Memories of Japanese Input: Appeal to Readers

I wonder whether Nobumi Iyanaga might know:

http://www.bekkoame.ne.jp/~n-iyanag/index.html
http://www.bekkoame.ne.jp/~n-iyanag/researchTools/researchtoolsindex.html

It’s still possible, although extremely cumbersome, to use a similar method on a Mac now. You just click on the input method label (shown near the top right of the screen) and choose Character Viewer from the drop-down menu. You can then choose Hanzi (Simplified), Hanzi (Traditional), or Kanji, which allow you to look up characters ordered by radical and number of strokes. (You can also access lots of other things like Greek, and Mongolian, etc. I seem to remember that Thai, Burmese, Hebrew, Devanagari and many others could also be accessed, but strangely enough they’re missing from my system at the moment.)

I really don’t remember the input system that Helen mentions, although that’s possibly because I never saw it as having been phased out. Instead, I think I saw it as somehow “seguing” into the current method.

However, I know that it was always possible to look up characters using this method on a Mac. That is because when I was using a Mac in China in the late 90s, some Chinese staff said that since the character 町 doesn’t exist in Chinese, they had problems printing text with Japanese place names in it. They then proudly showed me how they’d managed to cobble together a makeshift 町 to fill the gap. It took me less than five minutes to locate the character Simplified Chinese character 町 in the Character Viewer, based on using radicals, not pronunciation.

Of course, I’m not sure whether this is the same as the input system that Helen describes. My feeling is that it was simpler in those days to look up characters by this method. The Character Viewer is pretty slow on my computer, and while it allows you to access a huge range of Unicode characters, I have a vague recollection that the old system was divided up by language. That is, when accessing Japanese characters, you were only shown kanji, when showing Chinese characters, you were only shown Hanzi, etc. The current system is more comprehensive but less convenient for the user.

I’m afraid I can’t remember more than that. Memory plays tricks on you when things are constantly changing. (I’m now trying to figure out why I used to be able to see Thai, Hebrew, Burmese, etc. on my Character Viewer — or at least I think I could — but can’t see them now.)

On my Mac purchased a few years ago I don’t see “Character Viewer” either under my input menu or when I search under the magnifying glass.

Keyboard preferences, keyboard tab. There’s a checkbox there to enable access in the input menu to the keyboard viewer and the character viewer (“Emoji & Symbols” since OS X 10.11).
If you type IPA, you will enjoy having the IPA Palette from blugs.com.

The reason that the hanzi in OS 9 were segregated by language is that they belonged to separate and incompatible non-Unicode encodings, so they only worked in applications that could handle files with multiple encodings, or using the ISO 2022 meta-encoding (of which ISO-2022-JP is a special case).

IICORE is a collection of about 10,000 widely used characters intended to bridge CJK, so that systems which implement it can do a good enough job. Predictably, it’s only Chinese systems that take this seriously.

Try System Preferences/Language and text/Input sources. There’s a checkbox at the very top of the input sources list. Then go and type something in Syriac or Linear A. I dare you.

Y, it’s clear that what you are proposing would require a system update for me (10.8.5), which I’m reluctant to do since I can do what I need to do now and am not sure it will be clear how to do it afterwards. But for Helen DeWitt, it’s probably what is needed.

I reread Y’s comment and was able to turn on Character Viewer and Keyboard Viewer. Character Viewer always only gives “Arrows”, “Parentheses”, “Punctuation”, “Currency Symbols”, “Pictographs” (not Kanji), “Bullets/Stars”, “Math Symbols”, “Letterlike Symbols” (no Kanji), “Emoji” and “Latin” (this is even with Traditional Chinese switched on and a Chinese input method selected). However, with Traditional Chinese switched on and Cangjie or Sucheng selected, Keyboard viewer gives me radicals. I don’t know how to do the rest. I also get radicals in the Keyboard Viewer with Simplified Chinese on and Wubi Xing selected. My computer did not come with Syriac or Linear A, I’m certain.

It seems to me that only someone who had never personally tried writing a text in Japanese (or, for that matter, Chinese) would replace the features we used to have with one which presupposed you always knew how to pronounce what you wanted to write – an interesting (if exasperating) mistake to make.

I think this is a bit unfair. The decision to change the Japanese text input method was almost certainly made at least in part by someone who personally had a great deal of experience writing text in Japanese — that is, a Japanese person. The fact is that when you are writing a text in Japanese, the vast majority of the time you do know how to pronounce what you want to write. There are obvious exceptions (transcribing names that are new to you, etc.) but they are very rare. So it makes sense to prioritize sound-based input once you have a system that can do it well enough.

(If they really did remove the radical search entirely, that I would agree deserves vigorous protest, but I can confirm the other reports here that it is in recent versions of OS X at least and I strongly suspect that it’s been in there from the beginning — perhaps tucked away somewhere too difficult to find, but that’s another kind of mistake.)

(1) Begging our host’s indulgence…
To Helen DeWitt, if you’re reading this, let me just say how much I enjoyed The Last Samurai, which I found out about through languagehat (and read while in Japan, as it happens). Can there be any other writer who has succeeded not only in combining Homeric Greek and Japanese in the same book, but making them both smoothly integral to the plot? So many fond passages… No need for any reply, only know that whatever you’re working on, I look forward to it.

(2) The sort of thing it would interest me to know is whether it could also be used in Word or WordPerfect.

Fortunately, the answer is yes. I’m not a Mac user, so can’t competently answer your questions about that, but if you use the standard Microsoft IME, absolutely, and it should work throughout Windows without regard to application. Just right click the spot on the language bar that toggles between “A” and あ, then select the “IME Pad” option. The small window that comes up will open default to a drawing utility (which I highly recommend over the radical-lookup method), but if you select the “部” (for bushu部首, “radical”) option on the window’s left-edge side, you should have what you’re looking for.

(3) It seems to me that only someone who had never personally tried writing a text in Japanese (or, for that matter, Chinese) would replace the features we used to have with one which presupposed you always knew how to pronounce what you wanted to write – an interesting (if exasperating) mistake to make.

Matt is of course completely right about this. I’d even go further: I’m sure there are hobbyist adepts out there, but for the average Japanese user, the radical-input method has got to be among the worst (and surely least-used) input-methods available. This is because Japanese users are almost always trying to write not characters, but words, words they already know. The vocabulary used to discuss input methods reflects this. When you type, you’re not so much “writing” characters as “converting (into characters)” (henkan変換). “Autocorrect” is accordingly called “predictive conversion” (yosoku henkan 予測変換). And “misconversions” (henkan-misu (E. miss) 変換ミス) are an amazingly common species of typo (and infuriatingly hard to spot in your own writing).

Likewise, when you can’t produce the character you want, it’s not that you can’t write the character, only that you can’t “get it to come up (lit. out)” (dasenai出せない). This usually means that the speaker has either learned/guessed the pronunciation wrong in the first place, or is drawing a blank trying to think of a word that contains the desired character (and thus can’t use the most common method for inputting single characters–typing words that contain them and deleting the extra kana or kanji). But as Matt said, this doesn’t regularly happen with literate speakers writing about topics they’re familiar with. Even in the most common case of input difficulty, that of personal and place names containing rare characters (or rarer forms of common characters), a host of frequently updated, internet-connected input programs are making things easier all the time. And when you do hit a wall, there’s always Google/Wikipedia and cut-and-paste.

(4) But what about the <3-5% of cases where even (gasp) Google-san fails us? This is where I can see the radical-lookup method coming in handy even for the native speaker, at least 5-10 years ago. Nowadays it's much quicker–especially with touch screens–to use the "drawing pad" option most input systems provide. This is my own go-to when all else fails. From personal experience at least, though I do a lot of ridiculously specialized J-word-processing, involving transcripts of materials with either (a) non-standard character forms or (b) characters that have little or no citizenship outside Chinese-Chinese texts, I don't think I've used the radical-lookup method twice in two months. Yet I have to admit: part of it's just laziness. Upping my radical-parsing/stroke-counting game would occasionally make things faster, but (to my shame) I'd rather google up the Shijing for the thousandth time and copy-paste from it than slog though the radical lists for something like 簋.

Thanks everybody! And especially Steve H. – I did in fact manage to find the radical listings buried deep in Special Characters. So I’m glad it is not gone after all. With regard to typing something whose pronunciation you don’t know, I can certainly believe this is more common for users who are not native Japanese speakers – but the old version was convenient for anyone who was not a native speaker. (The radical listing simply came up as one of the input options.) But I wonder whether even a Japanese speaker might not sometimes find him- or herself wanting to copy a passage from a book, say, in which not all the characters were known – perhaps one from an older text, for example.

For Chinese it is even more important to have an alternative to phonetic input methods. A method of entering handwriting is obviously one of the best solutions, and I’ve seen a number of Chinese religiously using handwritten input, but constituent-based methods (mostly using radicals) also have a place. Some of the fastest but most laborious-to-learn Chinese input methods are based precisely on inputting the components of the character.

Be that as it may, as Victor Mair pointed out recently on Language Log (sorry, I can’t locate it in the waterfall of somewhat unfocussed posts in LL recently), something like 99% of Chinese texters input using pinyin.

With regard to typing something whose pronunciation you don’t know, I can certainly believe this is more common for users who are not native Japanese speakers – but the old version was convenient for anyone who was not a native speaker.

No argument there! Apple in particular is very big on streamlining for the most common use case rather than putting a bunch of options up front and center, though. I suppose they decided that it was more important to prioritize native speakers over non-.

But I wonder whether even a Japanese speaker might not sometimes find him- or herself wanting to copy a passage from a book, say, in which not all the characters were known – perhaps one from an older text, for example.

Well, it’s not unheard of, but it is rarer than you might think, because of course what people type isn’t determined for them randomly. If a passage from a book is meaningful enough for someone to copy out, then they probably know how to pronounce it. If it’s their job to type up things that aren’t meaningful to them, then as a job skill they probably know enough to at least guess at virtually everything thrown at them, and so on. (And of course ever since Google became a thing, as Elessorn says, it’s almost always easier just to find the characters online and copy+paste.)

Looking forward to the new story!

(henkan-misu (E. miss) 変換ミス)

I’ve long harbored a suspicion that this misu is actually a shortening of misteeku rather than a borrowing of E. miss directly, but never found a good way to prove it. (There are definitely some instances of misu that do seem to be from E. miss directly, though, e.g. verbified with -suru and used in contexts like “miss the train”… and of course there’s the verifiably separate loan of the honorific “Miss”, used as a nominal element sometimes, e.g. in orudo misu “old miss” = “old maid”.)

Type some hiragana, hit space, and it’ll convert to kanji.

Or it won’t. No matter how hard I may try to input だに to get 壁蝨, I just can’t (I’m talking Windows here, don’t know about Mac). Even typing かべ and しらみ doesn’t help: the result is 壁虱.

But I wonder whether even a Japanese speaker might not sometimes find him- or herself wanting to copy a passage from a book, say, in which not all the characters were known – perhaps one from an older text, for example.

I’m not a Japanese speaker, but a few years ago there was a project of digitizing a Japanese-Russian (paper( dictionary. The conversion problem kept cropping up incessantly, and the only way out was to draw characters by hand. I wish there was a way to input them by combining elements, like 犭and 美 for きょん (which I can’t find anywhere else, not even in the Unihan database).

壁蝨 comes up fine on a Mac. In fact, 壁虱 doesn’t even come up as a choice.

I’m not familiar with 犭+ 美. I did manage to find 猐 (in Chinese) but I assume that’s not what you’re looking for. If, as I suspect,犭+ 美 is not in Unicode, then you’ll have to create the glyph yourself. I’m not sure how this is done but I believe it’s possible.

I’ve long harbored a suspicion that this misu is actually a shortening of misteeku rather than a borrowing of E. miss directly, but never found a good way to prove it.

Or it could be more subtle than that: the ‘miss’ in ‘miss’ and the ‘mis-’ in ‘mistake’ could have reinforced each other or (to put it another way) have become subtly confused — the meanings are quite similar.

Interesting! The “mistake” possibility never occurred to me. I’ve always thought it was from miss as in hit/miss, slotting smoothly into the Japanese atari 当たり / hazure 外れ mind-groove, so to speak. (Even had the occasional Lakoffian musing that it hinted at an external, “target-rightness” concept-metaphor different from the internal, “essence-rightness” concept-metaphor familiar from English, but that’s neither here nor there.) Still, I bet you guys are right that it’s more complicated than that. Definitely could be overdetermined, with “miss” (v. “hit”), “mistake”, and the “mis-” prefix all coalescing. This might help explain why, say, misu-riidoミスリード and misu-macchi ミスマッチ work slightly differently in Japanese.

“target-rightness” concept-metaphor

Well, hit is not reconstructible beyond Germanic, but it seems to start out life as ‘meet’. Hit it off, hit upon ‘discover’, hit on reflect much older states of meaning, and the first at least is 17C. Then from the idea of a missile (perhaps more likely a stone than an arrow, but I can see either one) it becomes ‘strike a target’ and then just ‘strike’ in general. Outside English it goes a different way semantically and becomes ‘find’ in the Scandinavian languages.

Miss has a PIE root *mei- ‘change, go move, send’, but always seems to mean ‘lack’ in Germanic, both as a separate word and as the native and borrowed prefixes mis-, which have converged semantically. (One of the characteristic errors made by other Germanics in writing English is to try to use miss directly where ‘lack’ is meant rather than something like not have, since lack is too formal and archaic.) In a few French words like mischief, miscreant it represents the other PIE *mei- that leads to minus and its friends.

(Life in the hospital still involves being awakened at 3 AM, though not as often as used to be the case. But at least now there is the Internet, and the potential comfort of friends in other time zones or time-delayed. There is also the new, or newish, danger of the IV line getting imbrangled with the mouse cord, however.)

the secret of codepoint 2AEC4

A remarkable find! (I was going to look in my volume of CJVK language processing but it’s off the bookshelves and ready to be packed for my move back to Oz.)

I was able to emulate KanjiTalk 7 and it looks like you can do a radical search by pressing Option-Shift-B to show the character palette (文字パレット表示) and then selecting the table to look up kanji by radical (漢字表（部首引き）).

Does this look similar to what you remember?