[Edu-sig] Learning (some more) programming

[Paul D. Fernhout]

John-

An excellent post; and I'll have to agree with most of it, including your 
conclusions at the end, especially in relation to choosing educational 
strategies based on empirical research.

Some minor comments below.

John Zelle wrote:
> On Thursday 28 December 2006 12:51 pm, Ian Bicking wrote:
> 
>>Paul D. Fernhout wrote:
>>>Ian Bicking wrote:
>>I offer keyboarding as a counterexample.  
> 
> 
> I'll be keeping this reply very short, as I never learned to touch type. My 
> keyboarding style is, at best, idiosyncratic.  I'd score that one for Paul, 
> based on my experience. I never "learned to type." I just do it.
> 

Actually, and unfortunately, I am a self taught keyboarder. I started on 
the original Commodore Pet chiclet calculator style keys.
   http://en.wikipedia.org/wiki/Commodore_PET
I've tried a few typing tutors since (twenty years or so into typing) but 
never stuck with them. I can type at 60+ words per minute, but, I still 
mostly need to look at the keyboard, and my accuracy is not 100% (so, more 
revision is needed, especially if I am excited and type faster). 
Fortunately, spelling checkers help with some of my loss of accuracy 
(though sometimes introduce semantic mistakes if I don't read over things 
carefully, so another cost of being self-taught here).

So, I'll have to agree with both you guys -- you can be self-taught and 
really good at something, but you may still have lost something. A typical 
example from music is classical pianist training versus jazz pianists who 
picked it up on their own (and have quirky styles and can never play as 
complex pieces).  On the other hand, today's typing tutor programs have 
become so interesting that there is little question in my mind that if I 
had been offered access to one from the start on full sized keyboards, I 
would have learned as a young kid (the value would have been obvious to 
me, and the experience interesting). I think my best bet at this point if 
I wanted to learn to touch type (having a previously learned system to 
overcome) would be to use a different keyboarding modality. I tried a 
chord keyboard, but the one I picked (the twiddler) was unergonomic. So, I 
should either get another one or perhaps get a Dvorak lettered keyboard.
   http://www.maltron.com/maltron-press.html

On the other hand, and perhaps I am wrong i  thinking this, it seems quite 
a few people do not have my ability to take somewhat-legible written notes 
without looking at the paper (except occasionally). And that just emerged.

Still, there is another issue here that for *writing* like emails, touch 
typing can be a big win. But for *programming* (especially in some 
languages with lots of symbol characters or numbers) touch typing is not 
so much of a big win. Programmers spend most of their time reading code 
(why Python is such an innovation), and when they write code it often has 
symbols in it or unusual words (so conventional keyboarding training 
focusing on letters is limited). Also, the task of browsing and editing 
code is much more mouse intensive than just writing emails, and using the 
mouse is disruptive to touch typing. So, when writing code, lowered 
keyboarding performance isn't quite the problem as with writing text. 
Still, even their, I'll agree, compared to someone who is a touch typist, 
like my wife, it still would be better to touch type, to improve the flow 
from ideas to screen.

Now, to go on the offensive here, Doug Engelbert and others clearly showed 
even in the late 1960s and early 1970s  that a set up with a chord 
keyboard in one hand and a mouse in the other is much father than a full 
keyboard and a mouse when using a typical computer application.
   http://en.wikipedia.org/wiki/Chord_keyset
Then why doesn't everyone use this settup? My only conclusion is that 
there seems to be a general problem with people investing in IT skills and 
technology. :-) And that was a big part of Doug Engelbart's point: that 
people have to be willing (or able or aware) to invest in technologies 
that make them more productive.

Note that court stenographers do use chord keyboards:
   http://www.slate.com/id/2119534/
and othter compression practices, resultign in accordign to the following 
link approximately 225 words per minute at very high accuracy (many users 
of this machine can even reach 300 words per minute):
   http://en.wikipedia.org/wiki/Stenotype
Why don't we all make this investment to multiply our text input speed by 
a factor of five? Especially the people here advocating touch typing. :-) 
I know for stenographers it is a very high investment (two to three years, 
thought I am not sure how much time per day is spent on typing).
   http://stats.bls.gov/oco/ocos152.htm
So perhaps another issue comes up here of diminishing returns.

But in any case, I don't think learning keyboarding skills (which is 
perhaps the one most important thing a kid can learn in the internet age, 
until voice recognition accuracy improves) can justify thirteen years of 
compulsory education. Maybe as a compromise, we can think about which of 
these things are *really* important and whittle the compulsory part down 
to a few years. :-)

>>>Nothing is more likely to make children not want to read or write than
>>>following standard pedagogical advice and breaking reading into a series
>>>of incremental hoops (learn letters, learn words, learn simple sentences,
>>>and so on) which is just going to bore most kids out of their skulls. Can
>>>you imagine if we tried to teach kids to listen to spoken language and to
>>>talk that way? Thankfully, kids learn to listen and talk on their own by
>>>just absorbing language in their environment and trying to use it to
>>>accomplish  goals meaningful to themselves.
> 
> 
> Here, things get a little silly. Is there really evidence that children who 
> are motivated to read (say because their parents read to them) are somehow 
> then de-motivated by having that skill taught to them in a sane way? And yes, 
> I'm afraid there is research showing that it's sane. Children who already 
> know their letters pick up reading faster than those who don't. Starting with 
> simple sentences leads to faster learning than starting with Shakespeare, and 
> the best way to quickly enhance reading is to tackle "graded" texts that are 
> at about 75% comprehension. Any easier than that, and you are not learning 
> anything new. Harder than that, you don't have enough scaffolding to figure 
> out new constructs from the context (since the context makes no sense).  Not 
> only is that common sense, I believe it's supported by 
> _actual_reading_research_.  I might also add that around my house those 
> incremental "hoops" are themselves bringers of much delight. Successfully 
> learning the alphabet is a source of joy to children.
> 
> However, I really want to weigh in on the second point of this paragraph, as 
> it brings up a faulty analogy I often see in educational debates (even those 
> surrounding programming ;-). I have considerable background in the areas of 
> learning and language acquisition (specifically machine learning for natural 
> language). The consensus of modern linquists is that learning to speak is 
> almost nothing like learning to read (or driving or programing or...you name 
> it). Learning to speak is in your genes. We are adapted through evolution to 
> be a speaking species. Our ability to learn language is so innate and acute 
> that it's arguably best thought of as a "speech organ" or "language instinct" 
> (see Steven Pinker's excellent book of that title). No normal child fails to 
> learn speech in _any_ culture, even those that Ian points out below do not 
> necessarily encourage it. Learning to speak is an inevitable developmental 
> phenomena that requires only exposure to speech at an appropriate age. 
> Just like children don't need to be taught to recognize their mother's face, 
> they do not have to be "taught" to speak.

Overall, I'll have to agree with you based on the facts that humans are 
wired for language. And you are right to say the specifics of language 
learning are different because of that from many other human learning 
tasks. (My undergraduate advisor, George A. Miller,
   http://wordnet.princeton.edu/~geo/
would also probably be very displeased if I said otherwise. :-)

Still, clearly if we did teach language like we taught other things, the 
basic machinery would probably have a hard item with it, you have to admit 
that -- we would be going against the grain of how humans are set up to 
learn language. So, clearly, there is precedent for saying, humans 
naturally learn well a certain way, and trying to structure such learning 
in certain unnatural ways may actually reduce progress.

Still, even for foreigh language, the "puberty" argument by itself is 
being discredited (even as it is less common to learn languages later, and 
certainly much harder to speak them without an accent). On accent, what 
seems to be more the case is that kids learn to *not* distinguish certain 
sounds in the first year or so of life (that is, the processing ability 
for , say, distinguishing between "r" and "l" is lost early on for Chinese 
speaking natives). We can get that discrimination back somewhat later, but 
it is very hard. As for "puberty", as kids get older, lots of other things 
get more interesting. See:
   http://ivc.uidaho.edu/flbrain/latelang.htm
"Merrill Swain (1979), a leader in the field of foreign language learning, 
believes that early immersion students enter into the process of learning 
a second language at a time when it does not compete with other interests, 
as it is an integral part of their normal school activity.  Older 
students, on the other hand, quickly recognize that learning a second 
language involves considerable time, dedication and effort, consequently 
preferring to spend their time and energy elsewhere.  In other words, 
older students may excel in their initial rate of second language learning 
as input is more comprehensible for them because of their background 
knowledge--they are faster acquirers as well as faster learners and 
because of this they have a greater ability to consciously learn grammar 
rules (Krashen & Terrell, 1983), while younger students excel in long-term 
second language achievement.  However, it is a myth to think that children 
find the process totally painless (Hakuta, 1986).  The most difficult 
learning task for children and adults alike may be the attempt to acquire 
second language proficiency in school environments (Asher, 1982).  It is 
simply not true that young children learn a new language more easily and 
quickly than adults because the many variables that are directly involved 
in the process of learning a language such as specific situations, input, 
interactions and most importantly, the amount of time invested in language 
learning in a quality program make language learning hard work for both 
groups."

I think there is some other research somewhere that says when adults are 
taught the same way as kids are taught, that is with pointing and naming 
of objects, and short social interactions, that adults learn language 
faster than through conventional adult means of text books and repetition 
of dialogues. Which makes sense if you think some basic wired machinery is 
being properly activated. Throughout human history tribes with differing 
languages have been neighbors and learning multiple human languages 
informally has been an important survival skill.

Almost everyone agrees that if you want to become fluent in a language, 
the best way (and perhaps only thorough way) is to live somewhere it is 
spoken as a part of normal life.

Having said that, almost every academic likes to think their specialty is 
of special importance. Linguists are more likely to be right, of course. 
:-)  Still, many people communicate via sign language, or touch codes, are 
via typing and reading, even from birth, and so clearly "language" like 
capacities are also applicable in other modalities than speech and hearing.

> Skills such as reading, writing, mathematics, and programming are not innate. 
> In fact, even with great effort, many people do not learn to read well. There 
> are precious few self-taught readers. The analogy to speech just doesn't cut 
> it. It's even less appropriate for considerations of writing and mathematics. 
> Even learning a second language after a certain age (around puberty) simply 
> does not (and cannot) happen the way we learn our first language as children. 
> Second language learning draws on different cognitive mechanisms. So all 
> those language courses that promise you can learn effortlessly the way a 
> child learns language are just blowing smoke.
> 

Now, agreeing their may be special wiring for language, consider how 
people learn so many things -- playing the piano, how Unix directory 
structures work, really big philosophical ideas, or learning some aspect 
of programming. Often times we need to return to something over and over 
again (in slightly different contexts) before one day it just "clicks" and 
becomes clear. That is a function of how a neural network is processing 
information. And it is clearly not just "hierarchical". I'm sure everyoen 
here has had "aha" moments like this abotu a lot of things. Sometimes they 
are not obvious -- like when you get new glasses and a fewer days later 
you stop noticing them.

Hierarchical presentation may help some in some cases (depending on the 
skill). Maybe even a lot in a lot of cases. I have no problem arguing for 
the value of well written tutorials, innovative learning games, texts of 
increasing structured difficulty, or whatever. It is more the compulsory 
aspect and the timetable aspect I have problems with. And those are 
aspects which deny that we don't really understand what is going on in 
that neural net, and also often deny the role *intrinisic* or what I might 
call "extrinsic direct value" motivation plays in learning. By extrinsic 
direct value I mean you see directly the value of what you are learning to 
solve a problem of importance to you (e.g. touch typing or driving); it 
isn't just about a grade. We don't know all the details about motivation 
and learning, but we do know motivation is a really important part of 
learning; and we know that frustrating and boring kids repeatedly turns 
them off learning in an area.

One can, even for language, distinguish between "immersion" and 
"submersion" :-) see:
   http://en.wikipedia.org/wiki/Language_immersion


>>Teaching is central to the idea, but it is decentralized insofar as it
>>doesn't expect or rely upon a professionalized set of teachers, and that
>>success in the system depends on children becoming teachers themselves.
>>  Part of the community structure is to help those children become
>>*good* teachers, by teaching each other how to teach.
> 
> 
> Teaching at its heart is really just good communication. Part of communication 
> is making sure you have a receptive audience, that's where I see all the 
> various discussion of things like enthusiasm and internal vs external 
> motivations. And I think Arthur's original post about taking on interested 
> learners, vs. getting them interested is an important distinction to think 
> about. BUT. The second part of communication is providing the right 
> information that will most effectively communicate. Teachers have been 
> teaching for many centuries now. We know a lot about how children learn, and 
> many ideas that have been demostrated through both experience and research 
> are now standard parts of educational curriculum. Now, I'd be the first to 
> admit that I have very little patience with a lot of so-called educational 
> research, but finding better teaching methods can (and should be) an 
> empirical, scientific process. I have _no_ patience for the "everything we're 
> doing now is just plain wrong" crowd. These sorts of claims are seldom backed 
> up by any evidence except vague hand-waving and appeals to ideology 
> (centralized, hierarchical BAD, unstrutured GOOD, "coercion" BAD, voluntary 
> GOOD, lecture BAD,  contructionism GOOD), or false analogies (kids should 
> learn to read the same way they learn to speak).
> 
> Can we improve education? Almost certainly. But I'm sure we can also make it 
> much, much worse.  Before we throw out the bathwater, let's see what's 
> actually in it.

I'll have to agree. Still, even the results of scientific studies can be 
biased, based on the funder, the assumptions, the researcher, the peer 
review process, and so on.

All the best. Wish I had more time right now to reply more fully.

--Paul Fernhout