Lambda the Ultimate

An essay like this is worthless..

I remember for example that previously on LtU, a teacher made two variants of a language: with space as indentation (like Python) or with a C-like indentation.
He found that beginners learnt much more easily with the Python-like indentation: this is very interesting, a scientific way to check what is best for introductory language.

But just arguing random reason to support language X is worthless.

I'd rather the dude just said "I really like Perl and think it should be taught as a first language" rather than pretending like he came to this conclusion after measured consideration and dispassionate application of logic and reason!

Indeed, and here is why:

Now, if I had to summarise [Linda McIver's language] in one word it would be this: sexless. It's incredibly limited, not flexible, and not fun. It has no pointers or references and instead relies on nested structures and arrays. There are two basic data types - a number and a string. The language does not have functions as first-order objects, closures, or objects and classes in the Object Oriented Programming sense. Furthermore, it has very few ways for one to express oneself. As a result implementing many algorithms would be very difficult in it. [...] McIver's approach is flawed in the sense that she is trying too hard to save the students from all possible problems they may encounter in trying to understand their introductory language.

The author does not believe in an incremental approach to introducing new concepts. The availability of the kitchen since is paramount

...it seems that he cannot even tell first-order from first-class functions, although these are almost opposites.

The author does not believe in an incremental approach to introducing new concepts. The availability of the kitchen since is paramount

I do believe in an incremental approach to introducing new concepts. For example, I think that it is very possible Object-Oriented Programming is not necessary to be taught at first, and its discussion delayed for later on. However, the language that McIver proposes is too limited, limiting and sexless.

If we choose a programming language, then the entire language cannot be taught at once. However, it is necessary for the students to like the language, be able to experiment on their own, and be able to make use of it later on.

This was the case for BASIC for me, back in the late 80's and PC XT era (and I continued to use QBASIC even after I learned C), and this is the case for people getting their feet wet with Perl, PHP, Python, Ruby, etc. today.

Most of the SICP exercises are about number theory, recursion, and a lot of other relatively abstract stuff, and too few are about real world...

I hereby propose James's law: when somebody uses the phrase "real world" he or she probably means "my world." Authors should be encouraged to use the later, more accurate phrase.

All I can say is that any analysis that comes to the conclusion that Perl would be the best choice for first teaching language must be deeply flawed. I mean deep, as in a hole so deep that its bottom sticks out at the opposite side of the planet.

Well, it would be useful if you told us why you think so. You've made an assertion, yet you didn't give evidence for it. Furthermore, if you believe my analysis is flawed, perhaps you'd like to point its flaws.

And finally, I gave several possible languages as good choices for introductory programming languages, not just Perl, and just said that Perl would be my preference, in part because that's what I'm most familiar with.

All of the languages reviewed are dynamically typed (Perl, Python, Ruby), or weakly typed (C). What about strongly typed languages like SML, OCaml, or Haskell? What about other long-standing functional languages, like Scheme or Lisp that have been used in CS programs for decades? What about a dynamically typed object-oriented language, like Smalltalk? What about long-standing logic languages like Prolog? The analysis simply isn't comprehensive enough not to be flawed.

As a personal judgment, there seem to be far too many concepts for beginners to grasp in even basic Perl programs. Simple SML programs are far simpler than simple Perl programs, for instance.

then you're asking the wrong question.

If your answer is Perl, you wouldn't recognize the right question.

As I often point out, Perl is a write-only language. In contrast to APL (the other famous write-only language), it is hypothetically possible to write maintainable programs in perl. I'm not aware of any examples that exist in the wild, however, and there is a reason for that...

I'd like to know what didactic goals Logo is supposed to support. Exploratory learning?

I suppose that is addressed to me.

Logo is good to learn concepts such as sequencing, loops, procedures etc. According to Wikipedia "Logo was a primary influence on the Smalltalk programming language."

The graphical interaction encourages exploratory learning. I think it is an excellent choice as a first introduction to computer programming.

My question wasn't directed just at you, though it was your comment that provoked me to ask it. In fact several regular LtUers have expressed admiration for Logo, including Ehud and Peter van Roy. I'm curious to know more. Is it a programming language just for secondary education, or would it make sense as an introductory undergraduate course? If we want to apply Pappert's ideas to constructing a constructionist pedagogy for a course aimed at teaching general PL principles, how would we go about this. I really haven't a clue

Brian Harvey's three volume series, "Computer Science Logo Style", might help answer these questions. It's a "three-volume series intended mainly for teenagers and their teachers, or for adults who aren't professional programmers." Info and links are on his home page.

Is it a programming language just for secondary education, or would it make sense as an introductory undergraduate course?

My impression (based on little more than the web pages linked in this comment) is that Dr. Harvey considers it more the former (he's "interested in the use of computers in K-12 education") than the latter. He teaches an SICP-based course at Berkeley, using Scheme.

Indeed, background is important, which is why McIver's analysis is good. It basically assumes syntax and semantics similar to high school level algebra and mathematics that most people starting a course in programming are already familiar with. I think that's fairly reasonable, and transitioning from such a language to SML should be fairly smooth, so this author's contention that it's not "sexy" or "powerful" enough, while technically correct, is not that big a deal IMO.