Re: The Promise of Forth

John Doty <jpd@[EMAIL PROTECTED]
> writes:
> Bruce McFarling wrote:
>> On Mar 24, 4:16 pm, John Doty <j...@[EMAIL PROTECTED]
> wrote:
>>> But can one call a notation that lacks the properties of a language,
>>> "language"?
>> What in your view is the properties of a language?
>
> The universals linguists identify in languages: syntax, tree
> structured grammar, parts of speech, ...

The following trivial gramar (using EBNF) defines an idealised
concatenative notation (I'll refrain from calling it a language) :-

  expr := symbol | number | '[' expr ']' | { expr } .
  program := expr .

The following trivial grammar defines a idealised Lisp :-

  expr := symbol | number | '(' { expr } ')' .
  program := expr .

Either both define a language or neither do.  If the former, then
Forth can have a grammar if having one is important, if the latter
then Lisp isn't a language.

Clearly grammar alone does not define a language, without semantics it
is all just meaningless symbols.  Lisp, APL, and concatenative
languages share a trait that they have a trivial grammar and push
almost everything into the semantics.  Thus grammar is mostly
irrelevant and semantics are what is important.  That approach appeals
to some, but others abhor it.  For example, Larry Wall, a trained
linguist, has nothing good to say about the Lisp approach and thinks
that the Perl approach is the right way.  I'm with
McCarthy/Iverson/Moore but it seems that the majority is with Wall.