Jun 11, 2026 · Learning Hiragana

The hiragana characters that look almost identical

You can get through あ, い, う without trouble. Then you hit the pairs that share strokes, and everything starts to blur. It's not random — specific characters consistently trip people up, and there's a reason for it.

← Back to blog

Why your brain groups them together

When you learn something new, your brain doesn't store it in isolation. It compares it to everything already in memory and files it alongside similar things. That's usually helpful. For hiragana, it creates a specific problem.

Characters that share visual features get stored in the same "neighborhood" of memory. That means when you try to retrieve one, you get interference from the others near it. Psychologists call this proactive interference — older memories competing with newer ones that live in the same category. The more visually similar two characters are, the harder your brain has to work to keep them separate.

This isn't a memory weakness. It's memory doing exactly what it was designed to do. The issue is that hiragana has several pairs where the visual overlap is substantial, and your brain hasn't yet built up enough exposure to each character individually to separate them cleanly.

The pairs that cause the most trouble

These are the characters people consistently report mixing up, with the specific difference that makes each pair distinct.

は (ha) and ほ (ho) — The left side of both characters is identical. The difference is entirely on the right: は has a single vertical stroke with a curve; ほ has two curves, almost like a loop. If you're mixing these up, it's because you're storing the left side and not encoding the right side precisely enough.

ぬ (nu) and め (め) — Both are loops. ぬ has a tail that hooks outward at the bottom right; め closes more completely and has a small crossing stroke inside the loop. People often describe ぬ as "messier" and め as "rounder." That informal description is actually a useful encoding cue.

わ (wa) and ね (ne) — Same curve on the left, same general proportions. ね has a loop on the right that わ doesn't have. Once you see the loop as a distinguishing feature rather than a variation, the mix-ups stop.

き (ki) and さ (sa) — Both have a horizontal stroke crossing a vertical one. き has four strokes and feels more complex; さ has three and looks simpler. き also has a distinctive small stroke above the crossing that さ doesn't have. Counting strokes actively — not just recognizing the shape — helps encode this pair correctly.

り (ri) and い (i) — These look nearly the same at a glance: two vertical strokes, both curving at the bottom. The difference is subtle. In い, both strokes are roughly the same height. In り, the right stroke is taller and has a longer sweep. When these appear in actual words, context usually makes them clear, but in isolation people still hesitate.

What doesn't work

The natural instinct when you mix up two characters is to look at them more carefully. Study the chart. Stare at は and ほ until they feel different. This doesn't produce reliable memory.

Recognition — seeing a character and feeling like you know it — and recall — seeing a character and producing the correct sound without help — are different things. You can recognize both は and ほ as familiar hiragana and still blank out on which one is which. Familiarity isn't accuracy.

What builds accurate recall is being tested on the distinction. Not looking at both characters together. Being shown one, trying to produce the correct answer, and finding out whether you were right. That moment of retrieval effort — especially when you get it wrong — is what actually encodes the difference.

The technique that works

Hunt and Worthen (2006) found that memory improves when you encode how items differ from each other, not just what they look like individually. They call it distinctiveness processing. Applied to hiragana: you need to study the similar-looking characters together, not separately.

The method is straightforward. When you get は wrong and confuse it with ほ, don't just note the correct answer and move on. Look at both characters side by side and articulate, out loud or in your head, exactly what makes them different. "は has one curve on the right. ほ has two." Name the specific visual feature. Then do a few more retrieval attempts on that pair before continuing.

This is slower than running through a chart. It's also why people who use this approach stop mixing characters up within a few sessions, while people who review the chart keep mixing them up for weeks. The effort is the point.

SimplyHiragana shows each character with a visual mnemonic that highlights its distinguishing feature, then schedules the characters you keep missing more frequently. The algorithm handles the spacing; you just do the retrieval attempts.

How long until the mix-ups stop

For most people, a few targeted sessions on the confusable pairs is enough. Not hours — a few short sessions where you're specifically drilling the pairs you mix up, with active retrieval and comparison.

The characters that feel almost identical today won't feel that way after you've successfully retrieved each one twenty times. Repetition through retrieval gradually separates them in memory. The pairs stop feeling similar and start feeling obviously different. That shift usually happens faster than people expect.

After that, reading real Japanese accelerates everything. Every word you read that contains は or ほ is another retrieval attempt. Once you move from isolated characters to actual text, the mix-ups become rare.

References

Helpful links