What could be the possible advantages of teaching Scheme as first programming language?

Question

When I started my Computer Science studies, Scheme was the first thing we learned in our first programming course. Almost 8 years later, I ask a friend of mine, and it seems that they are still teaching Scheme for beginners. Besides the annoying amount of parenthesis, I found Scheme helpful, and I was able to understand important concepts as how interpreted languages work, anonymous functions, recursion and so on.

My question comes, because, except for that semester, I have never used Scheme again, and even though it was helpful I have always thought why not teach another real world use language.

Wouldn't, for example, Python help in the same way? Why Scheme?

Answer 1

I think that the purpose of such a course is not to teach you a language. After all, Scheme, with its abstract syntax, is pretty minimal as a language. The purpose of a course like that is to teach you to think abstractly. If you can do that now, several years later, you can probably thank the course for getting you started. Abstraction, after all, is the big idea about computer languages.

A course in Python or another language can do similar things, but not in quite the same way, since in those languages you also need to learn the language syntax, idioms, and other structures. Scheme pushes all that aside for the opportunity to focus on abstraction. The things that are not in the language make it easier to focus on the big ideas.

---

This isn't really related to your question, but I note that many language designers start out with a lisp- (scheme-) like syntax to try out ideas in the language before they think very hard about its concrete syntax. It turns out to be very good for such experimentation.

Answer 2

There's one more reason I'd like to add to those here, less high-minded, but also a genuine consideration. One of the practical difficulties of teaching an introductory course is that the students come in at vastly different levels. Some are quite competent already, and some are brand new. It is rare for kids to come in with any experience outside of imperative labguages, though.

By utilizing a functional language in your first course, you get to have a classroom full of kids who have a much more similar footing. And when you move on, later, to imperative programming, you no longer have any kids with no experience. It's a simple trick to keep your cohorts more on-level without wasting the time of the advanced students.

Answer 3

Every second you spend explaining a programming language is a second you are not teaching programming, software development, software engineering, software design, or computer science.

You can teach the entirety of Scheme in a single lesson. You can probably teach the entirety of Scheme in 10 minutes. Python is significantly more complex than Scheme, so, you will spend significantly more time explaining Python and not teaching CS.

Oh, and BTW: the world's most widely-deployed programming language is essentially Scheme with syntax. (And an object system inspired by NewtonScript, Act-1, and Self.) Personally, I didn't understand ECMAScript until I learned Scheme and Self. Teaching Scheme will get your students halfway there. So, in that sense, Scheme is one of the most "real world" languages out there.

Answer 4

With Scheme, you start teaching programming concepts on day 1 - and also implementing them as working code on day 1.

With a typical procedural language (C++, Java, etc) you first have to crawl through the swamp formed around 20+ years of accumulated obsolescent and deprecated syntax. The survivors from that experience might get to learn some concepts eventually, but only if the course doesn't run out of time first!

Recent languages like Python have the great educational disadvantage of features like duck-typing, etc. Of course that makes them very effective tools in the hands of good programmers, but not so effective as teaching languages it's too easy to make subtle mistakes. (And don't even mention the mess that Python got itself into over syntax changes in the upgrade from Python 2 to Python 3...)

Answer 5

Simplicity

You can write the definition of scheme on the back of a postage stamp. Therefore as @Buffy says, you don't have to learn the language at the same time as learning the concepts.

It is a pure functional language.

You will be a better programmer, because you learnt functional. Better to do it first, few people learn functional second.

Education vs training

Education is about learning the concepts, training is about leaning a language. Education is long term, training is short term. Training is about learning, Education is about learning how to learn (meta learning).

A personal experience

I leant scheme late, after basic,pascal,C,C++,C#,python,tcl, et al. I have never done it again, but I have not stopped using it, in the way I program in other languages.

Answer 6

Scheme is not that widely used in the industry, but that is not the point. The purpose of CS is not to teach you any particular language, but to teach fundamental concepts. When you know the fundamental concepts, learning any particular language is pretty easy. Anybody with a solid CS background would be able to learn Python in a few hours of self-study.

Scheme is well-suited to teach fundamental concepts for a few reasons:

• Built on very simple and logical concepts.
• Syntax may be unconventional and somewhat tedious, but it is also totally logical and consistent, which means you don't waste a lot of time on surface issues.
• Extremely flexible which means you can implement almost any CS concept and paradigm in it.

You can consider scheme more of a "blank slate" than Python. For example the SICP book teaches object oriented programming by having the student implement an object system from scratch in Scheme. Now in industry you might prefer a language which already have a solid built-in object system (like Python), but for teaching it gives a much deeper understanding to implement one yourself.

Answer 7

Procedural approach. Not a joke: with functional programming students are very quickly faced with the necessity to write their code as a set a (reusable) functions with a clearly defined role.

With the traditional approach of imperative/OO languages, they learn functions after a lot of other concepts : loops, arrays, etc. Probably too late.

For example: if you ask them a program to compare the average value of an array, a lot of them will write a loop to sum the values, not use a function for the sum of the array. Even if it was a previous exercice.

Answer 8

With many programming languages, you don't see the computer's view of your input: the tokenized form, the parse trees. All those are part of a computer language's universe and are kept out of the programmer's hands and access.

In Scheme, your input is the parse tree. That puts you more level with the computer than many other languages. In some manner, Scheme does not even have a programming language rather than just a convenient way to enter parse trees.

In contrast, the distinction (or at least its naming) procedural/functional is sort of a red herring: the programming style employed in Scheme quite often is rather procedural, doing one action after another.

Answer 9

We don't have Scheme in our entire country (here in India) but that honour goes to C programming language.

For the purpose of my answer, substituting C for Scheme, I keep asking myself, why bother with C. After that 1 semester of C, I have never used C professionally anywhere. In fact, once I step out of educational purposes, C becomes useless. That begs the question, after almost 25 years, why are they still teaching C? Why not something else?

I think, why they are still teaching C (in my country) and Scheme (in yours) is probably because it is ubiquitous. there are any number of books written on C. More importantly, the faculties who teach them ( I think most faculties change their jobs once in 60 years) can teach the same thing for the rest of their lives, use the same lab manual, and use the same code, and ask the same questions, use the same evaluations. From a strictly logistical point of view, it makes sense to teach C 30 years ago, and 30 years hence.

Can you image the mess, if every five years, the universities change their syllabus? Thats a logistical nightmare (not to forget other challenges) no one wants to deal with. For instance, one university could decide to go with C sharp, and another with Java. That creates all kind of problems. However, if all universities agree (which is how things are now) to simply keep using Scheme (in your case) and C (in my case), life becomes simple.

Further, the foundation that you get by learning an ancient language is that, everything that came after it, be default, would be easier. I learnt to ride a bike (as in motorcycle, not bicycle) in my dads old bike which was, well not that good. However, because I drove the tough one, all of today's bikes seem like child's play. Although I don't use C itself, anytime I learn a new language, I unconsciously compare it to C, and learn from that.

Update 1 : To add some context, I have been part of the IT industry for 11 years, and 5 of them as a developer/trainer/educator who travels widely. I am sure C is used somewhere, I haven't seen one single enterprise usage of C and I have worked for startups, small companies (less than 100 employees) and for MNCs (more than 2000 employees) and never ever has any division, anywhere talked about C, used C or implemented C.

Again, no disrespect to C (or its legion of fans) but I request that my opinion be taken in this context.