Having finished the first part of Heisig’s Remembering the Kanji a little while ago, I’ve had some time to think about the experience.  Not that it’s entirely over, mind you; there’s still usually a hundred or so reviews a day, and I’m still not passing 100%.  Usually about 85%.  That should go up in fairly short order as the troublesome characters get concentrated toward the front of the queue and get dealt with.

You won’t find too many people who have used RTK that will advise others against it, and with good reason.  Still, there are some significant imperfections, or for most such points one could say incompleteness, in the system.

For those new to the idea (and I imagine that won’t be many of you) RTK is, briefly, this.  The concept is to assign a single keyword to each “primitive” (similar to what is usually called a “radical”, but not identical) and use these keywords to build an imaginative story for each kanji, assigning each one its single keyword in turn, with the idea that an interesting incident or amusing image is a great deal easier to remember than a random arrangement of marks on a page.  Furthermore, the order in which Heisig has arranged the kanji goes by related primitive element, and builds on previously introduced elements, using known kanji as primitives in their own turn as they form compounds.  Reviewing is done strictly from English keyword to kanji writing, under the assumption that if the writing is known surely the recognition will be trivial.

There are a great many advantages to this system.  With the story mnemonics, acquiring new kanji is very quick and retention is excellent.  Previously I had attempted the more usual brute-force method of massive repetition, but this didn’t get me past 500 kanji at best and most of those were very quickly forgotten.  I didn’t use an SRS system though, so I was handicapped there.  (I didn’t use SRS with RTK either; I used the Leitner box system on the RTK website.)  I have heard of more than one person going from zero kanji knowledge to the full jouyou kanji in two weeks with RTK; clearly, that is a feat both of memory and extremely hard work, but I can’t see how such a thing would ever be remotely possible without the use of this type of mnemonic.

The fact that the stories rely on breaking down the kanji into its component parts is also a great benefit, not only for ease of memory, but also for ease of getting the stroke order correct, and further down the path also for learning other kanji not covered in RTK.  This gives the student a good understanding of how kanji actually work, which makes learning unfamiliar ones a much simpler process.  Instead of learning 20 strokes all you need to learn is three primitives, for example.

The order of learning is unconventional, but a great improvement on either going by Japanese school grade level or frequency of occurrence. Since the thinking is that in order to read fluently, a knowledge of only a part of the kanji isn’t terribly helpful, no matter if it’s a rather large part, Heisig instead ordered the kanji by related primitive elements.  This leads to much improved retention in early reviews, since you know that all the kanji you’ll be reviewing that day have the same element, or one of only a few elements, however many you learned at a chunk the previous day.  It almost seems like cheating, but quick retention in the early stages saves a great deal of time overall.  Moreover, because of this order, you are never dealing with more than one unfamiliar primitive element at a time.

The emphasis on writing is also a strong point, but a flawed one.  The muscle memory does greatly aid retention, especially in the medium and long term.  And writing the kanji makes very certain that you fully know it, because you can’t miss even one stroke, or get the stroke order incorrect; whereas if one attempted recognition only, missing details is rather a lot easier.  The flaw in this, though, is what has been wryly dubbed “Heisig syndrome”, in which the student finds himself in the odd position of being able to write the kanji but not readily recognize it.  This is much easier to fix than the opposite problem though, and a relatively brief time of reviewing in the opposite direction, not to mention encountering the kanji in use, will soon have this sorted.

It must also be mentioned that the website dedicated to RTK at kanji.koohii.com is a tremendous user-generated help in getting through the kanji the RTK way.

The greatest single annoyance with RTK is synonymous keywords, due to the system of single keyword per kanji.  At first, these are not an issue.  As you work your way through, more and more pairs and even triplets of keywords will begin to crop up and trip up your reviews.  At the  moment I find myself constantly writing the kanji for “wedding” instead of “marriage”, for instance, and the group of “heir”, “inherit”, and “bequeath” are another of many such trouble spots.  It seems until around maybe 1300 or 1500 characters, this won’t even register as a problem.  After that, it’s a serious nuisance, more annoying because you actually do know the kanji, the problem was in the English keywords that you eventually won’t even be using at all!  As I mentioned at the beginning, at the moment I usually fail around 15% of my reviews; I would estimate around half these failures are caused by this issue.  I can only see the problem compounding as one tackles the next thousand kanji in the third volume of RTK, too.

Another issue, which is probably not of great concern to many, is the frequent inaccuracy of meaning.  Since the system is designed to facilitate memory, not to educate an etymologist, primitive elements are often assigned an arbitrary meaning that is easy to remember.  I wish there were a fusion of RTK and the accurate etymology found in such books as The Key to Kanji.  But kanji etymology is a subject of some fascination to me, whereas most people want simply to be able to read the character and don’t much care how it developed its current form.

In sum, even with the detractions mentioned, I would whole-heartedly recommend RTK to any student.  In fact I agree with Khatzumoto that learning the meanings and writing of the kanji should be the very first thing a beginning student does, because having done that any word and any sentence is open for your learning the actual language, not just its orthography.  And once you’ve finished the first 2042 characters, you will intimately understand how kanji actually work.  Be prepared for synonymous keyword troubles though.  It is mainly because of this that I don’t plan to use RTK for any more kanji, although I’ll certainly adapt much of the method as it fits, and will likely use Heisig’s order in RTK3 as well.