Skip to main content

Yōon: Contracted Sounds in Hiragana (きゃ, しゅ, ちょ and the Full 33)

Yōon (contracted sounds) in hiragana are digraphs formed when a small ゃ, ゅ, or ょ is attached to an i-column kana. Two written symbols fuse into a single palatalized beat, such as きゃ kya, しゅ shu, or ちょ cho.1 The 46 base hiragana plus the 25 dakuten and handakuten kana give the reader most of the chart; the 33 yōon are the final block that completes it.12

Overview

Yōon, written 拗音 in kanji and ようおん in kana, is the cover term for every hiragana digraph that bends an i-column consonant onto a following y-glide.1 Most learner-facing pages call the same thing contracted sounds or combination sounds. Linguists call it palatalization.13 All three names point at the same 33 cells.

What "yōon" means

The kanji 拗 means "twisted" or "bent," and 音 means "sound," so 拗音 reads literally as "twisted sound."1 The standard linguistic definition is "a feature of the Japanese language in which a mora is formed with an added [j] sound, i.e., palatalized."1

The reader will meet three labels for this phenomenon. Yōon is the formal Japanese term and the one the linguistics literature uses.1 Contracted sounds is the most common English pedagogical gloss.1 Combination sounds is the phrase that dominates beginner sites.3

Why this is pre-N5 mechanics, not vocabulary

Yōon is a hiragana-system feature, not a JLPT level. Official N5 vocabulary lists already assume you can decode contracted-sound kana like きょう, しゃ, and ちゅ inside everyday words.13

An obsolete sibling category called 合拗音 (gō-yōon), built on くゎ and ぐゎ, exists in historical sources but is no longer part of the standard modern inventory.1 This article covers only the modern 33-cell yōon system.

Where yōon sits among the hiragana extensions

After the 1900 script reform, the base hiragana inventory is 46 productive kana. Two further historical kana (ゐ and ゑ) are restricted to proper names.2 On top of those bases, the dakuten ゛ and handakuten ゜ marks add roughly 25 precomposed voiced and semi-voiced kana across the g-, z-, d-, b-, and p-rows.2 Yōon adds another 33 standard combinations.1

Use the companion article The Complete Hiragana Chart (Gojūon) and How to Read It to review the inventory itself, and Dakuten and Handakuten: The Voicing Marks to review the voicing system.2 Stroke Order for Hiragana: Why It Matters covers handwriting mechanics that apply to the host kana under each yōon.

Yōon is not sokuon

The small つ (sokuon, 促音) is also a half-size kana, but it marks held silence and gemination of the following consonant, not palatalization. The two conventions share a visual cue and nothing else. See The Small つ (Sokuon) and Geminate Consonants for the parallel rules.14

Why only ゃ, ゅ, ょ and only on i-column kana

Yōon's structural recipe is fixed: the second slot must be one of the three small y-kana, and the first slot must be from the i-column.13 MIT OpenCourseWare describes the same pattern as C + y + V. Here, C is the consonant of an i-column kana, y is the palatal glide, and V is the a, u, or o vowel of the small kana.3

The phonological reason is that i-column consonants already carry a high-front (palatal) coarticulation in Japanese, meaning the tongue is already near the hard palate. Attaching a /j/-glide and reopening on a, u, or o therefore produces a phonotactically licit, or sound-pattern-allowed, palatalized mora.5 No other column offers a host consonant that this recipe accepts.

The y-row itself contributes only や, ゆ, よ to the small set: there is no /yi/ or /ye/ mora in the modern y-row, which is why no fourth small kana exists in this paradigm.1

A single diagram captures the derivation.

How yōon is written

The yōon digraph is a visual unit on the page, even though it is built from two separate kana codepoints. It helps to pin down the spelling mechanics before any pronunciation work.

Size and position of ゃ, ゅ, ょ

The small kana ゃ, ゅ, ょ are roughly half the height of full-size kana and sit next to their host on the baseline in horizontal text.14 In vertical writing, the small kana sits in the upper-right of its own cell, mirroring the small つ convention.4

The small kana is never written alone in modern orthography; it always follows an i-column host.1 Each of the three has its own Unicode codepoint: U+3083 for ゃ, U+3085 for ゅ, and U+3087 for ょ.6

The official spelling rule that codifies writing 拗音 with small ゃ, ゅ, ょ is the Agency for Cultural Affairs text 文化庁「現代仮名遣い」, published as Cabinet Notice No. 1 of 1986.4 Before the 1946 現代かなづかい reforms, contracted readings were written with full-size や, ゆ, よ. The yōon reading had to be inferred from context: "In historical kana orthography, yōon were not distinguished with the smaller kana, and had to be determined by context."1

Small ゃ ゅ ょ versus full-size や ゆ よ

The size difference is the only orthographic signal that separates a one-mora palatalized syllable from a two-mora vowel sequence.1 Miss the size, miss the word.

The standard minimal pair is きょう versus きよう.1 The first, written with a small よ, is one mora of palatalization plus a long vowel and reads kyō, "today." The second, written with full-size よ, is three morae and reads kiyō, "skillful."

今日きょう東京とうきょうきます。7
"Today, I go to Tokyo."

器用きようです。1
"Skillful."

The same contrast appears in a hundred-versus-runaway-vowel pair.

ひゃくえんです。7
"It's a hundred yen."

ひやく (full-size や) would be three morae and a different word entirely. The size of the second kana is the entire signal.

Treat font size as load-bearing information

In some serif and condensed fonts the gap between よ and ょ shrinks to a few pixels. Until the size contrast is automatic, prefer reading materials at a comfortable display size; misreading the small kana as full-size rewrites the word.1

Typing the small kana on a Japanese IME

For standard yōon combinations, typing the contracted romaji directly produces the digraph on a Japanese IME (input method editor): kya yields きゃ, sha yields しゃ, cho yields ちょ, ryo yields りょ.3 The IME treats the contracted romaji as a unit and emits a host plus a small kana in a single conversion step.

To force a bare small kana on its own, type the prefix x (or lt): xya yields ゃ, xyu yields ゅ, xyo yields ょ.3 This is the route to use when documenting yōon itself, since the digraphs above will not produce a standalone small kana.

Typing kiya does not give you きゃ

A common beginner attempt is to type kiya expecting きゃ. The IME instead produces きや, two morae and the wrong word. The romaji must be entered as the contracted digraph kya; the IME does not auto-contract a C + iya sequence.3

How yōon is pronounced

Pronunciation has two parts: how long the yōon lasts, and what the consonant actually does. The length rule is short and unambiguous. The consonant rule benefits from a one-line linguistic name.

One mora, one beat

A yōon counts as one mora, the same length as a single base kana.13 MIT OpenCourseWare states the rule plainly: "These contracted syllables are always considered to be one syllable, not two."3 The small ゃ, ゅ, or ょ is not a separate beat; it bends the preceding consonant within a single mora.

This rule makes the kyō-versus-kiyō contrast a counting problem, not only a spelling one.

The mora count matters as soon as you meet pitch accent, song lyrics, or haiku scansion. NHK's 発音アクセント新辞典 (Pronunciation and Accent Dictionary) and the NINJAL corpora both treat a yōon syllable as a single mora unit when transcribing pitch-accent contours.7

Palatalization in plain language

Palatalization is the technical name for what the small kana does to the host consonant. The place of articulation stays the same: the tongue starts where it would for the bare i-column kana. A brief /j/-glide is added on release, and the mora opens onto the a, u, or o of the small kana.5

For き plus ゃ this means one beat that begins where ki begins and ends on /a/, not the two beats of "ki-ya."53 Some older textbooks call the same phenomenon "softening"; the linguistic term is palatalization.5

Historically, palatalized syllables entered Japanese phonology during the heavy borrowing of Sino-Japanese vocabulary from Middle Chinese in the Old and Early Middle Japanese periods. They brought phonotactic patterns that the native language did not have.8 The yōon inventory is, in effect, a fossil of that contact.

Voiced yōon (ぎゃ, じゃ, びゃ, ぴゃ and the rest)

The dakuten and handakuten rows form yōon by the same recipe: a voiced or semi-voiced i-column kana plus small ゃ, ゅ, or ょ.1

RecipeResultReading
ぎ + ゃぎゃgya
じ + ゃじゃja
び + ゃびゃbya
ぴ + ゃぴゃpya

The voicing rule is the same one you already learned for the base dakuten and handakuten kana, so no new mark needs to be memorized. See Dakuten and Handakuten: The Voicing Marks for the underlying transformation.2

The d-row counterpart ぢ + ゃ → ぢゃ is "theoretically possible in rendaku, but [is] nearly never used in modern kana usage."2 Modern orthography uses じゃ, じゅ, じょ in nearly all positions where ぢゃ, ぢゅ, ぢょ might historically have appeared.24

The complete 33 yōon combinations

The standard learner inventory is 33 cells: 21 unvoiced, 9 voiced, and 3 semi-voiced. The two tables below cover every combination you will meet in standard modern text.

Unvoiced yōon (21 combinations)

Seven i-column hosts (き, し, ち, に, ひ, み, り) each combine with the three small y-kana for a total of 21.

Host+ ゃ+ ゅ+ ょ
きゃ kyaきゅ kyuきょ kyo
しゃ shaしゅ shuしょ sho
ちゃ chaちゅ chuちょ cho
にゃ nyaにゅ nyuにょ nyo
ひゃ hyaひゅ hyuひょ hyo
みゃ myaみゅ myuみょ myo
りゃ ryaりゅ ryuりょ ryo

One example per row makes the inventory easier to read and remember.

京都きょうとんでいます。7
"I live in Kyoto."

社長しゃちょう会議かいぎちゅうです。7
"The president is in a meeting."

ちゃみます。7
"I drink tea."

にわ蝶々ちょうちょうがいます。7
"There are butterflies in the garden."

ひゃくえんです。7
"It's a hundred yen."

苗字みょうじおしえてください。7
"Please tell me your surname."

旅行りょこうきです。7
"I like travel."

Voiced and semi-voiced yōon (12 combinations)

Three voiced hosts (ぎ, じ, び) and one semi-voiced host (ぴ) each combine with the three small y-kana for a further 12 cells.

Host+ ゃ+ ゅ+ ょ
ぎゃ gyaぎゅ gyuぎょ gyo
じゃ jaじゅ juじょ jo
びゃ byaびゅ byuびょ byo
ぴゃ pyaぴゅ pyuぴょ pyo

Voiced yōon are no rarer in daily text than the unvoiced set. ぎゅう, じゃ, びょう, and ぴょう appear constantly in everyday vocabulary.

牛乳ぎゅうにゅういます。7
"I'll buy milk."

じゃあ、また明日あした7
"Well then, see you tomorrow."

病院びょういんはどこですか。7
"Where is the hospital?"

発表はっぴょうきました。7
"I heard the announcement."

The 33 count versus the 36 count

Some charts present 36 cells rather than 33 because they include the ぢゃ, ぢゅ, ぢょ row alongside じゃ, じゅ, じょ. The arithmetic for 33 is 7 unvoiced hosts × 3 small kana = 21, plus 3 voiced hosts (ぎ, じ, び) × 3 = 9, plus 1 semi-voiced host (ぴ) × 3 = 3. The 36 figure restores the ぢ row, which survives only in rendaku spellings such as 鼻血 (はなぢ). Otherwise, modern 現代仮名遣い orthography displaces it with じゃ, じゅ, じょ. Both counts are correct. They differ only on whether ぢゃ ぢゅ ぢょ are listed or collapsed.124

Romanization conventions: Hepburn versus Nihon-shiki

The same kana can be romanized in more than one system. The i-column irregulars are where the systems visibly diverge.

KanaHepburnNihon-shiki / Kunrei-shiki
しゃshasya
しゅshusyu
しょshosyo
ちゃchatya
ちゅchutyu
ちょchotyo
じゃjazya
じゅjuzyu
じょjozyo

Hepburn maps to English spelling intuition, which is why it dominates outside Japan and in most learner-facing materials.1 Nihon-shiki and Kunrei-shiki render the same kana with the systematic C + y + V spellings. You may see those spellings in older Japanese-government textbooks, JIS keyboard tables, and many train-station signs.1 MIT OpenCourseWare lists both side by side: しゃ is "sya (sha)," ちゃ is "tya (cha)."3

All three systems describe the same kana. The kana is canonical; the romanization is a notation choice.5

Yōon in everyday vocabulary

Yōon is not a marginal corner of the system. It saturates Sino-Japanese vocabulary, where most JLPT compound words come from. Three high-frequency examples show why.

旅行 (ryokō), travel

旅行 reads as りょこう and breaks down into three morae: りょ (ryo, a yōon, one mora), こ (ko, one mora), and う (the long-vowel marker, one mora).1 You cannot read the word correctly without the yōon mechanic.

旅行りょこうたのしいです。7
"Travel is fun."

Both characters carry on'yomi readings, readings of kanji borrowed from Chinese: 旅 ryo and 行 .8 This makes 旅行 a Sino-Japanese compound, the vocabulary type where yōon clusters. The long-vowel convention interacts with the yōon in Hepburn: ry + o + ō becomes ryokō with a macron over the second o.1

教科書 (kyōkasho), textbook

教科書 reads as きょうかしょ and breaks down into four morae with two yōon: きょ (kyo, one mora), う (long-vowel, one mora), か (ka, one mora), and しょ (sho, one mora).1

教科書きょうかしょみます。7
"I read the textbook."

The compound 教 (kyō) + 科 (ka) + 書 (sho) is a pattern-perfect on'yomi word, with palatalization in the first and third characters.8 If you know the 33 yōon, you can decode the whole word from the kana spelling without consulting a dictionary.

学校 is sokuon, not yōon

学校 (がっこう, gakkō, "school") is the canonical confusion case: its small kana is the small つ that marks sokuon, not one of the small y-kana. The word does not contain a yōon and should not be used to demonstrate yōon mechanics. See The Small つ (Sokuon) and Geminate Consonants for the parallel convention.14

写真 (shashin), photograph

写真 reads as しゃしん in three morae: しゃ (sha, a yōon, one mora), し (shi, one mora), and ん (the syllabic nasal, one mora).13

写真しゃしんります。7
"I take a photograph."

Both characters carry on'yomi readings: 写 sha and 真 shin.8 The yōon sits in the first character. The second is plain consonant-plus-vowel plus syllabic nasal.

Why yōon is everywhere in kanji vocabulary

The historical answer is that yōon entered Japanese through the Sino-Japanese loan stratum borrowed from Middle Chinese. Frellesvig observes that "the number of loanwords was considerable and brought new phonological phenomena to the language such as palatalization."8 The native (wago, kun'yomi) vocabulary largely lacks palatalized syllables. On'yomi compounds saturate them.18

For a pre-N5 reader, the practical result is that learning yōon now makes much JLPT vocabulary easier to read across all levels. 旅行, 写真, 教科書, 自由 (jiyū), 宿題 (shukudai), and 社会 (shakai) are not edge cases. They are the typical shape of Sino-Japanese compounds.8

宿題しゅくだいをします。7
"I do homework."

Good to know

Reading a small kana as full-size

The single most damaging mistake is reading a small ょ as a full-size よ. That converts a one-mora palatalized syllable into a two-mora vowel sequence and changes the word.13 The classic case is きょう, two morae and "today," misread as ki-yo-u, three morae and "skillful." The correct reading is:

今日きょう1
"today"

In some fonts the size difference between よ and ょ is subtle, so until the contrast registers automatically, prefer a comfortable display size and zoom in when in doubt.

Typing "kiya" expecting きゃ on an IME

Typing kiya on a Japanese IME produces きや, two morae and the wrong word. The correct input for きゃ is the contracted digraph kya.3 The IME treats kya as a unit and emits one host plus one small kana in a single conversion step:

きゃく3
"guest"

To force a bare small kana when documenting yōon itself, use the x (or lt) prefix: xya yields ゃ, xyu yields ゅ, xyo yields ょ.

Treating 学校 as a yōon example

学校 (がっこう, gakkō) is a famous false friend for yōon practice because its second kana is half-size. But the half-size kana is the small つ that marks sokuon (held silence and consonant gemination), not one of the small y-kana.14 Yōon palatalizes the preceding consonant. Sokuon doubles the following one. They share a visual cue and nothing else. The genuine yōon example to anchor the section is:

教科書きょうかしょ1
"textbook"

Romanization-system whiplash across older textbooks

If you studied with an older Japanese-government textbook, you may see sya, tya, zya for what later resources call sha, cha, ja.13 Both are valid romanizations of the same kana. The kana itself is canonical; the romanization is a notation choice. J-Compass uses Hepburn throughout because it maps to English spelling intuition and is what most learners encounter outside the classroom.

Why yōon dominates kanji compounds

Yōon entered the language through Sino-Japanese borrowings from Middle Chinese during the Old and Early Middle Japanese periods, importing palatalized syllables that native Japanese phonotactics did not permit.8 Native (wago, kun'yomi) vocabulary largely lacks them. On'yomi compounds saturate them. Time spent on yōon at pre-N5 pays dividends across every later JLPT level because the contracted-sound mechanic underwrites most compound vocabulary.

The small kana was standardized only in the 20th century

Historical kana orthography did not visually distinguish yōon from a two-mora vowel sequence. The reading was inferred from context.1 The half-size convention was standardized through the 1946 現代かなづかい reforms and codified in the 1986 Cabinet Notice 「現代仮名遣い」.4 Unicode encodes the three small kana as U+3083 (ゃ), U+3085 (ゅ), and U+3087 (ょ) in the Hiragana block.6

Handwriting yōon in genkō-yōshi

Write the small kana noticeably smaller than the host and shift it down-right slightly in horizontal text. In genkō-yōshi (squared writing paper), the small kana sits in the upper-right quadrant of its own cell, sharing the row with the host.4 The same positioning rule covers the small つ of sokuon, so the muscle memory transfers across both conventions.

"Small means glued" as one mnemonic

A single rule covers both yōon and sokuon: when the second kana in a sequence is small, it fuses to the first into a single beat.1 For yōon, "glued" means the small や, ゆ, or よ supplies the palatalizing glide. For sokuon, "glued" means the small つ supplies held silence. The small kana always belongs to the mora before it.

See also

References

Footnotes

  1. "Yōon." Wikipedia. https://en.wikipedia.org/wiki/Y%C5%8Don 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39

  2. "Hiragana." Wikipedia. https://en.wikipedia.org/wiki/Hiragana 2 3 4 5 6 7 8

  3. MIT OpenCourseWare. "Hiragana: Consonant + y + Vowel; Syllabic /n/." RES.21G-01 Kana. https://ocw.mit.edu/courses/res-21g-01-kana-spring-2010/pages/hiragana/hiragana-consonant-y-vowel-syllabic-n/ 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

  4. 文化庁. 「現代仮名遣い」(内閣告示第一号, 1986). https://www.bunka.go.jp/kokugo_nihongo/sisaku/joho/joho/kijun/naikaku/gendaikana/ 2 3 4 5 6 7 8 9 10

  5. Vance, Timothy J. The Sounds of Japanese. Cambridge University Press, 2008. 2 3 4 5

  6. Unicode Consortium. Hiragana block (U+3040–U+309F). https://www.unicode.org/charts/PDF/U3040.pdf 2

  7. 国立国語研究所 (NINJAL). 『現代日本語書き言葉均衡コーパス』(BCCWJ). https://clrd.ninjal.ac.jp/bccwj/ 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

  8. Frellesvig, Bjarke. A History of the Japanese Language. Cambridge University Press, 2010. 2 3 4 5 6 7 8