Game Guide
Kana Scout hides one odd katakana in a grid of identical ones. Here are the five demo pairs, why the last two are hard, and how to scan a 25 cell grid.
Kana Scout fills a grid with one katakana repeated and hides a single different glyph among them. You tap the odd one out, and the game reports how long it took you in hundredths of a second. It tells you which glyph to look for before each round, so this is not a reading test: you can play it knowing no Japanese at all.
The five free browser levels grow the grid from nine cells to twenty five and, more importantly, shrink the visual difference between the two glyphs. That second axis is what makes the last two rounds hard.
Here is the MyRin short for Kana Scout. This one is almost purely visual, so the clip conveys the difficulty of the later grids better than any description of stroke differences.
The pairs are not chosen at random. They are drawn from the sets that katakana learners genuinely confuse, which means they are pairs whose members share most of their strokes. That is exactly the property a visual search task needs: if the target differed obviously, finding it would be effortless regardless of grid size.
A three by three grid. These two glyphs share a diagonal stroke but ア has an enclosing horizontal at the top and a distinct crossbar, while イ is just two strokes. At this size and with this much difference, you should find it in about a second.
Use the round to calibrate. Nine cells is small enough to take in almost at once, so it shows you what the easy end feels like and gives you a baseline to compare your later times against.
The grid grows to twelve, and the difference collapses to a dakuten: two small diagonal marks at the upper right of the glyph. Everything else is identical.
This is a much bigger jump in difficulty than nine to twelve cells suggests, and it is the round that teaches the real lesson of the game. You are no longer looking for a different shape. You are looking for the presence of a small mark in a specific position, which means you should scan that position rather than the whole glyph. Let your eyes travel across the upper right corners of the cells and ignore the rest of the character entirely.
A four by four grid, and a difference of closure rather than addition. コ is open on the right side; ロ is a closed box. The strokes are otherwise the same.
Closure is a feature the visual system handles fairly well, so despite the larger grid this round is usually faster than level two. It also demonstrates that grid size is the weaker of the two difficulty levers: sixteen cells with a structural difference beats twelve cells with a diacritic difference nearly every time.
Twenty cells, and the classic katakana confusion. Both glyphs are two strokes, a short mark and a long sweep, and they differ in the angle and direction of those strokes: ソ leans one way, ン the other.
Orientation differences are genuinely hard for the visual system to detect in a crowded field, and unlike a diacritic there is no fixed corner to scan. This is the round where a systematic pattern matters most. Use a boustrophedon sweep: left to right across the first row, right to left across the second, left to right across the third. Reversing direction each row means you never lift your gaze back across ground you covered, which is the main way people waste time and re-scan the same cells.
A five by five grid and the dakuten problem again, now at the largest size in the demo. This is the hardest round and the times reflect it.
Combine both techniques. Scan by corner, since the difference is confined to the upper right of the glyph, and sweep in alternating directions so you cover the grid once. Do not try to read the characters. Reading engages a slower process than the feature detection you actually need, the same interference the Stroop effect exploits.
Do not focus hard. The instinct under time pressure is to stare intently at each cell in turn, which narrows your attentional window to roughly one glyph and guarantees that you search serially through all twenty five.
A soft, defocused gaze across the whole grid uses a wider window, and detecting the presence or absence of a small mark does not require high acuity. It requires that the mark falls somewhere in your field while you are attending broadly. Most players who improve at this game do so by relaxing their gaze, not by moving their eyes faster.
The second technique is to trust discomfort. Something that differs from a uniform field often registers as a faint sense of wrongness in a region before you can say what is wrong. When that happens, investigate the region instead of dismissing it and continuing the sweep. That feeling is a partial detection, and it is usually correct.
Kana Scout is a visual search task with distractors that share features with the target, which makes it a measure of selective attention rather than of eyesight. What improves with practice is the scanning strategy, the ability to attend to a diagnostic feature rather than a whole object, and the willingness to use a wide attentional window instead of a narrow one.
There is one genuine side benefit for anyone learning Japanese. Extended exposure to exactly the pairs that learners confuse, under conditions that force you to attend to the distinguishing stroke, is close to ideal practice for telling them apart. You will not learn to read from this game, but ソ and ン will stop looking alike, and that is worth something on its own.
The installed game runs the same mechanic across 600 levels, with larger grids and, at higher difficulties, differences of shade and size rather than stroke. The demo covers the two hardest kinds of difference already, so it is a fair sample of what the game asks.