be surprised it this technique was badly applied. Do you have any examples of misuses? Dodoïste (talk) 03:22, 11 October 2010 (UTC)

No specific misuses except for the example in the guideline itself. Chris Cunningham (user:thumperward: not at work) - talk 03:29, 11 October 2010 (UTC)

OK. Well, sure. The example basically says "insert descriptive caption here", instead of giving an example of what would be a descriptive enough caption. If you can improve that - without removing, by rephrasing only - I'd be grateful. :-) Dodoïste (talk) 03:38, 11 October 2010 (UTC)

Has anybody noticed if you use alt= it no longer works but Alt= does.REVUpminster (talk) 00:15, 8 December 2010 (UTC)

• I might be wrong on this but some work, some do not.REVUpminster (talk) 00:26, 8 December 2010 (UTC)

See Wikipedia:Wikipedia Signpost/2011-05-02/Technology report. I was the one who reported bugzilla:18682 a long time ago, and I'm very thankful to the developers for finally fixing it.

We'll soon be able to add alt text to our galleries. It's not live yet, though. It should go live a the next MediaWiki update, and I'm not sure when this will happen. Anyway, keep an eye on it, and see if it gets enabled in a near future. Yours, Dodoïste (talk) 07:52, 3 May 2011 (UTC)

Cool! :-) Graham87 09:36, 3 May 2011 (UTC)

Yeeeepee, it's live now! Altought it may have been live for quite some time already. Probably since MediaWiki 1.18 deployment in October. At any rate, we can now discuss its use, update best practices, and all. :-) Dodoïste (talk) 01:17, 7 December 2011 (UTC)

Well, on a second thought, since en.wiki uses mostly templates to generate galleries, best practices may not need be updated. However, it may be beneficial to update the templates. Templates currently create a table containing a list of images which have an alt text. We can now change it for a standard gallery. It's more semantic and accessible (no more table read aloud by screen readers, but silent divs). Dodoïste (talk) 01:48, 13 December 2011 (UTC)

One of your examples is "alt=Tony Blair and George W. Bush shaking hands at a press conference. " I'm in the habit of removing the period from such alt captions because it isn't a complete sentence, according to

) 14:35, 8 July 2011 (UTC)

A screen reader generally will not read a period, but merely pause. So the question is whether it already pauses enough after the link. I would think they would, but I don't know. — trlkly 21:24, 15 October 2011 (UTC)

Unless the alt text provides a pixel-by-pixel description of the image, the above statement, taken from the introductory paragraph, is simply not possible. A description can serve the same general purpose as the image, but it is not technically accurate to say that it conveys the same information. In addition, having no images, whether because of visual impairment or other reasons, is already a loss of functionality outside our control. Effective alt text can reduce its impact, but we should not pretend that we can eliminate it. Feezo (send a signal | watch the sky) 20:27, 30 July 2011 (UTC)

Yes, it should be reworded. The point is, for images such as File:Proteinsynthesis.png alt text is crucial to understand the content of the article (where this image is inserted). Alt text is needed when the image contains an information that is important to understand the subject of the article. If the information is already provided textually in the article, like in Cell (biology)#Creation, alt text is almost useless. Did my explanation helped ? Cheers, Dodoïste (talk) 01:58, 2 August 2011 (UTC)

If a browser tries to obtain an image, but fails, and the image has alt text, the browser may display the alt text instead. I think we should mention that.

To illustrate the display of alt text, I have created File:Metropolitan Railway, Praed Street Junction alt text.PNG. This is a screenshot that I took from Metropolitan Railway when the lead image File:Metropolitan Railway, Praed Street Junction.jpg failed to load because of a slow server. --Redrose64 (talk) 21:10, 26 April 2012 (UTC)

At the accessibility workshop I went to last year at Wikimania, one of the chief complaints from blind users was our over-use of alt text. According to these users, in most cases the captions are sufficient to explain the gist of the image, and having to listen to both the alt text and the caption is often redundant and tedious. For years, we've been emphasizing alt text so much that it has become a de facto requirement, even when it serves essentially no purpose. If we're really interested in serving the needs of blind readers we should pay attention to this feedback and change our guidelines to emphasize that alt text should only be added when the caption does not sufficiently describe the image. Otherwise we're just creating invisible cruft. Kaldari (talk) 21:43, 6 June 2012 (UTC)

That's right. But it's more complicated, because MediaWiki does not handle images without alt text as it should. If we do not duplicate some information in the alt text, MediaWiki is going to place the filename in the alt text. So we are saying that the should be minimum duplicate, but cannot do without it at the moment. Dodoïste (talk) 22:38, 13 July 2012 (UTC)

It's even more complicated, or perhaps even more simple, than you suggest Dodoïste. Many seem to forget that the purpose of alt text isn't to describe the image, but to stand as an alternative for it, for those who for whatever reason cannot or choose not to view the images. Often the caption is sufficient, so I very commonly use something like "alt=Photograph" where's that's the case, because as you say, the way the software works means that there has to be some alt text. But there's no excuse for duplicating any of the caption and even less for the long and rambling alt text that's become all too common here.

Fatuorum

23:05, 13 July 2012 (UTC)

That's right Malleus Fatuorum. :-)

As I'm on a break, and I rarely edit articles, I haven't seen examples of very long alt text that duplicates the caption or the surrounding text. Could you provide such examples?

Ultimately, when bugzilla:34750 will be solved it will make it really easier for our editors. But I have no idea when this is going to happen.

@Kaldari: was there an accessibility workshop at Wikimania 2011 in Haïfa? Or are you referring to Poland 2010? Dodoïste (talk) 23:50, 13 July 2012 (UTC)

Black and white: stylized coiffure, jeweled necklace & earrings, fur wrap around bare shoulders.

Were any proceedings kept of the Accessibility meetings? We should incorporate specific suggestions for Alt= tags for all classes of images. Portraits, still lifes, landscapes, technical, synthetic visualizations, etc., etc. I mean, for biographical articles, as in Anne Francis, there are 3 photos, including this one: what text suits?

• "alt = Head & shoulder shot", or
• "alt = Black & white studio publicity glamour portrait of Francis, with stylized hairdo, bared shoulders, a jewelled necklace, earrings, and a fur wrap."

I mean, what's important? Does this vary based on gender/ethnicity/age? I'd rather know than guess, or assume. Data, and discussion are important, plus, I don't want to waste time on unwanted writing. --Lexein (talk) 02:18, 23 August 2012 (UTC)

You need to remember that alt text is an alternative to the image, not a description of the image, and that it and the caption need to work together. "Black and white photograph" would be sufficient, but the caption probably ought to be boosted as well, maybe something like "publicity glamour shot, c. 1955".

Fatuorum

02:41, 23 August 2012 (UTC)

(Aw, please don't condescend with "you need to remember that".) I observe that alt text, as used in the wild, uneasily serves two masters: as alternative for non-loading images, and as supplementary info for the visually impaired. I agree about caption and alt working together. So if caption is "Publicity glamour shot, c. 1955 (age 25)", then the alt text should tend to the details which would be redundant in the caption. After the Istanbul example you offer below, the alt text might be "Francis with a stylized coiffure, wearing jeweled necklace and earrings, and a fur wrap around bare shoulders. Black and white.". No? --Lexein (talk) 08:22, 23 August 2012 (UTC)

In short, no. Malleus Fatuorum pointed out Istanbul as a bad example. That is the biggest problem with the inclusion of alt text in editorial practices right now. It turns out to be more complicated than expected. Alt text is intended to contain "functional information", or "information needed to understand the article", or "data shown in the image". In most cases, it could be empty - we insert one word only to compensate a current software bug. In cases of pictures such as in Anne Francis and buildings and landscapes such as in Istanbul, the caption is sufficient. Often, it is good to improve the caption though. People with sight issues that can't be corrected by glasses (

low vision

...) will appreciate to have the crucial information in the image shown in the caption. In those cases, alt text is useless. It might be irritating for a screen reader user when we depict a simple portrait of a baseball player in every detail like a major piece of art. It might feel out of place. We'll have to provide detailed good practices on improving captions, that is my next big task on my to-do list.

At least this is how I understand WebAim's best practices on alternative text. But it would be good to have a detailed review from an actual screen reader user. I'm going to ask Graham87. Cheers, Dodoïste (talk) 13:12, 23 August 2012 (UTC)

Thanks, Dodoïste, for letting me know about this discussion. I'll respond in detail tomorrow. Graham87 14:23, 23 August 2012 (UTC)

There seems to be a variety of published opinion about how the image should be replaced (with no text, or some, or a lot). See below.--Lexein (talk) 17:59, 23 August 2012 (UTC)

Dodoïste, take a look at pretty much any FA nomination in the last year or so.

Fatuorum

02:52, 23 August 2012 (UTC)

Thanks for sharing your thoughts. There is a vast problem to solve, here. The sooner the better. But let's do this in a sensible manner. We don't want our editors to feel disgusted to see all the work they put into alt text thrown away with a "undesirable" tag on it. Truth is, lengthy descriptions of images such as provided in Istanbul have their place on Commons. They will have to be improved, but they will ultimately be very useful to everyone there. Not as alt text, but as a detailed caption on Commons. I believe that makes sense, and that this approach might save us from a drama. Cheers Dodoïste (talk) 13:12, 23 August 2012 (UTC)

Screenreaders can optionally ignore alt text, can't they? Anyways, might be good to understand the complaints - desired alt text length, which articles were trouble spots, etc. Might also be good to select some master sources on accessibility standards (probably not only WebAIM), and even usability studies. WebAim's suggested practices on alternative text, as illustrated in their George Washington portrait examples, seem to prefer to convey no functional information about photographs. This advice contradicts Maximum Accessibility: Making Your Web Site More Usable for Everyone by John M. Slatin, Sharron Rush starting with page 259. In the end, it may not be possible to simultaneously satisfy accessibility goals and the dissatisfied screenreader users mentioned at the top. Forced to choose, I'd support accessibility goals. In compromise, I'd say "keep caption + alttext less than 10-15 words" or something to that effect. Out of a 2000 word article, is that a perceived burden? --Lexein (talk) 17:59, 23 August 2012 (UTC)

The "master source" on this subject is the

WCAG 2.0

, nor without the advice of someone with experience on the field.

In this case, the source you found does not contradict my saying. It applies to different cases, such as when the image convey important information in context. For example, we would provide a lengthy description of important pieces of art on a page about art. But when an image is decorative, such as the portrait of a baseball player shot in a hurry by a Commons volunteer, no alt text is needed.

As for WebAim's source, it is quoted several times in W3C's guidelines as a good reference (H67, alt techniques, FLASH 11). This is rare, and it means WebAim is regarded as a good reference. Cheers, Dodoïste (talk) 20:20, 23 August 2012 (UTC)

Thanks for the excellent W3C links & info about WebAIM. Yes, different cases require different treatment; I suggested (above) having WP guidance for different classes of images, in different situations. Biographies and portraits in them were the reason I came here for guidance. At W3C we read "Provide the same informational content as the image", "according to context", "be succinct", "avoid redundancy". All fair enough, but nothing explicit about portraits in biographies. There is this at WebAIM: "We could go on with different scenarios, but I think you get the point. There is no one right alt text for any particular image. It all depends upon the context and the purpose of the image. This is a judgment call that the page's author must make.. So here's another attempt for approval/reasoned disapproval (for the above Anne Francis image):

caption: Publicity shot, circa 1955 (age 25)

alt=Black and white: stylized coiffure, jeweled necklace & earrings, bare shoulders, fur wrap

We may agree that a person's appearance is important to sighted and unsighted persons, via different sensory experiences, and to differing degrees. For biography portraits, perhaps describing clothing style and accessories does not serve - then what does? I am trying to avoid "excruciating detail". It's hard to see how an empty alt text helps anyone - in fact, at W3C Images which enhance... there are these two links:

• about screenreader actual use. The author debunks complaints about excessive alt text, by noting that users skip around.
• Blind Ambition: Pictures into words (full PDF) Strongly advocates descriptive text, for the benefit of blind readers who wish to "read" it.

These are not presented as an excuse for long, redundant, inappropriate, or wrong-in-image-context alt text: to the contrary, I merely advocate effective, concise alt text.

I reiterate for discussion: Screenreaders can optionally ignore alt text, can't they? Those users bothered by WP compliance with accessibility practices can switch it off, or skip it. And/or WP can implement turning off alt text in user preferences, too. And/or we can absolutely seriously address cases of incorrect use of alt text, after we agree on what can be correct. Addendum: Sorry for being long-winded. I'll back off. --Lexein (talk) 23:55, 23 August 2012 (UTC)

Thanks for the long research and efforts you are putting into this. Now you are bringing interesting thoughts and points in the discussion. First I'll wait for Graham's opinion on this matter, as I feel we cannot take a decision by ourselves. Indeed, an extended survey and usability testing would be helpful here. Unfortunately, I don't have the necessary resources to do that as of now. That is something the WMF should help us doing. Dodoïste (talk) 08:32, 24 August 2012 (UTC)

Yes, it is possible for screen reader users to easily turn off or skip alt text.

Re: The images in Istanbul, I really like the alt text for the map images, as they give me a sense of the locations of the features depicted on the map. However I'm not so keen on the alt texts for some of the buildings – the one about the acropolis is IMO one of the better examples because it describes what is actually left of the structure (however I think we could cut out the "in the Islamic style" bit, as that's fairly obvious). However I'm not sure how the yellow car described as being in front of the post office is relevant to the city of Istanbul – IMO it'd be better to cut out such details from alt text. I feel similarly about the constant mentions of Turkish flags ... I wouldn't be surprised at all to find them in front of government buildings in Turkey.

I like the idea of the current alt text at the Anne Francis article, but I'd prefer it if it were written like a full sentence as the caption is.

Even *I* am sometimes not too sure what to write as alt text, because I don't know what I'm missing due to not being able to see the image. Sometimes it's clear-cut; an image of a sign written in English probably shouldn't contain the text of the sign in the caption, but it should definitely be in the alt text. For photos of things such as buildings and portraits, it depends on the image's relevance to the article and how "special" the image is, per Dodoïste's comments above. Graham87 09:31, 24 August 2012 (UTC)

Thanks for this very interesting feedback Graham. So there are different views on the matter, as Lexein suggested. I believe actual blind users should have the final word on what alt text are supposed to contain. Graham87, if you know other blind Wikipedia users, blind people or communities, perhaps you could ask them to share their opinion on the matter too? The more input we get, the better we will be able to provide useful alt text. Thanks in advance! Dodoïste (talk) 10:40, 24 August 2012 (UTC)

Very encouraging, Graham. I agree with Dodoiste on getting input from as many as possible vision-impaired users/editors. I had a wild idea for an ongoing project, the Alt-text Noticeboard, where existing or proposed caption/alt text is evaluated and refinements proposed, a bit like the

WP:DYK

refinement process, only much lighter weight. This could be done bottom up (jumping in and evaluating/!voting; eventually the general shape of the preferred standard may take shape, and be codified in this essay), or top down (setting some criteria first such as MOS, if that helps, then testing against articles, and revised as needed.)

Random question: do screenreaders read Wikipedia infoboxes in the order,title, then image caption, then alt text? If so, then maybe the "full sentence" goal is a very usable one for building cap/alt text that scans, such as, perhaps, "Anne Francis / Publicity shot circa 1955 (age 25) / Black & white, with stylized hairdo, jewelled necklace and earrings, and fur wrap around bare shoulders." --Lexein (talk) 11:43, 24 August 2012 (UTC)

I'll try to get into contact with some blind users about alt text on Wikipedia. I'm not really well connected with the blindness community, however. Screen readers always read the alt text before the caption, so they both should be written in a similar format. I'm not sure of a merit of a separte alt text noticeboard ... this talk page seems to function quite nicely for that purpose. Graham87 15:13, 24 August 2012 (UTC)

Thanks for the reading order info. Okay, so maybe not a noticeboard, but how about a (hidden) Category:Articles needing alt text attention with an edit summary? This could help accumulate and semi-organize work. --Lexein (talk) 21:01, 24 August 2012 (UTC)

Hmmm, I'm not sure now! The main disadvantages of a category are that it can't be annotated and users can't check to see which articles have previously been in the category. An actual onwiki iscussion would be better in this respect ... but I still don't understand the advantages of making an entire noticeboard out of it when we have this talk page (which is archived and readily searchable). Graham87 03:47, 25 August 2012 (UTC)

(arbitrary break for editing)

Ok, I shall start with the Joan Baez alt text I just added (it had none). What's the threshold of "too long" (annoying to both blind readers, and sighted readers viewing with images off). At what point should too-long alt text should be moved to the image's file page description, or alt text there? If moved, should it remain alt-text, or be spelled out in the description, even if it's redundant for sighted users? --Lexein (talk) 07:03, 25 August 2012 (UTC)

Once again, please use full sentences in the alt text. "Standing behind too-tall podium bristling with microphones, wearing plaid sleeveless top, longish hair in feather cut" should be "Baez Standing behind a too-tall podium bristling with microphones, wearing a plaid sleeveless top and longish hair in a feather cut" or something. And quotes do not work in alt text; they display as """; perhaps that's a JAWS oddity? Graham87 15:29, 25 August 2012 (UTC)

Fair enough! Didn't know quotes cause trouble. Thanks. Length ok? Content helpful? I was working against the pallidly functional, "A woman stands behind a podium holding a guitar". --Lexein (talk) 21:36, 25 August 2012 (UTC)

The quotes issue sure looks like a JAWS bug. They are proper " entities. —TheDJ (talk • contribs) 13:30, 26 August 2012 (UTC)

Yep, the length and content are fine. Graham87 03:42, 26 August 2012 (UTC)

Actually, I take issue with the alt text that's been added the Baez article. The lead image has this alt text for instance, "Baez at The Neighborhood Theatre, Charlotte, North Carolina, wearing orange sleeveless top", in conjunction with a caption that reads "Baez performing in 2003". First of all the top doesn't look orange to me, and why is the colour of her top significant anyway? Secondly the alt text ought not to give information unavailable to those looking at the image and reading the caption, such as that the picture was taken at The Neighbourhood Theatre, and thirdly what concept would a blind person have of "orange" anyway?

Fatuorum

05:01, 26 August 2012 (UTC)

Ok, in this version the top image venue info is from the image's page on Commons, trusting the uploader. No, it's not verifiable from the photo, so that's bad - agreed, good point, and removed. I stated a color ("orange" due to the viewing conditions, see below) because alt text has one function (describe image if missing, without redundancy with the caption), but four or five masters: the blind, blind but previously sighted, partially sighted, colorblind (maybe) and sighted (who may have images off, or a balky image server). Who wants color descriptors, and who doesn't?
Colors as colors have meaning to the previously sighted, partially sighted, colorblind (mostly) and sighted ("PSPSS"), and it's a color image.
Colors as words have meaning as adjectives, as classifiers: (red light, blue moon).
"Orange"The jacket falls in the "yellow, orange, red, brown" earthy color band, and Joan performed a lot in that color band. Do blind people know or care about color, abstractly? Sorry, I don't know (gotta find out), but the "PSPSS" might. Given the two articles linked above [1] and [2](PDF) advocating descriptive text, I felt somewhat justified including color. The specific name of color can be adjusted to be less specific: reddish? Does that help? I've edited the Joan Baez top image alt text in response to this discussion. --Lexein (talk) 11:59, 26 August 2012 (UTC)

Totally blind people who have never seen have no understanding of colours and would not understand them. But there are plenty of totally blind people who used to have at least some sight, and IMO alt text should cater for them by describing the colour (where relevant) exactly as one would, for example, when describing the image over the phone to a sighted person. Graham87 14:24, 26 August 2012 (UTC)

In what sense is the colour (which doesn't look like orange to me anyway) significant in this specific instance?

Fatuorum

14:32, 26 August 2012 (UTC)

I already asked if reddish is better - if not, what color is? I already answered you sufficiently generally, on the basis of a class of photo, for a class of audience, for (I now realize) this class of article (BLP musician). I think we can decide on a class basis, rather than having to worry over individual cases. Color can be argued to have descriptive relevance for photos of this type based on: 1. it's a color photo (alt text can concisely describe, to replace a photo), 2. a class of viewer would see color in the color photo, and so might expect color to be described in alt text, 3. it's at a specific time in Baez' career, when her color palette tended a certain way, 4. she's a performer, so the whole package matters (and since the whole package is illustrated the photo), and 5. there is (above) documented preference for richer descriptive text.
I'm keenly aware that when the great bunches of accessibility project editors get here, the whole thing, and all my edits could be summarily reverted as not concise, and "nobody cares about clothing on singer/songwriters" and we'll be back to "Woman with microphone on pole." --Lexein (talk) 15:35, 26 August 2012 (UTC)

@User:Kaldari - can you send the disgruntled vision-impaired users our way to participate in this discussion? --Lexein (talk) 21:09, 24 August 2012 (UTC)

Learning from previous discussions

There have been lengthy discussion about the correct use of alt text in the past, and specifically alt text length. Please see the archive 4 of this talk page, and especially the sections

. Several advices were requested, notably from a Webaim professional if I remember correctly. Since then, this guidelines was significantly changed. We now encourage shorter alt text.

This past discussion harmed heavily our project at the time. I wish we could learn from it, and avoid another controversy. For now we do not have enough information and feedback to take a decision in favor of short alt text or long alt text. Pushing in favor either choice is not a good idea, what we need is a basis for our decision. This is specifically intended for Lexein: debates and points of views are welcome, but pushing your point of view in the articles is not a healthy thing for the project. I suggest we read these archived discussions, and request more input from screen readers and experts. I'll try to contact Webaim about this issue. Dodoïste (talk) 10:27, 1 September 2012 (UTC)

Reluctant side note: some discussions were less productive than others, but harm is a strong word. I do not doubt that the project will be retemplated upon consensus as prescriptive and properly founded on standards and usage preferences by the target audience(s). The sooner, the better, I think, given such photo initiatives as "Wiki Loves Monuments" (which has some fine alt text IMO), and

WP:BRD). IMHO, the best place to evaluate alt-text is in the target article, in context.
--Lexein (talk

) 08:50, 2 September 2012 (UTC)

I never doubted you are acting in good faith. And there is no damage here. Please don't be offended, as I do consider that you are a valuable participant in this discussion. It might just be an excessive enthusiasm, or something similar that is linked to good faith. I still believe that you are hastily changing alt text in several pages - while the guideline was not changed. Making a few testcases for the purpose of discussion and evaluation is good, but this might be a little to early. Well, I'm only thinking aloud here, it's not that much of a big deal. Don't worry Lexein, I like your involvement. :-) Dodoïste (talk) 17:53, 2 September 2012 (UTC)

I found the results of the surveys made by Webaim, in early 2009, late 2009, 2010, and 2012. Concerning image alt texts in early 2009, they reported that 59% of users prefers to get a long description of a decorative image, where 31% preferred no alt text. This was further analyzed in the late 2009 survey. The conclusion they came to was that writing alt text that pleases everyone is difficult, and that no great solution was yet found in this case. It was also suggested that the request from several screen reader for long descriptions of images may be due to a lack of descriptions in general. If all images had long and descriptive alt text, screen reader users might change their mind. My conclusion is that this is a complicated topic, and no hasty decisions should be made. Until we find a better approach, I would suggest to let editors choose themselves if they prefer a long or short description. Dodoïste (talk) 13:28, 1 September 2012 (UTC)

Nice work! Per the May 2012 study Problematic Areas chart, alt text is a top-three current problem. I think a metastudy, building on your note above, from all the papers cited in all

WP:ALT like ==Usage research==, or in the article Alt attribute (which needs work anyhow), or at least as a 'sticky' text block at the top of this Talk page, so it doesn't scroll off into the archive. Again, let me reassure everyone, meaning you, that I'm not out to damage anything. I think we tread between two disservices: oversimple alt text, and spamming the screenreader user. Though we've determined that the screenreader can skip. --Lexein (talk

) 08:50, 2 September 2012 (UTC)

Changes in thumbnail alt text

A patch has been submitted based on a review of accessibility that might benefit from your input. Please see https://bugzilla.wikimedia.org/show_bug.cgi?id=34750. —TheDJ (talk • contribs) 07:55, 3 July 2012 (UTC)

Yes. But instead of making uncoordinated comments at bugzilla, let's discuss it among ourselves first. Rather at Wikipedia_talk:WikiProject_Accessibility#Changes_in_thumbnail_alt_text, since it is not meant to affect the practices of editors and focus on improving the code. Dodoïste (talk) 18:42, 3 July 2012 (UTC)

The FireFox "Popup Alt Attribute" add-on is helpful when mousing over article images, for sighted users. --Lexein (talk) 07:03, 25 August 2012 (UTC)

• Category:Unclassified articles missing image alternative text. (~ 251 articles, 1 subcategory about the UK.) For some infoboxes, if the alt= parameter is missing, or is empty, the article is added to the above category. It is not known many infobox templates support this.
• {{Alternative text missing}}. Adds article to the above category, with optional parameter (see doc) --Lexein (talk) 12:44, 26 August 2012 (UTC)

Papers

• "Making Wikipedia editing easier for the blind". 2008. M.Claudia Buzzi, Marina Buzzi. IIT-National Research Council. DOI:10.1145/1463160.1463210
• "Is Wikipedia Usable for the Blind". 2008. Marina Buzzi (IIT), Barbara Leporini (ISTI). p. 15-22.

• "Wikipedia, the open encyclopaedia: is it really open to blind users?" (Conference paper). 2008. j. ACM. Barbara Leporini. DOI:10.1145/1368044.1368049 (Derived from parent work)

--Lexein (talk) 11:48, 27 August 2012 (UTC)

Discussion of articles with recently changed alt text, focused on the merits of the alt text, and the changes to it.

Template:Selected picture

Added longer alt text in examples. (Note: their "caption" is alt, their "text" is caption). Comments? --Lexein (talk) 16:25, 26 August 2012 (UTC)

Mir

Pre-existing example of very long alt texts. In a table of space station component modules, the alt text is mostly replicated in pairs line-drawings and photographs. Comments?--Lexein (talk) 04:16, 28 August 2012 (UTC)

Hmmm, perhaps the repetitive alt text could be replaced by something like "photograph of <station component>"? The alt text is long in places, but it does a very good job of describing some quite complex graphics. Graham87 15:16, 28 August 2012 (UTC)

Thanks for reviewing. Can alt text reference other alt text? "See alt text in image to left"? In a table, positional references do make sense, as opposed to flowing layouts, unless tables are reversed in some languages (doh!). --Lexein (talk) 02:24, 29 August 2012 (UTC)

"See alt text of previous image" would probably be better, as it's clear where the "previous image" is in this case. Graham87 07:48, 29 August 2012 (UTC)

I'll try it. --Lexein (talk) 13:47, 30 August 2012 (UTC)

The Pirate Bay

In

) 13:47, 30 August 2012 (UTC)

One Million Years B.C.

Added restrained alt text to the two images. --Lexein (talk) 13:14, 31 August 2012 (UTC)

At WT:Route diagram template#About disabling link, there is a discussion on what the alternative text for little icons (such as , ) should be, or if they should be there at all. The discussion arose after we decided to remove a hard-to-manage automatic alternative text template. Please take time to give your comment there, thanks. – PeterCX&Talk 12:12, 28 March 2013 (UTC)

This page fails the accessibility check using the Wave tool provided by WebAIM.- see http://wave.webaim.org/report#/http://en.wikipedia.org/wiki/Wikipedia:ALT.

Several of the errors seem to relate to missing Alt text on the little "Enlarge" icon at the bottom right of each image. Can something be done about this? rossb (talk) 11:25, 25 June 2013 (UTC)

The "enlarge" text is part of the image title, which is read out by screen readers. Alt text is not a high priority there, but I suppose it wouldn't hurt. Graham87 14:59, 25 June 2013 (UTC)

What kind of alternative text should be provided with a logo? A section discussing this rule should be added to the guideline. - Rahat | Message 13:52, 19 October 2013 (UTC)

It should describe the logo, especially if the article doesn't already do that. By the way, new sections should always go at the end of talk pages; I've moved your comment. Graham87 13:57, 19 October 2013 (UTC)

Per

12:40, 10 September 2013 (UTC)

It would make the alt text sound different with a

speech synthesizer because it lowers its pitch at the end of a sentence like people do when they speak. Graham87

06:15, 11 September 2013 (UTC)

I'm asking because of this edit. Should we therefore add a notice in this regard to

my talk

06:19, 11 September 2013 (UTC)

Oops, I didn't really read your initial post properly. The above-linked edit is fine, and alt text should be punctuated in exactly the same way as captions. Graham87 15:27, 11 September 2013 (UTC)

Thank you. --

my talk

18:13, 12 November 2013 (UTC)

Something that came up at Taylor Swift discography [3]. The Alt text was "A blond-haired Caucasian woman smiles." This is of course ridiculous and serves no purpose at all. How about a change that says when showing a picture of someone the article is about, mention their name, don't just give vague descriptions of their race and hair color. Although I don't see how having any alt text at all in cases like this would help blind people. Dream Focus 02:19, 1 February 2014 (UTC)

It might be useful to blind people who have seen before to know that her hair is blond ... but that detail is irrelevant to the Taylor Swift discography article. You're right, there probably can't be any sensible alt text there. Graham87 05:54, 1 February 2014 (UTC)

Hello, I just came upon this concept of providing alt descriptions for photos. I've attempted to do it for the article, Phoenix, Arizona, but am unsure I did it correctly. Could someone who is more conversant in this check that article, and let me know any corrections which might be needed? Thanks. Onel5969 (talk) 17:45, 26 February 2014 (UTC)

At a quick glance (I really don't have much time at the moment), they sound good except that there's no need to write things like "A photograph of ...". Graham87 00:33, 27 February 2014 (UTC)

I've noticed that when I move my cursor over an image that's part of an infobox that has "alt=|" in the "image=" portion of the infobox, the "alt" text displays in my browser. However, for images that just use the [[File:...]] method, even if the "alt" text is provided, it doesn't display when I move my cursor over the image. Any idea why it works when it's included in a template, but not in text? For examples, see

17:28, 4 May 2014 (UTC)

@D'Ranged 1: the image in the infobox is generated by Module:InfoboxImage which adds the alt text in two places, (1) in the alt= and (2) in the 'caption/title' field. the equivalent standard image syntax would be something like [[File:Bad Title Example.png|220px|alt=some alt text|some alt text]]. note that the last chunk of text would be for the image caption if 'thumb' is used, and the caption/title text is not used for the hover text. hope this makes sense. Frietjes (talk) 15:45, 5 May 2014 (UTC)

Specifically, what shows tooltip text is the infobox code, with the "alt=" parameter added: {{Infobox civilian attack | title=Virginia Tech massacre | image= Virginia Tech massacre candlelight vigil Burruss.jpg | alt=A crowd of people holding candles. | caption= The candlelight vigil the night of April 17, <br>taken in front of Burruss Hall.| (etc.) }}

What doesn't show tooltip text is the standard image syntax: [[File:Norris WestAJ Map scale.jpg|thumb|alt=Building locations depicted in red.|Aerial photo showing location of [[Campus of Virginia Tech#Norris Hall|Norris]] and [[Campus of Virginia Tech#Ambler Johnston Hall|West Ambler Johnston]] Halls.]]

As another example of it not working, the first photo in this article displays no tooltip when I move my cursor over the photo: [[File:Jacques-Louis David 017.jpg |thumb |160px |alt=Painting of Napoleon Bonaparte in His Study at the Tuileries |''[[The Emperor Napoleon in His Study at the Tuileries]]'' by [[Jacques-Louis David]]]]

In this article, the "<imagemap>" example in "Unusual examples of alternative text" shows the names "William J. Bryan" and "Arthur Sewall" when I hover over those parts of the image, and "1896 Democrats Website" when I hover over the center of the image. This is documented in the explanation for that image, but no reason for the display of "1896 Democrats Website", which is part of the url given in the image map, rather than the alt text, which is "1896 Democratic campaign poster".

Likewise, another example in that section, [[File:Commons-logo.svg|frameless |upright=0.23 |border |center |link=Commons:Special:Search |Search Wikimedia Commons]] results in tooltip text of the caption, "Search Wikimedia Commons".

In the Wikipedia:Extended image syntax article, the example under How to display an image with no caption or formatting, [[File:Westminstpalace.jpg|Alt text]], renders the tooltip, "Alt text". In the New syntax for images section, in the examples of images rendering inline, [[File:Westminstpalace.jpg|150px|alt=A large clock tower and other buildings line a great river.|The Palace of Westminster]] gives the caption as the tooltip, "The Palace of Westminster", rather than the alt text. The same is true for the other example, [[File:tst.png|100px|alt=Tiny globe|This is a globe.]], which gives "This is a globe." as the tooltip, rather than "Tiny globe". In the Overlaying annotations on an image section, the tooltip for the image displays the caption even if I add "alt=" to the template used. In the Compatibility considerations section, the caption, rather than the alt tet, is again displayed as the tooltip. {| align="right"| [[File:Westminstpalace.jpg|160px|alt=A large clock tower and other buildings line a great river.|Example]]|}

These results are consistent between the Google Chrome (Version 34.0.1847.131 m) and Firefox (Version 29.0) browsers. There seems to be some inconsistency in how tooltips operate between templates and various forms of the image syntax; I don't know where to point this out to the developers, if that's what's needed. Thanks for any help you can give!—D'Ranged 1 ^talk 19:40, 5 May 2014 (UTC)

@Frietjes: I've just noticed the example images you added. In the case of [[File:Bad Title Example.png|frameless|100px|right|alt=this is the alt text|this is caption/title text]], the tooltip displays the caption, "this is caption/title text", rather than the alt text; while in the case of [[File:Bad Title Example.png|thumb|right|100px|alt=this is the alt text|this is the caption/title text]], the tooltip shows nothing at all. There seems to be some problem with the use of "thumb", perhaps? And "frameless" ignores the alt text for the tooltip, but shows the caption? There are too many variables for me to figure out which variable is causing what text to display; if I remove the parameter "frameless", the caption text still displays, rather than the alt text. I've also experimented with no other parameters other than "alt" and the caption (e.g., [[File:Bad Title Example.png|alt=Alt text|Caption text]]), and the caption displays as the tooltip instead of the alt text.—D'Ranged 1 ^talk 20:06, 5 May 2014 (UTC)

@D'Ranged 1: the image in the infobox is generated by Module:InfoboxImage which adds the alt text in two places, (1) in the alt= and (2) in the 'caption/title' field. when the 'thumb' option is used, the caption/title text is used for the caption. when 'thumb' is not used, then this text is used for the tooltip. Frietjes (talk) 20:19, 5 May 2014 (UTC)

I've always struggled with writing alt text, and most of the time I find myself either leaving it blank or just writing "refer to caption." I understand the examples used on this page, but I have a hard time putting it into practice. Could someone please help me? What would be appropriate alt text for

) 10:56, 9 October 2014 (UTC)

Redundancy

"The alt text is read out by screen readers just before the caption, so try to avoid having the same details in both." Exactly that mistake is demonstrated in the Napoleon example ;-) --Mosmas (talk) 08:43, 19 October 2014 (UTC)

We have the following sentence in the section

:

• Where the caption does not describe or identify the image, but the image is merely decorative, the alt text should contain the shortest possible description or identification of the image.

I can see no place at all for captions on merely decorative images - and [H67] tells us that decorative images should have null alt text (|alt=""). I'm going to remove that sentence. If there is some authoritative guidance on dealing with captions on merely decorative images, then it would be worth mentioning, but I suggest that the existing guidance in the

) 13:27, 20 April 2014 (UTC)

User:RexxS, I was thinking about decorative images and came looking for people's thoughts. I suspect that what this sentence meant was, if you're going to put up a picture of a butterfly as eye candy, with a caption that is irrelevant (perhaps "People feel happier when they look at pretty objects" or "Growth can transform you"), then it might be appropriate to describe the image. However, Wikipedia shouldn't have such images in encyclopedia articles. So I think it was a fine suggestion, just one that wasn't very relevant to this particular website. Our version would read something like Where the caption does not describe or identify the image, but the image is merely decorative, then remove that image!

Here's my question: I wonder whether using CSS to mark decorative images as metadata would also make unimportant decorations 'disappear' (<span class="metadata">[[File:Foo.jpg]]</span>). This might help deal with the case of an image that needs to be linked (because it's not public domain) but has no real value to people using a screen reader. What do you think? WhatamIdoing (talk) 05:35, 4 August 2014 (UTC)

@WhatamIdoing: I'm fairly sure the original idea of a 'decorative image' dates back to when we used to use fancy (foreground) images to create borders around webpages by using tables as the means of layout. Those sort of images clearly shouldn't have alt text. In that sense, the advice not to have alt text wasn't intended to apply to the picture of the butterfly, because as soon as you get a caption, the screen reader will want to know what it refers to. In an ideal Wikipedia, we wouldn't have eye candy; but in the version we have, we have to pragmatically accept that editors will want to add images to their articles, even if the pictures don't replace the thousand words that they should be worth.

With that in mind, I think your idea of using css is good in principle. I would want to reserve the .metadata class for other things, but we could have a definition like this:

@media aural {
  .decorative {
    display: none;
  }
}

That would allow things like

• <span class="decorative">[[File:PrettyPicture.jpg|People feel happier when they look at pretty objects.]]</span>

which should be ignored by screen readers. Perhaps we should still set |alt= so that any screen reader that didn't recognise the "aural" media type would still be able to deal with the alt text. How does that sound? --RexxS (talk) 15:28, 4 August 2014 (UTC)

My CSS-fu is so weak as to be nonexistent. The "metadata" class was recommended as a way to get pointless images skipped by Media Viewer's slideshow feature (like the letter-i-in-a-blue-dot image that deserves neither link nor alt text in the examples on this page). I don't know what other options are available or what else Media Viewer might skip (which could always be changed, if we created a new one).

My general bias is to go with something that will be fairly widely recognized and understood, because making a change that helps almost no one wouldn't be very helpful. Of course, my minimum requirement is something that works reliably for User:Graham87, so perhaps he'll tell us what he thinks. WhatamIdoing (talk) 22:13, 4 August 2014 (UTC)

Nope, it doesn't work; I tested it at User:Graham87/sandbox21 and added the above code to my monobook.css. JAWS generally isn't very good with the aural media type. Graham87 02:03, 5 August 2014 (UTC)

I think I remember you telling me about JAWS and the aural media type some time ago, Graham, so I'm not surprised. Oh well, perhaps it's something to keep in mind for the future and thank you anyway. Back to the drawing board ... --RexxS (talk) 14:26, 5 August 2014 (UTC)

Ah, I see you said, "I can see no place at all for captions on merely decorative images - and [H67] tells us that decorative images should have null alt text (|alt="")." I just edited the page to make this more clear. All images should have alt text. Decorative images should have alt="" so that a screen reader never reads the filename. Banaticus (talk) 19:08, 3 November 2014 (UTC)

I've reverted you there. The code alt="" is equivalent to no alt text, as far as screen readers are concerned. Graham87 04:59, 4 November 2014 (UTC)

Banaticus : RexxS mentioned that decorative images should have null alt text AND no link, if possible. Because W3C guidelines tells us that an image inside a link cannot have an empty alt text, otherwise the filename is read aloud F89: Failure due to using null alt on an image where the image is the only content in a link). On Wikipedia, the link is often required to respect the creative commons licence and to credit authors. It's complicated, and that's the reason why the guideline might seem confusing. Dodoïste (talk) 23:09, 4 November 2014 (UTC)

Graham, alt="" is not the same as no alt text. Dodoïste, I see where the confusion came from. I left the sentence regarding linked images and put back in the following paragraph which refers solely to images used aesthetically, i.e. for padding, rounded corners, etc. Yes, I know those things can and likely should be done with CSS, but some web designers just slice images in Photoshop and drop them in. Linked images serve a navigation purpose -- the images discussed in the last paragraph of the header (my most recent edit at this time) are completely nonfunctional and thus should have alt="" so that a screen reader ignores them. Banaticus (talk) 03:45, 5 November 2014 (UTC)

I've clarified the text to say that it only applies to images that can be unlinked. By the way, in wiki-markup, alt="" produces the alt text of two quotation marks, as shown at [[User:Graham87/sandbox22]. Graham87 08:46, 5 November 2014 (UTC)

Banaticus: There is another confusion. In HTML syntax, the code alt="" is an empty alt text - I mean no alt text at all. In wiki syntax, alt="" produces the following alt text : "". Which reads 'Quote quote'. It is most unhelpful.

Banaticus, I believe you are mistaken on several other points. the article written by Jim Thatcher is not a reference than can apply to Wikipedia. Wikipedians are not web designers. They do not produce round padding images for decoration, they do not use photoshop, they write articles.

WP:ALT

is a reference for wikipedians who want to write an article.

I see you have good intentions and you want to help. That's good. But you are new here, so please don't edit this guideline without gaining consensus. And please learn more about what we are doing here and how. I can provide much more explanations if you are interested. Cheers, Dodoïste (talk) 11:14, 6 November 2014 (UTC)

Dodoïste , yes I know that alt="" produces the alt text of "". That is not "quote quote", however, that is how you tell a screen reader to ignore the decorative image. The page currently says that purely decorative images should have no alt text, and that's not correct because a screen reader that sees an image without an alt attribute will read the filename of the image. All images should have alt text and if the image serves no purpose, then it should have alt text of "". Look, you've been around on Wikipedia long enough to be able to recognize that I put in text based on a verifiable quote from an authority on the subject. That quote says alt="" is what should be used for purely decorative images. Unless you can find another good source that says otherwise, then you'll just have to go with that reputable quote, because (and I know you already know this) that's how Wikipedia works. On a completely unrelated note, why would you think I'm new? ;) Banaticus (talk) 08:15, 8 November 2014 (UTC)

Banaticus: Sorry for the harsh tone earlier, it did not help. I do not like such edit warring and disputes. I thank you for your calm response. :-)

I know you have been a volunteer on Wikipedia for several years. I meant that you are new on the project and guidelines related to accessibility. And it surely seems to me that you are also new in the field of web accessibility in general. There are several people here who are experienced in the field. RexxS is a professional web designer, Graham87 is a blind user with a lot of experience.

Hopefully I have more time to explain myself today. We have a very different understanding of the advice written by Jim Thatcher. Don't you think he could have meant to say: "[In HTML], use alt="" on images that don't carry information" ? Because I do, and I'm certain that others here have the same understanding as me. Jim Thatcher did not write this article for wikipedians who use wikisyntax, he wrote it for web designers who use HTML. The result is very different.

Usually, we do not use the website of a professional as a source of information. We use the official

WCAG 2.0

published by the Web Accessibility Initiative (WAI) of the World Wide Web Consortium (W3C). It is an international standard, used by accessibility expert around the world.

I can not find a source that says otherwise, because it is true that "alt="" is what should be used for purely decorative images". But it's only true when you write

Wiki markup

. That's what I've been trying to explain.

There are other things I would like to explain. But first we have to agree on the understanding of alt="". I'm going to ping User:RexxS, maybe he can explain it better than me. I'm not a native english speaker after all. Cheers, Dodoïste (talk) 10:59, 8 November 2014 (UTC)

Posted at

17:02, 30 August 2015 (UTC)

Altviewer tool down

It looks like Dispenser (talk · contribs)'s Altviewer tool is offline. Not sure if this is a temporary or permanent development, but is there an alternative in place that users can be directed to?-- Patrick, _oѺ^∞ 14:11, 27 October 2015 (UTC)

It's at http://dispenser.homenet.org/~dispenser/view/Altviewer. I've reopened T116757 to get the old URLs to redirect. — Dispenser 13:40, 30 October 2015 (UTC)

Thanks!-- Patrick, _oѺ^∞ 16:50, 8 November 2015 (UTC)

How does one apply this guidance to film posters? As art pieces they generally attempt to pack a lot of visual imagery which requires rather lengthy explanation to provide full context. For example, the poster for

15:07, 28 November 2015 (UTC)

Does

WP:CAPFRAG

apply to alt texts?

All the

17:10, 24 May 2016 (UTC)

Alt text not appearing on mouseover?

Hi, for some reason alt text on Wikipedia isn't working when I mouse over images when they appear in the body of the text (including the ones on this help page). Mouse over seems to work fine for infobox images though--is this a problem with my browser, or is it a known issue on Wikipedia? Thanks, Rachel Helps (BYU) (talk) 16:15, 6 June 2018 (UTC)

A comment at the top of this old FAC prompted to look at WP:ALT, and I see the page does not agree with the advice given there, which is that "photograph" or "picture" is often the right alt text given that a screen reader will follow that with the caption. This seems more sensible than the current recommendation to provide a description. For example, the picture of Napoleon on WP:ALT has alt text of "Painting of Napoleon Bonaparte in His Study at the Tuileries" and a caption of "The Emperor Napoleon in His Study at the Tuileries by Jacques-Louis David". As I understand it, a screenreader would read out:

Painting of Napoleon Bonaparte in His Study at the Tuileries The Emperor Napoleon in His Study at the Tuileries by Jacques-Louis David

which is more verbose and no more helpful than

painting The Emperor Napoleon in His Study at the Tuileries by Jacques-Louis David

Am I missing something? Mike Christie (talk - contribs - library) 12:44, 21 October 2018 (UTC)

Thanks, I've tweaked it per your suggestion. The page does say that details shouldn't be repeated in both the alt text and the caption. Graham87 01:22, 22 October 2018 (UTC)

Great, glad I hadn't misunderstood. The basics section still seems to give advice that's contrary to this approach, though. I'd like to suggest changing

Alternative text should be short, such as "A basketball player" or "Tony Blair shakes hands with George W. Bush". If it needs to be longer, the important details should appear in the first few words, allowing the user of a screen reader to skip forward once the key points are understood.

to

Alternative text should be written with the understanding that a screenreader will follow it with the caption, so if the caption contains all the useful information that a reader might need, the alt text should be simply "photograph", "painting", "diagram", or something similar. If there is no caption, or the caption omits useful information, use concise alt text such as "A basketball player" or "Tony Blair shakes hands with George W. Bush". If it needs to be longer, the important details should appear in the first few words, allowing the user of a screen reader to skip forward once the key points are understood.

-- Mike Christie (talk - contribs - library) 01:48, 22 October 2018 (UTC)

That edit needs more refinement. I've reverted for now under the impression that it will indeed be refined. --Izno (talk) 13:03, 22 October 2018 (UTC)

@Mike Christie and Izno: Sounds good to me. In hindsight, my edit might've been a bit premature. Graham87 14:41, 22 October 2018 (UTC)

OK, here's an attempt to outline a coherent set of edits that address this in a couple of places on the page.

1. Change the example markup in the lead to change the alt text to "Painting".
2. At the end of the paragraph describing how a screen reader will read out the filename in the absence of alt text, add the following sentence: "The alt text provided in the example, 'painting', avoids this problem, and is all that is needed because the caption provides sufficient information for a screen reader".
3. In the "Basics" subsection, I suggest repeating the definition of "alternative text", since for those who only occasionally refer to this page it is easy to get confused between alt text and alternative text. I suggest starting that paragraph with "Alternative text (that is, the alt text followed by the caption) should be short..."

-- Mike Christie (talk - contribs - library) 22:17, 22 October 2018 (UTC)

I'd like to check to see what is being recommended on the WWW right now regarding the alt attribute, given the presence of the HTML 5 <figcaption>, which will probably someday be used to mark our captions up. EEng gets into some of the confusion in the next thread down. --Izno (talk) 00:11, 23 October 2018 (UTC)

Getting into confusion is my natural state of being. EEng 00:19, 23 October 2018 (UTC)

I'm confused

I'm completely confused. At Wikipedia:Manual_of_Style/Images#How_to_place_an_image we're told that the dog image should have

alt=A white dog in a harness playfully nuzzles a young boy.

But here on this page, we're told that

Unless it appears in an article on fashion, the alt text should not be "an elderly woman wearing a black hat"

and the image itself carries

alt=Elizabeth II speaking to the public.

It seems to me that the dog image is done correctly: the alt describes what a sighted reader would see, but a non-sighted cannot see i.e. alt = A white dog in a harness playfully nuzzles a young boy. It also seems to me that for the QE image, the alt should be (and here I'm elaborating it a bit) alt=An elderly woman in a green dress and black fur hat smiles at a small group of people and the caption (not alt) should be Elizabeth II speaking to the public. Someone enlighten me. EEng 22:11, 22 October 2018 (UTC)

You're correct re the first image. Re the second one, we only need to describe it in such detail if it's relevant to the article. Graham87 01:44, 23 October 2018 (UTC)

You're probably right, though as you say it will depend on context and purpose of the image, and of course judgment will be involved. OK, but then what about the Napoleon example currently on the page? It recommends

[[File:Jacques-Louis David 017.jpg |thumb |alt=Painting of Napoleon Bonaparte in His Study at the Tuileries |''[[The Emperor Napoleon in His Study at the Tuileries]]'' by [[Jacques-Louis David]]]]

This to me makes no sense. How does alt = Painting of Napoleon Bonaparte in His Study at the Tuileries help a non-sighted reader, or indeed anyone? First of all, it's just a repeat of a subset of the caption. Second, even if was not a repeat of the caption, it tells the reader nothing unless he either (a) is familiar with that particular painting or (b) has at least some idea what Napoleon looked like. The whole point of showing the painting is the details of how the subject is presented, so it seems to me the alt should be something like alt = A neo-classical portrait of Napoleon, showing him as a dark-haired man in his early 40s wearing the uniform of a French admiral, his hand with etc etc. (I made all that stuff up) -- exactly what you'd say is a complicated question, especially since there's so much you could say about this particular painting, it has its own article, various elements have symbolic or political significance, different readers (non-sighted readers, remember) will have different levels of familiarity with Napoleon and his time, and much more. So it's not an easy example for this discussion. But surely the alt recommended on the page now is completely senseless. EEng 03:10, 23 October 2018 (UTC)

Yeah it's not the best as it is right now. I'm not sure what to do with it ... but I agree it could cause confuzzlement. Graham87 14:43, 23 October 2018 (UTC)

Follow up on this discussion

Izno, EEng, Graham87, any more thoughts on this? The advice in the lead still seems quite wrong to me. Why would we leave in place a recommendation for alt text that would lead to a screen reader repeating itself? I take Izno's point about making sure the caption is useful before cutting the alt text to just "painting", but in this case it is useful, and we shouldn't be recommending what we currently show in the lead. Mike Christie (talk - contribs - library) 12:12, 31 January 2019 (UTC)

Well until someone can answer my query about the top-level purpose of alts, I don't know what to do. EEng 21:03, 31 January 2019 (UTC)

At least we could remove the obviously bad example. If nobody objects in the next day or so, I'll have another go at this; I'll reinstate Graham87's edit but add a caveat to address Izno's comment. Mike Christie (talk - contribs - library) 10:46, 1 February 2019 (UTC)

I think we should read and absorb the WHATWG living developer's specification regarding alt, which I find fairly readable on the point of alt texts. We should echo or mirror the specification, given that it's been written with the other HTML elements in mind (specifically figure/figcaption, which we don't use today but which we will use at some point in the future). --Izno (talk) 14:15, 1 February 2019 (UTC)

Well, if there's any hope of the intent of that spec being followed by editors all over the project, someone's gonna have to do a lot of work to distill it into a far more usable form. Also, I find some of its pronouncements problematic. At one point it says

One way to think of alternative text is to think about how you would read the page containing the image to someone over the phone, without mentioning that there is an image present. Whatever you say instead of the image is typically a good start for writing the alternative text.

but then elsewhere it says

If the picture is just a face, there might be no value in describing it. It's of no interest to the reader whether the individual has red hair or blond hair, whether the individual has white skin or black skin, whether the individual has one eye or two eyes.

But see, if I were reading an article to someone over the phone, I would, in fact, say, "And here the article gives a picture of Smith lecturing in a large crowded auditorium. He's a large, imposing man in his 50s with graying hair, and he's smiling as a student asks a question. His tweed coat has smudges of chalk here and there." If he only has one eye, I'd certainly say that. And in one case the page recommends this as an alt:

A high forehead, cheerful disposition, and dark hair round out Alex Salmond's face.

...which is somewhat along the lines of what I described just a second ago, but contradicts what I quoted earlier. So I find the spec confusing. EEng 22:19, 1 February 2019 (UTC)

EEng, just reading your comment above about how you'd describe a picture over the phone. It seems to me that your description is how you'd describe it to someone who can see a picture, but just can't see that particular picture at that time. How would you describe it, though, if you were talking to a person who had been blind from birth, and thus wouldn't know what a person in their 50s looks like as opposed to a person in their 20s, wouldn't know what a color looks like, would have never seen anyone smile, and would have never seen a smudge? I think that's how to think of an alt. I'd be very curious about your answer to that particular question (but will otherwise discuss the broader issue in the new thread below). For my part, I think I'd just say, "And here the article gives a picture of Smith lecturing in a large crowded auditorium." I might add something like, "The crowd is listening attentively." Leviv ich 23:03, 19 February 2019 (UTC)

This engages some complex issues that border on the philosophical. As you say, some blind readers will be late-blinded, so will know what color and so on is, and (and here I conjecture) even those blind from birth have some understanding that objects have these surface properties called brightness and color which are perceptible from a distance, that color arrangements may be perceived as pleasing or unpleasant or may have cultural significance, and so on. More tangibly, a blind person will certainly know from tactile experience that old faces are lined and young ones smooth, so if I describe "an elderly man with a deeply lined face" they'll get something from that. In sum, it seems to me that a comprehensive description (ideally in a kind of pyramid form, working from overview down to little details) serves the spectrum of all readers. EEng 00:19, 20 February 2019 (UTC)

The context for this comes from a BLP/N discussion on Ben Shapiro but the BLP factors don't matter, just the question that if we have an infobox picture of a person on their BLP article, with a known timestamp (eg like a photo taken in 2016), what are the appropriate alt and captions should be for that? It essentially comes down to the caption, some feeling that having the name in the caption ("X in 2016") duplicates the name at the top of the infobox, but then the question of how screenreaders "see" this came up. What generally should be preferred here for MOS and accessibility? --Masem (t) 21:48, 19 February 2019 (UTC)

As a screen reader user, I instinctively strongly prefer ""Shapiro in 2016" but I can't really articulate *why*, except that we've always done it that way and anything else sounds really weird to me. Not really an accessibility-related justification, but there you have it. Graham87 01:46, 20 February 2019 (UTC)

Also, screen readers tend to put captions and alt text on separate lines. Perhaps my revulsion just stems from the fact that it's weird to have a line in that context start with a preposition. Graham87 01:51, 20 February 2019 (UTC)

• The alt text should somehow pertain to why the image is being used in the article. The suggested alt makes about zero sense. It is not an image being used because it is a middle aged man (which might make sense on Middle age) or because he is using hand gestures (which might make sense on List of gestures). You may as well have the alt say A mammal on the planet Earth. They're both technically true, but equally irrelevant. The article is on an individual person, and the image has been chosen because it depicts that person, and because that person is doing something they are well known for, such as public speaking engagements. GMG^talk 13:58, 24 February 2019 (UTC)

I'm probably late to this particular train, but I just discovered the capability to add alternative text for images. I attempted to do so for the images in

19:13, 12 March 2019 (UTC)

ONUnicorn, it works fine for me. I have the Navigation Popups gadget enabled from preferences, which shows alt= for images (and doesn't show them if they're not coded correctly–that's how I check my alts usually), and all the alts on that article show fine (the lead image and the gallery). I checked all the images with a screenreader (VoiceOver) and it read the alts just fine, too. Leviv ich 20:41, 12 March 2019 (UTC)

Thank you. I'm glad it worked. Now that I know this exists I'll start using it!~

problem solving

03:04, 13 March 2019 (UTC)

Alt text for mission insignias

I made an edit to a couple of astronauts articles (1), adding alt text to the mission logos that are in the infobox. Should I have alt=[Mission] logo, alt=[Mission], or alt=""? It seems decorative to me since the mission names are just above it. I suppose I do not know how the infobox is navigated with software anyways. Thoughts? Kees08 (Talk) 22:50, 29 May 2019 (UTC)

I think the way you added them (i.e. "Gemini 8 logo" and "Apollo 11 logo" is probably best, especially because there are two logos in this case. Graham87 02:53, 30 May 2019 (UTC)

Alt text for astronaut group articles

Taking NASA Astronaut Group 4 as an example, there is typically a group photo and individual portraits of each astronaut.

For the group photo I was thinking maybe "Official group portrait" or "Astronaut group 4 portrait", but that runs a bit afoul of "All readers will be aware this element is an image, so adding "photograph of" isn't usually necessary." So maybe just "Astronaut group 4"?

For the individual portraits, I suppose "Firstname Lastname" would be sufficient, but that seems stupid since their name will be read very shortly afterwords. Is there anything logical I can do about that? Appreciate any feedback, I would like to apply this to a few articles so thought it would be worth asking ahead of time. Kees08 (Talk) 01:29, 16 June 2019 (UTC)

Noting that it's a portrait is fine because not everybody will know that each image is a portrait. The alt text "Official group portrait" would be best for the group, and just "portrait" would be alright for the individual images. Graham87 05:54, 16 June 2019 (UTC)

Alt text for album artwork

Is this a good way to write alt text for album art? Diff Kees08 (Talk) 17:22, 3 August 2019 (UTC)

I don't think for either of those that the alt is too far off the mark, but they can be shorten. For example, for the base album cover: "A photograph of two men in business suits on a movie studio lot shaking hands. The right man is on fire. The right edge of the photo is singed." For the other cover: "A circular logo on a black background. The circle is divided into four multi-colored quadrants, each with unique abstract backgrounds, and an indistinct mechanical device crosses it from the bottom left to top right. The band's and the album name are in the top left quadrant." --Masem (t) 17:31, 3 August 2019 (UTC)

Grammar correction: Instead of singed I think you mean sang or sung EEng 20:03, 3 August 2019 (UTC)

Sorry for the confusion, I changed it to the much shorter "Album artwork". I think the previous way it was written would be fit for a longdesc attribute (which mediawiki does not support). Thoughts? Kees08 (Talk) 17:47, 3 August 2019 (UTC)

Alternative Text for Formulae

Hello! The function to add alternative text to a mathematical formula does not work as described. Supporting information: [4] and an old discussion here: [5] --Lpd-Lbr (talk) 09:27, 17 September 2019 (UTC)

TFA alt text

I recently had a discussion with a few main page curators, and decided I was going to try to consistently add alt text to TFA images. Many TFAs feature similar images (battleships, profile photos, animals), so the alt text should be pretty similar. I was hoping to propose a few for this month to see what you all think before I start adding them en masse to blurbs. November blurb list, which is where the photos for the below are located. I know some dates have passed, this is more an exercise in determining good alts for TFAs.

• Ships
  • HMS Royal Oak (November 17). Proposed alt: HMS Royal Oak at sea or HMS Royal Oak underway
  • T7 (November 23) Proposed alt: Torpedo boat T3 underway
• Coins
  • Second Fitna (November 7) Proposed alt: ??
  • Pilgrim Tercentenary half dollar (November 28) Proposed alt: Obverse of Pilgrim Tercentenary half dollar
• Planes
  • McDonnell Douglas F/A-18 Hornet in Australian service (November 2) Proposed alt: Three F/A-18A Hornets in formation
  • No. 33 Squadron RAAF (November 15) Proposed alt: Parked KC-30A tanker aircraft

Pinging Graham87 due to past useful input. Kees08 (Talk) 23:50, 3 November 2019 (UTC)

@Kees08: I don't think this would usually be necessary. In the case of Main Page images, the alt text is automatically made exactly the same as the image title, so where the latter is descriptive enough, there's no need for additional alt text. If the alt text provided is different from the caption, the screen reader will typically read the alt text and not the caption. So at Wikipedia:Today's featured article/November 17, 2019, for example,there's no need to add additional alt text at all. Graham87 03:33, 4 November 2019 (UTC)

Alt text for images that incorporate text

At John Brown's raid on Harpers Ferry#Other casualties, civilian and military there is an image incorporating both a drawing and a caption (from the original source) that describes it. I've transcribed the entire caption into the alt text, although I understand alt text is supposed to be short. Is this an appropriate approach, or would it be better to crop the caption out and render it as text in the main caption? Hairy Dude (talk) 04:22, 30 May 2020 (UTC)

I for one think it's a good approach; I don't have any problem with the length in this case. Graham87 16:05, 30 May 2020 (UTC)

Gravestone with inscription: "Martin Cherry, aged 58 years, died 28th June 1880 at the siege of Glenrowan"

From an accessibility view, cropping the original caption and reproducing it in our image caption is an improvement (perhaps along with |alt=See caption) over no alt text. But transcribing the original caption into alt text is also fine.

Nevertheless, in this case, I think the current caption is UNDUE detail for John Brown's raid on Harpers Ferry. I'd prefer the cropped image with the original caption transcription as the image caption in that article.

On the other hand, if you look at the Heyward Shepherd monument article, the extra information about the image (the current caption) is much more likely to be DUE, so the uncropped image with full alt text and extra information in the caption seems to be right to me. Hope that makes sense.

In general, it's good to include even lengthy transcribed text as alt text when it's important to understanding the image. For example, if you used File:Abigail Overman Pike Headstone.png, you would want to include the inscription in alt text, but not in the caption (because it's already obvious to any sighted reader). Whereas File:Benalla cemetery Cherry.JPG is unlikely to be legible at normal thumbnail size, so putting the transcription into our image caption is more useful.

If you think there's something worth adding to our guidelines, please feel free to go ahead. --RexxS (talk) 16:57, 30 May 2020 (UTC)

Templates

I am adding an image with text in French and thought that I could add the text as alt text. Can I use {{lang}} to mark the text as French? What about other templates? --Error (talk) 22:57, 13 September 2020 (UTC)

@

tl}} did though ... perhaps it's because the lang template tries to override the title HTML tag. Graham87

08:34, 14 September 2020 (UTC)

It sort-of works in a way. I didn't think of reading the source. In this case:

<img alt="French: Dubo, Dubon, Dubonnet. Vin tonique au quinquina." src="//upload.wikimedia.org/wikipedia/commons/thumb/5/51/R%C3%A9clame_Dubonnet.jpg/220px-R%C3%A9clame_Dubonnet.jpg" decoding="async" class="thumbimage" srcset="//upload.wikimedia.org/wikipedia/commons/thumb/5/51/R%C3%A9clame_Dubonnet.jpg/330px-R%C3%A9clame_Dubonnet.jpg 1.5x, //upload.wikimedia.org/wikipedia/commons/thumb/5/51/R%C3%A9clame_Dubonnet.jpg/440px-R%C3%A9clame_Dubonnet.jpg 2x" data-file-width="2304" data-file-height="3072" style="width: 220px; height: 293px;" width="220" height="293">

It does not seem to mark it as French for screen readers. I don't know if HTML allows that.

--Error (talk) 20:05, 14 September 2020 (UTC)

The alt text attribute cannot have internal HTML, only alphanumeric characters (and dashes etc.). --Izno (talk) 20:58, 14 September 2020 (UTC)

Deprecating CC-BY-SA?

When a blank alt attribute is desired, consider replacing CC-BY-SA images with public domain equivalents. I think it is highly problematic to deprecate CC-BY-SA in any way. This could easily lead to lower quality images being used in articles when a perfectly acceptable CC image is available. If the public domain image is an improvement, it can and should be replaced anyway without any reference needed to accessibility. SpinningSpark 13:06, 11 January 2022 (UTC)

I'm not even sure where the idea that alt text should be used for CC-BY-SA attribution - that attribution is a function of the File: space page (if here on WP) or the image's page at Commons, not the alt text. Even if we are only talking decorative images, it seems to be something that would be unhelpful to the screen reader to suddenly have a "image by so-and-so" read to them. --Masem (t) 13:09, 11 January 2022 (UTC)

As I read it, having both |link= and |alt= blank usefully suppresses the screenreader from reading out the file name, but not so usefully, prevents clicking the image linking to the image page. Linking to the image page is a requirement for attribution of copyrighted images, therefore this must not be done on such images and consequently alt text must exist to suppress the file name being read. However, the text is opaque to say the least and I might have got that completely wrong. SpinningSpark 13:36, 11 January 2022 (UTC)

Indeed. If the best decorative image available is CC-by-SA, then use it, but use a generic alt text like, say, "Stub icon". If a PD image is the best fit, having no alt text/link is a good course of action. Graham87 14:38, 11 January 2022 (UTC)