One Language at a Time

Sep 01, 2017

Miles Berry

Most of us working in computer science education seem to be agreed that coding is not the end in itself, but is the means through which our students get useful things done, express themselves creatively and develop their understanding of the fundamental principles of computer science. Whilst some of our students will go on to be the software engineers and computer scientists of the future, almost all will find themselves using computers, and an understanding of how these machines work and how to think about problems so that the computers can help solve them, is what’s important for these students.

Where some of us disagree though is the best means to this end. On the one hand, there are those who argue that these principles, and indeed the constructs of programming (such as sequence, selection and repetition) are best learnt generically and then applied in as many programming languages as time, teacher expertise and student interest permits. Others take the view that actual, practical experience of coding in one language is the best way to learn to program, and through this to acquire, either through explicit instruction or guided discovery, an understanding of the underlying principles.

The arguments for the ‘many languages’ approach seem, to me, similar to those used in the ICT teaching past: that to avoid death by PowerPoint, we should allow students to choose between it, Prezi, Google Slides etc, as the software skills themselves are not what matters. The ‘many languages’ advocates argue that there’s more to programming than Scratch, or even than Python; and to avoid children becoming bored with Scratch, they should also learn Logo, Kodu, Hopscotch, Tynker, etc.

Somehow though, programming seems different from making a presentation. If we’re serious about students becoming adept at solving problems and expressing their creativity through programming, then they need to develop some sort of mastery of the medium: in short, they need to become fluent (or at least conversant) in one, or two, or three, particular programming languages.

By analogy with learning human languages, whilst we might be concerned that students get to grips with comparative linguistics and deep structure, we’re much more concerned that students learn a language, are able to hold productive conversations in it and, perhaps later, become able to write stories, essays or poems in the language.

There seems relatively little research comparing the one language and the many language approaches to teaching programming: mainly, I suspect, because this question rarely arises in undergraduate CS education: it’s simply assumed that introductory programming courses teach a particular language, although this still allows plenty of argument over the choice of paradigm and language.

Whilst the brightest and best will, I think, be able to transfer the concepts from one language to another, I fear we do a disservice to those who struggle to express their ideas in a first programming language by rapidly introducing them to many more: changes in syntax, grammar and vocabulary doing little more than adding to their already high cognitive load. Let’s learn the lesson from the mathematics education community, that mastery comes through depth of experience not superficial acceleration.

Originally published in Hello World #3 as Many Languages