Spaced Exposure with SRS

Flashcards are among the most traditional and widely utilized language study tools. In my own experience they are also one of the most ineffective. Look at the front, recall translation, confirm by looking at the back, then repeat. Brute force memorization is effortful, time intensive, and boring.

To lessen the review burden of flashcards, spaced repetition algorithms were developed (Sebastian Leitner was an early pioneer). If a student has a deck of 100 cards then it is likely a certain percentage of those cards are well known and do not require constant review. These "known" cards are put into a separate box and reviewed at larger intervals. More challenging cards are kept in a "difficult" box and reviewed more frequently. Algorithms vary on spacing and categorization, but the basic principal holds throughout: flashcards are more effectively studied by spacing out reviews.

A wide array of computerized flash card systems (SRS) is available to make the utilization of spaced repetition simple and convenient. The core weakness of SRS is that it remains rooted in the realm of deliberate memorization (it is just a more efficient way to go about it). Efficiently scheduled brute force memorization is still effortful and boring; hardly an ideal way to acquire a foreign language.

An alternative paradigm is spaced exposure (SE): using SRS as a means to ensure routine exposure to interesting, comprehensible input. The purpose of SE is not to memorize or "retain" information, but rather to gradually internalize it over time. In that sense it is not optimal for time compressed study demands (i.e., exam preparation over the short term) but I have found it highly effective and pleasurable for long term study.

Making the Schedule Easy

A key element to SE is using an SRS system that has a scheduling algorithm which can be altered by the user. As an example, let us consider Anki – it's scheduling algorithm can be modified by following these instructions. The spacing between grades is different for each learner; I don't enjoy spending hours reviewing flashcards so I scheduled liberal gaps in between grades. Anki has four grades that can be given to a card: Again, Hard, Good, Easy. For "Again" I have a minimum of three days between viewings and for "Hard" it is in the neighborhood of one to two weeks. The remaining grades have much larger intervals: "Good" is half a month to one month, while "Easy" is one to two months.

Grading for Convenience

SE is about exposure to input, not about memorization or meeting expected learning benchmarks or goals. If a word is forgotten, it is forgotten. That does not necessarily mean I grade it as "hard" or "again". If I want to push a card out simply because I believe I may pick up the word over a month in the course of other exposures then I will mark the card as "Easy". If I really want to see it again soon then I will use "Again" or "Hard". If I am ambivalent then I will probably mark the card as "Good". If the card takes too much time to read or is uninteresting – even though it may have been interesting at one time – or I just feel like giving up, then I will delete it.

The central goal of an SRS is memorization, whereas the central goal of SE is exposure. The more time you spend with real, comprehensible, interesting input the more rapidly and accurately you will acquire the language.

Good Cards, Bad Cards

A good flash card is:

1. Comprehensible: at least 80% known words/patterns.
   
2. Meaningful: the real language (native composed) in a real context.
   
3. Interesting: concerning a topic that stimulates you.
   
4. Contains hints and/or visual aids if necessary: pictures, related words, explanation notes, etc.
   
5. Helps you notice something about the word, grammar, kanji, etc.
   

A bad flash card is:

1. A word and then a translation with no context.
   
2. A definition: dictionaries are not often read for pleasure.
   
3. A single kanji: characters, like words, are best learned in the context of a full sentence or phrase.
   
4. Boring: unreadable, contrived, or otherwise just not pleasurable to read.
   
5. Deleted from your deck, quickly.
   

Card Examples

The following are examples of cards I have or at one point had in my deck. What constitutes an optimal card will vary from person to person, but the above rules hold true throughout. Especially critical is comprehensibility. If you cannot at a meaningful level comprehend the card then it has little or no utility.

Unless otherwise noted the backs of all the presented cards are nothing but ruby readings.

Front: エールフランス機が緊急着陸、爆弾脅迫で　ブラジル
緊張　救急車　協力　白人

One trick is to use known kanji to assist in reading unknown kanji. If you study about Chinese characters you will discover that the majority are　形声文字 (keisei moji) characters; kanji that contain elements which give a clue to both its meaning and reading. In this sense, kanji is quite consistent and logical. It just takes repeated exposure to learn the base characters.

The first line in the above card is a news article headline which contains many words I had trouble reading, but the meanings I was reasonably certain on from the context and the kanji. All of the helper words I read and understand comfortably. Some kanji match perfectly in both lines (緊、急) while others do not but the reading is the same in that case (白>>迫　協>>脅).

Front: ローマ帝国: 定食　×携帯電話

The above is a simple card that is just a word. It violates my rule of always providing a larger context but I am interested in the word and the card helps me notice some things: (1) the correct reading of a character I have trouble with (note the use of color), and (2) another character I sometimes confuse the character in question for (帝 帯). Note the use of the "batsu" or "X".

Front: 中ソ対立とは、1960年代から表面化した中華人民共和国と旧ソ連のイデオロギー対立、軍事対立、政治対立を指す。

I can read this card completely, but the meaning of the word in red is not 100% yet. Perhaps when I read this card I will look up the meaning or search for another sentence containing the word, or perhaps not. I may lookup up the meaning of the word and put it on the back of the card along with the ruby reading for future reviews.

Front: 消費者　　消費税金

Once again, here are words without a larger context but their presentation helps me notice something so the card is useful.

Review Time

Not a great deal of time each day is spent reviewing my cards, but it does fill in lots of dead time at the computer. I find it easy to keep up with my cards because they are pleasurable to read, I can comprehend almost everything, I can read them quickly, and the gaps in between reviews are long.

Dealing with backlogs is easy because they hardly every happen, and even if 100 cards are waiting to be reviewed it does not take much concerted effort to get through them because I can read them so quickly (they are at my level). If the backlog was greater or I just felt unmotivated to do anything about it I would probably just delete or suspend a large portion of the due cards. I am more concerned with enjoying the real language than trying to measure up to artificial metrics.

I use spaced exposure because I find it pleasurable and I think it is helping to improve my acquisition of Japanese. If and when it becomes boring or tedious - as of this writing I have been doing SE for almost one year - then I will stop.