Re: [xml-dev] Semantic entities versus syntactic entities

[Piotr Bański <bansp@o2.pl>]

Hi Roger,

I think you are freely juggling the terms "semantic" and "syntactic"
here, not even minding their shift of status between the linguistics and
programming language design.

On 2010-05-29 15:19, Costello, Roger L. wrote:
> Compare these two sentences:
> 
>     1. Say your name out loud.
> 
>     2. Say "your name" out loud.
> 
> In the first sentence we expect to hear that person's name. In the second sentence we expect to hear the words, "your name."
> 
> In the first sentence the words "your name" are semantic entities. In the second sentence the words "your name" are syntactic entities.

No. In both cases, these words constitute a single syntactic entity, the
object of "say". Due to their different interpretive status (the latter
being a metalanguage expression), you arrive at their denotations
differently. But that is not a syntax vs. semantics difference, you get
this effect when you look at objects whose meaning is compositional (as
in the phrase "your name" in (1)), vs. those whose meaning is to various
degrees non-compositional ("idioms", proper names, metalinguistic
expressions like the quoted "your name" in (2)).

All in all, I'd say the opposition is false here. This is not a syntax
vs. semantics issue.

> By quoting the words we have turned off the normal semantic interpretation of the words.
> 
> In XML there is an analogous situation:
> 
>     1. <altitude>
> 
>     2. &lt;altitude>
> 
> In the first case we expect an XML parser to process it as an element. In the second case we expect an XML parser to process it as literal text.
> 
> In the first case <altitude> is a semantic entity to an XML parser. In the second case &lt;altitude> is a syntactic entity to an XML parser.
> 
> By escaping the less-than symbol we have turned off its normal semantic interpretation.

I may be risking running against the established terminology, but I
wouldn't see this as a syntax-sematics opposition either, even
remembering that we are moving away from natural language here.

I don't see how "&lt;altitude>" is a syntactic entity here -- it's just
element or attribute content, no syntactic constraints apply to it.
Semantic constraints may -- those defined in the schema.

Likewise, I'm not sure I'd want to call "<altitude>" a semantic element.
I'd be most happy calling it a formal element or even a syntactic
element -- an identifier of a node in the XML tree.

Best,

  Piotr