[SCL] Re: Formal_Semantics =/= Semantics

[Jon Awbrey]

o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o

JS = John Sowa

JS: You are absolutely 100% correct in your assessment:

JA: What you are doing here is treating the "language of sets" (LOS)
    as a canonical language and prospectively providing translations
    from various other languages to LOS.  Furthermore, you are trying
    to do this by purely syntactic means, that is, without the boost
    of "collateral acquaintaince" with objects, the way your average
    human translator might.

JS: But as you should note, we are not translating human languages.

Let me mark that statement for future comment and move on for now.

JS: We are translating formalized languages, each of which has been
    formalized in terms of a set-based model theory.  To translate
    those languages from one to another it is absolutely essential
    that we use the dreaded LOS, and any other kind of translation
    that considered or dealt with any "collateral acquaintance"
    with objects would be totally inappropriate.

I have no dread of LOS, and I long ago got accustomed to using the term
"formal language" according to its formal definition, in lieu of whatever
prior intuitions I might have had about what makes up a language.  I am only
trying to make sure that we all make use of a proper description of the process
in question, so as not to confuse the masses, among which masses we frequently
number ourselves.  We appear to agree that the process is properly described
as a transduction from one formal language to another, a purely syntactic
translation, indeed, an arrow f : X -> Y, and thus a 2-adic relation.

> I am glad that you approve:
> 
> > Now, if you want to call these syntactic translations by the name
> > of "formal semantics" or "model theory", well, you are in a large
> > if not so good company, and so you can probably keep doing that.

It is not so much approval as resignation to common usage --
all it means is that I put the term "formal semantics" in
the same lexical category of non-analytic portmanteaus as
"formal language", "formal power series", ..., "formal X",
none of which necessarily has any clear semantic relation
to any of the more common senses of either "formal" or "X".
If there are any Fregeans about, I am sure that it would
bother them to violate compositional semantics like that,
but freedom from compositional semantics is a property of
languages that are higher up in the Chomsky-Schutzenberger
hierarchy than we like to think about in computer science.

With that reservation noted, we may move on.

JS: And I also agree with the following point:

JA: But I think that it would be a service to all concerned to clear up
    this long-standing confusion once and for all, and to state up front
    that formal semantics, done in this vein, has nothing semantic to it,
    not in the sense of objective reference, the way most folks understand
    the term.

JS: I did talk about that matter in my CS book and my KR book, but many
    people missed the point.  But I intend to put those issues up front,
    center stage in my forthcoming LLKW book.  I have no intention of
    doing that in connection with the CL/SCL standard.

I look forward to the appropriate footnote.
But I am puzzled that you appear to think
that a proper description of the process
in question ought to be kept sub rosa,
and not an integral feature of the
way that Common Logic represents
itself to the common populace.

Jon Awbrey

o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o
http://www.cs.bsu.edu/homepages/mighty/history.html
o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o