W3C home > Mailing lists > Public > w3c-wai-ig@w3.org > April to June 2000

Re: Linearizing Tables

From: Charles McCathieNevile <charles@w3.org>
Date: Fri, 14 Apr 2000 17:01:02 -0400 (EDT)
To: Steven McCaffrey <smccaffr@MAIL.NYSED.GOV>
cc: kford@teleport.com, w3c-wai-ig@w3.org
Message-ID: <Pine.LNX.4.20.0004141658100.9178-100000@tux.w3.org>

a good analysis of the problem, although it makes a fe assumptions that don:t
reflect my own experience.

The value of a table is being able to analyse data in two dimensions
reasonably easily. But they are not really easy to inspect - in many cases a
linearised version is a better place to look for maxima and minima, and some
other fetures.

This is why sighted people also use graphics a lot - the shape of a curve
constructed from data points makes some things very easy to find.


Charles McCN

On Mon, 10 Apr 2000, Steven McCaffrey wrote:

  Hi Kelly:
   A very good example illustrating many existing problems.
       As pointed out in a previous thread, 
  the example table in the techniques document of the table of cups of coffe consumed by each senator does *not*
  transform  as described, using JFW 3.2.  I don't know if JFW 3.5 transforms it as described.
  Even if it does, developers should not assume all users have JFW,
  nor that all JFW users have the latest version. 
  If the row and column headings were  spoken before the cells, this would be an adequate linearized version.  However, I still maintain that an adequate  linearized version of a table does not provide equivalent access, although it 
  does provide  a barely minimum degree of access.
  The following applies to      data tables.
  Linearized tables do not provide equivalent access.  Otherwise, why do sighted people use a two dimensional visual presentation and not a linear list of the cells? 
  In other words, if a person were to read a large data table to a friend over the phone, one cell at a time, would the person listening to the verbal enunciation of the table feel that he/she is getting the information out that she/he wants from the table?
  Clearly not.  What would happne in this case?
  The person listening would no doubt ask to have certain cells or even entire rows read out again, without having to listen to the entire table again, one cell at a time.  For data tables containing numerical data,(unlike the landsend.com example),  
  the listener might ask the reader  at the other end of the phone to answer a question like
  "How does the values of X change over time?  Is there a pattern?  Where does it reach a maximum..."
  In a table like that in the landsend.com example, if prices are also listed, one might want to ask,
  "What is the cheapest , most expensive etc." just to name a few questions I might want to ask.
  Another class of question might be "Is there product x in size range Y?" 
  All this, in a slightly different context, I have raised before in the How to describe flowcharts, ... thread.
  all this information could be provided in a long describption ahead of time, or by
  some interactive database-like query, (XML to the rescue?).
       Until such descriptions or interactivity exists, access to data tables is not equivalent,
  linearized or not.  And the fact of the matter is that still,
  many, if not most, screen readers do not even provide a correct linearized version for all tables.
      I accept a correctly linearized version of data tables as only an interim solution
  and hope the W3C WAI is still working on truly equivalent access to data tables.
  Tables are going to be used more and more, especially in data intensive areas like statistical or budgetary information.
  The overall principle boils down to the fact that a two dimensional visual representation 
  of a table is incorrectly assumed to be the definition of a table, and is thus,
  by its very nature, confusing presentation with logical structure.
  The definition used in the WCAG of "tabular information" is, very roughly, correct, and yet
  the WCAG and techniques documents still insist that a linearized version is accessible.
  As I said, a linearized version provides a bare minimum degree of access but falls far short of equivalent access.
Received on Friday, 14 April 2000 17:01:06 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 13 October 2015 16:21:08 UTC