W3C home > Mailing lists > Public > w3c-wai-gl@w3.org > April to June 2005

Re: Definition of "semantics"

From: Jason White <jasonw@ariel.its.unimelb.edu.au>
Date: Mon, 13 Jun 2005 12:12:39 +1000
To: w3c-wai-gl@w3.org
Message-ID: <20050613021239.GB3159@jdc>

On Mon, Jun 13, 2005 at 01:07:12AM +0200, Jens Meiert wrote:
> 
> "Semantics" lexically is "the philosophical and scientific study of meaning"
> [1] (or try an analogy [2]). Cool.
Correct.
> 
> "Semantics in the domain of Web standards" not only seems to imply that
> elements (I'd narrow it to Markup languages) actually have got a meaning,
> but that they need to be used accordingly - everybody uses semantic Markup,
> but only a few use Markup according to its semantics (which is a problem).
> 
This is also correct.

I have argued in earlier contributions to this discussion that the term 
"semantics" as applied to markup languages is consistent with its 
meaning in respect of natural languages; the difference lies only in the 
type of language involved. The definitions offered in this most recent 
thread confirm that view.

WCAG 2.0 applies to formats which are not markup languages, and for that 
reason we need to employ the concept of semantics in a broader sense 
than that restricted to markup alone. Also, I think it would be mistaken 
to suggest that the meaning of "semantics" as used in connection with 
markup languages is in any way different from that applicable to natural 
languages, as Joe Clark's comments in this thread and elsewhere appear 
to imply.

The main problem so far as guideline 1.3 is concerned is that it doesn't 
define which structures, and which semantics, need to be separated from 
presentation. In particular, the requirements of guideline 1.3 are not 
subject to the baseline, with the result that any semantics capable of 
being captured in a markup language or other content format are 
potentially required to be so expressed in order to fulfill the success 
criteria of guideline 1.3, depending on how these are worded and 
interpreted.

Defining "document semantics" or a similar expression in terms of the 
semantics of markup languages does nothing to solve this problem, and it 
would artificially restrict the application of guideline 1.3 to 
technologies designed as markup languages. Pertinent semantics, on the 
other hand, can be supplied via API's or file formats not amounting to 
markup languages.

I propose:

1. That guideline 1.3 be made explicitly dependent on whatever 
technologies are included in the author's chosen baseline: structure is 
separated from presentation so far as is possible using the technologies 
in the baseline.

2. That we include a requirement whereby the features, including, where 
applicable, the markup language constructs, of each of the technologies 
employed in the content, be used in accordance with the semantics 
prescribed in the specification for that technology.

Importantly, in this proposal we should be careful to specify "the 
semantics of the technology", rather than using the term "semantics" 
without qualification; it is the unqualified use of "semantics" that 
leads to the problems I have outlined and accounts for the difficulties 
that have been cause for controversy in this discussion.
Received on Monday, 13 June 2005 02:13:02 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 5 February 2014 23:39:37 UTC