3.1.2 nonvisual does not have "No information is available").

[GeorgeKerscher]

In reviewing the guidelines in 3.1.2 in the note it says:
This display field should be rendered even if there is no metadata (See the examples where "No information is available").

In the compact and descriptive there is no such statement. In the example, there is no statment either.

Seems like we either change the note or add it in the statements. It would take another example to show no information.

What do we want to do?

[mattgarrish]

Seems like we either change the note or add it in the statements.

I remember asking how this couldn't end in no information available if there was no metadata and the preference was to end with "May not be fully readable in read aloud or dynamic braille" because you don't have to have metadata to know there is content that may or may not be readable.

I'd amend the note, in other words, since there's no way to output a no information available statement even if there was one.

[GeorgeKerscher]

If we do not have the statement of not available,
Then are we saying that "May not be fully readable in read aloud or dynamic braille"
is coming from the publisher?

We need to finesse this.

[GeorgeKerscher]

I reviewed the EPUB techniques, and the final else statement depends on two values. However, it seems that if there is no value for images and no value on descriptions, or if there is no accessibility metadata, we should be able to make the no information available statement.

[mattgarrish]

That's the thing, I think you can only swap in "no information" for "may not be". I don't think we can have both output, as what metadata would the publisher use to explicitly say it may not be readable?

But let me ponder it a bit.

[mattgarrish]

FYI, I think the test for visual content not represented by text is currently flawed. It only looks at the visual access mode descriptors and ignores if there is text in videos, for example (e.g., instructional videos with text descriptions on screen).

You generally know if text equivalents are available by finding a sufficient access mode of textual, although this isn't a perfect test, either, as I mentioned in another issue because we didn't require extended descriptions in 1.0. It also fails if there is only an access mode of textual because publishers are not required to repeat this as a sufficient access mode.

What we probably should be doing is checking if there is a conformance claim. Any claim at level A has to have textual equivalents for non-visual content (except, again, for content that conforms to the 1.0 specification, but that's hopefully a small sample of content).

Making a claim about not being readable is only safe if an accessModeSufficient property is set and does not have the single value textual. In that case, we know the publisher has checked the sufficient access modes and is verifying that there isn't a purely textual one.

We might then be able to say the content may be readable if there is a textual access mode but no sufficient access modes and no conformance claim.

If there is no conformance claim, no access modes, and no sufficient access modes, then no information is available.

(This relies heavily on the accessMode and accessModeSufficient properties, so I don't know how well this translates to ONIX.)

[GeorgeKerscher]

If there is a metadata statement that there are images, and there is no information about alt text or extended descriptions, then the publisher has not provided accessibility information about this accessibility feature. Should we then say that no information was provided?

It would be correct to say that this may not be accessible, but this is not a publisher's statement.

I am leaning towards the no information approach rather than this may not be accessible.

[mattgarrish]

The current algorithm will also say that audiobooks, purely visual works, and tactile publications may not be fully readable, when it should say they are not readable at all, or, in the case of braille, fully readable in braille but not using read aloud. It's a bit naive in terms of what it can handle properly.

[gregoriopellegrino]

I remember that this decision was made to avoid the fact that most of the EPUBs on the market without accessibility metadata, although largely readable through assistive technology, would result with the statement saying that there was no information (which seems a bit negative), so we had opted for that "may not be fully readable".

[clapierre]

Could we combine the two statements saying, "No information is available, may not be fully readable." or something to that effect.

[GeorgeKerscher]

I am still concerned about making statements the publisher has not maid. We talk about publisher claims. May not be fully readable appears to be a publisher's claim, which we are infering.

For that reason, I think the statement should indicate that no information is available. This has the added benefit of publishers seeing that statement which encourages them to put in proper metadata.