So I was going over some Japanese sites and I've begun to notice that certain Katakana phrases are coming to me as I see them rather than as I sound them out or read each individual kana because the 'look' of the word is being recognized by my brain or something, kind of like in English how you glaze over words and get the idea of the paragraph. It's really amazing and makes me feel warm and fuzzy inside when it happens.

I was just wondering, for you Japanese language successes, how long was it until you noticed this happened? When you did Sentence SRS, did this start happening to you with the words in those sentences too?

I guess you mean reading by the shape of the word rather than the actual characters. That will depend on how often you see the word. Obviously the more you read the better your reading will get, so if you want it to happen faster, read more.

Sentence SRSing will help very little with this. You'll be able to read them very fluidly when in the specific context of the sentence you SRS'd, but not so much outside of it.

This skill will develop through a lot of reading. Just read more and more, and it will get easier and easier.

This topic makes me curious about something. Eventually, any advanced reader has to start to recognize whole words by shape, due to the physical limitations of the human eye. There's only a small area in your center of vision where you can see fine details (the fovea), so you need to move your eyes in order to take in the details of a larger region. However, your brain is wired to halt visual processing during saccades (eye movements) and resume during fixations (pauses).

So the average English reader recognizes words mostly by shape, and moves rapidly across the page by jumping from word to word, fixating on one spot on the word to get its shape, then saccading across whitespaces to fixate on the next word.

So that being the case, how do the eyes of Japanese readers, who aren't cued by spaces, fixate and saccade?

Well, I think that's where Kanji and particles comes in, right? I always thought Kanji was Japan's way of visual cues where spaces are the English way.

Mushi Wrote:This topic makes me curious about something. Eventually, any advanced reader has to start to recognize whole words by shape, due to the physical limitations of the human eye. There's only a small area in your center of vision where you can see fine details (the fovea), so you need to move your eyes in order to take in the details of a larger region. However, your brain is wired to halt visual processing during saccades (eye movements) and resume during fixations (pauses).

So the average English reader recognizes words mostly by shape, and moves rapidly across the page by jumping from word to word, fixating on one spot on the word to get its shape, then saccading across whitespaces to fixate on the next word.

So that being the case, how do the eyes of Japanese readers, who aren't cued by spaces, fixate and saccade?

zachandhobbes Wrote:Well, I think that's where Kanji and particles comes in, right? I always thought Kanji was Japan's way of visual cues where spaces are the English way.

Yep. Mostly kanji are used to give the reader visual cues of word separation and the like. This is why Pokemon games (which are all in hiragana) are written with spaces, and why children's books that don't use spaces are so hard to read.

Although I always wonder about older Western writing system that had no spaces or any sort of different character sets, like certain scripts in Greek or Latin. Hebrew was written with no spaces and only the consonants were shown. Were readers able to process those as quickly as we can today? Unfortunately it's hard to ever prove this because there are no native speakers of any of those languages, but it's an interesting question.

zachandhobbes Wrote:I always thought Kanji was Japan's way of visual cues where spaces are the English way.

Yeah, and wars are God's way of teaching Americans geography.
(Sorry, feeling lighthearted today.)

I agree that this is how comprehension works, but a potential problem with this method is that it probably requires fixation on the particle to determine that it is, indeed a particle. I wish I had some eye motion studies on Japanese readers, but everything I know is based on English readers.

Mushi Wrote:Eventually, any advanced reader has to start to recognize whole words by shape, due to the physical limitations of the human eye.

Actually most researchers in this field would disagree with you. The word-shape theory of reading, also called "bouma shapes" was discredited in this quite fascinating paper:
http://www.microsoft.com/typography/ctfo...ition.aspx

The theory that reading psychologists seems to favor now is called "parallel letter recognition."

However, this is based on studies that all use the Roman alphabet. I would love to find out if there are any reading studies on Japanese reading, or research comparing Roman to Japanese reading process.

DavidZ Wrote:The word-shape theory of reading, also called "bouma shapes" was discredited in this quite fascinating paper:
http://www.microsoft.com/typography/ctfo...ition.aspx

That paper is awesome. So what does this mean for kanji i wonder? Are kanji on the same level as english letters? Are we recognizing multiple kanji simultaneously when reading? Or are primitives/radicals on the level of english letters, and kanji on the level of english words? Fascinating.

I doubt primitives/radicals are on the level of english letters since they don't actually amount to anything.

zigmonty Wrote:That paper is awesome. So what does this mean for kanji i wonder? Are kanji on the same level as english letters? Are we recognizing multiple kanji simultaneously when reading? Or are primitives/radicals on the level of english letters, and kanji on the level of english words? Fascinating.

I agree, this really a very, very interesting paper. The first two thirds of the paper describes in very good detail the theory *I* was taught, and I wished they spent more time explaining their neural network theory. It was the penultimate paragraph before the neutral network section, when they briefly go over why they feel that the word shape theory is refuted, which I thought made it worth the price of admission.

I also wonder how all that applies to Japanese. It might be interesting to study Korean Hangul readers as well, since that that writing system has a block construction somewhat similar to kanji primitives, except with precise phonetic meanings.