...Notice that the Koreans arrange their phonetic symbols in a standard square format like hanzi. Perhaps they realized long ago that this makes for faster reading.
I know considerably more than the average about vision and very little about languages, but this speed for square array may be due to the fact that the foveae (in humans, but not most birds of prey) is round and the fact that approximately 80 or more percent of your reading time is not spent seeing any symbols. (Amount depends on the nature of the material being read and how familiar you are with the symbols.)
Most of the time you are moving your direction of gaze (Making saccades) and then perception is suppressed to avoid making everything be a motion blur. It is interesting to look alternatively at your eyes close up in your mirror image - you see one, then the other, but never any eye in angular motion.
The typical line of text you are reading now utilizes only a small part of the fovea due to geometric miss match between the line and the high resolution circular fovea.
It is amazing how much faster (in some cases, faster than the framing rates available in the fastest refresh computer monitors available)* one can read with full comprehension and few errors IF you do not need to move your eyes. I.e. small sections of the text are rapidly displayed at the same location on the display screen (No time wasted in saccades).
It would be interesting to test this suggested reason, if possible, by converting the symbols of the Korean symbols from their standard square format
hanzi into a horizontal (or vertical) linear format. I predict that if this is done, the speed advantage will disappear. (Probably be slower, until the reader becomes very familiar with the new format.
BTW all orientations for these lines and directions of reading are essentually equally good. In some poor countries when books were very scarce, four students were taught to read from the same book. If you learn to read "upside down" (and right to left), or "sideways" you will do so for the rest of your life, but not be significantly slower reading that way.
What the Koreans seem to have discovered is how to match the information rich square symbol to the fovea better, if I understood what you are stating about their symbol's geometric arrangement for text.
-----------------
*Each text frame was followed by a blank frame. - I forget why. Some subjects still had no increase in error rate etc. at fast rate of display possible. (The monitors were too slow in refresh rate to determine how fast they could read.) Patentable idea: Make this display in electronic book machine and you could read "gone with the wind" in about 5 or 10 minutes, I think.
It is also true that you can hear with full comprehension much faster than anyone can speak. This too is only demonstrable with computer pre-processing of the speech. I.e. much of normal speech has the vowels sounds of much longer duration than required. (In many words they can essentually be absent -only a tiny fraction of normal duration, if present at all.) Even in normal speech there is no separation between the words. You insert that perceived "temporal separation" if it is your "native tongue." You will easily recognize this fact when listening to a foreign language you have not heard before. - If you have heard it and it is related to yours, you will still be able to parse it into words (assuming it has words) that you do not understand at all. Thus, with computers you can also start the sounds of the "next word" before the "current word" is even half verbalized and that is not perceived as two overlapped words - Your perception is still of distinct separated words.
Sorry - this is all from memory. No references, but much of the Rapid Sequential Display, (RSD or RSR it was called, I think, if you want to search) reading work was first done at Johns Hopkins Hospital (Wilmer Research Center, as I recall) while I was working with doctors there on other technical problems they had.