Xref: utzoo sci.philosophy.tech:3160 sci.psychology:3176 comp.cog-eng:1747 sci.lang:7056
Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!swrinde!zaphod.mps.ohio-state.edu!uakari.primate.wisc.edu!uflorida!stat!sun13!gw.scri.fsu.edu!pepke
From: pepke@gw.scri.fsu.edu (Eric Pepke)
Newsgroups: sci.philosophy.tech,sci.psychology,comp.cog-eng,sci.lang
Subject: Re: Computer Languages and the Sapir/Whorf hypothesis
Message-ID: <489@sun13.scri.fsu.edu>
Date: 23 Aug 90 15:41:28 GMT
Sender: news@sun13.scri.fsu.edu
Followup-To: sci.philosophy.tech
Organization: Florida State University, but I don't speak for them
Lines: 63
References:<5137@munnari.oz.au> <11606@pucc.Princeton.EDU>

In article <11606@pucc.Princeton.EDU> EGNILGES@pucc.Princeton.EDU (Ed 
Nilges) writes:
> The lesson of pseudocode (the use
> of a structured form of natural language in system specification)
> is that skill at expressing algorithms may not be tied to knowledge
> of ANY programming language whatsoever.

Most of the examples of pseudocode I have seen omit the details of 
programming languages but comprise many assumptions about the structure of 
code.

My own informal opinion, based on a great deal of learning and some 
teaching, is that programming languages can serve as conduits for learning 
various programming techniques which can then be readily transferred to 
other languages.

I wrote my first recursive descent top-down parser as part of a freshman 
programming exercise in Pascal.  Up until then, I had not really grokked 
the power of recursion.  After then, it was trivial to do the same thing 
in BASIC using a push-down stack.  Pascal is a good way of teaching 
recursion, and BASIC is not, but once taught, the technique is widely 
applicable.

Similarly, learning SNOBOL teaches one about pattern matching, the idea of 
success or failure as an additional piece of information apart from the 
value of an expression, and associative programming through table lookup 
and the ability to create new variables on the fly.  APL teaches one about 
certain kinds of data abstraction and operator overloading.  LISP teaches 
one about treating programs as data and vice versa.  Even FORTRAN and 
COBOL have lessons.  FORTRAN teaches about numerical data types such as 
COMPLEX and a primitive kind of operator overloading, and even COBOL 
enforces some sort of organizational discipline.

Learning as many languages as possible, even the "bad" ones, teaches an 
intuitive understanding of Turing completeness.  It also teaches the 
mistakes of language designers, which are as important to learn as their 
successes.  The process of learning involves extracting from a wealth of 
information that which is most useful.  I have no evidence that humans are 
generally incapable of doing this, so I view Dijkstra's pedantic runs as 
counterproductive.  People who can easily jump from one programming 
language to another and who can learn a new one in a week's time are 
stronger programmers than those who waste their time in immature squabbles 
about whether C or Pascal is "better" or "more structured."

Such an overview, however, only comes after a person has learned several 
programming languages.  Perhaps a similar thing occurs in human languages, 
but most people just don't learn enough of them to have the effect.  I 
don't know what is typical, but I only know a mere three human languages, 
all of them European, compared to about 30 programming languages,
including different Assemblers but excluding other dialects.

In summary, while various programming language can expand one's ability to 
use programming techniques, the idea that they can constrain one's ability 
is very shaky.  

Eric Pepke                                    
INTERNET:pepke@gw.scri.fsu.edu
Supercomputer Computations Research Institute MFENET:   pepke@fsu
Florida State University                      SPAN:     scri::pepke
Tallahassee, FL 32306-4052                    BITNET:   pepke@fsu

Disclaimer: My employers seldom even LISTEN to my opinions.
Meta-disclaimer: Any society that needs disclaimers has too many lawyers.