Things we shouldn’t have to worry about with screen readers:
- Failing to:
- detect that a filename ending in
.pdf
or any link annotated withtype="application/pdf"
is actually a PDF - differentiate visited and unvisited links
- find the obvious main content of a page, clearly indicated by a heading element followed by paragraphs of text
- detect that a filename ending in
- Ignoring:
ins
/del
longdesc
- empty form fields
- semantically significant punctuation, including quotation marks, parentheses, and brackets
-
q
-
map
link
- Mispronouncing:
- “skip to content”
»
or other pi characters when obviously used as separators- acronyms (detectable by change in case in various languages, viz. GUI, ATypI, GmbH, Srl), including plurals (CMSs)
- delimited numerals (2005.10.10)
- Misunderstanding:
display: none
andvisibility: hidden
- any sequence that is unambiguous in the markup, including
<a></a><a></a>
or<a></a> <a></a>
- nested table headers
- imagemaps
- frames and iframes
- MathML
- Unicode
- Not reading:
title
s- CSS3 generated content
- PDFs
- Switching languages in mid-page
- Activating form elements without an explicit keystroke (as with pull-down menus lacking a Go button)
- Screwing up even very simple JavaScript
- Turning off sounds that automatically play