Home Previous Columns Book Reviews Other Articles Reader Response Links

LEXIKAN, version 2

Reviewed by Shira Coffee

Let me introduce myself. I'm a homeschool mom from the Chicago area with a fifteen-year old daughter. Both of us have chosen to learn Japanese in the past year. Aside from a weekly tutoring session with a Japanese native speaker, we have been working largely on our own, using many computer-based tools.

We began learning kanji not quite six months ago. In that time, I have learned over 600 kanji and my daughter has learned nearly 150. I attribute much of this progress to a Windows program called Lexikan, which is our primary tool for learning new kanji.

What follows is a review of Lexikan's various features: the drawing features, the flashcard features, and the data-entry features along with some thoughts about the process of learning kanji and ways in which we have used Lexikan to make that process easier.

The most common and basic way we use Lexikan is to draw kanji correctly on the screen. Here is what the process is like:

Every Thursday morning I sit down in front of my computer to begin learning 25 new kanji using Lexikan. For each character, I get a square with dotted gridlines to draw the character in, with a list of the character's reading(s) below the drawing box. As I encounter each unfamiliar character for the first time, I first bring up a ghost or stroke-outline of the kanji. Using my mouse, I trace the strokes in the order I expect them to be made. Of course I make mistakes at first. Sometimes I mistake one stroke for two, or two for one; the program lets me know this by drawing incomplete strokes in light green and excessive ones in red. Sometimes I pick the wrong stroke order, and again, the program warns me by drawing that stroke in blue. Correct strokes are drawn in black and prettified so that when I have correctly followed the outline, I have what appears to be a printed rendition of the new kanji in front of me. I then promptly erase the character (with a single mouse-click) and draw it again, and yet again, until I am drawing it easily. Then I move on to the next character.

My daughter prefers to work a little differently. When she is first learning a new character, she clicks on a demo button. A second drawing box is opened next to the first, and the character is drawn slowly, stroke by stroke, as many times as she wants to watch it before trying it herself.

Nearly every aspect of the drawing mode is customizable. The drawing box may be blank or have one of two dotted grids superimposed. The outline of the character may be shown or not. The user can request a hint (beginning and end points of the next stroke) at any time. By default, every stroke the user makes is evaluated, feedback is offered, and the stroke is prettied up so that the final character looks perfect. However, these defaults can be overridden, in which case the program shows the unedited strokes until the user asks for feedback. In addition, the user can specify how precisely his or her strokes must match the program's model of the character in order to count as correct.

This kind of flexibility is very helpful, because in my experience, people differ in their tolerance for frustration. My daughter, for instance, is discouraged relatively easily. In order to get her to practice, I find it best to set the precision fairly low so that she is not sweating the details of every stroke. If I require too much precision, she either avoids practicing or practices fewer repetitions, and as a result, learns the characters more slowly. For myself, I tend to set the precision fairly low when I am first learning characters, but when I am reviewing well-known kanji, I find it profitable to set the precision high and turn off instant feedback that way, I catch myself in small errors that the feedback might mask.

One concern I had when I began working with Lexikan was whether the on-screen writing practice would translate into the ability to write the kanji using pen and paper. Obviously, using a mouse involves different muscles than writing with a pen, but I am happy to report that this has not been a problem. If I really know a kanji in Lexikan, I am able to reproduce it even the first time I need to write it freehand. Our Japanese tutor has never had trouble reading our kanji, and only on a couple of occasions has she offered suggestions for writing a particular character more legibly.

Besides the drawing box options, it is also possible to customize the box that contains readings and meanings. Readings can be shown in romaji or in kana. As I'll explain when I discuss the data-entry features of the program, it is possible to add or remove readings from this box, making the cues simpler or more complex. In addition, it is possible to add a user-defined comment box to the bottom of the screen. I use the comment box when I am learning a particular kanji primarily because it is in a compound that I wish to learn. Reminding myself of the compound whenever I am drawing the kanji is very helpful!

Besides the kanji drawing mode, Lexikan also has drawing tools for hiragana and katakana. These work exactly the same way the kanji drawing tool does. I mention them only briefly here, because I assume that readers of the Kanji Clinic column are primarily interested in learning kanji.

The second set of features of interest to Kanji Clinic readers involves flashcards and quizzes. There are essentially four kinds of flashcards that Lexikan can test on. First are kanji cards, which consist of a single kanji with its readings and meanings. (Also, invisibly attached to the kanji card is such information as how to draw the character, stroke count, stroke order, and reference information.) A second kind of cards hold kanji compounds, which consist of one or more kanji plus okurigana if needed, associated with a pronunciation in kana and an English definition. Third are cards for kana words, which consist entirely of kana characters and are associated with an English definition. Finally, the user can define multiple cards for kanji that function as counters, easily generating individual cards for as many numbers as the user chooses to define.

Two kinds of quizzes are provided for any kind of flashcard: fill-in-the-blank or multiple-choice. Both quiz modes display the kanji, kanji compound, kana word or counter expression (number plus counter) at the top of the screen. In fill-in-the-blank mode, the user must supply the romaji for the kana representation of the word any reading for the given kanji is accepted, but the other three card types have only one correct answer. In multiple-choice-mode, the user chooses the correct reading or meaning from four possible answers offered by the program. The quizzes can be timed or untimed, with timing values set by the user. Statistics are kept for both individual flashcards and groups of flashcards (called word groups).

The way the multiple-choice quiz works is highly customizable: it's possible to be tested on on-readings, kun-readings, or both, or on English definitions. As in other parts of the program, readings can be displayed in kana or romaji. In addition, if a kanji has verb forms, a random verb inflection can be tested instead of the simple kanji.

In the interest of being complete, I should mention here that there are similar quiz modes for hiragana and katakana, which work pretty much like the quiz modes for other flashcards. There is also a no-quiz mode similar to some other flashcard programs, in which the user tells the program whether he got the card wrong or right.

Finally, let me describe the data-entry modes that Lexikan offers. Each Lexikan flashcard belongs to a word group a group of cards that will be drilled at the same time. One or more word groups can be stored in a physical file. To begin using the kanji draw mode or any of the quiz modes (except kana quiz), a file with at least one word group containing the proper sort of flashcards has to be open and selected.

There are several ways to generate word groups of flashcards to begin working with. One way is to download preformatted files for each grade-level's kanji from the Lexikan website. That is how my daughter and I began. To supplement or replace files available on the website, a new data-entry mode for kanji has been added to Lexikan version 2. It is now possible to auto-generate lists of kanji in the order specified by grade level or by any of 10 common kanji reference texts . Anyone wanting to learn or teach kanji using one of these books should find this very useful. Personally, I don't use these references, so I can't comment further on this feature.

I began learning kanji in grade-level order, but after grade 2 I found it more convenient to select each week's kanji based on various sources: whatever manga I'm reading, songs, words our tutor gives us, etc. For that reason, I have a great deal of experience using the basic data entry features of Lexikan. Every week I add 30-35 kanji (my 25 and Jen's 5-10 per week) and at least 50 compounds to our vocabulary files. I also add kana words and counters frequently as I come across them.

As with the other features, there is a great deal of flexibility in the ways one can enter data into Lexikan. The basic utility used for data entry is the kanji-lookup dialog, which is quite convenient. I generally look up by reading, but it's also possible to look up by stroke count or by meaning. You can look up by combinations of criteria as well, which reduces the number of kanji hits that pop up from the search. It's even possible to look up a kanji by drawing it and having the program search the database for similar characters. I have been amazed at how well the handwriting recognition search works: even when I've left out a stroke or misdrawn a component, the kanji I was looking for generally popped up somewhere on the results list of ten most likely matches.

For kanji flashcards, the information that comes out of the database is quite complete. I only occasionally have to add readings to the ones supplied. It's also possible to delete readings (from the word group version of the card, not from the database), which is useful, for instance, if you have a young learner and only want to present a single reading initially.

For compound flashcards, the kanji are selected using the same kanji-search dialog as for kanji cards. Any needed okurigana are added by the user. An interesting feature, and one that I have come to appreciate, is that compounds that use the on-readings as combining forms are stored by default as katakana, while those that use kun-readings are stored as hiragana. Since I learn on-readings after the fact (by learning compounds), this is a useful mnemonic device for me. It is also another feature that is under user control you can override the defaults and store any compound's pronunciation in whichever syllabary you prefer.

One new feature (still incomplete, but growing) is a database of kanji compounds. As of the release of version 2, there are about 500 compounds in this database, all of them based on grade 1 kanji. The database will be expanded as the developers' time allows, and should eventually make the process of entering kanji compounds into vocabulary lists much simpler.

Kana words are added by specifying the pronunciation (which is translated from romaji to kana) and the meaning. Counters are specified by selecting the kanji using the kanji-search dialog and then specifying the pronunciation of some or all of the first ten numbers with that counter appended.

As I mentioned earlier, I add 35 kanji and 50 compounds from scratch, which takes about an hour each week. This doesn't count the time I spend during the week sticking words that I encounter into lists for selection on Tuesday night, when I enter the next week's vocabulary lists. Since my mind is working out strategies for the coming week's vocabulary practice while I enter the new kanji, I consider the data-entry time well-spent.

I'd like to offer two practical tips about using Lexikan. First, it is possible to mix all four kinds of cards into a single word group, but from personal experience, I advise against it. Since the preferred quiz modes are stored with the files, I try to put word groups that I want to be quizzed the same way in the same file. For that reason, I enter weekly kanji and compound word lists (one type of card in each word group) into monthly files. Kana words go in a separate file, because I always quiz them in multiple-choice, pick the English meaning mode. By storing words this way, I spend less time changing quiz options and more time doing quizzes.

Also, the size of word groups is something that requires a bit of experimentation. The smallest unit that can be drawn or quizzed is a word group (though one can easily combine several word groups for a given drawing or quiz session.) The number of kanji to be learned in a sitting depends, I think, on individual psychological factors. The smaller the word group, the more often each kanji will be presented. But the more kanji in the word group, the more common elements will turn up, allowing one to group kanji mentally as having, say, a particular component in common. For myself, 25 kanji has turned out to be an ideal number to memorize at once. My daughter prefers a much smaller number of new kanji to deal with at once. For familiar kanji, I like to review in large groups of 150 or more, while my daughter prefers to keep review groups under fifty characters. So, while data-entry concerns may seem excessively technical to folks who just want to learn kanji, in fact choosing a data-entry strategy can make learning easier. Lexikan's flexibility makes it very easy to work out an optimum individualized strategy.

I have tried in this review to outline the features of Lexikan from the perspective of my own experience. I should mention, for the sake of completeness and fairness, one other dimension of this product. Since late May, when I began learning kanji in earnest, I have been a rather active user of Lexikan. I use the program at least one hour every day, and I have not hesitated to email the developers with enhancement requests and (occasional!) bug reports. I have found the developers to be responsive to complaints and open to user suggestions. Many of my enhancement requests made it into v.2, and I was one of several users selected to beta-test the new version before its release. As a result, I can no longer say (as I did in my initial letter to Kanji Clinic) that I have no relationship to the developers of this product.

What I can say is that I have no financial incentive to push this product. I am simply a user: a fairly intensive user, and a very satisfied user. The fact that the developers have been responsive to user input is, I think, something that puts this product ahead of a many otherwise decent pieces of software developed by faceless corporations. This is a piece of software I depend on; it is nice to know that the people who wrote it, and who keep it running, are not only dependable, but responsive.

To summarize the basics: Lexikan v.2 is a Windows program requiring Windows 98 or better, 4MB of hard disk space and 32MB of RAM to run. The program (and a demo for those who wish to test it) are available from www.lexikan.com. The cost of the full program (to be labeled the Advanced Edition) is $49. There is a Standard Edition (with a smaller database containing only 881 essential kanji ) available for $39.