Link to Original Source
Link to Original Source
Here is a programming analogy of teaching kids to read and wrie in English. Let's say you need to program a computer to perform a task that seems logical, but has an enormous set of exceptional cases. Fortunately you know a lot about how this weird machine works and have known the problem domain intimately for years. Most developers would start here: pick a language, design the program and start hacking.
But Engleman saw a more efficient approach for this quirky machine. He experimented with then-recently-learned shortcuts and had the insight that it would be possible to bootstrap a slow but powerful, high-level interpretter that could understand a few primitives. Together with another scientist colleague, they spent a lot of effort analyzing and deconstructing the crazy program that had to be written — realizing along the way that they were handicapped by their intimate knowledge of the problem domain for so many years. With the help of many field trials and data mining, they experimented with a series of successive "programing languages" on a number or "real machines", each language bootstraps the next language, but eventually implements the program quite thoroughly in record time.
Selected features of the Engelmann/Becker DI reading/writing method...
- Limit "I/O" to/from the child. Always use the same simple "prompts" when asking questions and do it consistently. When getting answers, encourage steady paced answers (to a metronome beat!) at a rate which initially seems too slow, but steadily increases over time.
- English spelling is even less consistent than you realize. Not only should you bootstrap from a consistent subset, but try to engineer consistency even in non-obvious things such as the relationship between a letter's name and the possible sounds it makes. For example, with the vowels, "a", "e", "i", and "o", one can make use of the fact that their names say one of the sounds they represent (the "long" sounds, although they never confuse kids with that "label")...
- A child has to encounter a rule N times before it gets internalized. If they get it at N-5, and get bored at the seemingly extra encounters, there's no harm done in acquiring a "that's to easy" mentality. Conversely, you can take advantage of the fact that a temporarily remembered interim rule will be forgotten if not re-encountered
- To keep a child brain engaged, ask questions at a high, several-times-per-minute, pace, 80-90% of whose answers they automatically know. Sometimes this means you immediately tell them the answer before you ask the question. "This word is 'go'. What word? [go] Right! 'go'. There are 2 sounds in 'go'. The 1st sound is 'guh'. What is the 2nd sound? [oh] Right! 'oh'"
- One approach to bootstrapping English spelling in a consistent way would be to introduce a symbol for every phoeneme. So "go", "pot", and "do" use 3 different symbols for "o". But it would be better to avoid such a symbol explosion. One of Engelmann's symbol compression techniques is the "blue" letter. It is a letter that itself is silent, but causes another letter to "say it's name" (e.g. determine that a vowel uses the "long" pronunciation). So he starts off making what we remember as "silent-e", blue. Yet there are other blue letters. In the "combination" oa, the "a" is blue. That means "a" is silent, but "oa" "says the name of the not blue letter". There are many others "ai" where "i" is blue, etc."
- Englemann also uses a limited set of other compressed symbol phoenemic helpers e.g. squiggled underlining means that the sound for this letter or combination doesn't do what you'd expect from the rules you've learned up until now. These symbolic helpers last only a few lessons to help bootstrap recognition of a pattern until it gets pushed down to automaic memory, then the helping notation is dropped. When that happens, the child is told that "from now on, this will appear in 'adult' type 'without the squiggled line'".
- Kids don't mind inconsistency if it is constrained. The phoenemes only have to be close enough that when you are sounding out a word, you can get it. For example, the sound for the "qu" combination is taught as "kuh". So when you "say it, sound at a time" (e.g. sing the word in slow motion but not pitch shifted), then the word 'queen' is "kuh ee n" which would seem to be one syllable (and vowel) too many. But kids are never tripped up on that.
- The lessons were extensively field tested on kids. Data from these tests let them know when they were trying to go too fast or be too inconsistent.
Some of you may have experienced this kind of instruction before — leading foreign language programs, such as the Pimsleur method, use similar instruction techniques. But if you've never witnessed or otherwise experienced this kind of instruction, it is tricky to get the pacing, prompts, and encouragements/corrections right (even though everything is written for you as a script) while simultaneously monitoring and adjusting to your child's responses.
Because of this, I can highly recommend the (seemingly, costly) software-based tutoring implementation of a slightly-updated version of the book method, Funnix. Its level of interactivity resembles a DVD, but this allows you to be a facilitator instead of both tutor/facilitator. Therefore you can focus on your child's responses — and better know when to pause, repeat, take a break, etc. Additionally, you don't have to read ahead and/or prepare in advance for the slight changes in approach that come every 2nd or 3rd lesson.
When you experience the impressive results of seeing your child more logically (than you) apply rules s/he is learning about decoding English, your first question will be why these techniques are not standard practice. Especially, if you, like me, have tried other methods, including more popular best-selling children's reading software.
But as a slashdotter, you might realize most people are not open to weird, compulsively deconstructed, rule-based, fully-scripted (e.g. robotic) "instruction". Where slashdotters see the potential to scale up such a consistent, teacher-independent, assembly-line approach to instruction, there are just as many non-slashdotters who would fight the "McDonalds-ization" of the education system.
Until that changes, we at least have the method available, as either "Teach your Child to Read" in book form, or "Funnix" in software tutor form, to use at home.
During the summer, every restaurant that can, puts tables/chairs outside because the swiss strongly favor it.
except the one that links directly to your article about the best method for teaching kids to read. (Which apparently doesn't seem to exist.)
BTW, The article about the best method for teaching kids to read is Chapter 7 of "Super Crunchers" and its footnotes which link several academic data-driven field studies over the course of 40 years that back this up. e.g. 70'000(?) kids over the 10 year(?) "Project Follow Through" project. This book chapter is not online - I unfortunately can't link to it. I'm also surprised nobody has done a slashdot book review on "Super Crunchers" in general. It is basically practical case studies in data mining.