Hello Paul,

Several screen readers and talking web browsers have support for some sort
of table navigation already.  They tend to allow the user to move from
cell to cell, down an indivdual column, or through any chosen row.  At a
minimum these packages will take the information at the top of a column or
start of a row and use that as the default header for the given
column/row.  In some cases they will go beyond this and allow the user to
define what is to be used as a header and even choose how much of that
header is verbalized as one navigates around.

So, while AT is by no means perfect, I think from what you are describing
doing the script creation you are talking about might not be necessary.

Kelly



On Thu, 28 Jun 2001, Paul Chapin wrote:

> I've been working on making sure a table I'm adding to a web page is
> accessible.  My first instinct was to hit all the markers set by the WAI
> such as adding all the headers and captions and the like.  Then it occurred
> to me that there was still a problem.  No matter how good I was a doing
> this, the linear presentation of the material would require that the user go
> through the entire table and remember the appropriate column headings while
> listening to the contents of each row.  This is long, tedious and a
> potential source of errors.
>
> When a sighted person reads a table, they often don't look at every cell.
> By using the row and column headers, they cut to the cell or cells that they
> want.  It struck me that if the table was properly constructed according to
> the WAI guidelines, it should be possible to create a script that could be
> run against a web page with a table or tables that would automatically
> create descriptive pages for the table that permitted more intelligent
> browsing by breaking the table apart and creating a linked approach to the
> table.
>
> For example, the table I'm working with deals with Cyrillic email.  Whether
> the results are readable depends on the client used to write the email and
> the client used to read it.  An intelligent description would first present
> the user with three links,
>
> Read the entire table
> Pick a mail reader client
> Pick a mail writer client
>
> Assuming the person chose to Pick a mail reader they would be asked,
>
> Netscape Messenger
> Outlook Express
> Web Mail
> Outlook Web Client
>
> Now assuming that they selected Outlook Express they would be asked,
>
> Pick a mail writer
> Read the entire column
>
> If they chose to pick a mail writer they would get a new list.  Picking one
> of the items on the list would result in getting the result for the that
> cell.  Reading the entire column would result in text something like
>
> With Netscape Messenger, No
> With Outlook Express, Yes
> With Web Mail, No
> With Outlook Web Client, Yes
>
> Actually, the table is more complicated that this.  The result vary with the
> settings selected in the writer and also according to which browser you use.
>
> My question to this group is whether this would be a useful way to present
> table data to somebody using a screen reader or am I just making a
> complicated problem worse? As a sighted user I don't trust my instincts on
> something like this.  This looks to me to be a doable script, although it
> might require some rigid rules about headers, but I don't want to take the
> time to work on the script if it's not a problem worth solving or if there's
> an existing solution I don't know about.
>
> ------
> Paul Chapin
> Curricular Computing Specialist
> Amherst College
> 413 542-2144
> (The attached file is an electronic business card.  If you don't know what
> that is, ignore it.)
>
>