Myndex
commented
3 years ago Hello @sdw32 thank you for commenting.
I am well aware of all of these issues. From what I gather you may have not read the actual guidelines and specifications, as most of what you stated is described there, or in the whitepaper materials which you have previously been given links to, but for the record, here: https://www.w3.org/WAI/GL/task-forces/silver/wiki/Visual_Contrast_of_Text_Subgroup
I believe you are referring to the "single line title" which is intended as the simplest description of the objective goal, but being simple leaves out all of the minutiae. That is part of how the total WCAG 3.0 document is structured, agnostic of content.
The "deepest" discussion was moved out for development into several white papers that are in progress, some of which is covered in the link I provided above. While I do have reservations regarding making some things too simple, in this case, the single line descriptor:
Provides adequate luminance contrast (lightness/darkness difference) between background and text colors in order to read the text easily.
Characterizes the objective in the simplest single line manner possible. You should study WCAG 2.x and all of WCAG 3.0 to see how these structures are applied, and the manner that context is used for displaying farious aspects of the standard. The overall structure is a completely separate discussion, and not related to content, for instance, your first several paragraphs are very clearly illustrated in the guidelines.
Also, the APCA tool directly addresses most of the rest, in an objective and testable manner.
To address some of your other assertions:
Emphasis added:
It might be feasible for a developer to generate paragraphs of text containing these 15 random words, in their chosen style of text (font-name, font-size, font-weight, line spacing, foreground colour, background colour). The guideline could specify a reference text style, and the developer could test the reading speed of a person with vision loss, for both the reference text and their text, and their text passes the guideline if the rate of reading speed for their text is not less than the reference text.
No. This is not objective, not testable in an automated manner, not "feasible" in that it's presupposed that either authors or testers create mock ups of the testable content within the narrow definition of the specific reading test, for each and every variation of style (not at all practical ina typical webpage which has 15 to 40 sizes, styles, colors) and THEN ALSO employ a "person with vision loss", and moreover fails as a single-test-subject-with-impairment for reading speed is not an instructive metric.
The statement "test the reading speed of a person with vision loss" is unsupported. There is no "one type" of "vision loss," and one impairment type's needs may (and often does) interfere with another impairment type's needs. Nothing useful is learned through this approach, nor is it consistent, nor is it feasible from a practical implementation viewpoint.
Reading speed tests
Wilkins is two and a half decades old, suffers many of the same issues of most reading speed tests, and worse, is not a simulation of actual content. MNREAD is the current widely accepted method, and there are some other newer methods and hybrid methods that are pushing the envelop farther.
-
Reading speed tests are very useful in research. In research, typically using a large number of test subjects to eve- out the individual differences in cognitive and neurological processing that are semi-independent axes of the reading speed evaluation.
-
Reading speed tests are very useful in a clinical setting, for evaluating issues with a particular patient.
-
Reading speed tests can not be directly applied as a "testable" method for evaluating a page design for the reasons mentioned above.
How we are using reading speed
The extensive work of Dr. Lovie-Kitchin and separately Dr Legge defines important metics that affect reading speed, including the key concepts of contrast reserve, critical contrast, acuity reserve and critical acuity.
I suggest looking through that research, as it is the current reference standard, and used extensively in the development of WCAG 3.0 visual contrast and SAPC/APCA.
Thank you,
Andy
Andrew Somers W3 AGWG Invited Expert Myndex Color Science Researcher https://www.myndex.com/APCA/simple
alastc
bruce-usab
ChrisLoiselle
sdw32
From https://www.w3.org/WAI/GL/WCAG3/2020/outcomes/luminance-contrast-between-background-and-text
Luminance contrast is not a useful outcome measure in its own right. Extremely large text may be acceptable at extremely low levels of luminance contrast, whereas small text needs much higher levels of luminance contrast.
I would suggest the outcome measure for text should be based on ensuring that text can be read at a maximum speed (readability), specified in a way that can be objectively measured.
Readability measures that I am aware of include the 'Maximum Reading Speed' from the MNREAD acuity chart, and the reading speed from the 'Wilkins rate of reading test' (https://www1.essex.ac.uk/psychology/overlays/rrt%20OC4.htm). Of these, the latter might be more useful, because it uses continuous repeats of 15 random common words, so is not influenced by learning effects and can be repeatedly administered for the same participants.
It might be feasible for a developer to generate paragraphs of text containing these 15 random words, in their chosen style of text (font-name, font-size, font-weight, line spacing, foreground colour, background colour). The guideline could specify a reference text style, and the developer could test the reading speed of a person with vision loss, for both the reference text and their text, and their text passes the guideline if the rate of reading speed for their text is not less than the reference text.
It might even be possible to use a method of simulating impairment as a proxy for running this test with participants who have some degree of vision loss. For example, the guideline could specify that the test is run with a participant who has reasonably good vision, but the viewing distance is manipulated to represent a specified degree of vision loss.
The Wilkins rate of reading test is covered by existing IP arrangements, but I have been in conversation with Arnold Wilkins, and he is open to the possibility of contributing a version of this test to the WCAG guideline on a royalty free basis. I will look forward to following up on this in due course, in whatever manner would be appropriate for this.