Is Python the Esperanto of programming languages?

Steven Taschuk staschuk at telusplanet.net
Sat Mar 22 21:54:06 CET 2003


Quoth Carl Banks:
> Steven Taschuk wrote:
> > I don't, however, think that these add up to the morpheme "saying
> > nothing".
> 
> Ok, let's say it means nothing in the same sense that sin x = x.  How
> about that?

When I first read this, I was truly baffled; now I'm just unsure
what you mean.  My best guess is that you mean "it means nothing"
has an implicit free variable indicating the context, i.e., "it
means nothing in context X", and that by not specifying that
variable when you said simply that the morpheme means nothing, you
intended an implicit quantifier "in most contexts X".

If that's accurate, I think we agree on the substantial points and
are just using the words differently.  (I'd prefer to say "adds no
information in context X".)

> > In my personal theory of meaning (idiosemantics?), it
> > always means "third person singular", and its absence always means
> > "not third person singular"; this explains why
> >   1.   The man goes to the store.      is good,
> >   2.  *The man go to the store.        is bad, and
  [...]
> I don't agree with the reasoning here.  Things that are paradoxical or
> nonsensical don't always sound bad, and things that sound bad are not
> always nonsensical or paradoxical.

Well, no.  I hadn't intended to imply so general a rule.

However, I don't see any bright line between "sounding bad" and
"semantically incoherent".  For example:
    4a.   The man goes to the store.
     b.  *The man go to the store.
    5a.   John ceased to have fits of anger.
     b.  *John ceased to have a fit of anger.
    6a.   That sentence is false.
     b.  *This sentence is false.
(4b) sounds bad, and is unacceptable for that reason whether the
meaning is clear or not.  (6b) sounds fine, and is unacceptable
only for high-level semantic reasons.  I borrow (5) from McCawley,
who analyzes (5b) (and other examples) as unacceptable because
"cease to" can be applied only to things which "can be interpreted
as denoting a state" as opposed to an event -- a semantic
restriction.  But (5b) causes (in me) a mental hiccup that feels
very similar to the one caused by (4b).

I also consider rejecting (5b) as part of language competence,
while rejecting (6b) is not; thus I want a theory of English to
account for the former, but not the latter.  This is so whether or
not the theory includes a model of language processing, and if it
does, whether that model is stated in terms of pattern matching or
inference or other mechanisms.  (Indeed, it seems possible to me
that all inference actually occurs by pattern matching.)

My sketch of a model can be extended naturally to cover (5); I
think it could be restrained from covering (6) by restricting the
types of inferences the model allows.

I don't claim that this sketch of a model is a faithful
representation of how language is actually processed; I have no
idea how it's processed.  Your fascinating example of the little
French expatriate shows that it is possible to be competent enough
in a language to reject things like (4b) but not so competent as
to reject things like (5b) (I assume she wouldn't); this might, as
you suggest, reflect the operation of two distinct faculties, but
it might just be a weird level of competence, or, say, a
preservation due to her unusual circumstances of a state that all
children pass through before they are able to speak.

-- 
Steven Taschuk                  staschuk at telusplanet.net
"Telekinesis would be worth patenting."  -- James Gleick





More information about the Python-list mailing list