WebAIM - Web Accessibility In Mind

E-mail List Archives

Re: soft hyphens hard coded

for

From: Mallory
Date: Jan 12, 2021 1:28AM

> A browser with an auto-hyphenation mechanism like Firefox would wrap at these hyphenation points if the word wouldn't fit in the line as a whole.

Since mechanical separation of words isn't something browsers are capable of (computers are still terrible at deciphering context of language), I cannot imagine a browser correctly hyphenating better than a human.

Additionally, where hyphenation happens and is acceptable seems to differ between languages. For example, Dutch newspapers seem to have no issue with hard word-breaks in random positions of words, which I think is gross and sloppy, but there it is. I don't think Dutch language pages ever use &shy.

So instead of browsers trying and failing, I think the speech engines should simply code in a full ignore on &shy characters. They should never expose them unless the browser is actually inserting a "-" character visually (and then do whatever they normally do with a hyphenated word). So far as I know, there is no requirement to pause at a visible hyphen. That depends on the speech engine AND the user's own settings, not user agents. NVDA doesn't generally even announce dashes within content at the default verbosity level (this makes date ranges sound... interesting :)

cheers,
_mallory

On Tue, Jan 12, 2021, at 9:09 AM, <EMAIL REMOVED> wrote:
> I try it again with an example:
> Given the word *hyphenation*. These are the hyphenation points:
> Hy‧phen‧ation
> A browser with an auto-hyphenation mechanism like Firefox would wrap at
> these hyphenation points if the word wouldn't fit in the line as a whole.
> I found no problems in pronunciation by any speech synthesizer, whether the
> word was wrapped or not.
>
> Imagine, there was an island called *hyphe*. Their inhabitants could be
> called the hyphenation with the following hyphenation point:
> Hyphe‧nation
> The library for auto-hyphenation in the browser has no exception for the
> Hyphe-Nation. So, the author must manually code the hyphenation point in
> HTML:
> Hyphe&shy;nation
> The word is correctly wrapped visually and:
> The word is correctly pronounced by speech synthesizers. (There should be a
> small pause between *hyphe* and *nation* to hear the difference). I didn't
> hear any aural problems for that usage. So, the pause is not a bug of
> synthesizers.
>
> Now, because not every browser uses auto-hyphenation yet, some authors hard
> code the hyphenation points manually (probably using some tools to do so):
> hy&shy;phen&shy;ation
> No big problem for visual layout, except that the CSS *hyphens: manual*
> setting shows more often hyphenations.
> But it is a significant problem, when listening to the text. Just paste the
> code in HTML and hear the difference with all that pauses.
>
> So, if there was an auto-hyphenation in every browser, excessive manual
> hyphenation could be deprecated.
>
> Therefore, there should be a requirement for UA:
> Provide auto-hyphenation.
> And there should be a SC for authors:
> Provide manual hyphenation only for meaningful differentiation.
>
> Wolfgang
>
> -----Original Message-----
> From: WebAIM-Forum < <EMAIL REMOVED> > On Behalf Of
> Patrick H. Lauke
> Sent: Monday, January 11, 2021 10:19 PM
> To: <EMAIL REMOVED>
> Subject: Re: [WebAIM] soft hyphens hard coded
>
> On 11/01/2021 20:13, <EMAIL REMOVED> wrote:
> > Thanks Patrick for your estimation,
> >
> > I thought of that too, but my tests with different German speech
> > synthesizers in different AT (jaws & nvda) and browsers showed similar
> > experiences.
>
> So it's a fault with most synthesizers/ATs then? Or how browers output it to
> the accessibility APIs/expose it in the accessibility tree? If the latter, a
> bug to file against browsers.
>
> Again, soft hyphens, CSS-based hyphenation, etc are visual properties that
> should not impact how something is aurally announced.
>
> > And since I didn't find any spec, I thought, there should be one and
> > where if not in UAAG?
>
> UAAG 2.0 1.4.6 is about the visual presentation, nothing to do with how it's
> output as speech by AT.
>
>
> --
> Patrick H. Lauke
>
> https://www.splintered.co.uk/ | https://github.com/patrickhlauke
> https://flickr.com/photos/redux/ | https://www.deviantart.com/redux
> twitter: @patrick_h_lauke | skype: patrick_h_lauke
> > > > >
> > > > >