Japanese Kana Dakuten & Handakuten Cheatsheet

Japanese marks voicing with two small diacritics placed at the top right of a kana. The dakuten, two short strokes that look like a quotation mark, turns an unvoiced consonant into its voiced pair: か becomes が, さ becomes ざ, た becomes だ, は becomes ば. The handakuten, a small circle, applies only to the h-row and produces the p sounds: は becomes ぱ. Together these two marks generate a large part of the kana inventory without adding new base shapes.

The mapping is systematic, which makes it easy to learn. K becomes g, s becomes z, t becomes d, and h becomes b with the dakuten or p with the handakuten. A few entries in the table deserve a note: じ and ぢ are both pronounced ji in modern standard Japanese, and ず and づ are both zu. They are not interchangeable in spelling. Modern orthography uses じ and ず by default, reserving ぢ and づ for cases where the sound arises from a compound or a repeated syllable, as in はなぢ (nosebleed, from はな plus ち) and つづく (to continue). Choosing the wrong one is a spelling error that Japanese readers notice.

There is a Unicode point worth understanding if you handle Japanese data. Voiced kana exist both as single precomposed characters and as a base kana followed by a combining dakuten. が can be one character (U+304C) or two (U+304B plus U+3099). They look identical and are different byte sequences, so string comparison and search will fail across the two representations. Unicode normalisation form NFC composes them into the single-character form, and running Japanese text through NFC is the standard fix. The standalone ゛ and ゜ at the end of the table are the spacing versions of these marks, used when the mark is discussed on its own rather than applied to a kana.

Half-width katakana adds another layer to the same problem. In half-width form the dakuten is always a separate character, so ガ is stored as two half-width characters. Half-width katakana appears in older systems, bank statements and some point-of-sale data, and normalising it to full-width is usually necessary before matching against normal text.

For typing, an IME handles all of this transparently: type ga and you get が, type pa and you get ぱ. You will rarely enter the marks separately. Where the table helps is editing text outside a Japanese input environment, or fixing a document where the composed and decomposed forms have got mixed. On mobile Japanese keyboards, the flick input includes a dedicated key that cycles a kana through its plain, dakuten and handakuten forms, which is why Japanese mobile typing needs so few keys.