Japan DAISY Consortium
JDC101, 1-1-61 Wakaba-cho, Chofu-shi,
Tokyo 182-0003, Japan
TEL: +81-3-5384-7207 FAX: +81-3-5384-7207
October 11, 2021
Open Letter to Browser Vendors, W3C, and WHATWG: Towards
Accessible Ruby
Dear Browser Vendors, W3C, and WHATWG,
The Japan DAISY Consortium has promoted DAISY textbooks for those Japanese students who
have print disabilities. We have been providing organizational and technical support to those
who produce DAISY textbooks in Japan. We are thus practical rather than theoretical or political.
The historical purpose of introducing ruby to Japanese typesetting practice was
probably to help those who could not read difficult CJK ideographic characters. Ruby
indeed helps such people. However, we have observed that ruby also hampers
accessibility. First, existing text-to-speech implementations do not work well for
documents containing ruby. Second, the typical rendering of ruby is hard to read for
low-vision and dyslexic people.
Making ruby accessible is a challenge. It is necessary to improve browser
implementations, and it is also necessary to improve HTML documents and CSS style
sheets.
We believe that CSS of W3C will play a critical role in making ruby accessible. We also
believe that some features of W3C HTML Ruby not present in WHATWG HTML are
crucial for making ruby accessible. The Japan DAISY Consortium cordially requests
browser vendors, W3C, and WHATWG to study accessibility problems around ruby.
These problems have been sadly overlooked, as we see them.
We hope that this letter will raise awareness on the accessibility issues with ruby, and
that in the near future relevant stakeholders can coordinate and take up the necessary
technical work, both on specifications and more importantly on implementations, to
improve the accessibility of Japanese on the Web.
Current status of ruby accessibility
1) T2S
As of 2021, T2S does not work well for documents containing ruby. We created a wiki
page documenting this problem and hope to create a W3C Note from it.
That page is restricted to user requirements and thus does not consider possible
enhancements to current implementations. Here we would like to point out that the
generation of accessibility trees requires special treatment of ruby. Simply stripping tags
representing ruby while keeping text leads to double reading of base and ruby, which is
very annoying and, in the worst case, completely misleading.
Ideally, each base-ruby pair should be clearly represented in the accessibility tree or
DOM tree. It then becomes easier to use both the base and ruby to choose the most
appropriate phonetics. There should also be some way for authors to indicate if the ruby
is used for phonetic annotations (in which case reading both is inappropriate), or as
supplemental information (in which case both should be read), as heuristics cannot
reliably determine that.
2) Dyslexia
Some dyslexic people cannot separate base characters and ruby. Thus, a base character
with associated ruby looks like an unknown character having a strange radical to them.
To overcome this problem, it is important to widen the gap between base characters
and ruby. But this is not possible in most browsers except Firefox. The use of a different
color for ruby is also helpful and is fortunately supported by most browsers.
3) Low vision
Ruby is much smaller than normal text. Ruby is thus difficult for low-vision people to
read.
In printed materials, it is quite common to avoid smaller characters (e.g., U+3063) and
use similar-but-larger characters (e.g., U+3064) instead. However making this kind of
substitution changes the phonetics of the word, because these smaller and larger
characters, while visually similar, are in fact distinct letters. But IT technology allows
other possibilities. By converting smaller characters to larger characters at rendering
rather than in the document source (see the text-transform property of CSS, currently
only supported in Firefox), we can make ruby easier for low-vision people to read
without hampering T2S.
It is also possible to change the font size of ruby depending on user preferences. Prof.
Nakano of Keio University reported that low-vision students prefer ruby long in height.
4) Textual fallback
Although visual rendering of ruby is expected to put the ruby in parallel to the base, the
textual or linear fallback of ruby is often required. Such fallback is used for metadata,
user interface components (see an EPUB issue), and so forth. When mono ruby is
attached to a sequence of base characters, it is unfortunately too common to see ruby
after each base character (e.g., 東とう京きょう), rather than combined after the compound
word (e.g., 東京とうきょう). Such awkward results are a natural consequence of the
inappropriate structure of WHATWG ruby markup (e.g., <ruby><rt>とう</rt><rt>きょ
</rt></ruby>), a problem W3C ruby markup extensions address.
5) Switching of rendering style
Given Japanese text having CJK ideographic characters, one can choose to present ruby-
annotated text in many ways: showing or hiding some or all annotations; inlining them
in addition to, or in place of, the base text; and varying their typesetting style (mono vs
jukugo) and spacing. Different presentations can be appropriate for different reading
levels, reading preferences, and disabilities.
Unlike printed materials, IT technology, in theory, allows on-the-fly switching between
rendering styles, but such switching is not practically realized.
6) Conversion to braille
When assistive technology converts HTML or EPUB to Japanese braille, just as in the
text-to-speech situation, it would be helpful if each base-ruby pair is clearly represented
in the accessibility tree or DOM tree. It then becomes easier to use both the base and
ruby for choosing the most appropriate braille.
Sincerely,
Hiroshi Kawamura
Chairman of the Steering Committee of the Japan DAISY Consortium
Vice President, Assistive Technology Development Organization (ATDO)
Board Member of the DAISY Consortium (one of the founders)
Member of the WAI/W3C Steering Council
Standing Committee Member of IFLA/LPD Section
Member of the Committee on Services for Persons with Disabilities, Japan Library Association
Global Chair of the ICTA/Rehabilitation International