WebAIM - Web Accessibility In Mind

E-mail List Archives

Re: PDF, Accessibility and Quality Control


From: Zwack, Melanie C
Date: Jan 29, 2007 8:40AM

Here are some key responses I got the the questions about PDF,
Accessibility, and Quality Control:

* If time is not a critical issue and the PDF was not
well-formed from the start (i.e, no headings, etc.), then I do suggest
using the TouchUp Reading Order tool to reclassify the content where
appropriate. While the TouchUp tool is a major improvement to how
things had to be done in earlier versions of Acrobat, a proper workflow
is a better option IMO as touching up even short document can take more

* In terms of QC/QA, I generally do not recommend the use of a
specific screen-reader as what works in screen-reader version "X" does
not mean that the same function will exist in version "Y" (makes for
some tedious testing!). I try out a few screen-readers when I am trying
something new in the PDF conversion process, but I have that luxury.
The best results I have had with respect to evaluating reading order is
to save out a tagged PDF as "Text(Accessible)" from the Save As... menu.
You should get the proper text flow and additional items that a
screen-reader will also announce are typically displayed in brackets.
The drawback to this method is if data tables are present (linearizes
the data).

* When you test with a particular screen reader it can be
problematic, because as you say, a different version will give you
different results. Also, this way people who use a different screen
reader might also be excluded. It is best to follow good practices, be
as maticulous about tagging as possible, and let assistive technologies
worry about reading a properly written document. It pays off on the long

These are all interesting responses, and I really appreciate the input.
My points about the above are:

* For MS Word and InDesign, I recognize the fact that in those
sorts of PDFs, I have control over how the tags will be outputting.
These sorts of PDFs are much easier to deal with. However, I guess my
real concern is for those PDFs, such as those developed in Quark or
other programs, where the output tags is not so clean, and there's no
control over the source-files. At this point, we've just been making it
work - - getting the content in the correct sequential order and
applying alt tags to images. This is a pretty basic goal, but given the
fact that the PDF is a difficult one to work with, and the goal of
trying to tag it perfectly could really literally take 2-3 times as
long, we've made the decision that this basic goal - - getting the
content in the correct sequential order is good enough. I've disregarded
whether the tags correctly associated with <h3> <p> <figure> - - so far
this has worked. Do you feel that this is acceptable in terms of quality
control? Or, should we try to actually tag these correctly?

* Another question I have, You speak of well-formed documents.
What exactly do you mean by that. I guess I'm not totally clear about
what the meaning of a well-formed document is?

* We are currently using Acrobat 7 Pro, and possibly soon
working with Acrobat 8 Pro. Since Acrobat 5, it was just amazing to get
a document tagged in the first place. But, now several years later, I
think it is important to re-evaluate these standards, and figure out if
there are better ways that we need to do things.

Melanie Zwack

-----Original Message-----

From: Zwack, Melanie C [mailto: <EMAIL REMOVED> ]

Sent: Friday, January 26, 2007 12:16 PM


Subject: [WebAIM] PDF, Accessibility and Quality Control

Can people share what they do to ensure Quality Control for Accessible
PDFs they produce?

My main question: How vigilant are you that the tags used in the
tagging, accurately represent the content? For example, in the past we
have placed content in sequential order as best we can, but pretty much
ignored the tags. For example, even if a paragraph text is tagged with a
header tag, we would just let it go, because mainly due to the fact of
how much time that it would take to correct this. Now, with Jaws 7.0
that we just got, it is actually reading 'Graphic: Then the content
here' - - so it is specifying the tag explicitly. The fact that in the
past we did not pay much attention to this, is now a problem when we use
Jaws 7.0 to Qc. Jaws is specifying each tag, before reading the
content, so it is obvious in the case when an incorrect tag is used. Any
thoughts on this?

Also, I am evaluation our Quality Control process in general for
Accessible PDFs, and would be very interested to hear of any other
people's procedures for this area. I would like to make sure our process
is where it needs to be.

I'll also start off with specifying our process:

1) We apply tags to a PDF and touchup the tags as necessary (ie.,

adding ALT tags to images, re-arranging the order of the tags as


2) Quality Control staff spot checks the PDFs to see if the tags

are reading right. (We do not use the automated Accessibility Checker to
verify correctness, but instead we have Quality Control check each file
with a Screen Reader.)