WebAIM - Web Accessibility In Mind

E-mail List Archives

PDF tables (again): NVDA, VO not reading headers

for

From: Alan Zaitchik
Date: Jan 2, 2020 5:50PM

I have tried every combination I can think of to get NVDA and VO to read
all column headers and row headers in a table I am remediating, but to no
avail. JAWS handles the table nicely, and the table shows no errors in PAC
1.3 or PAC 3. But NVDA and VO have (different) problems. Suggestions most
welcome!
The table has 2 columns, each with a simple column header cell in row 1.
Also, each data row in the table has a row header in the row's column 1
cell. But the problem comes from the fact that every group of rows is
introduced by a 2-column-wide (table spanning) row that describes the
"group" of rows that follow. For example, the main column headers are
"Fund" and "Fee", but the group "titles" are like "Mutual funds" or
"Stocks", etc.
I have tried setting up these group titles as column headers for the rows
that follow, using generated unique IDs and the data cells' "Headers"
attribute. I have also tried using the Scope attribute. I have also tried
mixing both the above. I have also simply left these titles as (colspan=2)
data cells rather than making them header cells!
No matter how I twist and turn NVDA and VO have problems (but not JAWS).
Specifically, VO will read the column headers in row 1 ("Fund" and "Fee)
but ignore row headers. NVDA announces row headers but not the column 2 row
1 header "Fee". (It does announce the column 1 row 1 header "Fund"!)
Neither screen reader announces the group title column headers, but truly
that matters less to me (and I'm happy to leave them as data cells with
content tagged as a Heading to support easy navigation through the table.)

It seems to me that the mere colspan of these title cells interferes
somehow with NVDA and VO. Is that so?

Any ideas?
A