Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!mhuxr!mhuxt!houxm!whuxl!whuxlm!akgua!gatech!gitpyr!chen
From: chen@gitpyr.UUCP
Newsgroups: net.cse
Subject: Re: various recently-discussed topics (learning languages)
Message-ID: <1549@gitpyr.UUCP>
Date: Sat, 15-Mar-86 14:19:56 EST
Article-I.D.: gitpyr.1549
Posted: Sat Mar 15 14:19:56 1986
Date-Received: Sun, 16-Mar-86 10:45:12 EST
References: <1092@burl.UUCP> <1093@burl.UUCP> <117@andromeda.UUCP>
Distribution: net
Organization: The Clouds Project, The School of ICS, Georgia Tech
Lines: 45

In article <117@andromeda.UUCP>, marco@andromeda.UUCP (the wharf rat) writes:
>In article <1093@burl.UUCP>, rcj@burl.UUCP (Curtis Jackson) writes:
>>   ... He said, "Oh, yes,
>>   get this book [holding up a copy of K&R] from the bookstore, $12.  You
>>   have 2 weeks."  Now THAT is the way to teach a computer language!
> 
>      Why is that the way to learn a language ?  Admittedly, you can
> pick up a new language in a few weeks, but to learn a new language
> *and* complete tricky projects at the same time seems a bit much.
> Also, I find that learning a language that way leaves a lot of
> gaps-I learned macro-32 that way, and anyone who's seen my postings
> to net.decus will know how many gaps I have in my knowledge of the subject.
> To me,the only way to learn a language well is to write lots and lots
> of stuff in it, not just figure out how to do one thing. 

By the second year, a CS major had better realize that there's more to
CS than just "learning languages".  I agree with Curtis.  CS people
should be able to learn languages on their own.  This is because
languages allow programmers to express abstract concepts in a concrete
way.  Languages differ in the concepts they allow for (semantics) and
the exact mechanics of expressing them (syntax).  Learning a language
is a matter of figuring out what concepts the language is built around
and then figuring out the nitty-gritty syntax.  So, if the languages
are similar enough (and most procedural languages are) one non-trivial
programming assignment should be enough to get people acquainted with
what the language can do.

Unfortunately, it's hard to "teach" people this.  You can lecture them
all you want on similar concepts/constructs in langauges, but in order
for it to catch on, the students not only have to learn the facts about
languages, they also have to learn a new way of looking at languages
and *that* is impossible to guarantee.  You can't push people down
that road, the best you can do is show them the way and hope they
follow.

The one time students might need help learning a language is when the
language embodies concepts the students haven't seen before.  Things
like fp, and prolog probably fall into this category.  Then the
students might need some help picking up the concepts (not the
language).  After all, in the case of prolog say, how many
undergraduates know what a backward-chaining, resolution theorem
prover is?

	Ray Chen
	gatech!chen