2009年2月7日土曜日

How I use Anki

Anki is an awesome program. It basically takes memory out of the equation when learning. The problem is that there's a bit of a learning curve to using it. Anki is unique from other spaced repitition programs because it uses models and distinguishes between a 'fact' and a 'card' for data organization. Many cards can be generated from a single fact. Then there's also the question on how to use it effectively, what you should put it, how you should put data in, etc. Here I'll start off with how I've setup my models and card generation in Anki and move to how I put data in it.

Models and cards
Starting of with my Japanese, I've created a model to put in sentences:

Japanese Sentence Recognition:
%(Expression)
A:
%(Reading)s
%(Context)s [font color set to white in display properties]
%(Japanese notes)s

In my normal everyday Japanese sentence model, it quizzes me from the sentence to the readings. In theory I only need the "Expression" field and "Reading" field. However, I use the context and notes field in case I forget something. Context is there to remind me of the context and situation of where the sentence was, including the surrounding sentences, while Japanese notes are where I put definitions from my dictionary.

When I want to remember things other than Japanese, I use a Knowledge model and created some card models. In the first model, Question Quizzing, I take a sentence using cloze deletion and quiz on the term I want to remember in the Info answers field. In this model, the image will show what is being defined to aid memory.

Question Quizzing:
Q:
Answer the question or fill in the blank:
%(Info question)s
%(Images)s
A:
%(Info answer)s
%(Anecdotes & Notes)s
Source & Context:
%(Source & Context)s

Here you can see the result of this:


In the opposite model, Definition Quizzing, it shows the word and I have to define it. The image is showed on the opposite side, because that would give it away. Sometimes, like when I'm quizzed on remembering things like what a Methyl group (CH3) is, I'll draw out the picture.

Definition Quizzing:
Q:
Define or explain:
%(Info answer)s
A:
%(Info question)s
%(Images)s
%(Anecdotes & Notes)s
Source & Context:
%(Source & Context)s

Result:


Card creation
This is all good and well, but probably more important than this is how I convert text into cards.



"Protons, neutrons, and a host of other exotic particles are now known to be composed of six different varieties of particles called quarks, which have been given the names of up, down, strange, charm, bottom, and top."

This is a perfect example because it shows the source, an image, a simple definition, and the creation of both cards. You can see how the original sentence was long and weighty, not suitable for card creation. Here's the thought process of how I trimmed it down:

", which have been given the names of up, down, strange, charm, bottom, and top." - Too much information. I don't want to have to remember what a quark is, AND remember a six item list to associate with it. I also won't make a separate card for this, as I don't know what they are. I'd just be memorizing useless data without learning about them.

"and a host of other exotic particles" - 'Etc.' will do much better until I learn about these exotic particles.

" are now known to be" - No shit? Delete.

Each card has a single purpose. Make sure you always know what that purpose is.

Selecting information
Now, getting going to an even lower (or higher?) level, selecting information to make cards out of. In textbooks, often information will be built on what was presented earlier. For example, in my Biology textbook it started with a basic introduction to atoms, protons, electrons, neutrons, covalent bonds, etc. The next chapter it went into more detail and presented hydrocarbons, carbon skeletons, molecules, isomers, etc. At the end it presented a chart showing six functional groups and their structers which I memorized (Hydroxide - -OH, Carboxyl - carbonyl with hydroxide, that sort of thing).

As I gradually learned and understood the above concepts presented, the more important it became to have mastered the previous material. Recently I learned that a fatty acid is a carboxylic acid with a carbon skeleton attached to it. That a tryacylglycerol has three fatty acids esterified to a glycerol molecule.

Unlike my classmates, who had to look up these terms or forfeit understanding, I understood it immediatly. I don't need to tell you how much more effective it is for learning to have a concrete understanding without the trouble of memory.

Here are some tips on how to convert data to cards

-Be selective:
From my Physics book:
"A new era in physics, usually referred to as modern physics, began near the end of the 19th century."

Is this a fact you could make a card out of? Sure. Am I reading a physics book for a history lesson? No.

I used to SRS every little fact that came along. But it was useless. I don't need to know a term's Latin word it came from. I was spending too much time reviewing useless information when I could have been spending time learning new things that are important.

-SRS hard facts, not conclusions:
This rule is a continuation from the last, but not so clear cut. For example:
"The existence of neutrons was verified conclusively in 1932. A neutron has no charge and a mass that is about equal to that of a proton. One of its primary purposes is to act as a “glue” that holds the nucleus together. If neutrons were not present in the nucleus, the repulsive force between the positively charged particles would cause the nucleus to come apart."

The only piece of information in this that was new to me was how neutrons prevent the protons from coming apart in a nucleus. However, since I am already very familiar with atom's protons/neutrons and their nuclear charges, this immediately made sense to me. I debated putting it in, but I really don't think I'll forget this for a long time. It's a result of facts I already have memorized.

3 件のコメント:

  1. so, do you always do reading recognition for cards for sentences? Do you ever go kana to kanji or anything? I've been wondering about that for a while. I want to do the regular recognition cards, but...I don't want to create any gaps in my writing ability.

    返信削除
  2. I only use recognition. The way I see it is, I can can either have 6000 unique Japanese sentences, or 3000 unique Japanese sentences. The extra time an effort isn't worth it. I'll deal with my writing ability after I'm really good at Japanese.

    返信削除
  3. So, you just read the sentences aloud without copying them down or anything? I've done dictation cards up to now, and admittedly, it takes quite a while to get through, and they are tiring. Listening to the audio carefully over and over, writing everything out, making sure the Kanji are written correctly and hearing and writing the right particles (sometimes it's very difficult to hear the difference because Text-to-Speech isn't perfect by any stretch). Then essentially shadowing the audio that plays and getting the speed, intonation, etc. down. This is all a lot of work for each rep as you can imagine. And, to get through 100 of these usually takes about two hours. And it makes me wonder how people can even get through 100 reps a day and still have time to do other things.

    Granted, my writing ability with the Kanji has increased quite a bit with the ones that are in my deck, and I'm sure my listening ability is much better than it would be. Not to mention by shadowing the audio, I'm likely to be building good pronunciation and intonation, and wherever Misaki (text-to-speech) fails me, I'm certain listening to news, dramas, and music in Japanese will cover the gaps. And, I haven't done this long enough to really say, but I think in the long run I'll have a better connection to sounds and meaning, especially since I use pictures with the audio.

    But all of this kind of contradicts the idea of "quantity of reps" and getting to the 10,000 goal requires a lot more work. Not to mention the time it takes for me to learn new words using a monolingual dictionary (I'm still new to it).

    However, I am giving myself much more time to finish the goal of 10,000 (about two more years from now). So, I think I can make the sacrifice here for a higher quality of repetition (that includes listening, writing, shadowing, and reading to check the answer).

    And, I actually find that, after I've written a sentence out by hand from the audio. Once I flip it to see the answer, I read the Japanese sentence and I only need that to confirm whether I understood it correctly or not since I have such strong connections to the Kanji being used (RTK and other Sentences that use the same Kanji).

    So, if you were at all curious about how that process was different from what you're doing then I hope this helps. Otherwise, sorry for the ramble. :|

    返信削除