Re: Off Topic: metaphor in programming languages?

[Christopher Palmer <reid@pconline.com>]

On Fri, 19 Feb 1999, xod wrote:

> > Not so. Libraries are additions to the 'dictionary' of primitive 
> > functions; they are lexical, not metaphoric.
> 
> If I say to you "a stitch in time...", you know what I mean. I didn't make
> it explicit in my text; it was commonly understood.

'A stitch in time saves nine' (or whatever) is not a metaphor.

> If I use a library call to process a JPEG, the reader of the source
> code, whether human or compiler, knows what I mean. I didn't make it
> explicit in my source; it was commonly understood.

If

#include <jpeglib.h>

isn't explicit, I don't know what is. ;^)

Even if it were left implicit, it would not be a metaphoric construction.


> Everything is context-independent if your scope is broad enough.

That's not true. Context dependence is not necessarily about scope;
context dependence speaks to logical form. Take two examples from SPE-era
generative phonology (just because it's a clear example):

1. /s/ --> /t/

'/s/ shifts to /t/.'

2. /s/ --> /t/ { [+ cont. -voc.] __ }

'/s/ shifts to /t/ after a continuant consonant.'

Rule 2 is context-dependent, rule 1 is not.

Early theories of syntax (which were originally taken also to be complete
theories of language, including semantics) were entirely context-free,
which is why they failed in MT, among other applications. Most current
mainstream computational linguistics work is still being done in the vein.

It's quite sad, really.



---------(( Christopher Reid Palmer : www.pconline.com/~reid/ ))---------

              the characters i am, made into a word complete
                              -- Meshuggah