RE: Is XML a Language? (was RE: [xsl] XSLT Architecture: Next Step)

Subject: RE: Is XML a Language? (was RE: [xsl] XSLT Architecture: Next Step)
From: Didier PH Martin <martind@xxxxxxxxxxxxx>
Date: Mon, 07 Jul 2003 08:37:02 -0400
Hi Michael

Michael said:
I thought a meta language was a kind of language.

Didier replies:
Thinking about this subject this week end, I even more strongly believe that
a meta language like XML is not a language per se but a structure used to
create a language. Let's take an analogy from structuralism (Note: for
example works from Ferdinand de Saussure and successors). A grammar is a set
of rules that a language should conform to, it's a structure. Subjects are
placed in a certain position within a named phrase etc... Modern linguistic
is based to a certain point on Structuralism. 

Can we say that grammatical rules are a language? Probably not since the
words are missing. Without the words, the structure is not significant, it's
a structure without meaning. For example, I may have two languages with
similar structure and different words. Are we facing the same language in
both cases? Off course not. Is HTML the same language than SVG? I let you
provide the answer. Is XML without element and attribute a language? Let's
continue our exploration...

A structure without words (i.e. keywords, elements, etc...) doesn't mean
anything. Moreover, in XML the grammar specified by recommendations 1.0 and
up is incomplete, you need to specify, in addition, the position of
elements, their occurrences, their attributes with an external language like
for instance, Relax, DTD, XSchema (not specified in XML specifications but
the recommendations are referring to only certain ones). Thus, we can say
that XML specifications (either 1.0 or more recent ones) provide only one
part of the structure, the other part is provided by other structure
definition languages not defined in the XML recommendations version 1.0 and
up. Some of these structure definition languages are even not defined by the
vendors' consortium also known are W3C. Some where defined prior to XML and
more specifically defined within the ISO process.

The other part missing in XML as a stand alone entity are the words, the
meaning, the elements. These latter are defined in the structure definition

In conclusion, if XML is a language then, we can reasonably say that
grammatical rules are also a language. However, linguists will argue that a
language is comprised of the grammatical rules and the words (i.e. structure
and meaning). This last statement can be intuitively resolved by answering
to the question: Is Latin the same language as German because both share a
common set of grammatical rule? Would a Roman coming back to life today be
able to understand German even if both languages share some common
grammatical rules? Is a language without words a language? Would our Roman
perceive both Latin and German languages as similar? To get an answer we
have just to ask the question to our re-incarnated Roman :-) 

XML only provide a part of the structure (i.e. grammatical rules). This
implies that, at this stage, it is barely a language or maybe a phantom
language or in philosophical terms an intention toward a language. To create
a real language, you need the help of a XSchema, DTD, Relax and tutti quanti
structure definition languages. To get a real language, you need to add the
words (semantics), how these words are placed (grammar), how many
occurrences are allowed in different contexts (grammar), what are the
allowed attribute attached to words (semantics). Nope, XML per se, or at
least as defined in the recommendations is too incomplete to be called a
language. It is however a very useful tool to create ones. 

Didier PH Martin

 XSL-List info and archive:

Current Thread