Closed StephDriver closed 5 months ago
traversing from top to bottom of page.
<a itemprop="email" href="[redacted]" class="fa fa-envelope email-link"></a>
<a href="/issue/910/info/">Volume 10 • Issue 1 • 2023</a>
<title>Janeway-Logo-Final</title>
These are only notes problems not already described.
traversing from top of page to bottom
:x: User Content "Glossa al of general linguistics, link, image, navigation, landmark"
:x: #4226 home, link, list start - but nothing visible on the page and is a duplicate of the link above.
:x: #4226 Visually, at phone screen width there is a hamburger menu, but to the screen reader, no such thing exists, but instead the whole uncollapsed menu is read out despite being hidden. These menu items are not identified as a menu or as navigation.
:x: #4226
items in the menu (articles, submissions, journal policies, about) are read followed by "arrow underscore drop underscore down" and identified as links. <i class="material-icons right">arrow_drop_down</i>
✅ Search, link
✅ Login, link
✅ Register, link, list end
:x: #4226 "Menu, link, end, navigation, end, banner". This is the hamburger menu, and when activated the menu does pop up, but you can't navigate the list, as you already have - what pops up visually is the list that I've just described above, but it has already been navigated, when advancing one reaches the article title instead. Further, there is no way now to dismiss this menu without refreshing the whole page, activating the menu link again does nothing.
:x: #4232 Article title is read out correctly but identified as an image - so is the alt for the image, then when moving on, one gets "Research Article" followed by the title again (!) none of which is identified as a heading (should be H1).
:x: #4232 There is no heading level 1 at all.
✅ abstract, heading level 2
✅ paragraph text is fine, until...
:x: #4238
<i>ɛna</i>
which has been put in like that, with the symbol, and so voiceover reads it as 'na' unable to parse the epsilon.
✅ Keywords, heading level 2
✅ How to cite, heading level 2
:x: #4221 doi link should be contextual
✅ Download, heading level 2
✅ Download PDF, link
✅ Download XML, link
✅ known structural issues with metrics have just been addressed in https://github.com/BirkbeckCTP/janeway/issues/3912
:x: #4215
The 'eye' icon next to views has no description at all, found by voice over, but has no description to give. <i class="fa fa-eye"></i>
:x: #4215
likewise, the 'download' icon. <i class="fa fa-download"></i>
:x: #4232 One Introduction, heading level 2 - This puts the article start on the same heading level as the metadata above it. Structurally this doesn't make much sense. I can see this being on the same level as abstract, i.e. level 2, in which case there should be an H2 for meta under-which keywords, how to cite etc are H3, or There should be an H2 here for 'article' and the Introduction should be an H3. Noting this here, but going down the list I am now looking for consistency of headings with this introduction as it is, at H2.
:x: #4243 examples here are being shown visually indented and with just a number in brackets to differentiate them, no label that this is an example or when the example ends and normal paragraph begins. There should be some non-visual differentiation of these sections. Likewise references to these examples should be more than just numbers in brackets (eg. marked up as 'example 1' not just '(1)' in both the example itself and in references to that example).
x: #4245 When following the link to a reference (link goes to References section), there is no way to return to the original position.
:x: #4245 Endnote indices do not indicate that they are endnotes rather than just a number - contextual links needed.
:x: #4238 The symbols aren't read by voiceover, making, for example, this section unintelligible. This issue continues throughout the article.
:x: #4245 Missing internal cross-ref. "see discussion in §3 below" without any link.
:x: #4221 Link text of "2011" which is not descriptive, but makes sense visually with how referencing is done. Screen-reader users use link navigation as one way to traverse the page, so contextual, differentiated links are important as part of the user's concept of page layout and where they are within that page. Use Aria-label to give additional context to such links.
:x: #4243 Two Tone in Akan, heading level 2. The screenreader doesn't differentiate that the '2' at the start of this is section two. And with this particular heading there is clear potential for misunderstanding (2. Tone.. or Two Tone..) So I wonder whether the aria markup should add a prefix 'section' especially as in the cross referencing these are referred to as sections.
✅ Section 3 heading is consistent as an H2
✅ Section 3.1 heading is consistent as an H3
✅ Section 3.1.1 heading is consistent as an H4
:x: #4243 this formatting causes problems because the implementation of the small caps formatting is with span classes that break up the words, so voice over reads the initial letter on its own, and then the rest of the word.
:x: #4232 3.2.1.1 Focus and Linking is inconsistent at H4 (should be H5) - looks like the divs may be wrong too, this is not being coded as within the 3.2.1 div, but at the same level.
✅ 3.2.1.2 is consistent with other headings (i.e. not same error as 3.2.1.1) and shown as an H5
:x: #4243
using small caps for this doesn't change the tone of the voiceover, which is how markup such as strong and em are indicated audibly. So there is no audible signal that this 'fm' is different, the formatting is purely visual. <span class="small-caps">fm</span>
✅ of the [±D-linked] feature
the ± symbol here is correctly being read out, and the [ ] ensure a change of pace indicating the insertion.
:grey_question: No changes required because these phases are in common usage within English, see #4218 The use of 'in situ' is not marked up as Latin, similarly 'ex-situ'. The page language is set to English, so any text which is not in English should be marked up.
:x: [for later phase of A11y work] #4209
abbreviation not marked up. This is just written as text: but “there is no reference (in the strong sense) to a specific moment” (p. 6). This seems to us to be a more adequate
but the p.
is an abbreviation for page, probably should be marked so that it gets read out as page, to avoid confusion.
:x: #4243 example (27) is followed by (27') for comparison, but these are read out identically. Where punctuation is used to differentiate there needs to be additional markup for screenreaders.
:x: BirkbeckCTP/bau#38 something strange going on with this ordered list - its not all one list, and so voiceover doesn't read it out correctly, it does not identify these as list items.
✅ Abbreviations is H2
:x: #4244
The definitions for abbreviations should be a list, not comma separated values within a paragraph. A list can be traversed. They should also be marked up with<abbr>
tags.
✅ Acknowledgements is H2
✅ Competing Interests is H2
✅ Notes is H2
✅ Endnotes have links back to their position within the text, but (see next)...
:x: #4245
[']
used with the carat as the only link text within notes are problematic. All of them end with these, so there is nothing to differentiate them, and repeated 'carat, link' is unhelpful when traversing the document by links.
✅ References is H2
:x: #4221 doi in references are exposed, rather than as contextual links
:x: [for later phase of A11y work] #4209
DOI
is read out as a word to rhyme with boy, this needs to be marked up as an abbreviation and to be read as D O I
? #4244 references are a list and should be coded as such, rather than as individual paragraphs? A list can be traversed.
:x: #4218
in Bearth, Thomas. 2002. Fokus im Akan – Outline. Handout of a
the Fokus im Akan
needs language to be specified. Similarly other reference titles that include phrases in other languages.
:x: #4232 Share is an h4, indicating that it is a subsection of References
✅ lack of alt text for share icons already being addressed in #3936
✅ Authors is consistent with Share, at H4, (see above for issue with Share at H4)
:x: #4221 ORCiD logo, read out as 'ORCiD logo, link' but no differentiation as to which author or that this link leads to the ORCiD site - i.e. same description for both links which means they lack context when traversing by link.
:x: #4215
envelope icon following first author is missing alt text, such that is is read simply as 'link' with no description at all. <a itemprop="email" href="mailto:[redacted]" class="fa fa-envelope email-link"></a>
? Download is consistent at H4 - but could be improved by 'download article' or similar, to avoid confusion.
✅ Download PDF, link
✅ Download XML, link - could be improved by 'download article as XML' or similar, to make completely clear what is to be downloaded.
✅ Issue is H4, consistent
:x: #4211 "2024 bullet Volume 9" is confusing, mark the bullet as decorative.
✅ Dates is H4, consistent
:x: #4055 dates are given in a table. This has no column header. I suggest "Date" as the header, or not to use tables here.
? #4210 consider using dates in a format where the month is written out, because that reduces ambiguity between US and UK orderings (and sounds nicer when read aloud!) see examples from UK Government stylesguide
✅ License is H4, consistent
:grey_question: #4221 License was done better in Sample 1, where there was a brief description of what the license meant.
:x: #4221
The external link to the license info is an icon without alt-text, but it does have a rel=license
attribute. But the licence name is in a <p>
tag with no id - so it doesn't link to that licence.
✅ Identifiers is H4, consistent
:x: #4221 DOI issues as already described above
✅ Peer Review is H4, consistent
:x: #4215
<i class="fa fa-check-circle-o" style="color: green"></i>
no alt for the icon, but also not marked as decorative.
✅ File Checksums (MD5) is H4, consistent
:x: #4211
"Vertical line ISSN" the |
is being read aloud in the footer area.
:x: #4248 English En selection - reconsider position - would a user get all the way down to the bottom of the page in a language they didn't understand? Also, when there is only one language on this selector, should it be there at all?
I.
Roman numeral denoting it as the first section does not get read out as "one" but "I" get later for section 2, the "II." is read out as 2. This is a problem with the screenreader not being able to distinguish for single 'I' that it is a roman numeral. Similar issues are known with other single letter Roman numerals. See notes.dramatis personae
should have language markup“[t]here are three degrees
is read out as "t here are three degrees" and there needs to be better way to distinguish [t]here
from there
which doesn't end up hiding there resulting word from the audio. Not a bit issue in this example, but it could introduce ambiguity elsewhere.That Night a Fire did ...
should be marked up as a blockquote, not as a list. Currently this is mostly distinguished from neighbouring text by visual layout.Sections VIII-IX.
should include an internal linkfind himself involved in ... such things...
prior to endnote 25. Blockquote without a cite attribute (suspect this is because there is the endnote for citation, but that information should be in cite attribute too.<em>perjurii poena divina, exitium; humana, dedecus</em>
It might refer (i) to actual results, i.e., to the empirical impact—if that can be measured—that each lie actually has on the enterprise of human communication. Or (ii) it might refer
and so on. The initial (i)
is read as i
and then the (ii)
as 2
- see notes on Roman numerals.front matter - article - end matter
make this more navigable.Login>
This has been written with reference to sample 1, only adding in differences (not identified before, or different behaviour).
<a href="/mmb/funder_articles/http://dx.doi.org/10.13039/100016537">National Cattlemen's Beef Association</a>
#7440.8
is understandably read as "number seven thousand four hundred and forty dot eight". Might be worth looking at why this is not 'point eight', but the main thing is working (the hash being read as number).<a class="xref-bibr" href="#r39">1999</a>
this is a known issue, already being addressed in #3939−80°C
read correctly.A 2.54-cm-wide cut
not read correctly - cm not read as centimetres, but as c m
I think this is to do with the use of hyphens, as later in the same paragraph, 2.54 cm
is read as "two dot five four centimetres"3-min
is read as "3 min" rather than "3 minutes"standardized to 2,000 μg/mL of protein.
the mu was not read as micro. Note the mu is read out in a different voice (Greek voice?) and so it is very difficult to recognise. I expected "micrograms per millilitre" It is being read as "mu g slash m L". Slash is generally not being read as 'per'.mL
is being correctly read as millilitres. An α of 0.05 was set for a level of significance.
this is read as 'An greek letter alpha of ..." is that the desired behaviour?<strong>
tags rather than <th>
<span class="table-label">Table 1.</span>
.3222–3228.
they are interpreted as phone numbers in ios safari - and should be marked up as not phone numbers.pages: 1-13
is read as "pages one thirteen" because the -
is considered invisible punctuation as it could in say 'copy-editing` - this should read as "Pages one to thirteen" or similar. It should be described to be read that way."
following the class attribute, which prevents the alt=""
attribute being recognised <img src="/media/cover_images/d78e50a8-3542-4429-b049-20261739ee56.svg" class=" alt=" "="">
.ignoring navigation, header/footer etc if already appears on article sample. Looking for other content not yet tested.
ERR_CONNECTION_RESET
. This is not the same link address as appears in the footer on articles, so it seems to be a user-content error. <hr>
following the about and before news is read aloud, but this is decorative.[...]
at the end of a news article summary sets up an expectation of a link, but this is plain text.Volume 15 • Issue 1 • 2023
the bullets are read aloud.When navigating away from Sample 1 page, e.g. following a link to another page, the back button ends up returning to the article search page (from which this article was chosen) and not to the article itself. This is highly confusing.
headings seem to be split when ever there is another tag inside them. I'm not sure whether this is a peculiarity of voiceover or a problem with our HTML. Need to investigate further. It is highly annoying! Because in this article there is a lot of use of <em>
inside headings due to the subject matter, and so single headings end up being split as if three different by consecutive headings.
despite both being H2, the "Download" heading is larger text than the "1 Introduction" heading. Is there consistency in how these styles are being applied, because otherwise they have the same font, and as font size is indicative of heading level, having H2 of inconsistent size is confusing.
is the use of [ ]
within quotations enough to indicate to screen reader users that such words have been inserted into the original quote? Would a user expect to change their punctuation settings when this detail is needed, or would they want quotations marked up in such a way to make it more audible regardless of punctuation settings?
using small identifying marks such as 27 to 27' to differentiate between two items is difficult visually as well as for screen readers as described above.
<em>
or <strong>
but as the html appears to be correct, this may be an issue with my use of voiceover or my settings, so need to look at whether I have same problem with other sites.in the footer where it is Published by Open Library of Humanities should that be Published by the Open Library of Humanities
navigation menu - Login and Register appear side by side, but all other links have their own line
further to Sample 3 - item 14, "contact" text itself is not formatted, such that it default to the user's browser settings, possibly the only text on the page that doesn't have a font set. If user has a very different default font to the theme, this looks very odd!
Roman numerals are not just a screen reader A11y issue. There doesn't seem to be an accepted solution, with general advice being to limit their use. Avoid the use of Roman Numerals and unfamiliar symbols in text where possible.
WC3 guidance
is lang
markup needed for Latin phrases in common use within English?
review use of <blockquote>
and <q>
tags, and their cite
attributes.
e.g.
and i.e.
are automatically read out as for example
and such that
respectively. While not a problem here, might be worth considering highlighting this behaviour to editors.should the 'bullet' in lists be read out? or marked as decorative?
something that has become annoying very fast has been navigating by headings, when there are lots of subheadings - need to learn whether this is about my knowledge - i.e. there is a simple way to navigate to the next same level heading, or whether this is purely about page navigability in which case some way to move between same level headings might be good for us to implement.
login
link at the bottom of the article page, which then arrives on the press logo at the bottom of the login page.navigation between pages - there are no breadcrumbs, no sitemap!
navigation within pages - a skip to main content would be nice and something that makes it clear one has reached the footer. For long lists of articles, some way to know how long the list - either pagination or a 'showing results x of y would be helpful.
dates written as "YYYY-MM-DD" are not very readable.
clean bug - Where the news article is being truncated for the summary on the home page, if there is a line break this is not being translated into a space when concatenating with the next line, they are being run together.
The results here are now integrated into the rest of the audit.
Context
Manual Testing of a sample of pages for screen reader capacity. This is for functionality, not user experience. The aim is to identify 'low hanging fruit' issues which can be addressed before future more exhaustive testing including direct user feedback.
Testing is non-exhaustive.
[edit] It was going to include multiple OS and browsers, and several different journals, with the first tests on iOS Voiceover on Safari because that's the one I know the best so I am more likely to catch such issues and reduces lead time on training on a new system. Early results from this after 2 sample articles show that the many of the errors identified are then clearly visible in the code. They are not 'I wonder how a different screen reader would handle this' but rather 'this heading should be an H5 instead of an H4' which would affect all users/screenreader softwares. So this exercise has been revised: it will be a representative sampling just with iOS Voiceover, then after the identified issues have been addressed, we will have another go at the original plan for testing with multiple different screen readers, browsers etc.
Approach