In all cases, when I asked JAWS to read the whole page, I heard: “Using background dash image to replace text vertical bar example one hello world exclaim“. [Replacing "one" by "one point two", etc. as appropriate] I am using Internet Explorer 6.0.
Each of the examples mentioned above uses a different method to hide the raw text. All of them appear to get their hidden text read aloud in JAWS without a hitch. (The words preceding the “hello world” come from the document title.) It appears the Fahrner Image Replacement (FIR) technique may be complying directly with its intent summarized at the end of the tutorial: With responsible use and careful testing, this method can be used reliably to enhance the presentation of type on the Web, while preserving accessibility for image-exclusive browsing and indexing applications.
I’m still unsure of behavior in other screen readers, such as IBM HPR, or WindowEyes. But it appears we may be able to put to rest our readability concerns of hidden text, at least for JAWS.
Now I need to go hit up some of the naysayers who originally pointed out this technique’s critical flaw. What say they now? Ahem… Mr. Clark? Nice to meet you in Austin this past weekend. But we seem to be slowly proving this technique still works?